a fuzzy system for evaluating the risk of change in construction projects

16
1 A fuzzy system for evaluating the risk of change in construction projects Dr. Ibrahim A. Motawa* 1 ; Prof. Chimay J. Anumba 2 ; and Dr. Ashraf El-Hamalawi 2 1 Department of Structural Engineering, Mansoura University, EGYPT 2 Department of Civil & Building Engineering, Loughborough University, LE11 3TU, UK ABSTRACT A major source of risk in construction is the potential changes occurring during the project lifetime. Changes in construction projects often result from the uncertainty associated with the imprecise and vague knowledge of much project information at the early stages of projects. IT systems for change management largely focus on managing reactive changes, in which changes are recorded and then propagated to the concerned project members. However, proactive change management is hardly dealt with. Proactive change management requires estimating the likelihood of occurrence of a change event as well as estimating the degree of change impacts on project parameters. A fuzzy model is proposed in this paper to maintain these requirements. The model simulates the relationships between change causes and effects, and is intended to facilitate proactive change management on projects. Keywords: change management, fuzzy models, risk assessment Introduction Risks are the uncertain outcomes or consequences of activities or decisions when they are manageable. The identification of risk sources and estimates is an essential function of risk management. While many major decisions are made early in the project life that may take risks into account, the realism of risk estimates increases as the project proceeds. Despite difficulties associated with the identification of risk sources and estimates, they are required to be identified as early as possible. The risk impact on any of the project parameters (such as time or cost) is always modelled on a likelihood distribution function that represents the risk value and the likelihood of its occurrence, as shown in Figure 1. The likelihood distribution of risk is produced from cumulative impacts of risk sources on a specific project. A change is considered one of these sources; its contribution to form the likelihood distribution is also illustrated in Figure 1. The proposed model in this paper will help in determining this part of risk, based on the available information at the early stages of projects. Sources of change in construction include the uncertainties associated

Upload: strathclyde

Post on 24-Apr-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

1

A fuzzy system for evaluating the risk of change in construction projects

Dr. Ibrahim A. Motawa*1; Prof. Chimay J. Anumba2; and Dr. Ashraf El-Hamalawi2

1Department of Structural Engineering, Mansoura University, EGYPT

2Department of Civil & Building Engineering, Loughborough University, LE11 3TU, UK

ABSTRACT

A major source of risk in construction is the potential changes occurring during the

project lifetime. Changes in construction projects often result from the uncertainty

associated with the imprecise and vague knowledge of much project information at the

early stages of projects. IT systems for change management largely focus on managing

reactive changes, in which changes are recorded and then propagated to the concerned

project members. However, proactive change management is hardly dealt with. Proactive

change management requires estimating the likelihood of occurrence of a change event as

well as estimating the degree of change impacts on project parameters. A fuzzy model is

proposed in this paper to maintain these requirements. The model simulates the

relationships between change causes and effects, and is intended to facilitate proactive

change management on projects.

Keywords: change management, fuzzy models, risk assessment

Introduction

Risks are the uncertain outcomes or consequences of activities or decisions when they are

manageable. The identification of risk sources and estimates is an essential function of

risk management. While many major decisions are made early in the project life that may

take risks into account, the realism of risk estimates increases as the project proceeds.

Despite difficulties associated with the identification of risk sources and estimates, they

are required to be identified as early as possible. The risk impact on any of the project

parameters (such as time or cost) is always modelled on a likelihood distribution function

that represents the risk value and the likelihood of its occurrence, as shown in Figure 1.

The likelihood distribution of risk is produced from cumulative impacts of risk sources on

a specific project. A change is considered one of these sources; its contribution to form

the likelihood distribution is also illustrated in Figure 1. The proposed model in this paper

will help in determining this part of risk, based on the available information at the early

stages of projects. Sources of change in construction include the uncertainties associated

2

with the imprecise and vague knowledge of much project information at the early stages

of projects. Other sources of change may also exist but cannot be determined at the early

stages of projects, therefore are not considered in the proposed modelling. The modelling

technique used will enable the project team to trace the source of this risk and therefore

facilitate the development of a proactive action plan to deal with this risk before the

change actually occurs.

