You are on page 1of 20

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1741-0401.

htm

IJPPM 59,1

Modelling construction project productivity using systems dynamics approach


Michael J. Mawdesley
School of Civil Engineering, University of Nottingham, Nottingham, UK, and

18
Received May 2008 Revised March 2009 Accepted March 2009

Saad Al-Jibouri
Department of Civil Engineering, University of Twente, Enschede, The Netherlands
Abstract
Purpose Improvement in productivity will not be achieved without bearing in mind that there is an enormous number of factors affecting productivity and that there is a necessity to locate the most inuential ones among them. Doing so will enable researchers as well as practitioners to pinpoint the areas where efforts are to be directed in order to reach the optimum productivity of the studied project. The work described in this paper is based on data collected from the construction industry in the UK. In collecting the information, there are three initial aims: to determine what factors affected productivity at site level, to determine how these factors interacted and to determine the signicance of the factors. Design/methodology/approach The research method undertaken is to model productivity in construction using system dynamics. In particular, it concentrates on the use of system dynamics and project level productivity. The literature identies 34 factors affecting productivity but based on a survey of professionals, ve of these are recognised as important. They form the basis of a systems model whose development is described. Findings The results of testing the developed model have suggested that investments in planning and control have most benets on project productivity and that investments in safety, motivation and reduction of disruptions are benecial. Originality/value The use of system dynamics to model productivity represents untraditional and novel approach in research in construction. The developed model is valuable in that it can be used to evaluate management strategies and their effects on project productivity. Keywords Construction industry, Construction operations, Productivity rate, Modelling, United Kingdom Paper type Research paper

International Journal of Productivity and Performance Management Vol. 59 No. 1, 2010 pp. 18-36 q Emerald Group Publishing Limited 1741-0401 DOI 10.1108/17410401011006095

Introduction In many different aspects, construction holds the key to the prosperity of emerging and industrialised countries. It is the worlds largest and most challenging industry and also a large user of national resources and accounts for a sizeable proportion of gross domestic product (GDP) of most developed and developing countries (Tucker, 1986). Typically, in developed European countries, for example, construction might account for 10 per cent of the GDP and even higher in many developing countries (Harvey and Ashworth, 1997; Hassan and McCaffer, 2002; Chen, 1998). The provision of an infrastructure is an essential part of the development; the maintenance and change of the infrastructure to cater for the changing demands of the rapidly changing world are essential to maintain developed status. Construction, however, is expensive, dangerous

and affected by many factors both within and outside the control of the parties involved. At all levels, the client will need to balance the cost, time and quality of construction with the benets which will accrue as a result of the work. The greater the capital investment in construction is, the less the amount of money remaining to operate and maintain the facilities and to spend on other things. The lower the amount spent on construction, the slower the development, the lower the quality of the facilities provided, the smaller the benets and the smaller the opportunity for improvement. The one opportunity to break the cycle is to improve the productivity of the total construction process. By doing this, clients at all levels will benet including the economy of the countries and the world as a whole. It is not surprising then that construction productivity has formed a major research area for a considerable time. In addition to research and studies of productivity itself, the whole of construction and project management might be considered to be aimed at the improvement of productivity from some aspects. Despite all this research effort, there appears to be a broad consensus that there has been a decline in productivity in the construction industry. It is also generally accepted that there is a great deal of improvement in productivity to be achieved. Improvement in productivity will not be achieved without bearing in mind that there is an enormous number of factors affecting productivity and that there is a necessity to locate the most inuential ones among them. Doing so will enable the researchers to pinpoint the areas where efforts are to be directed in order to reach the optimum productivity of the studied project. This paper describes an attempt to model productivity of construction projects after completion of design and all the factors, which affect it as a system. The work was carried out by collecting data from civil engineering projects. However, it is thought that other types of projects will follow a similar pattern to those identied here. The sources of data are outlined and initial results from the model are presented. A considerable amount of experimentation has been carried out with the model, however, only a limited number of typical experiments are described. Construction productivity and the use of system dynamics There are many factors that affect productivity in construction projects and considerable efforts have been made to try to identify and classify these factors (Oglesby et al., 1989). The attempt to dene, measure and compare productivity of construction is considerably more difcult than might be imagined. At national level, the construction industry is different from country to country as are the construction methods, the resource availability, the laws, which govern it and the requirements made of it. These factors also change over time as expectations change. It is, therefore, not sufcient, or indeed reasonable, to measure productivity in such terms as the cost of erecting a house or the cost per kilometre of road. It is also not reasonable to remove the resource costs by dening productivity in terms of labour hours per house or kilometre of road since the expectations of the users of the facilities under construction affect their costs and the amount of work included in them. Reduction of the size of the unit under consideration from the nished article (the house, road, etc.) to the component elements is an obvious method to overcome some of these problems. The productivity might, therefore, be considered in terms of the cost of pouring a cubic metre of concrete or the number of hours taken to lay 1,000 bricks.

