Benchmarking, Benchaction, and Benchlearning: Rework Mitigation in Projects

Transcription

1 Benchmarking, Benchaction, and Benchlearning: Rework Mitigation in Projects Peter E. D. Love 1 and Jim Smith 2 Abstract: The Australian government has issued several calls to improve the performance of the construction industry. A lack of available benchmark metrics has made it difficult, if not impossible, for organizations to identify areas to target for process improvement. A significant factor that has been found to contribute to poor organizational and project performance is rework. Using the results of a questionnaire survey, the paper presents and discusses a series of benchmark metrics for the causes and costs of rework for 161 construction projects. A generic framework for benchmarking rework at the interfaces of a project s life cycle is proposed, and unstructured interviews are used to subject the proposed framework to validation by industry practitioners. DOI: / ASCE X :4 147 CE Database subject headings: Australia; Bench marks; Construction industry; Quality control; Government; Project management. Introduction The concept of benchmarking has received widespread application in the construction industry as a technique for identifying ways to improve organizational and project performance for example, CIDA 1995 ; Mohamed 1996 ; Marosszekey and Karim 1997 ; Love et al ; Lietal Types of benchmarking that have been used by construction organizations include project cost, occupational health and safety, labor productivity, and customer service/satisfaction Morehead et al Limited published data have been made available, which project managers and construction organizations have been able to use for comparative purposes. According to Love et al. 1999, construction organizations have been reluctant to make data obtained from benchmarking available for fear of identifying inefficiencies within their processes. While the Australian government has issued several calls for improving the performance of the construction industry Gyles 1992; DIST 1998; DISR 1999, the lack of available benchmark metrics has made it difficult, if not impossible, for organizations to identify areas to target for process improvement. A major area that has been identified as contributing to poor organizational and project performance is rework Love 2002b; Love and Sohal Previous studies have shown that it is a major contributing factor to time and cost overruns on construction projects Chan and Kumaraswamy 1997; Kaming et al. 1997; Love 2002b. Numerous studies have attempted to quantify the direct costs of rework, or its constituent components, in building 1 Professor, We-B Centre, School of Management Information Systems, Edith Cowan Univ., Churchlands, Perth, WA 6018, Australia. p.love@ecu.edu.au 2 Faculty of Architecture, Building and Planning, Univ. of Melbourne, Parkville, Victoria 3502, Australia. smithjj@unimelb.edu.au Note. Discussion open until March 1, Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and possible publication on June 6, 2002; approved on October 2, This paper is part of the Journal of Management in Engineering, Vol. 19, No. 4, October 1, ASCE, ISSN X/2003/ /$ and engineering projects for example, Burati et al ; Abdul-Rahman 1993 ; Nylén 1996 ; Josephson and Hammarlund 1999 ; Love and Li 2000 ; Josephson et al These direct costs of rework have been found to be as high as 23% of contract value Barber et al In addition, the scope of these studies has been limited to a small sample, often with only several projects, so the costs that have been reported are by no means representative. A number of factors contribute to such large variance in these reported findings. These include the extent of quality management practices implemented, the type of project, the form of procurement method used, and the level of project complexity. Moreover, there is a lack of uniformity in the way in which rework cost data have been collected due to the various interpretations of what constitutes rework Gluch and Josephson Until a degree of consistency in the definition and calculation of rework is obtained, rework costs reported in the normative literature should not be considered indicative, but used as a source for reference e.g., Barber et al Measurement of rework in itself does not cause improvement it is merely the starting point for establishing new knowledge Love and Holt The success of project management is based upon an ability to become scientific, where knowledge is characteristically acquired through systematic observation, experiment, and deductive reasoning. This paper presents benchmark metrics derived from a questionnaire survey for the causes and costs of rework on 161 Australian construction projects. While the research was conducted in an Australian context, the research outcome is envisaged to be widely applicable in other locations. A generic framework for benchmarking rework at the interfaces of a project s life cycle is proposed, and the authors use unstructured interviews to subject the framework to validation by industry practitioners. Prior to the presentation of the research in this paper, a brief review of the causes and costs of rework in projects is presented to provide the context for the research. Rework in Projects Several definitions of rework can be found in the literature. For example, Ashford 1992 defines rework as the process by which JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003 / 147