Changes in construction projects are considered among the major sources of risks for

projects due to various reasons. They are common, likely and not often immediately well

defined, but time has to be taken to consider the full information on changes. The nature

of change has been addressed in both risk analysis and change management studies. The

following section gives brief on previous related work on the subject.

Previous related work on change management in construction

Change management is an integral process that relates all internal and external factors that

influence project changes. Research projects on managing change in construction have

tended to focus on the identification of factors affecting the success of a change process,

and to introduce guidance for best practice in change implementation. Examples of such

research include: a concept for project change management (Construction Industry

Institute [1]), best practices for managing change efficiently (Construction Industry

Institute [2]), a generic procedure for issuing a change order request (Cox et al [3]), an

analysis method to reduce the overall rate of construction change orders (Stocks and

Singh [4]), best practice recommendations for the effective management of change on

projects (CIRIA [5]), and an advanced project change management system (Ibbs et al [6]).

Several other researchers have investigated the evaluation of change effects on certain

project elements. Hester et al [7] studied construction change order impacts on labour

productivity at the craft level. Ibbs [8] investigated the effect of the size and the time of

change on a project. Hanna et al [9] developed a linear regression model that predicts the

impact of change orders on labour productivity. Williams [10] studied the risk of changes

to safety regulations and its effect on a project. Lee et al [11] developed decision tree

models to classify and quantify productivity losses caused by change order impacts.

Change management has also been the focus of different IT systems. An integrated

environment for computer-aided engineering was developed by Ahmed et al [12], which

is a blackboard representation that integrates a global database, several knowledge

modules, and a control mechanism to systemize object changes. Peltonen et al [13]

proposed an engineering document management system for changes that incorporated

document approval and release procedures. Spooner and Hardwick [14] developed a

system with rules for coordinating concurrent changes and for identifying and resolving

conflict modifications. Ganeshan et al [15] developed a system to capture the history of

the design process, initiate backtracking, and determine the decisions that might be

affected when changes are made in the spatial design of residential buildings.

Krishnamurthy and Law [16] presented an interesting change management model that

3

supports multidisciplinary collaborative design environments. Another change

management system was proposed by Mokhtar et al [17] for managing design change in a

collaborative environment. The model is capable of propagating design changes and

tracking past changes. Soh and Wang [18] proposed a constraint methodology based on a

parametric technique to coordinate design consistency between different geometric

models and to facilitate managing design changes. Hegazy et al [19] introduced an

information model to facilitate design coordination and management of design changes.

Important dependencies between building components were represented by this model to

help identify the ripple effect of changes between components. Also, a reporting system

was used to view the history of all changes made by all disciplines. A more generic IT

system was presented by Karim and Adeli [20] which is an object-oriented (OO)

information model for construction scheduling, cost optimisation, and change order

management. Charoenngam et al [21] developed a Web-based change order management

system that supports documentation practice, communication and integration between

different team members in the change order workflow.

It can be seen from the above that previous work on change management and evaluation

mainly focused either on the identification of the change process, best practice

recommendations for managing change during the project life cycle, or evaluation of the

change effects on a single project parameter. Identification and recommendation are not

enough for managing change effectively, as change management should help in

forecasting possible changes, planning defensive impacts; and coordinating changes

across the entire project. The IT systems developed for change management are, in

general, integrated systems that represent design information, record design rationale,

facilitate design co-ordination and changes, and notify users of file changes. These

systems were developed mainly to deal with reactive changes, particularly design changes.

On the other hand, systems for the evaluation of change effects were developed to

estimate certain change effects, such as productivity loss. However, evaluation of change

effects without relating multiple change causes to multiple change effects (which is the

case for many change events) does not consider the whole case of change.

