1、Measuring the Impact of Rework on Construction Cost Performance Bon-Gang Hwang1, Stephen R. Thomas, M.ASCE2, Carl T. Haas, M.ASCE3, and Carlos H. Caldas, M.ASCE4 1 Assistant Professor, Dept. of Building, National Univ. of Singapore, Singapore 117566. E-mail: bdghbgnus.edu.sg 2 Associate Director, Co
2、nstruction Industry Institute, Austin, TX 78759-5316. E-mail: sthomasmail.utexas.edu 3 Professor, Dept. of Civil Engineering, Univ. of Waterloo, Waterloo ON, Canada N2L 3G1. E-mail: chaascivmail.uwaterloo.ca 4 Assistant Professor, Dept. of Civil, Architectural, and Environmental Engineering, Univ. o
3、f Texas at Austin, Austin, TX 78712-0276. E-mail: caldasmail.utexas.edu (Accepted 23 September 2008) Introduction Construction projects often experience cost and schedule overruns and rework is a significant factor that directly contributes to these overruns. Research by the Construction Industry In
4、stitute (CII) reveals that direct costs caused by rework average 5% of total construction costs (CII 200511). Considering that the U.S. construction industry expended $1,502 billion in 2004 for total installed costs (Bureau of Economic Analysis 20065), almost $75 billion was wasted on direct costs c
5、aused by rework in that year alone. Therefore, rework must be considered a significant factor affecting cost performance in the construction industry.Several research efforts (OConner and Tucker 198623; CII 19897; Davis et al. 198912; Burati et al. 19924; Love et al. 1999a20, b21; Love 2002b16; Faye
6、k et al. 200313; Love and Edwards 200417) have attempted to identify and classify the root causes of rework, and to quantify its overall extent. Employing the metric, total field rework factor, and the classification of rework sources developed by CII, this paper assesses the direct impacts of rewor
7、k on construction cost performance using data from 359 actual projects. More specifically, the objectives of the research described in this paper were: (1) to identify the impacts of rework on construction cost performance for various characteristics of projects; (2) to determine the impacts of diff
8、erent sources of rework on construction cost performance; and (3) to isolate the root causes of rework and recommend possible solutions for those causes.By comparing the impacts of rework according to project characteristics and by measuring sources of rework, those projects most affected by rework
9、are identified. Additionally, those sources of rework having the biggest impact on construction cost performance are discussed. After the analysis of the cost impact of rework is summarized, the root causes of rework will be assessed and possible solutions can be suggested.The recognition of the var
10、ious impacts of rework is important for project managers. For those projects on which cost tends to be more affected by rework, project managers should focus on minimizing rework by developing systems for addressing the sources of rework. Preproject and quality management plans should be drafted wit
11、h an understanding of the causes of rework in order to minimize its impact. This paper provides an understanding of the impact of rework on construction cost performance, thus helping to reduce rework and improve project cost performance.Background topAccording to Love (2002b)16 rework has various d
12、efinitions and interpretations within the construction management literature: terms for it include “quality deviations” (Burati et al. 19924), “nonconformance” (Abdul-Rahman 19951), “defects” (Josephson and Hammarlund 199914), and “quality failures” (Barber et al. 20003). Love et al. (2000)22 charac
13、terize rework as the unnecessary effort of redoing a process or activity that was incorrectly implemented the first time. Similarly, field rework is defined as activities that have to be done more than once or activities that remove work previously installed as part of a project (CII 20018). Based u
14、pon CIIs definition, Fayek et al. (2003)13 proposed a definition of rework that adds the constraint that rework caused by scope changes and change orders from owners should not be classified as rework. In the sense of conformance, there are two main definitions of rework (Love 2002b16; Fayek et al.