2 an item is made to conform to the original requirement by completion or correction. The Australian Construction Industry Development Agency CIDA, however, defined rework as doing something at least one extra time due to non-conformance to requirements CIDA A repair, which is the process of restoring a nonconforming characteristic to an acceptable condition even though the item may not still conform to the original requirement Ashford 1992, can also be included as rework. Bearing in mind these definitions of rework, a repair in some instances may not actually constitute rework if it remains a nonconforming item. Changes, errors, omissions, quality failures, and defects can result in work having to be rectified Farrington 1987; Love and Li Yet previous research has tended to report only specific attributes of rework, such as those relating to defect, changes, error, or failure, rather than investigating it as a totality. In fact, some researchers have used terms such as defect, quality failure, and error interchangeably to denote rework Josephson et al even though these terms have distinct differences in definition. Causes and Costs of Rework The Building Research Establishment BRE in the United Kingdom found BRE 1981 that 50% of errors in buildings had their origin in the design stage and 40% in the construction stage. The National Economic Development Office NEDO conducted a survey NEDO 1987 that aimed at identifying ways of improving quality control in building work. The main factors deemed to influence quality were design e.g., lack of coordination of design, unclear and missing documentation and poor workmanship for example, lack of care and knowledge. Findings of a further study NEDO 1988 were almost identical to those of the previous year s study. Research undertaken by Cnudde 1991 determined the failure costs in construction by investigating the amount of nonconformance that occurred on site, finding that nonconformance cost was between 10 and 20% of the total project cost; furthermore, 46% of total deviation costs were created during design, compared to 22% ascribed to construction deviations, which were due to poor execution of work. Cnudde 1991 suggests that clients are responsible for ensuring quality and economy in their projects, and in particular that clients can improve the quality of the project delivery process by Clearly defining their requirements; Selecting competent project personnel; Selecting a suitable procurement method; and Monitoring the quality of all project participants. Cusack 1992 suggested that the major cause of rework in a project was related to documentation errors, noting that projects without a quality system in place typically experience a 10% cost increase because of rework. Burroughs 1993 reported that a major Australian contractor had experienced rework costs of 5% of contract value in one of its major projects that were attributable to poor documentation by design consultants; furthermore, the concreting subcontract experienced a cost increase of 31% due to rework. Gardiner 1994 estimates that the costs related to the rework of design consultants could be as high as 20% of their fee for a given project. Koskela 1992 suggests that it sometimes seems that the wastes caused by design are larger than the cost of design itself, and he further states that even if there is a lack of data on internal waste in design, it can be inferred that a substantial share of design time is consumed by redoing or waiting for information and instructions. Rounce 1998 has suggested that much of the design-related rework generated in projects is attributable to poor managerial practices of architectural firms. In particular, Powell 1997 has suggested that the insularity and aversion of architectural firms to management has resulted in poor service quality and their marginalization within the industry. While architectural firms have been identified as the primary source of the design-related rework experienced in projects, other firms, such as consulting engineers, contractors, and project managers, who are integral to the procurement of projects, are also prone to implementing poor managerial practices Love and Sohal Hammarlund and Josephson investigated the sources of quality failures in a building project and found the costs of correcting failures to be 6% of production costs. Moreover, the time taken to rectify these errors was estimated to be 11% of the total working hours allocated for the project. Hammarlund and Josephson suggested that a large part of the failure costs were attributable to the poor skills of site management. Their study found the major causes of quality failures in order of precedence to be defective workmanship, defects in products, insufficient work separation, inadequate construction planning, disturbances in personnel planning, delays, alterations, and failures in setting out and coordination. Quality failures that have occurred after a project has been completed have been estimated by Hammarlund and Josephson to be as high as 4% of the actual project s production cost. Interestingly, 51% of these failure costs were found to be design related, while 26% were related to poor installation of materials and 10% to material failure. Abdul-Rahman 1993, in a study of nonconformances in a water treatment plant, identified 62 nonconformances that accounted for 2.5% of contract value, although not all nonconformances could be identified due to study resource constraints and availability of site research personnel he therefore stated that, assuming that the rate at which the cost of nonconformances occur is constant throughout construction, then the total cost of nonconformance is estimated to be 6% of the estimated project cost. This figure, however, did not reflect the full extent of rework that occurred, as many client-initiated variations could not be included in the analysis. Abdul-Rahman 1993 also revealed that design errors or omissions contributed to 30% of the cost of nonconformances; the major construction-related costs of nonconformance were ascribed to the subcontractor, coordination, planning, and construction. The same study examined the incidence and costs of nonconformance in a highway project; 72 nonconformances were identified, 