Construction project productivity 19

IJPPM 59,1

20

Whilst these measures might be useful to a contractor working in a specic market, they are not helpful to a government wishing to ensure that they achieve value for money for their overall construction. For example, bricks come in different sizes; methods of pouring concrete are more or less mechanised, relying on more or less capital expenditure. These latter measures also take no account of the effectiveness of management of the project and the wastage, which occurs because of inefcient deployment of resources on a project or across projects, or the design of the works. In addition each project is, to a great extent, a prototype and the construction is, to a greater or lesser degree, affected by factors, which are outside the control of the people directly involved. A project in China or in the Middle East, where cheap labour is available, is different from a project in the west where labour charges are high. The laws on safety in the workplace, the requirements to provide facilities for the workers and the laws on employment differ widely even within apparently similar countries. Take, for example, the European Union countries. Whilst they might appear to the outsider to be at least becoming a homogeneous consortium of states, the conditions of employment vary considerably giving rise to signicant variations on the actions which contracting companies can take to improve productivity. The simple example of the notice required to sack or lay-off a worker can be used to illustrate this. In the UK, it is customary to operate on a months notice for site operatives with payment in lieu of this often enforced. In other member states, the notication of redundancy can be several months with the consequent increase in costs to the employer and apparent reduction in productivity. Studying the productivity of a project, a company or an industry, therefore, requires an understanding of how this productivity is dened. Because of the complexity of the problem described above, it is suggested here to use system dynamics approach to model construction productivity. System dynamics is a methodology for studying complex systems. It is an academic discipline created in 1960s by Dr Jay Forrester of the MIT. In this eld of study, a system is dened as a collection of elements that continually interact over time to form a unied whole. The underlying relationships and connections between the components of a system is called the structure of the system. The term dynamics refers to change over time. If something is dynamic, it is constantly changing. A dynamic system is, therefore, a system in which the variables interact to simulate changes over time. System dynamics is a methodology used to study and understand how systems change over time. The way in which the elements and variables composing a system vary over time is referred to as the behaviour of the system. There seems to be a general agreement in both the academic and industrial domains on the need to look at construction differently and to switch in the direction of system thinking (Blockely and Godfrey, 2000). The application of system dynamics in the eld of project management, however, is not new. The rst project management model was developed by Roberts (1964) and was intended to examine the dynamics of research and development projects. Since then, the usefulness of system dynamics in project management has been gaining recognition. A more detailed review of this can be found in Rodrigues and Bowers (1996). System dynamics has been dened by Coyle (1996) as a method, which deals with the time-dependent behaviour of the managed systems with the aim of describing and understanding its behaviour using qualitative or quantitative models. System dynamics technique has been applied in construction management eld for modelling issues, which are important for the organisation as well as the projects.

For example, Tang and Ogunlana (2003) have used system dynamics to model factors affecting the performance of a construction rm. Other research works such as that described by Chang et al. (1991), Ogunlana et al. (1995) and Love et al. (2000) have focused on projects. The work of Ng et al. (1998), for example, used system dynamics to forecast the completion of large basement construction. Love et al. (2000) by using system dynamics to identify factors that affect schedule delay caused by design error-induced reworks. Construction productivity is certainly complex and difcult to specify precisely and its components interact with each other. Labour, project, company and industry comprise one body and they interact with each other; if any part of this body suffers from a deciency then, as a result, the whole body will suffer. Therefore, construction productivity would appear eminently suited to systems treatment. Research methodology used The model developed for this work is based on data collected from the construction industry in the UK. In collecting the information, there were three initial aims: to determine what factors affected productivity at site level, to determine how these factors interacted and to determine the signicance of the factors. Identication of the factors was initially carried out through an intensive literature search and a set of interviews with key project participants. The experts interviewed included mainly planning, senior planning and senior project engineers. Using the factors identied in this rst step, two forms were constructed. A Factor matrix which shows how factors interact with each other and a Factors-productivity form which shows how factors affect the productivity of the project. These two forms were used to survey a number of engineers in the eld to get their opinions about the relations of the system. The returned forms were then analysed to identify factors, which have major inuences on other related factors and on the construction productivity as a whole. A theoretical system model was developed to simulate a real construction project using the group of factors considered to be the most inuential factors. This model was then tested to check for any sensitivity to errors in the technique used. The model is also developed for use to evaluate some of the strategies that management might adopt in order to improve productivity in construction. Selection of factors for inclusion in the model Low or high productivity in a construction project is not the effect of a single variable, or factor, but a set of variables interacting with each other to produce the nal result. Such complexity requires a thorough study of these factors in order to understand both the individual factors and their interaction. For the purpose of this paper, the most signicant of these is the contribution by Shaddad and Pilcher (1984). They proposed that a construction project should be thought of as a function of a complex system of inter-related variables: psychological, technical, structural, environmental, personal and managerial. Their contribution to productivity studies is signicant here in that they based their view on the idea of systems. Unfortunately, this view has been largely neglected by researchers and little real work has been done to employ such theory in practice. Table I shows some of the factors which the authors have recognised as being potentially important.

