See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/48353344

Expert system for solid/liquid separation equipment selection

Conference Paper · May 1993

Source: OAI

CITATION READS

1 22

2 authors, including:

E.S. Tarleton
Loughborough University
122 PUBLICATIONS   1,488 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Filtration Conference and Exhibition View project

All content following this page was uploaded by E.S. Tarleton on 13 August 2015.

The user has requested enhancement of the downloaded file.

 Loughborough University
Institutional Repository

Expert system for

solid/liquid separation
equipment selection
This item was submitted to Loughborough University's Institutional Repository
by the/an author.

Citation: TARLETON, E.S. and WAKEMAN, E.S., 1993. Expert system

for solid/liquid separation equipment selection. World Filtration Congress 6,
Nagoya, Japan.

Additional Information:

• This is a conference paper.

Metadata Record: https://dspace.lboro.ac.uk/2134/5296

Version: Not specied

Publisher: The Filtration Society / The Filtration Society Japan Association

Please cite the published version.

 This item was submitted to Loughborough’s Institutional Repository
(https://dspace.lboro.ac.uk/) by the author and is made available under the
following Creative Commons Licence conditions.

For the full text of this licence, please go to:

http://creativecommons.org/licenses/by-nc-nd/2.5/
 1

EXPERT SYSTEM FOR SOLID/LIQUID SEPARATION EQUIPMENT SELECTION

E.S. Tarleton1 ([email protected]) and R.J. Wakeman2

1
Department of Chemical Engineering, Loughborough University, Loughborough LE11 3TU, UK.
2
School of Engineering, Exeter University, Exeter EX4 4QF, UK.

ABSTRACT

Commercially available expert systems on their own are inadequate for solid/liquid separation
equipment design and selection. Software which utilises a well-chosen mix of algorithm,
graphics, expert system, and interactive input from the engineer is more effective. An example of
the application of the combined approach to equipment selection is given to demonstrate that
efficient software can save time and enable decisions to be made rapidly. Equipment selection
using the pC-SELECT software is shown.

INTRODUCTION

Artificial intelligence has achieved considerable success in the development of expert systems
which can now be found in many areas of technical endeavour1-7. The area of expert systems
uses techniques such as rapid prototyping, backward and forward chaining and knowledge
representation through production rules, semantic networks and frames to construct man-machine
systems with specialised problem solving expertise. Figure 1 illustrates in a general sense the
components of an expert system; the knowledge base contains the factual and empirical
knowledge of experts in the subject, the inference mechanism simulates the problem-solving
strategy of a human expert and the explanatory interface determines how the expert system
interacts with the user.

Expertise comprises knowledge about a particular topic, understanding of the problems in the
topic, and skill at solving some of these problems. Knowledge in this sense is usually of two sorts,
public and private. Public knowledge includes published definitions, facts and theories of which
textbooks are typically composed. But expertise in a technical area usually involves rather more
than just this public knowledge; human experts generally possess private knowledge that has not
found its way into the published literature. Private knowledge may consist of unpublished
techniques or approaches, or it may take the form of rules of thumb that have come to be called
heuristics. Heuristics enable the human expert to make educated guesses when formalised or
algorithmic solutions are difficult or impossible to determine, to recognise promising approaches to
problems and to deal in the best practicable way with errors or incomplete data.

Filtration and separation technology contains numerous heuristics, which can be evidenced by
consulting industrial reference books such as Solid/Liquid Separation Equipment Scale-Up8. A
majority of industrial process engineers need to possess wide ranging knowledge covering many
unit operations and different types of plant, and rarely have the opportunity to gain in-depth
specialist knowledge of filtration and separation technology. Consequently the large number of
heuristics that have evolved in the technology must be confusing and probably presents to the non-
expert a picture of a technology in disarray. This is compounded by published textbooks which are
frequently remarkably similar to previous ones in their technical content and generally convey the
impression that developments in the understanding and formulation of more rational and reliable
approaches to such factors as equipment selection and design have stagnated since about 1930.
It is also the case that better design methods exist in published literature, but all too often there is
the problem that book authors have not assimilated what is available and they may not have a
good enough grasp of either fundamental or of heuristic knowledge to advance the technology
through the written medium.

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 2

Solid/liquid separation technology, whether it be in areas of selection or design, is best dealt with
by software designed to run interactively, so that the engineer can input data and receive a result
rapidly. The expert system can be used to ensure the correctness of input data as far as this is
possible, and it can utilise interactive graphics facilities to show effects of changes in variables or
to allow the engineer access to calculations to make value judgements where these are peripheral
to the expert system. To be most effective the software must be a well-chosen mix of algorithms,
expert system and input information from the engineer9.

