North Dakota State University
Fargo, North Dakota
of this paper is to demonstrate a computer methodology that might be used by the construction industry to improve the effectiveness
of its decision- making processes. The ultimate goal is the development of a software system that will bring sophisticated
management science techniques to the construction professionals and at the same time will utilize their knowledge and expertise.
The process of construction is highly dependent on the
decisions made by different people at different stages of the project. The quality of the project, both during the construction
process, and during its life as a constructed facility, is a direct consequence of the quality of these decisions. Expertise
and judgment of the decision makers play critical roles in the life of a construction project since many of the decisions
are based on incomplete knowledge, uncertain situations, and imprecise data. Due to the unique nature of each construction
process, inherent uncertainties, and incomplete scope definition, it is almost impossible to have all the needed information
at the time of decision-making. This paper is based on the conviction that decisions of improved quality can still be obtained
by improving the process of decision-making. Application of analytic management-science tools can greatly improve the decision-making
process. However, there use is not very common in the construction industry. Critical Path Method or CPM is a classic example.
CPM is the only analytic technique that is generally accepted in the industry. Yet its use, in most cases, is forced by regulation
or contractual agreement. The real purpose of using CPM as a schedule and cost control tool is most often ignored even when
it is used.
The objective of this paper is to look into the specific
needs of the decision makers in the construction industry and to outline a framework that might be useful for the improvement
of its decision-making processes.
DECISION-MAKING PROCESS IN THE CONSTRUCTION INDUSTRY
The human decision-making process is normally implicit.
The rationale and the reasoning of the expert decision makers are not always clearly known to others. Most decision problems
are solved by expert judgment, where knowledge is subjective. This is even evident in the construction industry. The process
of construction is highly dependent upon the experience and expertise of the people involved. Financial, contractual and scheduling
decisions are all made by experts using their subjective and implicit knowledge. Some of these decisions are too complex to
be handled by one individual implicitly and should be solved explicitly. The explication will allow the decision makers to
use analytic decision-making tools and will improve communication among the many experts involved. This way decision-making
will be a structured process. Moreover, explicit decision-making methods can be documented and computerized.
According to Keeney and Winterfeldt (1989) the main advantages
of an explicit decision-making process are as follows:
procedure typically breaks an implicit thought process into smaller parts and applies logic to integrate these parts.|
in an explicit thought process can be clearly documented to improve communication and facilitate peer review.|
is an absolute necessity when the problem is complex, when information is required from a range of technical disciplines,
or when justification of the experts' thought processes or their implications are important.|
The inputs to any decision-making process, implicit or
explicit, are the data or the facts. Decision makers make decisions on the basis of their knowledge about the facts. In complex
problem situations, it is very difficult to apply knowledge on the unprocessed data or facts. Therefore, in explicit methods,
analytical models are used to convert "data" into "information." Information, thus obtained, becomes available for use by
the decision makers. However, their "knowledge" is still needed to process the information. In other words, explicit decision-making
processes must also be able to incorporate the knowledge, expertise and judgment of the decision makers in them. Otherwise,
the extent of their use in the construction industry will remain as low as it is now.
COMPUTER-BASED DECISION-MAKING SYSTEMS
A decision support system, DSS for short, is a computer-based
information system specially designed for use by the decision makers in the business and industry. In specific terms, A DSS
is an algorithmic computer program that can access a database to locate the necessary data, can utilize a repertoire of mathematical
and/or statistical models, and can produce the desired information at the user's terminal. It has been argued that (Cooper
1986) these softwares have not been able to bring management science and management together. The main criticism of the DSS
products is that they are intended for specialists and therefore are not directly useful to decision makers. Most decision
makers are not necessarily computer literate or skilled in mathematics. Decision makers in construction are generally averse
to the use of advanced mathematics. They neither have the training nor the inclination to use analytic management science
tools on a day-to-day basis. However, these tools can provide useful information. The need is to be able to extract information
using these tools without putting any demand on the decision maker's mathematical expertise. At the same time there should
be a way to blend the domain-specific knowledge and experience of the decision makers with the information obtained from the
databases and the models.