59 of which were used in the analysis because those excluded were of a trivial nature and did not carry costs. Nonconformances were found to account for 5% of the contract value and were primarily attributable to subcontractor-, construction-, and design-related issues. Burati et al collected data on quality deviations from nine industrial engineering projects and found that the cost of quality deviations could be as high as 12.4% of project cost; furthermore, 79% of total deviation costs were created during design, compared to 17% for construction deviations. Such deviations in cost could be even higher because they do not represent schedule delays, litigation costs, and other intangible costs of poor quality. Davis and Ledbetter 1987 have noted that rework costs are only the tip of the iceberg, as many other impact costs are associated with rework and lack of quality management. CIDA 1995 found that poor communication, the use of traditional lump-sum procurement methods, and the lack of a formal quality management system were major factors that contributed to 148 / JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003

3 rework in Australian projects. They reported that rework costs for traditional lump-sum methods exceeded 15% of a project s contract value, and, moreover, those projects that had a formal quality management system in place recorded significantly lower levels of rework. While previous research has sought to determine the direct tangible costs of rework, the indirect intangible costs remain unexplored in construction Love 2002b. This is because it is difficult, if not impossible, to quantify such costs in pure monetary terms. Bowersox et al estimates that the cost of rectifying a poor quality product or service can be more than eight times its original cost. Accordingly, Love 2002a reported that a multiplier effect of at least five was directly related to the indirect effects of rework incidents in construction. Josephson and Hammarlund 1999 found the cost of defects for seven Swedish building projects ranged from 2.2 to 9.0% of the contract value. They also found that 50% of the total costs of defects originated on site, with a further 32% originating from the client or design inconsistencies, and that the root cause for 40% of defect costs was poor motivation levels in the site workforce. Benchmarking Before rework can be significantly reduced and the overall performance and output of a project improved, a series of benchmark metrics for rework should be established to act as performance measures. For benchmarking to be effective at a project, at industry, or competitive level, organizations should view it as a process of improvement rather than of exposing an organization s weakness. Furthermore, benchmarking should be viewed as an operational process of continuous learning and adaptation that results in development of an improved organization. At a project level this can be a difficult and arduous task because of the temporary nature of projects. Organizations should essentially try to take the lessons learnt from one project and transfer them to the next project they undertake, as the processes by which projects are procured are not subject to constant change. It is merely the product that differs, not the process. Benchmarking is a suitable starting point for construction organizations embracing change and for learning, as it is a process that can be used to systematically acquire knowledge about the causes, costs, and impact of rework. Benchmarking focuses on the importance of understanding the tasks and activities in a process that produce an outcome and how improvements in processes can take place Schaffer It is used to continuously measure an organization s products and services against those of its competitors so that targets and priorities for gaining a competitive advantage can be established and the search for industry best practice initiated Oakland and Sohal According to Zairi 1996 a benchmark can be Anything taken or used as a point of reference or comparison; Something that serves as a standard by which others may be observed or compared; Anything or something that is comparatively measurable; and A physiological or biological reference value against which performance is compared. The establishment of a series of rework benchmarks at specific interfaces in a project could be used to enable those industry clients who procure buildings on a regular basis and project participants to compare their own rates against what has been found to be best practice performance. Research Methodology The research presented in this paper is part of a wider study that sought to determine the influence that project attributes, organizational management practices, and project management practices have on rework costs in construction projects. For the purposes of the research presented in this paper, and as noted earlier, rework is defined as the unnecessary effort of re-doing a process or activity that was incorrectly implemented the first time Love 2002b. Rather than developing a questionnaire survey that sought respondents general opinions about rework, respondents were asked to select a recently completed project most familiar to them and to answer questions about the perceived causes of rework, the associated costs, and the project management practices implemented. In essence, each project identified by respondents was treated as a separate case. Questionnaire Survey A questionnaire was developed that contained 116 variables with which to examine the influence of project characteristics, organizational management practices, and project management influences on rework costs in projects. The research reported in this paper focuses on the factors that caused rework and its costs. Respondents were asked to provide an estimate of the rework costs direct and indirect incurred