You are on page 1of 8

ICCBT2008

A Review on Expert Systems for Concrete Mix Design

M. Abdullahi*, Universiti Tenaga Nasional, MALAYSIA


H. M. A. Al-Mattarneh, Universiti Tenaga Nasional, MALAYSIA
A. H. Hassan, Universiti Tenaga Nasional, MALAYSIA
M. H. Abu Hassan, Universiti Tenaga Nasional, MALAYSIA
B. S. Mohammed, Universiti Tenaga Nasional, MALAYSIA

ABSTRACT

Expert systems for concrete mix design are presented. These systems were developed to
reduce the time and material wasted in the process of determining optimum mixture
proportions using conventional methods. Mix design codes are derived from data obtained
from experience with concrete materials in a particular region or country and may not be
applicable to other parts of the world. Expert systems with its interactive, user-interface tend
to alleviate these problems. The knowledge bases were derived from design codes, national
standards, technical literatures and interview or survey of experts. The expert systems were
developed using expert system shells such as EXSYS Professional, level5, level5 object, and
kappa-PC. Previous studies indicate that knowledge-based approach for the proportioning of
concrete mixes is quite effective and efficient. However, the general consensus among expert
system developers is that the outcomes of their work are prototype. The systems must be
nurtured and maintained and therefore can not claim completeness. For instance, some codes
used for expert system development have been updated in literature to be in line with current
research findings. In this wise, some of these systems may no longer be applicable to the mix
design domain as intended except where the knowledge base is also updated.

Keywords: Concrete, Mix design, Expert systems, Knowledge-based.

*Correspondence Authr: Abdullahi Mohammed Civil Engineering Department, Universiti Tenaga Nasional,
Malaysia. Tel: +60173511985. E-mail: abdulapai@yahoo.com

ICCBT 2008 - A - (21) - pp231-238


A Review on Expert Systems for Concrete Mix Design

1. INTRODUCTION

Personal computers have had significant impact on the business environment. Spreadsheets,
word processors, and databases provided the software tools for manipulating data and
information. Expert systems were not well known until 1980s. While conventional software
process data, expert systems process knowledge [1]. Over the past few decades, expert
systems have been employed in mix design of concrete [2, 3, 4, 5, 6]. The main purpose of the
expert systems was to emulate human experts in concrete mix design by attempting to reason
with knowledge. Concrete mix design is the process of selecting suitable ingredients of
concrete and determining their relative quantities with the objective of producing the most
economical concrete while retaining the specified minimum properties such as strength,
durability, and consistence [3].

The easiest way to do mix design is to use proportions established for similar concrete using
the same materials. Previous experience with concrete material is also of immense advantage
in concrete mixture proportioning and adjustment. Where these are lacking the only possible
option is to proportion the ingredient by a trial and error process. Several methods and codes
are available to serve as guide for mix design of normal concrete [7, 8], high performance
concrete [9, 10, 11, 12], and lightweight concrete [13, 14). However, these are just guide to
arrive at first trial mix. Optimum mix proportions are obtained through testing of trial mixes
and making adjustment accordingly [15, 16, 17]. This is because these codes were developed
based on experience with materials in certain parts of the world and may not be applicable to
mix design in other parts of the world [16]. Also, these codes do not address all issues
regarding concrete mix design such as admixtures, transportation, and temperature effect [4].
These methods of design can be followed and put in a computer program. These programs
have the capability of manipulating the numerical data to obtain mixture proportions of
concrete. [17]. These programs are not interactive and lack heuristic reasoning. They may not
be of much assistance in providing guidance in concrete mixture proportioning.

The rapid increase in construction work couple with a renewal of interest in wide range of
concrete performance has created more challenges in the construction industry [18, 19, 20,
21]. Concrete with several properties may be desired such as high workability, medium
workability, high strength, lightweight, insulation etc. These challenges cannot be met by
designing mix proportions based on existing codes and methods of concrete mix design.
Concrete mix design and adjustment is complex and the correct way to perform this can be
achieved with expert’s advice and experience [2]. Experts in concrete mix design are limited
in number and can be very expensive on demand. The manual process of mix design using
design codes is also tedious and time consuming. Expert systems have the capability of
reducing the ambiguity associated with concrete mix design. The expert’s knowledge is
captured mostly in a window, interactive environment. Expert systems only serve to
compliment the engineer’s knowledge. They can also replace experts where experience
personnel are not available to give guidance in mix proportioning and adjustment. [23, 4].