15、200313). The first definition is that rework is the process by which an item is made to conform to the original requirements by completion or correction (Ashford 19922). The second definition given by the Construction Industry Development Agency (1995)6 holds that rework involves doing something at
16、least one extra time due to nonconformance to requirements. Although the wording of the definitions and interpretations of rework vary, there is a common themerework means having to redo work due to nonconformance with requirements.Several studies have explored the cost of rework in the construction
17、 industry. Research conducted by CII reports that direct costs caused by rework average 5% of total construction costs (CII 200511). Josephson and Hammarlund (1999)14 estimated that the cost of rework on residential, industrial, and commercial building projects ranges from 2 to 6% of contract values
18、. Similarly, Love and Li (2000)19 found that the costs of rework for residential and industrial building projects are on average 3.15 and 2.4% of the contract values, respectively. The nonconformance costs (excluding material wastage and head office overhead) of a highway project are estimated to be
19、 5% of the contract value (Abdul-Rahman 19951). These authors suggest that nonconformance costs may be significantly higher on projects where poor quality management is found. The potential for such significant losses make it critical that rework costs should not be overlooked in efforts to improve
20、project cost performance.To manage rework, it is first necessary to identify and classify its causes. Many analysts have suggested that rework is often due to the complicated characteristics of the construction processes. By distinguishing between engineering rework and construction rework, OConner
21、and Tucker (1986)23 have argued that engineering rework is caused by owner scope and specification changes, design errors, or procurement errors and that construction rework is a result of poor construction techniques or poor construction management policies. Focusing on the origins of rework, Davis
22、 et al. (1989)12 reported that there are five origins of rework: owner, designer, vendor, transporter, and constructor. Similarly, CII (1989)7 and Burati et al. (1992)4 identified five major areas of rework: design, construction, fabrication, transportation, and operability. Each of these areas was
23、further subdivided by type of deviation, i.e., change, error, or omission. These classifications differ in perspective from those proposed by Love et al. (1999a20, b21) and Fayek et al. (2003)13. These authors argue that rework occurs as a result of uncertainty, poor leadership and communications, a
24、nd ineffective decision-making.CIIs Benchmarking and Metrics Committee has built on these previous studies to define a set of metrics appropriate for the industry sector that CII serves and also to examine how construction cost performance is affected by rework. The following two hypotheses were est
25、ablished in this study. 1. 1. There are statistically significant differences in the impacts of rework on construction cost performance for the various project groups.2. 2. There are statistically significant differences in the rank orders of rework sources.The research methodology, including the st
26、atistical methods used to test these hypotheses, is described in the next section.Methodology topData Collection and Presentation topThe CII Benchmarking and Metrics (BM&M) program collects capital project data by means of an online questionnaire. At the time of this study, the CII BM&M database was
27、 composed of data from 1,057 projects completed by 41 owner and 35 contractor companies. Although the database contained 1,057 projects, rework costs were not reported for 229 of these projects and of the remaining 828 projects, 469 projects did not report either direct rework costs or construction
28、phase costs. As it is desirable to measure direct rework costs as a portion of actual construction costs, the projects not reporting these costs were excluded from this study. Three hundred fifty-nine projects were finally selected and depending on project characteristics, the data were categorized
29、by industry group, nature, size, location, and work type (contractor projects only) as shown in Table 1. Detailed types of projects included in the industry group category are provided in the Appendix.Total Field Rework Factor topCII developed a metric for quantifying the impact of rework on constru
30、ction cost performance. The metric is defined as the total field rework factor (TFRF) and its formula is as follows: In the formula, the TFRF is expressed as a ratio of the total direct cost of rework to the total construction phase cost. The construction phase cost includes all costs associated wit
31、h the construction phase. Fig. 1 provides an example interpretation of the TFRF. The costs used for the example are not derived from real data, but are for illustrative purposes only. The total construction phase costs in the first and second example projects are $10 million each, with the total dir
32、ect rework costs of $1 million and $0.1 million, respectively. The TFRF are thus 0.1 for Project 1 and 0.01 for Project 2. If rework had not occurred on either project, the construction phase costs of the projects would have been $9 million and $9.9 million, respectively. In other words, due to rewo
