Ninth International IBPSA Conference

Montréal, Canada
August 15-18, 2005

INTEGRATED HEAT, AIR AND MOISTURE

MODELING TOOLKIT IN MATLAB

A.W.M. (Jos) van Schijndel and J.L.M. (Jan) Hensen

Technische Universteit Eindhoven
Department of Building and Architecture, Building Physics and Systems (BPS)
VRT 6.29, P.O.Box 513 5600 MB Eindhoven, the Netherlands
email: [email protected]

and validation results of the developed HAM models.

ABSTRACT Third, Section 6 presents applications of integrated
This paper gives an overview of recent developments HAM models. Finally, in Section 7, the key
and results of a new integrated heat, air and moisture questions will be revisited and answered.
(HAM) modeling toolkit in Matlab named HAMLab.
The recent developments include integration of a 2. RESEARCH METHODS
whole building model with building systems and The research consisted of four parts. First, a review
controllers, 2D/3D HAM transport in constructions of integrated HAM models was carried out. The aim
and 2D airflow respectively. The results include a of this review was to obtain a group of recent
short review on HAM models, a motivation of the simulation tools, based on the criteria provided in
selected simulation environment Matlab and Section 3. The goal of the review is to specify
extensive verification/ validation results. common modeling problems that are to be solved and
Furthermore, the integration capabilities are to have some reference tools for the discussion of the
demonstrated by applications. It is concluded that the benefits and drawbacks. Second, a single
simulation environment HAMLab is capable of computational environment was selected, based on a
solving a large scale of integrated HAM models. criterion presented in Section 3. Third, a range of
Limitations, benefits and drawbacks are discussed. HAM models were modeled simulated and validated
using the selected software. Fourth, case studies were
1.INTRODUCTION performed concerning integrated HAM models for
It is widely accepted that simulation can have a possible solutions of the above mentioned common
major impact on the design and evaluation of modeling problems.
building and systems performances. Also, the
modeling and simulation of whole building heat, air 3. REVIEW
and moisture (further-on called HAM) responses in The first Section summarizes recent single simulation
relation to human comfort, energy and durability are environment HAM models. The second Section
relevant. IEA Annex 41 will focus on a holistic summarizes common modeling problems to be
approach of HAM transfer between the outside, the solved within a single simulation environment.
enclosure, the indoor air and the heating, ventilation
and air-conditioning (HVAC) systems. Therefore 3.1 HAM models
integrated HAM models capable of covering all these
In 2005, 14 different tools were used in an Annex 41
issues are sought-after. There is no single simulation
common exercise about simulating the dynamic
tool that covers all of the issues [Augenbroe 2002].
interaction between the indoor climate of a room and
One option is the external coupling of tools [Hensen
the HAM response of the enclosure. All tools model
et al. 2004; Djunaedy 2005]. Recent developments in
the indoor air and the enclosure. Six HAM models
general modeling and simulation tools give rise to
are stand-alone simulation tools and have promising
another option: the use of a single computational
capabilities for simulating HVAC systems: Bsim,
environment. What are the limitations if we use state-
IBPT, IDA-ICE, TRNSYS, EnergyPlus, and
of-art computational software? What are the benefits
HAMLab. The latter is presented in this paper. A
and drawbacks?
complete overview will be published with the final
The aim of this paper is to answer these key reports, expected in 2008. In [Schwab et al. 2004]
questions. The outline is as follows: Section 2 265 different energy related software tools at the
presents the methodology. Section 3 presents the Energy Tools website [U.S. Department of Energy
review results. Sections 4 through 6 present all U.S, 2005] are reviewed. In this study a search was
modeling and simulation results. First, Section 4 performed in order to determine what kind of
discusses the selection of the computation simulation simulations each tool could perform. After this, the
environment. Second, Section 5 gives verification list was narrowed down to 11 programs that might

