AQUASIM 2.

0 User Manual
Computer Program for the Identi cation and Simulation of Aquatic Systems

Peter Reichert

Swiss Federal Institute for Environmental Science and Technology EAWAG CH - 8600 Dubendorf Switzerland September 1998 ISBN: 3-906484-16-5

Preface
The ideas for the realization of the program AQUASIM described in this manual grew from the experiences made in a lot of interdisciplinary studies at the Swiss Federal Institute for Environmental Science and Technology EAWAG, CH-8600 Dubendorf, Switzerland, in which I have been involved. It is not possible to mention all persons, who contributed with the discussion of their data interpretation and modelling problems to the concepts of this program. By far the largest in uence to the concepts of this program are due to Oskar Wanner and Jurg Ruchti. The large number of common data analysis and parameter estimation projects with Oskar Wanner let us recognise the usuefulness of a more universal program than those available at that time. Jurg Ruchti raised my interest in object-oriented programming and for the programming language C++ that was used for the implementation. The discussions with him signi cantly improved the design of the program. The realization of the program BIOSIM speci cally designed for bio lm modelling together with Oskar Wanner and Jurg Ruchti had also an important in uence on this project. AQUASIM includes the functionality of BIOSIM as a special case. Version 1.0 of the AQUASIM was documented in a technical report that contained information on modeling in general, on the selection of program tasks, on numerical algorithms, on object-oriented implementation concepts and on examples of program application Reichert, 1994b. The user manual with a brief tutorial was given in the appendix of this report. Because of the addition of a new variable type probe variable, of several new compartments advective-di usive reactor, saturated soil column, lake, signi cant extensions of the bio lm reactor compartment, and several new features for simulation and batch processing, this user manual got out of date. In addition, a new user interface for the most widely used platform Microsoft Windows, made the use of the program more comfortable. Because most users are only interested in the use of the program and not in the implementation concepts, I decided to write a new user manual and, as a separate volume, a new, more attractive tutorial Reichert, 1998. In this new user manual the equations solved by the program are given in the same chapter as the program use is described. This should facilitate the understanding of what the program does. For persons interested in numerical methods or in the implementation concepts, the technical report is still the most complete source of information. In addition, a brief description of the major program features Reichert, 1994a and a summary of the implementation concepts Reichert, 1995 are also available. I would like to thank Jurg Ruchti for the implementation of the formula variables and the plotting facilities, and Werner Simon for the realization of the saturated soil column compartment. Furthermore, Oskar Wanner contributed to the design of the extensions of the bio lm compartment, Claudia Fesch and Stefan Haderlein to the design of the

ii soil column compartment, and Gerrit Goudsmit and Johny Wuest to the design of the lake compartment. G rard Mohler, Bouziane Outiti and Raoul Scha ner gave support e in solving technical problems involving di erent hardware platforms, and Martin Omlin A introduced me into the L TEX system used for typesetting this manuscript. In addition, many program users gave hints on program errors and on possibilities for improvements. Most parts of this manual have been newly written. I apologize for all errors that it may contain. If you detect errors or unclear paragraphs, please send a note to [email protected]. I will improve this manual during the next years and try to eliminate as many errors as possible. Peter Reichert, September 1998

Contents
1 Introduction 2 File Handling 3 Model Formulation
3.1 Variables . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 State Variables . . . . . . . . . . . . . . . . 3.1.2 Program Variables . . . . . . . . . . . . . . 3.1.3 Constant Variables . . . . . . . . . . . . . . 3.1.4 Real List Variables . . . . . . . . . . . . . . 3.1.5 Variable List Variables . . . . . . . . . . . . 3.1.6 Formula Variables . . . . . . . . . . . . . . 3.1.7 Probe Variables . . . . . . . . . . . . . . . . 3.2 Processes . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Dynamic Processes . . . . . . . . . . . . . . 3.2.2 Equilibrium Processes . . . . . . . . . . . . 3.3 Compartments . . . . . . . . . . . . . . . . . . . . 3.3.1 Mixed Reactor Compartment . . . . . . . . 3.3.2 Bio lm Reactor Compartment . . . . . . . 3.3.3 Advective-Di usive Reactor Compartment . 3.3.4 Saturated Soil Column Compartment . . . 3.3.5 River Section Compartment . . . . . . . . . 3.3.6 Lake Compartment . . . . . . . . . . . . . . 3.4 Links . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 Advective Link . . . . . . . . . . . . . . . . 3.4.2 Di usive Link . . . . . . . . . . . . . . . . . 3.5 Numerical Parameters . . . . . . . . . . . . . . . . 3.6 Deleting Calculated States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 7
. 12 . 14 . 15 . 17 . 18 . 21 . 23 . 26 . 27 . 28 . 30 . 32 . 34 . 41 . 59 . 70 . 85 . 98 . 125 . 126 . 129 . 132 . 135

4 Simulation and Data Analysis

4.1 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 4.2 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 4.3 Parameter Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 iii

137

5 Visualization of Results

161

iv

CONTENTS
6.1 Character Interface Version . . . . . . . . . . . . . . . . . . . . . . . 6.2 Batch Version . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Problem Self-Diagnosis . . . . . . . . . . . . . . . . . . . . . . 6.3.2 Finding Help in the AQUASIM User Group . . . . . . . . . . 6.3.3 Reporting Program Bugs and Suggestions for Improvements . . . . . . . . . . . . . . . . . . . . 171 . 173 . 176 . 177 . 187 . 188

6 Appendix

171

Chapter 1

Introduction
The program AQUASIM was designed for the identi cation and simulation of aquatic systems in the laboratory, in technical plants and in nature. This user manual describes the equations solved by this program, the methods of systems analysis that are available and program handling. An additional document contains a series of extensively documented tutorial examples Reichert, 1998. A brief survey on the capabilities of version 1.0 of the program Reichert, 1994a as well as a description of program implementation techniques Reichert, 1995 can be found in scienti c journals. Finally, there exists an extensive technical report on all aspects of version 1.0 of the program Reichert, 1994b. Examples of program applications can be found in the following publications: Mixed reactor systems: von Schulthess et al., 1994; Wild et al., 1994; Reichert et al., 1995; Siegrist et al., 1995; Wild et al., 1995; von Schulthess and Gujer, 1996; Uehlinger et al., 1996; Kuba et al., 1996; Murnleitner et al., 1997; Novack and Sigg, 1997; Glod et al., 1997a; Musvoto et al., 1997; Eisenbeis et al., 1997; Peeters et al., 1997; Glod et al., 1997b; Filipe and Daigger, 1997. Wanner et al., 1994; Janning et al., 1995; Horn and Hempel, 1995; Wanner et al., 1995; Wanner, 1996; Wanner and Reichert, 1996; Reichert and Wanner, 1997; Vrany et al., 1997; Horn and Hempel, 1997; Mirpuri et al., 1997; Arcangeli and Arvin, 1997b; Arcangeli and Arvin, 1997a; Sanderson and Stewart, 1997; Suci et al., 1998; Beaudoin et al., 1997. von Gunten et al., 1997. Simon et al., 1997; Fesch et al., 1998b; Fesch et al., 1998a. 1

Bio lm systems:

Advective-di usive reactor systems: Saturated soil column systems:

2 River systems: Lake systems:

CHAPTER 1. INTRODUCTION
Londong et al., 1994; Albrecht et al., 1995; Jancarkova et al., 1997; Maryns and Bauwens, 1997. .

An updated list of references of AQUASIM applications can be found at the EAWAG home page at http: www.eawag.ch.

Program Design

Comparison of measurements with model calculations is the most important method of testing theories in the natural sciences. Most mathematical models of environmental systems consist of a set of nonlinear ordinary or partial di erential equations. A computer program which solves these equations numerically is usually required for calculating model predictions. Most programs available for this purpose can be put into one of three categories: universal simulation software; environmental simulation programs; and system identi cation programs. Universal simulation software is very exible with regard to model formulation, but it is di cult to use, especially for non-specialists. Environmental simulation programs are much easier to handle, but they usually implement a speci c model selected by the designer of the program. This makes their use for the comparison of di erent models impossible. Finally, system identi cation programs provide important tools for model comparison and parameter estimation, but the class of models considered in these programs is in most cases restricted to linear or algebraic models, and models cannot be formulated in a way familiar to environmental scientists. Although the classi cation of simulation programs into these categories is not strict and there are also a few programs that cover tasks belonging to more than one of these categories, a universal identi cation and simulation program is not yet available. The intention behind

the design of the program AQUASIM was to provide a more universal identi cation and simulation tool for a class of aquatic systems important in the environmental sciences. An additional important program design criterion was user-friendliness, which was achieved not only by providing a graphical user interface, but also by utilizing a communication "language" familiar to environmental scientists. AQUASIM is extremely exible in allowing the user to specify transformation processes, and, in addition to perform simulations for the user-speci ed model, it provides elementary methods for parameter identi ability analysis, for parameter estimation and for uncertainty analysis.
Version 1.0 of AQUASIM was developed in the years 1991-1994 in the Computer and Systems Sciences department of the Swiss Federal Institute for Environmental Science and Technology EAWAG, CH-8600 Dubendorf, Switzerland, and it is maintained and extended since then. The program was designed mainly for internal use in research and teaching, but is also available to the public. Information on the newest developments of AQUASIM can be found at the EAWAG home page at http: www.eawag.ch.

Program Tasks

AQUASIM is a program for the identi cation and simulation of aquatic systems. It performs the four tasks of simulation,

3 identi ability analysis, parameter estimation, uncertainty analysis. Due to the similarity of the mathematical techniques involved, identi ability and uncertainty analyses are combined to yield sensitivity analysis. The rst task of AQUASIM is to allow the user to perform model simulations. By comparing calculated results with measured data, such simulations reveal whether certain model assumptions are compatible with measured data. The existence of systematic deviations between calculations and measurements provides a hint that additional important processes may have to be considered, or corrections must be made in the way processes are formulated. AQUASIM allows the user to change model structure and parameter values easily. AQUASIM's second task is to perform sensitivity analyses with respect to a set of selected variables. This feature allows the user to calculate linear sensitivity functions of arbitrary variables with respect to each of the parameters included in the analysis. These sensitivity functions help in assessing the identi ability of model parameters identi ability analysis. Furthermore, the derivatives calculated in sensitivity analyses allow the user to estimate the uncertainty in any variable according to the linear error propagation formula. The calculation of the contribution of each parameter to the total uncertainty facilitates the detection of major sources of uncertainty uncertainty analysis. The third important task of AQUASIM is to perform parameter estimations automatically for a given model structure using measured data. This is not only important for obtaining neutral estimates of parameters, but is also a main prerequisite for e ciently comparing di erent models. Several calculations, each of them describing a single experiment with the possibility for several target variables, as well as universal and experiment-speci c model parameters, can be combined to a single parameter estimation process. The quantitative measure of the deviation between model calculations and measurements, which is minimized by the parameter estimation algorithm, is useful for statistically assessing the adequacy of the model.

User Interfaces

Three versions of the program AQUASIM with di erent user interfaces are provided. The window interface version, aquasimw, uses the machines own graphical user interface. The use of this version is strongly recommended for editing models, for de ning sensitivity analyses and parameter estimations, for specifying plot de nitions, for performing short calculations and for viewing results. The second version is the character interface version, aquasimc. This version is intended for users having a simple terminal without graphical capabilities. It provides all features available in the window interface version except the capability of plotting results directly on the screen listing results and preparing plots for printing, however, is also possible with this program version. The third version is the batch version, aquasimb, which is designed for submitting long calculations as batch jobs it should be noted that simulations, and especially sensitivity analyses and parameter estimations, may require much computation time. This version allows the user to start a calculation for an AQUASIM system, de ned with one of the interactive program versions, by specifying one simple command line. It is also possible to specify a series of AQUASIM jobs on a command le so that the consecutive execution of calculations together with

CHAPTER 1. INTRODUCTION

listing and plotting results can be combined to a single batch job. In the batch version of AQUASIM, models cannot be modi ed.

Hardware Platforms and Operating Systems

AQUASIM is written in the standardized object oriented programming language C++. There is a strict separation between the core program and the di erent user interface layers. There exist two versions of the user interface layer for the window interface version of the program. The rst uses a graphical user interface library, which is available for various hardware platforms and operating systems. This program design makes AQUASIM highly portable. A second implementation of the user interface version of the window interface version is speci cally designed for the Microsoft Windows operating system. The character interface version and the batch version can be compiled on nearly any platform and operating system without the need of special libraries. Table 1.1 gives a survey on all currently supported computing platforms.

Hardware
Sun SparcStation IBM RS 6000 HP 700 series DEC Alpha series

Operating System Window System

Solaris 2.x OpenWindows Motif 1.2 Windows 3.1 Win32s native native native

Program Version w c b
X X X X X X X X X X X X X X X

AIX 3.2.x HP-UX 9.x VMS 6.x DEC Unix Intel 80486 Pentium MS-DOS 5 or 6 Windows 95 Windows NT Apple Power Macintosh MacOS 7.x

X X X X X X X X X

Table 1.1: Computing platforms supported by the current AQUASIM version.

Organization of this Manual

Throughout this user manual it is assumed that the user is familiar with system speci c handling of menus, windows, list boxes, buttons, etc.. The description mainly concentrates on the window interface version of the program. The functionality of the character interface version is the same with the exception that it is not possible to plot graphics to the screen with this program version. In section 6.1 a brief introduction to the character interface version is given. In section 6.2 it is shown how batch jobs can be submitted. Figure 1.1 shows the main window of AQUASIM with the header, the menu bar and a button bar which facilitates the access to the most important menu commands from left to right: File!New, File!Open, File!Save, Edit!Cut inactive, Edit!Copy inactive, Edit!Paste inactive, Edit!System, Edit!Delete States, Calc!Simulation, Calc!Sensitivity Analysis, Calc!Parameter Estimation, View!Results, View!Close Dialogs. The commands in the four menus File, Edit, Calc and View are described in the chapters 2-5 of this user manual. The menu `File' chapter 2 is used for saving, loading and printing the current AQUASIM system, which consists of

Figure 1.1: Main window of AQUASIM. the mathematical model, measured data, de nitions of sensitivity analyses and parameter estimations, plot de nitions and calculated states. With the aid of the menu `Edit' chapter 3, the mathematical model and measured data can be entered and edited. The menu `Calc' chapter 4 is used to de ne and perform simulations, sensitivity analyses and parameter estimations. Finally, the menu `View' chapter 5 is used to specify plot de nitions and to list and plot results. The appendix chapter 6 contains additional information speci c to the di erent program versions and some hints for troubleshooting.

How to Proceed

The following procedure is recommended for learning to use the program AQUASIM: 1. Read the introduction to this manual chapter 1 to obtain a general idea of program concepts and capabilities. 2. Skim over the chapters 2-5 to increase your knowledge of program concepts and to start learning to use the program interface. If you plan to work with the character interface version, read section 6.1 also. 3. Study the tutorial exercises described in a separate document Reichert, 1998, and carefully read the corresponding sections of the chapters 2-5 of this user manual if you have problems in understanding the solutions. 4. Start using the program as a scienti c and or didactic tool. It may be helpful to implement the rst model by modifying one of the example applications or one of the tutorial system les instead of starting from scratch. If problems arise, look at section 6.3 for hints on troubleshooting.

CHAPTER 1. INTRODUCTION

Chapter 2

File Handling
Fig. 2.1 shows the menu File of AQUASIM. The rst 6 items of this menu are used to save

Figure 2.1:

File

menu.

and load AQUASIM systems, consisting of the user-speci ed mathematical model, measured data, de nitions of sensitivity analyses and parameter estimations, plot de nitions and calculated states. The item `New' is used to free memory from the current system, to allow the user to enter a new system. With the item `Open', a system previously stored with Save or Save As can be reloaded. Clicking the item `Close' results in deletion of the current system from memory, however, saved versions are not changed. The item `Save' is used to save a system, overwriting its old version on the disk. With the item `Save As' the current system can be saved under a new name. By clicking the item `Revert to Saved', the saved version of a system edited interactively can be reloaded. Note, that the items Close, Save, Save As and Revert to Saved are inactive, if no system has been opened or interactively entered. For a system interactively entered and not yet saved, the item Save results in the same operation as Save As and therefore allows to specify a le name. Before speci cation of a le name by loading or saving an AQUASIM system, 7

CHAPTER 2. FILE HANDLING

the item Revert to Saved is inactive. Furthermore, as long as a loaded system is not yet modi ed, the items Save and Revert to Saved are inactive. AQUASIM system les should not be edited with other programs, because such an attempt can result in inconsistent or unreadable les. AQUASIM system les can be transfered between all supported platforms using text ASCII data transfer. The le format is also compatible with electronic mail; mailed system les can directly be opened on any platform together with their mail headers. To keep a reasonable le size, it is recommended to delete calculated states before saving to an AQUASIM system le that is planned to be included in a mail message. The menu item `Print Options' allows to select the print le format. As shown in Fig. 2.2, a long and a short format can be selected. In the long format, nearly

Figure 2.2: Dialog box for editing print options. all user speci cations are listed, whereas the short form allows to have a compact listing of the most essential model elements. To facilitate program portability, the menu item `Print to File' does not directly print the system de nitions, but only writes them to a text le, the name of which can be speci ed by the user. Such text les have then to be submitted to a printer by the user either directly or after loading them into an editor or a text processing program. Printing an AQUASIM system is very useful, because the clear arrangement of all system de nitions facilitates checking user input or understanding the meaning of objects loaded from an AQUASIM system le created by someone else. Note, however, that an AQUASIM system cannot be reloaded from a print le. The menu item `About' gives information on the installed program version. This information is also written to any output le written by AQUASIM. Finally, clicking the menu item `Exit' results in program termination. If system de nitions have been edited or if states have been calculated or deleted, the user is asked to save the changes. For each interactive AQUASIM session, a log le with the name aquasim.log is written to the startup directory of the program in the batch version, the name of the log le can be speci ed by the user. This log le contains information on the progress of calculations. In the case of normal program termination, the log le can be ignored and deleted; in the case of problems during calculation, the information provided in the log le may help locating the problem. Restarting AQUASIM in the same directory as before, leads to overwriting the old version of the log le, so that only the log les of the most recent sessions in di erent directories are available unless an old log le has been renamed.

Chapter 3

Model Formulation
In the program AQUASIM, a model consists of a system of ordinary and or partial di erential equations and algebraic equations, which deterministically describes the behaviour of a given set of important state variables of an aquatic system. The di erential equations for water ow and substance transport can be selected by the choice of environmental or technical compartments, which can be connected by links. The source terms of these equations, which describe the e ect of transformation processes, can be freely speci ed by the user. The de nition of such transformation processes follows closely the notation of biochemical processes, as it is familiar to environmental scientists and engineers. The de nition of processes, compartments and links is done with the aid of variables, which represent objects taking a possibly context-sensitive numerical value. Fig. 3.1 visualizes the mutual depedencies between the four subsystems of variables, processes, compartments and links. It is evident, that the variables form the basic subsystem required for

Links Compartments Processes Variables

Figure 3.1: Main elements of model structure. the formulation of processes, compartments and links. Processes must be de ned before they can be activated in compartments. Finally, links can be used to connect compartments that are already de ned. After a short overview of the four subsystems of the AQUASIM model structure, in this chapter, the de nition of objects of these subsystems is explained in detail sections 3.1 to 3.4. All these de nitions together form the model 9

10

CHAPTER 3. MODEL FORMULATION

used by AQUASIM for simulation and data analysis. The basic subsystem of the AQUASIM model structure is the system of variables. Variables are objects which are characterized by the property of taking a numerical value. This value may depend on the values of other variables. Six types of variables are distinguished: State variables are used to describe properties of water or of a surface in contact with water to be calculated by the model. Program variables make auxiliary quantities used in the program available to the system of variables. Constant variables and real list variables are used to provide measured quantities for use in the system of variables. In addition, constant variables are used as model parameters in sensitivity analyses and parameter estimations. Variable list variables and formula variables are used to build functional relations out of other variables. Finally, probe variables make the values of variables evaluated at a given location in a compartment globally available. The system of variables serves as a pool of variables for the formulation of the other subsystems. The next subsystem of the AQUASIM model structure is the system of processes. Two types of processes are distinguished: Dynamic processes implement transformation or transfer processes, which are characterized by a common process rate and by individual stoichiometric coe cients describing the relative e ect to di erent variables. Time evolution of variables a ected by dynamic processes is determined by the solution of di erential equations. The second type of processes are equilibrium processes, which determine the value of the corresponding variables by the solution of algebraic equations. Such processes are used to model processes which are so fast, that the corresponding variables can always be approximated to take their current equilibrium values. The variables of the system of variables may be used and are needed to formulate processes. The next subsystem of the AQUASIM model structure is the system of compartments. This subsystem is designed to spatially divide the system under investigation. The following types of compartments are implemented in the current version of the program: Mixed reactor compartments are used to describe systems that can be approximated by an arrangement of well-mixed domains e.g. stirred reactors, mixed lakes, etc., bio lm reactor compartments are used to describe the growth and population dynamics of bio lms in which substrate gradients over depth are important, advective-di usive reactor compartments can be used to describe systems with a longitudinal given water ow e.g. plug ow reactors, rivers with given water ow, etc., saturated soil column compartments are used to model advective-dispersive transport, exchange with stagnant pore volumes, adsorption and transformation of substances in saturated soil columns, river section compartments are used to describe hydraulics, transport and transformation processes in rivers, and lake compartments are used to model strati cation, mixing, transport and transformation processes in horizontally well-mixed lakes. The last subsystem of the AQUASIM model structure is the system of links. The objects of this subsystem are used to connect the compartments to the desired spatial con guration. To connect the compartments listed above, two types of links are distinguished: Advective links describe water ow and advective substance transport between compartments. These links can not only directly connect compartments, but also bifurcations and junctions can be built. Di usive links model di usive boundary layers or membranes between compartments. These elements can be di usively penetrated by certain substances. The menu Edit of AQUASIM shown in Fig. 3.2 is based on the model structure

11 described above. Each of the four menu items `Variables', `Processes', `Compart-

Figure 3.2:

Edit

menu.

ments' and `Links' opens or activates if it is already open a modeless dialog box

containing a list of objects already de ned and control elements for de ning new objects and for editing and deleting objects of the corresponding subsystem. If the screen is large enough, it is recommended to open all these dialog boxes together to accelerate editing the model this can be done by selecting the `System' item in the Edit menu. The hierarchy of dialogs controlled by each of these dialog boxes is described in the following sections 3.1 to 3.4. As additional options, the Edit menu allows the user to change the values of `Numerical Parameters' and to `Delete States' calculated by the program. These options are described in the sections 3.5 and 3.6, respectively.

12

CHAPTER 3. MODEL FORMULATION

3.1 Variables
The basic objects for the formulation of models are variables. Variables are identi ed by their name. They are characterized by the property of taking a possibly context-sensitive numerical value. There are four main ranges of application of variables: Variables can be used for quantities to be determined by the model e.g. by the solution of algebraic or di erential equations or which have a prede ned meaning in a compartment e.g. time and space coordinates, they can be used to provide data e.g. model parameters or measured data series, they can be used to build functions depending on other variables e.g. for the speci cation of process rates or stoichiometric coe cients or they can be used as probes which make the values of other variables evaluated at a given location in a compartment globally available. According to these four categories, seven types of variables are distinguished: State Variables represent concentrations or other properties to be determined by a model according to user-selected transport and user-de ned transformation processes. Program Variables make quantities such as time, space coordinates, discharge, etc. that are used for model formulation available as variables. Constant Variables describe single measured quantities that can also be used as parameters for sensitivity analyses or parameter estimations. Real List Variables are used to provide measured data or to formulate dependencies on other variables with the aid of interpolated data pairs. Variable List Variables are used to interpolate between other variables at given values of an arbitrary argument e.g. for multidimensional interpolation. Formula Variables allow the user to build new variables as algebraic expressions of other variables. Probe Variables make the values of other variables evaluated at a given location in a compartment globally available. Figure 3.3 shows the dialog box used for editing variables. This dialog box is opened with the Variables command in the Edit menu shown in Figure 3.2. It is of modeless type in order to facilitate the editing process. The names of all variables already de ned are listed alphabetically in the list box of this dialog box. The type of the currently selected variable is indicated at the bottom of the dialog box. The buttons of the dialog box allow the user to perform the following operations with variables: By clicking the button `New', new variables can be created from scratch. Alternatively, by clicking the button `Duplicate' the selected varible can be duplicated. With the button `Edit' or by double-clicking the variable name in the list box, a variable can be edited. The type of a variable can be changed by clicking the button `Edit Type'. During this procedure, type-speci c data of the variable gets lost. Furthermore, by selecting two variables and clicking `Exchange', it is possible to exchange two variables in all other variables, processes, compartments, links and de nitions of sensitivity analyses and parameter estimations, where they occur as arguments. This feature allows the user to quickly change models without losing data. Finally, the button `Delete' allows the program users to delete variables. Deletion of a variable is only possible, if the variable is not an argument of another variable, of a process, of a compartment, of a link or of a de nition of a sensitivity analysis or of a parameter estimation. The buttons Duplicate, Edit, Edit Type, Exchange and Delete are inactive

3.1. VARIABLES

13

Figure 3.3: Dialog box for editing variables. as long as no variable is selected. Clicking the `Close' button results in closing this dialog box. It can be reopened by clicking the Variables command in the Edit menu shown in Fig. 3.2. The variables de ned with the subdialogs to the dialog box shown in Fig. 3.3 serve as a pool of variables for use in other AQUASIM objects. A new variable may depend on any variables already de ned circular references are not allowed. It is important to de ne all necessary variables before starting to de ne an object of one of the other subsystems of processes, compartments and links. After clicking one of the buttons New or Edit Type in the dialog box shown in Fig. 3.3 the variable type can be selected in the dialog box shown in Fig. 3.4. The seven types of

Figure 3.4: Dialog box for selecting the type of a variable. variables shown in this selection box are described in more detail in the following seven subsections.

14

CHAPTER 3. MODEL FORMULATION

3.1.1 State Variables

State variables describe properties of water or of a surface in contact with water e.g. temperature, masses or concentrations of dissolved or suspended substances or of substances attached to a surface. State variables obtain their meaning indirectly by the processes in which they are involved. Figure 3.5 shows the dialog box used for de ning or editing a state variable. As each

Figure 3.5: Dialog box for editing a state variable. variable, a state variable needs a unique `Name' as an identi er. A name of a variable consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The following reserved names are not allowed as variable names: div, mod, and, or, not, if, then, else, endif, pi, sin, cos, tan, asin, acos, atan, sinh, cosh, tanh, deg, rad, exp, log, ln, log10, sign, abs, sqrt, min, max. To improve documentation of variables, a `Description' and a `Unit' can be speci ed. There are two main types of state variables: The values of `dynamic' state variables are calculated as solutions to di erential equations according to the transport processes determined by the choice of the compartment type and to the transformation rates speci ed by the user. `equilibrium' state variables are used to describe quantities, the transformation processes of which are much faster than those of other variables, so that they can always be approximated to take the value corresponding to the current equilibrium state of their transformation processes. These equilibrium states depend on the values of the other variables and are given as the solution to algebraic equations provided by the user of the program. Dynamic state variables are further divided into dynamic `volume' state variables and dynamic `surface' state variables. Dynamic volume state variables are used to describe concentrations of substances transported with the water ow and quanti ed as mass per unit volume of water, whereas dynamic surface state variables are used to describe substances which are not transported with the water ow. Usually, this type of state variables is used to describe substances attached to a surface, which are quanti ed as total mass, as mass per unit length or as mass per unit of surface area surface density. The distinction into volume and surface variables is not needed for equilibrium state variables. The edit elds `Rel. Accuracy' and `Abs. Accuracy' can be used to specify the precision of the numerical calculations. The integration algorithm uses the absolute accuracy plus the relative accuracy times the current value as an error criterion

3.1. VARIABLES

15

to control the size of the time step. Therefore, not both of these accuracies are allowed to be zero, but pure absolute or pure relative error criteria are possible. It is important to specify reasonable values for these accuracies in order to obtain good behaviour of the integration algorithm. Good behaviour of the numerical algorithms is usually achieved, if the absolute accuracy and the product of a typical value of the state variable times the relative accuracy both are 4 to 6 orders of magnitude smaller than typical values of the state variable. See section 3.5 for more information on parameters of the numerical algorithms used in AQUASIM.

3.1.2 Program Variables

Program variables refer to prede ned quantities of the modelled system. From a mathematical point of view, program variables can represent independent variables time or space, parameters calculation number, compartment index, etc., or solutions to differential-algebraic systems of equations discharge, reactor volume, etc.. The idea of program variables is to make the corresponding quantities, which anyway are present in model formulation, available for use in the system of variables. In some cases program variables can also be used to specify initial conditions within compartments cf. section 3.3. Besides program variables for time and space coordinates and for compartment speci c quantities, the set of program variables also includes a Calculation Number, which allows the user to distinguish di erent calculations. Figure 3.6 shows the dialog box used for de ning or editing a program variable. As

Figure 3.6: Dialog box for editing a program variable. each variable, a program variable needs a unique `Name' as an identi er. A name of a variable consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The following reserved names are not allowed as variable names: div, mod, and, or, not, if, then, else, endif, pi, sin, cos, tan, asin, acos, atan, sinh, cosh, tanh, deg, rad, exp, log, ln, log10, sign, abs, sqrt, min, max. To improve documentation of variables, a `Description' and a `Unit' can be speci ed. With the aid of the list selection box `Reference to' the user can select the quantity to which the program variable refers to. It is not possible to create more than one program variable referring to the same quantity. Table 3.1 lists the sigini cance of all program variables considered in the current program version. Note that not all program variables are available in all compartments and links. In section 3.3 for each compartment a list of available program variables is given. The program variable Calculation Number is

16

CHAPTER 3. MODEL FORMULATION

a non-negative integer used for distinguishing di erent calculations cf. section 4. The program variables Compartment Index, Zone Index and Link Index are used to make variables depend on compartments, zones within compartments and links. The values taken by the program variable Zone Index depends on the compartment, the values of the program variables Compartment Index and Link Index can be set in the dialog boxes for the de nition of compartments and link cf. sections 3.3 and 3.4. All other program variables have a physical meaning. Note that it depends on the compartment or link type, which program variables are de ned. Program variables always return current values of the corresponding physical quantity as a function of simulation time and space coordinate within a compartment.
Calculation Number Time Compartment Index

Zone Index Link Index Discharge Water Fraction Space Coordinate X

Space Coordinate Z Reactor Volume Bulk Volume Biofilm Thickness Growth Velocity of Biofilm Interface Velocity of Biofilm Detachment Velocity of Biofilm Attachment Velocity of Biofilm Water Level Elevation Cross Sectional Area

Perimeter Length

Identi er for calculations all compartments and links; value set in the dialog boxes shown in Figs. 4.5 and 4.17. Simulation time all compartments and links. Identi er for compartments all compartments; value set in the dialog boxes shown in Figs. 3.21, 3.29, 3.41, 3.53, 3.67 and 3.78. Identi er for zones within compartments all compartments; cf. section 3.3 for a description of possible values. Identi er for links all links. Volumetric ow rate all compartments, advective link. Volumetric fraction of water all compartments; in some compartments always equal to 1. Space coordinate along the compartment advective-di usive reactor compartment, saturated soil column compartment, river section compartment. Depth coordinate in the compartment bio lm reactor compartment, lake compartment. Total volume of the reactor mixed reactor compartment, bio lm reactor compartment. Volume of mixed water zone mixed reactor compartment, bio lm reactor compartment. Thickness of the bio lm bio lm reactor compartment. Advective velocity of bio lm solid matrix bio lm reactor compartment. Velocity of the interface between bio lm and bulk uid bio lm reactor compartment. Detachment velocity of particles from the bio lm surface bio lm reactor compartment. Attachment velocity of particles onto the bio lm surface bio lm reactor compartment. Elevation of water level above an absolute reference level river section compartment. Area of water body perpendicular to the ow direction advectivedi usive compartment, saturated soil column compartment, river section compartment. Length of the interface between water and the river bed perpendicular to the ow velocity river section compartment.

3.1. VARIABLES
Surface Width

17
Length of the interface between water and the atmosphere perpendicular to the ow velocity river section compartment. Nondimensional friction force: Friction force divided by gravity force river section compartment. Density of the water lake compartment. Gradient of the cross-sectional area as a function of depth lake compartment. Stability frequency of the water column lake compartment. Velocity of horizontal wind-driven ow lake compartment. Turbulent kinetic energy TKE per unit mass of water in the lake lake compartment. Production of TKE due to shear of horizontal velocity lake compartment. Production or loss of TKE due to density di erences lake compartment. Dissipation of turbulent kinetic energy lake compartment. Total energy stored in Seiche motion lake compartment.

Friction Slope Density Area Gradient Brunt Vaisala Frequency Horizontal Velocity Turbulent Kinitic Energy TKE Shear Production of TKE Buoyancy Production of TKE Dissipation Energy of Seiche Oscillation

Table 3.1: Signi cance of program variables.

3.1.3 Constant Variables

Constant variables can be used to describe single measured quantities consisting of a value and its accuracy characterized by the standard deviation. Alternatively, constant variables can be used as model parameters the values and standard deviations of which are estimated by the program. In this case the minimum and the maximum bound the legal range of values. For simulations, only the value of a constant variable is used. It remains constant during the simulation. The standard deviation together with the legal range is used during sensitivity analyses. Figure 3.7 shows the dialog box used for de ning or editing a constant variable. As each variable, a constant variable needs a unique `Name' as an identi er. A name of a variable consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The following reserved names are not allowed as variable names: div, mod, and, or, not, if, then, else, endif, pi, sin, cos, tan, asin, acos, atan, sinh, cosh, tanh, deg, rad, exp, log, ln, log10, sign, abs, sqrt, min, max. To improve documentation of variables, a `Description' and a `Unit' can be speci ed. The user has to specify the `Value' of the variable which is used for simulations. In sensitivity analyses the `Standard Deviation' is used to investigate the in uence of uncertainty of model parameters to simulation results. The `Minimum' and `Maximum' bound the range of legal values. These bounds also hold for internal changes during sensitivity analyses and parameter estimations. For each constant variable, it can be decided, if it is `active for sensitivity analysis' and if it is `active for parameter estimation'. These states can also be accessed with the aid of the dialog boxes shown in Figs. 4.11 and 4.16.

18

CHAPTER 3. MODEL FORMULATION

Figure 3.7: Dialog box for editing a constant variable.

3.1.4 Real List Variables

Quantities measured as a function of another variable, e.g. time series or spatial pro les, are represented by real list variables. For the de nition of a real list variable a variable representing the argument, a list of argument-value data pairs, the standard deviations of the data and an interpolation method must be speci ed. The standard deviations can be given as global relative and absolute standard deviations or as individual standard deviations for all data values. Real list variables are usually evaluated as follows: In a rst step, the variable given as the argument is evaluated. Then, the value of the variable is calculated by employing the selected interpolation method at the value of the argument as follows: Linear interpolation: If the value of the argument is smaller than the argument of the rst list element, the value of the rst list element is returned. If the value of the argument is larger than the value of the argument of the last list element, the value of the last list element is returned. If the value of the argument is within the range of the arguments of the list, the value on the connecting straight line between neighbouring data points corresponding to the argument is returned.

Cubic spline interpolation: If the value of the argument is smaller than the argument of the rst list element, the value of the rst list element is returned. If the value of the argument is larger than the value of the argument of the last list element, the value of the last list element is returned. If the value of the argument is within the range of the arguments of the list, the interpolated value is given by cubic polynomials between neighbouring data points, which are determined by the conditions of continuous rst and second derivatives at inner data points and by zero rst derivative at

3.1. VARIABLES

19

10 8 6 4 2 0 0 10 8 6 4 2 0 0 2 4 6 8 10 2 4 6 8 10

data linear spline smooth (1.0) smooth (1.5)

data linear spline smooth (0.4) smooth (1.0)

Figure 3.8: Comparison of interpolation and smoothing methods. the end points. Smoothing: The values are de ned by a curve smoothing the data points. This curve is given by the t of a parabola through neighboring data points. For this t, the data points are weighted with a normal distribution with a standard deviation chosen by the user smoothing width and centered at the actual value of the argument. The larger the width of this distribution, the smoother the behavior of the curve.

Fig. 3.8 shows interpolation of two real list variables with di erent interpolation and smoothing methods. Note that spline interpolation may lead to undesired oscillations in case of very abrupt changes in the data series. An alternative use of real list variables is their use as target variables for parameter estimations. This is only possible, if the argument of the real list variable is either the program variable for time or the program variable describing the space coordinate of the compartment in which the comparison takes place. In this case, no interpolation is performed, but the variable to be compared with the real list variable is evaluated at the positions of the data pairs and the di erences between the values of the two variables are

20

CHAPTER 3. MODEL FORMULATION

Figure 3.9: Dialog box for editing a real list variable. summed up according to equation 4.13 using the standard deviations speci ed for the real list variable. Figure 3.9 shows the dialog box used for de ning or editing a real list variable. As each variable, a real list variable needs a unique `Name' as an identi er. A name of a variable consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The following reserved names are not allowed as variable names: div, mod, and, or, not, if, then, else, endif, pi, sin, cos, tan, asin, acos, atan, sinh, cosh, tanh, deg, rad, exp, log, ln, log10, sign, abs, sqrt, min, max. To improve documentation of variables, a `Description' and a `Unit' can be speci ed. The `Argument' may be any other variable already de ned. Its value is used to determine where to interpolate the list. Standard deviations can either be given as `individual' standard deviations for all data values or as `global' `Relative Standard Deviations' and `Absolute Standard Deviations'. In the latter case, the standard deviation of a data value is calculated as the square root of the sum of the square of the absolute standard deviation plus the square of the product of the relative standard deviation times the current value of the variable. The absolute standard deviation may not be zero if some data elements of the list are zero. As for constant variables, the `Minimum' and `Maximum' bound the range of legal values. These bounds also hold for internal changes during sensitivity analyses. The list of data is built by elements consisting of a value of the argument, a value of the variable, and, in case of individual standard deviations, the

3.1. VARIABLES

21

Figure 3.10: Dialog box for reading data pairs from a text le. standard deviation of the value. The data pairs are sorted with increasing value of the argument. All data pairs have to di er in their arguments. It is possible to `Add', `Replace' and `Delete' data pairs, to `Read' them from text les tab, space or comma delimited; missing lines and values as well as text columns allowed and to `Write' them to text les. The radio buttons `linear', `spline' and `smooth' make it possible to select the interpolation technique. For each real list variable, it can be decided, if it is `active for sensitivity analysis' Fig. 3.10 shows the dialog box used for reading data pairs from a text le. This dialog box is opened by clicking the button Read in the dialog box shown in Fig. 3.9. The user can specify the data area by the `Start Row' and `End Row' and by the `Column Number of Argument', the `Column Number of Values' and the `Column Number of Standard Deviations' Standard deviation only if individual standard deviations are selected in the dialog box shown in Fig. 3.9. Furthermore, the user can choose to `Delete existing data pairs' or to add the data read from the le to the existing data.

3.1.5 Variable List Variables

Variable list variables are similar to real list variables, but instead of a value, another variable is given corresponding to each value of the argument. If these variables are variable list variables or real list variables, variable list variables can be used for multidimensional interpolation; if they are constant variables, parameter estimations of time series or of spatial pro les are possible. Figure 3.11 shows the dialog box used for de ning or editing a variable list variable. As each variable, a variable list variable needs a unique `Name' as an identi er. A name of a variable consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The following reserved names are not allowed as variable names: div, mod, and, or, not, if, then, else, endif, pi, sin, cos, tan, asin, acos, atan, sinh, cosh, tanh, deg, rad, exp, log, ln, log10, sign, abs, sqrt, min, max. To improve documentation of variables, a `Description' and a `Unit' can be speci ed. Similarly to

22

CHAPTER 3. MODEL FORMULATION

Figure 3.11: Dialog box for editing a variable list variable. real list variables, an `Argument' must be speci ed. It is possible to `Add', `Replace' and `Delete' argument-variable pairs. The `Interpolation Method' must also be selected. Look at the preceeding section on real list variables for a description of these interpolation techniques. Note that for variable list variables, spline interpolation and smoothing are not very e cient, because these methods need evaluation of all variables of the list for each evaluation of the variable list variable.

3.1. VARIABLES

23

3.1.6 Formula Variables

Formula variables allow the user to build functional relations as algebraic expressions using previously de ned variables cyclic references are not allowed. Figure 3.12 shows the dialog box used for de ning or editing a formula variable. As

Figure 3.12: Dialog box for editing a formula variable. each variable, a formula variable needs a unique `Name' as an identi er. A name of a variable consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The following reserved names are not allowed as variable names: div, mod, and, or, not, if, then, else, endif, pi, sin, cos, tan, asin, acos, atan, sinh, cosh, tanh, deg, rad, exp, log, ln, log10, sign, abs, sqrt, min, max. To improve documentation of variables, a `Description' and a `Unit' can be speci ed. An algebraic `Expression' using the previously de ned variables can be given to de ne the new variable. The formula syntax is given by an expression de ned as given below, where varident must be the name of a variable already de ned:
varident letter letter or digit digit expression relop simple expression term sign addop factor unsigned constant unsigned number

= = = = = = = = = = = = =

letter A | ... letter

f
|

letter or digit digit | | term

| Z | a | ...

| z |

0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 simple expression simple expression = |  | term term + | + | - | or varident | unsigned constant | function | not factor unsigned number unsigned integer unsigned integer | |  expression  | | | = | sign term simple expression = | | simple expression factor addop relop

factor

mulop

predefined constant . unsigned integer E sign

24
predefined constant unsigned integer mulop function arith func 1 arg ident func 1 arg

CHAPTER 3. MODEL FORMULATION

= = = = = = = = = =
pi digit ^ * | |

digit

g
| arith func 2 arg | if function  expression 

| div | mod | and

arith func 1 arg ident func 1 arg

sin | cos | tan | asin | acos | atan | sinh | cosh | tanh | deg | rad | exp | log | ln | log10 | sign | abs | sqrt ident func 2 arg min | max if condition endif expression then expression else expression  expression , expression 

arith func 2 arg ident func 2 arg if function condition

Note that this syntax makes it possible to specify algebraic expressions using variables, usual operations and elementary functions, and that even conditional branching with if then else endif constructions is possible. The trigonometric functions use radians as the unit of the argument. Table 3.2 gives an overview on the functions and constants available in formula variables.
abs acos asin atan cos cosh deg exp ln log log10 max min pi rad sign sin

Function with one argument returning the absolute value of the argument. Function with one argument returning the inverse of the cosinus function in units of radians evaluated at the argument. Function with one argument returning the inverse of the sinus function in units of radians evaluated at the argument. Function with one argument returning the inverse of the tangens function in units of radians evaluated at the argument. Function with one argument returning the cosinus function evaluated at the argument that must be given in radians. Function with one argument returning the hyperbolic cosinus function evaluated at the argument that must be given in radians. Function with one argument returning the argument converted from radians to degrees. Function with one argument returning e to the power of the argument. Function with one argument returning the natural logarithm of the argument. Function with one argument returning the natural logarithm of the argument. Function with one argument returning the logarithm with base 10 of the argument. Function with two arguments returning the maximum of the two arguments. Function with two arguments returning the minimum of the two arguments. Constant returning the value of  =3.1415926.... Function with one argument returning the argument converted from degrees to radians. Function with one argument returning the sign of the argument +1 for positive arguments, -1 for negative arguments, 0 for arguments equal to zero. Function with one argument returning the sinus function evaluated at the argument that must be given in radians.

3.1. VARIABLES
sinh sqrt tan tanh

25

Function with one argument returning the hyperbolic sinus function evaluated at the argument that must be given in radians. Function with one argument returning the square root of the argument. Function with one argument returning the tangens function evaluated at the argument that must be given in radians. Function with one argument returning the hyperbolic tangens function evaluated at the argument that must be given in radians.