in the project they had selected. Using a 5-point Likert scale, respondents were asked to indicate their extent of agreement as to how much the identified factors contributed to the occurrence of rework in their selected project. Respondents were also provided with an opportunity at the end of the questionnaire to provide supplementary information about why rework occurred in their selected project. Stratified random sampling was used to select the study sample from the telephone directory Yellow Pages. In addition, to increase the representativeness of the samples, stratified random sampling was a useful technique that made general statements about the portions of the population possible. Prior to determining the sample size for the main study, a pilot survey was undertaken with 30 selected firms, which comprised architects, project managers, and contractors from the Geelong and Melbourne region, in the state of Victoria, Australia. This was undertaken to test the potential response rate, suitability, and comprehensibility of the questionnaire. Each firm was contacted by telephone and informed of the aims of the research. On obtaining the firms consent, the questionnaire was mailed, with a stamped addressed return envelope enclosed, for respondents returns, comments, feedback, and completion. The respondents were also asked to critically review the design and structure of the survey. All comments received were positive, and as a result the questionnaire remained unaltered for the main survey. Twenty-five responses were received in the pilot survey, which equates to a response rate of 83%. The composition of respondents that returned the questionnaire was architects 30%, contractors 50%, and project managers 20%. In the main survey, 70 questionnaires were mailed to each of the aforementioned categories of respondents and to consultant engineers and quantity surveyors QSs throughout Australia, which equates to a total of 420 questionnaires distributed; 136 valid responses were received from the main survey. As no changes were required to the pilot questionnaires, they were added to the sample, which resulted in 161 valid responses and a total consolidated response rate of 36%. This response rate was considered acceptable for a survey focusing on gaining responses from industry practitioners Alreck and Settle Fig. 1 provides a breakdown of the valid responses by respondent type, showing that contractors, architects, and consultant project managers accounted for approximately 81% of the respondents. While QSs and structural, mechanical, and electrical engineers may JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003 / 149

4 seem underrepresented, it should be noted that many consultancy firms offer project management services and as a result may have undertaken the role of project manager for a project they selected. As response rates from the QSs and the engineering profession were low, the respondents were recategorized under the following headings for analysis purposes: Design consultants, consisting of architects, QSs, and structural, mechanical, and electrical engineers 44% ; Contractors 33% ; and Consultant project managers 23%. Data Reliability Data reliability is related to the data source and the identification of the position held by the person who completed the questionnaire Oppenheim It was therefore important that the personnel who had detailed knowledge about the procurement processes associated with a project answered the questionnaire. The questionnaire was mailed to the senior personnel within the organizations identified; whether they actually answered the questionnaire was impossible to determine, except in circumstances where respondents optionally supplied their details at the end of the questionnaire. From the 161 questionnaires used in the research, 87 respondents provided their business details, revealing that many respondents held senior positions within their organizations. Based on the respondents position titles, the direct mailing to individuals in organizations seemed to have achieved its objective of reaching those who were closely involved with delivering construction projects. In addition, as the questionnaires were mailed to organizations in different states in Australia, this minimized duplication of selected projects. Fig. 2 provides a breakdown by state of the respondents who answered the questionnaire. Considering the number of construction projects being undertaken in Australia at any given time, the likelihood of the respondents selecting the same project was significantly reduced because of the diversity of data sources from each state. Analysis Fig. 1. Respondents by profession The data collected were analyzed using the Statistical Package for the Social Sciences SPSS for Windows, Version Prior to undertaking the detailed analysis, each of the constructs used to Fig. 2. Firms by state identify the causes of rework within the research instrument was tested for reliability, and each research instrument was evaluated using Cronbach s coefficient alpha. An value of 0.70 or above indicates a reliable measurement instrument Nunnally 1978 ; the levels for each of the constructs examined in this paper are presented in Table 1. Descriptive statistics were used to provide benchmark metrics for rework costs for different project types and procurement methods. Information obtained from the respondents about the causes of rework in their selected projects was used to develop a rework index RI, which can be used by project managers and the like to provide benchmark metrics for evaluating the performance of projects at the post mortem phase. In calculating the RI, the mean and standard deviation of each individual factor were not considered appropriate statistics for evaluating the overall rankings, as no relationship was found between them. Thus all the numerical