- 1107 -
simulate whole building HAM transfer and energy (1) a whole building model for the simulation
consumption. From this group three tools met all of the indoor climate and energy amounts;
criteria: moisture storage in building materials,
(2) a high spatial resolution tool, for the
calculate indoor climate, moisture exchange in
simulation of HAM responses of building
HVAC system, access to source code. The three tools
constructions and internal/external airflow,
are Bsim, TRNSYS and EnergyPlus. These tools are
for example a 2D and 3D FEM PDE solver;
already mentioned. In [Gough 1999] tools were
reviewed with the focus on new techniques for (3) a high time resolution tool, for the accurate
building and HVAC system modeling. In this study simulation of HVAC systems and
four new simulation techniques were investigated. controllers, for example an ODE solver.
The equation-based method is most relevant for this
These three items will be presented in the following
work. The other three techniques: modal, stochastic
Sections, including already published and new
and neural networks are not based on physical
preliminary verification and validation data.
parameters and are therefore less suitable for solving
design problems. The next three equation based
4. MODEL IMPLEMENTATION
method techniques are already included in one or
more tools of the previous Section: Neutral Model MatLab/SimuLink including FemLab has been
Format and IDA solver are included in IDA-ICE and selected as the simulation environment. Justification
TRNSYS. The other 11 mentioned equation based of this crucial choice is explained below.
method techniques are simulation environments with
limited HAM models for building simulation. 4.1 Whole building model
[Clarke 2001] describes the approach to domain The model is based on ELAN, a computer model for
coupling as implemented with within the ESP-r building energy design. [Wit et al. 1988]. The ELAN
system. Some recent Matlab based tools are: model, together with an analog hygric model has
SIMBAD provides HVAC models and related been implemented in a Building Physics Toolbox in
utilities to perform dynamic simulation of HVAC MatLab (WaVo) [Schijndel and de Wit 1999]. A
plants. Although moisture transport capabilities are major recent improvement is the development of a
limited, this toolbox is useful as a starting point for model in SimuLink (HAMBase) [Wit 2004;
the development of HAM-HVAC models. IBPT is a HAMLab 2005]. The implemented numerical model
tool that focuses on 1D HAM transport in building consists of a continuous part, solved with a variable
constructions. It originates from the HAMSTAD time step and a discrete part, solved with a time step
project [Hagentoft et al. 2002]. of one hour. For the HVAC system and the room
response of indoor climatic variations, a continuous
3.2 Common modeling problems
model is used. For the external climate a discrete
Note that the aim of review is not to obtain an model is used. The main advantages of this numeric
exhaustive list of HAM models, but to obtain a hybrid approach are: (1) the dynamics of the building
representative group of HAM models that will be systems where small time scales play an important
used in this Section for identifying common role (for example on/off switching) are accurately
modeling problems and in Section 5 for comparison simulated; (2) the model becomes time efficient as
purposes. All HAM tools mentioned in previous the discrete part uses 1-hour time steps; (3) the
Sections, face at least one limitation that cannot be moisture (vapor) transport model of WaVo is also
solved by the tool itself. Either a problem occurs at included. With this feature, the (de-) humidification
the integration of HVAC systems models into whole of HVAC systems can also be simulated.
building models, or a problem occurs at the
integration of 2D and 3D geometry based models Already published verification and validation data
(for example airflow and HAM response of In [Schijndel 2003c], the heat transport part of
constructions) into whole building models. The first HAMBase is verified with a standard test method
problem is caused by the difference in time constants [ASHRAE 2001].
between HVAC components and controllers (order
of seconds) and building response (order of hours). In [Schellen 2002], the HAMBase model has been
This can cause inaccurate results and long simulation subjected to a validation study. The measured and
duration times; see also [Gouda et al. 2003, Felsman simulated air temperature and Rh, during one month
et al. 2002]. The second problem is caused by the in a church, are compared.
lack of lumped parameter tools that include internal
2D, 3D finite element method (FEM) capabilities;
see also [Sahlin et al. 2004]. In order to be able to New preliminary verification data
solve all above-mentioned problems in a single [Schijndel 2004] presents preliminary results for the
simulation environment, the toolkit should be able to Annex 41 exercise on the dynamic interaction of the
integrate: indoor climate of a room and the HAM response of

