RISK ASSESSMENT IN CONSTRUCTION SCHEDULES

By B. Mulholland1 and J. Christian2

ABSTRACT: Construction projects are initiated in complex and dynamic environments resulting in circum-
stances of high uncertainty and risk, which are compounded by demanding time constraints. This paper describes
a systematic way to consider and quantify uncertainty in construction schedules. The system incorporates knowl-
edge and experience acquired from many experts, project-specific information, decision analysis techniques, and
a mathematical model to estimate the amount of risk in a construction schedule at the initiation of a project.
The model provides the means for sensitivity analyses for different outcomes wherein the effect of critical and
significant risk factors can be evaluated. The paper focuses on lessons learned from past projects and describes
a risk assessment process involving typical inputs and expected outputs. The paper also briefly reviews the
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

information technology of HyperCard and Excel, which were used to develop the system.

INTRODUCTION The purpose of this paper is to discuss the development of

a computer-based system for the assessment of construction
A 1992 worldwide survey reported that the majority of con- schedule risk, which can increase the effectiveness of the tra-
struction projects fail to achieve the objectives of the schedule ditional project scheduling processes. The system includes the
(Cooper 1994). On many of these projects a schedule overrun following three key features:
did not seem probable at the beginning of the project. Some-
times schedule targets are missed because of unforeseen events 1. A hypertext information system for schedule risk iden-
that even experienced construction managers could not have tification
anticipated. However, schedule target dates are more often 2. A spreadsheet to describe and evaluate project uncer-
missed because of events, such as design problems and indus- tainty
trial disputes, that were predictable but their likelihood and 3. Direct pictorial information to assist the decision makers
effects are difficult to predict with any precision because no in selecting a realistic yet acceptable project completion
two construction projects are the same (Thompson and Perry time.
1992).
A survey by Laufer and Stukhart (1992) of 40 U.S. con-
Part of the development of the system utilized a Macintosh
struction managers and owners indicated that for scope and
PC and commercially available application programs called
design objectives only 35% of the projects considered had low
HyperCard and Excel. In the system, the HyperCard applica-
uncertainty and the remaining 65% had medium to very high
tion program provides an information module that can be used
uncertainty at the beginning of construction. The costs of the
in identifying schedule risks. The Excel spreadsheet is the tool
projects averaged $5,000,000. This finding was confirmed in
used for modeling the effects of the risks on the project per-
a more recent report by Laufer and Howell (1993). They con-
formance time.
cluded that approximately 80% of projects at the beginning of
Part of the paper presents the findings of a literature review
construction possessed a high level of uncertainty.
of articles dealing with problems in traditional scheduling pro-
The amount of uncertainty in the internal and external en-
cesses. The problems are grouped into five primary classifi-
vironments of a project is an important factor in determining
cations.
whether there will be a schedule overrun. However, attempting
Four dimensions of construction schedule uncertainty then
to consider realistically the uncertainty in construction sched-
are presented; within each dimension, specific factors are iden-
ules poses three challenges. The first challenge is that systems
tified that have lead historically to schedule risk. A computer-
are not endorsed professionally or available commercially,
based system to model the effects of risks on the project sched-
which can be used to structure project uncertainty and measure
ule then is presented.
the effects on the project schedule. The second challenge is
Then, explanations are given on how the risks are quantified
the lack of easily accessible information documenting the ex-
and analyzed. The application and details of the computer sys-
perience of the construction industry or the knowledge scat-
tem are described so that an example for modeling schedule
tered within a corporation. The third challenge is the difficulty
risk can be shown. Before drawing some conclusions, the ben-
motivating the involvement of the senior project management
efits of using the model are listed and the importance of the
team to address adequately schedule risks. Project teams gen-
responsibility for schedule risk assessment and the timing of
erally are too preoccupied with solving current problems in-
the assessment are emphasized.
volved with getting work done and therefore have insufficient
time to think about, much less carry out, a formal risk assess-
ment program (Oglesby et al. 1989). PROBLEM STATEMENT