Now let's take a look at another type of computer-based
decision systems, commonly known as knowledge-based expert systems. The idea is, not only information, but expert-knowledge
can be obtained from these automatic systems. In expert systems, many limitations that are typical of DSS products, have been
successfully eliminated. Expert systems are more user-friendly than conventional decision support systems. In expert systems,
knowledge can be represented symbolical!y making qualitative knowledge representation possible. Expert systems can also explain
the process of inference by tracing back the inference-chain. Although expert systems did not live up to the expectation
generated when the technology first emerged, there have been a lot of useful applications. For a few construction-related
examples the readers are referred to Allwood (1989). The technologies of machine-learning and natural language processing
are not yet developed to the stage of useful and meaningful applications m the construction industry. The biggest hindrance
is, however, that they are designed as a store of knowledge for use by the people who do not have that knowledge. In other
words, expert systems are generally developed for novices who are in need of expert knowledge. There are situations in construction
where these expert systems might be useful (Ahmad and Minkarah 1990), but in many cases, a typical decision maker is knowledgeable
about his or her domain of expertise. What such a user needs is not expert knowledge, but technical support for processing
In essence, the construction industry needs a type of
systems that has the best of both DSS and ES (expert system, in short) technologies. Realizing this necessity researchers
have started putting their effort in developing systems that are called expert decision support systems or EDSS.
EXPERT DECISION SUPPORT SYSTEMS
A recent development has been described by Fiksel and
Hayes-Roth (1989). According to them these systems (although they used the term IDSS or intelligent decision support systems)
"... primarily concerned with support of learning and
exploration, requiring continuous modifications to the conceptual model of the domain. They are designed to provide interpretation
Quoted from Fiksel and Hayes-Roth (1989).
Fig 1 below shows the major points of differences among
the DSS, ES and EDSS.
It is apparent that the EDSS is a result of further improvement
of DSS products but they are fundamentally different from the ES. An ES is designed for the use of a novice on the basis of
a fixed conceptual model of the problem domain. The knowledge in an ES comes from the experts in that domain. The EDSS, on
the other hand, are designed for the knowledgeable users. The EDSS has technical expertise, not in a narrow, specific area,
but wide-ranging expertise - the kind that can be applied to a variety of different problems. The EDSS will lead the user
through certain procedures in order to help him or her analyze problems and make decisions.
Comparison of Computer-Based Decision Making Techniques
If decision-making is viewed as an input-output system,
EDSS can be thought of as a processing mechanism as shown in Fig. 2. Its purpose is basically to blend the information with
the knowledge by providing support in the form of interpretation and insight. An EDSS may have conventional databases, DSS
and/or MIS (management information systems, an earlier variety of the DSS products) as components of the system, but the components
must be integrated in a way to provide meaningful support to its users. It should pace a user-friendly layer between the quantitative
models and the qualitative knowledge of the user.
Figure 2 Decision Making as an Input-Output System
The advantages of computerizing the decision analysis
techniques in expert decision support systems are:
maker who does not have special knowledge of the analytical tools will still be able to use them.|
if designed properly, will explain and teach how to use the tools or techniques. Thus incorporation of the analytic management-science
techniques will be meaningful and useful.|
be possible to reveal the underlying logic of the decisions made, since the EDSS can look back at the inference-chain and
explain the reasoning process.|
will be obtained rather quickly. Also, the user will be able to change the value of any input variable to see the effect on
the output (decision). Thus "what-if" and sensitivity analyses will be performed easily and quickly.|
SCHEDULING AND CONTROL WITH A CONCEPTUAL EDSS
Scheduling of construction projects is a complex process
and requires experience, technical expertise and judgment. It begins with preliminary planning and generating a list of activities
(depending on the desired degree of breakdown). Even during this initial stage, a responsible and experienced scheduler would
collect a lot of information. For example, project characteristics, workforce and location-related data, constraints or restrictions
imposed by various parties involved, such as owners, contractors, architects, and so on. In the next step, the generated list
of activities are arranged in a network that shows the interrelationship (in terms of dependency) among the activities. Being
able to draw and comprehend an activity network without any technical knowledge is impossible, since it requires special understanding
of the network scheduling technique. Many construction professionals neither have time nor inclination (or training) to learn
and apply the technical knowledge needed. Commercially available network scheduling softwares do not provide the kind of support
needed by the construction
managers. A great majority of these computer packages
require the network diagram (or the information contained in a diagram), along with the activity durations, as input. As output,
these programs will generate a series of tables containing early and late start dates, floats (or slacks), critical path(s),
etc. Usually, no support for interpreting these highly technical information is available from these softwares. As a consequence,
CPM- a analytical tools (even the computerized systems remained largely unused. The full potential of the network analysis
technique has never been realized by the construction industry.
