International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

DEVELOPING NEW APPROACH FOR MEASURING CONSTRUCTION

PROJECT PERFORMANCE

Khaled Aboulata1, Ahmed Saad Eldin2 and Moheeb El-Said3

1. Deputy Director, Internal Audit Department, Orascom Construction, Cairo, Egypt
2. Asociate Professor of Construction Engineering & Management, Structural Engineering
Department, Faculty of Engineering, Cairo University
3. Professor of Construction Engineering & Management, Structural Engineering
Department, Faculty of Engineering, Cairo University

ABSTRACT

Identifying construction project’s performance is considered very important; not

only for project stakeholders, but also throughout the markets worldwide, as it will
result in attracting future investment, increasing share value and attracting high
caliber employees.
This creates the need for periodical evaluation of project against pre-defined
criteria to measure project management effectiveness & efficiency in completion of
project goals to provide management with reasonable assurance of project success.
Therefore, it is important to consider how construction organization’s performance
is measured and how it can be communicated to the market, i.e. how it can be
interpreted and understood by the potential investors, customers and employees.
This research discussed six different approaches used for measuring performance
of construction projects to illustrate differences between approaches through
detailed study of each approach objective, theory, advantages and limitations. Also,
a practical example was provided to apply each approach on real project.
Finally and after combining two approaches together to gain their advantages and
reduce their limitations; the researcher concluded a new approach which may be
used to evaluate the current performance and estimate the future performance in
addition to its probability of occurrence.
Keywords: Project Management; Project Performance; Project Evaluation;
Construction management; Performance Measurement.

1. Introduction
The construction industry is considered one of the largest industries in the whole
world. Growth in this industry in fact is an indicator of the economic conditions of a
country. This is because of the wide employment circle of labor in addition to all
examples of resources consumed by the project such as materials, equipment and
capital investment [1].

1
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

What differentiates the construction industry from other industries is that its projects
are large, complex, built on-site and generally considered a unique result of non-
routine work, which creates difficulties in controlling resources.
Due to resources consumption and complexity, many of the construction projects
exhibit cost overruns, time extensions, and conflicts among parties. Also, quality,
safety and environmental issues usually have a great impact on the project success.
This creates the need for periodical evaluation of project against pre-defined criteria
to measure project management effectiveness & efficiency in completion of project
goals to provide management with reasonable assurance of project success.
This paper aims to present different approaches used for measuring performance of
construction projects and discuss the differences between each approach. On the other
hand, the researcher will attempt to develop new method to evaluate project
performance by combining two or more approaches, which will enhance the new
method capabilities and eliminate the existing limitation in each approach.
During the last few decades, many approaches were created to evaluate performance
of construction projects [3] [4] [5] [6] [7] [8], such as:
1. Earned Value Technique (EV)
2. Measuring Project Key Performance Indicators (KPI)
3. Project Risks and Opportunities Assessment
4. Data Envelopment Analysis (DEA)
5. Project Evaluation using Weibull Analysis
6. Performance Evaluation Based on Statistical Process Control

2. Measuring project performance using earned value

The main objective of (EV) technique is to calculate quantitative project performance
indicators and predict future performance, which allow project team to manage their
projects and proactively take corrective actions.
Earned value was defined in the Project Management Body of Knowledge [9], as
follows: “Earned value (EV) is the value of work performed expressed in terms of the
approved budget assigned to that work for an activity or work breakdown structure
component”.
Using of Earned Value technique requires the awareness of few project measurements,
which are presented in the following table [10].

Table 1: Earned Value Terms and Descriptions (Project Management Institute)

Term Full Name Meaning
PV Planned Value Estimated value of the work planned to
be done
EV Earned Value Estimated value of the work actually
accomplished
AC Actual Costs Actual Costs Incurred
BAC Budget At Completion Amount budgeted for total project
EAC Estimate At Completion Currently expected total for project
ETC Estimate To Complete How much More to finish
VAC Variance At Completion How much over/under we expect to be

These measurements are used to calculate the project performance indicators, which
are presented in the following table.