1
The complex characteristics of major construction projects
Sr. Engr., New Brunswick Power, 515 King St., Fredericton, NB, have created the need for improved management support, tech-
Canada E3B 4X1; Prin., Janus Consulting, 86 Colonial Heights, Freder-
icton, NB, Canada E3B 5M1.
niques, and tools. Many companies have recognized this need
2
Assoc. Dean of Engrg. and M. P. Gillin Chair in Constr. Engrg. and and are drawing from their past experiences when developing
Mgmt., Univ. of New Brunswick, P.O. Box 4400, Fredericton, NB, Can- project organizations, selecting management tools, deciding on
ada E3B 5A3. contracting strategy, and developing project schedules. How-
Note. Discussion open until July 1, 1999. To extend the closing date ever, schedule overruns still frequently occur. Past experiences
one month, a written request must be filed with the ASCE Manager of are not being analyzed, synthesized, and corrected sufficiently
Journals. The manuscript for this paper was submitted for review and
possible publication on April 29, 1997. This paper is part of the Journal
in order to prevent mistakes on past projects from being re-
of Construction Engineering and Management, Vol. 125, No. 1, Janu- peated.
ary/February, 1999. 䉷ASCE, ISSN 0733-9634/99/0001-0008 – 0015/ There is a lack of an accepted method of risk assessment
$8.00 ⫹ $.50 per page. Paper No. 15664. and management among professionals in the construction in-
8 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999

J. Constr. Eng. Manage. 1999.125:8-15.

 dustry compared with the financial and health professions. The TABLE 1. Traditional Project Scheduling Processes — Classi-
construction industry also does not seem to recognize, nor ac- fication of Problems
cept, that risk should be addressed formally and given more Sources of Problems
serious attention. If a method of risk assessment and manage-
Application
ment were installed in the construction industry it should be
of network Realities of
extended into areas such as engineering design and analysis
Classification Network methodol- construction
and life-cycle costs. of problems theory ogy processes
(1) (2) (3) (4)
CLASSIFICATION AND SOURCES OF PROBLEMS IN
(a) Nature of Engineering and Construction Work
TRADITIONAL SCHEDULING PROCESSES
Complex organization x
The construction industry has been using quantitative net- Activity dependencies x
work modeling techniques to simulate the construction process Other techniques more ap-
since the late 1950s. Network diagram techniques have helped propriate x
Critical activities cost sepa-
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