Construction project productivity 21

IJPPM 59,1

22

The factors recognised as important in the literature were taken as a starting point for the systems approach in this research. In total, 34 factors were chosen for consideration although not all of them are presented in this paper. In order to produce a systems model, it is necessary to determine the importance of the individual factors, their interrelationships with one another and their effect on productivity itself. The rst two of these objectives were achieved by asking construction professionals. Unfortunately, there is a large amount of information required and a survey using traditional questionnaires or interviews would be inappropriate. In an attempt to overcome the shortcomings of the more traditional methods, the Factor matrix form shown in Figure 1 was used. This shows the size of the problem with approximately 1,000 pair-wise comparisons being asked of each respondent. In order to minimise the time required by the respondents to complete the form, the construction professionals were asked to ll in the form indicating the strength (high, medium, low or none) of the relationships that they believed existed between the factors. In the sample form, the leading diagonal was not allowed to take any value although it is recognised, for example, that absenteeism could affect future absenteeism. This form is itself quite difcult to complete and, on average, respondents suggested that it took more than an hour to complete. Despite this effort, it only provides the numerical information required for modelling by implication. The procedure followed to elicit the experts knowledge can be summarised in the following steps: (1) Convert the qualitative scale to a quantitative one. The qualitative scale of high, medium, low and no relation was converted to a number scale of 4, 3, 2 and 1, respectively. The idea of considering a score of one for a blank cell is that even though the cell is left blank, in practice, there exists a relation however small. Here, it is represented as a minor effect, which is one and not zero. (2) Normalise the forms. A major problem faced when the forms were received from respondents is that each respondent considers his own benchmark when looking at the relations in the form. A low interaction considered by one respondent may equate to medium by another. Therefore, it is important to normalise all the forms in order to judge on an equal basis. The normalisation to be done here is to nd a normalisation index for each form. The procedure used to nd this index is as follows:

Productivity factors M.1. Planning M.2. Control M.3. Good communications M.4. Material management M.5. Material movement M.6. Crew interference

References Olomolaiye et al. (1998, 1996), Shaddad and Pilcher (1984) and Rau (1988) Olomolaiye et al. (1998), Rau (1988) and Lucas and Barstad (1983) Rau (1988) Dieterle and DeStephanis (1992), Arditi (1985) and Neil (1986) Thomas et al. (1990) and Neil (1986) Thomas et al. (1990) and Maloney (1981)

Table I. Management-based productivity factors

Work inspection by engineer

Acceleration of performance

Number of foreman on site

Differing site conditions

Level of skilled labour

Design & build-ability

Interference by owner

Material management

Control of the project

Length of work day

Use of equipment

Restricted access

Communications

Learning curves

- Fill in the matrix to show the interactions of the factors. For example if you think the 'weather' has a large effect on 'absenteeism' enter H (for high) in the cell in the weather column and absenteeism row Absenteeism - Show the effects as H for high, M for medium and L for low. For no effects leave the cell blank Absenteeism Disruptions Level of skilled labour Use of equipment Over time Length of work day Number of foreman on site Crew size Motivation Learning curves Communications Design & build-ability Interference by owner Restricted access Change orders Acceleration of performance Differing site conditions Safety Work inspection by engineer Material management Control of the project Planning Modern management systems Crew interference Disruptions

Construction project productivity 23


Planning L L L H H

Change orders

Motivation

Over time

Crew size

M L

L L L L

M L L

L L L L M M M L M L M L L L

Safety

L L L M L

L L M L M L L L L

L L

L L L L L L L L L L L L L L L L L L L L L M L L L L L L

L L

M L M L L L L L L L L

L L L M M L L

L L L L L L

L L

L L L M

L L L

L L

L L

M L L L L L

L L

L M L M L L

M L M H

M M L M L L

M M M M L M L L L L

Figure 1. Extract from form used to elicit relationship information

(1) Find the total score for each respondent. (2) Find the total score for the average Factors matrix. (The average Factors matrix is the matrix which contains all the interactions of the factors at an average.) (3) Calculate the normalisation index for the form as follows: Total of the average Factors matrix from Step 2 : Total score of the form from Step 1 The 20 participants in this pilot-study research exercise were asked to assign values to the interaction between the factors. Table II shows part of the normalised and averaged results of this exercise. It shows, for example, that planning has an effect of 5.6 on the control of a project. The values have a range between 0 and 10 with 0 indicating no interaction. Several possible problems were apparent from the analysis. For example, it can be seen from the table that the respondents considered that communication