2
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

Table 2: Project Performance Indicators (Project Management Institute)

Name Formulae Explanation
Cost Variance CV=EV-AC Negative: Over budget,
Positive: Under budget
Schedule Variance SV=EV-PV Negative: Behind schedule,
Positive: Ahead schedule
Cost Performance Index CPI=EV/AC We are getting _ cents out of every $
spent
Schedule Performance SPI=EV/PV We are progressing at _% of the rate
Index originally planned
Estimate At Completion EAC=BAC/CPI As of now, how much do we expect the
total project to cost
Estimate To Complete ETC=EAC-AC How much more to finish
Variance At Completion VAC=BAC-EAC How much over/under we expect to be

The above mentioned project performance indicators are used to evaluate the current
project performance and predict the future performance regarding scope, schedule and
cost.
It must be noticed that value of “Estimate at Completion” may be calculated in four
deferent ways, according to project circumstances.

3. Measuring project performance using key performance indicators (KPI’s)

A key performance indicator (KPI) is the measure of the process performance that is
critical to its success. The best performance achieved in practice is the benchmark.
The target is to understand the current activities and then identify an external point of
reference or standard by which that activity can be measured or evaluated. The
purpose of the KPI is to enable measurement of project and company’s performance
throughout a large number of projects and by then provide indications about the
performance of the whole projects or per business unit [12].
Generally, Key Performance Indicators includes measures for achieving project
objectives in several areas, such as Scope, Time, Cost, Quality, Health & Safety.

4. Measuring project performance using project risk assessment

The construction industry is subjected to more risk and uncertainty than many other
industries and does not have a good track record of coping with risks. Late completion
of projects, surpassing their estimated budgets and in some worse instances without
even achieving the desired quality and operational requirements, has given a bad
name to the industry. Thus, effective management of construction associated risks
remains a big challenge to the industry stakeholders.
Risk was defined in (PMBOK) ) [9] as: uncertain events or condition that, if it occurs,
has an effect on at least one project objectives. According to risk impact nature,
consequence of risk could be Positive (Opportunity) or Negative (Threat). The overall
project risk magnitude is the summation of both types in different signs [13].

5. Measuring project performance using data envelopment analysis (DEA)

Data Envelopment Analysis (DEA) has been recognized as a data-oriented approach
that is used for evaluating the performance of business organizations. DEA technique
was developed in late 70’s to be used in measuring and evaluating performance of
organizations. The theory of this technique was based on nonparametric linear
3
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

programming technique that produces an empirical efficiency score to be used for

benchmarking Cook [14].
Performance is measured through calculating efficiency by comparing the process
output to process input according to the following equation:
Output
Performance =
Input
When single input & single output are measured, it is easy to calculate performance.
In order to evaluate performance in case of two inputs and one output, we need to
normalize the output by setting it to unity [6].

6. Measuring project performance using Weibull analysis

Weibull analysis includes fitting a Weibull distribution to any measurement. It could
be used for modeling reliability and to predict the failure of systems Taylor [15]. It
provides accurate results in many previous studies Zacks [16]. It was discovered that
various measurements of systems (such as failure of specific products), can be
modeled as a Weibull distribution, due to the flexibility of this distribution.
Weibull distribution may be presented at many shapes, according to its parameters,
Philip [17]. The Weibull function is presented as the following equation:

 α is scale parameter α ˃ 0
Measure of spread in the measured data
 β is shape parameter β˃0
Shows the rate of measurement is increasing, constant or decreasing
 β ˂ 1.0 decrease
 β ˃ 1.0 increase

After fitting the measurements to the Weibull distribution, risks can be calculated.
Some Weibull distributions curves are presented in the following graph for several β
values while α =1.