In the light of the above, the focus of this research was on how to deal with

proactive/reactive changes taken the interdependency between multiple causes and effects

into consideration. A change management system was developed that consists of a model

to identify and forecast potential changes and evaluate their effects, which in turn will

help in estimating the risk value for change. The system also includes a workflow model

that deals with reactive changes. This paper focuses on the principles and structure of the

proactive model of the system. The research problem being addressed is first presented

and is followed by a description of the approach adopted in the development of the model.

The paper presents the IT system architecture of the proposed model and an application

example is used to demonstrate the data input and output of the proposed system. The

benefits of the model are outlined in the concluding part of the paper.

4

Research problem

During the early stages of projects, the amount and precision of qualitative information

available makes projects prone to many changes. A well-established methodology to

gather this information and use it for tracking potential changes is required for timely and

accurate prediction of changes throughout the project process. When changes occur, the

investigation for the change consequences relates them to several causes such as design

errors or changes in client requirements, which are actually the direct change causes.

However, these causes are usually based on the amount and precision of information

available at the early stages of the project. Such information often influences the number

of change cases. This is actually a major outcome from the case studies carried out as part

of this research.

The information for building a full cause and effect relationship becomes available as

changes occur and are assessed. However, very little is known about project change at the

very early phases of a project, and thus, traditional change models, that were reviewed in

literature, are inadequate for predicting change events under uncertain conditions.

Therefore, at such early phases it is a challenging task to show correlations and

relationships between the limited information available and the problems that arise in the

later phases of the project life cycle. Existing models for change assessment are largely

based on checklists and definitions of change causes and are mainly used to negotiate

compensation for any disruptive consequences caused by the change. However, prediction

models tend to make subjective assumptions to deal with the events, based on the

experience of the project team.

For a given project, different project characteristics may lead to change. Therefore, the

proposed model targets simulating the cause and effect relationships within change cases

taking these project characteristics into consideration. Figure 2 depicts a typical network

of cause and effect relationships for a change case.

The proposed methodology for the change prediction model includes the identification of

prediction set elements (namely: project characteristics, change causes, and change

effects); and the relationships between these elements. At this stage, the model assumes

that the elements of each set are independent from the other elements in the same set.

Project characteristics (F) are factors or aspects that have an influence on the project and

may lead to change. Change causes (C) are the direct causes of a specific change event

when it occurs; these are likely to be because of the existence of certain project

characteristics. Change effects (E) are the change consequences on project parameters.

Two measures, R1 and R2, have been targeted to represent the degree of dependency

between F’s and C’s and between C’s and E’s, respectively. They actually represent the

sensitivity of the impact of one set of elements to variations in another set of elements.

The model elements (F, C and E) are detailed elsewhere: Motawa et al [22].

The model then uses these prediction elements and their relationships to generate the

likelihood of change occurrence and change effects on project parameters. The approach

5

for modelling these relationships is illustrated in Figure 3 using an example of three

project characteristics (F), three change causes (C) and two change effects (E). The

analysis is undertaken for the expected change effects, considering all causes of change

under specific project characteristics. In order to estimate the likelihood of change

occurrence and change effects, which reflect the risk and uncertainty around the change

occurrence, it is proposed to represent the prediction elements as follows:

Fi = the degree of influence of project characteristics on a specific project which may lead

to change;

Rij = the sensitivity of a change cause occurrence to variations in the project

Characteristics;

Rjk = the sensitivity of the change impact to variations in the change causes.