in clarifying activity relationships and have assisted in the es- ration x

tablishment of durations and coordination of construction proj-
ects. However, much has been written on the problems con- (b) Resource Allocation Process
cerned with traditional scheduling processes (Birrell 1980). Networks inadequacy con-
For example, one of the problems with traditional scheduling sidering resources x
processes is that the critical-path method (CPM) approach is Equal importance to each
activity in resource allo-
deterministic in nature; the process thereby ignores the effects cation x
of uncertainty by using a single value for the time estimate of Effect on resource program
each activity and even the entire construction of the project. for each activity with sin-
Five classifications summarizing problems that have been gle value time estimate x
documented with traditional scheduling processes are shown Resource allocation algo-
in Table 1. The problems within each classification were cat- rithms use of priority
rules x
egorized into the following three sources of constraints, which Limited resource types in re-
can limit effective use of traditional scheduling processes: source allocation program x
Resource allocation pro-
1. Technical problems involved with network concepts cesses problem with proj-
2. Process problems in the application of network meth- ect scope details x
odology (c) Complexities of Project Operations and Scheduling Methods
3. Problems involved with making the process work in con- Schedules too complex for
struction. field x
Choice of logic in network
One of the more serious problems with traditional sched- models x
Specialized scheduling op-
uling processes is that they do not evaluate explicitly the un- eration team required x
certain and risky nature of the internal and external environ- Interpretation of excessive
ment of a project. Thereby, in contradiction with reality, computer output x
traditional project scheduling processes have treated uncer- (d ) Contractors Lack of Confidence in Network Scheduling Processes
tainty and risk as if they do not exist (Archibald and Lichten-
Credibility that construction
berg 1994). Consequently, for projects with any amount of process can be modeled x
uncertainty, network-based planning processes have been Many secondary business
proven inadequate for estimating a realistic project perfor- activities occur but not
mance time (Laufer and Howell 1993). shown on schedule x
(e) Activity Time Estimating
SCHEDULE UNCERTAINTY Lack of reliable and realistic
Uncertainty can affect the seemingly precise predictions of time estimates for activi-
ties x
traditional scheduling processes. To provide a framework for Nature of activity duration
a systematic and structured assessment of schedule uncertainty, requires experience x
four dimensions of schedule uncertainty were established. The Time estimates only margin-
following first three dimensions relate to the uncertainty in the ally address schedule risk x
project life-cycle phases: engineering design, procurement, and
site construction. The uncertainty in each phase is driven by
its own unique set of variables and is differentiated from the are required in the execution of the procurement process.
other phases by work content. The fourth dimension involves High-risk factors in the procurement phase include vendor/
uncertainty in the effectiveness of the project management contractor selection processes and delivery on time.
function in completing the project as scheduled (Diekmann et One of the critical factors for success during the construc-
al. 1988) (Fig. 1). tion phase is the need to develop an effective contracting strat-
In the engineering design phase, uncertainty problems are egy that fairly assigns risk and thereby encourages owner and
encountered that include designers with varying degrees of contractor teamwork. One reason for failures in construction
experience, the fragmentation of the engineering and construc- projects has been caused by the selection of a contracting for-
tion processes, the trend to lump-sum bidding, economic and mat that did not fit the risk characteristics of the project. For
liability issues, constructability, and changes in design. example, the use of a lump-sum contract on a fast-track project
Sequential steps are involved in the procurement phase, and can lead to many contract disputes and the diversion of man-
although simple in nature, the procurement process is more agement’s attention from the critical fieldwork issues.
complex than it appears. There are a myriad of activities at Poor management practices also create problems. The ef-
various locations in which many widely dispersed personnel fectiveness of the project management function significantly
JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999 / 9

J. Constr. Eng. Manage. 1999.125:8-15.

 and Perry (1992); and (5) the report, ‘‘Confronting Uncertainty
in Risk Management,’’ by A. D. Finkel (1990) of the Center
for Risk Management, Washington, D.C.
Classic risk analysis is undertaken in the following three
iterative phases: (1) Risk identification; (2) risk measurement;
and (3) risk management (Diekmann et al. 1988). The subject
of this paper, risk assessment, is involved with the first two
phases of risk analysis. Risk identification involves determin-
ing which variable are likely to affect the schedule. Risk mea-
surement involves evaluating and quantifying the probability
FIG. 1. Dimensions of Uncertainty in Schedule of Construc- of the occurrence of a risk and the effects on the schedule
tion Project (Fig. 2).
To demonstrate the ideas presented in this paper, an assess-
TABLE 2. Characteristics of Successful and Unsuccessful ment of schedule risk is performed on a simulated megaproject
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