Figure 0-1: Weibull distributions with different scale and shape parameters

4
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

7. Measuring project performance using statistical process control

Statistical Process Control (SPC) is a technique that applies statistical tools to control
processes. Statistical tools include Control Charts, Scatter Diagrams, Histograms, Bar
Charts, and Pareto Charts …etc [18].
These tools, if successfully applied, can produce a significant impact not only in
evaluating the current performance, but also in process control and improvement.
A control chart is a graphic representation of the variation in the computed statistics
being produced by the process. It consists of a central line and two control limits
(upper and lower limit). If all process data are plotted within the control limits and
without any particular tendency, the process is regarded as being in the controlled
state. Generally, there are seven SPC control chart types known around the world,
each one has its unique characteristics and specific application.
A control chart based on one observation is called a chart for individuals or an X chart.
Since we are not using sample averages, we do not have the benefit of the central
limit theorem, which tells us that the distribution of averages is approximately normal
regardless of the underlying distribution. Therefore, an X chart is much more sensitive
to a lack of normality of the underlying distribution than is an X-bar chart [8].
The theoretical principles underlying X-Rm are three:
1. Samples are from normal distribution
2. Samples are independent and identically distributed;
3. Samples are randomly selected.
Since we are only taking one observation, we do not have an obvious source of an
estimate of the standard deviation that we can use to determine the control limits for
an X chart. We have two possible choices:
1. Use a moving range (difference between two successive observations
2. Calculate the standard deviation of 20 or more successive observations.
If the process does not shift, both statistics will give about the same answer. However,
if the process does shift, the moving range will minimize the effect of the shift.
Therefore, that method is usually used.
The basic procedure for developing X charts is as follows:
1. Select the measurable characteristic to be studied.
2. Collect enough observations (20 or more) for a trial study. The
observations should be far enough apart to allow the process to be
potentially able to shift.
3. Calculate control limits and the centerline for the trial study using the
formulas given later.
4. Set up the trial control chart using the centerline and limits, and plot the
observations obtained in step 2. If all points are within the control limits
and there are no unnatural patterns, extend the limits for future control.
5. Revise the control limits and centerline as needed (by removing out-of-
control points or observing trends, etc.) to assist in improving the
process.
6. Periodically assess the effectiveness of the chart, revising it as needed.
Upper and lower control limits of the individual control chart may be calculated
according to the following equations:
Upper control limit = X-bar + 2.66 R-bar
Lower control limit = X-bar - 2.66 R-bar
Where R-bar is the moving range between every two consecutive observations

5
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

8. Introducing new method to measure project performance

A recommended method for evaluating the project performance may be developed by
combining some of the previously discussed methods. This combination is intended to
maintain the main advantages and overcome most of limitation in all discussed
methods.
Two methods were chosen to be combined:
 Earned Value Technique
 Statistical Process Control
As discussed earlier, traditional Earned Value Technique provides an assessment of
cost and schedule performance through the use of CPI and SPI or CV and SV. CPI
and SPI relate purely to cost and schedule performance only at the report date and no
account is taken of any time sequence. This is a serious restriction, because the
consecutive performance variations and their assignable causes are not directly
evaluated in most cases [19]. With the aid of the consecutive performance variation
analysis, some assignable causes that affect project performances can be identified as
early as possible. SPC techniques are adopted to overcome the limitations of
traditional methods using the control chart.
This method may be used for evaluating the performance of the construction project
using the results of Cost and Schedule Performance Index.
In this example we will use the measurements obtained from Earned Value Technique
and apply the statistical process control tools of Histogram & individuals-moving
range control chart to evaluate project performance.
Data of Cost Performance Index (CPI) was collected by Nassar [7] for evaluation of
project performance. The following table shows the values of measured Cost
Performance Index (CPI) collected during 24 weeks.