24

IJPPM 59,1

Control Planning Disruptions Motivation Safety Acceleration Learning Build-ability Work inspections Communications Change orders 0 3.58 3.37 0 0 0 0 3.73 0 0 0 0 0 0 0 0 3.49 0

Table II. Results of analysis of the collected data Change orders 4.72 4.43 5.06 4.08 0 0 0 0 4.90 4.98 0 3.62 3.7 0 0 0 4.56 4.19 4.48 3.62 0 0 0 0 4.23 4.08 0 3.78 4.37 0 3.43 0 3.70 3.93 0 4.6 0 3.47 0 0 Project size Communication Training Disruptions Technology 0 0 0 0 4.15 0 3.91 4.4 0 0 0 Buildability 3.75 3.73 0 0 0 3.45 0 0 0 0 4.67 Motivation Safety 0 0 0 0 0 0 3.7 0 0 3.56 0 4.29 3.41 0 3.36 0 0 0 0 3.31 0 0 5.60 0 4.91 3.86 4.41 3.77 0 0

Control Planning

0 4.75 4.1 3.57 3.32 0 0 0

0 0 3.88

had no effect on safety. Whilst experts in the eld might argue with this the results were used as a true reection of the industry. The respondents were also asked to provide a numerical value between 2 10 and 10 for the effect of the factors on productivity. These were also normalised and averaged. Table III shows the ve most signicant factors, which were chosen for inclusion in a system dynamics model of construction productivity. An outline system diagram for the developed system is shown in Figure 2. It can be seen that four of the factors have a positive effect an increase in planning increases productivity and only one of the signicant factors has a negative effect. Once again, experts might argue with these gures but they were used as a true reection of the views of the industrial participants. The absolute total effect of 27.3 represents approximately 20 per cent of the total effects of the factors on the productivity. However, not all the factors, which affect the productivity, can be inuenced by site management (for example, the weather affects productivity but cannot be affected by management actions). The ve signicant factors represent approximately 45 per cent of the inuence, which can be achieved by management action. All other factors are treated together as a single compound factor.

Construction project productivity 25

Signicant factors 1. Control (P1) 2. Planning (P2) 3. Motivation (P4) 4. Safety (P5) 5. Disruptions (P3) Absolute total effect

Effects on productivity 6.12 7.24 5.41 3 2 5.53 27.3

Table III. Selection of factors and their effects on productivity

Productivity Other facors +

P5 2,4,5

Safety + 1,2, 3,4 Motivation +

1,2,3, 4,6 Planning + P2

+ +

P4

Disruptions

+ Control

1,3 P3 1,2,3, 4,6 + P1

Notes: Other factors: 1 change orders; 2 size of project; 3 communications; 4 training; 5 construction technology; 6 design

Figure 2. Outline system diagram for selected factors

IJPPM 59,1

26

One further point, which should be made about the signicant factors, is that the rst four are totally controllable by site management. The fth, Disruptions is often largely outside the control of the site management. Indeed, it is often totally outside the control of the contractor and depends on the design and the clients certainty about his requirements. These factors, their interactions and their effects on productivity were used to form a systems model, which was programmed in excel. All the relationships were formulated as linear. The model was calibrated using data collected from site and from published productivity data such as the Spons series (Langdon, 2008). System representation of productivity No generally accepted satisfactory method existed for the representation of the production problem in systems style. It was, therefore, necessary in this work to develop a system representation of the problem and the required relationships in equations. This is done as follows. Each factor that affects productivity is itself affected by a number of factors. For example, in the proposed model, control at any time is a function of control in the previous period and planning, safety, disruptions and other, non-signicant variables from the current period. The value of any factor at any period of time is the sum of the following two components: (1) a percentage from the previous period of that factor; and (2) the inuence of other factors on that factor from the preceding period. This is shown in Figure 3(a). This is logical because it means that at any period of time, the value of any factor is a cumulative effect of previous periods. Planning at the current period, i.e. period t, for example, is not separated from the previous effects, and therefore, whatever value is obtained from a period, part of that value is a contribution of a preceding effect. There is no standard for how great a percentage is to be considered but it is a matter of judgment. For the purpose of the developed model, the assumed percentage is 80 for all the factors. Translating this into equation form gives the following general relationship: factorR t e factort 2 1 1 2 e factort where: factorR e factor(t) resultant value of the factor at period (t); percentage of the considered factor from the preceding period; value of the factor at the present period inuenced by other factors from the preceding period.

factor(t 2 1) value of the factor from the preceding period; and

Each factor that affects productivity is itself affected by a number of factors. This means that each factor is a function of a number of factors. On the basis of this, the equations representing the various factors are formulated accordingly. For example, in the proposed model, control at any time is a function of control in the preceding period and planning, safety, disruptions and other, non-signicant variables from the current period:

Period (t) Period (t 1) Output Factor (t 1) Input FactorR (t) Input Output Output All other factors (t 1) Input Factor (t)

Construction project productivity 27

(a) Resultant values of factors Period (t) Period (t 1) Output ProductivityR (t 1) Input ProductivityR (t) Input Output Productivity (t)

(1) Control (t) (2) Planning (t) (3) Safety (t) (4) Motivation (t) (5) Disruptions (t) (b) Inputs for productivity from preceding period

Figure 3. The concepts of the resultant values of factors and productivity from preceding periods

ControlR t ec Controlt 2 1 1 2 ec Controlt where: ControlR ec the resultant Control at period (t); a percentage of Control from the preceding period; and

Control(t) the Control value at period t as a result of other factors from the preceding period. Thus: Controlt bc ba1;2 Planning1 2 t a1;4 Safetyt 2 1 2 a1;5 Disruptionst 2 1c 1 2 bc Other factorst 2 1 where: bc a coefcient which relates the signicant factor, in this case Control, with the other factors; and 2

a1,2, a1,4 and a1,5 coefcients which relate the signicant factors, with each other.

IJPPM 59,1

Combining equations (1) and (2) gives: ControlR t ec Controlt 2 1 1 2 ec bc a1;2 Planningt 2 1 a1;4 Safetyt 2 1 2 a1;5 Disruptionst 2 1 1 2 bc Other factorst 2 1 Similarly: PlanningR t ep Planningt 2 1 1 2 ep bp a2;1 Controlt 2 1 a2;4 Safetyt 2 1 2 a2;5 Disruptionst 2 1 1 2 bp Other factorst 2 1 MotivationR t em Motivationt 2 1 1 2 em bm a3;1 Controlt 2 1 a3;2 Planningt 2 1 a3;4 Safetyt 2 1 2 a3;5 Disruptionst 2 1 1 2 bm Other factorst 2 1 SafetyR t es Safetyt 2 1 1 2 es bs a4;1 Controlt 2 1 a4;2 Planningt 2 1 1 2 bs Other factorst 2 1 DisruptionsR t ed Disruptionst 2 1 1 2 ed bd a5;1 Controlt 2 1 a5;2 Planningt 2 1 1 2 bd Other factorst 2 1

28

where bc, bp, bm, bs and bd are coefcients which denote how much effect signicant factors have on the calculated factor. Coefcient bm, for example, relates the effect of signicant factors such as control, planning, safety and disruption on motivation. Productivity in the model is considered to be the result of the summation of: . a percentage from the preceding period (t 2 1); and . the inuence of factors on productivity from the current period (t). Figure 3(b) shows the above point. Based on this, the productivity equation could be written as follows: ProductivityR t c Productivityt 2 1 1 2 c Productivityt where: ProductivityR(t) resultant Productivity at period (t); Productivity(t) c Productivity at period t calculated from the effects of factors in that period; and percentage found by experimentation. Productivityt pc Controlt pp Planningt 2 pd Disruptionst pm Motivationt ps Safetyt 9 8

where pc, pp, pd, pm and ps are coefcients which relate signicant factors with productivity.

Substitution gives: ProductivityR t c Productivityt 2 1 1 2 cpc Controlt pp Planningt 2 pd Disruptionst pm Motivationt ps Safetyt: There is no consideration in the productivity equation for other factors other than the signicant ones. These factors have only an indirect effect, which is assumed to manifest itself throughout the ve signicant factors. Coefcients in the equations and their calculations As can be seen, the transition equations produced in the previous section require three sets of coefcients: a, b and p. They were all calculated based on an analysis of the data provided by the respondents and then normalised to ensure correct relative size. These coefcients are: . Type a coefcients: these are the coefcients produced for the equations of the signicant factors. The total effect of the ve signicant factors on productivity is found by adding the productivity of all of them, i.e. P1 P2 P3 P4 P5. Values of P are obtained from the results of the Factor-productivity form as shown in Table III. The total effect of factors on productivity is, therefore, 27.3. The total effect considered here is the sum of absolute values and no consideration is given to the minus sign in Table III as this effect is already considered in the transition equations (3)-(7). To demonstrate how the coefcients are calculated, part of the whole system diagram is shown in Figure 4. This part shows four elements: planning, control, safety and disruptions. It shows planning affected by control, safety
Safety (4) a2,4 = 0.12

10

Construction project productivity 29

Planning (1)

a2,1 = 0.17

Control (1)

a2,5 = 0.14 Disruptions (5) Notes: a2,1: coefficient relates to the interaction between planning factor no. 2 and control no. 1; a2,5: coefficient relates to the interaction between planning factor no. 2 and disruptions no. 5; and a2,4: coefficient relates to the interaction between planning factor no. 2 and safety no. 4