Projects in the following sections. The situation is one of a large-scale

Successful projects Unsuccessful projects owner organization, having the predominate role in the engi-
(1) (2) neering effort and in setting the master schedule for the entire
Well-defined scope Ill-defined scope project, while also acting in the role of construction manager.
Early, extensive planning Poor planning The design of the $1 billion facility will draw on resources
Good leadership and supervision Poor management and controls from three prime engineering firms producing designs that re-
Involved, positive owner relation- Poor communication between dis- sult in 115 major supply contracts and 85 prime on-site con-
ships ciplines struction contracts. Over the course of 5 years, approximately
Good communications and relation- Unrealistic schedules and budgets 11,000,000 h of engineering, manufacturing, on-site work, and
ships
Quick response to changes Poor personnel quality management will be required to complete the project. During
Engineering concerned with total Excessive changes this period, an average of over 1,000 people will be working
project each month in widely dispersed geographic locations.
Suppose you were responsible for establishing the duration
of this construction project; you have divided the project into
influences whether the planning project schedule duration will four phases with the description and duration associated with
be achieved successfully. Table 2 shows some of the principal each phase as follows: engineering design, procurement, con-
factors that influence the effectiveness of the project manage- struction, and commissioning estimated at 37, 35, 39, and 14
ment function (The Business Roundtable ‘‘Project’’ 1986). months, respectively. The schedule is to be of a fast-track na-
A list of other sources of risk that create schedule uncer- ture with optimum overlapping of engineering design, pro-
tainty can be found in Appendix I. For scheduling to be ef- curement, construction, and commissioning activities, which
fective in an uncertain environment, knowledge gained from means the preparation of the detailed civil engineering design
formal education, training, and substantial planning experience will be started even before the design descriptions have been
is required. However, the valuable knowledge and experience completed or equipment ordered and vendor shop drawings
that is acquired during the construction of a facility often are have been received.
not available for the construction of the next similar facility. Further, suppose that planning and design are early in the
A number of factors have contributed to the shortage in re- conceptual phase. Accordingly, the scope is not stabilized, all
cording past project experiences. The following are four of the parties have not found consensus, implementation methods
principal reasons: have not been decided yet, and parts of the project may even
need further research and development. As a result, one of the
1. The project team assumes each project is unique; there- biggest problems will be in establishing the interphase overlap
fore old records are often considered of little value, and and subsequent overall project duration.
the records are not maintained or are discarded (Sanvido The problem stems from the fact that quantities of work are
and Medeiros 1990). somewhat influenced by variables only known with any degree
2. At the end of the project there is a lack of interest or of certainty as the job progresses. Indeed, one survey revealed
funding to conduct postproject reviews. that 70% of the respondents had experienced schedule changes
3. A formal or convenient process does not exist to capture related to ‘‘differing site conditions’’ that were either not
and transfer readily knowledge to subsequent projects
(Ibbs 1986; Mohan 1990).
4. At the end of a project often there is not a similar project
in progress to which the professional and trades person-
nel can be transferred.

SCHEDULE RISK ASSESSMENT APPROACH

The schedule risk system, presented in this paper, is based
on experience gained with large power plant projects; how-
ever, the principles apply to any construction project. The de-
velopment of the system also draws from recent research and
advances in risk analysis. The following were the primary ref-
erence materials: (1) The successive principle (Lichtenberg
1988); (2) the concepts of conventional CPM/program evalu-
ation and review technique (PERT) and management science
techniques; (3) the research on risk management in capital
projects at the University of Colorado at Boulder; (4) the re-
port entitled ‘‘Engineering Construction Risks,’’ by Thompson FIG. 2. Assessment of Schedule Risk System

10 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999

J. Constr. Eng. Manage. 1999.125:8-15.

 known or not anticipated at the start of the project (Halligan
et al. 1987). For the project under consideration, the level of
sophistication used to develop each deterministic activity time
estimate ranged from the project management team’s experi-
ence to historical data. Therefore, because the actual execution
of the project will occur over a span of many years, the reli-
ability of activity time estimates is questionable.
Ultimately, more engineering design will resolve some of
the uncertainties surrounding the design effort and construc-
tion quantities, but at this point in time, an estimate of the
overall project duration must be produced based on this un-
certain data. To model the effects of uncertainty on the project
schedule, the execution of the following three recursive steps
is required: (1) Identify schedule risks; (2) evaluate their ef-
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