SELECTION CHARTS AND THEIR USE IN PC-SELECT

There exist a number of charts10,11 which can serve as a guide to the initial approach to equipment
selection, the better ones of which consider a variety of possible eventualities and indicate where
decisions must be made. These charts generally have been devised by experts to be fairly
comprehensive and are of values to the solid/liquid separation expert.

They also illustrate the near-impossibility of combining comprehensiveness with useability when so
much interacting is presented in the written form. Purchas8 introduced a general guide for the non-
specialist, which is a valuable aid to one confronted with this confusing and complex area. The
basis of this guide is adopted, suitably extended and adapted, for use in the software pC-SELECT12
which incorporates features of the type discussed in this paper to produce an expert system.

The essential steps in solid/liquid separation were clearly identified and laid out by Purchas8 and
are shown in a simplified form in Figure 2. There are three principal sets of data which
characterise the problem. The first set describes the requirements of the separation in the process
environment, the second set concerns the data obtained from leaf and/or jar tests to characterise
the filtration and/or sedimentation behaviour of the slurry, and the third set constitutes a data bank
which holds information about available solid/liquid separation equipment.

Information associated with more than fifty categories of equipment is held in the pC-SELECT data
bank. These data are analysed by pC-SELECT using public knowledge, heuristics and decision
making techniques such as production rules.

Results from the analyses are data sheets which detail both experimental and calculated results
from tests, and a list of recommended equipment which satisfies the process requirements and
slurry characteristics. The list may be sorted and ranked according to various relative operational
performance criteria or product quality demands. Any other screens displayed during the analysis
also may be printed for inclusion in reports.

EQUIPMENT SELECTION BY PC-SELECT

The following is a hypothetical problem illustrative of the type which the process engineer may face
in the selection of solid/liquid separation equipment. As part of the production cycle a plant needs
to recover the solids constituent in a washed form at the rate of 0.3 te h-1 from an aqueous feed
suspension carrying a solids mass fraction of 0.05%. The nature of the plant indicates that a
continuous type operation will be required; identify a preliminary list of equipment which may be
suitable for this separation. Leaf and jar data have been measured.

pC-SELECT allows entry into its rule based selection procedures at various levels, the level of entry
being determined by the amount and type of information available. In terms of the amount and
type of information available there are two important entry points. The first is the ability to enter for
an initial list of equipment without any form of test data, but with a knowledge of the process. Here
a list can be produced, but against each item in the list will be one or more warning messages
indicating the need to additional data of a particular type. The second important entry point is after

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 3

the analysis of leaf and/or jar test data, the results from which will enable a more reliable and
shorter list to be drawn up than was possible at the previous entry level.

Entry into the expert system with the low level of data specified above (excluding any test data) will
lead to a long list of equipment which might be capable of achieving the separation (Figure 3). The
list is divided into three parts. The top part is a summary of the information fed into the selection
procedure. The second part is a list of the selected equipment (in order of an overall performance
rating in this instance) that indicates for each item, through selection warnings, what further action
should be taken to check the equipment suitability and what limitations the equipment may
possess.

The suitability of the equipment is also related to typical particle size ranges and feed
concentrations. Although the latter information has been used in the selection, the values have
been implied through the use of composite data and it is at this point the engineer can check the
equipment information against values from the process. The third part of Figure 3 is the equipment
listed together with relative performance criteria. These are based on a scale from 0 to 9, with
larger numbers indicating better performance and showing whether the solid is generally
discharged in cake (C) or slurry (S) form. The technical expert will recognise that some items in
Figure 3 are not realistic alternatives. The selection warnings marked against each item in the list
reflect this and clearly it would not be acceptable to rely on Figure 3 for anything other than first
impressions.

Analysis of the leaf and jar test data, using pC-SELECT, provides additional data which the
selection procedure can utilise. The previous list of 20 items is reduced to the 6 listed in Figure 4,
which also shows that inclusion of the additional data has removed most of the selection warnings.

A full analysis using pC-SELECT uses algorithm and graphics software with interactive input from
the engineer to analyse test data followed by expert system software, again with engineer input. A
selection and a report if desired, are completed within minutes and ‘what-if?’ queries are easily
investigated.

It is not the purpose of the example to demonstrate all the facilities of pC-SELECT, but to illustrate
the general method of approach to equipment selection using computers. It is interesting to note
that avoidance of ‘pure’ expert system tools has been recommended12 in circumstances that arise
in evaporation process design where the need for an interface between heuristic based selection
and detailed calculation is similar to that required for solid/liquid separation equipment selection.