scores for each identified rework cause were transformed into indices to assess the relative rankings of the factors Okpala and Aniekwu 1988; Olomolaiye et al. 1987; Shash 1993; Holt 1997; Love et al The RI was calculated using the following formula: AN, 0 RI 1 where w weighting given to each factor by the respondent, which in this case ranged from 1 to 5, where 1 is not important and 5 is extremely important; A highest weighting 5 in this case ; and N total number of respondents. One-way analysis of variance ANOVA was used to compare the means of respondents estimates for rework costs and to determine if there were any reliable differences among them. The Kruskal Wallis test, a nonparametric equivalent to the ANOVA, Table 1. Cronbach s Coefficient Construct level Design cost causes Construction cost causes Project cost source Productivity Client causes Design team causes Site management causes Subcontractor causes / JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003

5 Table 2. Procurement Methods Used to Deliver Project Types Project type New build Refurbishment/ renovation Fit-out New build/ refurbishment Combination of all Total Procurement method Number % Number % Number % Number % Number % Number % Traditional lump sum Traditional cost plus Traditional with provisional quantities Design and manage Construction management Management contracting Design and build Novation Turnkey/package deal BOOT Total was undertaken to test whether there was a difference between group respondent type, project type, and procurement method rankings of the independent variables. This test was undertaken because variables had a continuous distribution and were measured using an ordinal scale of measurement. To interpret the output from the Kruskal Wallis test, it is important to look at the Chi-square and degree of freedom, which have been corrected for ties. These values are used to indicate whether there is a difference between groups; if the value of p 0.05, then the difference between groups is significant. Findings and Discussion Rework Costs Rework costs are very rarely, if ever, measured by Australian construction organizations Love et al. 2000a, so the estimates provided by respondents were based on their subjective evaluations and personal knowledge of the project. Tables 2 and 3 provide a summary of facility types and the procurement methods used in the projects sampled. The total rework costs were calculated by adding the direct and indirect estimates provided by the respondents. The mean M and standard deviation SD of total rework costs for the 161 construction projects were 12.0% and 13.56%, respectively. Respondents estimates for direct rework costs were M 6.4%, SD 7.78%, and for indirect rework costs, M 5.6%, SD 7.19% Table 4. The total costs of rework were found to vary considerably between projects Tables 5 and 6. Some respondents reported rework costs to be less than 1% of a project s original contract value, while others have reported them to be as high as 80%. The degree of variability in the estimates given by the respondents suggests that many respondents may be unsure about the actual costs of rework incurred in the projects they selected. To test whether differences between the estimates of the respondents for rework costs direct and indirect were significant, an ANOVA was undertaken. The descriptive statistics revealed differences between the design consultants (M 8.0%, SD Table 3. Type of Facility and Project Procured Project type New build Refurbishment/renovation Fit-out New build/refurbishment Combination of all Total Facility type Number % Number % Number % Number % Number % Number % Administrative authorities Administrative civic Banks Educational school Educational university Entertainment Hotel/motel/resort Hospitals/health Commercial recreational Commercial retail Commercial offices Industrial warehouses Industrial factory Residential Airport Total JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003 / 151

6 Table 4. Direct and Indirect Rework Cost Estimates by Respondent Type Respondent N Mean deviation Direct rework costs error Minimum Maximum Mean deviation Indirect rework costs error Minimum Maximum Design consultant Contractor Project manager Total ), contractors (M 5.8%, SD 7.2%), and project managers (M 4.3%, SD 3.9%) in their estimates of direct rework costs. Levene s test of homogeneity of variances was violated (p 0.05), which indicates that the population variances for each respondent type were not equal. The ANOVA revealed significant differences between respondents estimates of direct rework costs, F (2,158) 3.028, p The results of Tukey s HSD post hoc test indicated that the differences in the estimates for direct rework costs were between design consultants and project managers (p 0.05). Consultant project managers typically act as the client s representative for projects and thus would possess reasonable knowledge to be familiar with the direct rework costs in the form of change orders and defects as they invariably manage time and cost schedules. Project managers, however, may not be aware of the direct rework costs associated with redocumenting aspects of the project after clients have requested design changes and/or omissions. While client involvement in projects has been identified as a factor that can contribute to project success Walker 1994, their involvement may also contribute to the occurrence of rework. For example, drawing on the qualitative comments provided by the respondents, a project manager stated that the client was a decision-maker and actively involved in construction, resulting in scope and design changes throughout the construction. This in turn can lead to design consultants having to redocument or provide additional documentation, which can significantly affect their fee, as they are often not reimbursed for this service