- 1108 -
the enclosure. As mentioned before, the final results placed with its lower surface about 1 cm under water.
of this Annex will be published in 2008. The simulated moisture uptake profiles agree well
with the measurements of [Kunzel et al. 1996].
The geometry is identical to the one in the ASHRAE
standard test [ASHRAE 2001]. Further test In [Schijndel 2003a] a CFD model in FemLab for
conditions are as follows: (1) there is only one 1D indoor airflow is verified with the results of [Sinha
construction type with linear properties; (2) the 2000]. The problem is modeled by 2D
exposure is completely isothermal; (3) the outside incompressible flow using the Boussinesq
relative humidity (Rh) is 30%; (4) there are no approximation with constant properties for the
windows; (5) the internal gains equal 500g/hour; (6) Reynolds (Re) and Grasshof (Gr) numbers. In [Sinha
the ventilation rate is 0.5 ach. The exercise seeks to 2000] the problem is solved and validated with
provide the analytical solution as well as the measurements for several combinations of Re and
simulation results of the model. The model solution Gr.
is obtained by applying HAMBase. The analytical
New preliminary verification data
solution is obtained by integration of a FemLab
model of the construction into a first order model of In [Schijndel 2005a] a k-epsilon model in FemLab
the room modeled in SimuLink. In figure 1 the concerning outdoor airflow (wind) is verified by
corresponding SimuLink model is shown. All blocks another exercise of the mentioned IEA Annex 41
are standard blocks from the library. The FemLab exercise, on boundary conditions. The exercise seeks
model block of figure 1 is based on the following to provide for several building geometries: (1)
PDE: simulated wind velocity profiles around buildings.
∂w They are calculated using the standard k-epsilon
= ∇ ⋅ (D w (w)∇w) (1) turbulence model of FemLab. The turbulence
∂t
parameters are presented in Table I
where w is moisture content, t is time and Dw is Table I. The k-epsilon turbulence parameters
diffusivity. In figure 2 the analytical solution and the
Name Variable Value
model solution are shown. Preliminary results of all
models attending the exercise show that both the Cmu Cmu_ke 0.09
analytical solution obtained using FemLab and the Cd1 Cd1_ke 0.1256
HAMBase solution are satisfactory.
Cd2 Cd2_ke 1.92
4.2 High spatial resolution tool
sigmak sigmak_ke 0.9
HAM responses of building constructions and sigmad sigmad_ke 1.3
internal/external airflow are modeled and simulated
with FemLab. This is an environment for modeling Figure 3 shows the simulated wind profile around the
scientific and engineering applications based on lower building.
partial differential equations (PDE). (2) simulated raindrop trajectories. They are obtained
Already published verification and validation data from the equation of motion of a raindrop, moving in
a wind-flow field characterized by a velocity vector v
Two FemLab models are developed as part of the is:
benchmark study, concerning 1D HAM transport in
constructions: HAMSTAD [Hagentoft 2002]. One dr d 2r (2)
FemLab model shows an excellent agreement with g + f (Cd , Re) ⋅ (v − )= 2
dt dt
the solution of the analytical case. A second FemLab
model, simulating the insulated roof case, shows a where g is gravity, f is a function dependent on Re
good consensus with the other solutions. (Reynolds number) and Cd (drag coefficient), v is
wind velocity, t is time and r is the position of
In [Schijndel 2003a] a validation of the simulated raindrop. Equation 2 is solved with the ODE solver
water absorption in an initially dry brick cylinder of Matlab providing raindrop trajectories for the low
(length 24 mm) is presented. All sides except the building case. The result is shown in figure 4.
bottom are sealed. One side is submerged in water at
t=0 sec. The measured 1D water absorption process (3) simulated wind-driving-rain coefficients on the
of several brick materials based on [Brocken 1998] building facades are obtained from the locations of
are successfully simulated. raindrops hitting the building façade.

In [Schijndel 2003a] a validation of a combined heat The preliminary results are in good agreement with
and moisture transport for a 2D geometry is the results obtained by other models (Fluent)
presented. The problem is based on a 2D-problem
presented in [Kunzel et al. 1996]: a brick specimen,
initially dry and with two of the flanks sealed, is