Table 3.2: Functions and constants that can be used in formula variables.

26

CHAPTER 3. MODEL FORMULATION

3.1.7 Probe Variables

Probe variables are used to make the value of another variable evaluated at a given location in a compartment globally available.

Figure 3.13: Dialog box for editing a probe variable. Figure 3.13 shows the dialog box used for de ning or editing a formula variable. As each variable, a formula variable needs a unique `Name' as an identi er. A name of a variable consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The following reserved names are not allowed as variable names: div, mod, and, or, not, if, then, else, endif, pi, sin, cos, tan, asin, acos, atan, sinh, cosh, tanh, deg, rad, exp, log, ln, log10, sign, abs, sqrt, min, max. To improve documentation of variables, a `Description' and a `Unit' can be speci ed. Then a `Variable' must be selected and the `Compartment', `Zone' and the `Location' in the compartment, where the variable has to be evaluated, must be speci ed. If the check box `relative' is ticked, the location must be speci ed as relative coordinates between 0 and 1, otherwise the location must be given in absolute coordinates.

3.2. PROCESSES

27

3.2 Processes
Transformation processes can be de ned by a set of process rates, each of which describes the contribution of the process to the temporal change of the concentration of a given substance. Characteristic times of such processes may vary over several orders of magnitude. The consequence of the existence of widely varying time scales within a system is that the concentrations determined by fast processes converge so fast to their current equilibrium values which themself depend on slower processes that the transient phase is not important for the behavior of the system on the slower time scale. In such situations, a separation of time scales, which solves concentrations determined by fast processes directly for their equilibrium solution, can be advantageous. This leads to a replacement of di erential equations of fast processes by algebraic equations. Therefore, the following two types of processes are introduced: Dynamic Processes describe substance transformations the dynamics of which is important on the time scale of the simulation. Equilibrium Processes describe the e ect of very fast processes which lead to permanent equilibrium values of the corresponding state variables. Figure 3.14 shows the dialog box for editing processes. This dialog box is opened with

Figure 3.14: Dialog box for editing processes. the Processes command in the Edit menu shown in Figure 3.2. It is of modeless type in order to facilitate the editing process. The names of all processes already de ned are listed alphabetically in the list box of this dialog box. The type of the currently selected process is indicated at the bottom of the dialog box. The buttons of this dialog box allow the user to perform the following operations with processes: By clicking the button `New', new processes may be created from scratch. Alternatively, by clicking the button `Duplicate', the selected process can be duplicated. With the button `Edit' or by double-clicking the process name in the list box, a process can be edited. Finally, the button `Delete' allows the program users to delete processes. Deletion of a process is only possible, if it is not active within a compartment. The buttons Duplicate, Edit and Delete are inactive as long as no process is selected. Clicking the `Close' button results in closing of this dialog

28

CHAPTER 3. MODEL FORMULATION

box. It can be reopened by clicking the Processes command in the Edit menu shown in Fig. 3.2. For the de nition of transformation processes with the subdialogs assigned to the dialog box shown in Fig. 3.14, all variables listed in the dialog box shown in Fig. 3.3 may be used. The processes serve as a pool for use in compartments: Each process can be activated or inactivated in each compartment if it does not contain illegal dependencies; cf. section 3.3. After clicking the button New in the dialog box shown in Fig. 3.14, the process type can be selected in the dialog box shown in Fig. 3.15. The two types of processes shown in

Figure 3.15: Dialog box for selecting the type of a process. this selection box are described in more detail in the following two sections.

3.2.1 Dynamic Processes

A clear presentation of dynamic biochemical processes is very important to facilitate users of the program to obtain a survey of the interactions between the components of the system. The method of presentation used in AQUASIM was made popular for technical biochemical systems by the report of the IAWQ task group on mathematical modelling for design and operation of biological wastewater treatment Henze et al., 1986. It is based on work on chemical reaction engineering Petersen, 1965. Dynamic processes describe transformations by their contribution to the temporal rate of change of dynamic state variables. Usually, a biological or chemical process transforms several substances in xed stoichiometric proportions. Therefore, it is advantageous to separate a common factor as a process rate, and to describe a process by this rate and by stoichiometric coe cients for all substances involved in the process. The contribution of a process to the temporal change of the concentration of a substance is then given as the product of the common process rate and the substance-speci c stoichiometric coe cient. This decomposition of process rates into a common process rate and individual stoichiometric coe cients is not unique; to make it unique, one of the stoichiometric coe cients is usually set to unity. Physical processes or transfer processes which due to spatial averaging also have the mathematical form of a source term of the di erential equation can be integrated easily into this general scheme. The notion of a stoichiometric coe cient has then a more general meaning and can include geometric factors as well. With this concept, the total transformation rate of a substance sj is given by

rj =

X
i

i;j rpi

3.1

3.2. PROCESSES

29

where rj ML,3 T ,1  is the total transformation rate of the substance sj , i;j , is the stoichiometric coe cient of the substance sj for the process pi and rpi ML,3 T ,1  is the rate of the process pi . A clear presentation of a process model is given by writing the stoichiometric matrix i;j , supplemented by the process rates pi in an additional column. This results in a process matrix as shown in Table 3.3. The nonzero elements of a row of Process

p1 p2 p3
.. .

s1 1;1 2;1 3;1

.. .

s2 1;2 2;2 3;2

.. .

Substances

s3 1;3 2;3 3;3

.. .

::: ::: ::: :::

Rate

...

rp rp rp
.. .

1 2 3

Table 3.3: Representation of a process model with the aid of a process matrix. such a matrix show, which substances are a ected by a given process, whereas the nonzero elements of a column indicate, which processes have an in uence to a given substance. It is a useful convention to use positive process rates. In this case the signs of the stoichiometric coe cients indicate consumption , or production + of the corresponding substance. Figure 3.16 shows the dialog box used for de ning or editing a dynamic process. As

Figure 3.16: Dialog box for editing a dynamic process. each process a dynamic process needs a unique `Name' as an identi er. A name of a process consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of processes, a `Description' can be given optionally. The `Rate' contains the common factor of the transformation rates of all variables involved. As shown in Fig. 3.17, for each `Variable' involved in the process, an individual `Stoichiometric Coe cient', given

30

CHAPTER 3. MODEL FORMULATION

as an algebraic expression according to the syntax of formula variables cf. section 3.1.6 has to be speci ed. The list of stoichiometric coe cients can be edited using the buttons

Figure 3.17: Dialog box for editing a stoichiometric coe cient of a dynamic process.

`Add', `Edit' and `Delete'. The contribution of the process to the transformation rate

of a variable is given as the product of the common rate with the individual stoichiometric coe cient. During simulations, a dynamic process has only an e ect to variables, which are of the type of dynamic state variables. The fact, that in the list of stoichiometric coe cients, any type of variables is allowed, makes it easier to switch between di erent models e.g. if variables are changed from calculated dynamic state variables to measured real list variables, the processes have not to be changed.

3.2.2 Equilibrium Processes

Equilibrium processes are used for processes, which are much faster than the processes which determine the typical time scale of the simulation. A variable determined by such a process can be treated as taking always the value corresponding to its equilibrium state. Therefore, its value is given as the solution of an algebraic equation: req = 0 3.2 where req depends on the variable involved and on other variables in uencing the equilibrium value.

Figure 3.18: Dialog box for editing an equilibrium process. Figure 3.18 shows the dialog box used for de ning or editing an equilibrium process. As each process an equilibrium process needs a unique `Name' as an identi er. A name

3.2. PROCESSES

31

of a process consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of processes, a `Description' can be given optionally. For the selected `Variable', the `Equation' to be solved can be given as an algebraic expression, which is set equal to zero. The syntax of this algebraic expression is the same as that of formula variables described in section 3.1.6. The variable itself has to be an argument of this expression. In a similar way as in the case of dynamic processes, an equilibrium process has only an e ect, if the selected variable is of the type of an equilibrium state variable. Although equilibrium state variables and equilibrium processes can be de ned in AQUASIM, the di culty for the numerical algorithm to nd the solution to the system of nonlinear algebraic equations and limitations of the description of transport processes make the implementation of fast processes with dynamic state variables and fast dynamic processes often more advantageous as the implementation with equilibrium state variables.

32

CHAPTER 3. MODEL FORMULATION

3.3 Compartments
The geometrical con guration of an AQUASIM system consists of a set of compartments of given types. In order to be exible enough to describe the desired system, six types of compartments are distinguished: Mixed Reactor Compartments are used to describe well-mixed domains as e.g. stirred reactors, mixed lakes, etc.. Bio lm Reactor Compartments are used to describe growth and population dynamics of bio lms in which substrate gradients over the depth are important. Advective-Di usive Reactor Compartments are used to describe systems with a longitudinal given water ow such as plug ow reactors. Saturated Soil Column Compartments are used to model transport, adsorption and transformation of substances in saturated soil columns including exchange with dead zones or immobile pore volume. River Section Compartments are used to describe the hydraulics, transport and transformation processes in rivers. Lake Compartments are used to model strati cation, mixing, transport and transformation processes in horizontally well-mixed lakes. Figure 3.19 shows the dialog box for editing compartments. This dialog box is opened

Figure 3.19: Dialog box for editing compartments. with the Compartments command in the Edit menu shown in Figure 3.2. It is of modeless type in order to facilitate the editing process. The names of all compartments already de ned are listed alphabetically in the list box of this dialog box. The type of the currently selected compartment is indicated at the bottom of this dialog box. The buttons of this dialog box allow the user to perform the following operations with compartments: By clicking the button `New', new compartments may be created from scratch. Alternatively, by clicking the button `Duplicate', the selected compartment can be duplicated. With the button `Edit' or by double-clicking the compartment name in the list box, a compartment can be edited. Finally, the button `Delete' allows the program users to delete

3.3. COMPARTMENTS

33

compartments. Deletion of a compartment is only possible, if the compartment is not an argument of a link or of a de nition of a sensitivity analysis or of a parameter estimation. The buttons Duplicate, Edit and Delete are inactive as long as no compartment is selected. The buttons `Activate' and Inactivate can be used to activate and inactivate compartments from the calculation. The names of inactive compartments are indented in the list box of the dialog box shown in Fig. 3.19. Clicking the `Close' button results in closing this dialog box. It can be reopened by choosing the Compartments command in the Edit menu shown in Fig. 3.2. After clicking one of the buttons New in the dialog box shown in Fig. 3.19 the compartment type can be selected in the dialog box shown in Fig. 3.20. The six types of

Figure 3.20: Dialog box for selecting the type of a compartment. compartments shown in this selection box are described in more detail in the following six subsections.

34

CHAPTER 3. MODEL FORMULATION

3.3.1 Mixed Reactor Compartment

Overview
This simplest compartment of AQUASIM describes in ow, out ow and transformation processes of substances in a completely stirred reactor with constant or variable volume. If the volume is selected to be variable, the current volume of the compartment is calculated as the solution to the di erential equation dVR = Q , Q 3.3 in out dt where t is time, VR is the reactor volume, Qin the volumetric in ow, and Qout the volumetric out ow. Otherwise, the volume remains constant the out ow is then equal to the in ow. The temporal change of the concentration of substances dissolved or suspended in the water is given as dCi = Iin;Ci , Qin C + r 3.4 dt VR VR i Ci ; where Ci is the substance concentration represented by a dynamic volume state variable, Iin;Ci is the loading of the substance described by the concentration Ci into the reactor mass per unit of time, and rCi is the transformation rate of the substance described by the concentration Ci . This concentration rate is given as the sum of the products of the process rates times the stoichiometric coe cients of the substance described by the concentration Ci of all processes active in the compartment. The temporal change of substances attached to a surface is given by dSi = r 3.5 Si dt where Si is the concentration, surface density or mass of the attached substance represented by a dynamic surface state variable, and rSi is the transformation rate of the substance described by Si . This transformation rate is calculated analogously to the transformation rate rCi described above. The dimension of Si can be chosen by the program user who is responsible to make consistent process de nitions: If a substance is converted from dissolved Ci  to attached Si ; e.g. by an adsorption process the stoichiometric coe cients must be chosen in order to convert the units of Ci correctly to those of Si . For equilibrium state variables algebraic equations speci ed as equilibrium processes are solved in the compartment. Figure 3.21 shows the dialog box used for de ning or editing a mixed reactor compartment. The edit eld `Name' is used to specify the name of the compartment. Each compartment needs a unique name as an identi er. A name of a compartment consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. In the edit eld `Comp. Index' a nonnegative integer number can be speci ed as a compartment index. This value can be accessed with the aid of the program variable Compartment Index to make variables or process rates dependent on the compartment.

Equations Solved by AQUASIM

User De nitions

3.3. COMPARTMENTS

35

Figure 3.21: Dialog box for editing a mixed reactor compartment. To improve documentation of compartments, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a compartment. The buttons `Variables', `Processes', `Init. Cond.' and `Input' are used to activate and inactivate state variables, to activate and inactivate processes, to specify initial conditions and to de ne inputs to a compartment, respectively. These options are described later in this subsection. The next option allows the user to select the reactor type. Selection of the radio button `constant volume' leads to the use of a reactor the volume of which does not change; selection of the radio button `variable volume' leads to a reactor, the volume of which is calculated dynamically according to the equation given above. The edit eld `Volume' is used to specify the volume of the reactor. For a reactor with variable volume, the number given here is used as the initital volume at the start time of the simulation unless an initial condition for the program variable Reactor Volume is de ned. In the latter case, this initial condition is used. For a reactor with variable volume, in the edit eld `Out ow' the out ow of the reactor must be speci ed the in ow is given in the dialog box opened by clicking the button Input as described below. In this case, the change in reactor volume is given by the di erence in in- and out ow as given by equation 3.3 The check box `active for calculation' can be used to activate or inactivate the compartment from the calculations. This check box has the same functionality as the buttons Activate and Inactivate in the dialog box shown in Fig. 3.19. The button `Acc.' is used to specify the numerical accuracy of program variables as described later in this subsection. As for each compartment, the user has to select which state variables are active. This is done by clicking the button `Variables' of the dialog box shown in Fig. 3.21. This action opens the dialog box shown in Fig. 3.22. The two list boxes of this dialog box show the active variables and the available variables, respectively. The button `Activate' is used to activate available variables selected in the right list box, and the button `Inactivate' is used to inactivate active variables selected in the left list box. If an active variable is selected while activating another variable, the new active variable is inserted in the list of active variables immediately before the selected variable, otherwise it is appended to the

36

CHAPTER 3. MODEL FORMULATION

Figure 3.22: Dialog box for activating and inactivating state variables in a compartment. end of the list of active variables. This gives the user the possibility to in uence the order of active variables the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The list of active variables may contain variables of any type, but activation has only an e ect to state variables. Inactive state variables return a value of zero. The reason for allowing other types of variables in the list of active state variables is to facilitate switching between di erent models that do not contain the same state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of active variables of the compartments. Similarly to the activation of state variables, the user has to select which processes are active in a compartment. This is done by clicking the button `Processes' of the dialog box shown in Fig. 3.21. This action opens the dialog box shown in Fig. 3.23. The two list boxes in this dialog box show the active processes and the available processes, respectively. The button `Activate' is used to activate available processes selected in the right list box, and the button `Inactivate' is used to inactivate an active processes selected in the left list box. If an active process is selected while activating another process, the new active process is inserted in the list of active processes immediately before the selected process, otherwise it is appended to the end of the list of active processes. This gives the user the possibility to in uence the order of active processes the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Initial conditions for a mixed reactor compartment can be speci ed by clicking the button `Init. Cond.' of the dialog box shown in Fig. 3.21. This action opens the dialog box shown in Fig. 3.24. This dialog box shows a list of all initial conditions already speci ed. Each row of the list box contains the name of a variable, followed by the zone of the compartment for which the initial condition is speci ed in brackets and by the algebraic expression specifying the initial value. For each combination of a variable with a zone only one unique initial condition can be speci ed. The list of initial conditions can be edited using the buttons `Add', `Edit' and `Delete'. If an initial condition is selected while adding a new initial condition, the new initial condition is inserted in the list immediately before the selected initial condition, otherwise it is appended to the end

3.3. COMPARTMENTS

37

Figure 3.23: Dialog box for activating and inactivating processes in a compartment. of the list of initial conditions. This gives the user the possibility to in uence the order of the initial conditions the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The dialog box used for editing a single initial condition is shown in Fig. 3.25. In this dialog box the elds `Variable', `Zone' and `Init. Cond.' allow the user to select a variable and a zone and to specify an initial condition. In a mixed reactor compartment there is only one zone Bulk Volume. Initial conditions for any type of variables can be speci ed, but only initial conditions for state variables and, in the case of a reactor with variable volume, initial conditions for the program variable Reactor Volume, are used by the program. The reason for allowing to de ne initial conditions for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of initial conditions of the compartments. Initial conditions of a mixed reactor compartment may not depend on state variables and they can only depend on the program variables Calculation Number and Time. Input to a mixed reactor compartment can be speci ed by clicking the button `Input' of the dialog box shown in Fig. 3.21. This action opens the dialog box shown in Fig. 3.26. The edit eld `Water In ow' of this dialog box is used to specify the discharge of water into the bulk volume of the reactor, Qin , and the list box contains substance loadings, Iin;Ci , into the reactor. For each variable only one unique loading can be speci ed. The list of loadings can be edited using the buttons `Add', `Edit' and `Delete'. If a loading is selected while adding a new loading, the new loading is inserted in the list immediately before the selected loading, otherwise it is appended to the end of the list of loadings. This gives the user the possibility to in uence the order of the loadings the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.27 shows the dialog box used to specify a single input loading of a substance. In this dialog box the elds `Variable' and `Loading' allow the user to select a variable and to specify a loading, Iin;Ci . Note that all loadings must be speci ed as mass per unit of time. A loading of a mixed reactor compartment may depend on the program variable Discharge and on dynamic volume state variables. These variables return the discharge

38

CHAPTER 3. MODEL FORMULATION

Figure 3.24: Dialog box for editing initial conditions for a mixed reactor compartment.

Figure 3.25: Dialog box for editing a single initial condition for a mixed reactor compartment. and the concentrations of dynamic volume state variables resulting from all advective links connected to the inlet of the reactor. Loadings for any type of variables can be speci ed, but only loadings for dynamic volume state variables are used by the program. The reason for allowing to de ne loadings for variables of other types is to facilitate the user switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of loadings of the compartments. The button `Acc.' of the dialog box shown in Fig. 3.21 is used to specify the numerical accuracies of program variables. Figure 3.28 shows the dialog box used for this purpose. It allows the user to specify relative and absolute accuracies of the program variables `Discharge', Q, and reactor `Volume', VR , in the compartment. Good behaviour of the numerical algorithms is usually achieved if the absolute accuracy and the product of the relative accuracy times a typical value of the variable both are 4 to 6 orders of magnitude smaller than typical values of the variable.

3.3. COMPARTMENTS

39

Figure 3.26: Dialog box for editing inputs to a mixed reactor compartment. In Table 3.4 the program variables available in a mixed reactor compartment are summarized for a complete overview of all program variables see Table 3.1 on page 17.
Calculation Number Time Compartment Index Zone Index Discharge Water Fraction Reactor Volume Bulk Volume

Identi er for calculations value set in the dialog boxes shown in Figs. 4.5 and 4.17. Simulation time. Identi er for compartments value set in the dialog box shown in Fig. 3.21. Identi er for zones within compartments returns a value of 0. Volumetric ow rate. Volumetric fraction of water returns a value of 1. Total volume of the reactor. Volume of mixed water zone returns the value of the reactor volume.

Table 3.4: Program variables available in the mixed reactor compartment.

40

CHAPTER 3. MODEL FORMULATION

Figure 3.27: Dialog box for editing a single input to a mixed reactor compartment.

Figure 3.28: Dialog box for editing the accuracies of program variables of a mixed reactor compartment.

3.3. COMPARTMENTS

41

3.3.2 Bio lm Reactor Compartment

Overview
A bio lm consists of a solid matrix with pore water that can contain dissolved substances and suspended solids. The growth or decay of organisms forming the solid matrix leads to expansion or shrinking of the bio lm. The consumption of substrate of the organisms present at a high concentration in the bio lm solid matrix can lead to a growth limitation by the di usive mass transfer into the depth of the bio lm. The bio lm reactor compartment of AQUASIM describes a reactor with a completely mixed bulk water volume and with a bio lm growing on a substratum surface in the reactor. Solids can attach or detach at the surface of the bio lm and in its interior. Any transformation processes can be de ned. The description of the bio lm in AQUASIM is one-dimensional. Only the direction perpendicular to the substratum, which has the largest concentration gradients, is resolved. All variables are averaged over areas parallel to the substratum. Bio lm reactor compartments can be linked advectively or di usively to other AQUASIM compartments.

Equations Solved by AQUASIM

The equations solved in the bio lm reactor compartment are based on the development of a one-dimensional mixed-culture bio lm model since 1984 Wanner and Gujer, 1984; Wanner and Gujer, 1986; Gujer and Wanner, 1990; Wanner, 1994; Wanner, 1995; Wanner and Reichert, 1996; Reichert and Wanner, 1997. The equations as they are solved by AQUASIM are described in the last two papers of this list. In the following, a summary of these equations is given. In order to formulate the one-dimensional conservation laws compartment-speci c expressions for the one-dimensional density, ^ amount of conserved quantity per unit compartment length, for the one-dimensional ux, ^ amount of the j conserved quantity transported per unit time, and for the one-dimensional source term, ^ amount produced per unit compartment length and per unit time, must be derived. r Three zones are distinguished in the bio lm reactor compartment. Variables present in di erent zones are distinguished by the index `M' for the solid matrix, `P' for the pore volume, and `B' for the bulk volume. In addition, the index `F' is used for bio lm, when the distinction into solid matrix and pore volume is not relevant e.g. for the bio lm thickness, LF . The spatial dimension perpendicular to the substratum is resolved by the space coordinate z . This coordinate takes a value of zero at the substratum and it increases with increasing distance from the substratum up to the bio lm thickness, LF . In order to formulate the bio lm equations, 4 types of components of a conservation law must be distinguished: The array of one-dimensional densities of these types of components is given as follows:

j r @ ^ + @^ = ^ @t @z

3.6

0 AX B AXM;i P;i ^= B A C @
A

l;F P;i

1 C C A

3.7

42

CHAPTER 3. MODEL FORMULATION

The rst component of equation 3.7 describes particulate species in the bio lm matrix. The one-dimensional density of these species is given as the product of the area parallel to the substratum, A, and the concentration of particulate species in the bio lm matrix averaged over this area, XM;i . The second component of equation 3.7 describes the particulate species in the bio lm pore water. Similarly to the rst component, the one-dimensional density of these species is given as the product the area parallel to the substratum, A, and the concentration of these species averaged over this area, XP;i. Note that this implies a de nition of a concentration of particulate species as mass per unit of total volume and not only per unit of pore water volume. This de nition makes the concentrations XM;i and XP;i comparable with each other. Note, however, that this leads to a step in the concentration XP;i at the bio lm surface, because there is no solid matrix in the bulk volume and the concentration of suspended solids is continuous if de ned as mass per unit of water phase volume. The third component of equation 3.7 describes substances dissolved in the pore water of the bio lm. The concentration of dissolved substances is de ned as mass per unit of water phase volume. Therefore, their one-dimensional density is given as the product of the area parallel to the substratum, A, the liquid phase volume fraction, l;F , and the concentration of the dissolved species averaged over the area A, CP;i. All concentrations, Xi and Ci , are represented in AQUASIM as dynamic volume state variables. The liquid phase volume fraction is given as
nX X XP;i i=1 Xi

l;F =  ,

3.8

where Xi is the density of the particulate species described by the concentration Xi , nX is the number of particulate species, and  is the porosity of the bio lm given as unity minus the volume fraction of the solid matrix
nX X XM;i i=1 Xi

 =1,

3.9

The last component of equation 3.7 describes the porosity of the bio lm. The onedimensional density for this variable free volume per unit bio lm depth is given as the product of the porosity, , and the area parallel to the substratum, A.

The one-dimensional uxes of the substances with one-dimensional densities as de-

3.3. COMPARTMENTS
scribed by equation 3.7 are given as follows:

43

1 0 AuF XM;i , ADM;Xi @XM;i C B @z C B

XP;i  C B C B C B ,1 , Au XP;i , AD @  C B F  P;Xi @z C B
XP;k  C B B nX nX X DM;Xk @XM;k XP;i X DP;Xk @  XP;i C C B C B +A +A B @z  @z  C Xk Xk C B k=1 k=1 ^= B j B C C C B ,1 , l;F AuF CP;i , l;F ADP;Ci @CP;i B @z
XP;k  C C B C B nX nX X DP;Xk @  C B l;F X DM;Xk @XM;k l;F A B+ A CP;i +  CP;i C C B  @z @z C B k=1 Xk k=1 Xk C B nX C B X DM;Xk @XM;k A @ AuF + A
k=1 Xk

3.10

@z

In this equation, uF is the advective velocity of the bio lm matrix given by 1Z uF = A

0

z

! nX 1 X rM;Xk + r0 A dz 0  1,
k=1 Xk

3.11

0 where r is the excess growth of porosity de ned by 0 r = r ,

nX  X rM;Xk 1 ,  k=1 Xk

3.12

and r is the growth rate of free volume between the solid matrix. With this notation, 0 the porosity of a bio lm remains constant if r = 0, because the same fraction of free volume is produced as is of solid matrix volume. The rst component of equation 3.10 describes the advective and di usive ow of solids within the bio lm matrix. The second component of equation 3.10 describes advection and di usion of solids suspended in the pore volume of the bio lm. The rst term of this component describes advection with the water ow produced by growth and decay of organisms in the bio lm matrix. Because the organisms consist mainly of water, in a growing bio lm, the water contained in the organisms moving outward at the velocity uF in the solid matrix must be compensated by water owing through the pore water into the bio lm. The rst term describes advection of suspended solids with this water ow. The second term describes di usion or Brownian motion of the solids in the pore water. The third and fourth terms of the second component describe advection with the water ow induced by di usion of solids in the solid matrix and in the pore water, respectively. The third component of equation 3.10 describes advection and di usion of dissolved substances in the pore volume of the bio lm. The terms have the same meaning as those in the second component for suspended solids. The last component of equation 3.10 represents the ow of free volume in the bio lm. Free volume is transported advectively with velocity uF analogously to the solid matrix, but di usive transport in the solid matix has a reverse e ect on free volume.

44

CHAPTER 3. MODEL FORMULATION

The following one-dimensional source terms are required to complete the set of bio lm equations.

0 Ar , Ak 1 M;Xi de;vol;Xi XM;i + Akat;vol;Xi XP;i B C iX ^ = B ArP;Xi + Akde;vol;XAr M;i , Akat;vol;Xi XP;i C r @ A
Ci Ar

3.13

In these equations r are transformation rates and kde;vol;Xi and kat;vol;Xi are substancedependent volume detachment and attachment coe cients, respectively. The terms containing the detachment and attachment coe cients describe exchange of solids between the pore water and the solid matrix of the bio lm. Application of the general expression for di erential conservation laws 3.6 to the de nitions given by the equations 3.7 to 3.13 leads to the following set of 4 di erential equations. The rst equation describes the behaviour of the constituents of the bio lm solid matrix:

@XM;i = , u @XM;i + 1 @ AD @XM;i , r0 X F @z M;Xi @z  M;i @t A @z  ! nX XM;i X rM;Xk , k X + k X + rM;Xi , 1 ,  de;vol M;i at;vol P;i k=1 Xk

3.14

The second equation describes the behaviour of solids suspended in the pore water of the bio lm:

@XP;i =1 , u F @t

XP;i 
@z


P;i  1 0 @ X P;i 0 1 @ + 1 ,  X r + A @z @ADP;Xi @z A  X ! nX

1 @ , A @z A

nX @ XP;i + X rM;Xk XP;i , uF @z  k=1 Xk 

 1 0 nX X DP;Xk @ XP;k XP;i A  1 @ , A @z @A @z  X

k=1
k

DM;Xk @XM;k XP;i @z  k=1 Xk

3.15

+ rP;Xi + kde;vol;Xi XM;i , kat;vol;Xi XP;i The third equation describes the behaviour of substances dissolved in the pore water

3.3. COMPARTMENTS
of the bio lm:

45
n

@t

l;F C

X X + l;F @CP;i =1 , l;F uF @CP;i + rM;Xk l;F CP;i P;i @t @z k=1 Xk 

+ 1 ,  1 @ , A @z

l;F r0 C + 1 @  theta P;i A @z

 1 0 nX @ XP;k X 1 @ , A @z @ l;F A DP;Xk @z CP;iA + rCi X

k=1
k

! nX l;F A X DM;Xk @XM;k C  k=1 Xk @z P;i

@CP;i @CP;i l;F ADP;Ci @z @z

3.16 3.17

The fourth equation describes the changes in bio lm porosity:

nX @ = ,u @ , 1 @ A X DM;Xk @XM;k + 1 , r0 F @z A @z  @t @z k=1 Xk

In equation 3.16 @ l;F =@t can be calculated with the aid of the temporal derivative of equation 3.8:

nX @ , X 1 @XP;i @t = @t i=1 Xi @t l;F

3.18

The above equations must be combined with the equation dLF = u 3.19 L dt describing the temporal change of the bio lm thickness, LF . In this equation, uL is the velocity of the interface layer between bio lm and bulk volume. The velocity uL is given by

uL = uF LF  , ude + uat 3.20 where ude and uat are the detachment and the attachment velocities of solids at the bio lm surface, respectively. The detachment velocity, ude , can either be given globally as a function of the bio lm thickness, LF , the advective velocity at the bio lm surface, uF LF , time, t, surface shear stress, surf , etc. ude = udeLF ; uF LF ; t; surf ; ::: 3.21a or, in the case of a di usive bio lm matrix all DM;Xi 0, individual surface detachment rate coe cients, kde;surf;Xi , can be given alternatively. The latter case results in a
detachment velocity of
nX X kde;surf;Xk XM;k 1 ude = 1 , L  F k=1 Xk

3.21b

46

CHAPTER 3. MODEL FORMULATION

nX X kat;surf;Xk XL;k 1 uat = 1 , L  F k=1 Xk

The attachment velocity is calculated similarly to the equation 3.21b as 3.22

where kat;surf;Xi is the surface attachment coe cient for the solids described by the concentrations Xi . In all the above equations, XL;i is the concentration of solids of type i in the liquid boundary layer above the bio lm immediately adjacent to the bio lm surface. In order to make the solution of the above system of di erential equations unique, boundary conditions must be speci ed. For the case of a rigid bio lm matrix all DM;Xi = 0 no boundary condition is required for equation 3.14 at the substratum-bio lm interface. For the case of a di usive bio lm matrix all DM;Xi 0 equation 3.14 is solved with the following boundary condition at the substratum-bio lm interface:

@XM;i z = 0 = 0 @z

3.23a

The boundary condition for equation 3.14 at the bio lm surface is determined by attachment and detachment processes. For a rigid bio lm matrix all DM;Xi = 0 the boundary condition is given by ude , uat XM;i LF  uF , uL XM;i LF  = : u , u  kat;surf;Xi X L;i at de u
at

for ude uat for ude uat

3.23b

For a di usive bio lm matrix the boundary condition for equation 3.14 at the bio lm surface is given by uF , uL XM;i LF  , DM;Xi @XM;i LF  @z = kde;surf;Xi XM;i LF  , kat;surf;Xi XL;i 3.23c The boundary conditions for equation 3.15 that describes the behaviour of suspended solids in the bio lm pore water are given by a no ux condition at the substratum-bio lm interface:

@XP;i z = 0 = 0 @z XP;i LF  = LF XL;i

3.24a 3.24b

and by a continuity equation of the concentrations in the water at the bio lm surface Note that due to the de nition of XP;i as mass per total of bio lm volume including the solid matrix, this continuity condition for concentrations in the water phase leads to a discontinuity of the values of XP;i at the bio lm surface. The advantage of the de nition of XP;i as mass per unit of total bio lm volume is that the concentrations XP;i can directly be compared with the concentrations XM;i of solids in the bio lm matrix.

3.3. COMPARTMENTS

47

The boundary conditions of equation 3.16 that describes the behaviour of substances dissolved in the pore water of the bio lm are as follows. At the substratum-bio lm interface, the boundary condition is given as a continuity equation of the ow through the substratum The total ow through the substratum, Isubstr;Ci , is zero impermeable substratum unless a di usive link is connected to the bottom of the bio lm in order to model a permeable membrane. In the latter case the ow through the substratum is proportional to the di erence between the substance concentration in the pore water and the concentration in the other compartment connected to the di usive link the substance-speci c constants of proportionality are de ned in the link de nition. The boundary condition at the bio lm surface is a continuity condition for the concentration in the water

,A l;F DP;Ci @CP;i z = 0 = Isubstr;Ci @z

3.25a

CP;i LF  = CL;i

3.25b

The equation 3.17 does not require any boundary conditions. The total ows of solids out of the bio lm negative values for ows into the bio lm are given as

IL;Xi =AuF XM;i , ADF;Xi @XM;i LF  @z

P;i  @ X P;i , 1 , AuF X , ADP;Xi @z LF  3.26
 nX nX X DM;Xk @XM;k XP;i X DP;Xk @ XP;k XP;i  +A LF  + A @z  @z  LF  k=1 Xk k=1 Xk , AuLXM;iLF  This expression is the sum of the ow from the solid matrix and the pore volume at z = LF

plus the last term correcting for the movement of the bio lm surface. The total ows of dissolved substances out of the bio lm negative values for ows into the bio lm are given as

IL;Ci = , 1 ,  l;F AuF CP;i LF  , l;F ADP;Ci @CP;i LF   @z nX X DM;Xk @XM;k XP;i + l;F A  @z  LF 
+ l;F A 
k=1 nX X Xk k=1 Xk , AuL l;F CP;iLF 

DP;Xk @

XP;k 
@z


XP;i L   F

3.27

This expression is the ow at z = LF plus a correction that considers the movement of the interface. The equations for the bio lm described so far are connected to the bulk volume of the bio lm reactor through a liquid boundary layer. In AQUASIM the liquid boundary layer

48

CHAPTER 3. MODEL FORMULATION

is only roughly described with the aid of a mass transfer resistance Wanner and Reichert, 1996. The equation for particulate components is given as A similar expression describes the behaviour of the dissolved components  CL;i , CB;i = AL;Ci IL;Ci L

 XL;i , XB;i = AL;Xi IL;Xi L

F

3.28 3.29

F In these equations, L;Xi and L;Ci are the di usive resistances for the substances Xi and Ci , and XB;i and CB;i are the concentrations in the bulk volume. Note that for the

case of a resistance of zero, a continuity equation for the concentrations in the liquid layer adjacent to the bio lm and in the bulk volume results. As a last series of equations, the equations for the bulk volume must be given. The mass balance for particulate substances in the bulk volume is given as d 3.30 dt VB XB;i  = Iin;Xi , XB;i Qef + IL;Xi + VB rXi where VB is the bulk volume, Iin;Xi is the total input of the substance described by the concentration Xi into the bulk volume through advective and di usive links and as speci ed as input in the compartment de nition. The mass balance equation for dissolved substances in the bulk volume is given as d V 3.31 dt B l;B CB;i  = Iin;Ci , l;B CB;i Qef + IL;Ci + VB rCi where l;B is the liquid phase volume fraction in the bulk volume the porosity is equal to unity in the bulk volume. In AQUASIM, two types of reactors are distinguished: A con ned rreactor has a constant total reactor volume for the bio lm and the bulk water. For this reactor type the bulk volume is calculated as

VB = VR ,

L ZF
0

Az 0 dz0 ; VR = const

3.32a

where VR is the constant total reactor volume. The uncon ned reactor type describes a reactor with a constant bulk volume. For this reactor type, the total reactor volume, VR , is not constant but is given as the sum of the variable bio lm volume and the bulk volume as

VR =

L ZF
0

Az 0 dz0 + VB ; VB = const

3.32b

The e uent discharge, Qef , depends on the in uent and on the varying reactor volume. In the case of the con ned rector Qef = Qin 3.33a whereas due to the varying total volume for the uncon ned reactor Qef = Qin , AuL 3.33b

3.3. COMPARTMENTS

49

User De nitions
Within a bio lm reactor compartment three zones, the Bulk Volume, the Biofilm Matrix and the Pore Water are distinguished. Variables and process rates can be made dependent on the zone by using the program variable Zone Index which takes the value 0 in the bulk volume zone, 1 in the bio lm matrix zone and 2 in the pore water zone. Figure 3.29 shows the dialog box used for de ning or editing a bio lm reactor compartment. The edit eld `Name' is used to specify the name of the compartment. Each

Figure 3.29: Dialog box for editing a bio lm reactor compartment. compartment needs a unique name as an identi er. A name of a compartment consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The edit eld `Comp. Index' can be used to specify a nonnegative integer number as a compartment index. This value can be accessed with the aid of the program variable Compartment Index to make variables or process rates dependent on the compartment. To improve documentation of compartments, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a compartment. The buttons `Variables', `Processes', `Init. Cond.' and `Input' are used to activate and inactivate state variables, to activate and inactivate processes, to specify initial conditions and to de ne inputs to a compartment, respectively. These options are described later in this subsection. The buttons `Particulate Variables' and `Dissolved Variables' are used to de ne properties of dissolved or particulate substances for dynamic volume state variables. These options are also described later in this subsection.

50

CHAPTER 3. MODEL FORMULATION

The next option allows the user to select the reactor type. Selection of the radio button `con ned' leads to the use of a reactor with a constant total volume for bio lm and bulk uid. The value for this total reactor volume can be speci ed in the edit eld labelled `Reactor Vol.'. In such a reactor, growth of the bio lm leads to a decrease in bulk volume as described by equation 3.32a. Selection of the radio button `uncon ned' leads to the use of a reactor with a constant bulk volume. In this case, a growing bio lm leads to a growth in total reactor volume. The value of the bulk volume can be speci ed in the edit eld `Bulk Volume'. This option can be used to model free-surface water ow over a bio lm. The next option is used to select the properties of the pore volume. Selection of the radio button `liquid phase only' leads to a model without suspended solids in the pore water. Such a model is adequate to describe a dense bio lm with very small pores in which there is no relevant motion of suspended solids. Selection of the radio button `with suspended solids' leads to a model in which solids can penetrate a bio lm within the pore water. Such a model is suitable to describe bio lms with a high porosity. The radio buttons `rigid' and `di usive' allow the user to select how the bio lm matrix should be modelled. In a rigid bio lm matrix there is no di usive mass transport of solids, whereas in a di usive bio lm matrix a small e ective di usion is also allowed for solids in the bio lm matrix. This option requires the di usivities DM;Xi to be speci ed as properties of particulate variables as described later in this subsection. The treatment of surface detachment of solids from the bio lm can be selected with the aid of the radio buttons `individual rate' and `global velocity' If individual rate is selected, detachment is calculated with the aid of the substance-speci c detachment rate coe cient, kde;surf;Xi . This option can only be selected for a di usive bio lm matrix. If this option is selected, the detachment velocity, ude is calculated according to the equation 3.21b. If a global detachment velocity, ude , is speci ed, solids are detached from the bio lm according to their relative occurrence at the bio lm surface. The edit eld `Bio lm Area' is used to specify the area of the bio lm, A, as a function of the distance from the substratum, z , which is accessible by the program variable Space Coordinate Z. It is not allowed to use a time dependence or a dependence on state variables for the speci cation of the bio lm area. Some possibilities for this function are: A constant surface area

A = const
describes a planar bio lm, a surface area of the form

3.34a 3.34b

A = 2Lcy rcy + z  A = 2Lcy rcy , z 

describes a bio lm growing on the outside of a cylinder with radius rcy and length Lcy , a surface area of the form 3.34c describes a bio lm growing on the inside of a cylinder with radius rcy and length Lcy , and a surface area of the form

A = 4nsprsp + z2
describes bio lms growing on nsp spherical particles of radius rsp .

3.34d

3.3. COMPARTMENTS

51

0 r , de ned in equation 3.12. If this rate is set to zero and if no volume attachment or

The edit eld `Rate Porosity' is used to specify the excess growth rate of porosity,

detachment is active, the porosity of the bio lm remains constant. As last options, `Num. Grid Pts.', `Resolution' and `Acc.', in the dialog box used for the de nition of a bio lm reactor compartment Fig. 3.29 the user can select the number of grid points, the discretization order of the numerical algorithm and the accuracies of program variables. The number of grid points is used to specify by how many discrete points the continuous z -axis is approximated. If the number of grid points is set to ngp, the depth of the bio lm is resolved into 2 boundary points and ngp , 2 grid points located in the middle of ngp , 2 cells of equal thickness. Two additional grid points are used to describe the boundary layer and the bulk volume. For the division of the z -axis described above the `low resolution' method applies a simple rst-order discretization scheme that is very robust but can have signi cant numerical di usion. The `high resolution' method uses a second-order discretization scheme that applies uxlimiters to avoid oscillations of the numerical solutions Reichert, 1994b, chapter 6. The high resolution method is not yet implemented in the bio lm compartment of the current version of AQUASIM. The button `Acc.' is used to specify the numerical accuracies of program variables as described later in this subsection. The check box `active for calculation' can be used to activate or inactivate the compartment from the calculations. This check box has the same functionality as the buttons Activate and Inactivate in the dialog box shown in Fig. 3.19. As for each compartment, the user has to select which state variables are active. This is done by clicking the button `Variables' of the dialog box shown in Fig. 3.29. This action opens the dialog box shown in Fig. 3.30. The two list boxes of this dialog box show the

Figure 3.30: Dialog box for activating and inactivating state variables in a compartment. active variables and the available variables, respectively. The button `Activate' is used to activate available variables selected in the right list box, and the button `Inactivate' is used to inactivate active variables selected in the left list box. If an active variable is selected while activating another variable, the new active variable is inserted in the list of active variables immediately before the selected variable, otherwise it is appended to the

52

CHAPTER 3. MODEL FORMULATION

end of the list of active variables. This gives the user the possibility to in uence the order of active variables the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The list of active variables may contain variables of any type, but activation has only an e ect to state variables. Inactive state variables return a value of zero. The reason for allowing other types of variables in the list of active state variables is to facilitate switching between di erent models that do not contain the same state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of active variables of the compartments. Similarly to the activation of state variables, the user has to select which processes are active in a compartment. This is done by clicking the button `Processes' of the dialog box shown in Fig. 3.29. This action opens the dialog box shown in Fig. 3.31. The two list

Figure 3.31: Dialog box for activating and inactivating processes in a compartment. boxes in this dialog box show the active processes and the available processes, respectively. The button `Activate' is used to activate available processes selected in the right list box, and the button `Inactivate' is used to inactivate an active processes selected in the left list box. If an active process is selected while activating another process, the new active process is inserted in the list of active processes immediately before the selected process, otherwise it is appended to the end of the list of active processes. This gives the user the possibility to in uence the order of active processes the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Initial conditions for a bio lm reactor compartment can be speci ed by clicking the button `Init. Cond.' of the dialog box shown in Fig. 3.29. This action opens the dialog box shown in Fig. 3.32. This dialog box shows a list of all initial conditions already speci ed. Each row of the list box contains the name of a variable, followed by the zone of the compartment for which the initial condition is speci ed in brackets and by the algebraic expression specifying the initial value. For each combination of a variable with a zone only one unique initial condition can be speci ed. The list of initial conditions can be edited using the buttons `Add', `Edit' and `Delete'. If an initial condition is selected while adding a new initial condition, the new initial condition is inserted in the list immediately before the selected initial condition, otherwise it is appended to the end

3.3. COMPARTMENTS

53

Figure 3.32: Dialog box for editing initial conditions for a bio lm reactor compartment. of the list of initial conditions. This gives the user the possibility to in uence the order of the initial conditions the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The dialog box used for editing a single initial condition is shown in Fig. 3.33. In this dialog box the elds `Variable', `Zone' and `Init.

Figure 3.33: Dialog box for editing a single initial condition for a bio lm reactor compartment.