Tilley and McFallan Several design consultants articulated this point, with one stating that reworking of documentation is becoming a common occurrence on projects, which is not reflected in our fee. Another design consultant stated that a lot of rework had to be done on the documentation to reduce the scope of packages or substitute materials in an attempt to get the project within the budget. Redocumentation due to design changes and omissions initiated by clients and end-users appears to be a regular occurrence in Australian projects Love et al. 2000b; Tilley and McFallan The allocation of resources and planning of the documentation process are important points that need to be addressed if rework is to be reduced Love et al. 2000a. Yet understanding why estimates of direct rework costs differ is a major research task in itself and thus worthy of further investigation, although it would appear that consultant project managers might not fully understand how changes/omissions could affect the performance of design consultants particularly the way in which they manage the documentation process and thus explain the differences in the estimates for direct rework costs. The descriptive statistics revealed differences between the design consultants (M 6.77%, SD 8.34%), contractors (M 5.46%, SD 6.87%), and project managers (M 3.64%, SD 4.37%) in their estimates of indirect rework costs. Levene s test of homogeneity of variances was not violated (p 0.05), which indicates that the population variances for each group were equal. The ANOVA test revealed no significant differences between respondents estimates for indirect rework costs, F (2,158) 2.364, p It is noteworthy that design consultants estimates are almost twice those of project managers, which again demonstrates the variability associated with rework cost estimates, albeit not for the same projects. For both direct and indirect rework costs, contractors estimates are approximately midway between those of design consultants and project managers, which implies that contractors may well have a better understanding of actual rework as they invariably operate at the implementation interface between design and construction. When respondents were asked to compare the rework costs of their selected project with others in which they had been involved, 12% stated that the costs were comparable to a very large extent, 16% to a large extent, 37% to some extent, 26% to a minor Table 5. Direct and Indirect Rework Costs for Procurement Methods Used Procurement method N Mean deviation Direct rework costs error Minimum Maximum Mean deviation Indirect rework costs error Minimum Maximum Traditional lump sum Traditional cost plus Traditional with provisional quantities Design and manage Construction management Management contracting Design and build Novation Turnkey and package deal BOOT Total / JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003

7 Table 6. Direct and Indirect Rework Costs for Project Types Project type N Mean deviation Direct rework costs error Minimum Maximum Mean deviation Indirect rework costs error Minimum Maximum New build Refurbish/renovate Fit-out New build/refurbish Combination of all Total extent, and 9% not at all. The mode was found to be to some extent, and thus one can conclude that the estimated rework costs reported are generally representative of industry practice in Australia. The rework costs found in this research appear relatively comparable with those in previous studies Burati et al. 1992; Abdul- Rahman 1995; Nylén 1996; Josephson and Hammarlund 1999; Love and Li However, with the exception of Barber et al. 2000, such studies fail to differentiate or acknowledge a distinction between direct and indirect costs. Thus the mean rework costs direct and indirect reported are considered benchmarks that can be used to pursue best practices in the Australian construction industry. Design and Construction Cost Causes Table 7 identifies the mean rankings and rework index RI for design cost causes. It can be seen that the variable changes made at the request of the client is the principal design cost cause of rework in the project sampled (RI 0.645). Moreover, errors in contract documentation was ranked second (RI 0.599), omission of items from the contract documentation was ranked third (RI 0.571), and changes made at the request of an end-user/ regulatory body fourth (RI 0.477). The mean rankings and RI for construction cost causes are presented in Table 8. Here it can be seen that changes initiated by the client/occupier after work had been completed were the principal cost cause in the projects sampled (RI 0.549). In addition, changes initiated by the client/occupier after a product or process had been completed were ranked second (RI 0.478). Hence it can be concluded that during construction client and occupier changes are the primary cost source of rework, though in the cases sampled the extent to which these events occurred was not considered significant. However, changes in construction due to site conditions were ranked third (RI 0.471) and changes in the construction method due to site conditions fourth (RI 0.420), and thus these were also factors that contributed to rework costs. A significant difference was revealed between design cost causes for revisions, modifications of the design initiated by the contractor/subcontractor for different procurement methods used ( , p 0.04). Nontraditional methods, particularly design and construct D&C arrangements, tend to encourage the greater involvement of contractors and subcontractors in the design process, and additional changes may occur because of their input. Yet their involvement in the design process is often too late, and as a result their recommendations to improve constructability and save costs may in fact result in design-related rework being undertaken, which is