2. EXPERT SYSTEMS

An expert system is a computer system which emulates the decision-making ability of a


human expert [Giarratano and Riley, 1998]. An expert on the other hand is an individual who

232 ICCBT 2008 - A - (21) - pp231-238


M. Abdullahi & H. M. A. Al-Mattarneh

possesses a superior understanding of the problem. Through experience, the expert develops
the skills that enable him or her to effectively and efficiently solve the problem. Expert system
developers then assemble this expert in to expert system. The attributes of an expert that is
being modelled are the expert’s knowledge and reasoning. This is achieved using the
knowledge base and inference engine.

Expert system structure is composed of the followings: knowledge base, working memory,
inference engine, explanation facility and interface [23]. The knowledge base contains the
domain knowledge. The working memory contains the problem facts that are discovered
during the session. The inference engine is a processor that matches the facts contained in the
working memory with the domain knowledge contained in the knowledge base to draw
conclusions about the problem. The explanation facility provides an explanation to the user
about why it is asking a question and how it reached some conclusion. The user interface
stands between the user and the application program, and enhances the communication and
interaction between the two [24]. It is the channel where user can interact with the internal
process of the application program and vice versa. The expert system structure is shown in
Fig.1.

Knowledge Base
Domain Knowledge

Inference User
Engine Case Facts
Conclusions
Working Memory
Case/Inferred Facts
Conclusions

Fig.1: Expert system structure (Source: Durkin, 1994)

3. ADVANTAGES AND DISADVANTAGES OF EXPERT SYSTEMS

Despite the success in the application of expert systems, it has its associated disadvantages.
To appreciate this better, human expertise and expert systems are compared and their
advantages and disadvantages highlighted as follows [24, 25]:

3.1. Advantages of Expert Systems

1. Human expertise is perishable. For example, humans may change jobs, they can become ill,
and so on. Expert system, on the other hand, is permanent.
2. Human expertise is not always consistent. Human experts can have “off-days”, busy
schedules, and so on, all of which could adversely affect their normal performance. Expert
systems are not influenced by outside pressures and produce consistent results regardless of
the complexity of the problem.
3. Human expertise is difficult to transfer: a human being cannot be in two places at the same
time. Computer expertise is relatively easy to transfer. For example, an expert system running

ICCBT 2008 - A - (21) - pp231-238 233


A Review on Expert Systems for Concrete Mix Design

on a PC could be copied to another PC at a different site, or downloaded onto a network of


PCs, or even down-loaded from the internet.
4. Human expertise is expensive. Employee salaries are far greater than the cost of PC
hardware and software. Expert systems, even taking into account development costs, are much
more affordable.

3.2. Advantages of Human Experts

1. Humans are creative, often inspired; computers are uninspired.


2. Humans are flexible, and easily adapt or integrate their expertise with other domain
knowledge; computers are not very flexible in this respect.
3. Humans possess common sense. Expert systems cannot apply knowledge to a problem
beyond the domain knowledge, and thus have a narrow focus of the problem.
4. Expert system programs have capabilities for learning that transcend those available in a
conventional program. However, such capabilities are still very primitive compared with
human learning.

4. PREVIOUS WORKS

Expert systems have been employed in mix design by researchers. These systems are
knowledge-based containing expert’s domain knowledge which is referred by the working
memory through an expert system shell for solving problems. A description of some expert
systems in concrete mix design is as follows:

4.1 HPCMIX

HPCMIX is a rule and framed based expert system for mix design of high performance
concrete (HPC). It was developed using kappa-PC expert system shell at the Department of
Civil and Structural Engineering, Universiti Kebangsaan Malaysia, Malaysia. The mix design
procedure involves both object-oriented approach and production rule. The mix design is
based on the mix design procedure for HPC developed by Aitcin as a result of its wide
acceptance among the experts in Malaysia. It is capable of proportioning concrete mixes,
diagnose causes of mix performance failure, and give recommendations on mix performance
adjustment. The effects of air content, water contributed by superplasticizer and moisture
conditions of aggregates are considered in proportioning the concrete constituents (mixing
water, cement, supplementary cementitious materials, aggregates and superplasticizer). The
knowledge base was extracted by text analysis and interviewing experts. The system is user-
friendly, runs using a windows operating system, very interactive with explanatory facilities,
gives information of data at every stage of data input and has flexible flow of consultation
making it very useful for educational environment. HPCMIX can be used to assist the user in
the mix design of HPC where material proportions of concrete constituents and past
experience are limited or not available [2].