The procedures outlined above enable the non-expert to some to rational decisions based on
expert knowledge, without the need to consult an expert in the earlier stages of solving his
problem. This is important in solid/liquid separation, as the expert is often a representative of an
equipment manufacturing company whose job it is to sell a particular type of separator. Taking
filters as an example, many types usually will be capable of carrying out a particular filtration, but
probably only a few general types will be most suited to the task. It is wise to have an insight into
which types these are before consulting and expert.

Software such as pC-SELECT enables the rapid analysis of data and exploration of alternatives, it
puts the engineer in a position to ask more penetrating questions of whichever expert he or she
may consult, and reduces expenditure on unwarranted pilot scale testwork. These advantages are
gained without the need for extensive computer knowledge or high speed/capacity computers.

CONCLUDING REMARKS

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 4

The widespread use of heuristics, the lack of standard approaches to most aspects of design and
the limited information available to the design engineer in texts gives rise to several requirements
in solid/liquid separation. There is a need to

• standardise small scale tests

• rationalise the analysis of the data which come from the tests
• formalise the approaches to process design and scale-up of equipment.

Manufacturers of different equipment types tend to use different heuristic approaches to equipment
sizing, making it very difficult for a user engineer to check that the correct equipment is being
specified for the separation and that the size of the equipment is appropriate.

Computer software could be of considerable assistance not only in the areas identified above, but
also in design and scale-up14. Design procedures and know-how in many other branches of
engineering, for example in heat transfer, are documented widely and simulation and model
equations are freely available and well developed.

This is not the case in solid/liquid separation where design codes and other useful information are
effectively non-existent. Starting from the present overall position of the technology and the
current state of design procedures it would be unrealistic to expect to formalise all aspects of
design or scale-up, not least because of the wealth of rule-of-thumb knowledge which is not
available to the technologists. The main problem is that this information is available to a limited
number of technology experts, but to few others.

The use of an integrated software approach utilising a well-chosen mix of algorithm, graphics,
expert system and input information from the engineer could overcome the problem that probably
stems from deficiencies of communication between experts and non-experts, and between
academics and industrialists. To bring all of the technology up to such a level that it is equally
available to everyone that wishes to use it would be an enormous task, but perhaps it is a goal we
should aim to achieve.

REFERENCES

1. Shortliffe E.H., 1976, MYCIN: Computer-Based Medical Consultation, Elsevier, New York.

2. van Melle W., Shortliffe E.H. and Buchanan R.G., 1981, EMYCIN: A domain-independent
system that aids in constructing knowledge-based consultation programs, Machine Intelligence,
Infotech State of the Art Report 9, No. 3.

3. XCON: Customer-specific configuration of computer systems, Digital Equipment Corporation.

4. NAVEX: Monitoring the navigation of the space shuttle, NASA.

5. PLAN-POWER: Advising banks, insurance and investment companies, Applied Expert

Systems.

6. Nebendahl D., 1988, Expert Systems: Introduction to the Technology and Applications, Wiley.

7. Pressman R.S., 1988, Software Engineering: A Practitioners Approach, 2nd Edn., McGraw-Hill.

8. Purchas D.B. and Wakeman R.J., 1986, Solid/Liquid Separation Equipment Scale-up, Uplands
Press & Filtration Specialists.

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 5

9. Wakeman R.J. and Tarleton E.S., 1991, Solid/liquid separation equipment simulation and
design – An expert systems approach, Filtration and Separation, 28(4), 274.

10. Davies E., 1965, Selection of equipment for solid/liquid separation, Trans. IChemE, 43, T256.

11. Tiller F.M., 1974, Bench scale design of SLS systems, Chemical Engineering, S1(9).

12. Tarleton E.S. and Wakeman R.J., 1991, Solid/Liquid Separation Equipment Simulation &
Design: pC-SELECT – Personal computer software for the analysis of filtration and
sedimentation test data and the selection of solid/liquid separation equipment, Separations
Technology Associates.

13. Bell G., 1989, Application of expert systems in design of evaporation processes, Directory of
A.I. related engineering projects, Artificial Intelligence Applications Institute, University of
Edinburgh, November.

14. Wakeman R.J. and Tarleton E.S., 1990, Modelling, simulation and process design of the filter
cycle, Filtration and Separation, 27(6), 412-419.

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 6

TABLES AND FIGURES

Figure 1: The basic elements of an expert system.