The relationship between these elements are formed and combined to produce the

cumulative relation, as shown by Equation 1:

jk

lkmi

ki

mjni

jiijijkiji RRFRRFf

,

1,

,

1,

*,, ………………………….…. Eqn.(1)

The inner summation in the calculation of Eqn.1 gives the likelihood of change

occurrence while the overall calculation results in the change impacts. The model

attempts to predict change impacts on project parameters using data that are realistic to

obtain, while limited data (documented) are only available from previous projects and

many of the above elements can only be expressed in linguistic terms. This means that the

model will simulate the uncertainty, the subjectivity of the impact and variations of these

sets of elements. Different approaches have been investigated to develop this model,

which in turn will determine the amount of risk for change. The following section

discusses the most suitable one, and outlines why other approached were discarded.

Determining the risk of change

Modelling uncertainty and variations of subjective variables has been the focus of many

systems such as: Rule-based systems, Neural Networks-NN, and fuzzy systems. For the

problem in hand, it was concluded from the case studies undertaken as part of the research

project that:

Much of the information available for the change cases is always vague.

Change consequences might lead to dispute, therefore, detailed analyses and

investigation of the change causes and effects are required. This needs close

observation for the simulation route of cause and effect relationships.

These reasons make approaches such as rule-based systems and NN inappropriate for

modelling change prediction. Rule-based systems need the information on change events

to be represented in the form of well-defined rules. This is inappropriate in this case

because the available information on the recorded change events cannot be encapsulated

in rules explicitly and it is difficult to formulate a complete and accurate set of rules to

6

capture the expertise required and the issues involved. On the other hand, NN systems

were found inappropriate, as they neither provide details of variable manipulations nor the

rationale for the conclusions drawn. Therefore, fuzzy systems were found the most

suitable approach for modelling the problem domain.

Fuzzy Modelling

A fuzzy model represents a real problem using fuzzy variables, algorithm, and/or rules.

Fuzzy set theory was first developed by Zadeh [23] as a mathematical approach to

representing uncertain and imprecise information. It provides approximate but effective

descriptions for highly complex, ill-defined mathematical systems. It effectively supports

linguistic imprecision and vagueness. Fuzzy systems, based on fuzzy sets and fuzzy logic,

can be used for the purposes of modelling, prediction, classification, and control.

The quality of the fuzzy approximation depends predominantly on the quality of

engineering judgment (subjective knowledge) and human expertise used to identify the

fuzzy input variables. For this reason, the development of fuzzy sets for the identified

input variables is often considered the most important step in developing the fuzzy

system. Therefore, a fuzzy algorithm for change prediction using Equation 1 is

considered. The estimation of the problem variables (Fi, Rij, and Rjk) will be simulated as

fuzzy numbers on a scale of 0 - 10 (where ‘0’ represents the ‘no relationship/effect’ and

‘10’ represents the ‘very strong relationship/effect’). The membership function of this

numerical scale is shown in Figure 4-a. A linguistic scale is also proposed in case of using

linguistic judgment for the problem variables, shown in Figure 4-b. The sum and the

product of fuzzy sets are computed using the extension principle, described in Equations 2

and 3, where 'A' and 'B' are fuzzy numbers. For the defuzzification process to convert the

membership function of the final summation into a crisp (non-fuzzy) value, the centre of

area method is used. The fuzzy algorithm for the model is fully detailed elsewhere:

Motawa et al [24].

A · B = 1

0

).(

BA = 1

0

).(

BA ………………………………….…….………Eqn.(2)

A+ B = 1

0

).(

BA = 1

0

).(

BA ………………………………..……………Eqn.(3)

The fuzzy model described above was implemented in a software system for operational

purposes (Motawa et al [25]). The system is designed as a Java package of components

with their supporting classes, beans, and files. Figure 5 shows the architecture of the

system and the relationships between its various components. Briefly, the main

components of the system are:

1. Database: The system efficiency depends on the quality of a well-designed database.

The system database contains two kinds of information: (1) data collected from the

7

case studies carried out regarding the prediction elements of change. A set of tables

were designed to store the required information for this database as well as the system

outputs; and (2) case-base, which represents specific knowledge tied to specific

situations of a change case, also represents knowledge at an operational level; that is,

how a change case was carried out or how a piece of knowledge was applied or what

particular strategies for accomplishing a change case were used. Each case in the case-

base is composed of three major parts: change case description, system outcome and

actual case scenarios. The description specifies the situation in a change case (i.e. the

means by which the case was affected). The outcome indicates the analysis of the

fuzzy results after the change case is modeled. The actual case scenario outlines any

elements that have arisen during the implementation of change and have not been

taken into account for modelling.