4.2 EXMIX

EXMIX is a rule-based goal-driven knowledge-based system for the proportioning of normal


concrete mixtures. It contains both quantitative and qualitative data in its knowledge base and

234 ICCBT 2008 - A - (21) - pp231-238


M. Abdullahi & H. M. A. Al-Mattarneh

uses an expert system shell call level 5. It is capable of selecting proportions of mixing water,
cement and aggregates, with consideration being given to the effects of air-entraining
admixtures and moisture conditions of aggregates. It was developed in accordance with CSA
A23.1-M90 (Canadian Standards Association 1990) and ACI 211.1-89 using either mass or
volume method (ACI Committee 211 1989). It also conforms to building code ACI 318-89
(ACI Committee 318 1989) with respect to concrete quality requirements. A user’s manual
has been prepared for the current version of EXMIX and has been sent, together with the
program, to two community colleges in Canada for educational purposes [3].

4.3 Knowledge-Based Expert System for Concrete Mix Design

Knowledge-based expert system for concrete mix design, a rule-based expert system, was
developed to help the user in concrete mix design and adjustments. The system considers
criteria such as workability, consistency, strength, durability, density, admixtures,
transportation, and air temperature in concrete mix design. The developed system can handle
mix designs for normal-weight, normal-weight-mass, normal-weight no-slump, heavyweight,
and lightweight concrete. It uses a rule-based expert system shell known as EXSYS
professional and presents the concrete mix design on a spreadsheet (QUATTRO PRO). The
content of the knowledge base was derived from technical literatures (ACI manuals, ASTM
standards, e.t.c), interviews with experts and a survey of several experts. The system
incorporate a confidence factor of 0 – 10 to account for varying level of certainty in the result
[4].

4.4 Concrete Mix Designer

Concrete Mix Designer is a rule-based expert system that gives the proportion of the trial mix
of concrete in accordance with the ACI method. It was developed using the Personal
Consultant Plus expert system shell in the Department of Civil and Architectural Engineering
of the University of Miami[5]. Computer programs were also written in Basic to provide
question and answer capability. The system can justify its conclusions, can be expanded
incrementally and the knowledge base is ease to understand. The system is only applicable to
normal weight concrete with strength range between 17 MPa to 35 MPa without the provision
for using any admixture and thereby limiting the use of the expert system.

4.3 COMIX

COMIX is a rule and frame based expert system that gives recommendations on the design of
normal weight concrete mixes [6]. It was developed at the Central Laboratories in New
Zealand to provide expert knowledge to concrete technologists, design engineers and
consultants. The mix design employs the New Zealand concrete code "Specification for
Concrete Construction'. The system relates the type of structure to the consistency and the
placement method. The system chooses a specified strength to obtain a water-cement ratio and
calculates the amount of cement required. The volumes of coarse aggregate and sand are then
calculated. The system gives a display of the masses of the components of the concrete mix.

ICCBT 2008 - A - (21) - pp231-238 235


A Review on Expert Systems for Concrete Mix Design

5. FUTURE CHALLENGES

Expert systems have proved to be efficient and effective in providing guidance in concrete
mix design. However expert system developers must be proactive in ensuring the future
growth and development of this area of artificial intelligence. Efforts in this direction will
greatly enhance the acceptance of expert system in concrete mix design. Some of the issues
that need to be addressed are discussed as follows:

1. Some of the expert systems were developed based on earlier codes. These codes are being
improved upon based on the outcome of research findings. Expert systems are not creative
and cannot move with current mix design trends. To maintain the expert systems, the
knowledge base must also be upgraded otherwise they may not be useful in the mix design of
concrete. All system cannot claim completeness. The need to nurture and keep the system
current is quite pertinent.
2. Expert systems have narrow focus in their problem solving ability. Developers must be
willing to develop more expert systems in this area as may be demanded by future challenges.
For instance some of the expert systems are applicable for certain strength ranges, type of
concrete and based on certain codes and standards. It would be useful to explore other
concrete types like lightweight concrete, self compacting concrete, and foam concrete.
Various countries or regions could also use their codes or acceptable mix design techniques,
as may deem appropriate, to develop their expert system for desired requirement as may be
demanded in those countries or regions.
3. Most of the existing expert systems in concrete mix design are knowledge based. Future
research can be intensified to consider numerical methods (statistical method, neural network,
fuzzy logic, and genetic algorithms) in concrete mixture proportioning. These numerical
techniques have proved to be useful tools in system modeling and can be extended to concrete
mix design.