Cond.' allow the user to select a variable and a zone and to specify an initial condition.

In a bio lm compartment there are the zones Bulk Volume, Biofilm Matrix and Pore Volume. Initial conditions for any type of variables can be speci ed, but only initial conditions for state variables and for the program variable Biofilm Thickness are used by the program. The reason for allowing to de ne initial conditions for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of initial conditions of the compartments. Initial conditions of a bio lm reactor compartment may not depend on state variables and they can only depend on the program variables Calculation Number, Time and Space Coordinate Z. Input to a bio lm reactor compartment can be speci ed by clicking the button `Input' of the dialog box shown in Fig. 3.29. This action opens the dialog box shown in Fig. 3.34.

54

CHAPTER 3. MODEL FORMULATION

The edit eld `Water In ow' of this dialog box is used to specify the discharge of water

Figure 3.34: Dialog box for editing inputs to a bio lm reactor compartment. into the bulk volume of the reactor, Qin, and the list box contains substance loadings, Iin;Ci and Iin;Xi , into the bulk volume. For each variable only one unique loading can be speci ed. The list of loadings can be edited using the buttons `Add', `Edit' and `Delete'. If a loading is selected while adding a new loading, the new loading is inserted in the list immediately before the selected loading, otherwise it is appended to the end of the list of loadings. This gives the user the possibility to in uence the order of the loadings the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.35 shows the dialog box used to specify a single input loading of a substance. In this dialog box the elds `Variable' and `Loading' allow the user to select

Figure 3.35: Dialog box for editing a single input loading to a bio lm reactor compartment. a variable and to specify a loading, Iin;Ci or Iin;Xi . Note that this loading represents a mass per unit of time. A loading to a bio lm reactor compartment may depend on the program variables Discharge and on dynamic volume state variables. These variables return the discharge and the concentrations of dynamic volume state variables resulting from all advective links connected to the inlet of the reactor. Loadings for any type of variables can be speci ed, but only loadings for dynamic volume state variables are used by the program. The reason for allowing to de ne loadings for variables of other types is

3.3. COMPARTMENTS

55

to facilitate the user switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of input loadings of the compartments. The button `Particulate Variables' of the dialog box shown in Fig. 3.29 allows the user to specify properties of particulate variables. Figure 3.36 shows the dialog box used for this purpose. It contains a list box with the names of all de ned particulate variables.

Figure 3.36: Dialog box for editing properties of particulate variables in a bio lm reactor compartment. Note that a variable can only be in one of the lists of particluate and dissolved variables. The buttons `Add', `Edit' and `Delete' allow the user to edit this list. Figure 3.37 shows the dialog box used to specify the properties of a particulate variable in a bio lm reactor compartment. The eld `Variable' allows the user to select a variable for which

Figure 3.37: Dialog box for editing properties of a single particulate variable in a bio lm reactor compartment.

56

CHAPTER 3. MODEL FORMULATION

particulate properties are to be de ned. The edit eld `Density' is used to specify the density, Xi , of the variable. The expression for the density may not depend on time. The edit elds `Surf. Att. Coe .', `Surf. Det. Coe .', `Vol. Att. Coe .' and `Vol. Det. Coe .' are used to specify expressions for the surface and volume attachment and detachment coe cients, kat;surf , kde;surf , kat;vol and kde;vol , respectively. Note that the values entered for the surface detachment coe cient only has an e ect if the radio button individual rate of the option Surf. Detach. of the dialog box shown in Fig. 3.29 is selected. Similarly the expressions entered for volume attachment and detachment coe cients are only used if the radio button with free particles is selected for the option Pore Volume of the dialog box shown in Fig. 3.29. The edit eld `Bound. L. Res.' is used to specify the boundary layer resistance, L;Xi , for the particulate species described by the concentration Xi . The last two edit elds, `Pore Di .' and `Matrix Di .' are used to specify the coe cients for pore volume di usion, DP;Xi and DM;Xi , respectively. Note that the rst of these values has only an e ect if the radio button with free particles is selected for the option `Pore Volume' of the dialog box shown in Fig. 3.29 and that the second value only has an e ect if the radio button diffusive for the option Biofilm Matrix is selected. Otherwise the values entered here are ignored. The button `Dissolved Variables' of the dialog box shown in Fig. 3.29 allows the user to specify properties of dissolved variables. Figure 3.38 shows the dialog box used for this purpose. It contains a list box with the names of all de ned dissolved variables. Note

Figure 3.38: Dialog box for editing properties of dissolved variables in a bio lm reactor compartment. that a variable can only be in one of the lists of particluate and dissolved variables. The buttons `Add', `Edit' and `Delete' allow the user to edit this list. Figure 3.39 shows the dialog box used to specify the properties of a dissolved variable in a bio lm reactor compartment. The eld `Variable' allows the user to select a variable for which dissolved properties are to be de ned. The edit eld `Bound. L. Res.' is used to specify the boundary layer resistance, L;Ci , for the dissolved species described by the concentration Ci . The last edit eld, `Pore Di .' is used to specify the coe cient for pore volume di usion, DP;Ci of the dissolved species described by the concentration Ci . The button `Acc.' of the dialog box shown in Fig. 3.29 is used to specify the numerical accuracies of program variables. Fig. 3.40 shows the dialog box used for this purpose.

3.3. COMPARTMENTS

57

Figure 3.39: Dialog box for editing properties of a single dissolved variable in a bio lm reactor compartment. It allows the user to specify relative and absolute accuracies of the program variables `Discharge', Q, reactor or bulk `Volume', VR or VB , `Bio lm Thickness', LF , and `Water Fraction', l;F the last value is also used as the accuracy of the volume fractions of particulate species in the bio lm. Good behaviour of the numerical algorithmsis usually achieved if the absolute accuracy and the product of the relative accuracy times a typical value of the variable both are 4 to 6 orders of magnitude smaller than typical values of the variable. In Table 3.5 the program variables available in a bio lm reactor compartment are summarized for a complete overview of all program variables see Table 3.1 on page 17.

58

CHAPTER 3. MODEL FORMULATION

Figure 3.40: Dialog box for editing the accuracies of program variables of a bio lm reactor compartment.
Identi er for calculations value set in the dialog box shown in Figs. 4.5 and 4.17. Simulation time. Identi er for compartments value set in the dialog box shown in Fig. 3.29. Identi er for zones within compartments returns a value of 0 in the bulk volume zone, a value of 1 in the bio lm matrix, and a value of 2 in the bio lm pore volume. Volumetric ow rate. Volumetric fraction of water. Depth coordinate in the bio lm zero at the substratumbio lm interface. Total volume of the reactor. Volume of mixed water zone outside of the bio lm. Thickness of the bio lm. Advective velocity of bio lm solid matrix. Velocity of the interface between bio lm and bulk uid. Detachment velocity of particles from the bio lm surface. Attachment velocity of particles onto the bio lm surface.

Calculation Number Time Compartment Index Zone Index

Discharge Water Fraction Space Coordinate Z Reactor Volume Bulk Volume Biofilm Thickness Growth Velocity of Biofilm Interface Velocity of Biofilm Detachment Velocity of Biofilm Attachment Velocity of Biofilm

Table 3.5: Program variables available in the bio lm reactor compartment.

3.3. COMPARTMENTS

59

3.3.3 Advective-Di usive Reactor Compartment

Overview
The advective-di usive reactor compartment of AQUASIM can be used to describe onedimensional advective-di usive transport of substances in a ow-through reactor, exchange with settled, sorbed or sessile substances or organisms, and substance transformations. The inlet and outlet of the advective-di usive reactor compartment can be linked advectively to other AQUASIM compartments.

Equations Solved by AQUASIM

In order to formulate the one-dimensional conservation laws compartment-speci c expressions for the one-dimensional density, ^ amount of conserved quantity per unit compartment length, for the one-dimensional ux, ^ amount of the j conserved quantity transported per unit time, and for the one-dimensional source term, ^ amount produced per unit compartment length and per unit time, must be derived. r For the case of the advective-di usive reactor compartment, 3 types of components of a conservation law must be distinguished. The array of one-dimensional densities of these types of components is given as follows:

j @ ^ + @^ = ^ @t @x r

3.35

0 A 1 ^ = @ ACi A
Si

3.36

The rst component of equation 3.36 describes the conservation of water volume within the compartment water is approximated to be incompressible. The one-dimensional density of water volume in the compartment volume per unit length is given by the cross-sectional area, A, of the compartment. The second component of equation 3.36 describes substances transported dissolved or suspended with the water ow. Their onedimensional densities are given as the product of the cross-sectional area, rA, and the laterally averaged concentration, Ci . The last component of equation 3.36 describes substances settled to the bottom or sorbed to surfaces within the compartment. The value of the variable A must be speci ed as a function of the distance along the compartment, x. Its current value is then generally accessible by the program variable Cross Sectional Area. The variables Ci are represented by dynamic volume state variables, and the variables Si are represented by dynamic surface state variables. For equilibrium state variables algebraic equations speci ed as equilibrium processes are solved everywhere along the compartment. The one-dimensional uxes of the quantities with one-dimensional densities as described by equation 3.36 are given as follows:

0 Q B ^ = B QCi , AD @Ci j @
0

@x

1 C C A

3.37

60

CHAPTER 3. MODEL FORMULATION

In this equation Q refers to the volumetric discharge through the compartment, and D is the coe cient of longitudinal di usion or dispersion. The following one-dimensional source terms are required to complete the set of equations for the advective-di usive reactor compartment:

0
^ = B ArCi + r @

1+sign  2

q 
q qC lat;i + , rSi
1

sign  2

1 q  qC C i A

3.38

The rst component of equation 3.38 describes the lateral water in ow as volume per unit length of the compartment, q. In many cases, this longitudinal in ow is zero and the only in ow is that at the compartment inlet. This in ow is speci ed as a boundary condition below 3.42a. The second component describes the e ect of transformation processes in the compartment and the e ect of lateral in ows or out ows Clat;i is the concentration in the lateral in ow q. The last component describes the e ect of transformation processes on settled of sorbed substances or of growing organisms. Application of the general law for di erential conservation laws 3.35 to the de nitions given by the equations 3.36 to 3.38 leads to the following set of 3 di erential equations. The rst equation describes water ow through the compartment: The spatial gradient of the discharge, Q, is determined by the lateral in ow, q. Positive values of q in ow increase the downstream discharge, negative values out ow decrease the downstream discharge. The second equation describes the behaviour of substances transported in the compartment:

@Q = q @x

3.39

@Ci = , 1 @
QC  + 1 @ AD @Ci + r i Ci @t A @x A @x @x
+

1+sign  2

The concentration is a ected by advection with the water ow  rst term, dispersion or di usion second term, transformation processes third term and by lateral in ow or out ow fourth and fth terms. The third equation describes the behaviour of settled or sorbed substances or of sessile organisms: The concentration of such substances is only a ected by transformation processes. Note that settling or sorption must also be formulated as a transformation process transforming dissolved species, Ci , to settled or sorbed species, Si . In order to make the solution to the above system of di erential equations unique, one boundary condition for the ordinary di erential equation 3.39 and two boundary conditions for the partial di erential equation 3.40 are required. The ordinary di erential equation 3.41 that does not contain spatial derivatives does not require boundary conditions.

A Cin;i +

,signq 
2

q A Ci 3.40

@Si = r @t Si

3.41

3.3. COMPARTMENTS

61

The boundary condition for equation 3.39 that describes discharge through the compartment is given by

Qxs = Qin

3.42a

at the start point, xs, of the column. According to equation 3.39, due to the lateral in ow, q, this results in a discharge of

Qxe  = Qxs + q dx
xs

Zxe

3.42b

at the end of the compartment. The boundary conditions for equation 3.40 are given by the continuity of the substance loadings entering the compartment and by a `transmission boundary condition' Shamir and Harleman, 1967 at the end of the compartment:

QCi xs  , AD @Ci xs  = Iin;Ci @x

3.43a

@ 2 Ci = 0 3.43b @x2 where Iin;Ci is the total given mass input of the substance described by the concentration Ci per unit of time. The second of these boundary conditions 3.43b is omitted for
dispersion-free transport.

User De nitions
Figure 3.41 shows the dialog box used for de ning or editing an advective-di usive reactor compartment. The edit eld `Name' is used to specify the name of the compartment.

Figure 3.41: Dialog box for editing an advective-di usive reactor compartment.

62

CHAPTER 3. MODEL FORMULATION

Each compartment needs a unique name as an identi er. A name of a compartment consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The edit eld `Comp. Index' can be used to specify a nonnegative inter number as a compartment index. This value can be accessed with the aid of the program variable Compartment Index to make variables or process rates dependent on the compartment. To improve documentation of compartments, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a compartment. The buttons `Variables', `Processes', `Init. Cond.' and `Input' are used to activate and inactivate state variables, to activate and inactivate processes, to specify initial conditions and to de ne inputs to a compartment. These options are described later in this subsection. The edit elds `Start Coord.' and `End Coord.' are used to de ne the location of the compartment inlet, xs , and of the compartment outlet, xe , respectively. The values of variables are resolved continuously with the space coordinate x between these two locations. Although all equations in this subsection are formulated with the x-axis pointing in ow direction xs xe , the program works also correctly with an x-axis de ned in the reverse direction xs xe . The edit eld `Cross Sect.' is used to specify the wetted cross-sectional area of the compartment, A, as a function of the longitudinal spatial coordinate x, which is accessible by the program variable Space Coordinate X. Note that when the advective-di usive reactor compartment is used to model a porous medium, A is given as the product of the porosity and the total cross-sectional area. It is not allowed to use a time dependence or a dependence on state variables for the speci ction of the cross-sectional area. As a next option, `Di usion', in the dialog box shown in Fig. 3.41 the user can select how to model di usion or dispersion. If the user selects the radio button `without di usion', a purely advective equation with D = 0 is solved note, however, that due to the spatial discretization numerical di usion can occur. If the user selects the radio button `with di usion' a positive di usion or dispersion coe cient D must be speci ed. This coe cient can be made dependent on time, discharge or cross-sectional area by using the program variables Time, Discharge or Cross-Sectional Area. As last options, `Num. Grid Pts.', `Resolution' and `Acc.', in the dialog box used for the de nition of an advective-di usive reactor compartment Fig. 3.41 the user can select the number of grid points, the discretization order of the numerical algorithm and the accuracies of program variables. The number of grid points is used to specify by how many discrete points the continuous x-axis is approximated. If the number of grid points is set to ngp, the compartment is resolved longitudinally into 2 boundary points and ngp , 2 grid points located in the middle of ngp , 2 cells of equal thickness. For this division of the x-axis the `low resolution' method applies a simple rst-order discretization scheme that is very robust but can have signi cant numerical di usion. The `high resolution' method uses a second-order discretization scheme that applies ux-limiters to avoid oscillations of the numerical solutions Reichert, 1994b, chapter 6. The button `Acc.' is used to specify the numerical accuracies of program variables as described later in this subsection. The check box `active for calculation' can be used to activate or inactivate the compartment from the calculations. This check box has the same functionality as the buttons Activate and Inactivate in the dialog box shown in Fig. 3.19. As for each compartment, the user has to select which state variables are active. This is done by clicking the button `Variables' of the dialog box shown in Fig. 3.41. This action

3.3. COMPARTMENTS

63

opens the dialog box shown in Fig. 3.42. The two list boxes of this dialog box show the

Figure 3.42: Dialog box for activating and inactivating state variables in a compartment. active variables and the available variables, respectively. The button `Activate' is used to activate available variables selected in the right list box, and the button `Inactivate' is used to inactivate active variables selected in the left list box. If an active variable is selected while activating another variable, the new active variable is inserted in the list of active variables immediately before the selected variable, otherwise it is appended to the end of the list of active variables. This gives the user the possibility to in uence the order of active variables the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The list of active variables may contain variables of any type, but activation has only an e ect to state variables. Inactive state variables return a value of zero. The reason for allowing other types of variables in the list of active state variables is to facilitate switching between di erent models that do not contain the same state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of active variables of the compartments. Similarly to the activation of state variables, the user has to select which processes are active in a compartment. This is done by clicking the button `Processes' of the dialog box shown in Fig. 3.41. This action opens the dialog box shown in Fig. 3.43. The two list boxes in this dialog box show the active processes and the available processes, respectively. The button `Activate' is used to activate available processes selected in the right list box, and the button `Inactivate' is used to inactivate a active processes selected in the left list box. If an active process is selected while activating another process, the new active process is inserted in the list of active processes immediately before the selected process, otherwise it is appended to the end of the list of active processes. This gives the user the possibility to in uence the order of active processes the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Initial conditions for an advective-di usive reactor compartment can be speci ed by clicking the button `Init. Cond.' of the dialog box shown in Fig. 3.41. This action opens the dialog box shown in Fig. 3.44. This dialog box shows a list of all initial conditions

64

CHAPTER 3. MODEL FORMULATION

Figure 3.43: Dialog box for activating and inactivating processes in a compartment. already speci ed. Each line of the list box contains the name of a variable, followed by the zone of the compartment for which the initial condition is speci ed in brackets and by the algebraic expression specifying the initial value. For each combination of a variable with a zone only one unique initial condition can be speci ed. The list of initial conditions can be edited using the buttons `Add', `Edit' and `Delete'. If an initial condition is selected while adding a new initial condition, the new initial condition is inserted in the list immediately before the selected initial condition, otherwise it is appended to the end of the list of initial conditions. This gives the user the possibility to in uence the order of the initial conditions the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The dialog box used for editing a single initial condition is shown in Fig. 3.45. In this dialog box the elds `Variable' and `Init. Cond.' allow the user to select a variable and to specify an initial condition. The eld `Zone' is inactive, because the advective-di usive reactor compartment contains only one zone Water Body. Initial conditions for any type of variables can be speci ed, but only initial conditions for state variables are used by the program. The reason for allowing to de ne initial conditions for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of initial conditions of the compartments. Initial conditions of an advective-di usive reactor compartment may not depend on state variables and they can only depend on the program variables Calculation Number, Time and Space Coordinate X. Input to an advective-di usive reactor compartment can be speci ed by clicking the button `Input' of the dialog box shown in Fig. 3.41. This action opens the dialog box shown in Fig. 3.46. In this dialog box the user can select which type of input to edit. There exist two di erent types of inputs to an advective-di usive reactor compartment. The radio button `Inlet Input' can be used to describe water and substance ow at the inlet of the compartment and the radio button `Lateral Input' makes it possible to specify water and substance in ow along the compartment. In the following paragraphs, the dialog boxes used to de ne each of these input types are described.

3.3. COMPARTMENTS

65

Figure 3.44: Dialog box for editing initial conditions for an advective-di usive reactor compartment.

Figure 3.45: Dialog box for editing a single initial condition for an advective-di usive reactor compartment. Selection of the radio button `Inlet Input' in the dialog box shown in Fig. 3.46 opens the dialog box shown in Fig. 3.47. The edit eld `Water In ow' of this dialog box is used to specify the discharge of water into the compartment, Qin , and the list box contains substance loadings, Iin;Ci , into the compartment. For each variable only one unique inlet loading can be speci ed. The list of inlet loadings can be edited using the buttons `Add', `Edit' and `Delete'. If a loading is selected while adding a new loading, the new loading is inserted in the list immediately before the selected loading, otherwise it is appended to the end of the list of the loadings. This gives the user the possibility to in uence the order of the loadings the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.48 shows the dialog box used to specify a single inlet loading of a substance. In this dialog box the elds `Variable' and `Loading' allow the user to select a variable and specify an inlet loading, Iin;Ci . Note that this input loading represents a mass per unit of time. An inlet loading to an advective-di usive reactor compartment may depend on the program variables Discharge and on dynamic volume state variables. These variables return the discharge and the concentrations of dynamic volume state variables resulting from all advective links connected to the inlet

66

CHAPTER 3. MODEL FORMULATION

Figure 3.46: Dialog box for selecting an input type for an advective-di usive reactor compartment.

Figure 3.47: Dialog box for editing inlet inputs to an advective-di usive reactor compartment. of the reactor. Loadings for any type of variables can be speci ed, but only loadings for dynamic volume state variables are used by the program. The reason for allowing to de ne loadings for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of loadings of the compartments. Selection of the radio button `Lateral Input' in the dialog box shown in Fig. 3.46 opens the dialog box shown in Fig. 3.49. The edit eld `Water In ow' of this dialog box is used to specify the discharge of water per unit length into the compartment, q, and the list box contains substance concentrations, Clat;i , in the in owing water. A positive value of q represents a ow into the compartment, a negative value represents an out ow. According to the equations 3.38 and 3.40 the in ow concentration is irrelvant if q is smaller than or equal to zero out ow. For each variable only one unique lateral in ow concentration can be speci ed. The list of in ow concentrations can be edited using the buttons `Add', `Edit' and `Delete'. If an in ow concentration is selected while adding a new in ow concentration, the new in ow concentration is inserted in the list immediately before the selected in ow concentration, otherwise it is appended to the end of the list of in ow concentrations. This gives the user the possibility to in uence the order of the

3.3. COMPARTMENTS

67

Figure 3.48: Dialog box for editing a single upstream input to an advective-di usive reactor compartment.

Figure 3.49: Dialog box for editing lateral inputs to an advective-di usive reactor compartment. in ow concentrations the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.50 shows the dialog box used to specify a single lateral in ow concentration of a dynamic volume state variable. In this dialog box the elds `Variable' and `In ow Conc.' allow the user to select a variable and specify a lateral in ow concentration, Clat;i . A lateral in ow concentration of an advective-di usive reactor compartment may depend on the program variables Discharge and on state variables. These variables return the discharge and the current values of state variables in the compartment as a function of the location x. In ow concentrations for any type of variables can be speci ed, but only in ow concentrations for dynamic volume state variables are used by the program. The reason for allowing to de ne in ow concentrations for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of lateral in ow concentrations of the compartments. The button `Acc.' of the dialog box shown in Fig. 3.41 is used to specify the numerical accuracies of program variables. Fig. 3.51 shows the dialog box used for this purpose. It allows the user to specify relative and absolute accuracies of the program variable

68

CHAPTER 3. MODEL FORMULATION

Figure 3.50: Dialog box for editing a single lateral in ow concentration to an advectivedi usive reactor compartment.

Figure 3.51: Dialog box for editing the accuracies of program variables of an advectivedi usive compartment.

`Discharge', Q, and for the `Di usion' coe cient, D, in the compartment. Good
behaviour of the numerical algorithms is usually achieved if the absolute accuracy and the product of the relative accuracy times a typical value of the variable both are 4 to 6 orders of magnitude smaller than typical values of the variable.

3.3. COMPARTMENTS

69

In Table 3.6 the program variables available in a advective-di usive compartment are summarized for a complete overview of all program variables see Table 3.1 on page 17.
Calculation Number Time Compartment Index Zone Index Discharge Water Fraction Space Coordinate X Cross Sectional Area

Identi er for calculations value set in the dialog boxes shown in Figs. 4.5 and 4.17. Simulation time. Identi er for compartments value set in the dialog box shown in Fig. 3.41. Identi er for zones within compartments returns a value of 0. Volumetric ow rate. Volumetric fraction of water returns a value of 1. Space coordinate along the compartment. Area of water body perpendicular to the ow direction.

Table 3.6: Program variables available in the advective-di usive compartment.

70

CHAPTER 3. MODEL FORMULATION

3.3.4 Saturated Soil Column Compartment

Overview
The saturated soil column compartment of AQUASIM can be used to describe advectivedispersive transport of dissolved substances in a saturated soil column, exchange processes with immobile regions consisting of serially connected mixed zones, and transformations of dissolved and sorbed substances. Within the soil column, fast sorption processes can be used to describe equilibrium sorption and slower sorption processes to model the e ects of sorption kinetics. The use of any linear or nonlinear sorption isotherm is possible. The inlet and outlet of the soil column compartment can be advectively linked to other AQUASIM compartments.

Equations Solved by AQUASIM

In order to formulate the one-dimensional conservation laws

compartment-speci c expressions for the one-dimensional density, ^ amount of conserved quantity per unit compartment length, for the one-dimensional ux, ^ amount of the j conserved quantity transported per unit time, and for the one-dimensional source term, ^ amount produced per unit compartment length and per unit time, must be derived. r The soil column compartment has a variable number of zones. Variables present in more than one zone are distinguished by the index mob for the mobile zone and imjk for the mixed zone k of the immobile region j . Figure 3.52 shows schematically the arrangement and enumeration of the zones. The length of the soil column is resolved by
im m o b ile r e g io n 3 im m o b ile r e g io n 2 im
2 2

j @ ^ + @^ = ^ @t @x r

3.44

im

2 1

im m o b ile z o n e m o b im

im
3 1

im
3 2

im
3 3

im
3 4

3 5

c o m p le te ly m ix e d z o n e 2 o f th e im m o b ile r e g io n 3

1 1

im m o b ile r e g io n 1

Figure 3.52: Schematic cross-section through the soil column indicating the enumeration of the immobile zones in the example, there are 3 types of immobile regions consisting of 1, 2 and 5 mixed zones, respectively.

3.3. COMPARTMENTS

71

the space coordinate x. x can increase or decrease along the column, however, in the following the equations are formulated with increasing x in ow direction. In order to formulate the equations for the saturated soil column, 5 types of components of a conservation law must be distinguished. The array of one-dimensional densities of these types of components is given as follows:

0 Amob 1 B AmobCmob;i C B C ^ = B Aimjk Cimjk ;i C B @ Smob;i C A

Simjk ;i

3.45

The rst component of equation 3.45 describes the conservation of water volume within the mobile zone of the soil column water is approximated to be incompressible, and the volume of the immobile zones remains constant. The one-dimensional density of water volume in the mobile zone of the column volume in the mobile zone per unit length of the column is given by the product of the cross-sectional area, A, and the porosity of the mobile zone, mob of the column. The second component of equation 3.45 describes dissolved substances in the mobile zone of the saturated soil column. Their one-dimensional densities are given as the product of the cross-sectional area, A, the porosity of the mobile zone, mob , and the laterally averaged concentration in the mobile zone, Cmob;i . The third component of equation 3.45 describes dissolved substances in the mixed zone k of the immobile region j . Their one-dimensional densities are given as the product of the cross-sectional area, A, the porosity of the zone, imjk , and the laterally averaged concentration, Cimjk ;i . The two last components of equation 3.45 describe substances sorbed to surfaces in contact with the mobile zone or with the mixed zone k of the immobile region j , respectively. The value of the variable A must be speci ed as a function of the distance along the compartment, x. Its value is then accessible by the program variable Cross Sectional Area. The variables Cmob;i and Cimjk ;i are represented by dynamic volume state variables, and the variables Smob;i and Simjk ;i are represented by dynamic surface state variables. For equilibrium state variables, algebraic equations, speci ed as equilibrium processes, are solved in all zones everywhere along the column. The one-dimensional uxes of the substances with one-dimensional densities as described by equation 3.45 are given as follows:

0 Q B QCmob;i , Amob E @Cmob B @x ^= B j B 0 B @ 0

0

1 C C C C C A

3.46

In this equation Q refers to the volumetric discharge through the mobile zone of the saturated soil column, and E is the coe cient of longitudinal dispersion. There are no uxes in the direction along the column within the immobile zones and sorbed substances are not transported at all. The lateral di usive exchange processes of dissolved substances between the mobile zone and the immobile regions and within the immobile regions are described as source terms in the next paragraph.

72

CHAPTER 3. MODEL FORMULATION

The following one-dimensional source terms are required to complete the set of equations for the saturated soil column:

0 1 q B X C B nr C B , qex;imj ;iCmob;i , fex;imj ;iCimj ;i + Amob rCmob C B C B j=1 C B C


, q q qC B + qCmob;i C lat;i + B8 C B q C B ex;imj ;iCmob;i , fex;imj ;iCimj ;i C B C B C ,qex;imj ;iCimj ;i , fex;imj ;iCimj ;i for k = 1 B C B C +Aimj rCimj ;i C ^=B q r B C B ex;imjk;iCimjk, ;i , fex;imjk;iCimjk ;i C B B C ,qex;imjk ;iCimjk ;i , fex;imjk ;iCimjk ;i for 1 k nz;j C B C B C +Aimjk rCimjk ;i B C B qex;im Cim C B C jnz;j jnz;j , , fex;imjnz;j ;i Cimj;nz;j  B: C B C +Aimjnz;j rCimjnz;j ;i for k = nz;j B C B C B C rSmob;i @ A
1 1 1 1+sign  2 1 sign  2 1 1 1 2 1 2 2 1 1 1 +1 +1 +1 1

3.47 The rst component of equation 3.47 describes the lateral water in ow as volume per unit length of the column, q. The second component describes the e ect on the mobile zone concentrations of the exchange between the mobile zone and the rst mixed zones im11 to imnr ;1 of the nr immobile regions that are adjacent to the mobile zone, the e ect of transformation processes in the mobile zone and the e ect of lateral in ows or out ows Clat;i is the concentration in the lateral in ow q. The third component describes the e ect of the exchange with neighbouring zones of the nz;j serially connected mixed zones of the immobile region j . The rst row of this component describes the exchange of the rst mixed zone with the mobile zone and the second mixed zone, the second row the exchange of an intermediate mixed zone with the neighbouring mixed zones, and the last row the exchange of the last mixed zone with the second to last mixed zone of the immobile region j . All exchange processes are assumed to be proportional to the di erence between the concentration in one of the adjacent zones and a concentration multiplied with a conversion factor, fex;imjk ;i , in the other zone. The consideration of a conversion factor makes the description of systems with di erent solvents possible. If all zones are lled with the same solvent, all conversion factors are unity. The substance-speci c exchange coe cent is qex;imjk ;i represents the volume exchange rate per unit length of the soil column for the substance described by the concentration Ci . Application of the general law for di erential conservation laws 3.44 to the de nitions given by the equations 3.45 to 3.47 leads to the following set of 5 di erential equations. The rst equation describes water ow through the soil column: the spatial gradient of the discharge, Q, is determined by the lateral in ow, q. Positive values of q in ow increase the downstream discharge, negative values out ow decrease the downstream discharge.

rSimjk ;i

@Q = q @x

3.48

3.3. COMPARTMENTS

73

The second equation describes the behaviour of dissolved substances in the mobile zone of the column:

@Cmob;i = , 1 @
QC  + 1 @ A E @Cmob;i mob;i mob @t Amob @x Amob @x @x nr X qex;imj ;i , Cmob;i , fex;imj ;i Cimj ;i  + rCmob;i j =1 Amob

 q q q C + in;i + , q A Cmob;i 3.49 Amob mob
1 1 1 1+sign  2 1 sign  2

The concentration is a ected by advection with the water ow  rst term, dispersion second term, exchange with the immobile zones adjacent to the mobile zone third term, transformation processes fourth term and by lateral in ow or out ow  fth and sixth terms. The third equation describes the behaviour of dissolved substances in the mixed zone k of the immobile region j :

Aimj mob;i , fex;imj ;iCimj ;i  ex;im , qA j ;i Cimj ;i , fex;imj ;iCimj ;i for k = 1 imj +rCimj ;i qex;imjk;i Cimjk, ;i , fex;imjk ;i Cimjk ;i  @Cimjk ;i = Aimjk @t , qex;imjk ;i Cimjk ;i , fex;imjk ;iCimjk ;i for 1 k nz;j Aimjk +rCimjk ;i qex;imjnz;j ;i Aimjnz;j Cimjnz;j , ;i , fex;imjnz;j ;iCimjnz;j ;i  for k = nz;j : +rCimjn ;i z;j
1 1 1 1 2 1 2 2 1 1 1 +1 +1 +1 1

8 qex;imj ;i C

3.50

The rst row of this equation describes the behaviour of a dissolved substance within the rst mixed zone of an immobile region, the second row the behaviour in an inner zone, and the third row that in an end zone of the immobile region. The fourth and fth equations describe the behaviour of sorbed substances in the mobile zone and the immobile zones respectively:

@Smob;i = r Smob;i @t @Simjk ;i @t = rSimjk ;i

3.51 3.52

The concentrations are only in uenced by transformation processes. Note that sorption must also be formulated as a transformation process transforming the dissolved species, Ci , to the sorbed species, Si . When describing sorbing substances, the conventional notation

74

CHAPTER 3. MODEL FORMULATION

for the dissolved concentration the index of the zone, zo, must be replaced by mob or imjk , Ci , and Si : 1 3.53c for the sorbed concentration, Si. In these equations solid is the density of the solid material in the soil column, Seq;i Ci  is the equilibrium isotherm and the process describes relaxation of the actually sorbed concentration to the equilibrium concentration with a rate constant ki . If ki is set to a su ciently large value, this model is a good approximation to equilibrium sorption. In order to make the solution to the above system of di erential equations unique, one boundary condition for the ordinary di erential equation 3.48 and two boundary conditions for the partial di erential equation 3.49 are required. The ordinary di erential equations 3.50 to 3.52, that do not contain spatial derivatives, do not require boundary conditions. The boundary condition for equation 3.48 that describes discharge through the soil column is given by Qxs  = Qin 3.54a at the start point, xs , of the column. According to equation 3.48, due to the lateral in ow, q, this results in a discharge of

of C is mass of a substance per unit liquid volume and of S is mass of a substance per unit mass of the solid phase. If the concentrations are expressed as mass per unit of total column volume, C in zone zo must be converted to zoC multiplication of C with the porosity of the zone, and S must be converted to S solid 1 ,  multiplication with the density of the solid phase and with the volume fraction of the solid phase. Considering these conversion factors, sorption can be described by a dynamic sorption process with a process rate of
 ki Seq;iCi  , Si 3.53a and with stoichiometric coe cients of Ci : , solid 1,  3.53b
zo

Qxe  = Qxs + q dx
xs

Zxe

3.54b

at the column outlet. The boundary conditions for equation 3.49 are given by the continuity of the substance mass ows entering the column and by a `transmission boundary condition' Shamir and Harleman, 1967 at the outlet:
mob;i Qxs Cmob;i , Amob E @C@x = Iin;i

3.55a

@ 2 Cmob;i = 0 3.55b @x2 where Iin;i is the total given mass input of substance i per unit of time. The second of
these boundary conditions 3.55b is omitted for dispersion-free transport.

3.3. COMPARTMENTS

75

User De nitions
Within a saturated soil compartment a mobile zone and a user-de ned number of immobile zones are distinguished. Variables and process rates can be made dependent on the zone by using the program variable Zone Index, which takes the value 0 in the mobile zone and a user-de ned positive integer value in each of the immobile zones. Figure 3.53 shows the dialog box used for de ning or editing a soil column compartment. The edit eld `Name' is used to specify the name of the compartment. Each

Figure 3.53: Dialog box for editing a soil column compartment. compartment needs a unique name as an identi er. A name of a compartment consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The edit eld `Comp. Index' can be used to specify a nonnegative inter number as a compartment index. This value can be accessed with the aid of the program variable Compartment Index to make variables or process rates dependent on the compartment. To improve documentation of compartments, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a compartment. The buttons `Variables', `Processes', `Init. Cond.' and `Input' are used to activate and inactivate state variables, to activate and inactivate processes, to specify initial conditions and to de ne inputs to a compartment. These options are described later in this subsection. The edit elds `Start Coord.' and `End Coord.' are used to de ne the location of the column inlet, xs , and of the column outlet, xe , respectively. The values of variables are resolved continuously with the space coordinate x between these two locations. Although all equations in this subsection are formulated with the x-axis pointing in ow direction

76

CHAPTER 3. MODEL FORMULATION

xs xe , the program works also correctly with an x-axis de ned in the reverse direction xs xe . The edit eld `Cross Sect.' is used to specify the total cross-sectional area of the compartment, A, as a function of the longitudinal spatial coordinate x, which is accessible by the program variable Space Coordinate X. It is not allowed to use a time dependence or a dependence on state variables for the speci ction of the cross-sectional area. The edit eld `Mob. Vol. Fract.' is used to specify the porosity of the mobile zone, mob , of the soil column. Similarly to the cross-sectional area, A, mob can be a function of the space coordinate x and it is not allowed to use a time dependence in this variable. As a next option, `Dispersion', in the dialog box shown in Fig. 3.53 the user can select how to model dispersion. If the user selects the radio button `without dispersion', a purely advective equation with D = 0 is solved note, however, that due to the spatial discretization numerical di usion can occur. If the user selects the radio button `with dispersion' a positive dispersion coe cient, D, must be speci ed. This coe cient can be made dependent on time, discharge or cross-sectional area by using the program variables Time, Discharge or Cross-Sectional Area. The list box `Immobile Reg.' contains the names of all immobile regions index j in the equations above already de ned. The set of regions can be edited using the buttons `Add', `Edit' and `Delete'. This editing of immobile regions is described later in this subsection. As last options, `Num. Grid Pts.', `Resolution' and `Acc.', in the dialog box used for the de nition of a saturated soil column compartment Fig. 3.53 the user can select the number of grid points, the discretization order of the numerical algorithm and the accuracies of program variables. The number of grid points is used to specify by how many discrete points the continuous x-axis is approximated. If the number of grid points is set to ngp, the column is resolved longitudinally into 2 boundary points and ngp , 2 grid points located in the middle of ngp , 2 cells of equal thickness. For this division of the x-axis the `low resolution' method applies a simple rst-order discretization scheme that is very robust but can have signi cant numerical di usion. The `high resolution' method uses a second-order discretization scheme that applies ux-limiters to avoid oscillations of the numerical solutions Reichert, 1994b, chapter 6. The button `Acc.' is used to specify the numerical accuracies of program variables as described later in this subsection. The check box `active for calculation' can be used to activate or inactivate the compartment from the calculations. This check box has the same functionality as the buttons Activate and Inactivate in the dialog box shown in Fig. 3.19. As for each compartment, the user has to select which state variables are active. This is done by clicking the button `Variables' of the dialog box shown in Fig. 3.53. This action opens the dialog box shown in Fig. 3.54. The two list boxes of this dialog box show the active variables and the available variables, respectively. The button `Activate' is used to activate available variables selected in the right list box, and the button `Inactivate' is used to inactivate active variables selected in the left list box. If an active variable is selected while activating another variable, the new active variable is inserted in the list of active variables immediately before the selected variable, otherwise it is appended to the end of the list of active variables. This gives the user the possibility to in uence the order of active variables the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The list of active variables may contain variables of any type, but activation has only an e ect to state variables. Inactive state variables return a value of zero. The reason for allowing other types of variables in the list of active state

3.3. COMPARTMENTS

77

Figure 3.54: Dialog box for activating and inactivating state variables in a compartment. variables is to facilitate switching between di erent models that do not contain the same state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of active variables of the compartments. Similarly to the activation of state variables, the user has to select which processes are active in a compartment. This is done by clicking the button `Processes' of the dialog box shown in Fig. 3.53. This action opens the dialog box shown in Fig. 3.55. The two list

Figure 3.55: Dialog box for activating and inactivating processes in a compartment. boxes in this dialog box show the active processes and the available processes, respectively. The button `Activate' is used to activate available processes selected in the right list box, and the button `Inactivate' is used to inactivate a active processes selected in the left list box. If an active process is selected while activating another process, the new active

78

CHAPTER 3. MODEL FORMULATION

process is inserted in the list of active processes immediately before the selected process, otherwise it is appended to the end of the list of active processes. This gives the user the possibility to in uence the order of active processes the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Initial conditions for a saturated soil column compartment can be speci ed by clicking the button `Init. Cond.' of the dialog box shown in Fig. 3.53. This action opens the dialog box shown in Fig. 3.56. This dialog box shows a list of all initial conditions already

Figure 3.56: Dialog box for editing initial conditions for a saturated soil column compartment.

Figure 3.57: Dialog box for editing a single initial conditions for a saturated soil column compartment. speci ed. Each line of the list box contains the name of a variable, followed by the zone of the compartment for which the initial condition is speci ed in brackets and by the algebraic expression specifying the initial value. For each combination of a variable with a zone only one unique initial condition can be speci ed. The list of initial conditions can be edited using the buttons `Add', `Edit' and `Delete'. If an initial condition is selected while adding a new initial condition, the new initial condition is inserted in the list immediately before the selected initial condition, otherwise it is appended to the end of the list of initial conditions. This gives the user the possibility to in uence the order of

3.3. COMPARTMENTS

79

the initial conditions the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The dialog box used for editing a single initial condition is shown in Fig. 3.57. In this dialog box the elds `Variable', `Zone' and `Init. Cond.' allow the user to select a variable and a zone and to specify an initial condition. Initial conditions for any type of variables can be speci ed, but only initial conditions for state variables are used by the program. The reason for allowing to de ne initial conditions for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of initial conditions of the compartments. Initial conditions of a saturated soil column compartment may not depend on state variables and they can only depend on the program variables Calculation Number, Time and Space Coordinate X. Input to a saturated soil column compartment can be speci ed by clicking the button `Input' of the dialog box shown in Fig. 3.53. This action opens the dialog box shown in Fig. 3.58. In this dialog box the user can select which type of input to edit. There

Figure 3.58: Dialog box for selecting an input type for a soil column compartment. exist two di erent types of inputs to a saturated soil column compartment. The radio button `Inlet Input' can be used to describe water and substance ow at the inlet of the column and the radio button `Lateral Input' makes it possible to specify water and substance in ow along the column. In the following paragraphs, the dialog boxes used to de ne each of these input types are described. Selection of the radio button `Inlet Input' in the dialog box shown in Fig. 3.58 opens the dialog box shown in Fig. 3.59. The edit eld `Water In ow' of this dialog box is used to specify the discharge of water into the compartment, Qin , and the list box contains substance loadings, Iin;Ci , into the compartment. For each variable only one unique inlet loading can be speci ed. The list of inlet loadings can be edited using the buttons `Add', `Edit' and `Delete'. If a loading is selected while adding a new loading, the new loading is inserted in the list immediately before the selected loading, otherwise it is appended to the end of the list of input loadings. This gives the user the possibility to in uence the order of the input loadings the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.60 shows the dialog box used to specify a single inlet loading of a substance. In this dialog box the elds `Variable' and `Loading' allow the user to select a variable and specify an inlet loading, Iin;Ci . Note that this loading represents mass per unit of time. An inlet loading of a saturated soil column compartment may depend on the program variables Discharge and on dynamic volume state variables. These variables return the discharge and the concentrations of dynamic volume state variables resulting from all advective links connected to the inlet

80

CHAPTER 3. MODEL FORMULATION

Figure 3.59: Dialog box for editing inlet inputs to a soil column compartment.

Figure 3.60: Dialog box for editing a single inlet input to a soil column compartment. of the column. Input loadings for any type of variables can be speci ed, but only input loadings for dynamic volume state variables are used by the program. The reason for allowing to de ne loadings for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of loadings of the compartments. Selection of the radio button `Lateral Input' in the dialog box shown in Fig. 3.58 opens the dialog box shown in Fig. 3.61. The edit eld `Water In ow' of this dialog box is used to specify the discharge of water per unit length into the mobile zone of the column, q, and the list box contains substance concentrations, Clat;i , in the in owing water. A positive value of q represents a ow into the column, a negative value represents an out ow. According to the equations 3.47 and 3.49 the in ow concentration is irrelvant if q is smaller than or equal to zero out ow. For each variable only one unique lateral in ow concentration can be speci ed. The list of in ow concentrations can be edited using the buttons `Add', `Edit' and `Delete'. If an in ow concentration is selected while adding a new in ow concentration, the new in ow concentration is inserted in the list immediately before the selected in ow concentration, otherwise it is appended to the end of the list of in ow concentrations. This gives the user the possibility to in uence the

3.3. COMPARTMENTS

81

Figure 3.61: Dialog box for editing lateral inputs to a soil column compartment.