2. Project data: these are interface frames “JFC (Swing) Frames” (Java Foundation

Class) that provide users with interactive tools. They are used to locate the input data,

which are selected from the database and/or entered directly by the user.

3. Modelling change cases: this is a Java code that translates the fuzzy model for change

prediction and effects into the computer program. The system allows both numerical

and linguistic estimation for each element.

4. Output form: this is a JFC Frame that displays the system outputs.

5. The archive is a documentation system for all information on the change case and the

system output.

A set of links has been designed to navigate from page to page and to run the system’s

functions.

System application/operation

The example shown here is based on one of the case studies used in the development of

the model. The case study project involved extending a store building (to almost a

doubled size) and refurbishing the existing one. The case study focused on constructing

the floor of the building. The original construction plan for the floor had been completely

changed. During the construction phase, a floor survey was investigated by a new

European regulation on floor surfaces. The outcome of the survey was: the floor would

not meet the new standards and the floor was damaged in places and therefore either

needed to be replaced or repaired. Different options were proposed to deal with the case

as shown in Figure 6. The information regarding the change case, in terms of the project

characteristics, change causes and effects are presented in Tables 1, 2 and 3. The fuzzy

variables, in this example, are described in numerical format. In Table 1, the project

characteristics (Fi) and the fuzzy estimates for its influence on the likelihood of the

change occurrence are shown. The project team defined four characteristics for this

project that may have influence on the project. For example, ‘the external pressures

bearing on client’ (F1) was given the estimate of ‘7.0’ as a fuzzy weight for its influence.

The team also estimated the dependency values between each change cause and project

characteristics. For example, a fuzzy value of ‘8.0’ was given to the relationship between

the change cause (C1) and the project characteristic (F1). No relationship is given between

(F1) and (C5). These values express the sensitivity of the change causes occurrence to any

8

variations in the project characteristics. High values for these sensitivities can be noticed

in Table 2; such as between C1 & C2 and F1, all dependencies with F2, and between C4 and

F4. The team also defined the dependency between the change causes and effects in the

same way, as shown in Table 3.

The model output, shown in Tables 4 and 5, represent the likelihood of the change

occurrence and the amount of change effect on the defined project parameters (E1, E2).

The model output can be used in the following ways:

1. The project tasks of every change case should be modelled on a project management

program. This group of tasks is subjected to certain risks that are defined by the

project team at the planning stage. The model output for the defined change case has

given the likelihood of the change cause ‘Delays in decision making’ as 12.7%. This

risk value can be used to represent the effect of this risk on the project tasks of this

change case. For example, the durations of the tasks affected by ‘Delays in decision-

making’ will have a 12.7% increase on the original estimate.

2. For a successful project, the project parameter ‘Loss of workflow’ should have a zero

value (No loss). For the studied project, the change occurrence has given an effect of

36.71% on this parameter. This estimate can be used for measuring the project

success and also in the case of dispute resolution where the sources of the change

causes can be traced through the cause and effect relationships defined by this model.

The above model has been developed in a software application. A typical user of this

model is the project team who are expert in estimating the accuracy of the project

information at the early stages of projects. The model is best used at the planning stage.

The user can use this software following scenario below:

1. The user starts the application by connecting to the system database, which displays a

list of prediction elements or archived cases.