something that design consultants tend to dislike. In the case of construction cost causes, no significant differences were identified for different procurement methods ( p 0.05). Likewise, no significant differences between design and construction cost causes and project type were identified ( p 0.05). Differences found in respondents rankings for design cost causes of rework (p 0.05) were omission of items from the contract documentation ( , p 0.00) and errors made in contract documentation ( , p 0.01). Design consultants generally considered contract documentation adequate for its intended purpose, with one design consultant stating This project was successful as the architectural principal led the project, its design and documentation, and post contract administration. This included client briefing. The cost overrun was small as a result of a well designed project. In one of the selected projects, a design consultant apportioned the entire responsibility for causing rework to the contractor by stating Most rework resulted from the contractor not supervising tradesman who assumed they should do it work their normal way. The documentation was thoroughly prepared and clear.... In spite of the likely bias of designer consultants, both project Table 7. Design Cost Cause Index and Ranking for Rework Mean Contractor Project manager Design consultants Design cost causes Index Rank Index Rank Rank Index Rank Index Change s made at request of contractor during construction Change s made at request of client Change s initiated by end-user/regulatory bodies Revision s, modification s, and improvement s of the design initiated by the contractor or subcontractor Errors made in contract documentation Omission s of items from contract documentation JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003 / 153

8 Table 8. Construction Cost Causes Index and Ranking for Rework Mean Contractor Project manager Design consultants Construction cost causes Index Rank Index Rank Rank Index Rank Index Change s in method of construction to improve constructability Change s in construction methods due to site conditions Change s initiated by client or occupier after some work had been undertaken on site Change s initiated by client or occupier after some work had been completed Changes made during manufacture of product Change s initiated by contractor to improve quality Error s due to inappropriate construction methods Omission s of some activity or task Damage caused by subcontractor managers and contractors considered contract documentation to be problematic. For example, one project manager stated that the design documentation was pathetic and incomplete at the time of tender. In addition, a contractor stated The project had four to five months design period. Very little work was done in that time on cost planning, subcontractor practices, design coordination meetings, control of documentation and all the risk was put on to the contractor. The project overran over budget. It was estimated that we the contractor lost $225,000 on the project. The project that the contractor was referring to was a Design and Construct D&C project, and hence the contractor was deemed to bear the associated project risks. Yet, there is no obvious explanation of why designers produce contract documentation with errors and omissions. A plausible explanation as to why documentation is so poor may be attributable to the fees they are charging, as a design consultant stated The consultants fee was 0.6% of the total job cost for complex work. A low fee was needed to win the work. The correct fee including construction assistance should have been 2.5%. Irrespective of the fee, design consultants should act in a professional manner if they offer to do the required work voluntarily or by bidding and provide a quality service to their clients. For example, recent research undertaken by Hoxley 2000 in the United Kingdom has clearly demonstrated that competitive fee tendering does not have a negative impact on professional service quality. Client Causes Table 9 indicates the mean rankings and RIs for each client cause of rework: Here the lack of experience and knowledge of the design and construction process was ranked first (RI 0.605), payment of low fees for preparing the contract documentation second (RI 0.590), poor communication with design consultants third (RI 0.589), inadequate time and money spent on the briefing process fourth (RI 0.560), lack of funding for site investigations fifth (RI 0.475), and lack of client involvement in the project sixth. Poor communication between clients and design consultants was identified as a cause of rework; this is because most clients only ever build once and therefore would be unfamiliar with their role in the design and construction process. With this in mind, practitioners should be better at educating and informing their clients about the process involved in procuring a building, as well as managing their involvement in the project. In fact, greater involvement of clients, particularly in the design process, may enable them to understand the processes associated with the construction of their building, thus reducing the amount of changes that may occur. Clients who are inexperienced in procuring building often tend to pay low fees and also allow insufficient time to be spent on the briefing process Tilley and McFallan 2000, which may contribute to omissions and errors in documentation occurring in projects. Rework may not always result in a financial cost to a project; in some instances it can be seen in a positive light, as significant cost savings can also be made. For example, changes in the method of construction to improve constructability may result in Table 9. Client Cause Index and Ranking of Rework Mean Contractor Project manager Design consultants Client causes Index Rank Index Rank Rank Index Rank Index Lack of experience and knowledge of design and construction rocess Lack of funding allocated for site investigations Lack of client involvement Inadequate time and money spent on briefing process Poor communication with design consultants Payment of low fees for preparing contract documentation / JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003