Figure 3.62: Dialog box for editing a single lateral input to a soil column compartment. order of the in ow concentrations the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.62 shows the dialog box used to specify a single lateral in ow concentration of a dynamic volume state variable. In this dialog box the elds `Variable' and `In ow Conc.' allow the user to select a variable and specify a lateral in ow concentration, Clat;i . A lateral in ow concentration of a saturated soil column compartment may depend on the program variables Discharge and on state variables. These variables return the discharge and the current values of state variables in the compartment as a function of the location x. In ow concentrations for any type of variables can be speci ed, but only in ow concentrations for dynamic volume state variables are used by the program. The reason for allowing to de ne in ow concentrations for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of lateral in ow concentrations of the compartments. The buttons `Add', `Edit' and `Delete' of the dialog box shown in Fig. 3.53 allow to de ne immobile zones. Several immobile regions, each of which with an arbitrary number of serially connected mixed zones can be de ned. Figure 3.63 shows the dialog box used for de ning an immobile region. The edit eld `Name' is used to specify the name of the immobile region. Each immobile region needs a unique name as an identi er. A name of

82

CHAPTER 3. MODEL FORMULATION

Figure 3.63: Dialog box for editing an immobile region of a saturated soil column compartment. an immobile region consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of immobile regions, the edit eld `Description' can optionally be used to store comments on speci c implementation features of the zone. The list box of the dialog box shown in Fig. 3.63 shows the division of the immobile region into mixed zones. At least one mixed zone is required in order to specify a valid immobile region. The list of mixed zones can be edited using the buttons `Add', `Edit' and `Delete'. If a mixed zone is selected while adding a new zone, the new zone is inserted in the list immediately before the selected zone, otherwise it is appended to the end of the list of mixed zones. This gives the user the possibility to in uence the order of the mixed zones. The Fig. 3.64 shows the dialog box used for the de nition of mixed zone. The edit eld `Zone Index' can be used to specify a nonnegative inter number as a zone index. This value can be accessed with the aid of the program variable Zone Index to make variables or process rates dependent on the zone in the mobile zone, the program variable Zone Index returns zero. The edit eld `Vol. Fract.' is used to specify the volume fraction of the zone, imjk . Note that the sum of the volume fraction of the mobile zone and all volume fractions of the immobile zones must be smaller than unity. The list box of the dialog box shown in Fig. 3.64 shows the exchange coe cients, qex;imjk ;i, and the conversion factors, fex;imjk ;i , of the substances exchanged between the current mixed zone and the mixed zone immediately adjacent to the current mixed zone in direction to the mobile zone for the rst mixed zone this is the exchange coe cient with the mobile zone. The conversion factor can be used to describe transitions between di erent solvents in the pores. If all pores are lled with the same solvent, the conversion factors are unity. The list of exchange coe cients and conversion factors can be edited with the buttons `Add', `Edit' and `Delete'. If an exchange coe cient is selected while adding a new exchange coe cient, the new exchange coe cient is inserted in the list immediately before the selected exchange coe cient, otherwise it is appended to the

3.3. COMPARTMENTS

83

Figure 3.64: Dialog box for editing a mixed zone of a soil column compartment. end of the list of exchange coe cients. This gives the user the possibility to in uence the order of the exchange coe cients the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The Figure 3.65 shows the dialog box used for de ning exchange coe cients and conversion factors. In this dialog box the elds

Figure 3.65: Dialog box for editing an exchange coe cient of a mixed zone of a soil column compartment.

`Variable', `Exch. Coe .' and `Conv. Fact.' allow the user to select a variable and

specify an exchange coe cient, qex;imjk ;i , and a conversion factor, fex;imjk ;i. Exchange coe cients for any type of variables can be speci ed, but only exchange coe cients for dynamic volume state variables are used by the program. The reason for allowing to de ne in ow concentrations for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of lateral in ow concentrations of the compartments. The button `Acc.' of the dialog box shown in Fig. 3.53 is used to specify the numerical accuracies of program variables. Fig. 3.66 shows the dialog box used for this purpose.

84

CHAPTER 3. MODEL FORMULATION

It allows the user to specify relative and absolute accuracies of the program variable `Discharge', Q, and of the `Dispersion' coe cient, E , in the compartment. Good behaviour of the numerical algorithms is usually achieved if the absolute accuracy and the product of the relative accuracy times a typical value of the variable both are 4 to 6 orders of magnitude smaller than typical values of the variable.

Figure 3.66: Dialog box for editing the accuracies of program variables of a saturated soil column compartment. In Table 3.7 the program variables available in a saturated soil column compartment are summarized for a complete overview of all program variables see Table 3.1 on page 17.
Calculation Number Time Compartment Index Zone Index

Discharge Water Fraction Space Coordinate X Cross Sectional Area

Identi er for calculations value set in the dialog boxes shown in Figs. 4.5 and 4.17. Simulation time. Identi er for compartments value set in the dialog box shown in Figs. 3.53. Identi er for zones within compartments returns a value of 0 in the mobile zone and a value set in the dialog box shown in Fig. 3.64 in the mixed immobile zones. Volumetric ow rate. Volumetric fraction of water returns the value of the porosity. Space coordinate along the compartment. Area of water body perpendicular to the ow direction advectivedi usive compartment, saturated soil column compartment, river section compartment.

Table 3.7: Program variables available in the saturated soil column compartment.

3.3. COMPARTMENTS

85

3.3.5 River Section Compartment

Overview
The river section compartment of AQUASIM can be used to describe river hydraulics, advective-dispersive transport of substances dissolved or suspended in the water column, exchange of substances between the water column and the sediment, and transformation processes of substances in the water column or the sediment in a river section without abrupt hydraulic controls. In the presence of signi cant hydraulic structures or tributaries, several river sections can be linked advectively to model the river reach of interest. The description of the river section is one-dimensional. This means that all variables are averaged over the river cross section and the depth of the sediment is not resolved. Such a description is in many cases adequate when the main interest is to model transformation processes over relatively long distances, but it makes the application of the river section compartment to local mixing phenomena impossible.

Equations Solved by AQUASIM

One-dimensional river hydraulics can be described by a set of two partial di erential equations representing a mass and a momentum balance de St. Venant, 1871; Chow, 1959; Henderson, 1966; Yen, 1973; Yen, 1979; French, 1985. The two most important approximations to these so-called St. Venant equations, the kinematic and di usive wave approximations Yen, 1979, are implemented in AQUASIM to describe river hydraulics. The equations for river hydraulics are coupled with advection-di usion equations to describe transport of substances dissolved or suspended in the water. An empirical, substanceindependent di usion coe cient is used to describe the longitudinal mixing e ect due to dispersion. In order to formulate the one-dimensional conservation laws compartment-speci c expressions for the one-dimensional density, ^ amount of conserved quantity per unit compartment length, for the one-dimensional ux, ^ amount of the j conserved quantity transported per unit time, and for the one-dimensional source term, ^ amount produced per unit compartment length and per unit time, must be derived. r For the case of a river section compartment, 3 types of components of a conservation law must be distinguished. The array of one-dimensional densities of these types of components is given as follows:

j r @ ^ + @^ = ^ @t @x

3.56

0 A 1 ^ = @ ACi A
Si

3.57

The rst component of equation 3.57 describes the conservation of water volume within the river water is approximated to be incompressible. The one-dimensional density of water volume in the river volume per unit length is given by the wetted cross-sectional area, A, of the water body. The second component of equation 3.57 describes substances transported with the water ow along the river. Their one-dimensional densities are given as the product of the cross-sectional area, A, of the water body and the laterally averaged

86

CHAPTER 3. MODEL FORMULATION

concentration, Ci . The last component of equation 3.57 describes substances settled to the bottom or sorbed to surfaces of the river bed. The value of the variable A must be speci ed as a function of the distance along the river, x, and of the elevation of the water level, z0 . Its current value is then generally accessible by the program variable Cross Sectional Area. The variables Ci are represented by dynamic volume state variables, and the variables Si are represented by dynamic surface state variables. For equilibrium state variables algebraic equations speci ed as equilibrium processes are solved everywhere along the compartment. The one-dimensional uxes of the substances with one-dimensional densities as described by equation 3.57 are given as follows:

0 Q B ^ = B QCi , AE @Ci j @
0

@x

1 C C A

3.58

In this equation Q refers to the volumetric discharge through the compartment, and E is the coe cient of longitudinal dispersion. Calculation of river hydraulics requires the formulation of the cross-sectionally averaged friction force as an empirical function of averaged ow properties. Usually, instead of the friction force, the non-dimensional friction slope, Sf , the ratio of the friction force to the gravity force of the water body, is parameterized empirically as a function of wetted cross-sectional area, wetted perimeter and discharge. The discharge Q, for the kinematic or the di usive approximation to the St. Venant equations, is then given as the solution to one of the following equations, respectively:

Qkin : , @zB = Sf @x

3.59a

Qdiff : , @z0 = Sf 3.59b @x where zB and z0 are the elevation of the river bed and of the the water surface, respec-

tively. The applicability of the kinematic approximation given by the equation 3.59a is limited by the necessity of a geometry of the river bed that is monotonicly decreasing and simple enough that the bed slope ,@zB =@x can reasonably be de ned. This approximation assumes the driving gravity force to be equilibrated by the friction force everywhere along the river. In addition to the gravity and the friction forces, the di usive approximation given by the equation 3.59b considers longitudinal pressure gradients occuring when the water level is not parallel to the river bed. In contrast to the kinematic approach, the di usive approximation allows to describe backwater e ects of weirs or other hydraulic controls and it can be applied if the river bed is not monotonically decreasing. The following one-dimensional source terms are required to complete the set of equations for the river section compartment:

0 ^ = B ArCi +
r @

1+sign  2

q 
q qC lat;i + , rSi
1

sign  2

1 q  qC C i A

3.60

3.3. COMPARTMENTS

87

The rst component of equation 3.60 describes the lateral water in ow as volume per unit length of the compartment, q. This variable can be used to describe the integral e ect of small tributaries or the e ect of groundwater in- or ex ltration. Larger tributaries must be modelled as upstream input boundary conditions to a new river section compartment. The second component of equation 3.60 describes the e ect of transformation processes in the compartment and the e ect of lateral in ows or out ows Clat;i is the concentration in the lateral in ow, q. The last component describes the e ect of transformation processes on settled or sorbed substances or on growing organisms. Application of the general law for di erential conservation laws 3.56 to the de nitions given by the equations 3.57 to 3.60 leads to the following set of 3 di erential equations. The rst equation describes water ow through the compartment: The temporal change in the wetted cross-sectional area, A, is determined by the spatial gradient of the discharge, Q, and by the lateral in ow, q. A negative spatial gradient of the discharge causes an increase in the wetted cross-sectional area because more water ows into a river segment from upstream then leaves it downstream. Obviously, a positive lateral in ow also increases the wetted cross-sectional area. Note that the equation 3.61 must be solved simultaneously with one of the equations 3.59a or 3.59b. The second equation describes the behaviour of substances transported with the water ow:

@A + @Q = q @t @x

3.61

@ ACi  = , @
QC  + @ AE @Ci + Ar i Ci @t @x @x @x

1+sign  2

qClat;i +

,signq  qCi
2

3.62

The concentration is a ected by advection with the water ow  rst term, longitudinal dispersion second term, transformation processes third term and by lateral in ow or out ow fourth and fth term, respectively. The third equation describes the behaviour of settled or sorbed substances or of sessile organisms: The concentration is only a ected by transformation processes. Note that settling or sorption must also be formulated as a transformation process from the dissolved or suspended species, Ci , to the sorbed or settled species, Si . In order to make the solution to the above system of di erential equations unique, one or two boundary conditions are necessary for equation 3.61 depending on the choice of the kinematic or the di usive approximation, and two boundary conditions are required for the partial di erential equation 3.62. The ordinary di erential equation 3.63 that does not contain spatial derivatives does not require boundary conditions. The rst boundary condition for the equation 3.61 that describes discharge through the river section is given by

@Si = r @t Si

3.63

Qxs  = Qin

3.64a

88

CHAPTER 3. MODEL FORMULATION

at the start point, xs , of the river section. In the case of the di usive approximation according to equation 3.59b which makes it possible to describe backwater e ects, an additional downstream boundary condition is required. Such a boundary condition can be speci ed as a given discharge dependent water level

z0 xe  = zgiven Q

3.64b

at the end of the river section. Besides an arbitrary function describing a strategy of weir regulation, AQUASIM provides the options to select a normal z0;N , or a critical z0;C  end water level. These water levels are given as the solutions to the following algebraic equations:

z0;N : Sf = ,@zB =@x

2 z0;C : Q w = 1 gA3

3.65a 3.65b

The equation 3.65a describes the water level as used in the kinematic approximation to the St. Venant equations. This water level is also approached in the di usive case with a prismatic river bed at distances far away from hydraulic controls. This boundary condition can be used if the calculation ends within a uniform reach where the water level is not in uenced by hydraulic structures. The critical depth is a boundary condition reasonable at drops. The boundary conditions for equation 3.62 are given by the continuity of the substance mass ows entering the river section and by a `transmission boundary condition' Shamir and Harleman, 1967 at the end of the compartment:

Qxs Ci , AE @Ci = Iin;Ci @x

3.66a

@ 2 Ci = 0 3.66b @x2 where Iin;Ci is the total given mass input of substance described by the concentration Ci per unit of time. The second of these boundary conditions 3.66b is omitted for
dispersion-free transport.

User De nitions
Figure 3.67 shows the dialog box used for de ning or editing a river section compartment. The edit eld `Name' is used to specify the name of the compartment. Each compartment needs a unique name as an identi er. A name of a compartment consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The edit eld `Comp. Index' can be used to specify a nonnegative inter number as a compartment index. This value can be accessed with the aid of the program variable Compartment Index to make variables or process rates dependent on the compartment. To improve documentation of compartments, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a compartment.

3.3. COMPARTMENTS

89

Figure 3.67: Dialog box for editing a river section compartment. The buttons `Variables', `Processes', `Init. Cond.' and `Input' are used to activate and inactivate state variables, to activate and inactivate processes, to specify initial conditions and to de ne inputs to a compartment. These options are described later in this subsection. The edit elds `Start Coord.' and `End Coord.' are used to de ne the location of the upstream and downstrem ends of the river section, xs and xe, respectively. The values of variables are resolved continuously with the space coordinate x between these two locations. Although all equations in this subsection are formulated with the x-axis pointing in ow direction xs xe , the program works also correctly with an x-axis de ned in the reverse direction xs xe . The edit eld `Grav. Acceleration' is used to specify the value of the gravitational acceleration, g, in the units as used by the user of the program. Typical values of the gravitational acceleration are 9:81m s2 , 1:27m h2 or 7:32
1010 m d2 . The gravitational acceleration is required in order to calculate the critical water level 3.65b that can be selected as a downstream boundary condition if the di usive wave approximation is selected to calculate river hydraulics. The three edit elds `Cross. Sect.', `Perimeter' and `Width' are used to specify the geometry of the river bed described by the cross-sectional area, A, the wetted perimeter, P , and the surface width, w, as functions of the coordinate x along the river and the elevation of the water level, z0 . These two variables are accessible by the program variables Space Coordinate X and Water Level Elevation. A cross-sectional area

90

CHAPTER 3. MODEL FORMULATION

that depends on a state variable or does not depend on the water level elevation is illegal. As an example of a simple river geometry, a rectangular channel with slope S0 can be described by the three functions

Ax; z0  = z0 + S0xw

for the cross-sectional area

3.67a 3.67b 3.67c

P x; z0  = w + 2z0 + S0x

for the wetted perimeter and

wx; z0  = w

for the constant water surface width. The edit eld `Frict. Slope' is used to specify an empirical formula for the nondimensional friction slope. The two most widely used expressions for the friction slope are 2 2 S = 1 1 Q = n2 1 Q 3.68a
f

according to Manning-Strickler, and f 1 2 Sf = 8g R Q2 3.68b A according to Darcy-Weisbach Chow, 1959; Henderson, 1966; French, 1985. In these equations Kst is the friction coe cient according to Strickler, n is the friction coe cient according to Manning, is the hydraulic radius, and f is the non-dimensional friction factor. As a next option in the dialog box shown in Fig. 3.67 the user can select how to model `Dispersion'. If the user selects the radio button `without dispersion', a purely advective equation with E = 0 is solved note, however, that due to the spatial discretization numerical di usion can occur. If the user selects the radio button `with dispersion' a positive dispersion coe cient, E , must be speci ed in the edit eld below the radio buttons. An estimation of the dispersion coe cient can be calculated as follows Fischer et al., 1979:

2 Kst R4=3 A2

R4=3 A2

A R= P

3.69

where

w2 Q2 E  0:011 uA d u  gdSf

3.70 3.71 3.72

is the shear velocity and

d= A w

3.3. COMPARTMENTS

91

is the mean river depth. The next option of the dialog box shown in Fig. 3.67 is to select the method used for calculating river hydraulics. The user has the choice between the the radio buttons `kinematic' and `di usive', which refer to the kinematic wave approximation according to equation 3.59a and the di usive wave approximation to the St. Venant equations according to equation 3.59b. In the case of the di usive approximation, an end water level must be speci ed as a downstream boundary condition. The user can select between the radio buttons `normal', `critical' and `given' which refer to a normal end water level according to equation 3.65a, a critical end water level according to equation 3.65b or an end water level speci ed by the user in the edit eld below the radio buttons as a function of time and discharge. As last options, `Num. Grid Pts.', `Resolution' and `Acc.', in the dialog box used for the de nition of a river section compartment Fig. 3.67 the user can select the number of grid points, the discretization order of the numerical algorithm and the accuracies of program variables. The number of grid points is used to specify by how many discrete points the continuous x-axis is approximated. If the number of grid points is set to ngp, the river section is resolved longitudinally into 2 boundary points and ngp , 2 grid points located in the middle of ngp , 2 cells of equal thickness. For this division of the x-axis the `low resolution' method applies a simple rst-order discretization scheme that is very robust but can have signi cant numerical di usion. The `high resolution' method uses a second-order discretization scheme that applies ux-limiters to avoid oscillations of the numerical solutions Reichert, 1994b, chapter 6. The button `Acc.' is used to specify the numerical accuracies of program variables as described later in this subsection. The check box `active for calculation' can be used to activate or inactivate the compartment from the calculations. This check box has the same functionality as the buttons Activate and Inactivate in the dialog box shown in Fig. 3.19. As for each compartment, the user has to select which state variables are active. This is done by clicking the button `Variables' of the dialog box shown in Fig. 3.67. This action opens the dialog box shown in Fig. 3.68. The two list boxes of this dialog box show the

Figure 3.68: Dialog box for activating and inactivating state variables in a compartment.

92

CHAPTER 3. MODEL FORMULATION

active variables and the available variables, respectively. The button `Activate' is used to activate available variables selected in the right list box, and the button `Inactivate' is used to inactivate active variables selected in the left list box. If an active variable is selected while activating another variable, the new active variable is inserted in the list of active variables immediately before the selected variable, otherwise it is appended to the end of the list of active variables. This gives the user the possibility to in uence the order of active variables the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The list of active variables may contain variables of any type, but activation has only an e ect to state variables. Inactive state variables return a value of zero. The reason for allowing other types of variables in the list of active state variables is to facilitate switching between di erent models that do not contain the same state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of active variables of the compartments. Similarly to the activation of state variables, the user has to select which processes are active in a compartment. This is done by clicking the button `Processes' of the dialog box shown in Fig. 3.67. This action opens the dialog box shown in Fig. 3.69. The two list

Figure 3.69: Dialog box for activating and inactivating processes in a compartment. boxes in this dialog box show the active processes and the available processes, respectively. The button `Activate' is used to activate available processes selected in the right list box, and the button `Inactivate' is used to inactivate a active processes selected in the left list box. If an active process is selected while activating another process, the new active process is inserted in the list of active processes immediately before the selected process, otherwise it is appended to the end of the list of active processes. This gives the user the possibility to in uence the order of active processes the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Initial conditions for a river section compartment can be speci ed by clicking the button `Init. Cond.' of the dialog box shown in Fig. 3.67. This action opens the dialog box shown in Fig. 3.70. This dialog box shows a list of all initial conditions already speci ed. Each line of the list box contains the name of a variable, followed by the zone of the compartment for which the initial condition is speci ed in brackets and by the

3.3. COMPARTMENTS

93

Figure 3.70: Dialog box for editing initial conditions for a river section compartment.

Figure 3.71: Dialog box for editing a single initial condition for a river section compartment. algebraic expression specifying the initial value. For each combination of a variable with a zone only one unique initial condition can be speci ed. The list of initial conditions can be edited using the buttons `Add', `Edit' and `Delete'. If an initial condition is selected while adding a new initial condition, the new initial condition is inserted in the list immediately before the selected initial condition, otherwise it is appended to the end of the list of initial conditions. This gives the user the possibility to in uence the order of the initial conditions the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The dialog box used for editing a single initial condition is shown in Fig. 3.71. In this dialog box the elds `Variable' and `Init. Cond.' allow the user to select a variable and to specify an initial condition. The eld `Zone' is inactive, because the river section compartment contains only one zone Water Column. Initial conditions for any type of variables can be speci ed, but only initial conditions for state variables and for the program variable Discharge are used by the program. An initial condition for this program variable is required. The reason for allowing to de ne initial conditions for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of initial conditions of the

94

CHAPTER 3. MODEL FORMULATION

compartments. Initial conditions of a river section compartment may not depend on state variables and they can only depend on the program variables Calculation Number, Time and Space Coordinate X. Input to a river section compartment can be speci ed by clicking the button `Input' of the dialog box shown in Fig. 3.67. This action opens the dialog box shown in Fig. 3.72. In this dialog box the user can select which type of input to edit. There exist two di erent

Figure 3.72: Dialog box for selecting an input type for a river section compartment. types of inputs to a river section compartment. The radio button `Upstream Input' can be used to describe water and substance ow at the upstream end of the river section and the radio button `Lateral Input' makes it possible to specify water and substance in ow along the river. In the following paragraphs, the dialog boxes used to de ne each of these input types are described. Selection of `Upstream Input' in the dialog box shown in Fig. 3.72 opens the dialog box shown in Fig. 3.73. The edit eld `Water In ow' of this dialog box is used to specify

Figure 3.73: Dialog box for editing upsteam inputs to a river section compartment. the discharge of water into the compartment, Qin , and the list box contains substance loadings, Iin;Ci , into the river section. For each variable only one unique upstream input loading can be speci ed. The list of upstream substance loadings can be edited using the buttons `Add', `Edit' and `Delete'. If a loading is selected while adding a new loading,

3.3. COMPARTMENTS

95

the new lowding is inserted in the list immediately before the selected loading, otherwise it is appended to the end of the list of input loadings. This gives the user the possibility to in uence the order of the loadings the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.74 shows the dialog box used to specify a single upstream loading of a substance. In this dialog box the elds `Variable'

Figure 3.74: Dialog box for editing a single upstream input to a river section compartment. and `Loading' allow the user to select a variable and to specify an upstream loading, Iin;Ci . Note that this loading represents mass per unit of time. An upstream loading to a river section compartment may depend on the program variables Discharge and on dynamic volume state variables. These variables return the discharge and the current values of dynamic volume state variables resulting from all advective links connected to the upstream end of the river section. Input loadings for any type of variables can be speci ed, but only input loadings for dynamic volume state variables are used by the program. The reason for allowing to de ne loadings for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of input loadings of the compartments. Selection of the radio button `Lateral Input' in the dialog box shown in Fig. 3.72 opens the dialog box shown in Fig. 3.75. The edit eld `Water In ow' of this dialog box is used to specify the discharge of water per unit length into the river section, q, and the list box contains substance concentrations, Clat;i , in the in owing water. A positive value of q represents a ow into the compartment, a negative value represents an out ow. According to the equations 3.60 and 3.62 the in ow concentration is irrelvant if q is smaller than or equal to zero out ow. For each variable only one unique lateral in ow concentration can be speci ed. The list of in ow concentrations can be edited using the buttons `Add', `Edit' and `Delete'. If an in ow concentration is selected while adding a new in ow concentration, the new in ow concentration is inserted in the list immediately before the selected in ow concentration, otherwise it is appended to the end of the list of in ow concentrations. This gives the user the possibility to in uence the order of the in ow concentrations the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.76 shows the dialog box used to specify a single lateral in ow concentration of a dynamic volume state variable. In this dialog box the elds `Variable' and `In ow Conc.' allow the user to select a variable and specify a lateral in ow concentration, Clat;i . A lateral in ow concentration of a river section compartment may depend on the program variables Discharge and on state variables. These variables return the discharge and the current values of state variables in the compartment as a function of the location x. In ow concentrations for any type

96

CHAPTER 3. MODEL FORMULATION

Figure 3.75: Dialog box for editing lateral inputs to a river section compartment.

Figure 3.76: Dialog box for editing a single lateral input to a river section compartment. of variables can be speci ed, but only in ow concentrations for dynamic volume state variables are used by the program. The reason for allowing to de ne in ow concentrations for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of lateral in ow concentrations of the compartments. The button `Acc.' of the dialog box shown in Fig. 3.70 is used to specify the numerical accuracies of program variables. Fig. 3.77 shows the dialog box used for this purpose. It allows the user to specify relative and absolute accuracies of the program variables `Discharge', Q, `Cross-Sectional Area', A, `Water Level' elevation, z0 , and for the `Dispersion' coe cient, E , in the compartment. Good behaviour at the numerical algorithms is usually achieved if the absolute accuracy and the product of the relative accuracy times a typical value of the variable both are 4 to 6 oders of magnitude smaller than typical values of the variable.

3.3. COMPARTMENTS

97

Figure 3.77: Dialog box for editing the accuracies of program variables of a river section compartment. In Table 3.8 the program variables available in a river section compartment are summarized for a complete overview of all program variables see Table 3.1 on page 17.
Calculation Number Time Compartment Index Zone Index Discharge Water Fraction Space Coordinate X Water Level Elevation Cross Sectional Area Perimeter Length Surface Width Friction Slope

Identi er for calculations value set in the dialog boxes shown in Figs. 4.5 and 4.17. Simulation time. Identi er for compartments value set in the dialog box shown in Fig. 3.67. Identi er for zones within compartments returns a value of 0. Volumetric ow rate. Volumetric fraction of water returns a value of 1. Space coordinate along the compartment. Elevation of water level above an absolute reference level. Area of water body perpendicular to the ow direction. Length of the interface between water and the river bed perpendicular to the ow velocity. Length of the interface between water and the atmosphere perpendicular to the ow velocity. Nondimensional friction force: Friction force divided by gravity force.

Table 3.8: Program variables available in the river section compartment.

98

CHAPTER 3. MODEL FORMULATION

3.3.6 Lake Compartment

Overview
The lake compartment of AQUASIM can be used to describe the strati cation of the water column, vertical mixing and advection of substances dissolved or suspended in the water column, sedimentation and resuspension of particles, exchange of dissolved substances between water column and pore water of the top sediment layer, advective and di usive exchange between an arbitrary number of sediment layers, and transformation processes in the water column as well as in the sediment layers. A one-dimensional description is used that averages all variables over horizontal cross sections. This limits the applicability of this compartment to situations in which the dimensions of the lake, the strati cation and the time scales of the investigated processes make a horizontally averaged description reasonable. The current version of the lake compartment has no connections to advective or di usive links, so that it can only be used to describe a single lake with given inputs and processes.

Equations Solved by AQUASIM

The lake equations solved by AQUASIM consist of a combination of a conventional advection-di usion equation for the water column Ulrich et al., 1995 with a sediment model describing an arbitrary number of sediment layers, and with a k- turbulence model Rodi, 1980; Rodi, 1987; Burchard and Baumert, 1995 that has been extended by a simple model of energy storage in seiche motion in the lake basin. The user can specify the coe cient of vertical turbulent di usion as a given function of time and space, but it is also possible to use a parameterization depending on the stability of the water column and on turbulent kinetic energy and dissipation. In this subsection, the full set of equations is described. However, in order to make the use of simpler models easier and faster, the sediment submodel and the turbulence submodel can be inactivated independently of each other. In order to formulate the one-dimensional conservation laws

j r @ ^ + @^ = ^ @t @z

3.73

compartment-speci c expressions for the one-dimensional density, ^ amount of conserved quantity per unit compartment length, for the one-dimensional ux, ^ amount of the j conserved quantity transported per unit time, and for the one-dimensional source term, ^ amount produced per unit compartment length and per unit time, must be derived. r Several zones are distinguished in the lake compartment: one zone for the water column and one zone for each sediment layer. Variables present in all zones are distinguished by the index L for the lake water column and the index Sj for the sediment layer j . The vertical dimension of the lake is resolved by the space coordinate z . In the following, the lake equations are formulated with the z -axis pointing upwards, however, the program runs with both de nitions of the direction of the z -axis. In order to formulate the lake equations, 8 types of components of a conservation law must be distinguished. The array of one-dimensional densities of these types of components

3.3. COMPARTMENTS
is given as follows:

99

dz hsed;j XSj ;i The rst component of equation 3.74 describes the conservation of the water volume water is approximated to be incompressible. The one-dimensional density of water volume volume per unit of depth is given by the cross-sectional area, A, of the lake. The second component of equation 3.74 describes a horizontal water ow induced by the surface shear of the wind. This component is used in mixing models to calculate the production of turbulent kinetic energy by shear forces of wind induced water ow. The one-dimensional density of horizontal water discharge is given as the product of the cross-sectional area, A, and the horizontally averaged horizontal ow velocity, U . The third component of equation 3.74 describes turbulent kinetic energy. The one-dimensional density of turbulent kinetic energy is given as the product of the cross-sectional area of the lake, A, the density of water, , and the turbulent kinetic energy, k turbulent kinetic energy per unit mass of water. The fourth component of equation 3.74 is an equation for the dissipation, , of turbulent kinetic energy. This quantity together with k can be used to estimate the coe cient of turbulent di usion, Kz , of substances dissolved or suspended in the water column see below. The fth and the sixth components of equation 3.74 describe dissolved and suspended substances in the water column of the lake. Both of these one-dimensional densities are given as the product of the cross-sectional area of the lake, A, and the horizontally averaged volumetric concentration of the substance. In the following, concentrations of dissoved substances are described with the symbol C , concentrations of particles with the symbol X . The index L refers to the water column of the lake and the index i indicates, that an arbitrary number of such substances can be used simultaneously. The last two components of equation 3.74 describe the concentrations of dissolved substances and of particles in the pore volume of sediment layers of the lake. The one-dimensional density of dissolved substances is given as the product of the gradient of the cross-sectional area of the lake, dA=dz , the thickness of the sediment layer , hsedj , the porosity of the sediment layer, j , and the concentration of the dissolved substance in the pore water of the sediment layer averaged over the depth of the sediment layer and over the lake border at a given depth, CSj . The one-dimensional density of particles is given as the product of the gradient of the cross-sectional area of the lake, dA=dz , the depth of the sediment layer, hsedj , and the concentration of the particles in the sediment layer averaged over the depth of the sediment layer and over the lake border at a given depth, XSj . The gradient dA=dz in these expressions speci es the sediment area per unit depth of the lake this area is approximated by its horizontal projection. Note that the sediment submodel of AQUASIM does not resolve the depths of the sediment continuously. Instead, an arbitrary number of sediment layers can de ned by the user to roughly resolve

1 0 A B C AU C B C B C B Ak C B B C A B C B ACL;i C C ^= B C B AX C B L;i B C B dA C B hsed;j j CSj ;i C C B dz C B A @ dA

3.74

100

CHAPTER 3. MODEL FORMULATION

the sediment depths. These sediment layer always describe the top of the sediment. The in uence of the sediment underneath these layers can be described by substance uxes into and out of the bottom sediment layer described by AQUASIM. The variables A, U , k and in the rst four components of the one-dimensional density array 3.74 are accessible by the program variables Cross-Sectional Area, Horizontal Velocity, Turbulent Kinetic Energy and Dissipation. The variables CL;i , CS;i , XL;i and XS;i are represented by dynamic volume state variables with dissolved CL;i and CS;i or particulate XL;i and XS;i properties. The one-dimensional uxes of the substances with one-dimensional densities as described by equation 3.74 are given as follows:

0 1 Q B C B C ,At @U + QU B C @z B C B C B C B C ,A t @ @zk + Q k B C k B C B C t @   + Q B C ,A @z B C ^= B j B C C @CL;i + QC B C B C ,AKz @z L;i B C B B ,AK @XL;i + QX , Aw X C C B L;i z @z sed;i L;i C B C B C B C 0 @ A
0 where

3.75

t = PrKz

3.76

is the turbulent viscosity, k and are factors to change di usion of k and from turbulent viscosity, Kz is the coe cient of turbulent di usion of substances dissolved or suspended in the water column, Pr is the Prandtl number, Q is the vertical discharge induced by water in- or out ows in the lake depth positive upwards, and wsed;i is the sedimentation velocity of the particles Xi positive downwards. Terms of equation 3.75 that contain t or Kz describe vertical mixing by turbulent di usion, those containing Q describe vertical advection with water ow, and the term containing wsed;i vertical movement due to sedimentation. There is no vertical ux in the sediment, because the z -axis describes the vertical coordinate in the lake and not the depth within the sediment. The exchange between sediment layers is speci ed with the aid of source terms as given below. The following one-dimensional source terms are required to complete the set of lake

3.3. COMPARTMENTS
equations:

101

0 1 q B C B C  @U  ! dA @U B C  @z B C , t @z + ArU B C dz B C B C dA P B A P + A G , A + bottom + A pint C B C dz @ k 1 B C 0 B @z A dA t @  k  C B C ,@ B dz k @z C B C @ k 1 B C 0 B C dA t @   @z A B C A r ,@ B C dz @z C ^=B r B B , dA Di1 C , C  + Ar +

q  qC +
, q  qC C C B dz hsed; L;i S ;i CL;i L;i C lat;i B C B  2 C B C dA k X + Ar B, C wsed;i dA wsed;i XL;i + dz res S;i B C XL;i B C dz

, q q  qX B + qXL;i C lat;i + B C B C B C B C dA i dA h r B C ex;CSj ;i + dz sed;j CSj ;i B C dz B C @ A dA dA
1+sign 2 1+sign   2 1+sign   2 1 1 1+sign  2 1 sign  2 1+sign  2 1+sign  2 1 sign  2

dz iex;XSj ;i + dz hsed;j rXSj ;i

3.77

where

P = t @z

 @U 2

3.78

is the shear production of turbulent kinetic energy,

 , Kz;min g @ G = : Pr @z 0
t

8


t for Pr else

Kz;min

3.79

is the buoyancy production or destruction of turbulent kinetic energy buoyancy production and destruction are set to zero if the turbulent di usion coe cient is very small; this avoids negative values of turbulent kinetic energy in a stably strati ed lake with no wind forcing, iex;CSj ;i is the ux of the dissolved component i into the sediment layer j

102 given as

CHAPTER 3. MODEL FORMULATION

8 Di 1 h CL , CS ;i
2
1 1 1

 + Di minh 1 ; 2  CS ;i , CS ;i  h+
2
2 2 1

for j = 1

iex;CSj ;i =

Di minj ,1 ; j  C
hj,1 +hj 2

 + Di minhjj ; j +1  CSj ;i , CSj ;i  hj +

2
+1 +1

Sj,1 ;i , CSj ;i 

for 1 j for j = nsed

nsed

Di minnsed ,1; nsed  C

hnsed ,1 +hnsed 2 +ised;Ci

Snsed ,1 ;i , CSnsed ;i

3.80 assuming the number of layers, nsed, to be larger than 2 and neglecting advection due to sedimentation, and iex;XSj ;i is the ux of the particulate component i into the sediment layer j given as

8 X w ,k X L;i sed;i  res S ;i

,
1 1+sign

Fvol;tot;2  Fvol;tot;2
2 2

,signFvol;tot;2 

1+sign

Fvol;tot;j  Fvol;tot;j XS
2

, 1 XS ;i F1vol;tot;2 1 , 2 XS ;i
1 2

for j = 1

,
iex;XSj ;i =

1+sign

,signFvol;tot;j 
2 2

1 , j ,1

1+sign

Fvol;tot;j +1
2

,signFvol;tot;j +1 

for j = nsed 3.81 with the total volume ux of particles from layer j , 1 to layer j , Fvol;tot;j layer 0 corresponds to the water column, layer nsed + 1 to the sediment below the modelled sediment layers where the particle concentrations are assumed to be equal to those in the sediment layer nsed, given as

Fvol;tot;nsed  Fvol;tot;nsed 2 1 , n XSnsed ,1 ;i  1,signFvol;tot;n sed,1 Fvol;tot;nsed sed , 2 1 , nsed XSnsed ;i  , 1+signFvol;tot;nsed+1 Fvol;tot;nsed +1 XSnsed ;i 2 1 , nsed  1,signFvol;tot;n +1  Fvol;tot;nsed +1 sed + 2 1 , nsed +1 XSnsed ;i

Fvol;tot;j X , j Sj ;i 1Fvol;tot;j +1 1 , j XSj ;i Fvol;tot;j +1 1 , j +1 XSj ;i

+1

j ,1 ;i

for 1 j

nsed

Fvol;tot;1 =

X XL;iwsed;i
i Xi

+ Fvol , 1 , 1 kres

3.82a

3.3. COMPARTMENTS
Fvol;tot;j+1 = Fvol;tot;j + hsed;j

103

X rXSj ;i
i Xi

3.82b

The rst component of equation 3.77 describes the lateral water in ow as volume per unit lake depth, q. The second component of equation 3.77 describes the loss of horizontal velocity, U , at the sediment surface during vertical di usive downward transport the factor 1 + signdU=dz =2 makes this term only to be e ective if velocity is transported downwards and it includes a user-de ned source term rU for velocity that can be interpreted as a pressure gradient per unit of water mass. The third component of equation 3.77 describes the one-dimensional source terms for turbulent kinetic energy. The rst term describes production of turbulent kinetic energy by the shear forces induced by a gradient of the wind-driven horizontal velocity, U . The second term describes production or consumption of turbulent kinetic energy by turbulent mixing in the presence of strati cation. This buoyancy production term is negative for the case of a stable strati cation @ =@z 0, because mixing increases the potential energy of the water column. In contrast, during phases of convective mixing @ =@z 0, turbulent kinetic energy is produced. The third term of the third component of equation 3.77 describes loss of turbulent kinetic energy by dissipation, . The fourth and fth terms describe production of turbulent kinetic energy by bottom friction, Pbottom , and internal shear, pint , of seiche oscillations, respectively, and the last term describes loss of turbulent kinetic energy, k, at the sediment surface during vertical di usive downward transport the factor 1 + signd k=dz =2 makes this term only to be e ective if turbulent kinetic energy is transported downwards. The fourth component of equation 3.77 describes the one-dimensional source terms for dissipation of turbulent kinetic energy. The rst term is a general source term to be de ned by the program user. The expression for this term for the standard k- model is given as with the nondimensional model parameters c1 , c2 and c3 Rodi, 1980; Rodi, 1987; Burchard and Baumert, 1995. The second term of the fourth component describes loss of dissipation, , at the sediment surface during vertical di usive downward transport the factor 1 + signd k=dz =2 makes this term only to be e ective if dissipation is transported downwards. The fth component of equation 3.77 describes the e ect of the exchange of dissolved substances between water column and the pore water of the rst sediment layer, the e ect of transformation processes, and the e ect of water in- or out ows on the mass of dissolved substances in the water column. In this source term, Di is the molecular di usion coe cient of the dissolved substance i in the sediment pore water, rCL;i is the production rate per unit time and unit water volume due to transformation processes, and Clat;i is the concentration of the dissolved substance i in the lateral in ow. The terms 1  signq=2 switch between the concentration in the in ow, Clat;i , for an in ow q 0 and the concentration in the lake, CL;i, for an out ow q 0. The sixth component of equation 3.77 describes the e ect of sedimentation, resuspension, transformation processes and in- or out ows on the mass of particles in the water column. The rst two terms describe the loss and gain of particles by sedimentation and

r = c1 P + c3 G k , c2 k

3.83

104

CHAPTER 3. MODEL FORMULATION

resuspension, respectively. In the second term, kres describes the resuspension velocity of the sediment. The third term of the fth component of equation 3.77 describes production of particles by transformation processes and the last two terms describe changes in particle concentration in the lake by in- and out ows, analogously to the dissolved substances in the fourth component. The seventh component of equation 3.77 describes changes in mass of dissolved substances in the sediment layer j due to di usive exchange with neighbouring sediment layers and due to transformation processes. The e ect of advection on dissolved substances is neglected because always the top sediment layers are described by AQUASIM, changes in sediment depth leads to an advective exchange between the sediment layers. The eighth component of equation 3.77 describes changes in mass of particulate substances in the sediment layer j due to advection caused by sedimentation or by transformation processes. In addition to the continuous conservation laws de ned above, the energy stored in seiche oscillations is calculated by AQUASIM according to the following equation:

Z dEseiche = P , Az P Apint dz 0 bottom , wind dt

z0 zB

3.84

where Eseiche is the energy stored in seiche oscillations and Pwind ist the power of wind excitation of the seiche motion. Application of the general law for di erential conservation laws 3.73 to the de nitions given by the equations 3.74 to 3.77 leads to the following set of 8 di erential equations. The rst equation describes vertical water ow in the lake: The spatial gradient of the discharge, Q, is determined by the lateral in ow per unit lake depth, q. Positive values of q in ow into the lake increase the upward discharge, negative values out ows decrease the upward discharge. The second equation describes the behaviour of the horizontal ow velocity: @U = 1 @ A @U ,  @U  ! 1 dA  @U , 1 @
QU  + r @z 3.86 t @z U @t A @z A dz t @z A @z The horizontal ow velocity spreads di usively  rst term, it is dissipated if di usive vertial downward transport ends at the sediment second term, it follows the advective motion of the water column third term, and an arbitrary pressure gradient, rU , can be speci ed by the program user fourth term. The third equation describes the behaviour of turbulent kinetic energy:  @k = 1 @ A t @k ,  @k  ! 1 dA t @k @z @t A @z @z A dz k @z k 1 @
Qk + P + G , + 1 dA Pbottom + p 3.87 , A @z int A dz
1+sign 2 1+sign 2

@Q = q @z

3.85

In order to derive this equation from the equations 3.74 to 3.77, the gradient of the density, , was neglected. Turbulent kinetic energy spreads di usively  rst term, it is

3.3. COMPARTMENTS

105

dissipated if di usive vertial downward transport ends at the sediment second term, it follows the advective motion of the water column third term, it is produced by the shear forces of the horizontal ow fourth term, it is produced in an unstable water column or consumed in a stable water column by turbulent mixing  fth term, it is dissipated sixth term, it is produced by bottom friction of seiche oscillations seventh term, and it is produced by the shear of seiche oscillations produced within the water column eighth term. The fourth equation describes the behaviour of dissipation:  @ @ = 1 @ A t @k ,  @z  ! 1 dA t @ , 1 @
Qk + r 3.88 @t A @z @z A dz @z A @z In order to derive this equation from the equations 3.74 to 3.77, the gradient of the density, , was neglected. Dissipation spreads di usively  rst term, it is eliminted if di usive vertial downward transport ends at the sediment second term, it follows the advective motion of the water column third term, and it is produced by a source term fourth term. It is usual to use equation 3.83 in order to parameterize this source term of the equation. The fth equation describes the behaviour of dissolved substances in the water column:
1+sign 2

Dissolved substances spread di usively  rst term, they follow the advective motion of the water column second term, they are exchanged with the pore water of the rst sediment layer third term, they can be converted by transformation processes fourth term, and they are in uenced by in ows and out ows  fth and sixth terms, respectively. The sixth equation describes the behaviour of particles in the water column:

@CL;i = 1 @ AK @CL;i , 1 @
QC  , 1 dA Di 1 C , C  z @z L;i @t A @z A @z A dz hsed; L;i S ;i 2

q q  qC + rCL;i + lat;i + , q A CL;i 3.89 A
1 1 1+sign  2 1 sign  2

@XL;i = 1 @ AK @XL;i , 1 @
QX  z @z L;i @t A @z A @z   @
1 + @z wsed;i XL;i + , wsed;i dA wsed;i XL;i + A dA kresXS ;i dz dz
q q Xlat;i  +
, + rXL;i + A
1 sign  2 1 1+sign  2 1

sign  2

q q XL;i  A

3.90

Particles spread di usively  rst term, they follow the advective motion of the water column second term, they sink through the water column with their sedimentation velocity third term, they can be resuspended from the sediment fourth term, they can be converted by transformation processes  fth term, and they are in uenced by in ows and out ows sixth and seventh terms, respectively. The seventh equation describes the behaviour of dissolved substances in the pore water of sediment layer j :

@CSj ;i iex;CSj ;i rCS;i @t = j hsed;j + j

3.91

Substances dissolved in the pore water of the sediment layer are exchanged with the neighbouring layers  rst term and they are changed by transformation processes second

106

CHAPTER 3. MODEL FORMULATION

term. Note that the process rates must be formulated as mass per unit of time and per unit of total sediment volume and not only per pore water volume. This facilitates the formulation of mass conversion for transformation processes from the particulate to the dissolved phase. The last equation describes the behaviour of particles in the sediment:

@XSj ;i iex;XSj ;i @t = hsed;j + rXS;i

3.92

The particle concentrations in the sediment layer are exchanged with the neighbouring layers  rst term and they are changed by transformation processes second term. In order to make the solution of the above system of di erential equations unique, one boundary condition for the ordinary di erential equation 3.85 and two boundary conditions for each of the partial di erential equations 3.86 to 3.90 are required. The ordinary di erential equations 3.91 and 3.92 that do not contain spatial derivatives do not require boundary conditions. The boundary condition for the equation 3.85 describing vertical discharge at the lake bottom is given by

QzB  = 0

3.93a

According to the equation 3.85 this results in a discharge of

Qz0  = q dz
zB

Zz

3.93b

at the lake surface. This discharge is assumed to leave if it is positive or to enter if it is negative the lake at its suface in order to keep the lake volume constant. The boundary conditions for the equation 3.86 describing the wind-driven horizontal ow are given by

U zB  = 0
and

3.94a
surf

where surf is the wind shear at the lake surface. The boundary conditions for the equation 3.87 describing the turbulent kinetic energy are approximated to be no- ux conditions: and

t @U z0  = @z

3.94b

@k z  = 0 @z B @k z  = 0 @z 0

3.95a 3.95b

These boundary contitions neglect changes in turbulent kinetic energy due to e ects of the lake surface and the lake bottom.