2. The user selects an element for the change case by clicking on its name from the

appropriate list. This action causes the application to build a list of elements for each

specific change case. Figure 7 shows the change case elements of constructing the

floor of a store building. Input data required for the fuzzy model include fuzzy

weights for the selected project characteristics. The data also include fuzzy values for

the sensitivity of a change cause occurrence to variations in the project characteristics

and fuzzy values for the sensitivity of the change impact to variations in the change

causes (the relationship between the case elements).

3. The user can add more elements, which are not included in the database, where

applicable, to the studied change cases. The system has the capability either to update

the original database by the additional elements or to save them only for this specific

change case.

4. For a new change case, system users can integrate the case base to obtain information

regarding a similar change case together with related parameters and elements. This

enables further estimation and analysis.

9

5. The results are shown in Figure 8, which demonstrates the likelihood of occurrence of

each cause of change and the amount of impact that the change will have on each

project parameter.

Conclusions

Change, as a source of risk, requires to be estimated to help in determining the overall risk

value on a certain project parameter. The focus of change management in construction

was on the identification of the change process, the best practice of change

implementation within a project, the evaluation of change effects on specific project

parameters, and IT systems for managing reactive changes and facilitating co-ordination

of design changes. However, modelling construction change should consider the link

between: 1) project characteristics that lead to change 2) causes of change, 3) the

likelihood of change occurrence, and 4) the change consequences. Relating multiple

change causes to multiple change effects is a major role for modelling change in

construction. This research has focused on identifying and forecasting potential changes

and evaluating the impact of changes before they actually occur. A fuzzy model was

proposed in this research to estimate the likelihood of occurrence of a change event and to

predict the effect of change on project parameters using data available at the early stages

of projects. This paper presented the main components and functionalities of this model.

The fuzzy model relates the set elements of project characteristics to the occurrence of

change causes, and hence determines the overall change impact on project parameters.

These elements are treated as fuzzy set numbers or linguistic variables, combined and

processed through a fuzzy relation formula to map their impact on project parameters.

An example was presented to illustrate the working of the model, which includes a

simulation of the effects of multiple change causes. The model was developed as a

software application for operational use. This modelling approach is intended to enable

the project team to test the sensitivity of the likelihood of change occurrence and the

effects of change on project parameters due to variations in these sets of elements. This is

useful in alerting the project team to changes so as to take corrective action and to

minimize the disruptive effect of changes. The model is useful for dealing with proactive

changes by identifying/tracking the main project characteristics that have an influence on

change causes. It also gives the contribution of change to the overall risk estimate that

could be used for risk analysis.

As a limitation for the model use, the model assumes at this stage of development that the

elements of each set are independent from the other elements in the same set. The model

considers only the sources of change that can be determined at the early stages of projects.

Therefore, information that may cause changes such as irrational behaviours of the project

stakeholders and also information resulted from the project progress, are not included.

The estimations, given by the project team, to the accuracy of the project information at

the planning stage and how these estimations are linked with change events, will affect

the accuracy of the model results.

10

References

[1] Construction Industry Institute (CII), 1994. Project change management Research

team, Special publication 43-1., The University of Texas at Austin, US.

[2] Construction Industry Institute (CII) conference, 1996. Project change management –

implementation feedback report.

[3] Cox, I.D., Morris, J.P., Rogerson, J.H., and Jared, G.E., 1999. A quantitative study of

post contract award design changes in construction. J. Constr. Mgmt. And Economics, 17,

pp. 427-439.

[4] Stocks, S. N. and Singh A., 1999. Studies on the impact of functional analysis concept

design on reduction in change orders. J. Constr. Mgmt. And Economics, 17, pp. 251-267.

[5] Construction Industry Research and Information Association (CIRIA), 2001.

Managing project change – A best practice guide.

[6] Ibbs, C. W., Wong, C. K., and Kwak, Y. H., 2001. Project change management

system. J. Mgmt. in Engrg. ASCE, 17 (3), pp. 159-165.