9 Table 10. Design Team Cause Index and Rank for Rework Mean Contractor Project manager Design consultants Design team causes Index Rank Index Rank Rank Index Rank Index Ineffective use of quality management practices Inadequate client brief to prepare detailed contract documentation Ineffective use of information technologies Poor coordination between design team members Time boxing Poor planning of workload Lack of manpower to complete required tasks Staff turnover/allocation to other projects Incomplete design at time of tender Insufficient time to prepare contract documentation cost savings for the client. In fact, in many D&C projects, where a shared saving incentive scheme is in place between contractor and client, the contractor will invariably aim to identify cost savings so they can maximize their profit. However, rework in this instance may result in design consultants having to redocument, which may not have a direct impact on project cost but on the costs to the design firm instead, as fees for redocumentation are difficult to recover Love 2002a. No significant differences were revealed between client-related causes and procurement methods and respondents perceived causes of rework (p 0.05), but there were, however, significant differences between client-related causes, namely the lack of funding allocated for site investigations and the project types ( , p 0.03). The significant differences were attributable to fit-out projects, as respondents in these projects generally disagreed that inadequate funding had been allowed for site investigations. Design Team Causes Table 10 indicates the mean rankings and RIs for each of the design-team causes of rework. The variables considered the primary causes of design team-related rework were ineffective use of IT (RI 0.652); incomplete design at the time of tender (RI 0.614); insufficient time to prepare contract documentation (RI 0.614); inadequate client brief to prepare detailed contract documentation (RI 0.591); and time boxing curtailing documentation due to lack of time (RI 0.583). The results in Table 8 show differences between respondents as to the perceived design team causes of rework; notably in respondents rankings for the ineffective use of information technologies. Other differences were respondents perceptions as to the poor coordination of design team members in the projects they had selected. Both designers and contractors ranked poor coordination of design team members first, whereas project managers ranked it fourth. Similarly, differences can be seen among respondents in the ranking of time boxing and incomplete design at the time of tender. The analysis revealed significant differences between respondents with respect to the ineffective use of quality management practices ( , p 0.01), time boxing ( , p 0.02), incomplete design at the time of tender ( , p 0.02), and poor coordination between design team members ( , p 0.02). The ineffective use of quality management practices was ranked lower than the aforementioned variables and thus was not considered a significant cause of rework. It would appear that the design causes of rework in projects are exogenous to design consultants, and consequently their quality procedures and systems are unable to control these variables. Despite the changes imposed on firms, they do have control over the quality management systems and management practices used to produce contract documentation. Incomplete design at the time of tender was identified as a prominent cause by contractors and project managers, but not by design consultants. This would, however, appear to be the case in most projects, as there is increasing pressure for firms to maximize their fees and thus revert to time boxing. This practice places enormous pressure on staff, particularly if the workload is not well planned and the project has a high degree of staff turnover. Naturally, if such practices prevail, rework will be inevitable. No significant differences were found between procurement methods and design team causes of rework (p 0.05). Thus the design processes associated with the procurement of projects whether they are sequential or overlapped through fast tracking are not responsible for causing rework, but rather the ability of design consultants to adapt effectively to their external environment. Essentially, effective client management is the key to reducing change. A project manager stated that, in order to reduce rework Good project management of the design team and the contractors are essential. Contract terms and conditions can minimize the effect of rework by placing the responsibility on the contractor, but it is the client instigated change that is the most disruptive. Arguably, good project management is fundamental, but the problem here is defining what good project management means in this context. Placing all the risk and responsibility on the contractor for rework will neither eliminate the problem nor be considered equitable, as most rework originates in the design phase of projects. Site Management Causes Table 11 identifies the mean rankings and RI for the site management causes of rework. The mean rankings reveal that poor planning and coordination of resources was ranked first (RI 0.593) and thus was considered the primary cause of rework. Ineffective use of quality management practices was ranked second (RI 0.576), setting out errors third (RI 0.527), staff turnover/ JOURNAL OF MANAGEMENT IN ENGINEERING ASCE / OCTOBER 2003 / 155