3.3. COMPARTMENTS

107

The boundary conditions for the equation 3.88 describing the dissipation of turbulent kinetic energy are approximated to be no- ux conditions: and

@ @z zB  = 0 @ z  = 0 @z 0

3.96a 3.96b

These boundary contitions neglect changes in dissipation of turbulent kinetic energy due to e ects of the lake surface and the lake bottom. The boundary conditions for the equation 3.89 describing the behaviour of dissolved substances in the water colun of the lake are given by the continuity of the uxes over the boundary: and

Kz @CL;i zB  = kdiff;i CL;i zB  , CS;izB  @z Kz @CL;i z0  = isurf;Ci @z

3.97a 3.97b

where isurf;Ci is the given ux of the dissolved substance i through the lake surface as mass per unit lake surface area and per unit time positive. The boundary conditions for the equation 3.90 describing the behaviour of particles in the water column of the lake are given by the continuity of the uxes over the boundary:

and

Kz @XL;i zB  = ,kr esXS;i @z Kz @XL;i z0  = isurf;Xi @z

3.98a 3.98b

where isurf;Xi is the given ux of substance Xi through the lake surface as mass per unit lake surface area and per unit time positive.

108

CHAPTER 3. MODEL FORMULATION

User De nitions
Within a lake compartment two zones, the water column and the sediment layer, are distinguished. Variables and process rates can be made dependent on the zone by using the program variable Zone Index which takes the value 0 in the water column and 1 in the sediment. Figure 3.78 shows the dialog box used for de ning or editing a lake compartment. The

Figure 3.78: Dialog box for editing a lake compartment. edit eld `Name' is used to specify the name of the compartment. Each compartment needs a unique name as an identi er. A name of a compartment consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The edit eld `Comp. Index' can be used to specify a nonnegative integer number as a compartment index. This value can be accessed with the aid of the program variable Compartment Index to make variables or process rates dependent on the compartment. To improve documentation of compartments, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a compartment. The buttons `Variables', `Processes', `Init. Cond.' and `Input' are used to activate and inactivate state variables, to activate and inactivate processes, to specify initial conditions and to de ne s to a compartment. These options are described later in this subsection. The buttons `Particulate Variables' and `Dissolved Variables' are used to de ne properties of dissolved or particulate substances for dynamic volume state variables. These options are also described later in this subsection. The edit eld `Grav. Acceleration' is used to specify the value of the gravitational acceleration, g, in the units as used by the user of the program. Typical values of the

3.3. COMPARTMENTS

109

gravitational acceleration are 9:81m s2 , 1:27
108 m h2 or 7:32
1010 m d2 . The gravitational acceleration is required in order to calculate the stability or Brunt Vaisala frequency, N 2 , de ned by equation 3.99 below. The edit elds `Top Coord.' and `Bottom Coord.' are used to de ne the location of the water surface, z0 , and of the deepest point of the lake, zB , respectively. Although all equations in this subsection are formulated with the z -axis pointing upwards z0 zB , the program works also correctly with a reverse z -axis z0 zB . Both of these locations, z0 and zB , are xed numbers. It is not possible to describe varying water levels or changes in volume due to sedimentation with the current version of the lake compartment of AQUASIM. The edit eld `Cross. Sect.' is used to specify the cross-sectional area of the lake, A, as a function of the vertical coordinate z which is accessible by the program variable Space Coordinate Z. For most applications to real systems the cross-sectional area A will be de ned as a real list variable with the argument z and the name of this real list variable is then entered in this eld. However, it is also possible to specify the crosssectional area as an algebraic expression. Note that the cross-sectional area must be a non-increasing function with increasing depth of the lake and that the value of the surface cross-sectional area at zB must still be positive. It is not allowed to use a time dependence in the cross-sectional area. The edit eld `Density' is used to specify the density, , of water as a function of temperature and concentrations of dissolved and suspended substances. An irrelevant constant value can be entered here, if the coe cient of turbulent di usion, Kz , is given without reference to the strati cation of the water column and if the value of the stability or Brunt Vaisala frequency is not used. In the case of a freshwater lake, the following formula can be used to approximate the density of the lake water Buhrer and Ambuhl, 1975:

N2 = ,g @ @z

3.99

 999:84298 kg m3 + 10,3 kg m3
65:4891 o C,1 T 

 , 8:56272 o C,2 T 2 + 0:059385 o C,3 T 3
1 + 20 3.100
where T is temperature,  is given as 3.101   0:705
10,6 cm S and 20 is the electrical conductivity at 20o C that can be calculated from the conductivity at temperature T , T , according to

20  T
1:72118 , 0:0541369 o C,1 T  + 1:14842
10,3 o C,2 T 2 , 1:222651
10,5 o C,3 T 3 3.102 The edit eld `Di usion' is used to specify the coe cient of turbulent di usion, Kz . Kz can be given as any function of time, space, stability etc.. Some formulations for Kz
are described in the following. The coe cient of turbulent di usion can be parameterized as a simple function of space and time: Kz = given function of z and t. 3.103a

110

CHAPTER 3. MODEL FORMULATION

The parameters of such a function can be estimated by AQUASIM using measured data of temperature or substances without or with known transformation processes. An alternative can be to parameterize the coe cient of vertical turbulent di usion using the stability or Brunt Vaisala frequency, N 2 3.99 that is accessible as the program variable Brunt Vaisala Frequency. A parameterization often used in the literature is given by

8 :

Kz =

N Kz;max

a2b

if

N if N 2  0 or
a b N2

a2 b  Kz;max

Kz;max

3.103b

The parameters a and b of this expression can be estimated by AQUASIM using several temperature pro les of the lake. An additional option is to use the turbulence submodel to estimate the coe cient of vertical turbulent di usion. A possible parameterization is 1 3.103c Kz = c k
Pr The factor 1=Pr is necessary in order to convert from the turbulent viscosity, t , to the
2

di usion coe cient for substances and temperature. There are four `Modes' of calculation for the lake compartment that can be selected with the radio buttons `without Sediment', `with Sediment', `without TKE' and `with TKE'. If the sediment submodel is turned on, its options must be speci ed in a dialog box appearing after clicking the button `Sed. Prop.'. Similarly, if the submodel for the calculation of turbulent kinetic energy TKE is turned on, its options must be speci ed by clicking the button `TKE Prop.'. The dialog boxes for the speci cation of these submodels are discussed later in this subsection. As last options, `Num. Grid Pts.', `Resolution' and `Acc.', in the dialog box used for the de nition of a lake compartment Fig. 3.78 the user can select the number of grid points, the discretization order of the numerical algorithm and the accuracies of program variables. The number of grid points is used to specify by how many discrete points the continuous z -axis is approximated. If the number of grid points is set to ngp, the depth of the lake is resolved by 2 boundary points and ngp , 2 grid points located in the middle of ngp , 2 cells of equal thickness. For this division of the z -axis the `low resolution' method applies a simple rst-order discretization scheme that is very robust but can have signi cant numerical di usion. The `high resolution' method uses a secondorder discretization scheme that applies ux-limiters to avoid oscillations of the numerical solutions Reichert, 1994b, chapter 6. The button `Acc.' is used to specify the numerical accuracies of program variables as described later in this subsection. As for each compartment, the user has to select which state variables are active. This is done by clicking the button `Variables' of the dialog box shown in Fig. 3.78. This action opens the dialog box shown in Fig. 3.79. The two list boxes of this dialog box show the active variables and the available variables, respectively. The button `Activate' is used to activate available variables selected in the right list box, and the button `Inactivate' is used to inactivate active variables selected in the left list box. If an active variable is selected while activating another variable, the new active variable is inserted in the list of active variables immediately before the selected variable, otherwise it is appended to the end of the list of active variables. This gives the user the possibility to in uence the order

3.3. COMPARTMENTS

111

Figure 3.79: Dialog box for activating and inactivating state variables in a compartment. of active variables the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The list of active variables may contain variables of any type, but activation has only an e ect to state variables. Inactive state variables return a value of zero. The reason for allowing other types of variables in the list of active state variables is to facilitate switching between di erent models that do not contain the same state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of active variables of the compartments. Similarly to the activation of state variables, the user has to select which processes are active in a compartment. This is done by clicking the button `Processes' of the dialog box shown in Fig. 3.78. This action opens the dialog box shown in Fig. 3.80. The two list

Figure 3.80: Dialog box for activating and inactivating processes in a compartment.

112

CHAPTER 3. MODEL FORMULATION

boxes in this dialog box show the active processes and the available processes, respectively. The button `Activate' is used to activate available processes selected in the right list box, and the button `Inactivate' is used to inactivate a active processes selected in the left list box. If an active process is selected while activating another process, the new active process is inserted in the list of active processes immediately before the selected process, otherwise it is appended to the end of the list of active processes. This gives the user the possibility to in uence the order of active processes the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Initial conditions for a lake compartment can be speci ed by clicking the button `Init. Cond.' of the dialog box shown in Fig. 3.78. This action opens the dialog box shown in Fig. 3.81. This dialog box shows a list of all initial conditions already speci ed.

Figure 3.81: Dialog box for editing initial conditions for a lake compartment. Each row of the list box contains the name of a variable, followed by the zone of the compartment for which the initial condition is speci ed in brackets and by the algebraic expression specifying the initial value. For each combination of a variable with a zone only one unique initial condition can be speci ed. The list of initial conditions can be edited using the buttons `Add', `Edit' and `Delete'. If an initial condition is selected while adding a new initial condition, the new initial condition is inserted in the list immediately before the selected initial condition, otherwise it is appended to the end of the list of initial conditions. This gives the user the possibility to in uence the order of the initial conditions the order is irrelevant for the program, but it may be convenient for the user to have a certain order. The dialog box used for editing a single initial condition is shown in Fig. 3.82. In this dialog box the elds `Variable', `Zone' and `Init. Cond.' allow the user to select a variable and a zone and to specify an initial condition. Note that in the lake compartment the two zones Water Column and Sediment Layer are available. Initial conditions for any type of variables can be speci ed, but only initial conditions for state variables and for the program variables Brunt Vaisala Frequency N 2 , Horizontal Velocity U , Turbulent Kinetic Energy and Energy of Seiche Oscillation are used by the program. The reason for allowing to de ne initial conditions for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement

3.3. COMPARTMENTS

113

Figure 3.82: Dialog box for editing a single initial conditions for a lake compartment. ot editing the lists of initial conditions of the compartments. Initial conditions of a lake compartment may not depend on state variables and they can only depend on the program variables Calculation Number, Time and Space Coordinate Z. Input to a lake compartment can be speci ed by clicking the button `Input' of the dialog box shown in Fig. 3.78. This action opens the dialog box shown in Fig. 3.83. In this

Figure 3.83: Dialog box for selecting an type for a lake compartment. dialog box the user can select which type of input to edit. There exist four di erent types of inputs to a lake compartment. The radio button `Surface Input' can be used to describe water and substance exchange over the lake surface, the radio buttons `Lateral Input' and `Point Input' are two alternative ways of describing water and substance in- or out ow distributed over the depth of the lake, and the radio button `Sediment Input' is used to describe substance exchange between the sediment layer described by the model and deeper sediment layers. In the following paragraphs, the dialog boxes used to de ne each of these input types are described. Selection of the radio button `Surface Input' in the dialog box shown in Fig. 3.83 opens the dialog box shown in Fig. 3.84. This dialog box contains a list box for de ning the substance mass uxes, isurf;Ci and isurf;Xi , accross the lake surface. For each variable only one unique surface input ux can be speci ed. The surface input uxes can be edited using the buttons `Add', `Edit' and `Delete'. If an input ux is selected while adding a new input ux, the new input ux is inserted in the list immediately before the selected input ux, otherwise it is appended to the end of the list of input uxes. This gives the user the possibility to in uence the order of the input uxes the order is irrelevant for

114

CHAPTER 3. MODEL FORMULATION

Figure 3.84: Dialog box for editing surface inputs to a lake compartment.

Figure 3.85: Dialog box for editing a single surface input to a lake compartment. the program, but it may be convenient for the user to have a certain order. Figure 3.85 shows the dialog box used to specify a single surface input ux of a dynamic volume state variable. In this dialog box the elds `Variable' and `Input Flux' allow the user to select a variable and to specify a surface input ux, isurf;Ci or isurf;Xi . Note that this input ux represents a mass ux per unit lake surface area and per unit of time. A positive value of a surface ux represents a ux into the lake, a negative value a ux out of the lake. A surface input ux of a lake compartment may depend on dynamic volume state variables. These variables return the values of state variables in the lake immediately at the surface. The use of state variables makes it possible to model di usive uxes through the surface boundary layer. Input uxes for any type of variables can be speci ed, but only input uxes for dynamic volume state variables are used by the program. The reason for allowing to de ne input uxes for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of input uxes of the compartments. Selection of the radio button `Lateral Input' in the dialog box shown in Fig. 3.83 opens the dialog box shown in Fig. 3.86. The edit eld `Water In ow' of this dialog box is used to specify the discharge of water per unit depth into the lake, q, and the list box contains substance concentrations, Clat;i or Xlat;i , in the in owing water. A positive value

3.3. COMPARTMENTS

115

Figure 3.86: Dialog box for editing lateral inputs to a lake compartment.

Figure 3.87: Dialog box for editing a single lateral input to a lake compartment. of q represents a ow into the lake, a negative value represents an out ow. According to the equations 3.77, 3.89 and 3.90 the in ow concentration is irrelvant if q is smaller than or equal to zero out ow. For each variable only one unique lateral in ow concentration can be speci ed. The list of in ow concentrations can be edited using the buttons `Add', `Edit' and `Delete'. If an in ow concentration is selected while adding a new in ow concentration, the new in ow concentration is inserted in the list immediately before the selected in ow concentration, otherwise it is appended to the end of the list of in ow concentrations. This gives the user the possibility to in uence the order of the in ow concentrations the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.87 shows the dialog box used to specify a single lateral in ow concentration of a dynamic volume state variable. In this dialog box the elds `Variable' and `In ow Conc.' allow the user to select a variable and specify a lateral in ow concentration, Clat;i . A lateral in ow concentration of a lake compartment may depend on the program variables Discharge and on state variables. These variables return the discharge and the current values of state variables in the compartment as a function of the depth z . In ow concentrations for any type of variables can be speci ed, but only in ow concentrations for dynamic volume state variables are used by the program. The reason for allowing to de ne in ow concentrations for variables of other types is to

116

CHAPTER 3. MODEL FORMULATION

facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of lateral in ow concentrations of the compartments. Selection of the radio button `Point Inputs' in the dialog box shown in Fig. 3.83 opens the dialog box shown in Fig. 3.88. This dialog box contains a list of the names

Figure 3.88: Dialog box for editing point inputs to a lake compartment. of all point inputs to the lake typically rivers or technical discharges. This list can be edited with the aid of the buttons `Add', `Edit' and `Delete'. Figure 3.89 shows the dialog box used to specify a single point input to a lake. The edit eld `Name' is used to specify the name of the point input. Each point input needs a unique name as an identi er. A name of a point input consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of point inputs the edit eld `Description' can optionally be used to store comments on speci c implementation features of a compartment. The edit elds `Mean Depth' and `Range' are used to determine the vertical loaction of the input. The input is equidistributed over the depth range from Mean Depth minus one half of Range to Mean Depth plus one half of Range the input depth range is bounded by the lake bottom and the lake surface. The de nitions of Mean Depth and Range can depend on time in order to simulate temporary events of higher density of a river during a storm event. The edit eld `Water In ow' is used to specify the discharge of water of the point in ow. A positive value of this variable represents an in ow into the lake, a negative value an out ow. The list box `Concentrations' contains the concentrations of substances represented by dynamic volume state variables. For each variable only one unique in ow concentration can be speci ed. The infow concentrations speci ed here are irrelevant if the water in ow if negative or zero. The list of in ow concentrations can be edited using the buttons `Add', `Edit' and `Delete'. If an in ow concentration is selected while adding a new concentration, the new in ow concentration is inserted in the list immediately before the selected concentration, otherwise it is appended to the end of the list of in ow concentrations. This gives the user the possibility to in uence the order of the in ow concentrations the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.90 shows the dialog box used to specify an in ow concentration for a selected state variable to a lake. In this dialog box the

3.3. COMPARTMENTS

117

Figure 3.89: Dialog box for editing a single point input to a lake compartment.

Figure 3.90: Dialog box for editing a point input concentration for a lake compartment. elds `Variable' and `In ow Conc.' allow the user to select a variable and specify an in ow concentration. In ow concentrations for any type of variables can be speci ed, but only in ow concentrations for dynamic volume state variables are used by the program. The reason for allowing to de ne in ow concentrations for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement of editing the lists of in ow concentrations of the compartments. Selection of the radio button `Sediment Input' in the dialog box shown in Fig. 3.83 opens the dialog box shown in Fig. 3.91. This dialog box contains a list box for de ning the substance mass uxes, ised;Ci and ised;Xi , into the lake sediment. For each variable only one unique sediment input ux can be speci ed. The sediment input uxes can be edited using the buttons `Add', `Edit' and `Delete'. If a sediment input ux is selected while adding a new sediment input ux, the new input ux is inserted in the list

118

CHAPTER 3. MODEL FORMULATION

Figure 3.91: Dialog box for editing sediment inputs to a lake compartment. immediately before the selected input ux, otherwise it is appended to the end of the list of sediment input uxes. This gives the user the possibility to in uence the order of the sediment input uxes the order is irrelevant for the program, but it may be convenient for the user to have a certain order. Figure 3.92 shows the dialog box used to specify a single sediment input ux of a dynamic volume state variable. In this dialog box the

Figure 3.92: Dialog box for editing a single sediment input to a lake compartment. elds `Variable' and `Input Flux' allow the user to select a variable and to specify a surface input ux, ised;Ci or ised;Xi . Note that this input ux represents a mass ux per unit sediment area and per unit of time. A positive value of a sediment ux represents a ux into the sediment, a negative value a ux out of the sediment. A sediment input ux of a lake compartment may depend on dynamic volume state variables. These variables return the values of state variables in the lake sediment. The use of state variables makes it possible to model di usive uxes from deeper sediment layers to the sediment layer described by the lake model. Input uxes for any type of variables can be speci ed, but only input uxes for dynamic volume state variables are used by the program. The reason for allowing to de ne input uxes for variables of other types is to facilitate the users switching between models with di erent state variables. A user can change a state variable to a variable of another type without the requirement ot editing the lists of input uxes of the compartments. The button `Particulate Variables' of the dialog box shown in Fig. 3.78 allows the

3.3. COMPARTMENTS

119

user to specify properties of particulate variables for dynamic volume state varibles. Figure 3.93 shows the dialog box used for this purpose. It contains a list box with the

Figure 3.93: Dialog box for editing particulate variables of a lake compartment. names, densities and sedimentation velocities of all de ned particulate variables. Note that a variable can only be once in both lists of particluate and dissolved variables. The buttons `Add', `Edit' and `Delete' allow the user to edit this list. Figure 3.94 shows the dialog box used to specify the properties of a particulate variable in a lake. The edit elds

Figure 3.94: Dialog box for editing a single particulate variable for a lake compartment.

`Variable', `Density' and `Sedimentation Velocity' allow the user to select a variable
and to specify a density, Xi , and a sedimentation velocity, wsed;i. The density Xi is used in equation 3.82 to calculate the volume ux of particles between the sediment layers. A positive sedimentation velocity, wsed;i, describes downward movement. According to the di erential equations desciribed above, this downward movement includes movement through the water column and settling to the sediment if the lake cross-sectional area decreases with increasing depth of the lake. If for some special purposes, a negative sedimentation velocity is used, this indicates upward movement. In this case, an increase of the lake cross-sectional area with decreasing depth leads to a dilution of the moving particles. Note that properties of particulate variables can be speci ed for any type of

120

CHAPTER 3. MODEL FORMULATION

variables, but that only properties of dynamic volume state variables have an e ect. The larger class of variables is allowed in order to facilitate the users to switch between models with di erent state variables. The button `Dissolved Variables' of the dialog box shown in Fig. 3.78 allows the user to specify properties of dissolved variables for dynamic volume state variables. Figure 3.95 shows the dialog box used for this purpose. It contains a list box with the names and

Figure 3.95: Dialog box for editing dissolved variables of a lake compartment. molecular di usion coe cients of all de ned dissolved variables. Note that a variable can only be once in both lists of particluate and dissolved variables. The buttons `Add', `Edit' and `Delete' allow the user to edit this list. Figure 3.96 shows the dialog box used to specify the properties of a dissolved variable in a lake. The edit elds `Variable' and

Figure 3.96: Dialog box for editing a single dissolved variable of a lake compartment.

`Mol. Di usion' allow the user to select a variable and to specify a molecular di usion
coe cient, D. This di usion coe cient determines the exchange of dissolved substances between the sediment layers and between the top sediment layer and the water column of the lake. Note that properties of dissolved variables can be speci ed for any type of variables, but that only properties of dynamic volume state variables have an e ect. The larger class of variables is allowed in order to facilitate the users to switch between models with di erent state variables.

3.3. COMPARTMENTS

121

The button `Sed. Prop.' of the dialog box shown in Fig. 3.78 is used to specify the properties of the sediment submodel of the lake compartment. The de nitions of this submodel are only relevant if the radio button `with Sediment' is selected. Fig. 3.97 shows the dialog box used for this purpose. The edit eld `Vol. Flux' is used to specify a

Figure 3.97: Dialog box for editing the sediment submodel of a lake compartment. volume ux of particles to the sediment. This volume ux makes it possible to consider a particle ux to the sediment of particles that are not modelled as state variables. In the edit els `Resusp. Veloc.' a resuspension velocity can be speci ed. The resuspension velocity can be used to model resuspension events. The list box `Sediment Layers' shows the properties of all sediment layers currently de ned. The sediment layers can be edited with the aid of the buttons `Add', `Edit' and `Delete' below the list box. The dialog box shown in Fig. 3.98 is used to edit the properties of a single sediment layer. For

Figure 3.98: Dialog box for editing a sediment layer. each sediment layer a `Zone Index', the sediment layer `Thickness' and the `Porosity' must be speci ed. The button `TKE Prop.' of the dialog box shown in Fig. 3.78 is used to specify the properties of the submodel for the calculation of turbulent kinetic energy. The de nitions

122

CHAPTER 3. MODEL FORMULATION

of this submodel are only relevant if the radio button `with TKE' is selected. Fig. 3.99 shows the dialog box used for this purpose. The edit eld `Prandtl Number' is used to

Figure 3.99: Dialog box for editing the turbulent kinetic energy submodel of a lake compartment. de ne the value of the Prandtl number, Pr. The numbers `Sigma TKE' and `Sigma Diss.' are used to de ne the values of the modi cation parameters k and for di usion of turbulent kinetic energy and dissipation, respectively. The edit eld `Minimum Di usion Coe cient for buoyancy production of TKE' is used to specify the value of Kz;min used in equation 3.79. The edit eld `Surf. Shear' is used to specify the shear at the lake surface, surf . The surface shear is used to drive the horizontal motion described by the velocity U . It appears in the boundary condition 3.94b. The surface shear can be estimated as follows Amorocho and de Vries, 1980:
surf
2 = air cz Uw;10m

3.104

with

0:001 for Uw;10m  7 m s cz  : 0:001 + 0:00015 s m Uw;10m , 7 m s for 7 m s Uw;10m  17 m s 0:0025 for 17 m s Uw;10m  20 m s

3.105

The edit eld `Press. Grad.' is used to specify the source term rU for the horizontal velocity used in equations 3.77 and 3.86. The edit eld `Prod. Diss.' is used to specify the production of dissipation of turbulent kinetic energy. The expression used in the standard k- model is

c1 P + c3 G k , c2 k

3.106

3.3. COMPARTMENTS

123

The edit elds `Wind Excitation', `Bottom Frict.' and `Internal Frict.' are used to specify the parameters Pwind , Pbottom and pint . Empirical formulations for these parameters are required Goudsmit et al., 1996.

Figure 3.100: Dialog box for editing the accuracies of program variables of a lake compartment. The button `Acc.' of the dialog box shown in Fig. 3.78 is used to specify the numerical accuracies of program variables. Fig. 3.100 shows the dialog box used for this purpose. It allows the user to specify relative and absolute accuracies of the variables `Discharge', Q, `Horizontal Velocity', U , `Turbulent Kinetic Energy', k, `Dissipation', , and `Seiche Energy', Eseiche, in the compartment. Good behaviour of the numerical algorithms is usually achieved if the absolute accuracy and the product of the relative accuracy times a typical value of the variable both are 4 to 6 orders of magnitude smaller than typical values of the variable. The check box `active for calculation' can be used to activate or inactivate the compartment from the calculations. This check box has the same functionality as the buttons Activate and Inactivate in the dialog box shown in Fig. 3.19.

124

CHAPTER 3. MODEL FORMULATION

In Table 3.9 the program variables available in a lake compartment are summarized for a complete overview of all program variables see Table 3.1 on page 17.
Calculation Number Time Compartment Index Zone Index

Discharge Water Fraction Space Coordinate Z Cross Sectional Area Density Area Gradient Brunt Vaisala Frequency Horizontal Velocity Turbulent Kinitic Energy TKE Shear Production of TKE Buoyancy Production of TKE Dissipation Energy of Seiche Oscillation

Identi er for calculations value set in the dialog boxes shown in Figs. 4.5 and 4.17. Simulation time. Identi er for compartments value set in the dialog box shown in Fig. 3.78. Identi er for zones within compartments returns 0 in the water column and the value speci ed in the dialog box 3.98 in the sediment layers. Volumetric ow rate. Volumetric fraction of water returns a value of 1. Depth coordinate in the compartment. Area of water body perpendicular to the ow direction returns the horizontal area of the lake. Density of the water. Gradient of the cross-sectional area as a function of depth. Stability frequency of the water column. Velocity of horizontal wind-driven ow. Turbulent kinetic energy TKE per unit mass of water in the lake. Production of TKE due to shear of horizontal velocity. Production or loss of TKE due to density di erences. Dissipation of turbulent kinetic energy. Total energy stored in Seiche motion.

Table 3.9: Program variables available in the lake compartment.

3.4. LINKS

125

3.4 Links
The compartments introduced in section 3.3 can be connected by links in order to model water and substance exchange between the compartments. Two types of links are distinguished: Advective Links are used to describe water ow and advective substance transport from one compartment to another. Di usive Links model di usive boundary layers or membranes between compartments which can be di usively penetrated by certain substances. Figure 3.101 shows the dialog box for editing links. This dialog box is opened with the

Figure 3.101: Dialog box for editing links. command in the Edit menu shown in Figure 3.2. It is of modeless type in order to facilitate the editing process. The names of all links already de ned are listed alphabetically in the list box of this dialog box. The type of the currently selected compartment is indicated at the bottom of this dialog box. The buttons of this dialog box allow the user to perform the following operations with links: New links may be created directly or old links may be duplicated in both cases the new link needs a new name. The data items of a link can be edited. Finally, it is possible to delete links. The buttons Duplicate, Edit and Delete are inactive as long as no link is selected. Clicking the Close button results in closing this dialog box. It can be reopened by choosing the Links command in the Edit menu shown in Fig. 3.2. After clicking the button New in the dialog box shown in Fig. 3.101 the link type can be selected in the dialog box shown in Fig. 3.102. The two types of links shown in this dialog box are described in more detail in the following two subsections.
Links

126

CHAPTER 3. MODEL FORMULATION

Figure 3.102: Dialog box for selecting the type of a link.

3.4.1 Advective Link

Overview
An advective link connects an out ow connection of a compartment with an in ow connection of another compartment. An advective link can have an arbitrary number of bifurcations through which any fractions of water and substance ow can bifurcate to a further compartment. To each out ow connection of a compartment at most one advective link can be connected. To each in ow connection of a compartment an arbitrary number of out ows or bifurcations of advective links can be connected. Instead of connecting out ows or bifurcations of advective links to an in ow connection of a compartment, the connection can be left open so that the ows through such a link or bifurcation leave the modelled part of the system.

Equations Solved by AQUASIM

The water in ow, Qin, as well as the mass ow of substances into the link, Iin;i , are given by the out ow from the compartment to which the in ow of the link is connected. The water and substance ows through the bifurcation j , Qbif;j and Ibif;j;i, can be speci ed by the user as functions of the water in ow, Qin , and of the substance in ow concentrations. The index i runs over all dynamic volume state variable which model concentrations of substances transported with the water ow. The water and substance out ows of the link are then calculated by AQUASIM according to the equations

Cin;i = Iin;i Q
in

3.107

Qef = Qin ,
and

nbif X j =1

Qbif;j

3.108

Ief;i = Iin;i ,

nbif X j =1

Ibif;j;i

3.109

respectively. The program user is responsible that the sum of the bifurcating water ows which all must be nonnegative does not exceed the water in ow to the link. In contrast, the bifurcating substance ows may be negative modeling substance input from the bifurcation

3.4. LINKS

127

and the sum of all bifurcating uxes may be larger than the input ux this leads to a negative input ux to the compartment to which the link is connected; in some cases, this may be meaningful, e.g. to model the e ect of an oxygen saturation de cit.

User De nitions
Figure 3.103 shows the dialog box used for de ning or editing an advective link. The edit

Figure 3.103: Dialog box for editing an advective link. eld `Name' is used to specify the name of the link. Each link needs a unique name as an identi er. A name of a link consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The edit eld `Link Index' can be used to specify a nonnegative number as a link index. This value can be accessed with the aid of the program variable Link Index to make variables dependent on the link. To improve documentation of links, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a link. The list button `From Compartment' allows the user to select the compartment from which the advective uxes feed into the link. If the selected compartment has more than one advective output connection, the list button `Connection' to the right of the list button From Compartment allows the user to select the connection. Because the out ow from the connection of the compartment selected here determines water and substance uxes into the link, a compartment must be selected with the aid of the list button From Compartment. The list buttons `To Compartment' and `Connection' to the right of the list button To Compartment allow the user to select to which input connection of which compartment the water and substance uxes not bifurcating at the link are owing. The eld To Compartment can be left blank in which case these uxes leave the modelled system. The list box `Bifurcations' contains the names of all bifurcations that have been de ned for the link. These bifurcations can be edited using the buttons `Add', `Edit' and `Delete'.

128

CHAPTER 3. MODEL FORMULATION

The dialog box used for editing a bifurcation is shown in Figure 3.104.

Figure 3.104: Dialog box for editing a bifurcation of an advective link. The edit eld `Name' is used to specify the name of the bifurcation. Each link needs a unique name as an identi er. A name of a link consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of links, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a link. The list buttons `To Compartment' and `Connection' allow the user to select to which input connection of which compartment the water and substance uxes of the bifurcation are owing. The eld `To Compartment' can be left blank in which case these uxes leave the modelled system. The bifurcating `Water Flow', Qbif , can be speci ed as an algebraic expression of globally available variables, the program variables, Time, Calculation Number and Link Index and of the program variable Discharge which evaluates here to the in ow to the link. Flows of substances the concentrations of which are represented by dynamic volume state variables, can either be selected to divide `with water ow' or to be `as given below' in the list box of mass `Fluxes'. In the rst case the mass uxes of all dynamic volume state variables divide proportional to the division of water ow as it is typical for dissolved substances, in the second case the mass uxes of all substances transported through the bifurcation have to be speci ed. This second option can be used to de ne mass separation e.g. increase in concentrations of particulate substances at the bottom outlet of a settler. If the mass uxes are selected to be given, they are displayed in the list box and can be edited using the buttons `Add', `Edit' and `Delete'.

3.4. LINKS

129

Figure 3.105 shows the dialog box used for de ning a mass ux for a substance the

Figure 3.105: Dialog box for editing a mass ux within a bifurcation of an advective link. concentration of which is represented by a dynamic volume state variable. The list button `Variable' is used to select the state variable, in the edit eld `Flux' the mass ux, Ibif;i can be speci ed as an algebraic expression. In this expression, globally available variables, the program variables Time, Calculation Number and Link Index, the program variable Discharge representing the in ow to the link not the bifurcating ow, and dynamic volume state variables representing link-in ow concentrations, can be used. Bifurcation mass uxes can be de ned for any type of variable, but only those for dynamic volume state variables have an e ect.

3.4.2 Di usive Link

Overview
A di usive link connects two di usive connections of compartments. It is used to model diffusive mass exchange of substances between compartments through membranes or boundary layers. There is no water ow through a di usive link.

Equations Solved by AQUASIM

The mass ux of a substance i through a di usive link from compartment 1 to compartment 2 is given by the equation

Ii = qex;ifiCi;1 , Ci;2 3.110 where Ci;1 and Ci;2 are the concentrations of substance i in the compartment 1 and 2, respectively, qex;i is the exchange coe cient, and fi is a conversion factor that allows for
the description of phase transitions. This conversion factor is unity, if both compartments 1 and 2 contain the same solvent. If compartment 1 is a mixed reactor compartment modelling a gas phase, compartment 2 models a water phase, the conversion factor is the inverse of the non-dimensional Henry coe cient of substance i: 3.111 f= 1
i

In many cases the mass exchange coe cient, qex;i, is given as the product of the surface area of the membrane or boundary layer, A, and a mass transfer coe cient, ki

Hi

qex;i = Aki

3.112

130

CHAPTER 3. MODEL FORMULATION

which itself is often given as the ratio of the di usion on coe cient Di of the substance in the boundary layer or membrane and the thickness LM of the boundary layer or membrane

ki = Di LM
Instead of these expressions, an empirical exchange coe cient can be used.

3.113

User De nitions
Figure 3.106 shows the dialog box used for de ning or editing a di usive link.

Figure 3.106: Dialog box for editing a di usive link. The edit eld `Name' is used to specify the name of the link. Each link needs a unique name as an identi er. A name of a link consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. The edit eld `Link Index' can be used to specify a nonnegative number as a link index. This value can be accessed with the aid of the program variable Link Index to make variables dependent on the link. To improve documentation of links, the edit eld `Description' can optionally be used to store comments on speci c implementation features of a link. The list buttons `Compartment 1' and `Compartment 2' allow the user to select the two compartments to be linked di usively. If a selected compartment has more than one di usive connection, the list button `Connection' can be used to select one of these. An arbitrary number of di usive links can be connected to the same di usive connection of a compartment. If both compartments contain the same solvent, it is irrelevant which compartment is connected as compartment 1 and which as compartment 2. This is only relevant if the conversion factor of a substance is not unity. In this case it has to be noted that the concentration in compartment 1 is multiplied with the conversion factor in equation 3.110.

3.4. LINKS

131

and `Delete'. Figure 3.107 shows the dialog box used for de ning the exchange coe cient and the

qex;i, and conversion factors, fi. This list can be edited using the buttons `Add', `Edit'

The list box in the dialog box shown in Fig. 3.106 shows the list of exchange coe cients,

Figure 3.107: Dialog box for editing an exchange coe cient of a di usive link. conversion factor of a substance. The list button `Variable' is used to select the state variable, in the edit eld `Exchange Coe cient', the exchange coe cient, qex;i, can be speci ed and in the edit eld `Conversion Factor 1', the value of the conversion factor, fi, for the substance concentration in compartment 1 can be given. The program variables Time, Calculation Value and Link Index can be used for these de nitions in addition to all globally available variables. Exchange coe cients and conversion factors can be de ned for any type of variables, but only exchange coe cients for dynamic volume state variables have an e ect.

132

CHAPTER 3. MODEL FORMULATION

3.5 Numerical Parameters

In order to integrate the di erential equations of the user speci ed model in AQUASIM, as a rst step, the partial di erential equations are discretized in space. Then, the spatially discretized partial di erential equations together with the ordinary di erential equations and the algebraic equations are integrated numerically in time with the algorithm DASSL Petzold, 1983 which is based on the implicit backward di erencing variablestep, variable-order Gear integration technique Gear, 1971b; Gear, 1971a; Gear, 1971c. In the dialog box Edit Numerical Parameters shown in Fig. 3.108, which is opened with the Numerical Parameters command in the Edit menu, the general numerical parameters of the time integration algorithm DASSL Petzold, 1983 can be edited.

Figure 3.108: Dialog box for editing numerical parameters. The rst parameter, `Maximum Internal Step Size', makes it possible to bound the step size used internally by the numerical integration algorithm. The output step size speci ed in the dialog box Edit Calculation Definition shown in Fig. 4.5 only de nes the points of time at which the model results are stored in memory to be available for plotting and postprocessing of results. The internal step size used by the integration algorithm is chosen dynamically in order to optimise the integration e ciency while maintaining the requested integration accuracy see below for how to specify the integration accuracy of state and program variables. The maximum size of these internal time steps is ten times the output step size. In addition, the step size is bounded by the number speci ed in the eld Maximum Internal Step Size of the dialog box shown in Fig. 3.108. A too small value of this numerical parameter makes the integration ine cient slow because too many steps must be performed by the algorithm. Under most circumstances, a very large value has no negative e ect because the step actually used is then bounded by the accuracy requirement for the solution. There is one very important exception from this general rule: If a system has a very smooth temporal behaviour that is interrupted by very short excitations inputs or process rates, then, due to the large time step selected during the smooth period, the integration algorithm may step over some of the excitations. If one integration step leads to an evaluation of the input or process rate during an excitation period, the algorithm recognises the problem and repeats the step by a sequence of smaller steps in order to resolve the time course of the excitation accurately. However, if no step leads to an evaluation of the input or process rate during the excitation period, the algorithm steps over the excitation and fails to model the system correctly. In such

3.5. NUMERICAL PARAMETERS

133

situations, the numerical parameter Maximum Internal Step Size can be used to bound the internal step size to a value slightly smaller than the duration of the excitation period in order to avoid this problem. The second parameter, `Maximum Integration Order', in the dialog box shown in Fig. 3.108 makes it possible to bound the order of the temporal discritization scheme. By default the integration order is chosen dynamically between one and ve in order to optimise the integration e ciency while maintaining the requested integration accuracy see below for how to specify the integration accuracy of state and program variables. A smaller maximum integration order leads to smaller time steps required to maintain the integration accuracy. For this reason it is usually advisable to leave the value of the numerical parameter Maximum Integration Order at its default value of ve. The third parameter, `Number of Codiagonals of the Jacobian Matrix', can be used to increase the calculation e ciency for systems with a linear geometry. The implicit integration algorithm applied by DASSL Petzold, 1983 requires the evaluation of the matrix of partial derivatives of the components of the right hand side of the system of ordinary di erential equations obtained after spatial discretization with respect to the components of the state vector of the system, the so-called Jacobian matrix. The state vector of the system consists of the sequence of all active state variables and those program variables that are calculated by time integration or as the solution of algebraic equations at all grid points. For systems with a linear geometry e.g. a river, the time derivative of a component of the state vector depends only on the components representing other state or program variables at the same or at the neighbouring grid points. This leads to a Jacobian matrix in which all elements that are not zero are close to the main diagonal of the matrix. The limitation of the number of codiagonals to twice the number of state or program variables at one grid point then increases the e ciency of the algorithm considerably because all elements further away from the main diagonal which are zero are not evaluated by the program. It is evident that such a limitation to a banded Jacobian by selecting a small number in the edit eld Number of Codiagonals of the Jacobian Matrix leads to problems in the case of recirculation because then the time derivative at the rst grid point depends on the values in the last grid point, what leads to nonzero elements of the Jacobian matrix far away from its main diagonal. If the Number of Codiagonals of the Jacobian Matrix is given a too small value, the algorithm will fail to converge and the integration is stopped with an error message. A value at least equal to the number of equations minus one leads to the use of the full Jacobian matrix. It is not dangerous to try to nd a good value of this numerical parameter because if convergence is obtained, convergence is always to the correct solution even if the Jacobian matrix is not correct this is a very useful property of the Gear algorithm which makes it possible to use the same Jacobian matrix for several time steps and update it only if convergence becomes slow. The recommended value for this parameter is twice the number of state and program variables integrated in the compartment for systems with a linear geometry, and a value larger than the number of equations for a system with recirculations this leads to the use of the full Jacobian matrix. The fourth parameter, `Maximum Number of Internal Time Steps for One External Time Step', is useful for the detection of integration problems. In many cases, the fact that a huge number of steps is required for one output time step, is an indication of an integration problem. For this reason, this number can be limited in the edit eld Maximum Number of Internal Time Steps for One External Time Step. If a solution is calculated with a very large output time step for relaxation followed by a

134

CHAPTER 3. MODEL FORMULATION

period with higher numerical resolution, this numerical parameter can lead to unwanted interrupts of the integration during the initialization phase. In this case, the value of this parameter can be increased to avoid this behaviour. In most other cases, however, an interruption of a simulation caused by this parameter is a strong indication for numerical integration problems. In addition to the general numerical parameters of the time integration algorithm shown in Fig. 3.108, the required integration accuracies of the state variables and of those program variables that are integrated by AQUASIM, are additional important numerical parameters as well. The integration accuracy of the state variables can be speci ed in the edit elds Rel. Accuracy and Abs. Accuracy of the dialog box shown in gure 3.5. In the dialog boxes used for the de nition of the compartments see Figs. 3.21, 3.29, 3.41, 3.53, 3.67 and 3.78, the button Acc. allows the program users to specify the integration accuracy of program variables which are integrated in the current compartment see Figs. 3.28, 3.40, 3.51, 3.66, 3.77 and 3.100. As a last point, in the dialog boxes used for editing compartments see Figs. 3.29, 3.41, 3.53, 3.67 and 3.78 the Number of Grid Points and the Resolution of the algorithm can be selected by the user. Low Resolution corresponds to a rst-order discretization in space, High Resolution corresponds to a discretization of at least second order combined with a ux limiter that is used to avoid numerical oscillations.