Figure 4. Part of the systems model diagram

IJPPM 59,1

30

and disruptions. For the sake of simplicity in referring to coefcients, numbers are allocated to elements as can be seen in Figure 4 and Table IV. To calculate the coefcient a2,4, for example, the quantity of interaction between safety and planning, which is 3.41 (Table I) shown in the small box above the arrow, is divided by the total effect on productivity, i.e. a2,4 3.41/27.3 0.125. Type b coefcients relate the effects of other factors total factors, as shown in Figure 1, to the signicant factors. The procedure to calculate these coefcients involve dividing the total score of each signicant factor called Factorsignicant b the overall sum of all interactions between the signicant factors, which will be called Factortotal. For example, the coefcient relating to motivation is:   Motivationsignificant 1 15:4 1 0:5: Motivation coefficient 100 30:5 100 Motivationtotal Table IV shows the values of b coefcient for the ve signicant factors. Type p coefcients: these are the coefcients produced for the productivity equation. They relate signicant factors to productivity. The data available from the respondents in the Factor-productivity form, discussed earlier has been used for this purpose. The coefcients are obtained by dividing the effect of each signicant factor on productivity by the total effect of all signicant factor on productivity. For example, Pc coefcient for control can be calculated as 6.12/27.3 0.22 as seen in Table IV.

In addition to modelling the productivity using the equations indicated above it is necessary to determine that the behaviour of the model in known situations before it can be used in a predictive mode. However, it is impossible to check that the model is correct and it was decided to check that it gave sensible results in controlled experiments. This was done by producing a spreadsheet model and evaluating control situations, which could then be presented to and discussed with practicing engineers. Using the model to evaluate management strategies Management strategies are complex. They relate to the way the individual manager decides to operate and are affected by the managers personal qualities as well as the technical aspects of the management style. Not all these aspects are included in the strategies considered in this paper. The management of construction projects is, in general, concerned about what effect more funding has on productivity. A manager would normally seek to obtain the
1. Control 1. Control 2. Planning 3. Motivation 4. Safety 5. Disruptions 0 a2,1 0.17 a3,1 0.13 a4,1 0.12 a5,1 0.15 2. Planning a1,2 0.21 0 a3,2 0.14 a4,2 0.16 a5,2 0.18 3. Motivation a1,3 0 a2,3 0 0 a4,3 0 a5,3 0 4. Safety a1,4 0.16 a2,4 0.12 a3,4 0.12 0 a5,4 0 5. b p Disruptions coefcients coefcients a1,5 0.14 a2,5 0.14 a3,5 0.17 a4,5 0 0 bc 0.38 bp 0.36 bm 0.50 bs 0.39 bd 0.48 0.22 0.27 0.11 0.20 0.20

Table IV. The equation coefcients

maximum increase in productivity at the correct time whilst minimising the expenditure incurred. The management expenditure could, in theory, be incurred on any of the factors recognised in the data collection and analysis. In this paper, the terms funding, spending and expenditure refer to the magnitude of nancial funding, effort and time allocated by the project management in improving factors. Spending on planning, for example, does not merely imply putting more money into the project to purchase new equipment, such as new personal computers for the head ofce, but also devoting more time and effort to the planning process and giving it a higher prole within the project organisation. A management strategy, in this paper, is how and when a manager chooses to expend funds related to the management of the project. Furthermore, the strategies are limited to expenditure on any combination of the ve signicant factors. Thus, for example, a manager could operate a strategy in which planning is the only factor on which expenditure is incurred but the expenditure is maintained throughout the duration of the project. A second strategy could be to spend on planning but to only make that expenditure at the start of the project. All of the strategies are assessed in terms of the productivity predicted in the model. The productivity is assessed on a scale between 0 and 10 where 0 represents minimum productivity, 5 the average expected productivity and 10 the maximum productivity for any type of work. These three points are estimated by the manager for each type of work being considered. Knowing the amount of each type of work to be carried out, the productivity from the model can be used to determine the total true cost for carrying out the work. Illustrative experiments Two types of experiments have been carried out. The rst selected several strategies and tested their effect on productivity. These experiments were performed to determine if it was possible to give general advice to managers as to where to concentrate expenditure. The second used actual project work to examine if it was possible to determine the real savings available through the use of any strategy. The two types were called the General advice and the Savings available experiments. General advice experiments. Three sets of strategies were examined as examples of possible uses of the model: . Group 1. Considering all spending on factors to be concentrated at the start of a project: Spending on initial values of factors. . Group 2. Considering spending on factors to be in all periods: Time-based spending on factors. . Group 3. Considering more spending on situations where disruptions occur in the project at arbitrary chosen periods: Disruptions in construction projects. Each of these is intended to model a different but realistic scenario. For example, some managers expend a lot of effort on planning at the start of a project but thereafter do no planning at all. This is a Group 1 strategy. There are obviously a very large number of experiments, which could be performed to examine the behaviour of a project. Three are presented here as illustrations. An average project was used in which the initial productivity was assumed to be 5 and the project duration was considered to be 40 weeks.