33、rk, the cost of Project 1 grew by $1 million and that of Project 2 increased by $0.1 million. Therefore, it can be concluded that the rework that occurred on Project 1 contributed more to the increase of the actual construction phase cost and thus had a relatively greater impact on construction cost
34、 performance. The higher the value, the greater impact on actual construction phase cost.Fig 1. Examples for total field rework factorView first occurrence of Fig. 1 in article.To quantify the impacts of rework by various project characteristics, statistics for each group shown in Table 2 (1. Projec
35、t Characteristics) is calculated using the aforementioned TFRF formula. A group, for example, may be any one of buildings, heavy industrial, infrastructure, or light industrial for industry group, or add-on, grass roots, or modernization for project nature. By averaging and comparing the values calc
36、ulated by the formula by group, mean TFRFs for each group can be obtained and those types of projects most affected by rework can be identified.As shown in Table 2 (2. Sources of Rework), sources of rework were classified as owner change (OC), design error/omission (DE), design change (DC), vendor e
37、rror/omission (VE), vendor change (VC), constructor error/omission (CE), constructor change (CC), transportation error (TE), and other (OS), and their definitions are provided in the Appendix. The sources of rework having the most impact on cost performance can be also identified using the same form
38、ula. One difference in the numerator is that the total direct rework cost for a single source of rework is used. Each of the nine sources of rework may be plugged into the formula.Statistical Analysis Methods topThe one-way analysis of variance (ANOVA) or t-test was applied to test for Hypothesis 1,
39、 which was introduced earlier in the section entitled “Background.” The ANOVA and t-test are the commonly used methods to evaluate the differences in means between two groups and more than two groups, respectively. The levels of significance for the ANOVA and t-test were 0.05. For significant differ
40、ences, a post hoc test was performed as the second stage of the ANOVA procedure to determine specific groups that were different. This later test identified statistically different means by checking the 95% confidence intervals which is equivalent to a level of significance of 0.05. For Hypothesis 2
41、, also presented in the previously, the Spearman rank-order correlation was calculated and statistically tested. The Spearman rank-order correlation is a method of computing a correlation between the ranks of scores on two variables. The correlation is calculated on the ranks of scores, not the scor
42、es themselves. As a result, without the consideration of normality or equal variance of data, this statistical method can be used focusing on difference in rank orders of data rather than difference in means. The coefficient equals 1 for a perfect positive correlation and 1 for a perfect negative co
43、rrelation. When the correlation is not perfect, the coefficient lies between 1 and 1. A level of significance of 0.05 was also applied for this analysis.Data Analysis topThe rework data from the 359 projects were analyzed separately for owners and contractors. The impacts of rework by project charac
44、teristics are first discussed, and then sources of rework are compared.Owner Reported Projects: Rework Impact by Project Characteristics topTable 3 shows the results for the owner reported projects by project characteristic. Table 3 is composed of two parts: one part describes results from the ANOVA
45、 or t-test and provides the total number of projects (N), average total field rework factor (mean TFRF), standard deviation (SD), and p-value (p). The other part of Table 3 summarizes the post hoc test indicating the group for which the mean TFRF was significantly different from those other groups w
46、ithin each category. The mean TFRF for each group was calculated by Formula 1 by dividing the sum of the TFRF of each project in a group by the total number of projects within the group. The mean TFRF of the “All” category was the sum of the TFRF of all projects divided by the total number of all pr
47、ojects.In the industry group category, the mean TFRF for light industrial (0.093) was highest and that of heavy industrial (0.044) was lowest, indicating that for this sample, the cost impact of rework in light industrial projects is significantly greater than that of buildings or heavy industrial p
48、rojects (p = 0.0021). According to project nature, rework in modernization projects contributed to the increase of the actual construction phase cost almost twice as much as it did in add-on projects and this finding is also significant (p = 0.0130). Although modernization projects reported on avera
49、ge approximately 50% more rework than grass roots projects, this finding lacks statistical significance. Based on project size, the mean TFRF for projects between $50 million and $100 million was calculated as being the highest at 0.073, however, this is based on a small sample of 12. The lowest mean TFRF (0.049) was recorded for projects costing less than $15 million, but again, these findings lack significance. Finally, results by project location reveal that the mean TFRF for domestic (
陕公网安备 61072602000132号 违法和不良信息举报：0916-4228922