- 1109 -
4.3 High time resolution tool construction details within reasonable simulation
time: 1 year simulation, takes 12 hours of
HVAC systems and controllers are modeled and
computation time.
simulated using SimuLink. This is a platform for the
simulation of dynamic systems. 6. CONCLUSIONS
Already published verification and validation data
6.1 Limitations, drawbacks and benefits
In [Schijndel 2003c], results of ODE based models
of a heat pump, an energy roof and a thermal energy It may be concluded that the simulation environment
storage (TES) are compared with measurements. HAMLab is capable of solving a large range of
integrated HAM problems. Furthermore, it is
New preliminary verification data promising in solving common modeling problems of
In [Schijndel 2005b], a room of a famous museum in other HAM models such as problems caused by the
the Netherlands is simulated, using HAMBase. The difference in time constants between HVAC
HVAC system and controller are modeled by components and building response and problems
standard blocks in SimuLink. This model is caused by the lack of lumped parameter tools that
presented in figure 5. After completing the models include internal 2D, 3D FEM capabilities. However,
with the necessary input data, the integrated HAM there are some limitations: (1) some specific solvers
model in SimuLink, consisting of the room, HVAC such as time-dependent k-epsilon turbulence solvers
system and controller was validated by are not available yet. This means that, for example,
measurements, see figure 6. The model was time dependent 3D airflow around buildings cannot
successfully used to design a new HVAC controller be solved; (2) although it is possible to construct a
strategy. full 3D integrated HAM model of the indoor air and
all constructions in FemLab, the simulation duration
5. APPLICATIONS time would probably be far too long to be of any
practical use; (3) the modeling of radiation is just
Already published applications recently available in FemLab and is not included in
this research.
[Schijndel 2003d] shows the integration of an airflow
model, using FemLab, into a controller model of Possible drawbacks are: (1) the software package
SimuLink. In [Schijndel 2003b] the set point MatLab itself and basic knowledge of Matlab are
operation of the HVAC system of a church is required to use the models; (2) at this moment
optimized by the use of MatLab, FemLab and HAMLab is research tool. Therefore it lacks facilities
SimuLink models. The goal was the preservation of for design-oriented users such as user-friendly
the monumental church organ. Evaluation after a interfaces and user guides. (3) although state-of-art
year showed that the operation strategy functioned solvers are present, the simulation of FEM based
satisfactory. integrated models can easily become very
computation time consuming.
New application
The main benefits of HAMLab are: (1) it takes
[Schijndel 2004] demonstrates the integration of a
advantage of the facilities of the well maintained
3D HAM model of a roof/wall construction into a
Matlab/SimuLink and FemLab simulation
whole building model. The geometry of the building
environment such as the state-of-art ODE/PDE
is identical to one in [ASHRAE 2001]. HAMBase is
solvers, controllers library, graphical capabilities etc;
used as whole building model. Figure 7 shows the
(2) all presented HAMLab models in this paper are
input/output structure of the integrated model. The
public domain; (3) although not explicitly shown in
output (air temperature and Rh) of this model and
this paper, compared to other HAM models, it is
climate data are used as input for the FemLab model
relative easy to integrate new models that are based
of a 3D roof/wall construction. The output of the
on ODEs and/or PDEs. (4) the simulation facilitates
FemLab model (heat and moisture flow) is connected
open source modeling and if desired, models can be
to the input of the building model. The FemLab
compiled into stand-alone applications.
models solves the following PDE:
6.2 Current research projects
∂w
= ∇ ⋅ (D w (w)∇w + D F (T, w)∇T)
∂t (3) At this moment we are working on: (1)
∂T implementation of web based models for designers
= ∇ ⋅ (D T (w)∇T) using the Matlab web server Toolbox to meet
∂t
drawbacks 1 and 2; (2) speed up of computation
where T is temperature, Dw, DF DT are diffusivities. duration times by the use of system identification
Figures 8 and 9 demonstrate the 3D temperature and modeling approximations of complex models to meet
vapor distributions at a specific time. The integrated drawback 3. (3) development of a full 2D integrated
model is capable of simulating the indoor climate and HAM model for indoor air, building constructions