Construction project productivity 31

IJPPM 59,1

32

As a control, the project was run assuming no spending on any of the signicant factors. This models the situation, which might be expected to occur when a project was not effectively managed. The results are shown as the original line. Following this, the model was run with expenditure on a number of the signicant factors. Figures 5 and 6 show the results of experiments with expenditure on the factors only at the start of the project. It can be seen that no matter what the spending on the management factors, the productivity reduces over the rst few weeks of a project. This describes a common situation at the start of many projects where the resources do not perform effectively in the initial stages of a project because of the novelty of the organisation and lack of knowledge of the project. As the project progresses, the productivity increases. This would be expected from the learning curve models. It can be seen that spending on any individual factor improves the productivity over the rst half of the project with the effect being greatest between approximately weeks three and 13. Spending on planning has the greatest effect although all factors have similar effects. It can be also seen from Figure 6 that it is possible to overcome the initial reduction in productivity by spending on all the signicant factors. Once again, however, the effect is negligible in the latter half of the project. Figure 7 shows the effect on productivity of spending on the signicant factors throughout the project. It shows that the effects are similar initially to the single initial spending situation. However, the effect lasts throughout the project and the nal
7 6 Productivity 5 4 3 2 1 0 0 5 10 15 20 Time 25 30 35 40 Control Planning Motivation Safety Original line

Figure 5. Effect on productivity of initial spending on individual signicant factors

7 6 Productivity 5 4 3 2 1 0 0 5 10 15 20 Time 25 30 35 40 Original

Figure 6. Effect on productivity of initial spending on all signicant factors

productivity is, therefore, considerably higher than that obtained with a single initial expenditure. Savings available experiments. These experiments were designed to determine if it would be possible to use the model to quantify the savings available through expenditure on the signicant factors. In simple project models, there are two types of saving available: a reduction in direct cost and a reduction in project duration. Taking the rst of these as an example, a measure of the benet to be gained by altering a factor might be considered to be:
iX 40 i1

Construction project productivity 33

P ai 2 P oi W i

where: Pai productivity in period i where the factor is altered; i number of periods; Poi productivity in period i where no factors are altered; and Wi average work in period i. So, if for example, an average of 100 m2 of formwork are to be installed in a period and the productivity goes from 2.2 to 3.02 as it does in period 6 on the planning line in Figure 1, the benet in period 6 would be: 3:02 2 2:2 8:2 m2 ; 10 100 which at y/m2 would give a benet of 8.2 y in period 6. (The divisor of 10 is required to convert the model productivity into the published productivities.) This would imply that in fact 108.2 m2 could be placed by the allocated resources. This idea can be extended over many periods or until the required amount of work has been completed. Similar experiments have been performed for projects in which there are signicant uncertainties and disruptions. The results are similar.
10 8 Productivity 6 4 2 0 0 5 10 15 20 Time 25 30 35 40 Control Planning Motivation Safety Original

Figure 7. Effect on productivity of continuous spending on signicant factors

IJPPM 59,1

34

Conclusions The research has demonstrated that it is possible to produce a systems dynamics model for construction productivity. In order to test the validity of the model, the coefcients produced for the system transition equations were tested for a variation in their values to see how the model behaved. Up to ^ 40 per cent variations in these coefcients were tried but no signicant change was reported. The model has been extensively tested and has shown that it is possible to help construction project managers to evaluate alternative strategies and consequently to determine the one to be adopted in their own particular situation. There is some evidence to suggest that the best factors on which to spend money, time and effort are planning and control. However, all the signicant factors have similar effects. There is also some evidence to suggest that it is more advantageous to spread any expenditure throughout the project rather than concentrate it at the beginning. A typical contracting organisation would have to change their modus operandi to achieve these savings and it is suggested that company and project specic experimentation is required before such action is taken. Although this work was carried out with construction contractors, it should be possible to extend it to other project-based industries.