[7] Hester, W.T., Kuprenas, J. A. and Chang, T. C., 1991. Construction changes and

change orders: their magnitude and impact. Construction Industry Institute (CII), Source

document 66, CII, Austin, Tex.

[8] Ibbs, C. W., 1997. Quantitative impacts of project change: Size issues. J. Constr.

Engrg. and Mgmt. ASCE, 123 (3), pp. 308-311.

[9] Hanna, A. S., Russell, J. S., and Vandenberg, P.J., 1999. The impact of change orders

on mechanical construction labour efficiency. J. Constr. Mgmt. And Economics, 17, pp.

721-730.

[10] Williams, T.M., 2000. Safety regulation changes during projects: the use of system

dynamics to quantify the effects of change. I. J. of Project Mgmt, 18 (1), pp.23 – 31.

[11] Lee, M., Hanna, A. S., and Loh, W., 2004. Decision tree approach to classify and

quantify cumulative impact of change orders on productivity. J. Computing in Civil

Engrg. ASCE, 18 (2), pp. 132-144.

[12] Ahmed, S., Sriram, D., and Logcher, R., 1992. Transaction-management issues in

collaborative engineering. J. Computing in Civil Engrg. ASCE, 6 (1), pp. 85-105.

[13] Peltonen, H., Mannisto, T., Alho, K., and Sulonen, R., 1993. An engineering

document management system. Proc. ASME Winter Annu. Meeting.

[14] Spooner, D. and Hardwick, M., 1993. Using persistent object technology to support

concurrent engineering systems. Concurrent engineering: Methodology and applications,

Elsevier Science, Amsterdam, pp. 205-234.

[15] Ganeshan, R., Garrett, J. and Finger, S., 1994. A framework for representing design

intent. Design Studies, 15 (1), pp 59-84.

[16] Krishnamurthy, K. and Law, K., 1995. A data management model for design change

control. Concurrent engineering: Res. and applications, 3(4), pp. 329-343.

[17] Mokhtar, A., Bedard, C, and Fazio, P., 1998. Information model for managing design

chnages in a collaborative environment. J. Computing in Civil Engrg. ASCE, 12 (2), pp.

82-92.

[18] Soh, C. and Wang, Z., 2000. Parametric coordinator for engineering design. J.

Computing in Civil Engrg. ASCE, 14 (4), pp. 233-240.

11

[19] Hegazy, T., Zaneldin, E. and Grierson, D., 2001. Improving design coordination for

building projects – I: Information model. J. Constr. Engrg. and Mgmt. ASCE, 127 (4), pp.

322-329.

[20] Karim, A. and Adeli,. H., 1999. CONSCOM: An OO construction scheduling and

change management system. J. Constr. Engrg. and Mgmt. ASCE, 125 (5), pp. 368-376.

[21] Charoenngam, C., Coquince, S. T. and Hadikusumo, B. H. W., 2003. Web-based

application for manging change orders in construction projects. J. Construction

Innovation, 3 , pp. 197-215.

[22] Motawa, I.A., Anumba, C. J., El-Hamalawi, A., Chung, P. and Yeoh, M., 2003a. An

innovative approach to the assessment of change implementation in construction projects.

Proceeding of the Second International Conference on Innovation in Architecture,

Engineering and Construction, Loughborough University, UK, June 25-27, 2003, pp 729-

740.

[23] Zadeh, L., 1979. A theory of approximate reasoning. Machine Intelligence, Halstead

Press, New York, pp. 149-194.

[24] Motawa, I. A., Anumba, C. J., El-Hamalawi, A., Sun, M., Chung, P. W. H. and

Yeoh, M., 2003b. A fuzzy model for change prediction in construction projects.

Proceedings of the 10th EG-ICE International Workshop on Intelligent Computing in

Engineering, (European Group for Intelligent Computing in Engineering), Delft

University of Technology, Delft, The Netherlands, July 3-4, 2003, pp 44-52.