3.6. DELETING CALCULATED STATES

135

3.6 Deleting Calculated States

Calculations, as described in chapter 4, lead to the storage of calculated states of the user-de ned system. In this section it is described, how the user can check which states are currently stored in memory and how selected states can be deleted. In the dialog box Delete Calculated States shown in Fig. 3.109, which is used to check which states are available and to delete selected states, is opened with the Delete States command in the Edit menu. Selective deletion of calculated states may be advan-

Figure 3.109: Dialog box for deleting calculated states. tageous to save memory during program operation or disk space if the calculated states are saved together with the system de nitions if no calculated states are to be saved, it is not necessary to delete them before saving, because the user is asked during the saving process if the calculated states are to be saved in addition to the system de nitions. As an example, the user can delete all states with the exception of the last one in order to be able to continue the simulation from a relatively small saved le calculated states may enlarge the le size considerably. The left list box of the dialog box shown in Fig. 3.109 shows the calculation numbers for which calculated states are currently available. The right list box of this dialog box shows the states that are available for the calculation number selected in the left list box the right list box is empty if in the left list box no or more than one calculation numbers are selected. The button `Delete' below the left list box makes it possible to delete all calculated states for all selected calculation numbers, the button `Delete' below the right list box can be used to delete the selected calculated states for the calculation number selected in the left list box. Note that initializing a simulation for a given calculation number also leads to the deletion of all calculated states for this calculation number and that editing operations which lead to a change in the number of di erential equations to be solved e.g. addition deletion of a compartment, activation inactivation of state variables, changes of the numbers of

136

CHAPTER 3. MODEL FORMULATION

grid points of comartments etc. also leads to the deletion of all calculated states for all calculation numbers.

Chapter 4

Simulation and Data Analysis

Fig. 4.1 shows the menu
Calc

of AQUASIM. The item `Simulation' of this menu is

Figure 4.1:

Calc

menu.

used to de ne and execute simulations for the user-de ned model, the item `Sensitivity Analysis' is used to de ne and execute sensitivity analyses, and the item `Parameter Estimation' is used to de ne parameter estimation procedures and to estimate model parameters. These three items of the menu Calc are described in detail in the three sections of this chapter. During all calculation operations, information on the system of equations currently solved, on the progress of the integration, sensitivity analysis or parameter estimation algorithm, and on numerical problems is written to the log le aquasim.log in the program directory of AQUASIM. In the case of numerical problems this information may be very useful in order to help to locate the source of the error. Some help for the interpretation of the information contained on this le is given in section 6.3. The le is overwritten every time AQUASIM is started. 137

138

CHAPTER 4. SIMULATION AND DATA ANALYSIS

4.1 Simulation
Execution of a simulation is equivalent to numerically integrating a system of ordinary and partial di erential equations in time and simultaneously solving the algebraic equations. In AQUASIM, as a rst step, the partial di erential equations are discretized in space. Then, the spatially discretized partial di erential equations together with the ordinary di erential equations and the algebraic equations are integrated numerically in time with the algorithm DASSL Petzold, 1983 which is based on the implicit backward di erencing variable-step, variable-order Gear integration technique Gear, 1971b; Gear, 1971a; Gear, 1971c. Spatial discretization of partial di erential equations is done using conservative nite di erence schemes LeVeque, 1990. The di erential conservation laws 3.6, 3.35, 3.44, 3.56, and 3.73 are discretized as j j d ^x ; t = ,^num xi+1=2 ; t , ^num xi,1=2 ; t + ^x ; t r i 4.2 i dt xi+1=2 , xi,1=2 using the division of the spatial coordinate axis in grid points as shown in Fig. 4.2. The user
c e ll 1 c e ll 2 c e ll i-2 c e ll i-1 c e ll i c e ll n -2

@ ^ = , @^ + ^ j r @t @z

4.1

x
1

x
2

x
3

i-1

x
i

x x
i+ 1 /2

i+ 1

n -1

x
n

i-1 /2

Figure 4.2: Division of the spatial coordinate axis in cells and grid points for spatial discretization of partial di erential equations note that the number of cells is equal to the number of grid points minus 2 and that the grid spacing is smaller at the boundaries; n is the number of grid points speci ed in the dialog boxes shown in Figs. 3.29, 3.41, 3.53, 3.67 and 3.78. has the choice between a low resolution  rst-order and a high resolution second-order with ux limiter numerical ux ^num . For the high resolution ux, the van Leer ux j limiter is applied van Leer, 1974; Sweby, 1984; LeVeque, 1990. Because time integration uses a high order discretization, numerical di usion for the rst-order technique can be estimated to be Dnum  v
x 4.3 2 where v is the advective transport velocity and
e
x = xn , xs ,2

4.4

4.1. SIMULATION

139

is the grid spacing xe , xs is the length of the compartment and n the number of grid points. The main advantage of the Gear integration technique for the solution of systems of ordinary di erential equations Gear, 1971b; Gear, 1971a; Gear, 1971c; Hindmarsh, 1983 is its property of being sti y stable. The implementation DASSL Petzold, 1983 of this technique, which is used in AQUASIM, has the advantage that not only systems of ordinary di erential equations, but di erential-algebraic systems can be solved. Sti systems of di erential equations are systems for which the transients towards the solution to be calculated approach this solution on a much shorter time scale as the time scale for variations of the solution of interest. The main idea of achieving a sti y stable integration technique is using backward di erencing schemes. The advantage of backward versus forward di erencing can easily be illustrated with a comparison of the explicit and the implicit Euler algorithm. If the system of ordinary di erential equations is written in the form dy = Fy; t 4.5 dt the explicit Euler technique is given as

ytj+1  = ytj  + tj+1 , tj Fytj ; tj 

and the implicit Euler technique as

4.6 4.7

ytj+1  = ytj  + tj+1 , tj Fytj+1 ; tj+1

Figure 4.3 shows the numerical solution of a sti system of di erential equations with the explicit technique 4.6 left and with the implicit technique 4.7 right. Because the
f f

Figure 4.3: Numerical solution of a sti system of di erential equations using the explicit left and the implicit right Euler algorithm with the same size of the time step. numerical solution with the implicit technique is tangent to the true solution at the end of each time step and not at the beginning, the technique can use much larger time steps. For sti systems of di erential equations this advantage leads to a faster integration although a implicit system of nonlinear equations must be solved at each time step the explicit

140

CHAPTER 4. SIMULATION AND DATA ANALYSIS

technique requires only the evaluation of explicit functions at each time step. Whereas Fig. 4.3 demonstrates the main idea of achieving a sti y stable integration algorithm by backward di erencing, the Gear algorithm Gear, 1971b; Gear, 1971a; Gear, 1971c uses a more sophisticated generalization of this idea to a multi-step, variable step size and variable order originally up to 6, in the implementation DASSL up to 5 algorithm. The rst disadvantage of the Gear integration technique and the implementation DASSL Petzold, 1983 is its inability to step over discontinuities of inputs or process rates. Such step discontinuities can be approximated by a linear increase or decrease of the quantity of interest over a very short time interval compared to the typical time scale of the problem of interest. With this approximation, the algorithm still has to decrease the time step considerably at the points where the time derivative is discontinuous the integration order is also reduced to 1. As an alternative, a smoothed step function with continuous derivatives can be used, e.g. a step approximated by third order spline polynomials for which, in addition to the value of the function, also the rst and second derivatives are continuous. A second disadantage of the Gear algorithm is that it requires the evaluation of the jacobian matrix for solving the nonlinear system of algebraic equations at each time step. The implementation DASSL Petzold, 1983 allows only the choice between using the full or a banded jacobian matrix. For linearly arranged systems of compartments and, in AQUASIM, if the alphabetic order of compartments corresponds to the sequence of ow the option of a banded jacobian matrix, with a number of codiagonals cf. section 3.5 equals to twice the number of equations solved at one grid point, leads to a very e cient solution procedure. However, in the case of branched systems or systems with recirculations, due to typically only a few nonzero elements of the jacobian matrix far from its main diagonal, the option of a full jacobian matrix must be used. The evaluation of the jacobian matrix then consumes much computation time and can make the integration ine cient. Figure 4.4 shows the dialog box used for de ning and starting simulations. This dialog box is opened with the Simulation command in the Calc menu shown in Figure 4.1. It is of modeless type in order to facilitate performing simulations. A simulation consists of one or more calculations. The names of all active calculations are listed in the left list box of the dialog box shown in Fig. 4.4, those of all available calculations in the right list box. The buttons between these two list boxes allow the user to perform the following operations: By clicking the button `New', new calculations can be de ned. Alternatively, by clicking the button `Duplicate', the selected calculation can be duplicated. With the button `Edit', or by double-clicking the calculation name in the list box, a calculation can be edited. The button `Delete' allows the program users to delete calculations. The buttons Duplicate, Edit and Delete are inactive as long as no calculation is selected. With the button `Activate', calculations selected in the right list box can be activated, with the button `Inactivate', calculations selected in the left list box can be inactivated. Active calculations must have di erent values of their calculation numbers. This means that activation of a calculation is impossible, if there is already an active calculation with the same calculation number. Execution of a simulation leads to the storage of calculated states of the user-de ned model. Clicking the button `Initialize' deletes the calculated states for the calculation

J = @F @y

4.8

4.1. SIMULATION

141

Figure 4.4: Dialog box for editing calculation de nitions and executing simulations. numbers of all active calculations and generates newly calculated initial states. Clicking this button is only necessary before a simulation, for which there exist already calculated states, is to be redone. The button `Start Continue' is used to continue the active calculations from the last calculated states if there are no calculated states, the simulation is automatically initialized before the dynamic simulation is started. The buttons Initialize and Start Continue are inactive as long as there is no active calculation. Clicking the Close button results in closing this dialog box. It can be reopened by choosing the Simulation command in the Calc menu shown in Fig. 4.1. As mentioned above, the buttons New, Edit and Duplicate of the dialog box shown in Fig. 4.4 are used to de ne calculations for the user-de ned model. Figure 4.5 shows the dialog box used for de ning and editing a single calculation. The edit eld `Name' is used to specify the name of the calculation. Each calculation needs a unique name as an identi er. A name of a calculation consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of calculations the edit eld `Description' can optionally be used to store comments on the purpose of the calculation de nition. The edit eld `Calc. Number' allows the program user to identify a calculation by a nonnegative integer number. Variables and thus also processes, etc. can be made dependent on this calculation number with the aid of the program variable Calculation Number. Active calculations must have di erent values of their calculation numbers. Each calculation must be initialized before it can be started. In the edit eld `Initial Time' the time at which a calculation is initialized can be speci ed. The user has the choice between two types of the `Initial State' to be used by the program. The default option `given, made consistent' applies the initial state as given by the user a value of zero is the default initial condition for all state variables for which the user did not explicitly specify an initial condition. If this initial state violates constraints on boundary conditions or algebraic equations of equilibrium processes, the program tries to ful ll these constraints by small modi cations to the initial state given by

142

CHAPTER 4. SIMULATION AND DATA ANALYSIS

Figure 4.5: Dialog box for editing a calculation de nition. the user. This procedure is called to make the initial state consistent. With the selection of the alternative option `steady state' the program tries to nd the steady state solution of the user-de ned model under external parameter values evaluated at the initial time and uses this steady state solution as the initial condition. Note, however, that not all models have a steady state solution and that even if such a solution exists, the numeric algorithm used by AQUASIM may fail to nd it. In such a situation, the convergence can be improved by a good choice of initial conditions which are used as starting values for the iterative search process for the steady state solution. If this does not help the steady state solution must be found by relaxation, i.e. by executing a dynamic simulation with constant boundary conditions. In the list box `Output Steps' the user can de ne sequences of time steps that are executed consecutively. To do this, the `Step Size' and `Number of Steps' must be entered in the edit elds below the list box and then one of the buttons Add or Replace must be clicked. If no row of the list box is selected, `Add' leads to insertion of the new step sequence at the end of the list, otherwise the new step sequence is inserted before the selected row. `Replace' leads to the replacement of the selected row by the step sequence speci ed in the edit elds below the list box. Clicking the button `Delete' leads to the deletion of the selected rows of the list. Note that the output steps de ned for a calculation specify the points of time at which the calculated states are stored in memory and are available later on for plotting results and for exporting results to other programs. The internal step size used by the integration algorithm, however, is determined according to the requirement of ful lling the accuracy requirements of the state and program variables as described in section 3.5. This internal step size is usually much smaller than the output step size, however, it can increase to ten times the output step size if all variables

4.1. SIMULATION

143

change only smoothly in time the solution at the output steps is then determined by interpolation. The size of the internally used time step can be bounded as described in section 3.5 this may be useful in order to avoid that the algorithm steps over short excitations. Finally, the check boxes `active for simulation' and `active for sensitivity analysis' are used to activate the calculation for the speci ed task. During the execution of a simulation, the dialog box shown in Fig. 4.6 shows the

Figure 4.6: Dialog box for interrupting a simulation. current calculation and the current simulation time and it allows the user to interrupt the simulation.

144

CHAPTER 4. SIMULATION AND DATA ANALYSIS

4.2 Sensitivity Analysis

As mentioned in the introduction, sensitivity analysis combines the tasks of identifyability analysis and uncertainty analysis. The goal of identi ability analysis is to check if model parameters can be uniquely determined with the aid of the available data and to estimate the uncertainty of the parameter estimates. This can be done by estimating the standard errors and correlation coe cients of parameters during the parameter estimation procedure as described in section 4.3. An additional source of information on the identi ability of model parameters are the sensitivity functions discussed in this section. The following four sensitivity functions are distinguished by AQUASIM:
a;a y;p = r;a y;p = a;r y;p

@y @p y @p
1 @y

4.9a 4.9b 4.9c 4.9d

y
ya , ,p r
y

= p @y @p

r;r y;p =

p @y y @p

y (p )

ya , ,p r

2 p

a;r Figure 4.7: Interpretation of the absolute-relative sensitivity function y;p = p@y=@p.

In these functions, y is an arbitrary variable calculated by AQUASIM and p is a model parameter represented by a constant variable or by a real list variable. The absoluteabsolute sensitivity function 4.9a measures the absolute change in y per unit of change

4.2. SENSITIVITY ANALYSIS

145

in p, the relative-absolute sensitivity function 4.9b measures the relative change in y per unit of change in p, the absolute-relative sensitivity function 4.9c measures the absolute change in y for a 100  change in p, and the relative-relative sensitivity function 4.9d measures the relative change in y for a 100  change in p. All these changes are calculated in linear approximation only. The most useful sensitivity functions are the absolute-relative sensitivity function 4.9c and the relative-relative sensitivity function 4.9d, because their units do not depend on the unit of the parameter. This makes quantitative comparisons of the e ect of di erent parameters p on a common variable y possible. Because the relative-relative sensitivity function 4.9d is non-dimensional, this sensitivity function can not only be used to compare the e ect of di erent parameters on a common variable, but also the e ects of di erent parameters on di erent variables. However, the disadvantage of this relative-relative comparison is that it gives not very useful results if the value of the variable y becomes small during a simulation because large relative changes of a value that is close to zero are not relevant. Fig. 4.7 illustrates the meaning of the absolute-relative sensitivity function 4.9c. The linear approximation to the change in y for a 100  change in p can easily be constructed with the aid of the tangent to the function yp in p. The larger the values of sensitivity functions and the more pronounced the di erences in shape of the sensitivity functions within the range of available data, the more accurately are the parameters identi able. Figure 4.8 shows an example of the time dependence of the sensitivity functions of

Figure 4.8: Example of a plot of sensitivity functions of a calculated concentration with respect to three model parameters. a calculated concentration with respect to three model parameters K , rmax and Cini look at the tutorial for a complete description of this example. It is evident that the parameter Cini is identi able from measured concentrations, because for small values of the time, t, the calculated concentration is only sensitive to this parameter. The sensitivity functions of the parameters K and rmax have a similar shape the signs indicate that the calculated concentrations increase with increasing values of K , but they decrease

146

CHAPTER 4. SIMULATION AND DATA ANALYSIS

with increasing values of rmax . This leads to a correlation between the estimates of these parameters changes in calculated concentrations caused by a change in K can approximately be compensated by an appropriate change in rmax . Furthermore, the much smaller sensitivity of the calculated concentration to the parameter K in comparison to the sensitivity to the parameter rmax leads to a larger uncertainty of the estimate of K than that of rmax . In uncertainty analysis, the uncertainty of model parameters is propagated to the uncertainty of model results model structure uncertainty is not addressed as an automated task of AQUASIM, however, predictions with di erent model structures can be compared manually. In the current version of AQUASIM only the simplest error propagation method is implemented, the linearized propagation of standard devitions of uncorrelated parameters. The error propagation formula using the linearized model and neglecting the parameter correlation is given by

v m  2 uX @y u y=t
i=1

@pi

2 pi

4.10

where pi are the uncertain model parameters, pi are their standard deviations, yp1 ; :::; pm  is the solution of the model equations for a given variable at a given location and time, and y is the approximate standard deviation of the model result. Figure 4.9 shows a

Figure 4.9: Example of a plot of simulation results with estimated error bounds. plot of measured data and the simulation result with error bounds limiting the range of values of the result plus and minus one standard deviation. The error contribution of each parameter is given as
err y;p

= @y p @p

4.11

4.2. SENSITIVITY ANALYSIS

147

The error contributions of the three parameters of the example shown in the Figs. 4.8 and 4.9 are shown in Fig. 4.10. It is interesting that the magnitudes of the error contributions

Figure 4.10: Example of a plot of error contributions of three model parameters on a calculated concentration. due to the parameters K and rmax are of a similar size despite the di erence in sensitivity shown in Fig. 4.8. The reason for this e ect is the higher uncertainty in the estimate of K resulting form the smaller sensitivity. The derivatives required for calculating the sensitivity functions 4.9a to 4.9d, the standard deviations of calculated variables according to equation 4.10, and the contributions of parameter uncertainties to the total uncertainty according to equation 4.11 are calculated using the nite di erence approximation @y  ypi +
pi , ypi  4.12 @pi
pi where
pi is chosen to be 1  of the standard deviation pi of the parameter pi . In order to make the program as exible as possible, the solutions for the basic parameter values p1 ; :::; pm  and for all sets of parameter values, where one of the pi 's is replaced by pi +
pi are calculated one after the other and stored simultaneously. Using these m + 1 stored states, all the expressions 4.9a to 4.9d, 4.10, and 4.11 can be evaluated easily for any variable y. Figure 4.11 shows the dialog box used for de ning and starting a sensitivity analysis. This dialog box is opened with the Sensitivity Analysis command in the Calc menu shown in Figure 4.1. The two upper two list boxes in this dialog box show the active and the available `Parameters' constant variables and real list variables, respectively. With the aid of the buttons `Activate' and `Inactivate', parameters selected in the right list box can be

148

CHAPTER 4. SIMULATION AND DATA ANALYSIS

Figure 4.11: Dialog box for editing a sensitivity analysis. activated and parameters selected in the left list box can be inactivated this is equivalent to toggle the check box active for sensitivity analysis in the dialog boxes shown in Figs. 3.7 and 3.9. The two lower list boxes show the active and the available `Calculations', respectively. These are the same calculation de nitions as are needed to execute simulations as described in section 4.1. The buttons `New', `Duplicate', `Edit', and `Delete', used for editing calculations, have the same functionality as the same buttons described for the dialog box shown in Fig. 4.4 in section 4.1. The calculations are edited as described in section 4.1 with the aid of the dialog box shown in Fig. 4.5. With the aid of the buttons `Activate' and `Inactivate', calculations selected in the right list box can be activated and calculations selected in the left list box can be inactivated this is equivalent to toggle the check box active for sensitivity analysis in the dialog box shown in Fig. 4.5. The button `Start' is used to start the sensitivity analysis, the button `Close' to close this modeless dialog box. After clicking the button Start the user is asked to specify the name of a le for a sensitivity ranking. If the user clicks Cancel in the le open dialog box, no sensitivity ranking is produced this may be advantageous to save computation time if the ranking is not required. During execution of the sensitivity analysis the dialog box shown in Fig. 4.12 is displayed. This dialog box shows the progress of the calculation and it allows the user to interrupt the sensitivity analysis. When all simulations have been performed and if a sensitivity ranking le has been speci ed, the dialog box shown

4.2. SENSITIVITY ANALYSIS

149

Figure 4.12: Dialog box for interrupting a sensitivity analysis during calculation. in Fig. 4.13 appears. Now the program calculates a ranking of the sensitivity functions as

Figure 4.13: Dialog box for interrupting a sensitivity ranking without loosing the other sensitivity analysis results. described below. This process can be interrupted loosing the ranking without loosing the calculated states. The sensitvity ranking le looks as follows:
*********************************************************************** AQUASIM Version 2.0 win batch - Sensitivity Analysis File ************************************************************************ Date and time of listing: 07 28 1997 09:01:42

Ranking of mean absolute sensitivities and error contributions: Calculation Number: 1 Compartment: Reactor Zone: Bulk Volume Variable: C Parameter: 1 2 3 4 5 Cini1 rmax1 K Cini2 rmax2

Sens AR: mg l 5.606 3.166 0.7936 0 0

Parameter: rmax1 K Cini1 Cini2 rmax2

Error Contr.: mg l 0.1093 0.1069 0.1032 0 0

150
Calculation Number: 2 Compartment: Reactor Zone: Bulk Volume Variable: C Parameter: 1 2 3 4 5 Cini2 rmax2 K rmax1 Cini1

CHAPTER 4. SIMULATION AND DATA ANALYSIS

It gives a ranking of the averages of the absolute values of the absolute-relative sensitivity functions 4.9c and of the error contributions 4.11 for all state variables in all zones of all compartments and for all calculations. For models with many parameters, this ranking is very useful to quickly nd out which are the most important parameters in uencing the state variables in the di erent compartments.

Sens AR: mg l 0.0769 0.0632 0.0530 0 0

Parameter: K rmax2 Cini2 rmax1 Cini1

Error Contr.: mg l 0.0071 0.0031 0.0030 0 0

4.3. PARAMETER ESTIMATION

151

4.3 Parameter Estimation

Model parameters represented by constant variables can be estimated by AQUASIM by minimizing the sum of the squares of the weighted deviations between measurements and calculated model results
2 p =

n X  ymeas;i , yip 2 i=1 meas;i

4.13

In this equation ymeas;i is the i-th measurement, meas;i is its standard deviation, yi p is the calculated value of the model variable corresponding to the i-th measurement and evaluated at the time and location of this measurement, p = p1 ; :::; pm  are the model parameters, and n is the number of data points. The measurements ymeas;i for i = 1; :::; n must be represented by real list variables with the argument either the program variable Time or the program variable corresponding to the space coordinate of the compartment in which the comparison takes place. The standard deviations meas;i can be de ned individually for each data point or globally for all data points of each real list variable in the dialog box used for editing real list variables shown in Fig. 3.9. The sum 4.13 extends over all data points of all real list variables speci ed as t targets as shown below. Simultaneous comparisons of data for measurements corresponding to di erent variables, compartments and zones are possible. AQUASIM performs a minimization of the sum of squares 4.13 with the constraints

pmin;i  pi  pmax;i 4.14 where pmin;i and pmax;i are the minimum and maximum of the constant variable representing pi . The values for pmin;i and pmax;i are speci ed in the dialog box used for editing
constant variables shown in Fig. 3.7. Due to the possible nonlinearity of the model equations and due to the numerical integration procedure, the sum 4.13 must be minimized numerically. The user has the choice between two numerical minimization algorithms: The simplex algorithm Nelder and Mead, 1965 and the secant algorithm Ralston and Jennrich, 1978. Both of these techniques are well-suited for the minimization of numerically integrated equations, because they avoid the calculation of derivatives of the solutions with respect to the parameters. The simplex technique starts with a set of m + 1 m is the number of parameters arrays of parameter values buildung a simplex with nonzero volume in parameter space in AQUASIM, this set consists of one array with the user-de ned start values and m arrays in which exactly one parameter value di ers from its start value. At each iteration step, the simplex of arrays of parameter values is replaced by a new simplex according to the following strategy Nelder and Mead, 1965 illustrated in Fig. 4.14 for the case of two parameters to be estimated. The parameter array leading to the highest value of 2 is rst re ected at the plane de ned by the other parameter arrays see Fig. 4.14 1; this is a rough guess for the downhill direction. If the value of 2 at this re ected array of parameter values is between the lowest and the highest value of the old simplex, this re ected array of parameter values is accepted and replaces the array leading to the highest value of 2 . If the value of 2 at the re ected parameter array is smaller than the lowest value on the old simplex, a new parameter array with an increased step size is calculated see Fig. 4.14 2 by an expansion of the simplex. If the value of 2 at this

152
(1 ) re fle c tio n

CHAPTER 4. SIMULATION AND DATA ANALYSIS

(2 ) re fle c tio n a n d e x p a n s io n

h ig h lo w

h ig h lo w

(3 ) c o n tra c tio n in o n e d im e n s io n

(4 ) c o n tra c tio n in a ll d im e n s io n s

h ig h lo w

h ig h lo w

Figure 4.14: Basic steps of the simplex algorithm for function minimization illustrated for the case of two parameters. The solid line simplex is replaced by one of the hashed simplexes. new array of parameter values is smaller than the lowest value on the old simplex it is accepted, otherwise the re ected array without expansion is accepted in both cases the new array replaces the array leading to the highest value of 2 of the old simplex. If the value of 2 at the re ected point is higher than the highest value at the old simplex, a new array of parameter values is constructed by contracting the point with the highest value of 2 of the old simplex in direction to the other points of the simplex see Fig. 4.14 3. If this leads to a value of 2 that is smaller than the highest value at the old simplex, this array is accepted, otherwise, all arrays of parameter values of the old simplex with exception of that leading to the lowest value of 2 are contracted in direction to the array leading to the lowest value see Fig. 4.14 4. This procedure is repeated until all values of 2 at the points of the simplex are close enough together in order to ful ll the convergence criterion. This unconstrained simplex technique is extended to constrained minimization by the coordinate transformation of the parameters  0 = tan  2pi , pmax;i , pmin;i 4.15 pi 2 pmax;i , pmin;i shown in Fig. 4.15. Application of the unconstrained technique to the parameters p0i is equivalent to constrained minimization of the parameters pi . If the unconstrained estimates of the parameters p0i are found the parameters pi can be calculated by the inverse formula 0 1 pi = 2 pmax;i + pmin;i  + pmax;i , pmin;i arctanpi 4.16  Similarly to the simplex technique described above, the secant technique starts with a set of m + 1 arrays of parameters. But now the secant technique uses the special form of the function 2 4.13 to be minimized. There is exactly one linear function in the parameters that passes through all function values yip for all m + 1 parameter sets. For this linear function, the expression 4.13 is a quadradic form with exactly one wellde ned minimum. The idea of the secant technique Ralston and Jennrich, 1978 is to

4.3. PARAMETER ESTIMATION

p i'

153

p i' p p p
i

m in ,i

m a x ,i

p
i

Figure 4.15: Scaled tan-function used to transform the bounded interval pmin;i; pmax;i of the parameter pi to the whole real axis for the parameter p0i. replace one of the m + 1 parameter arrays with the array leading to the minimum for the linear approximation of the model functions yi p at each iteration step. In order to take into account the constraints of the parameters, the secant technique was combined with the active set technique Gill et al., 1981 for the implementation in AQUASIM. In the active set technique the inequality constraints 4.14 are replaced by a number of active equality constraints for which some of the parameters are exactly on one of their boundary values. After each iteration step, this set of active constraints is checked for validity and changes to this set are made if this seems to be necessary. Because it looks for a downward direction in a very robust way, the simplex technique may be applied even to a poorly de ned parameter estimation process with starting values of the parameters far from those leading to the minimum of 4.13. In contrast, the secant method has more problems with bad starting values and poorly de ned minima of 2 , but it leads to much faster end convergence close to a well-de ned minimum. Estimates for the standard errors of the estimated parameters and for the parameter correlation matrix are only calculated by the secant method, and only if no parameter estimated is on one of the bounds pmin;i or pmax;i of the parameter. Figure 4.16 shows the dialog box used for de ning and starting a parameter estimation. This dialog box is opened with the Parameter Estimation command in the Calc menu shown in Figure 4.1. The two upper list boxes in this dialog box show the active and the available `Parameters' to be estimated constant variables, respectively. With the aid of the buttons `Activate' and `Inactivate', parameters selected in the right list box can be activated and parameters selected in the left list box can be inactivated this is equivalent to tog-

154

CHAPTER 4. SIMULATION AND DATA ANALYSIS

Figure 4.16: Dialog box for editing a parameter estimation. gle the check box active for parameter estimation in the dialog box used for editing constant variables shown in Fig. 3.7. The two lower list boxes show the active and the available `Calculations', respectively. The calculation de nitions for parameter estimation can be edited with the aid of the buttons `New', `Duplicate', `Edit' and `Delete' as discussed below. With the aid of the buttons `Activate' and `Inactivate', calculations selected in the right list box can be activated and calculations selected in the left list box can be inactivated this is equivalent to toggle the check box active for parameter estimation in the dialog box shown in Fig. 4.17. The radio buttons `secant' and `simplex' can be used to select the numerical minimization `Method' above. The basic ideas and references for both possible selections are given above. In the edit eld `Maximum Number of Iterations', the number of iterations can be bounded. A value of zero leads to the calculation of the value of 2 without trials to improve this value. The button `Start' is used to start the parameter estimation, the button `Close' to close this modeless dialog box. After clicking the button Start the user is asked to specify

4.3. PARAMETER ESTIMATION

155

the name of a le on which the parameter and 2 values for all iteration steps are reported and which contains a nal summary of parameter estimates, estimated standard errors and correlation coe cients for the secant method not leading to estimates on the bounds of the constraining interval only, and a list of the contributions of each data series to the total value of 2 see below for an example of such a le. As discussed above, the buttons New, Edit and Duplicate in the dialog box shown in Fig. 4.16 are used to de ne calculations for parameter estimation. Figure 4.17 shows the dialog box used for de ning and editing a single calculation. The items Name, Description,

Figure 4.17: Dialog box for editing a parameter estimation calculation.

Calc. Number, Initial Time

and Initial State have the same meaning as for calculations for simulation and sensitivity analysis in the dialog box shown in Fig. 4.5: The edit eld `Name' is used to specify the name of the calculation. Each calculation needs a unique name as an identi er. A name of a calculation consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of calculations the edit eld `Description' can optionally be used to store comments on the purpose of the calculation de nition. The edit eld `Calc. Number' allows the program user to identify a calculation by a nonnegative integer number. Variables and thus also processes, etc. can be made dependent on this calculation number with the aid of the program variable Calculation Number. Active calculations must have di erent values of their calculation numbers. Each calculation must be initialized before it can be started. In the edit eld `Initial Time' the time at which a calculation is initialized can be speci ed. The user has the choice between two types of the `Initial State' to be used by

156

CHAPTER 4. SIMULATION AND DATA ANALYSIS

the program. The default option `given, made consistent' applies the initial state as given by the user a value of zero is the default initial condition for all state variables for which the user did not explicitly specify an initial condition. If this initial state violates constraints on boundary conditions or algebraic equations of equilibrium processes, the program tries to ful ll these constraints by a minimal number of modi cations to the initial state given by the user. This procedure is called to make the initial state consistent. With the selection of the alternative option `steady state' the program tries to nd the steady state solution of the user-de ned model under external parameter values evaluated at the initial time and uses this steady state solution as the initial condition. Note, however, that not all models have a steady state solution and that even if such a solution exists, the numeric algorithm used by AQUASIM may fail to nd it. In such a situation, the convergence can be improved by a good choice of initial conditions which, under this option for the initial state, are used as starting values for the iterative search process for the steady state solution. If this does not help, the steady state solution must be found by relaxation, i.e. by executing a dynamic simulation with constant boundary conditions. After these ve items which have the same meaning as those discussed for simulation and sensitivity analysis Fig. 4.5, instead of step sizes and numbers of steps, the user must specify Fit Targets the step sizes are selected automatically by the program to meet the points in time at which data are available. All t targets already de ned are listed in the list box of the dialog box shown in Fig. 4.17. Each t target consists of a data series and a speci cation of which variable evaluated in which zone of which compartment and at which point in time or space should be compared with the data. The list of t targets is edited with the buttons `Add', `Edit' and `Delete' below the list box. Finally, the calculation can be made `active for parameter estimation', if no other calculation with the same calculation number is already active. The Fig. 4.18 shows the dialog box used to specify a t target. This dialog box is

Figure 4.18: Dialog box for editing a t target. opened by clicking one of the buttons Add or Edit of the dialog box shown in Fig. 4.17. In the rst eld `Data' of this dialog box a data series can be selected. This data series is represented by a real list variable with the argument either the program variable Time or the program variable corresponding to the space coordinate of the compartment in which the comparison with the calculated variable takes place.

4.3. PARAMETER ESTIMATION

157

The next four elds of the dialog box shown in Fig. 4.18 are used to de ne which calculated value of the model should be compared with the data series speci ed in the rst eld. In the eld `Variable', a variable must be selected. In many cases measured data is compared with values of a state variable of the model, but in AQUASIM, any variable can be used. E.g. a formula variable can be used for conversion of units, for building combinations of state variables e.g. a sum of state variables that is measured, etc.. In the eld `Compartment' the compartment must be selected in which the variable to be compared with the data has to be evaluated. This eld is inactive if there exists only one compartment. In the eld `Zone' the zone of the compartment must be selected in which the variable to be compared with the data has to be evaluated. This eld is inactive if there exists only one zone in the compartment selected above. In the last eld Time Space the time must be speci ed at which the comparison of the spatial pro le data with the calculated variable, or the spatial location at which the comparison with the time series data must take place. Note that the value entered here is interpreted as a spatial location if the argument of the real list variable selected in the eld Data is the program variable Time, and as a point of time if the argument of the real list variable selected in the eld Data is the program variable corresponding to the spatial location of the compartment selected in the eld Compartment. If the check box `relative space' is ticked, spatial locations must be given in relative coordinates, otherwise in absolute coordinates this check box has no in uence for time speci cations which are always absolute. In the case of a comparison with time series data, the value entered in the eld Time Space must be within the legal coordinate range of the compartment. For absolute spatial locations, the legal range is de ned in the de nition of the compartment selected in the eld Compartment. Relative locations must be between 0 and 1. If the compartment is a mixed reactor compartment which has no spatial coordinate, the value entered in the eld Time Space is ignored. During execution of the parameter estimation the dialog box shown in Fig. 4.19 is displayed. This dialog box shows the initial value of 2 , the best smallest value already

Figure 4.19: Dialog box for interrupting a parameter estimation. found, and the value for the current parameter set. This dialog box allows the user to interrupt the calculation.

158

CHAPTER 4. SIMULATION AND DATA ANALYSIS

In case of normal termination of the parameter estimation algorithm, the dialog box shown in Fig. 4.20 is displayed to give the user a brief survey of the t results. The rst

Figure 4.20: Dialog box for displaying a summary of t results. row of this dialog box shows the termination status of the algorithms, the second row the number of iterations performed. In the following list box for each parameter that was active for parameter estimation, its initial value, its nal value found by the parameter estimation algorithm, and its minimum and maximum is given. Finally, below the list box, the initial and the nal value of 2 are given in order to give the user a quick overview on the improvement that could be achieved by the parameter estimation procedure. More details on the performance of the parameter estimation algorithm is given on the le speci ed after clicking the button Start in the dialog box shown in Fig. 4.16. This le looks as follows:
*********************************************************************** AQUASIM Version 2.0 win mfc - Parameter Estimation File ************************************************************************ Date and time of listing: Number of parameters Number of data points Estimation method Parameters: Name Cini1 Cini2 K rmax1 rmax2 Unit mg l mg l mg l mg l h mg l h Start 12 0.8 2 2 2 Minimum 0 0 0 0 0 Maximum 20 2 10 10 10 = = = 07 29 1997 13:56:09 5 22 secant

4.3. PARAMETER ESTIMATION

Calculations: Cini1 mg l 12 12.2 12 12 12 12 11.9411 11.8094 Cini2 mg l 0.8 0.8 0.82 0.8 0.8 0.8 0.80557 0.82432 K mg l 2 2 2 2.1 2 2 2.3 2.9 rmax1 mg l h 2 2 2 2 2.1 2 2.01325 2.06794 rmax2 mg l h 2 2 2 2 2 2.1 2.01687 2.13019 Chi^ 2

159

. .

10.1604 10.1531

. .

1.00113 1.00421

. .

1.04662 1.03489

. .

1.04293 1.03858

. .

1749.41 1700.74 1725.51 1656.36 1928.34 1795.93 1516.94 1191.95

0.51529 0.51681

. .

22.6634 22.5086

Parameter estimation successfully finished convergence criterion met Cini1 mg l Cini2 mg l K mg l rmax1 mg l h rmax2 mg l h

Estimated values of the parameters: 10.1531 1.00421 1.03489 1.03858 0.51681

Estimated standard errors of the parameters: 0.22133 0.03057 0.27284 0.07084 0.09739

Estimated correlation matrix of the parameters: 1 0.17923 0.53889 0.77737 0.52576 0.17923 1 0.34117 0.30991 0.46943 0.53889 0.34116 1 0.92990 0.98009 0.77737 0.30991 0.92990 1 0.90166 0.52576 0.46943 0.98009 0.901655 1

Contribution of data series to Chi^ 2 Calculation: Fit1 Fit2 Data Series: Cmeas1 Cmeas2 Chi^ 2 ini: 801.34 948.074 ------1749.41 = = 27 39 Chi^ 2 end: 12.1706 10.338 ------22.5086

Number of steps performed Number of simulations performed

After the header, on this le the number of parameters and the number of data points active for the current parameter estimation and the method selected for the numerical min-

160

CHAPTER 4. SIMULATION AND DATA ANALYSIS

imization algorithm are indicated. Then the le contains a listing of all active parameters with unit, start value, minimum and maximum. Then a listing of the parameter values and of the values of 2 for all calculations performed during the parameter estimation process is given. After this listing, the parameter estimates are given. If the estimation algorithm was the secant method and no parameter estimate was on the border of its legal range, an estimate of the standard error and of the parameter correlation matrix is also given. Then, in addition to the value of 2 which represents the sum of the deviations for all t targets, the contributions of all t targets to the nal value of 2 are given individually. Finally, the number of iterative steps and the number of simulations performed, are given.

Chapter 5

Visualization of Results
Fig. 5.1 shows the menu View of AQUASIM. The item `Results' of this menu is used

Figure 5.1:

View

menu.

to specify plot de nitions, to plot results to the screen or in PostScript or Encapsulated PostScript format to a le, and to list results to a text le for external postprocessing. The item `Toolbar' is used to activate or hide the AQUASIM toolbar, and the item `Status Bar' is used to activate or hide the status bar of AQUASIM the status bar is not used in the current version of AQUASIM. Figure 5.2 shows the dialog box used for editing plot de nitions and for plotting and listing results. This dialog box is opened with the command Results of the menu View shown in Fig. 5.1. The list box in this dialog box contains a list of plot de nitions. Each of these plot de nitions contains general plot speci cations such as meaning, labels and scaling of axes and a list of curves to be drawn in the plot. Each of these curves contains speci cations such as the type of the curve value, sensitivity function or error contribution of a variable, the variable, the calculation number, the compartment, the 161

162

CHAPTER 5. VISUALIZATION OF RESULTS

Figure 5.2: Dialog box for editing plot de nitions and plotting and listing results. zone, the spatial location if the abscissa of the plot is time or the time if the abscissa of the plot is the spatial coordinate of the compartment at which the curve should be drawn, and the signature of the curve. All these de nitions can be speci ed independently of the existence of the calculated states necessary for actually drawing the curves. The recommended strategy is therefore to spend some time to carefully specify a series of plot de nitions which can later on be used very e ciently to present an overview of the current simulation results. The buttons at the right of the list box of plot de nitions allow the user to perform the following operations: By clicking the button `New', new plot de nitions can be speci ed. Alternatively, by clicking the button `Duplicate', the selected plot de nition can be duplicated. With the button `Edit', or by double-clicking the plot de nition name in the list box, a plot de nition can be edited. The button `Delete' allows the program user to delete plot de nitions. The buttons Duplicate, Edit and Delete are inactive as long as no plot de nition is selected. The dialog boxes used for editing plot de nitions are discussed later in this section Figs. 5.3 to 5.5. The three buttons `Plot to Screen', `Plot to File', and `List to File' are used to draw a plot corresponding to the currently selected plot de nition to the screen, to write it in PostScript or Encapsulated PostScript format to a le, and to write it in text format to a le, respectively. The PostScript or Encapsulated PostScript le must be handled by the user. Typically, a PostScript le is sent to a printer with the aid of an appropriate program available as a shareware program, an Encapsulated PostScript le is included as a gure into a text processing program. All the three actions described above are only possible, if calculated data for at least one curve of the currently selected plot de nition exists and if no curve contains unde ned values as they can occur e.g. by division by zero or by taking logarithms of negative numbers in the variable to be plotted. The three buttons `Scr. Opt.', `File Opt.', and `List Opt.' are used to specify

163 general options for plotting to the screen, for plotting to a PostScript le, and for listing plot data to a text le, respectively. The dialog boxes used for specifying these options are discussed later in this section Figs. 5.6 to 5.8. Figure 5.3 shows the dialog box used for de ning or editing a plot de nition. This

Figure 5.3: Dialog box for editing a plot de nition. dialog box is opened by clicking one of the buttons New, Edit or Duplicate of the dialog box shown in Fig. 5.2. Each plot de nition needs a unique `Name' as an identi er. A name of a plot de nition consists of a sequence of letters A-Z,a-z, digits 0-9 and underline characters  . The rst character may not be a digit. To improve documentation of plot de nitions, a `Description' can be given optionally. The edit eld `Title' is used to specify the title of the plot. The radio buttons `Abscissa' can be used to select the meaning of the abscissa. The user has the choice between the abscissa `Time' which is used for plotting time series at given locations in the system described with AQUASIM and the absissa `Space' used for plotting spatial pro les at given points of time this last option is not meaningful for mixed reactors that have no spatial variation in state variables. The edit elds `Label' can be used to specify a label for the Abscissa and for the Ordinate, respectively. A plot de nition contains an arbitrary number of de nitions of `Curves'. All curve de nitions are listed in the list box of the dialog box shown in Fig. 5.3. The buttons `Add', `Edit' and `Delete' below this list box are used to edit the list of curve de nitions. If no

164

CHAPTER 5. VISUALIZATION OF RESULTS

curve de nition is selected, clicking the button Add leads to the insertion of a new curve de nition at the end of the list of curve de nitions. If a curve de nition is selected, the new curve de nition is inserted before the selected curve de nition. Finally, the button `Scaling' is used to choose options for bounds and ticks of the axes of the plot. Figure 5.4 shows the dialog box used for de ning or editing a curve de nition within a plot. This dialog box is opened by clicking one of the buttons Add or Edit of the dialog

Figure 5.4: Dialog box for editing the de nition of a curve within a plot. box shown in Fig. 5.3. The radio buttons `Type' are used to select the type of the curve. The user has the choice between the following options: The radio button `Value' is selected in order to plot the value of a variable. The value of all variables can be plotted if calculated states corresponding to the speci cations given below are available. If not only a simulation, but a sensitivity analysis has been performed, the curve for the value of the variable is supplemented by thin lines indicating the value plus and minus one standard deviation calculated according to the linear error propagation formula which ignores parameter correlations 4.10. Usually, if the abscissa of the plot is Time, the curve for the value of a variable is plotted as linear interpolation between values evaluated at each output time step speci ed in the dialog box shown in Fig. 4.5 for plots drawn after a simulation or a

165 sensitivity analysis or by the time steps of the data speci ed as t targets in the dialog box shown in Fig. 4.18 for plots drawn after a parameter estimation. If the abscissa of the plot is Space, similarly the grid points of the spatial resolution of the compartment are used. There is one important exception from this general rule: If the abscissa of the plot is Time and the variable speci ed in the curve de nition is a real list variable with the program variable Time as the argument, or if the abscissa of the plot is Space and the variable specifed in the curve de nition it is a real list variable with the program variable corresponding to the spatial extent of the compartment as the argument, the data pairs of the real list are used instead of the interpolated values at the simulation times or grid points. This makes it possible to plot measured data without interpolation see alsow Figs. 5.9 and 5.10. Selection of one of the radio buttons `Error Contrib.' and `Sens. Function' leads only to a curve if a sensitivity analysis corresponding to the speci cations below has been performed. Selection of the radio button Error Contrib. leads to the plot of the error contribution of a parameter p to the total error of a variable y given by equation 4.11. Selection of the radio button Sensitivity Function leads to the plot of a sensitivity function. Selection of the radio button `AbsAbs' leads to the plot of the absolute-absolute sensitivity function given by equation 4.9a, selection of `RelAbs' to the plot of the relative-absolute sensitivity function given by equation 4.9b, selection of `AbsRel' to the plot of the absolute-relative sensitivity function given by equation 4.9c, and selection of `RelRel' to the plot of the relative-relative sensitivity function given by equation 4.9d.