References Arditi, D. (1985), Construction productivity improvement, Journal of Construction Engineering and Management, Vol. 111 No. 1, pp. 1-13. Blockely, D. and Godfrey, P. (2000), Doing it Differently: Systems for Rethinking Construction, Thomas Telford, London. Chang, C., Ogunlana, S. and Saeed, K. (1991), Construction project management: a system dynamics approach, Proceedings of the 1991 International System Dynamics Conference, Bangkok, Thailand, pp. 108-15. Chen, J.J. (1998), The characteristic and current status of Chinas construction industry, Construction Management and Economics, Vol. 16 No. 6. Coyle, R.G. (1996), System Dynamics Modelling: A Practical Approach, Chapman & Hall, London. Dieterle, R. and DeStephanis, A. (1992), Use of Productivity Factors in Construction Claims, AACE Transactions, Orlando, FL, pp. C.1.1-C.1.7. Harvey, R.C. and Ashworth, A. (1997), The Construction Industry of Great Britain, 2nd ed., Butterworth-Heinemann, Oxford. Hassan, T. and McCaffer, R. (2002), Vision of large scale engineering construction industry in Europe, Automation in Construction, Vol. 11 No. 4. Langdon, D. (2008), Spons Civil Engineering and Highway Works Price Book, 22nd ed., Taylor & Francis Group, London. Love, P.E.D., Mandal, P., Smith, J. and Li, H. (2000), Modelling the dynamics of design error induced rework in construction, Construction Management and Economics, Vol. 18 No. 5, pp. 567-74. Lucas, D. and Barstad, P.E. (1983), Planning for increased project productivity, Project Management Institute Seminar/Symposium, Houston, TX, October , pp. IV-K-1-IV-K-6.

Maloney, W.F. (1981), Motivation in construction: a review, The Journal Construction Division, Vol. 107 No. 4, pp. 641-7. Neil, J.M. (1986), Improving construction cost effectiveness, AACE Transactions, Vol. A12, August, pp. A.12.1-A.12.9. Ng, W.M., Khor, E.L., Tiong, L.K. and Lee, J. (1998), Simulation modelling and management of large basement construction project, Journal of Computing in Civil Engineering, Vol. 12 No. 2, pp. 101-10. Oglesby, C.H., Parker, H.W. and Howell, G.A. (1989), Productivity Improvement in Construction, McGraw-Hill, New York, NY. Ogunlana, S., Lim, J. and Saeed, K. (1995), Civil engineering design management using a dynamic model, 1995 International System Dynamics Conference, Tokyo, Vol. II, pp. 757-65. Olomolaiye, P.O., Harris, F.C. and Price, A.F. (1996), The sensitivity of bricklayers output to changes in skill, Computers and Structures, Vol. 58 No. 2, pp. 419-28. Olomolaiye, P.O., Jayawardane, A.K. and Harris, F.C. (1998), Construction Productivity Management, Addison-Wesley Longman, Harlow. Rau, A.N. (1988), Management of productivity in construction, IE(I) Journal-CI, Vol. 69, pp. 113-16. Roberts, E. (1964), The Dynamics of Research and Development, Harper & Row, New York, NY. Rodrigues, A. and Bowers, J. (1996), System dynamics in project management: a comparative analysis with traditional methods, System Dynamics Review, Vol. 12 No. 2. Shaddad, M.Y.I. and Pilcher, R. (1984), The inuence of management on construction system productivity towards a conceptual system causal research model, Proceedings of the 4th CIB W65 International Symposium on Organisation and Management of Construction, Canada, Vol. 2, pp. 613-27. Tang, Y.H. and Ogunlana, S.O. (2003), Modelling the dynamic performance of a construction organization, Construction Management and Economics, Vol. 21 No. 2, pp. 127-36. Thomas, H.R., Maloney, W.F., Horner, R.M.W., Smith, G.R., Handa, V.K. and Sanders, S.R. (1990), Modeling construction labour productivity, Journal of Construction Engineering and Management, Vol. 116 No. 4, pp. 705-26. Tucker, R.L. (1986), Management of construction productivity, Journal of Management in Engineering, Vol. 2 No. 3, pp. 148-56. Further reading Tinbergen, J. (1967), Development Planning, World University Library, London. About the authors Michael J. Mawdesley is an Associate Professor in the Department of Civil Engineering at the University of Nottingham. He is a civil engineer by education and holds a PhD in construction management. His research interests are in all aspects of the provision, engineering and maintenance of infrastructure. He has been involved in research projects in decision-support systems for allocation of land to housing, systems thinking to improve infrastructure provision, infrastructure asset management especially related to ood defences, planning and risk and the importance of human factors, funding of development projects, genetic algorithms to help solve infrastructure problems and simulation. He lectures widely on the topics and has a keen interest in the use of games and simulations for education and training.

Construction project productivity 35

IJPPM 59,1

36

Saad Al-Jibouri is an Associate Professor in the Civil Engineering Department at the University of Twente, The Netherlands. He holds a PhD in construction management from the University of Nottingham in the UK. Before starting his PhD degree, he worked as a site engineer and a project manager for many large infrastructural projects. After nishing his PhD, he carried out consultancy works for major construction companies. During his academic career, he has taught construction management for more than 25 years at both undergraduate and graduate levels and he kept close relationship with the industry. His research interests include planning and control, risk management, simulation games, performance measurement and modelling of construction processes in general. Saad Al-Jibouri is the corresponding author and can be contacted at: s.h.s.al-jibouri@ctw.utwente.nl

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

You might also like