As a possible solution to the problems and difficulties,
the network analysis technique can be used as a component within an EDSS, as shown in Fig. 3.
Figure 3 Scheduling Decisions with a Conceptual EDSS
In fact, with proper interfacing, it is possible to use
commercially available project scheduling softwares within a broader system. In a similar manner, database systems, statistical
models, and other algorithmic programs could be incorporated as components of the EDSS. An additional module designed to be
a support system should be added to the overall system. The main task of this module would be to establish a dialog-system
between the user and the other modules of the EDSS.
A properly designed EDSS for scheduling should be able
to offer suggestions as shown below:
Activity M can be delayed by X number of days without
changing the project duration or the early-start dates of other activities; do you want to delay M by X (or any number smaller
than X) days? If yes, by how many days??? (Note: X is the free float for activity M.)
At this point, the user might want to know the implications
of delaying M and the system should be able to show the impacts on the cost, resource and other areas. Based on the user's
own knowledge and the information provided by the system a decision regarding the delay of activity M would not be too difficult
to reach at.
The paper is based on the observation that, in construction,
potential users of computerized decision-making systems are neither novice - as ideal expert systems (ES) users - nor quantitatively
oriented - as ideal decision support systems (DSS) users. Therefore, the need is to have a kind of systems that can provide
support in the form of interpretation and insight to the knowledgeable users. EDSS - or expert decision support systems -
is the term used to describe such systems. A conceptual EDSS in the context of scheduling decisions is outlined in the paper.
The main advantages of computerizing decision-making processes in the EDSS framework are pointed out below:
management-science tools (network analysis, for example) will be accessible to the construction managers; usefulness of these
tools will be revealed while their complexity will remain hidden - but not impenetrable.|
knowledge and expertise of the construction professionals will be utilized in the decision-making process since an EDSS relies
on the domain-specific knowledge of the user. The EDSS will be used to elicit the pertinent knowledge from the practitioners
and to blend this knowledge with other input variables so that an analytical tool can be used meaningfully.|
communication system among different construction professionals (owners, architects, engineers, foremen, estimators, schedulers,
and so on) would become feasible|
- Ahmad, I. and
Minkarah, I. (1990), “Decision Analysis and Expert System Technology: A Construction Industry Perspective,” Proceedings
of the CIB 1990 Symposium on Value in Building Economics and Construction Management, Sydney, Australia. To be held in
- Allwood, R.J. (1989). Techniques and Applications of Expert Systems in the Construction
Industry, Ellis Horwood Limited, England.
- Cooper, P. (1986) “Expert Systems in Management Science,” North-Holland
Future Generations Computer Systems, No. 2, pp.217-223.
- Fiksel, J. and Hayes-Roth, F.(1989). “ Knowledge Systems for Planning Support,”
IEEE Expert, Fall issue, pp.16-23.
- Keeney, R.L., and Winterfeldt, D.V. (1989). “On the Uses of Expert Judgment
on Complex Technical Problems,” IEEE Transactions on Engineering Management, Vol. 36, No.2, pp.83-86.