Figure 2: The steps involved in the selection of separation equipment.

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 7

Solid/Liquid Separation Equipment Simulation & Design, PC-SELECT

DATA SHEET FOR EQUIPMENT SELECTION

Specifications

Scale:- medium (1 m3/hr)

Duty Operation:- continuous
Objective:- washed solids recovery

Rate:- not specified

Settling Overflow clarity:- not specified
Sludge proportion:- not specified

Filtration Cake growth rate:- not specified

Selected Selection Particle Feed

equipment description warnings size (μm) conc. (% v/v)

Horizontal belt, pan or table filter 2,3 20-80,000 3-40

Bottom fed drum filter, knife dis. 2,3 1-200 3-30
Bottom fed drum filter, roll dis. 2,3 1-50 3-30
Bottom fed drum filter, string dis. 2,3 1-70 3-30
Continuous pressure filter 2,3 1-100 0.01-30
Variable volume filter 1h,2,3 1-200 0.1-25
Top fed drum filter │ │ │
Screen or sieve bend classifier │ │ │
Low shear crossflow microfilter │ │ │
Gravity thickener │ │ │
Drum, grid or belt magnetic filter │ │ │
High shear crossflow filter │ │ │
High gradient magnetic filter │ │ │
Scroll (decanter) centrifuge │ │ │
Hydraulic classifier │ │ │
Mechanical classifier │ │ │
Low shear crossflow microfilter 1h,2,3 │ <10

Selected F3: F4: F5: F6: F7:

equipment description index index index index index

Horizontal belt, pan or table filter 7C 7 9 8 31

Bottom fed drum filter, knife dis. 6C 7 7 8 28
Bottom fed drum filter, roll dis. 6C 7 7 8 28
Bottom fed drum filter, string dis. 6C 7 7 8 28
Continuous pressure filter 6C 7 6 7 26
Variable volume filter 8C 7 4 7 26
Top fed drum filter │ │ │ │ 22
Screen or sieve bend classifier │ │ │ │ 18
Low shear crossflow microfilter │ │ │ │ 18
Gravity thickener │ │ │ │ 17
Drum, grid or belt magnetic filter │ │ │ │ 17
High shear crossflow filter │ │ │ │ 16
High gradient magnetic filter │ │ │ │ 15
Scroll (decanter) centrifuge │ │ │ │ 14

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 8

Hydraulic classifier │ │ │ │ 14
Mechanical classifier │ │ │ │ 13
Low shear crossflow microfilter 1S 9 2 0 12

F3 index:- Solid product dryness

F4 index:- Liquid product clarity
F5 index:- Washing performance
F6 index.- Crystal breakage
F7 index:- Overall performance

Equipment listed in order of overall performance rating

Figure 3: Equipment selection with a low level of data.

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.
 9

Solid/Liquid Separation Equipment Simulation & Design, PC-SELECT

DATA SHEET FOR EQUIPMENT SELECTION

Specifications

Scale:- medium (1 m3/hr)

Duty Operation:- continuous
Objective:- washed solids recovery

Rate:- (0.1-5 cm/s)

Settling Overflow clarity:- good
Sludge proportion:- medium (2-20% vol)

Filtration Cake growth rate:- fast (greater than 1 cm/min)

Selected Selection Particle Feed

equipment description warnings size (μm) conc. (% v/v)

Horizontal belt, pan or table filter 2,3 20-80,000 3-40

Screen or sieve bend classifier 1cBG 45-100,000 20-40
Gravity thickener
Scroll (decanter) centrifuge
Hydraulic classifier None 50-2,000 4-40
Mechanical classifier 1h 100-3,000 4-40

Selected F3: F4: F5: F6: F7:

equipment description index index index index index

Horizontal belt, pan or table filter 7C 7 9 8 31

Screen or sieve bend classifier 5C 5 4 4 18
Gravity thickener │ │ │ │ │
Scroll (decanter) centrifuge │ │ │ │ │
Hydraulic classifier 3C 3 3 5 14
Mechanical classifier 4C 3 2 4 13

F3 index:- Solid product dryness

F4 index:- Liquid product clarity
F5 index:- Washing performance
F6 index.- Crystal breakage
F7 index:- Overall performance

Equipment listed in order of overall performance rating

Figure 4: Equipment selection with a high level of data.

th
Cite paper as: Tarleton E.S. and Wakeman R.J., 1993, Expert system for solid/liquid separation equipment selection, Proc. 6 World
Filtration Congress, pp.919-925, Japanese Society Chemical Engineers, Nagoya, Japan.

View publication stats