6. CONCLUSIONS

Concrete mix design is best done using proportions established for similar concrete using the
same materials. Rules-of-thumb and past experience is also applicable to concrete mix design
and adjustment wherever possible. In the event that such information is limited or not
available expert systems can be used to assist the user in concrete mix design. Expert system
approach is feasible, efficient and effective for the proportioning of concrete mix. The systems
are capable of selecting proportions of mixing ingredients based on the criteria set up by
various researchers and codes used. Some of these systems are being used by educational
institutions and industries. Future growth of expert system technology in concrete mix design
largely depends on continuous research and development of expert systems within the domain
and maintenance of the existing ones.

236 ICCBT 2008 - A - (21) - pp231-238


M. Abdullahi & H. M. A. Al-Mattarneh

REFERENCES

[1]. Darligton, K. The essence of expert system. Pearson education, England. 2000
[2]. Zain, M.F.M., Islam, M.N. and Basri, H. An expert system for mix design of high
performance concrete. Advances in engineering software. 2005. 36(5): 325 – 337.
[3]. Akhras, G. and Foo, H.C. A knowledge-based system for selecting proportions for
normal concrete. Expert systems with applications, 1994.7(2): 323-335.
[4]. Bai, Y. and Amirkhanian, S.N. Knowledge-based expert system for concrete mix
design. Journal of construction engineering and management, ASCE, 1994. 120(2): 357-
373.
[5]. Malasri, S. and Maldonado, S. (1988) Concrete mix designer. Computer application in
concrete technology, ACI; SP111-03: 33-42.
[6]. Smith, L.M. Interim report on COMIX: an expert system for concrete mix design.
Report No. M4.87/1. Central Laboratories, New Zealand, 1987
[7]. ACI Committee 211.1-91 Standard Practice for Selecting Proportions for Normal,
Heavyweight and Mass Concrete. Detroit, American Concrete Institute.
[8]. Teychenné, D C, Franklin, R E and Erntroy, H C. Design of normal concrete mixes.
Department of the Environment. HMSO, 1988.
[9]. 16. Aїtcin, P.-C. High performance concrete. London: EF&N SPON. 1998
[10]. ACI Committee 363. 1995. State-of-the-art report on high-strength concrete (ACI 363R-
92). Detroit: American Concrete Institute.
[11]. de Larrard, F. 1990. A method for proportioning high strength concrete mixtures.
Cement, Concrete, and Aggregates 12(1): 47-52.
[12]. Mehta, P.K. and Aїtcin P.-C. (1990) Principles underlying production of high-
performance concrete. Cement, Concrete and Aggregates, 12(2): 70-78.
[13]. ACI Committee 211.2-98 Standard Practice for Selecting Proportions for Structural
Lightweight Concrete. Detroit, American Concrete Institute.
[14]. ACI Committee 213R-87: Guide for structural lightweight concrete. Detroit, American
Concrete Institute.
[15]. Kett, I.. Engineered concrete: mix design and test methods. CRC Press, Boca Raton
London New York Washington, D.C. 2000
[16]. Neville, A. M. Properties of concrete, 4th ed. Addison Wesley Longman, England.
1995
[17]. Akhras, G., and Foo, H.C. An expert system for the proportioning of normal concrete
mixtures. Proceedings of the annual conference of Canadian society of civil
engineering. 1993. 4: 425-434.
[18]. Day, K.W. Computer control of concrete proportions. Concrete international.
1996.18(12).
[19]. Ganju,T.N.A method for designing concrete trial mixes. Concrete international.1996.
18(12).

ICCBT 2008 - A - (21) - pp231-238 237


A Review on Expert Systems for Concrete Mix Design

[20]. de Larrard, F. and Sedran, T. Computer aided mix design: predicting final results.
Concrete international. 1996. 18(12).
[21]. Popovics, S. and Popovics, J.S. Novel aspects in computerisation of concrete
proportioning. Concrete international. 1996. 18(12).
[22]. Giarratano, J., and Riley, G. Expert Systems Principles and programming, third edition.
PWS publishing Company, Boston, USA. 1998
[23]. Durkin, J. Expert system design and development. Macmillan publishing company,
New Jersey, USA. 1994
[24]. Wu, A.K.K. A knowledge-based application in engineering. Advances in engineering
software. 1997. 28: 469-486.
[25]. Schneider, M., Kandel, A., Langholz, G., and Chew, G. Fuzzy Expert System
Tools.John Wiley and Sons Ltd, West Sussex, England. 1996

238 ICCBT 2008 - A - (21) - pp231-238

You might also like