Figure 5.5: Dialog box for editing scaling of plot axes. As a next item in the dialog box used for the speci cation of a curve de nition shown in Fig. 5.4, a `Variable' must be selected. If the type of the curve is Error Contrib. or Sensitivity Function, a constant variable or a real list variable must be selected as a `Parameter'. The curve can then only be plotted, if a sensitivity analysis with this Parameter as an active parameter has been performed. This eld is inactive, if the curve is of the type Value. The edit eld `Calculation Number' is used to select a calculation identi ed by its calculation number speci ed in one of the dialog boxes shown in Figs. 4.5 and 4.17.

166

CHAPTER 5. VISUALIZATION OF RESULTS

Because calculations for di erent calculation numbers can be stored in memory simultaneously and because variables and process rates can be made depend on the calculation number with the aid of the program variable Calculation Number, curves for the same variable at the same location under di erent process hypotheses can be compared in the same plot. As a next speci cation, a `Compartment' within which the curve must be evaluated can be seleced. This eld is inactive if only one compartment is de ned. If the compartment selected above contains more than one zone, a `Zone' must be selected. This eld is inactive if the compartment contains only one zone.

Figure 5.6: Dialog box for editing plot to screen options. Finally, the edit eld `Time Space' is used to complete the speci cation of when and where to evaluate the curve. If the abscissa of the plot is Time, a spatial location valid for the compartment selected above must be given here. If the check box `rel. space' is ticked, relative coordinates between 0 and 1 must be given, otherwise absolute spatial locations must be speci ed. The values of the variable of function to be plotted are then linearly interpolated between the values at the neighbouring grid points of the compartment. In the case of a mixed reactor compartment, the value of this edid eld is ignored, because there is no spatial resolution. If the abscissa of the plot is Space the edit eld Time Space is used to specify the time at which the curve is evaluated. In this case the check box rel. space is ignored points in time must always be given as absolute values. The values are linearly interpolated between the neighbouring output time steps speci ed in one of the dialog boxes shown in Figs. 4.5 and 4.17. In the edit eld `Legend' the user can specify a legend entry for the curve. The appearance of the curve can be speci ed by `Line' and `Marker' attributes. The check boxes `active' are used to turn on and o the lines between data pairs and the markers at the data pairs, respectively. As described above, the data pairs of plots with the abscissa Time are usually determined by the output time steps of the calculation, the data pairs of plots with the abscissa Space by the grid points of the compartment. If the value of a real list variable with the program variable corresponding to the space coordinate of the compartment as its argument is plotted in a plot with the abscissa Time, or if the value of a real list variable with the program variable Time as its argument is plotted in a plot with the abscissa Space, instead, the data pairs of the real list variable

167

Figure 5.7: Dialog box for editing plot to le options. are used. In order to plot the actually used interpolated values that may be di erent from linear interpolation according to the selection of the Interpolation Method in the dialog box shown in Fig. 3.9, a formula variable with the name of the real list variable as its algebraic expression may be de ned and speci ed as the variable to be plotted this is in fact the way the data for the plot shown in Fig. 3.8 has been generated. Several line attributes, grouped as `Style', `Width' and `Color' attributes, can be used to obtain a clear distinction between plotted curves. Similarly, for markers, `Style', `Size' and `Color' attributes are available. Figure 5.5 shows the dialog box used to specify the scaling of the plot. This dialog box is opened by clicking the button Scaling of the dialog box shown in Fig. 5.3. For both, `Abscissa' and `Ordinate', a `Minimum', a `Maximum', a `Tick Position' and a `Tick Distance' can either be speci ed or selected to be automatically determined. The possibility of specifying a tick position makes the positions of the ticks independent of the bounds of the axis. Figure 5.6 shows the dialog box used to edit the options for plotting to the screen. This dialog box is opened by clicking the button Scr. Opt. of the dialog box shown in Fig. 5.3. It allows the program users to specify the `Width' and the `Height' of the plot window, the width of the `Margins' and the `Legend Width'. In addition, the user can select the `Font Size' for the plot `Title', for the axes `Labels', for the `Legend' and for the tick `Numbers'. All measurement units are pixels. Figure 5.7 shows the dialog box used to edit the options for plotting in PostScript

168

CHAPTER 5. VISUALIZATION OF RESULTS

Figure 5.8: Dialog box for editing list to le options. or Encapsulated PostScript format to a le. This dialog box is opened by clicking the button File Opt. of the dialog box shown in Fig. 5.3. It allows the program user to select the `Paper Size' and the `Orientation' and to specify the number of `Columns' and `Rows' of plots on the sheet, the `Margins', the `Legend Width' and the `Font Size' for the `Title', the axes `Labels', the `Legend' and the tick `Numbers'. Furthermore, the user can select to plot in `color' or `black white', to use `cm' or `inch' as measurement units for distances and to select `PostScript' or `Encapsulated PostScript' as the le format PostScript is suited for printing, Encapsulated Postscript for including AQUASIM plots in other documents.

Figure 5.9: Example of a plot window on the screen. Figure 5.8 shows the dialog box used to edit the options for listing calculated results corresponding to a plot de nition to a text le. This dialog box is opened by clicking the

169 button List Opt. of the dialog box shown in Fig. 5.3. The user can select the number of `Signi cant Digits' and the Separator between numbers on the output le. Figure 5.9 shows a plot window on the screen. This plot window is generated by clicking the button Plot to Screen in the dialog box shown in Fig. 5.2. It contains four curves. Two of them, represented by markers, are real list variables with the program variable Time as the argument. In this case, the data pairs of the real list variables are plotted. The other two curves, represented by a dashed and a solid line, respectively, are simulation results and are plotted with a high temporal resolution. Figure 5.10 shows the same plot written in Encapsulated PostScript format to a le and imported to this document. This le is obtained by selecting the option Encapsulated PostScript in the dialog box shown in Fig. 5.7 and then clicking the button Plot to Screen in the dialog box shown in Fig. 5.2. Finally, the following listing shows the plot of two of the curves shown above as obtained by clicking the button List to File in the dialog box shown in Fig. 5.2.
Production
0.6 0.5 Prod [mgO/l/h] 0.4 0.3 0.2 0.1 0 0 1e+003 2e+003 time [s] 3e+003 90uE/m2/s 1300uE/m2/s 90uE/m2/s 1300uE/m2/s

Figure 5.10: Example of an Encapsulated PostScript plot.

170

CHAPTER 5. VISUALIZATION OF RESULTS

************************************************************************ AQUASIM Version 2.0 win mfc - List File ************************************************************************ Date and time of listing: Name of plot: Val Prod Type: Variable: Parameter: CalcNum: Compart.: Zone: Time Space: Unit: Legend: Argument t 05 05 1998 11:47:00 Value Prod1 Argument t Value Prod

0 0

Reactor Bulk Volume s mgO l h 90uE m2 s s

1 0

Reactor Bulk Volume mgO l h 90uE m2 s

150 300 600 900 1200 1500 1800 2100

0.095 0.16 0.16 0.14 0.13 0.13 0.13 0.13

0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 . .

0 0.01492 0.02858 0.04106 0.05248 0.06291 0.07246 0.08118 0.08915 0.09644 0.1031 0.1092 0.1147 0.1198 0.1244 0.1286 0.1325 0.136 0.1392 0.1421 0.1447 0.1471 0.1493 0.1513 0.1531 0.1548 . .

Chapter 6

Appendix
6.1 Character Interface Version
Besides the window interface version of AQUASIM described in the other sections of this manual, there exists a character interface version with the same functionality exept that it is not able to draw plots to the screen. This version is designed for editing AQUASIM system les and for executing simulations, sensitivity analyses and parameter estimations via simple character-oriented terminal connections and for automating editing tasks with the aid of scripts of the operating system. For normal usage of the program, the window interface version is much more attractive. Because the menu and dialog structure of the character interface version is exactly the same as that of the window interface version, the other chapters of this manual can be used as a manual for the character interface version also. In the following, some speci c hints on the usage of the character interface version are given and a simple example demonstrates how to start and exit the character interface version. In the character interface version, user input is prompted with a sign ` ' and each user input must be followed by pressing `Return'. Menu items have to be selected by typing the number preceeding the item and pressing `Return'. Within menus and submenus, `Return' without giving an input before leads back to the previous menu level. Default values proposed in square brackets can be accepted by pressing `Return' without giving a new input before. The following sequence shows how to start and exit the character interface version of AQUASIM the two menus correspond to the menu bar of the window shown in Fig. 1.1 and to the menu File shown in Fig. 2.1:

171

172
aquasimc *********************************************** * * AQUASIM - Simulation of Aquatic Systems * * * *

CHAPTER 6. APPENDIX

*********************************************** AQUASIM - Main Menu ------------------1 2 3 4 1 AQUASIM - File Menu ------------------1 2 3 4 5 6 7 8 E B e End of program AQUASIM normal termination = = = = = = = = = = New Open Close Save Save As Revert to Saved Print About Exit Back = = = = File Edit Calc View

6.2. BATCH VERSION

173

6.2 Batch Version

The batch version of AQUASIM is designed for executing demanding calculations on a compute server as batch jobs. In contrast to the window and character interface versions of AQUASIM, the batch version does not provide full functionality. In particular, it is not possible to de ne or edit models. The idea of the batch version is to perform one of the following tasks for a model de ned with the aid of an interactive program version and stored on an AQUASIM system le: Perform a simulation. Perform a sensitivity analysis. Perform a parameter estimation. Calculate values of 2 . Calculate results for given parameter sets and given variables. Plot results to a PostScript or Encapsulated PostScript le. List results to a text le. On systems for which command line arguments can be speci ed, these jobs can be started with the following command lines
aquasimb jobfile aquasimb -s logfile loadfile savefile aquasimb -a logfile loadfile savefile scmdfile sensfile

aquasimb -e logfile loadfile savefile fitfile aquasimb -c logfile loadfile vcmdfile chifile aquasimb -r logfile loadfile vcmdfile rcmdfile resfile aquasimb -p logfile loadfile pcmdfile aquasimb -l logfile loadfile lcmdfile

The parameters used in these command lines have the following meaning: jobfile AQUASIM job le. This le can contain one or more command lines in the style of the other command lines. These command lines are executed one after the other so that a job speci ed in a command line can use the save le of one of the previous jobs as its load le. When used within such a job le, the program name aquasimb may be omitted and the line can directly start with -s, -a, -e, -c, -p or -l. Within a job le a sign terminates input for the current line and makes it possible to add comments. -s Perform a simulation. -a Perform a sensitivity analysis. -e Perform a parameter estimation. -c Perform a calculation of 2 -values. -r Calculate results for given parameter sets and given variables. -p Plot results to a PostScript le.

174
-l logfile

CHAPTER 6. APPENDIX
List results to a text le. Name of an output text le for job log information. The information written to this le may be useful in case of problems during calculation CAUTION: an existing le with this name is overwritten without warning!. Name of the input AQUASIM system le containing the model de nitions. This le must be created with one of the interactive program versions. Name of an output AQUASIM system le on which the results are stored. This le can be opened with an interactive program version or from a succeeding batch job for processing results CAUTION: an existing le with this name is overwritten without warning!. Name of an optional input text le simulation command le containing the instructions for initialization and step sizes of simulations. If no such le is provided, the active calculations speci ed in the simulation dialog box are executed. If a simulation command le is provided, for each calculation number for which simulations should be performed, a line providing the two numbers
calcnum inittime calcnum = calculation

loadfile savefile

scmdfile

number, inittime = initial time must be provided. This line is followed by an arbitrary number of lines of the form
calcnum timestep numsteps calcnum = calculation number, timestep = output step integration, numstes = numper of steps to be performed. sensfile

size for time

fitfile

vcmdfile

chifile

rcmdfile

Name of an optional output text le to which a ranking of mean absolute sensitivity and error functions is written CAUTION: an existing le with this name is overwritten without warning!. Name of an output text le to which detailed results of the parameter estimation are written CAUTION: an existing le with this name is overwritten without warning!. Name of an input text le parameter value le specifying the names and values of parameters. The rst line of this le must contain the names of the parameters constant variables, on succeeding lines values of these parameters must be speci ed, each line containing one set of values for all parameters the value of each parameter must be within the legal range of the parameter bounded by its minimum and maximum. Name of an output text le to which the results of the calculations of 2 -values are written. After a header, each line of this le contains the parameter values and the corresponding value of 2 CAUTION: an existing le with this name is overwritten without warning!. Name of an input text le result de nition le de ning when and where to calculate results. This le must contain lines of the following form:

var calcnum comp zoneind time space relabs where var is the name of a variable, calcnum is the calculation number of an active calculations, comp is the name of a compartment, zoneind is an allowed zone index for this compartment, time is a point of time within

6.2. BATCH VERSION

175

resfile

pcmdfile

the simulation time range, space is a location within the compartment, and relabs is equal to the letter `a' for absolute space coordinates or equal to the letter `r' for relative space coordinates. The space coordinate must also be speci ed for mixed reactor compartments, however, in this case its value is ignored. Name of an output text le to which the results of the calculations are written. After a header, each line of this le contains the parameter values and the corresponding values for all result de nitions speci ed in the result de nition le CAUTION: an existing le with this name is overwritten without warning!. Name of an input text le plot command le specifying the names of the plot de nitions that are used to generate plots and the le names to which the plots are written in PostScript format. A plot command le contains an arbitrary number of lines of the form where plot is the name of a plot de nition and psfile is the name of a PostScript le to which the plot is appended. Name of an input text le list command le specifying the names of the plot de nitions that are used to generate listings of results and the le names to which these results are written in ascii format. A list command le contains an arbitrary number of lines of the form
plot lisfile plot psfile

lcmdfile

where plot is the name of a plot de nition and lisfile is the name of a text le to which the results are appended. If the batch version of AQUASIM is started without command line arguments e.g. on systems that do not support this feature the program looks for a le aquasim.job in the current directory. If it nds such a le, this le is interpreted as an AQUASIM job le, if no le with this name exist, the programs terminates with an error message. As an example, an AQUASIM job le containing the lines
-e -e -e -s -p fit1.log test.aqu test res.aqu test1.fit fit2.log test res.aqu test res.aqu test2.fit fit3.log test res.aqu test res.aqu test3.fit sim.log test res.aqu test res.aqu sim.cmd plot.log test res.aqu plot.cmd

executes a parameter estimation based on the AQUASIM system le test.aqu, restarts this estimation twice, then for the tted parameter values starts a simulation according to the speci cation given in the simulation command le sim.cmd e.g. for improving the plot resolution and plots the results according to the speci cations given in the plot command le plot.cmd to PostScript les.

176

CHAPTER 6. APPENDIX

6.3 Troubleshooting
AQUASIM was developed in the computer and systems sciences department of EAWAG primarily for internal use. Since EAWAG is an institution for research and teaching and not a software company, it is not possible for us to provide end-user support. Instead, the following possibilities are o ered: In this chapter hints are given for self-diagnosis and solution of problems section 6.3.1. A user group is maintained as an e-mail list through which questions can be sent to other AQUASIM users. In this way, AQUASIM users can nd help within the user community of AQUASIM section 6.3.2. Bug reports and suggestions for program improvements are welcome at any time. Although there is no technical support on program use, we try to x all bugs and provide bug- xed program versions to licensed program users section 6.3.3. Courses on AQUASIM use are given occasionally at EAWAG. A topical overview of AQUASIM-related information, such as on the most recent program version, on references of program applications, on planned AQUASIM courses, and on known bugs or problems can be found on the EAWAG home page at http: www.eawag.ch.

6.3. TROUBLESHOOTING

177

6.3.1 Problem Self-Diagnosis

In this subsection, hints are given for how to nd the cause of problems in program use. In the rst subsection, problems of loading AQUASIM system les are discussed, in the second subsection problems of editing models, and in the third subsection problems during calculations.

Problems of Loading Files

Format of AQUASIM System Files Before discussing possible problems of loading les, in this paragraph some explanations to the le format used by AQUASIM are given. In order to guarantee compatibility between di erent operating systems, AQUASIM systems are stored as ASCII text les. Line breaks guarantee compatibility with le editors and with e-mail programs. During the loading process all special characters ASCII code below 32, e.g. line break and carriage return characters are stripped from the le and the le is converted to a simple character stream this is important because line breaks are represented di erently on di erent operating systems. AQUASIM items are structured with the aid of curly braces f and g. Text between two successive opening or closing braces is ignored, only text between an opening and a closing brace of the same level without any braces in between is interpreted by AQUASIM. A program convedit convert AQUASIM system le for editing with a text editor is provided with AQUASIM in the subdirectory bin which converts the originally more compact le format to a format that is much easier to read line breaks and indents are inserted at each opening brace and between closing braces of di erent level and is equivalent for AQUASIM there is no tool to reverse this process, because this is not necessary since AQUASIM can read the converted le; the original format can easily be restored by loading and saving the le with AQUASIM. This tool may be used to allow advanced AQUASIM users to analyse causes for le loading problems or even to edit an AQUASIM system le directly with a text editor or a programming language e.g. with Perl scripts. This last possibility may be e cient for the automation of repeated complicated editing processes. However, it is also dangerous, because with the aid of a text editor or a scripting language minor errors can make a le unreadable by AQUASIM. Problems During the Loading Process The most important cause for problems in loading les is that a le from one of the very old versions AQUASIM 1.0x or from a newer AQUASIM version than that used for loading is tried to be loaded. Files from AQUASIM 1.0x must be converted to the le format for AQUASIM 1.1 with the aid of the conversion program conv1011 which is delivered with AQUASIM in the subdirectory bin. Files from AQUASIM 1.1x and from pre-releases of AQUASIM 2.0 can then be read by AQUASIM 2.0, however, it is not possible to save a system in an older le format with AQUASIM 2.0. Files from newer versions than AQUASIM 2.0 cannot be loaded with AQUASIM 2.0. AQUASIM system les are compatible between all operating systems supported by AQUASIM. One problem may occur on the Mac if an AQUASIM system le is transferred to the Mac from another operating system. If the le type is not AQU , the le is not displayed in the le open dialog box of AQUASIM. For this reason, after the transfer to the Mac, the le type must be set to this value AQUASIM system les created on the Mac automatically obtain this le type. This can be done in various ways:

178

CHAPTER 6. APPENDIX
The le is moved on the application AQUASIM Typer which is delivered with the Mac distribution of AQUASIM. The le is loaded with the character interface version of the Mac and saved by this program version. The le type is changed to AQU with another tool, e.g. with the Mac resource editor.

After a parameter estimation, the estimated standard deviations may become undened due to inversion of an ill-de ned matrix. On some operating systems especially on the Mac, we did not succeed in nding a function to check for such unde ned values. If on these systems such an unde ned value is saved on the AQUASIM system le, AQUASIM is not able to read this value during loading. Because the save le format of AQUASIM is a text le, such an unde ned value can be found with a text editor and can be corrected see comments on the le format above. The string indicating an unde ned value may depend on the operating system e.g. NaN.

6.3. TROUBLESHOOTING

179

Problems of Editing a Model

Errors of two major types can occur during model editing: Errors assigned to an edit eld of the active dialog box and more general errors of incompatibility of the newly edited structure with the rest of the system de nitions. Errors assigned to an edit eld can usually easily be identi ed. Figure 6.1 shows an example of such an error. The error message indicates an error in the edit eld Expression.

Figure 6.1: Example of an error assigned to an edit eld of the dialog box. If this error message occurs it is recommended to rst check the existence of all variables used in the edit eld and to check the correctness of the algebraic expression. In the present case Fig. 6.1, as is seen in the dialog box Edit Variables in the background, the variable x used in the algebraic expression of the formula variable does not exist. If syntax and existence checks do not help uncovering the problem, it is recommended to carefully read the explanating text to the referenced edit eld in the user manual. Besides errors in algebraic equations, illegal dependencies may occur. Examples of illegal dependencies are use of a program variable that is not meaningful in the context of the current compartment or the absence of obligatory dependencies. Inconsistency errors of the newly de ned object with existing system de nitions are more di cult to nd. In many cases, as shown in the example shown in Fig. 6.2, the name of the object may already exist. Also in this case, it is recommended to carefully read the comments to the dialog box in the user manual.

180

CHAPTER 6. APPENDIX

Figure 6.2: Example of an error assigned to an edit eld of the dialog box.

6.3. TROUBLESHOOTING

181

Problems During Calculation

The Log File of AQUASIM

The most important source of information on problems during calculation is the log le of AQUASIM. For each interactive section, a log le aquasim.log is written to the program directory of AQUASIM the log le from the previous session is overwritten. As explained in section 6.2 for batch jobs, the user can specify the name of the log le in the command line or in the job le. The contents of the log le can usually be ignored. However, it may contain useful information in case of problems during calculations. The following listing shows a typical log le written during an interactive simulation of the activated sludge example delivered with AQUASIM and described in section 8.2 of the report on the concepts of AQUASIM Reichert, 1994b.
************************************************************************ AQUASIM Version 2.0 win mfc - Log File ************************************************************************ Start of Session: 05 12 1998 20:58:43

Calculation of consistent initial condition ------------------------------------------Number of codiagonals considered: Time of initial condition: 05 12 1998 22:03:04 Start of calculation 05 12 1998 22:03:04 End of calculation Number of equations: Number of integers needed for calculation: Number of reals needed for calculation: Number of steps taken: Number of evaluations of the jacobian: Dynamic calculation ------------------Number of codiagonals considered: Maximum internal step size: 05 05 05 05 05 05 05 12 12 12 12 12 12 12 1998 1998 1998 1998 1998 1998 1998 22:03:04 22:03:04 22:03:07 22:03:08 22:03:08 22:03:08 22:03:08 Start of calculation Integration at time -50 Integration at time -49.98 Integration at time -49.96 Integration at time -49.94 Integration at time -49.92 Integration at time -49.9

1000 -50

50 50 5400 0 0

1000 1

. .

182
05 05 05 05 05 05 05 12 12 12 12 12 12 12 1998 1998 1998 1998 1998 1998 1998 22:03:13 22:03:13 22:03:13 22:03:13 22:03:13 22:03:13 22:03:13 Integration at time Integration at time Integration at time Integration at time Integration at time Integration at time End of calculation 0.9 0.92 0.94 0.96 0.98 1

CHAPTER 6. APPENDIX

Number of equations: Number of integers needed for calculation: Number of reals needed for calculation: Current integration time: Current step size: Current order of integration: Number of steps taken: Number of function evaluations: Number of evaluations of the jacobian: Number of error test failures: Number of convergence test failures:

50 70 2990 1.0055 0.00915549 3 420 864 48 10 2

First, information on the initialization process is given. The value of the numerical parameter, Number of Codiagonals of the Jacobian Matrix cf. section 3.5, and the time at which the initial condition is evaluated, are given. Then, start time and end time of the initialization procedure are given. As a next group of parameters, the number of equations solved by the numerical algorithm and the memory requirements for the initialization process, expressed as a number of integer variables and a number of real variables, are listed. The number of steps and the number of evaluations of the jacobian matrix required for the calculation of the initial state conclude the information on the initialization process. Information on dynamic calculations starts with the values of the numerical parameters Number of Codiagonals of the Jacobian Matrix and Maximum Internal Step Size cf. section 3.5. The next block of information shows the time consumption of the simulation by giving the time at each output time step. As a next group of parameters, the number of equations solved by the numerical algorithm and the memory requirements for the application of the di erential-algebraic system solver DASSL Petzold, 1983, expressed as a number of integer variables and a number of real variables, are listed. The next block of information shows the current at the end of the simulation values of integration time, step size and integration order, gives a summary on the number of integration steps taken, of function evaluations and of evaluations of the jacobian matrix, and gives the numbers of error test failures and convergence test failures of the algorithm DASSL Petzold, 1983. Especially these last two numbers may be interesting for simulation performance improvements. An error test failure indicates that the algorithm tries to perform a too large time step and it has to repeat the step with a smaller step size in order to ful ll the accuracy criteria of all state and program variables cf. section 3.5. A change of the accuracies of state and program variables cf. section 3.5 may decrease the number of error test failures. A convergence test failure indicates that the algorithm had to decrease the size of the time step because it was not able to nd the solution of

6.3. TROUBLESHOOTING

183

the nonlinear system of algebraic equations for the solution after the time step the iterative equation solver did not converge; cf. section 4.1. One cause for a large number of convergence test failures may be a too low value of the numerical parameter Number of Codiagonals of the Jacobian Matrix.
Problems of Wrong Calculation Results Due to the robustness of the integration algorithm used by DASSL Petzold, 1983 it is very rare that wrong simulation results occur. To our knowledge the only cases in which wrong calculation results occur are cases related to the automatic selection of integration time steps. If the integration time step is large in comparison to a short term change in system behaviour, the algorithm may step over such a change in behavior. The consequence of this problem is that the change in behaviour is not recognised by the calculation. There are two very important examples of this problem. The rst is related to shortterm external excitation of a system, the second to short term changes in process rates due to changes in the values of variables calculated from the solution. a If the driving forces of a system process rates and inputs usually only vary slowly in time but show some short-term excitations, then, during the slow variation phase, the integration algorithm gradually increases the size of its time step. If then a short-term excitation with a duration shorter than the time step occurs, the algorithm may hit the excitation or it may step over the excitation. In the rst case the algorithm recognises the integration problem and it follows the excitation accurately by decreasing the integration step size appropriately. In the second case the algorithm steps over the excitation without recognising it and the simulation result becomes erroneous. The simplest way of solving this problem is to decrease the value of the maximum internal step size of the integration algorithm to a value smaller than the duration of the excitation as described in section 3.5. If this slows down the integration too much during the smooth behaviour periods, an alternative may be to decrease the output time step to a value smaller than one tenth of the duration of the excitation during the excitation only this also leads to better results for the plots, and let the output time step to be large during the smooth periods. This also works because the size of the internal time step is not allowed to increase to values larger than ten times the output time step. b If changes in reaction rates may occur abruptly as a function of a variable calculated by the program e.g. a calculated concentration, a similar problem as described under a may occur. The di erence of the present problem to the one described above is that the time at which it occurs is unknown. This makes the second solution procedure proposed above inapplicable decreasing the time step only in the critical ranges of time. The rst solution decreasing the internal time step may improve the results, however, the appropriate size of the time step is more di cult to estimate, because it depends on the time scale of changes of the values of system variables which are calculated by the program. For this reason and to make the integration faster, it is better to solve this problem by changing the formulation of the rate expressions in a way that makes the problem detectable by the integration algorithm. This procedure is illustrated with a simple example: A rate expression of Monod type

where r is the degradation rate of the substance described by the concentration C , rmax is the maximum degradation rate and K is the concentration at which the degradation

rmax r = K +C C

6.1

184

CHAPTER 6. APPENDIX

rate is half of its maximum value, may cause such a problem if the value of K is very small in comparison to typical initial values of C and if degradation continues to zero. For large concentrations C K  the degradation rate is approximately equal to rmax , for decreasing concentrations it tends to zero. However, for large negative concentrations, the degradation rate is also equal to rmax . If now the integration algorithm makes huge time steps because the decrease of C with a constant rate rmax is very simple to integrate, the concentration may jump over the whole range of small and negative rates to large negative values of C , where the rate is again approximately equal to rmax , and the degradation process continues to larger negative values of C . This problem only occurs if typical concentrations C are much larger than K and a fast decrease of C to zero is possible. An elegant way to solve this problem is to extend the function given above in a di erent way to negative concentrations to avoid the problem that again rmax is returned. If the rate is extended linearly by

r=

8 rmax C

the problem is avoided without changing the function in the domain of resonable positive concentrations C . If now the algorithm steps to negative concentrations, it obtains negative degradation rates that di er signi cantly from the positive value of rmax . Such a rapid change in the rate during a single time step is not acceptable for the error test criterion and the step size is decreased. For this reason, the correct smooth approach to zero is achieved and negative concentrations do not occur in the nal model results they may only occur during integration steps that are not accepted. In most cases, errors in the user input are the cause for wrong simulation results. It is advantageous to print out the system de nitions and to check these de nitions carefully in order to nd the cause of the problem. In addition it may be advantageous to plot important system variables that do not show correct behaviour and to turn o processes one after the other in order to nd out which process caused the problem. Then this process can be studied in more detail.
Problems of Integration Interrupts If the integration algorithm detects a serious problem that cannot be solved by reducing the integration step size and order, the integration process is interrupted with the message box shown in Fig. 6.3. If this dialog box appears, in most cases there will be a message on

K +C : rmax C K

for C  0 for C  0

6.2

Figure 6.3: Example of a dialog box indicating a numerical problem. the log le of AQUASIM that may give hints with respect to the nature of the problems.

6.3. TROUBLESHOOTING

185

Two types of error should be distinguished: Error messages by the integration algorithm DASSL Petzold, 1983 which start with the keyword `DASSL' and error messages by AQUASIM. Some examples of such error messages are discussed in the following two subsections. DASSL Errors DASSL errors are more di cult to interpret than AQUASIM errors because no speci c reference to AQUASIM system de nitions can be given. The DASSL error message
05 13 1998 15:27:39 Start of calculation 05 13 1998 15:27:39 Integration at time 0 DASSL AT CURRENT T = 0.00990773 1000 STEPS DASSL TAKEN ON THIS CALL BEFORE REACHING TOUT. 05 13 1998 15:27:39 End of calculation

indicates that more time steps are required to reach the next point of output time than was speci ed as the value of the numerical parameter Maximum Number of Internal Time Steps for One External Time Step in the present example, the value of this parameter was set equal to 1000; cf. section 3.5. This limitation of the number of internal steps for one external step was made in order to show numerical problems without loosing too much computing time. Under usual circumstances the requirement for more than 1000 internal steps for one external step indicates the existence of a numerical problem. There are two important exceptions from this general rule: If a dynamic simulations starts with a relaxation to an equilibrium state and this is done with one output step only because the relaxation dynamics is not of interest, a large number of internal steps may be required in order to calculate the single external output step. The message shown above may also appear during a parameter estimation, for which a simulation could be run without problems. This is the case if the temporal distance between data points is much larger than the output interval selected for the simulation. For such cases it is necessary to increase the numerical parameter Maximum Number of Internal Time Steps for One External Time Step as discussed in section 3.5. In other cases it may be more appropriate to try to nd out why the integration algorithm performs poorly with very small time steps. The DASSL error message that is most di cult to interpret is the following:
05 13 1998 15:32:44 Start of calculation 05 13 1998 15:32:44 Integration at time 0 05 13 1998 15:32:44 Integration at time 0.1 05 13 1998 15:32:44 Integration at time 0.2 05 13 1998 15:32:44 Integration at time 0.3 05 13 1998 15:32:44 Integration at time 0.4 05 13 1998 15:32:44 Integration at time 0.5 05 13 1998 15:32:44 Integration at time 0.6 05 13 1998 15:32:44 Integration at time 0.7 05 13 1998 15:32:44 Integration at time 0.8 05 13 1998 15:32:44 Integration at time 0.9 DASSL AT T = 1 AND STEPSIZE H = 2.06912e-024 DASSL THE ERROR TEST FAILED REPEATEDLY DASSL OR ABSH = HMIN. 05 13 1998 15:32:44 End of calculation

186

CHAPTER 6. APPENDIX

This or a similar error message results if the algorithm repeatedly tried to reduce the integration step size without being able to ful ll the error test criterion. It is then the di cult task of the program user to nd the cause of the numerical problem. The point in time at which the problem occurs may help to nd the cause. A rate or an input could be discontinuous at this point in time. It may be useful to look at time series plots of important system variables in order to detect strange behaviour at this point in time. It may also be useful to change the accuracy of program and state variables in order to nd out if the problem occurs always at the same time cf. section 3.5. AQUASIM Errors In most cases it is easier to nd the cause for AQUASIM error messages. The following example shows an error message that occurs when in the advective-di usive compartment a negative di usion coe cient is speci ed:
05 13 1998 15:44:46 Start of calculation ADVCOMP numerical problem: Di usion coe cient is not positive 05 13 1998 15:44:46 End of calculation

If the error message is not clear enough, it is recommended to read the section of the user manual in which the edit eld of the referenced quantity is described.

6.3. TROUBLESHOOTING

187

6.3.2 Finding Help in the AQUASIM User Group

To join the AQUASIM user group, send the mail message
subscribe aquasim-users mail-address

In order to facilitate communication on program use and application among AQUASIM users, an electronic user group is maintained at the EAWAG mail server.

to
[email protected]

In the mail message mail-address must be replaced by the mail address that should be added to the user group. Because the mail is processed automatically, it is very important that the mail message has exactly the format given above and does not include any further comments. The text must be given in the body of the message and not as a subject. In order to leave the AQUASIM user group send the mail message
unsubscribe aquasim-users mail-address

to
[email protected]

Your mail address is then deleted from the user group list. Questions to be distributed or responses to distributed questions that should again be distributed must then be sent to
[email protected]

All messages sent to this address are automatically distributed to all members of the list. Usually, it is recommended to send your responses to distributed questions also to the user group in order to allow all user group members to participate in the discussion. However, individual answers to the person who sent out the question, are also possible.

188

CHAPTER 6. APPENDIX

6.3.3 Reporting Program Bugs and Suggestions for Improvements

Please report program errors as well as suggestions for program improvements to Peter Reichert, EAWAG, CH-8600 Dubendorf, Switzerland, or by email to [email protected]. An accurate description of the error accompanied by an AQUASIM system le which leads to the problem is very important. We will provide information on known program errors, on workarounds and on bugxed program versions on the EAWAG home page at http: www.eawag.ch.

Bibliography
Albrecht, A., Reichert, P., Beer, J., and Luck, A. 1995. Evaluation of the importance of reservoir sediments as sinks for reactor-derived radionuclides in riverine systems. J. Environ. Radioactivity, 283:239 269. Amorocho, J. and de Vries, J. 1980. A new evaluation of the wind stress coe cient over water surface. J. Geophys. Res., 85:433 442. Arcangeli, J.-P. and Arvin, E. 1997a. Modelling of the cometabolic biodegradation of trichloroethylene by toluene-oxidizing bacteria in a bio lm system. Environmental Science & Technology, 31:3044 3052. Arcangeli, J.-P. and Arvin, E. 1997b. Modelling of the growth of a methanotrophic bio lm. Water Science and Technology, 361:199 204. Beaudoin, D., Bryers, J., Cunningham, A., and Peretti, S. 1997. Mobilization of broad host range plasmid from Pseudomonas putida to established bio lms of Bacillus azotoformans. II. modeling. Biotechnology & Bioengineering, 573:280 286. Buhrer, H. and Ambuhl, H. 1975. Die Einleitung von gereinigtem Abwasser in Seen. Schweizerische Zeitschrift fur Hydrologie, 37:347 369. Burchard, H. and Baumert, H. 1995. On the performance of a mixed-layer model based on the k- turbulence closure. J. Geophys. Res., 100C5:8523 8540. Chow, V. 1959. Open-channel hydraulics. McGraw-Hill, New York. de St. Venant, M. 1871. Th orie du mouvement non permanant des eaux crues des e rivires et  l'introduction des mar es dans leur lit. Comptes Rendus, 73:147 154 & e a e 237 240. Eisenbeis, M., Bauer-Kreisel, P., and Scholz-Muramatsu, H. 1997. Studies of the dechlorination of tetrachloroethene to cis -1,2-dichloroethene by dehalospirillium multivorans in bio lms. Water Science and Technology, 361:191 198. Fesch, C., Lehmann, P., Haderlein, S., Hinz, C., Schwarzenbach, R., and Fluhler, H. 1998a. E ect of water content on solute transport in aggregated porous media. J. Contaminant Hydrology, 331-2:211 230. Fesch, C., Simon, W., Reichert, P., Haderlein, S., and Schwarzenbach, R. 1998b. Nonlinear sorption and nonequilibrium transport of organic contaminants in saturated porous media: experiments, process identi cation and modeling. J. Contaminant Hydrology, 313-4:373 407. 189

190

BIBLIOGRAPHY

Filipe, C. and Daigger, G. 1997. Development of a revised metabolic model for the growth of phosphorus-accumulating organisms. Water Environment Research, 701:67 79. Fischer, H., List, E., Koh, C., Imberger, J., and Brooks, N. 1979. Mixing in Inland and Coastal Waters. Academic Press, New York. French, R. 1985. Open-channel hydraulics. McGraw-Hill, Stuttgart. New York. Gear, C. 1971a. Algorithm 407, DIFSUB for solution of ordinary di erential equations. Communications ACM, 143:185 190. Gear, C. 1971b. The automatic integration of ordinary di erential equations. Communications ACM, 143:176 179. Gear, C. 1971c. Numerical initial value problems in ordinary di erential equations. Prentice Hall, Stuttgart. Englewood Cli s, N.J. Gill, P., Murray, W., and Wright, M. 1981. Practical Optimization. Academic Press, London. Glod, G., Angst, W., Holliger, C., and Schwarzenbach, R. 1997a. Corrinoid-mediated reduction of Tetrachloroethene, Trichloroethene, and Trichloro uoroethene in homogeneous aqueous solution: Reaction kinetics and reaction mechanisms. Environ. Sci. Technol., 31:253 260. Glod, G., Brodmann, U., Angst, W., Holliger, C., and Schwarzenbach, R. 1997b. Cobalamin-mediated reduction of cis - and trans -dichloroethene, 1,1-dichloroethene, and vinyl chloride in homogeneous aqueous solution: reaction kinetics and mechanistic considerations. Environ. Sci. Technol., 31:3154 3160. Goudsmit, G.-H., Reichert, P., and Wuest, A. 1996. Modelling of physical and biogeochemical properties in lakes using AQUASIM. In Muller, A., editor, Hydroinformatics 96, pages 779 786. Balkema, Rotterdam. Gujer, W. and Wanner, O. 1990. Modeling mixed population bio lms. In Characklis, W. and Marshall, K., editors, Bio lms, pages 397 443. John Wiley & Sons, New York. Henderson, F. 1966. Open channel ow. Macmillan, New York. Henze, M., Grady, C., Gujer, W., Marais, G., and Matsuo, T. 1986. Activated sludge model no. 1. Scienti c and Technical Report 1, IAWPRC Task Group on Mathematical Modelling for Design and Operation of Biological Wastewater Treatment Processes, IAWPRC, London. Hindmarsh, A. 1983. ODEPACK. a systematized collection of ODE solvers. In Stepleman, R., editor, Scienti c Computing, pages 55 64. IMACS North-Holland. Horn, H. and Hempel, D. 1995. Mass transfer coe cients for an autotrophic and a heterotrophic bio lm system. Wat. Sci. Tech., 328:199 204. Horn, H. and Hempel, D. 1997. Substrate utilization and mass transfer in an autotrophic bio lm system: Experimental results and numerical simulation. Biotechnology and Bioengineering, 534:363 371.

BIBLIOGRAPHY

191

Jancarkova, I., Larsen, T., and Gujer, W. 1997. Distribution of nitrifying bacteria in a shallow stream. Water Science and Technology, 368-9:161 166. Janning, K., Harremoes, P., and Nielsen, M. 1995. Evaluating and modelling the kinetics in a full scale submerged denitri cation lter. Wat. Sci. Tech., 328:115 123. Kuba, T., Murnleitner, E., van Loosdrecht, M., and Heiijnen, J. 1996. A metabolic model for biological phosphorus removal by denitrifying orgnisms. Biotechnology and Bioengineering, 526:685 695. LeVeque, R. J. 1990. Numerical Methods for Conservation Laws. Birkhauser, Basel. Londong, J., Borchardt, D., Firk, W., Reichert, P., Stein, M., and Strotmann, U. 1994. Nitrit in Fliessgewassern. Korrespondenz Abwasser, 11 94:2069 2076. Maryns, F. and Bauwens, W. 1997. The application fo the activated sludge model no. 1 to a river environment. Water Science and Technology, 365:201 208. Mirpuri, R., Sharp, W., Villaverde, S., Jones, W., Lewandowski, Z., and Cunningham, A. 1997. Predictive model for Toluene degradation and microbial phenotypic pro les in at plate vapor phase bioreactor. Journal of Environmental Engineering, 6 97:586 592. Murnleitner, E., Kuba, T., van Loosdrecht, M., and Heiijnen, J. 1997. An integrated metabolic model for the aerobic and denitrifying biological phosphorus removal. Biotechnology and Bioengineering, 545:434 450. Musvoto, E., Wentzel, M., Loeventhal, R., and Ekama, G. 1997. Kinetic-based model for mixed weak acid base systems. Water SA, 234:311 322. Nelder, J. and Mead, R. 1965. A simplex method for function minimization. Computer Journal, 7:308 313. Novack, B. and Sigg, L. 1997. Dissolution of FeIIhydroxides by metal-EDTA complexes. Geochimica et Cosmochimica Acta, 615:951 963. Peeters, F., Kipfer, R., Hohmann, R., Hofer, M., Imboden, D., Kodenev, G., and Khozder, T. 1997. Modeling transport rates in lake Baikal: Gas exchange and deep water renewal. Environ. Sci. Technol., 31:2973 2982. Petersen, E. 1965. Chemical Reaction Analysis. Prentice-Hall, Englewood Cli s, New Jersey. Petzold, L. 1983. A description of DASSL: A di erential algebraic system solver. In Stepleman, R. e., editor, Scienti c Computing, pages 65 68. IMACS North-Holland, Amsterdam. Ralston, M. and Jennrich, R. 1978. DUD - a derivative-free algorithm for nonlinear least squares. Technometrics, 201:7 14. Reichert, P. 1994a. AQUASIM - A tool for simulation and data analysis of aquatic systems. Wat. Sci. Tech., 302:21 30.

192

BIBLIOGRAPHY

Reichert, P. 1994b. Concepts underlying a computer program for the identi cation and simulation of aquatic systems. Schriftenreihe der EAWAG 7, Swiss Federal Institute for Environmental Science and Technology EAWAG, CH-8600 Dubendorf, Switzerland. Reichert, P. 1995. Design techniques of a computer program for the identi cation of processes and the simulation of water quality in aquatic systems. Environmental Software, 103:199 210. Reichert, P. 1998. AQUASIM 2.0 Tutorial. Technical report, Swiss Federal Institute for Environmental Science and Technology EAWAG, CH-8600 Dubendorf, Switzerland. Reichert, P., von Schulthess, R., and Wild, D. 1995. The use of AQUASIM for estimating parameters of activated sludge models. Wat. Sci. Tech., 312:135 147. Reichert, P. and Wanner, O. 1997. Movement of solids in bio lms: Signi cance of liquid phase transport. Water Science and Technology, 361:321 328. Rodi, W. 1980. Turbulence models and their applications in hydraulics. State of the art paper, International Association for Hydraulic Research IAHR, Delft, Netherlands. Rodi, W. 1987. Examples of calculation methods for ow and mixing in strati ed ows. J. Geophys. Res., 92:5305 5328. Sanderson, S. and Stewart, P. 1997. Evidence of bacterial adaptation to monochloramine in Pseudomonas aeruginosa bio lms and evaluation of biocide action model. Biotechnology & Bioengineering, 562:201 209. Shamir, U. and Harleman, D. 1967. Numerical solutions for dispersion in porous mediums. Water Resources Research, 32:557 581. Siegrist, H., Krebs, P., Buhler, R., Purtschert, I., Rock, C., and Rufer, R. 1995. Denitri cation in secondary clarifyers. Wat. Sci. Tech., 312:205 214. Simon, W., Reichert, P., and Hinz, C. 1997. Properties of exact and approximate travelling wave solutions for transport with nonlinear and nonequilibrium sorption. Water Resources Research, 335:1139 1147. Suci, P., Vrany, J., and Mittelman, M. 1998. Investigation of interactions between antimicrobial agents and bacterial bio lms using attenuated total re ection Fourier transform infrared spectroscopy. Biomaterials, 19:327 339. Sweby, P. 1984. High resolution schemes using ux limiters for hyperbolic conservation laws. SIAM J. Numer. Anal., 215:995 1011. Uehlinger, U., Buhrer, H., and Reichert, P. 1996. Periphyton dynamics in a oodprone prealpine river: Evaluation of signi cant processes by modelling. Freshwater Biology, 36:249 263. Ulrich, M., Imboden, D., and Schwarzenbach, R. 1995. MASAS - A user-friendly simulation tool for modeling the fate of anthropogenic substances in lakes. Environmental Software, 103:177 198.