- 1110 -
and outdoor wind profiles & rain distributions. This Schijndel A.W.M. van 2003a. Modeling and solving
project is a pilot study for a full 3D model and fits building physics problems with FemLab,
into the IEA Annex 41. Building and Environment 38, pp 319-327
Schijndel A.W.M. van, Schellen, H.L., Neilen D.
REFERENCES & TOOLS 2003b. Optimal setpoint operation of the climate
ASHRAE 2001. Standard method of test for the control of a monumental church, 2ND
evaluation of building energy analysis computer international conf. on research in building
programs, standard 140-2001. physics, Leuven, Sept. 14-18 pp. 777-784
Augenbroe G. 2002. Trends in building simulation, Schijndel A.W.M. van, Wit M.H. de 2003c.
Building and Environment 37, pp. 891-902 Advanced simulation of building systems and
Brocken H. 1998. Moisture transport in brick control with SimuLink, 8TH IBPSA Conference
masonry, Ph.D. thesis, Technische Universiteit Eindhoven. pp. 1185-1192
Eindhoven, December 1998 Schijndel, A.W.M. van 2003d. Integrated building
Clarke J.A. 2001. Domain integration in building physics simulations with FemLab/SimuLink
simulation, Building and Environment 33, pp. /Matlab, 8TH Int. IBPSA Conference, Eindhoven,
303-308 pp. 1177-1184
Djunaedy E. 2005. External coupling between Schijndel A.W.M. van 2004. Advanced use of
building energy simulation and computational HAMLab on CE1, Annex 41 preliminary paper
fluid dynamics. PhD thesis, Technische TUE Oct 2004 Paper A41-T1-NL-04-5
Universiteit Eindhoven. Schijndel A.W.M. van 2005a. Application of
Felsmann C., Knabe G., Werdin H. 2002. Simulation HAMLab on Subtask 3, Annex 41 preliminary
of domestic heating systems by integration of paper TUE Oct 2004 Paper A41-T1-NL-05-1
TRNSYS in a MATLAB/SIMULINK model, Schijndel A.W.M. van, Schellen H.L. 2005b.
Proc. 6TH Int. Conference System Simulation in Application of an integrated indoor climate &
Buildings Liege, December 16-18 HVAC model for the indoor climate design of a
Gouda M.M., Underwood C.P., Danaher S. 2003. museum, submitted for the 7TH symposium on
Modelling the robustness properties of HVAC building physics in the Nordic Countries, 13-15
plant under feedback control, Building Serv. Eng. June 2005, Iceland
Res. Techn. 24,4 pp. 271-280 Schwab M., Simonson C.J. 2004. Review of building
Gough M.A. 1999. A review of new techniques in energy simulation tools that include moisture
building energy and environmental modelling, storage in building materials and HVAC systems,
BRE Contract No. BREA-42. Garston, Watford, Annex 41 preliminary paper A41-T2-C-04-6
UK: Building Research Establishment. Sinha S.L. 2000. Numerical simulation of two-
Hagentoft C.E. and 13 other authors. 2002. dimensional room air flow with and without
HAMSTAD – Final Report,, Report R-02:8, buoyancy, Energy and Buildings 32 pp121-129
Chalmers Univ. of Tech., Sweden U.S. Department of Energy, 2005
Hensen J., Djunaedy E., Radosevic M., Yahiaoui A. http://www.eere.energy.gov/
2004. Building performance simulation for better Wit M.H. de ,H.H. Driessen 1988, ELAN A Computer
design: some issues and solutions, 21TH PLEA Model for Building Energy Design. Building and
Conference Eindhoven. Environment, Vol.23,4, pp. 285-289
IEA Annex 41 2005. Wit M.H. de 2004. WAVO, A model for the simula-
http://www.ecbcs.org/annexes/annex41.htm tion of thermal and hygric performance of buil-
Kunzel H.M., 1996. IEA Annex 24 HAMTIE, Final ding and systems. http://sts.bwk.tue.nl/hamlab/
Report – Task 1
Sahlin P. Eriksson L., Grozman P., Johnson H., BSIM 2005. http://www.bsim.dk
Shapovalov A. Vuolle M. 2004. Whole-building Energyplus2005. http://www.eere.energy.gov
simulation with symbolic DAE equations and ESP-r http://www.esru.strath.ac.uk/
general purpose solvers. Building and FemLab 2005. http://www.comsol.com/
Environment 39, pp. 949-958. Fluent 2005 http://www.fluent.com/
Schellen H.L. 2002. Heating Monumental Churches, HAMLab, 2005. http://sts.bwk.tue.nl/hamlab/
Indoor climate and preservation of cultural IBPT 2005 http://www.ibpt.org
heritage, PhD thesis, Technische Universiteit IDA ICE, 2005. http://www.equa.se/ice/intro.html
Eindhoven. Matlab 2005. http://www.mathworks.com/
Schijndel A.W.M. van, Wit M.H., de 1999. A SIMBAD 2005. http://software.cstb.fr
Building Physics Toolbox in MatLab, 5th TRNSYS, 2005. http://sel.me.wisc.edu/trnsys/
Building Physics in the Nordic Countries
Conference, Goteborg, August 24-26, 1999, pp.
81-88

- 1111 -
Figure 1. The SimuLink model of the analytical solution

Figure 2. Simulated Rh of the indoor climate of HAMBase and the analytical solution using FemLab/SimuLink

Figure 3. The wind velocity profile: absolute value of the velocity [m/s] for each coordinate x, y [m].

- 1112 -
Figure 4. The raindrop trajectories

Figure 5. The HVAC and controller model in SimuLink

Figure 6 Simulated and measured room air temperatures (top) and Rh (bottom) for a week

- 1113 -
Figure 7 Input/ouput structure of the SimLink model.

Figure 8. The temperature distribution: temperature [oC] for each coordinate x, y, z [m].

Figure 9. The vapor pressure distribution: vapor pressure [Pa] for each coordinate x, y, z [m].

- 1114 -