Table 4: Data of Cost Performance Index (CPI)

Month CPI Month CPI Month CPI Month CPI
1 0.46 7 0.78 13 0.91 19 0.99
2 0.91 8 1.04 14 0.46 20 0.91
3 0.73 9 0.61 15 1.04 21 0.91
4 0.91 10 0.61 16 0.76 22 0.91
5 0.68 11 0.46 17 1.14 23 1.09
6 0.68 12 0.65 18 1.42 24 0.57

In order to apply the statistical process control tool of individuals-moving range

control chart; data set should be normally distributed [18]. To evaluate whether the
data set fulfills the principle of normal distribution, a normality test was adopted for
the evaluation. The level of significance for this case was defined as 0.05.
The obtained P-value = 0.433 was compared to level of significance = 0.05. Values of
the calculated level of significance were higher than 0.05, which means that data is
not normally distributed. If the requirement of normal distribution was not satisfied,
the analysis and calculated results of control charts may be affected. Therefore, a
transformation method was used to obtain normalized data by calculating the
reciprocal of data.
The following table shows the reciprocal of the measured Cost Performance Index
(CPI).

6
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

Table 5: Data of Reciprocal Cost Performance Index (1/CPI)

Month 1/CPI Month 1/CPI Month 1/CPI Month 1/CPI
1 2.173913 7 1.282051 13 1.098901 19 1.010101
2 1.098901 8 0.961538 14 2.173913 20 1.098901
3 1.369863 9 1.639344 15 0.961538 21 1.098901
4 1.098901 10 1.639344 16 1.315789 22 1.098901
5 1.470588 11 2.173913 17 0.877193 23 0.917431
6 1.470588 12 1.538462 18 0.704225 24 1.754386

The obtained P-value = 0.026 was compared to level of significance = 0.05.

After transformation, values of the calculated level of significance were lower than
0.05, which means that data is normally distributed.
The performance tendency should be analyzed based upon the charts. Control chart
patterns used in SPC provide the information about the state of a process. As stated
earlier, the points on a control chart that follows a completely random pattern is
regarded as natural pattern; which means that process is in statistical control.
The following figure shows the inverse of CPI in sample project with the relevant
month presented in control chart.

Figure 3: Inverse CPI control chart

It is obvious that unusual performance occurred during the 1st, 2nd, 3rd, 18th & 20th
month. The value of inverse CPI was higher than the upper limit of the control chart,
which indicates unfavorable performance.
Further investigation and analysis should be performed to explore the reasons behind
the assignable causes and propose the corrective action needed to prevent the
reoccurrence of this performance.
In addition, the histogram tool will be used as another statistical tool for calculating
the probability of occurrence of specific potential value of CPI in future. The
following figure shows Histogram plotted for reverse CPI.

7
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

Figure 4: Histogram of reverse CPI

The following statistics were calculated for the inverse CPI:

Mean = 1.334; Standard Deviation = 0.4199; Range = 1.4697
Using the Z-table, we could calculate the probability of obtaining a specific value in
future. In order to predict the probability of having CPI = 1 in upcoming month, we
will calculate the inverse of CPI, as follows:
Target CPI = 1; Inverse CPI = 1 / CPI = 1
Z-value = (CPI – Mean) / Standard Deviation
Z-value = (1 - 1.334) / 0.4199 Z-value = - 0.795
Using z-table to identify probability associated to Z-value = - 0.795; Probability = 21%
The previous results indicated that the probability of occurrence of the inverse value
of CPI “1” or less is equal to 21%.
Hence, 21% is the probability of occurrence of value of “CPI ≥ 1”.

9. Comparison between approaches used for measuring project performance

The following criteria may be used to evaluate each approach capabilities and
limitation:
Lead–Lag Capabilities: Some methods are used to evaluate the current performance,
while it is not able to predict the future performance.
Measurements Complexity: Some methods need simple measurements, while others
need complicated measures.
Analysis Complexity: Some methods are easy to use, while others need sophisticated
analysis.
Evaluation Independency: Some methods provide the evaluation of project
performance in absolute manner, while others need to benchmark the project
performance to other projects.