BIBLIOGRAPHY

193

van Leer, B. 1974. Towards the ultimate conservative di erence scheme ii: Monotonicity and conservation combined in a second order scheme. Journal of Computational Physics, 14:361 370. von Gunten, U., Elovitz, M., and Kaiser, H. 1997. Characterization of ozonation processes with conservative and reactive tracers: Prediction of the degradation of micropollutants. Analusis Magazine, 257:29 31. von Schulthess, R. and Gujer, W. 1996. Release of nitrous oxide from denitrifying activated sludge: Veri cation and application of a mathematical model. Water Research, 303:521 530. von Schulthess, R., Wild, D., and Gujer, W. 1994. Nitric and nitrous oxides from denitrifying activated sludge at low oxygen concentration. Wat. Sci. Tech., 306:123 132. Vrany, J., Stewart, P., and Suci, P. 1997. Comparison of Recalcitrance to Cipro oxacin and Levo oxacin exhibited by Pseudomonas aeruginosa bio lms displaying rapidtransport characteristics. Antimicrobial Agents and Chemotherapy, 416:1352 1358. Wanner, O. 1994. Modeling of mixed-population bio lm accumulation. In Geesey, G., Levandowski, Z., and Flemming, H.-C., editors, Biofouling and Biocorosion in Industrial Water Systems, pages 37 62. Lewis Publishers, Boca Raton. Wanner, O. 1995. New experimental ndings and bio lm modelling concepts. Wat. Sci. Tech., 32:133 140. Wanner, O. 1996. Modelling of bio lms. Biofouling, 101-3:31 41. Wanner, O., Cunningham, A., and Lundman, R. 1995. Modeling bio lm accumulation and mass transport in a porous medium under high substrate loading. Biotechnology & Bioengineering, 47:703 712. Wanner, O., Debus, O., and Reichert, P. 1994. Modelling the spatial distribution and dynamics of a xylene-degrading microbial population in a membrane-bound bio lm. Wat. Sci. Tech., 2910-11:243 251. Wanner, O. and Gujer, W. 1984. Competition in bio lms. Wat. Sci. Tech., 17:27 44. Wanner, O. and Gujer, W. 1986. A multispecies bio lm model. Biotechnology & Bioengineering, 28:314 328. Wanner, O. and Reichert, P. 1996. Mathematical modelling of mixed-culture bio lms. Biotechnology & Bioengineering, 49:172 184. Wild, D., von Schulthess, R., and Gujer, W. 1994. Synthesis of denitri cation enzymes in activated sludge: Modelling with structured biomass. Wat. Sci. Tech., 306:113 122. Wild, D., von Schulthess, R., and Gujer, W. 1995. Structured modelling of denitri cation intermediates. Wat. Sci. Tech., 312:45 54. Yen, B. 1973. Open-channel ow equations revisited. Journal of the Mechanical Engineering Division, ASCE, 99:979 1009.

194

BIBLIOGRAPHY

Yen, B. 1979. Unsteady ow mathematical modeling techniques. In Shen, H., editor, Modeling of Rivers, pages 13 1 13 33. John Wiley, New York.

Index
about le menu, 8 abs formula variable, 24 abscissa plot, 163 absolute accuracy program variable advective-di usive reactor compartment, 67 bio lm reactor compartment, 57 lake compartment, 123 mixed reactor compartment, 38 river section compartment, 96 saturated soil column compartment, 84 state variable, 14, 134 accuracy program variable, 132, 134, 142 advective-di usive reactor compartment, 62, 67 bio lm reactor compartment, 51, 56 lake compartment, 110, 123 mixed reactor compartment, 35, 38 river section compartment, 91, 96 saturated soil column compartment, 76, 83 state variable, 14, 132, 134, 142 acos formula variable, 24 active for calculation advective-di usive reactor compartment, 62 bio lm reactor compartment, 51 lake compartment, 123 mixed reactor compartment, 35 river section compartment, 91 saturated soil column compartment, 76 195 for parameter estimation calculation, 154 constant variable, 17, 153 for sensitivity analysis calculation, 143 constant variable, 17, 147 real list variable, 21, 147 for simulation calculation, 143 active for parameter estimation calculation, 154 constant variable, 17, 153 active for sensitivity analysis calculation, 143 constant variable, 17, 147 real list variable, 21, 147 active for simulation calculation, 143 active processes advective-di usive reactor compartment, 62, 63 bio lm reactor compartment, 49, 52 lake compartment, 108, 111 mixed reactor compartment, 35, 36 river section compartment, 89, 92 saturated soil column compartment, 75, 77 active state variables advective-di usive reactor compartment, 62 bio lm reactor compartment, 49, 51 lake compartment, 108, 110 mixed reactor compartment, 35 river section compartment, 89, 91 saturated soil column compartment, 75, 76 add argument-value pairs variable list variable, 22 add data pairs real list variable, 21

196 advective link, 10, 125 129 bifurcation, 126, 127 description, 128 name, 128 substance ow, 126, 128 water ow, 126, 128 connection, 127 description, 127 equations, 126 127 link index, 127 name, 127 overview, 126 state variable, 126, 128, 129 dynamic, 126, 128, 129 volume, 126, 128, 129 substance ow, 126 user de nitions, 127 129 water ow, 126 advective-di usive reactor compartment, 10, 32, 59 69 accuracy of program variables, 62, 67 active for calculation, 62 active processes, 62, 63 active state variables, 62 area, 59, 62 boundary condition, 61 compartment index, 62 cross section, 62 description, 62 di usion coe cient, 60, 62, 68 discharge, 60 dispersion coe cient, 60, 62 end coordinate, 62 equations, 59 61 initial conditions, 62, 63 inlet input, 64, 65 input, 62, 64 inlet, 64, 65 lateral, 64, 66 lateral in ow, 60 lateral input, 64, 66 name, 61 number of grid points, 62 overview, 59 resolution, 62 space coordinate x, 62 start coordinate, 62 state variable, 59, 62, 64, 65, 67

INDEX
dynamic, 59, 65, 67 equilibrium, 59 surface, 59 volume, 59, 65, 67 user de nitions, 61 69 applications of AQUASIM, 1 aquasim.log, 8, 137, 181 183 area advective-di usive reactor compartment, 59, 62 bio lm reactor compartment, 42, 50 lake compartment, 99, 109 river section compartment, 85, 89 saturated soil column compartment, 71, 76 area gradient lake compartment program variable, 124 program variable, 17 argument real list variable, 18, 20 variable list variable, 22 asin formula variable, 24 atan formula variable, 24 attachment bio lm reactor compartment, 41 attachment coe cient surface bio lm reactor compartment, 46, 56 volume bio lm reactor compartment, 44, 56 attachment velocity bio lm reactor compartment, 45, 46 attachment velocity of bio lm bio lm reactor compartment program variable, 58 program variable, 16 batch version, 3, 4, 8, 173 175 bifurcation advective link, 126, 127 bio lm area bio lm reactor compartment, 50 bio lm matrix di usive bio lm reactor compartment, 45, 50

INDEX
rigid bio lm reactor compartment, 46, 50 bio lm reactor compartment, 10, 32, 41 58 accuracy of program variables, 51, 56 active for calculation, 51 active processes, 49, 52 active state variables, 49, 51 area, 42, 50 attachment, 41 attachment coe cient surface, 46, 56 volume, 44, 56 attachment velocity, 45, 46 bio lm area, 50 bio lm matrix di usive, 45, 50 rigid, 46, 50 bio lm thickness, 41, 45 boundary condition, 46, 47 bulk volume, 41, 48, 50 compartment index, 49 con ned reactor, 48, 50 density, 56 description, 49 detachment, 41 global velocity, 50 individual rate, 50 detachment coe cient surface, 45, 56 volume, 44, 56 detachment velocity, 45 di usion coe cient bio lm matrix, 56 pore volume, 56 di usive solid matrix, 50 dissolved substances, 41 dissolved variables, 49, 56 equations, 41 48 free volume growth rate of, 43 initial conditions, 49, 52 input, 49, 53 liquid phase, 50 liquid phase volume fraction, 42 mass transfer resistance, 48, 56 name, 49 number of grid points, 51

197 overview, 41 particulate variables, 49, 55 pore volume, 41 porosity, 42 rate of change, 51 reactor type, 50 reactor volume, 48, 50 resolution, 51 rigid solid matrix, 50 solid matrix, 41 space coordinate z, 41 state variable, 42, 49 54 dynamic, 42, 49, 54 volume, 42, 49, 54 substratum, 41 surface attachment coe cient, 56 surface detachment coe cient, 56 suspended solids, 41, 50 uncon ned reactor, 48, 50 user de nitions, 49 58 velocity of bio lm matrix, 43 volume bulk, 48 reactor, 48 volume attachment coe cient, 56 volume detachment coe cient, 56 zones, 41 bio lm thickness bio lm reactor compartment, 41, 45 program variable, 53, 57, 58 program variable, 16 bottom coordinate lake compartment, 109 bottom friction lake compartment, 103, 123 boundary condition advective-di usive reactor compartment, 61 bio lm reactor compartment, 46, 47 lake compartment, 106, 107 river section compartment, 87, 88 saturated soil column compartment, 74 Brunt Vaisala frequency lake compartment, 109, 110 program variable, 110, 112, 124 program variable, 17 bulk volume

198 bio lm reactor compartment, 41, 48, 50 program variable, 57, 58 mixed reactor compartment program variable, 39 program variable, 16 buoyancy production of TKE lake compartment, 101, 103 program variable, 124 program variable, 17 calc menu, 5, 137 160 calculation, 148 parameter estimation, 155 156 calculation number, 155 description, 155 initial state, 155 initial time, 155 name, 155 sensitivity analysis, 140 143 calculation number, 141 description, 141 initial state, 141 initial time, 141 name, 140, 141 number of steps, 142 output steps, 142 step size, 142 simulation, 140 143 description, 141 initial state, 141 initial time, 141 name, 140, 141 number of steps, 142 output steps, 142 step size, 142 calculation number advective link program variable, 128 advective-di usive reactor compartment program variable, 69 bio lm reactor compartment program variable, 58 calculation parameter estimation, 155 sensitivity analysis, 141 simulation, 141

INDEX
curve, 165 lake compartment program variable, 113, 124 mixed reactor compartment program variable, 39 program variable, 15, 16 river section compartment program variable, 97 saturated soil column compartment program variable, 84 character interface version, 3, 4, 171 172 close le menu, 7 column number of argument real list variable, 21 column number of standard deviations real list variable, 21 column number of values real list variable, 21 compartment, 10, 32 124 advective-di usive reactor, 10, 32, 59 69 bio lm reactor, 10, 32, 41 58 curve, 166 t target, 157 lake, 10, 32, 98 124 mixed reactor, 10, 32, 34 40 probe variable, 26 river section, 10, 32, 85 97 saturated soil column, 10, 32, 70 84 compartment index advective-di usive reactor compartment, 62 program variable, 62, 69 bio lm reactor compartment, 49 program variable, 49, 58 lake compartment, 108 program variable, 108, 124 mixed reactor compartment, 34 program variable, 34, 39 program variable, 16 river section compartment, 88 program variable, 88, 97 saturated soil column compartment, 75 program variable, 75, 84 compartment type, 10, 32 compartments

INDEX
edit menu, 11 computing platforms, 4 conductivity lake compartment, 109 con ned reactor bio lm reactor compartment, 48, 50 connection advective link, 127 constant variable, 10, 12, 17, 21, 151 active for parameter estimation, 17, 153 active for sensitivity analysis, 17 description, 17 maximum, 17 minimum, 17 name, 17 standard deviation, 17 unit, 17 value, 17 constant volume mixed reactor compartment, 35 continue simulation, 141 conversion factor di usive link, 129, 131 saturated soil column compartment, 83 correlation coe cients, 144 cos formula variable, 24 cosh formula variable, 24 critical water level river section compartment, 91 cross section advective-di usive reactor compartment, 62 lake compartment, 109 river section compartment, 89 saturated soil column compartment, 76 cross-sectional area advective-di usive reactor compartment program variable, 62, 69 lake compartment program variable, 100, 124 program variable, 16

199 river section compartment program variable, 96, 97 saturated soil column compartment program variable, 76, 84 curve plot, 163 DASSL, 132, 133, 139, 182, 183, 185 data t target, 156 data pairs real list variable, 18 deg formula variable, 24 delete argument-value pairs variable list variable, 22 delete data pairs real list variable, 21 delete states, 135 136 edit menu, 11 density bio lm reactor compartment, 56 lake compartment, 109, 119 program variable, 124 program variable, 17 saturated soil column compartment, 74 description advective link, 127 bifurcation, 128 advective-di usive reactor compartment, 62 bio lm reactor compartment, 49 calculation parameter estimation, 155 sensitivity analysis, 141 simulation, 141 constant variable, 17 di usive link, 130 dynamic process, 29 equilibrium process, 31 formula variable, 23 immobile region saturated soil column compartment, 82 lake compartment, 108 mixed reactor compartment, 35 plot, 163

200 probe variable, 26 program variable, 15 real list variable, 20 river section compartment, 88 saturated soil column compartment, 75 state variable, 14 variable list variable, 21 design of AQUASIM, 2 detachment bio lm reactor compartment, 41 global velocity bio lm reactor compartment, 50 individula rate bio lm reactor compartment, 50 detachment coe cient surface bio lm reactor compartment, 45, 56 volume bio lm reactor compartment, 44, 56 detachment velocity of bio lm bio lm reactor compartment, 45 program variable, 58 program variable, 16 di usion coe cient advective-di usive reactor compartment, 60, 62, 68 bio lm matrix bio lm reactor compartment, 56 lake compartment turbulent, 99 molecular lake compartment, 103, 120 pore volume bio lm reactor compartment, 56 turbulent lake compartment, 100, 109 di usive approximation river section compartment, 86, 91 di usive link, 10, 125, 129 131 conversion factor, 129, 131 description, 130 equations, 129 130 exchange coe cient, 129, 131 link index, 130 name, 130 overview, 129 state variable, 131

INDEX
dynamic, 131 volume, 131 user de nitions, 130 131 di usive solid matrix bio lm reactor compartment, 50 discharge advective link program variable, 128 advective-di usive reactor compartment, 60 program variable, 62, 65, 67 69 bio lm reactor compartment program variable, 54, 57, 58 lake compartment, 100, 104 program variable, 115, 123, 124 mixed reactor compartment program variable, 37 39 program variable, 16 river section compartment, 86 program variable, 93, 95 97 saturated soil column compartment program variable, 76, 79, 81, 84 discretization rst order, 138 high resolution, 138 low resolution, 138 second order, 138 spatial, 138 dispersion coe cient advective-di usive reactor compartment, 60, 62 river section compartment, 86, 90, 96 saturated soil column compartment, 71, 76, 84 dissipation lake compartment, 99, 103, 105 program variable, 100, 123, 124 program variable, 17 dissolved substances bio lm reactor compartment, 41 dissolved variables bio lm reactor compartment, 49, 56 lake compartment, 108, 120 dynamic process, 10, 27 30 description, 29 name, 29 rate, 29 stoichiometric coe cient, 29

INDEX
dynamic state variable, 14 edit advective link, 126 129 advective-di usive reactor compartment, 59 69 bio lm reactor compartment, 41 58 calculation parameter estimation, 155 156 sensitivity analysis, 141 143 simulation, 141 143 compartments, 11, 32 124 constant variable, 17 delete states, 135 136 di usive link, 129 131 dynamic process, 28 30 equilibrium process, 30 31 formula variable, 23 25 lake compartment, 98 124 links, 11, 125 131 mixed reactor compartment, 34 40 numerical parameters, 11, 132 134 probe variable, 26 processes, 11, 27 31 program variable, 15 17, 40 real list variable, 18 21 river section compartment, 85 97 saturated soil column compartment, 70 84 state variable, 14 15 system, 11 variable list variable, 21 22 variables, 11 26 edit menu, 5, 9, 10, 12, 13, 27, 28, 136 electronic mail, 8 end coordinate advective-di usive reactor compartment, 62 river section compartment, 89 saturated soil column compartment, 75 end row real list variable, 21 energy of seiche oscillation lake compartment program variable, 112, 123, 124 program variable, 17 equation

201 equilibrium process, 31 equilibrium process, 10, 27, 30 31 description, 31 equation, 31 name, 30 equilibrium state variable, 14 error bounds of, 146 contribution to, 146 propagation of, 146 error contribution, 146, 165 plot, 161 error propagation, 146 exchange coe cient di usive link, 129, 131 saturated soil column compartment, 82, 83 exit le menu, 8 exp formula variable, 24 expression formula variable, 23 le menu, 4, 7 8 le options, 162, 168 rst order, 138 t target parameter estimation, 156 ux limiter, 138 format print le, 8 long, 8 short, 8 system le, 177 formula variable, 10, 12, 23 25 abs, 24 acos, 24 asin, 24 atan, 24 cos, 24 cosh, 24 deg, 24 description, 23 exp, 24 expression, 23 functions, 24 25 ln, 24

202 log, 24 log10, 24 max, 24 min, 24 name, 23 pi, 24 rad, 24 sign, 24 sin, 24 sinh, 24 sqrt, 25 tan, 25 tanh, 25 unit, 23 free volume growth rate of bio lm reactor compartment, 43 friction factor river section compartment, 90 friction slope program variable, 17 river section compartment, 86, 90 Darcy-Weisbach, 90 Manning-Strickler, 90 program variable, 97 Gear integration technique, 138 140 gravitational acceleration lake compartment, 108 river section compartment, 89 growth velocity of bio lm bio lm reactor compartment program variable, 58 program variable, 16 hardware platforms, 4 high resolution, 138 horizontal velocity lake compartment, 99, 103, 104 program variable, 100, 112, 123, 124 program variable, 17 identi ability analysis, 3, 144 146 immobile region saturated soil column compartment, 70, 76, 81, 82 in ow mixed reactor compartment, 34 initial conditions

INDEX
advective-di usive reactor compartment, 62, 63 bio lm reactor compartment, 49, 52 lake compartment, 108, 112 mixed reactor compartment, 35, 36 river section compartment, 89, 92 saturated soil column compartment, 75, 78 initial state calculation parameter estimation, 155 sensitivity analysis, 141 simulation, 141 initial time calculation parameter estimation, 155 sensitivity analysis, 141 simulation, 141 initialize simulation, 140 inlet input advective-di usive reactor compartment, 64, 65 saturated soil column compartment, 79 input advective-di usive reactor compartment, 62, 64 bio lm reactor compartment, 49, 53 inlet advective-di usive reactor compartment, 64, 65 saturated soil column compartment, 79 lake compartment, 108, 113 lateral advective-di usive reactor compartment, 64, 66 lake compartment, 113, 114 river section compartment, 94, 95 saturated soil column compartment, 79, 80 mixed reactor compartment, 35, 37 point lake compartment, 113, 116 river section compartment, 89, 94 saturated soil column compartment, 75, 79 sediment

INDEX
lake compartment, 113, 117 surface lake compartment, 113 upstream river section compartment, 94 interface velocity of bio lm bio lm reactor compartment program variable, 58 program variable, 16 internal friction lake compartment, 123 internal shear lake compartment, 103 interpolation multidimensional, 21 real list variable, 21 interpolation method linear, 18 real list variable, 18 19 smoothing, 19 spline, 18, 19 variable list variable, 22 kinematic approximation river section compartment, 86, 91 label plot, 163 lake compartment, 10, 32, 98 124 accuracy of program variables, 110, 123 active for calculation, 123 active processes, 108, 111 active state variables, 108, 110 area, 99, 109 bottom coordinate, 109 bottom friction, 103, 123 boundary condition, 106, 107 Brunt Vaisala frequency, 109, 110 buoyancy production of TKE, 101, 103 compartment index, 108 conductivity, 109 cross section, 109 density, 109, 119 description, 108 di usion coe cient molecular, 103, 120 turbulent, 99, 100, 109

203 discharge, 100, 104 dissipation, 99, 103, 105 dissolved variables, 108, 120 equations, 98 107 gravitational acceleration, 108 horizontal velocity, 99, 103, 104 initial conditions, 108, 112 input, 108, 113 lateral, 113, 114 point, 113, 116 sediment, 113, 117 surface, 113 internal friction, 123 internal shear, 103 lateral in ow, 103, 104 lateral input, 113, 114 molecular di usion coe cient, 103, 120 name, 108 number of grid points, 110 overview, 98 particulate variables, 108, 118 point input, 113, 116 porosity sediment layer, 121 Prandtl number, 100, 122 pressure gradient, 103, 122 production of dissipation, 122 resolution, 110 resuspension, 104 resuspension velocity, 121 sediment input, 113, 117 sediment layer, 99, 102, 105, 106 porosity, 99 thickness, 99 sediment submodel, 110, 121 sedimentation, 103 sedimentation velocity, 100, 119 seiche oscillation, 103, 104 shear production of TKE, 101 stability frequency, 109, 110 state variable, 100, 108, 110, 112, 114 116, 118 121 dynamic, 100, 108, 114 116, 118 120 volume, 100, 108, 114 116, 118 120 surface input, 113 surface shear, 122 temperature, 109

204 thickness sediment layer, 121 top coordinate, 109 turbulence submodel, 110, 121 turbulent di usion coe cient, 99, 100, 109 turbulent kinetic energy, 99, 103, 104 user de nitions, 108 124 volume ux, 121 wind excitation, 104, 123 zones, 98 lateral in ow advective-di usive reactor compartment, 60 lake compartment, 103, 104 river section compartment, 87 saturated soil column compartment, 72 lateral input advective-di usive reactor compartment, 64, 66 lake compartment, 113, 114 river section compartment, 94, 95 saturated soil column compartment, 79, 80 legend curve, 166 line curve, 166 linear interpolation real list variable, 18 variable list variable, 22 link, 10, 125 131 advective link, 10, 125 129 di usive link, 10, 125, 129 131 link index advective link, 127 program variable, 127, 128 di usive link, 130 program variable, 16 link type, 10 links edit menu, 11 liquid phase bio lm reactor compartment, 50 liquid phase volume fraction bio lm reactor compartment, 42 list options, 162, 168 list to le, 162 ln formula variable, 24 loading les, 7 problems, 177 178 location probe variable, 26 log formula variable, 24 log le, 8, 137, 181 183 log10 formula variable, 24 long print le format, 8 low resolution, 138

INDEX

mail electronic, 8 marker curve, 166 mass transfer resistance bio lm reactor compartment, 48, 56 max formula variable, 24 maximum constant variable, 17 formula variable, 24 real list variable, 20 maximum integration order, 133 maximum internal step size, 132, 182, 183 maximum number of internal time steps, 133, 185 maximum number of iterations parameter estimation, 154 menu calc, 5, 137 160 parameter estimation, 151 160 sensitivity analysis, 144 150 simulation, 138 143 edit, 5, 9 136 compartments, 32 124 delete states, 135 136 links, 125 131 numerical parameters, 132 134 processes, 27 31 variables, 12 26 le, 4, 7 8 view, 5, 161 170

INDEX
menu bar, 4 min formula variable, 24 minimum constant variable, 17 formula variable, 24 real list variable, 20 mixed immobile zone saturated soil column compartment, 70, 82 mixed reactor compartment, 10, 32, 34 40 constant volume, 34 accuracy of program variables, 35, 38 active for calculation, 35 active processes, 35, 36 active state variables, 35 compartment index, 34 constant volume, 35 description, 35 equations, 34 in ow, 34 initial conditions, 35, 36 input, 35, 37 name, 34 out ow, 34 overview, 34 reactor type, 35 state variable, 34 38 dynamic, 34, 38 equilibrium, 34 surface, 34 volume, 34, 38 user de nitions, 34 40 variable volume, 34, 35 volume, 34, 35 volumetric in ow, 34 volumetric out ow, 34 mobile zone saturated soil column compartment, 70 molecular di usion coe cient lake compartment, 103, 120 multidimensional interpolation, 21 name advective link, 127 bifurcation, 128

205 advective-di usive reactor compartment, 61 bio lm reactor compartment, 49 calculation parameter estimation, 155 sensitivity analysis, 140, 141 simulation, 140, 141 constant variable, 17 di usive link, 130 dynamic process, 29 equilibrium process, 30 formula variable, 23 lake compartment, 108 mixed reactor compartment, 34 plot, 163 probe variable, 26 program variable, 15 real list variable, 20 river section compartment, 88 saturated soil column compartment, 75 immobile region, 81 state variable, 14 variable list variable, 21

new

le menu, 7 normal water level river section compartment, 91 number of codiagonals of the jacobian matrix, 133 number of grid points advective-di usive reactor compartment, 62 bio lm reactor compartment, 51 lake compartment, 110 river section compartment, 91 saturated soil column compartment, 76 number of steps calculation sensitivity analysis, 142 simulation, 142 numerical algorithms rst order spatial discretization, 138 ux limiter technique, 138 Gear integration technique, 138, 140 high resolution spatial discretization, 138

206 low resolution spatial discretization, 138 secant minimization algorithm, 151 154 second order spatial discretization, 138 simplex minimization algorithm, 151, 153, 154 numerical parameters, 132 edit menu, 11 maximum integration order, 133 maximum internal step size, 132, 182, 183 maximum number of internal time steps, 133, 185 number of codiagonals of the jacobian matrix, 133 open le menu, 7 operating systems, 4 options list to le, 162, 168 plot to le, 162, 168 plot to screen, 162, 167 print to le, 8 out ow mixed reactor compartment, 34 output steps calculation sensitivity analysis, 142 simulation, 142 parameter, 165 parameter estimation, 3, 17, 19, 21, 151 160 calculation, 155 156 active for parameter estimation, 154 t target, 156 compartment, 157 data, 156 time space, 157 variable, 157 zone, 157 maximum number of iterations, 154 start, 153, 154 particulate variables bio lm reactor compartment, 49, 55 lake compartment, 108, 118 perimeter length

INDEX
program variable, 16 river section compartment, 89 program variable, 97

pi

formula variable, 24 platforms, 4, 8 plot abscissa, 163 curve, 163 calculation number, 165 compartment, 166 legende, 166 line, 166 marker, 166 time space, 166 type, 164 zone, 166 description, 163 label, 163 name, 163 scaling, 167 title, 163 plot to le, 162 plot to screen, 162 point input lake compartment, 113, 116 pore volume bio lm reactor compartment, 41 porosity bio lm reactor compartment, 42 mobile zone saturated soil column compartment, 71, 76 rate of change bio lm reactor compartment, 51 saturated soil column compartment, 71 sediment layer lake compartment, 99, 121 Prandtl number lake compartment, 100, 122 pressure gradient lake compartment, 103, 122 print le format, 8 long, 8 short, 8 print options

INDEX
le menu, 8 long, 8 short, 8 print to le le menu, 8 probe variable, 10, 12, 26 compartment, 26 description, 26 location, 26 name, 26 unit, 26 variable, 26 zone, 26 problem during calculation, 181 186 editing a model, 179 loading les, 177 178 process, 10, 27 31 dynamic process, 10, 27 30 equilibrium process, 10, 27, 30 31 process matrix, 29 process rate, 28, 29 process type, 10 processes edit menu, 11, 27, 28 production of dissipation lake compartment, 122 program variable, 10, 12, 15 17, 40 accuracy, 132, 134, 142 accuracy of advective-di usive reactor compartment, 62, 67 bio lm reactor compartment, 51, 56 lake compartment, 110, 123 mixed reactor compartment, 35, 38 river section compartment, 91, 96 saturated soil column compartment, 76, 83 area gradient, 17 lake compartment, 124 attachment velocity of bio lm, 16 bio lm reactor compartment, 58 bio lm thickness, 16 bio lm reactor compartment, 53, 57, 58 Brunt Vaisala frequency, 17 lake compartment, 110, 112, 124 bulk volume, 16

207 bio lm reactor compartment, 57, 58 mixed reactor compartment, 39 buoyancy production of TKE, 17 lake compartment, 124 calculation number, 15, 16 advective link, 128 advective-di usive reactor compartment, 69 bio lm reactor compartment, 58 lake compartment, 113, 124 mixed reactor compartment, 39 river section compartment, 97 saturated soil column compartment, 84 compartment index, 16 advective-di usive reactor compartment, 62, 69 bio lm reactor compartment, 49, 58 lake compartment, 108, 124 mixed reactor compartment, 34, 39 river section compartment, 88, 97 saturated soil column compartment, 75, 84 cross-sectional area, 16 advective-di usive reactor compartment, 62, 69 lake compartment, 100, 124 river section compartment, 96, 97 saturated soil column compartment, 76, 84 density, 17 lake compartment, 124 description, 15 detachment velocity of bio lm, 16 bio lm reactor compartment, 58 discharge, 16 advective link, 128 advective-di usive reactor compartment, 62, 65, 67 69 bio lm reactor compartment, 54, 57, 58 lake compartment, 115, 123, 124 mixed reactor compartment, 37 39 river section compartment, 93, 95 97 saturated soil column compartment, 76, 79, 81, 84 dissipation, 17

208 lake compartment, 100, 123, 124 energy of seiche oscillation, 17 lake compartment, 112, 123, 124 friction slope, 17 river section compartment, 97 growth velocity of bio lm, 16 bio lm reactor compartment, 58 horizontal velocity, 17 lake compartment, 100, 112, 123, 124 interface velocity of bio lm, 16 bio lm reactor compartment, 58 link index, 16 advective link, 127, 128 name, 15 perimeter length, 16 river section compartment, 97 reactor volume, 16 bio lm reactor compartment, 57, 58 mixed reactor compartment, 35, 37 39 reference, 15 shear production of TKE, 17 lake compartment, 124 space coordinate x, 16 advective-di usive reactor compartment, 62, 69 river section compartment, 89, 97 saturated soil column compartment, 76, 84 space coordinate z, 16 bio lm reactor compartment, 50, 58 lake compartment, 109, 113, 124 surface width, 16 river section compartment, 97 time, 16 advective link, 128 advective-di usive reactor compartment, 62, 69 bio lm reactor compartment, 58 lake compartment, 124 mixed reactor compartment, 39 river section compartment, 97 saturated soil column compartment, 76, 84 turbulent kinetic energy, 17 lake compartment, 100, 112, 123, 124

INDEX
unit, 15 water fraction, 16 advective-di usive reactor compartment, 69 bio lm reactor compartment, 57, 58 lake compartment, 124 mixed reactor compartment, 39 river section compartment, 97 saturated soil column compartment, 84 water level elevation, 16 river section compartment, 89, 96, 97 zone index, 16 advective-di usive reactor compartment, 69 bio lm reactor compartment, 49, 58 lake compartment, 108, 124 mixed reactor compartment, 39 river section compartment, 97 saturated soil column compartment, 75, 82, 84 rad formula variable, 24 rate, 28, 29 dynamic process, 29 reactor con ned bio lm reactor compartment, 48, 50 uncon ned bio lm reactor compartment, 48, 50 reactor type bio lm reactor compartment, 50 mixed reactor compartment, 35 reactor volume bio lm reactor compartment, 48, 50 program variable, 57, 58 mixed reactor compartment, 34 program variable, 35, 37 39 program variable, 16 read data pairs real list variable, 21 real list variable, 10, 12, 18 21, 151 active for sensitivity analysis, 21 add data pairs, 21 argument, 18, 20 column number of argument, 21

INDEX
column number of standard deviations, 21 column number of values, 21 data pairs, 18 delete data pairs, 21 description, 20 end row, 21 interpolation, 21 interpolation method, 18 19 linear, 18 smoothing, 19 spline, 18, 19 maximum, 20 minimum, 20 name, 20 read data pairs, 21 replace data pairs, 21 standard deviation, 18, 20, 21 absolute, 18, 20 global, 20 individual, 20, 21 relative, 18, 20 start row, 21 unit, 20 write data pairs, 21 reference program variable, 15 relative accuracy program variable advective-di usive reactor compartment, 67 bio lm reactor compartment, 57 lake compartment, 123 mixed reactor compartment, 38 river section compartment, 96 saturated soil column compartment, 84 state variable, 14, 134 replace argument-value pairs variable list variable, 22 replace data pairs real list variable, 21 resolution advective-di usive reactor compartment, 62 bio lm reactor compartment, 51 lake compartment, 110 river section compartment, 91

209 saturated soil column compartment, 76 resuspension lake compartment, 104 resuspension velocity lake compartment, 121 revert to saved le menu, 7, 8 rigid solid matrix bio lm reactor compartment, 50 river bed elavation river section compartment, 86 river section compartment, 10, 32, 85 97 accuracy of program variables, 91, 96 active for calculation, 91 active processes, 89, 92 active state variables, 89, 91 area, 85, 89 boundary condition, 87, 88 compartment index, 88 critical water level, 91 cross section, 89 description, 88 di usive approximation, 86, 91 discharge, 86 dispersion coe cient, 86, 90, 96 end coordinate, 89 equations, 85 88 friction slope, 86, 90 Darcy-Weisbach, 90 Manning-Strickler, 90 gravitational acceleration, 89 initial conditions, 89, 92 input, 89, 94 lateral, 94, 95 upstream, 94 kinematic approximation, 86, 91 lateral in ow, 87 lateral input, 94, 95 name, 88 normal water level, 91 number of grid points, 91 overview, 85 perimeter length, 89 resolution, 91 river bed elevation, 86 start coordinate, 89 state variable, 86, 89 91, 93, 95

210 dynamic, 86, 95 equilibrium, 86 surface, 86 volume, 86, 95 upstream input, 94 user de nitions, 88 97 water level elevation, 86 width, 89 saturated soil column compartment, 10, 32, 70 84 accuracy of program variables, 76, 83 active for calculation, 76 active processes, 75, 77 active state variables, 75, 76 area, 71, 76 boundary condition, 74 compartment index, 75 cross section, 76 density, 74 description, 75 immobile region, 82 dispersion coe cient, 71, 76, 84 end coordinate, 75 equations, 70 74 exchange coe cient, 82 immobile region, 70, 76, 81, 82 initial conditions, 75, 78 inlet input, 79 input, 75, 79 inlet, 79 lateral, 79, 80 lateral in ow, 72 lateral input, 79, 80 mixed immobile zone, 70, 82, 83 mobile zone, 70 name, 75 immobile region, 81 number of grid points, 76 overview, 70 porosity, 71 mobile zone, 71, 76 resolution, 76 sorption, 74 isotherm, 74 start coordinate, 75 state variable, 71, 75, 76, 79, 81, 83 dynamic, 71, 79, 83 equilibrium, 71 surface, 71 volume, 71, 79, 83 user de nitions, 75 84 volume fraction, 82 mobile zone, 76 zone index, 82 zones, 70

INDEX

save

le menu, 7, 8 save as le menu, 7 scaling plot, 167 screen options, 162, 167 secant algorithm, 151 154 second order, 138 sediment input lake compartment, 113, 117 sediment layer lake compartment, 99, 102, 105, 106 sediment submodel lake compartment, 110, 121 sedimentation lake compartment, 103 sedimentation velocity lake compartment, 100, 119 seiche oscillation lake compartment, 103, 104 sensitivity analysis, 3, 17, 20, 144 150, 164, 165 start, 147, 148 sensitivity function, 144, 165 absolute-absolute, 144, 165 absolute-relative, 145, 165 plot, 161 relative-absolute, 145, 165 relative-relative, 145, 165 sensitivity ranking, 148, 149 shear production of TKE lake compartment, 101 program variable, 124 program variable, 17 short print le format, 8 sign formula variable, 24 simplex algorithm, 151, 153, 154

INDEX
simulation, 2, 3, 138 143 continue, 141 start, 141 sin formula variable, 24 sinh formula variable, 24 smoothing real list variable, 19 variable list variable, 22 solid matrix bio lm reactor compartment, 41 sorption isotherm saturated soil column compartment, 74 saturated soil column compartment, 74 space coordinate x advective-di usive reactor compartment, 62 program variable, 62, 69 program variable, 16 river section compartment program variable, 89, 97 saturated soil column compartment program variable, 76, 84 space coordinate z bio lm reactor compartment, 41 program variable, 50, 58 lake compartment program variable, 109, 113, 124 program variable, 16 spatial discretization, 138 spline interpolation real list variable, 18, 19 variable list variable, 22 sqrt formula variable, 25 stability frequency lake compartment, 109, 110 standard deviation, 146, 151 absolute real list variable, 18, 20 constant variable, 17 global real list variable, 20 individual

211 real list variable, 20, 21 real list variable, 18, 20, 21 relative real list variable, 18, 20 standard error, 144 start parameter estimation, 153, 154 sensitivity analysis, 148 simulation, 141 start coordinate advective-di usive reactor compartment, 62 river section compartment, 89 saturated soil column compartment, 75 start row real list variable, 21 state variable, 10, 12, 14 15 absolute accuracy, 14, 134 accuracy, 14, 132, 134, 142 advective link, 126, 128, 129 advective-di usive reactor compartment, 59, 62, 64, 65, 67 bio lm reactor compartment, 42, 49 54 description, 14 di usive link, 131 dynamic, 14 advective link, 126, 128, 129 advective-di usive reactor compartment, 59, 65, 67 bio lm reactor compartment, 42, 49, 54 di usive link, 131 lake compartment, 100, 108, 114 116, 118 120 mixed reactor compartment, 34, 38 river section compartment, 86, 95 saturated soil column compartment, 71, 79, 83 equilibrium, 14 advective-di usive reactor compartment, 59 mixed reactor compartment, 34 river section compartment, 86 saturated soil column compartment, 71 lake compartment, 100, 108, 110, 112,

212 114 116, 118 121 mixed reactor compartment, 34 38 name, 14 relative accuracy, 14, 134 river section compartment, 86, 89 91, 93, 95 saturated soil column compartment, 71, 75, 76, 79, 81, 83 surface, 14 advective-di usive reactor compartment, 59 mixed reactor compartment, 34 river section compartment, 86 saturated soil column compartment, 71 type, 14 unit, 14 volume, 14 advective link, 126, 128, 129 advective-di usive reactor compartment, 59, 65, 67 bio lm reactor compartment, 42, 49, 54 di usive link, 131 lake compartment, 100, 108, 114 116, 118 120 mixed reactor compartment, 34, 38 river section compartment, 86, 95 saturated soil column compartment, 71, 79, 83 step size calculation sensitivity analysis, 142 simulation, 142 sti , 139 sti y stable, 139, 140 stoichiometric coe cient, 28, 29 stoichiometric matrix, 29 substance ow advective link, 126 bifurcation, 126, 128 substratum bio lm reactor compartment, 41 supported platforms, 4, 8 surface attachment coe cient bio lm reactor compartment, 56 surface detachment coe cient bio lm reactor compartment, 56

INDEX
surface input lake compartment, 113 surface shear lake compartment, 122 surface state variable, 14 surface widt river section compartmenth program variable, 97 surface width program variable, 16 suspended solids bio lm reactor compartment, 41, 50 system edit menu, 11 tan formula variable, 25 tanh formula variable, 25 tasks of AQUASIM, 2 3 temperature lake compartment, 109 thickness sediment layer lake compartment, 99, 121 time advective link program variable, 128 advective-di usive reactor compartment program variable, 62, 69 bio lm reactor compartment program variable, 58 lake compartment program variable, 124 mixed reactor compartment program variable, 39 program variable, 16 river section compartment program variable, 97 saturated soil column compartment program variable, 76, 84 time space curve, 166 t target, 157 title plot, 163 top coordinate

INDEX
lake compartment, 109 troubleshooting, 176 188 turbulence submodel lake compartment, 110, 121 turbulent di usion coe cient lake compartment, 99, 100, 109 turbulent kinetic energy lake compartment, 99, 103, 104 program variable, 100, 112, 123, 124 program variable, 17 type compartment, 10, 32 curve, 164 link, 10 process, 10 state variable, 14 variable, 10, 12, 13 uncertainty analysis, 3, 146 147 uncon ned reactor bio lm reactor compartment, 48, 50 unit constant variable, 17 formula variable, 23 probe variable, 26 program variable, 15 real list variable, 20 state variable, 14 variable list variable, 21 upstream input river section compartment, 94 user group, 176, 187 user interfaces, 3 4 batch version, 3, 4, 8, 173 175 character interface version, 3, 4, 171 172 window interface version, 3, 4 value constant variable, 17 plot, 161 variable, 10, 12 26 constant variable, 10, 12, 17, 21, 151 t target, 157 formula variable, 10, 12, 23 25 probe variable, 10, 12, 26 program variable, 10, 12, 15 17, 40 real list variable, 10, 12, 18 21, 151 state variable, 10, 12, 14 15

213 variable list variable, 10, 12, 21 22 variable list variable, 10, 12, 21 22 add argument-value pairs, 22 argument, 22 delete argument-value pairs, 22 description, 21 name, 21 replace argument-value pairs, 22 unit, 21 variable type, 10, 12, 13 variable volume mixed reactor compartment, 35 variables dissolved bio lm reactor compartment, 56 lake compartment, 120 edit menu, 11 13 particulate bio lm reactor compartment, 55 lake compartment, 118 velocity bio lm matrix bio lm reactor compartment, 43 horizontal, 17, 123, 124 velocity of bio lm matrix bio lm reactor compartment, 43 view menu, 5, 161 170 volume bulk, 16, 39, 41, 57, 58 bio lm reactor compartment, 48 constant mixed reactor compartment, 34, 35 mixed reactor compartment, 34, 35 pore, 41 reactor, 16, 35, 38, 39, 57, 58 bio lm reactor compartment, 48 mixed reactor compartment, 34 variable mixed reactor compartment, 34, 35 volume attachment coe cient bio lm reactor compartment, 56 volume detachment coe cient bio lm reactor compartment, 56 volume ux lake compartment, 121 volume fraction mobile zone

214 saturated soil column compartment, 76 saturated soil column compartment, 82 volume state variable, 14 volumetric in ow mixed reactor compartment, 34 volumetric out ow mixed reactor compartment, 34 water ow advective link, 126 bifurcation, 126, 128 water fraction advective-di usive reactor compartment program variable, 69 bio lm reactor compartment program variable, 57, 58 lake compartment program variable, 124 mixed reactor compartment program variable, 39 program variable, 16 river section compartment program variable, 97 saturated soil column compartment program variable, 84 water level elevation program variable, 16 river section compartment, 86 program variable, 89, 96, 97 width river section compartment, 89 wind excitation lake compartment, 123 molecular lake compartment, 104 window interface version, 3, 4 write data pairs real list variable, 21 zone curve, 166 t target, 157 probe variable, 26 zone index advective-di usive reactor compartment

INDEX
program variable, 69 bio lm reactor compartment program variable, 49, 58 lake compartment program variable, 108, 124 mixed reactor compartment program variable, 39 program variable, 16 river section compartment program variable, 97 saturated soil column compartment, 82 program variable, 75, 82, 84 zones bio lm reactor compartment, 41 lake compartment, 98 saturated soil column compartment, 70