[25] Motawa, I. A., Anumba, C. J., El-Hamalawi, A., Chung, P. W. H., Yeoh, M. and

Sun, M., 2004. Development of a fuzzy system for change prediction in construction

projects. Proceedings of the Xth International Conference on Computing in Civil and

Building Engineering (Xth ICCCBE), Weimar, Germany, June 02-04, 2004.

Table 1: Project characteristics of the floor case

ID Project characteristics Fuzzy estimates

F1 External pressures bearing on client (re-location, changing needs) 7.0

F2 Building and statutory regulations 8.0

F3 Quality issues 4.0

F4 Skills and Knowledge 3.0

Table 2: The dependency between project characteristics and change causes

ID Change causes F1 F2 F3 F4

C1 Revised client requirements 8.0 7.0 3.0 4.0

C2 Schedule pressure 8.0 8.0 4.0 4.0

C3 Lack in Design Information 3.0 9.0 4.0 2.0

C4 Delays in decision making 4.0 8.0 3.0 7.0

C5 Unforeseen events and site conditions 0.0 6.0 3.0 3.0

12

Table 3: The dependency between change causes and effects

ID Change effects C1 C2 C3 C4 C5

E1 Re-organise and schedule the work

methods, production schedules and

deliveries

8.0 9.0 8.0 8.0 7.0

E2 Loss of workflow 5.0 6.0 4.0 9.0 5.0

Table 4: The likelihood of occurrence of change causes

ID Change causes Likelihood of occurrence

C1 Revised client requirements 13.8

C2 Schedule pressure 15.0

C3 Lack in Design Information 11.7

C4 Delays in decision making 12.7

C5 Unforeseen Events and site conditions 7.3

Table 5: The change impact on project parameters

ID Change effects Value of impact

E1 Re-organise and schedule the work methods, production

schedules and deliveries

50.23

E2 Loss of workflow 36.71

Figure 1. Likelihood distribution function for risk impact on a project parameter

the risk impact on the project parameter

Change contribution to the risk value

Initial estimate of the project parameter

–ve % +ve %

1.0

0

13

Figure 2. Cause and effect relationships within change cases

Figure 3. Analysis of cause-and-effect relationships for a change case

Figure 4-a. The membership function of the change prediction elements (numerical scale)

Rjk

Rij

Project characteristics (F)

Change causes (C)

Change effects (E)

Rj1

Ri2 Ri3 Ri1

C1

F1

F2 F3

E1

C2

F1

F2 F3

C3

F1

F2 F3

Rj2

Ri2 Ri3 Ri1

C1

F1

F2 F3

E2

C2

F1

F2 F3

C3

F1

F2 F3

0 1 2 3 4 5 6 7 8 9 10

1.0

µ

14

Figure 4-b. The membership function of the change prediction elements (linguistic scale)

Figure 5. Architecture of the Change Prediction System

3. Modelling change cases

run the Fuzzy model of

change prediction

query tag prediction

elements

2. Project

data

5. Change

cases archive

populate prediction

elements/case

display prediction

elements/case

4. Output

form

display prediction

output

Update archive model experience

Case-base

User

1. Database

Generate

1.0

µ

0 1 2 3 4 5 6 7 8 9 10

very low low moderate high very high

15

Figure 6 The change options for the floor construction

(2) Overlay

tiles with new

(3) Patch

repair

Survey to confirm

assumption made

Ok

Failed

(1) Initial proposal (Replacement with Terrazzo)

Testing of the screed

below the existing tiles (1) Terrazzo floor

Failed Passed

Tiles testing

(4) Replace

with new

Implement

option (4) if the

survey confirm

the assumption

If not, implement

option (5)

‘patch/repair

screed’ lay new

tiles as overlay

This option was

expected as the

most likely

option from the

beginning

16

Figure 7. Data of the change case example

Figure 8. Results of the change case example