Table 6 presented at next page is designed to include all the studied methods for
evaluation of project performance in order to illustrate each method capabilities and
limitations. This table shows the advantage of applying statistical tools over data
obtained from earned value technique to calculate probability of occurrence for
specific value in future, which allows project team to predict and assess project risk.

8
 International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

Table 6: Comparison between Measuring Project Performance Methods

Lead–Lag Capabilities Measurements Analysis Complexity Evaluation
Complexity Independency
Evaluate the Probability Complicate
Predict the Simple Simple
current prediction d Sophisticat Individu Benchmarki
Method future project Measureme Calculatio
project Measureme ed Analysis al ng
performance nts n
performance nts
1 Earned Value
Technique     

2 Key Performance
Indicators (KPIs)    

3 Risk Assessment
   
4 Data Envelopment
Analysis (DEA)    

5 Weibull Analysis
   
6 Statistical Process
Control     

7 Proposed New
Approach      

9
International Conference on Structural and Geotechnical Engineering, Ain Shams University ICSGE 14

10. Conclusion
Based on the results obtained through this study, it is recommended that construction
projects use the combination of Earned Value Technique and Statistical Process
Control tools to evaluate the current project performance in addition to prediction of
future values and its probability of occurrence.
By using the new approach, we gain the advantages of both individual methods.
Special consideration should be maintained during the use of data, which should
fulfill the conditions of using the selected statistical process control tool.
Additional efforts should be conducted to explore more approaches used for
evaluating the performance of construction projects in future studies, which will allow
for combing more approaches together to increase the method’s advantages and
reduce its limitations.
Future studies should also involve more sophisticated approaches used for measuring
project performance such as artificial intelligent and M-TPSIS methods.

References
[1] Levine, 2009, Practical Project Management, John Wiley & Sons, Inc., 2002.
[2] Sumara, 1997, "Earned Value: The Next Generation, A Practical Application
for Commercial Projects," in Annual Seminars & Symposium Chicago, ,
Illinois.
[4] D. Louis, 2011, Project Performance Index and Benchmarking for Large
Construction Companies, MBA Dissertation,, Maastricht : Maastricht School
of Management - Netherlands.
[5] Dikmen, 2006, "An analytic hierarchy process based model for risk and
opportunity assessment of international construction projects," NRC Research
Press.
[6] C. e. al, 2007, Data Envelopment Analysis, A Comprehensive Text with
Models, Applications, References and DEA-Solver Software.
[7] Nassar, 2005, "Using Weibull Analysis for Evaluation of Cost and Schedule
Performance,," Journal of Construction Engineering and Management.
[8] J. Juran., 1999, Juran’s quality handbook, McGraw-Hill Companies.
[9] Project Management Institute (PMI), 2013, Project Management Body of
Knowledge (PMBOK) Fifth Edition, USA: Project Management Institute
(PMI).
[10] Project Management Institute, 2003, Construction Extension to Project
Management Body of Knowledge, USA: Project Management Institute.
[11] A. &. C. A. Chan, 2004, "Key performance indicators for measuring
construction success.," Benchmarking: An International Journal.
[13] E. Fayez, 2003, Construction Project Risks in Egypt, MSc. Thesis, Giza,
Egypt.: Cairo University, , Faculty of Engineering.
[14] W. Cook, 2005, Modeling Performance Measurement - Applications and
Implementation, Springer Science.
[15] G. C. Philip, 2007, "A Generalized EOQ Model for Items with Weibull
Distribution Deterioration," vol. 6, no. 2, pp. 159-162.
[16] S. Zacks, 1992, Introduction to reliability analysis: Probability models and
statistical methods,, Springer, New York.
[17] L. D. Taylor, 1974, Probability and mathematical statistics,, Harper & Row,
New York.
[18] J. Oakland, 2003, Statistical Process Control,, Butterworth Heinemann.

10