fects and the probability of occurrence; and (3) within the

proposed project schedule framework, model the risks and FIG. 3. Probability Density Function: Performance Time of
their effects to obtain the project’s schedule risk profile. For Project
the project under consideration, the results from the analysis
indicated that the expected project duration of 59.6 months is for modeling the effects of schedule risks on the performance
3% greater than the originally predicted deterministic duration time of a project. For the purposes of the modeling process,
of 58 months. the variance of the performance time distribution of a project
IDENTIFICATION OF SCHEDULE RISKS was used to measure schedule risk. Variance implies uncer-
tainty; it can be used as an objective measure of the ability to
During the planning of a project, it is not possible to mea- predict a reliable performance time for a project. The larger
sure directly the time required for the various phases or ob- the variance the greater the risk associated with the perfor-
serve directly how these phases interrelate. Thus, at the con- mance time of the project. Indeed, variance often has been
ceptual stage of a project a base case schedule model is used by the financial community to measure the riskiness of
developed to predict the performance time of the project. investment options (Keller et al. 1994).
However, uncertainty surrounds all facets of most projects and The PERT technique was used to develop a project perfor-
can confound any attempts to predict precisely the perfor- mance time distribution and as a basis to compute the proba-
mance time of a project with confidence. Therefore, at this bility that a project will be completed on or before a scheduled
stage, a good schedule analysis would demand explicit iden- time, without the need to take other than normal measures to
tification of the risks to the project schedule. expedite the activities of the project. The information from the
The HyperCard system (discussed later) can provide the ba- PERT calculation is summarized in Fig. 3. The probability
sis for risk identification by presenting most known schedule density function shown in Fig. 3 has the following two struc-
risks. The database provided by the HyperCard system should tural characteristics: (1) A normal distribution (bell-shaped
act as a stimulus for follow-up brainstorming sessions with the curve); and (2) symmetry about the mean, i.e., the expected
key members of the project team. The output from the review project performance time E. The variance VT of the distribution
of the HyperCard database and the brainstorming process answers the question as to how typical is the expected project
should produce a comprehensive list of potential schedule performance time E of all values in the data set.
risks, which then can be rewritten and reordered into the rel-
evant risks for each dimension of schedule uncertainty. Refer
ANALYSIS OF SCHEDULE RISK
to a sample of a database of engineering design risks in Table
3. The information from the PERT calculation can be used to
determine the lower (5%) and upper (95%) confidence limits
QUANTIFYING SCHEDULE RISK of the performance time distribution of a project and thereby
The guidelines for quantifying uncertainty, developed at the provide the variables for a schedule probability statement. The
Center for Risk Management (Washington, D.C.), were used probability statement is defined as follows: the largest proba-

TABLE 3. Sample of Database of Engineering/Design Risks

Independent
(I) or
Risk Confidence dependent
number Risk source Importance level (D) risks Description
(1) (2) (3) (4) (5) (6)
30 Productivity High Medium I Actual engineering hours greater than planned
31 Engineering centers Medium Medium D Geographic location of engineering staff
32 Engineering resource pool Medium High D Depth and qualifications of engineering replacement re-
sources
33 Engineering estimate High High D Accuracy of engineering estimate
34 Material substitution High Medium D Redesign due to material substitution
35 Qualified staff Medium Medium D Capability to perform work in allowable time frame
60 Design errors High Low I
90 Design criteria High Medium I Availability of well-documented design standards
91 Code changes Medium High D Changes to codes that affect project design criteria
92 Site investigation High Low D Problems or engineering changes due to insufficient site in-
vestigation
93 Design standards Medium Medium D Availability of well-documented design standards
94 Technology High Medium D Degree of uncertainty in technology

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999 / 11

J. Constr. Eng. Manage. 1999.125:8-15.

 sion. The Excel spreadsheet used for modeling schedule risk
requires as inputs the optimistic, likely, and pessimistic activity
times; the relative importance of risks; and calculates the ex-
pected performance time and the variance in the performance
time distribution, as well as other information and statistical
data.

HYPERCARD INFORMATION SYSTEM FOR

RISK IDENTIFICATION
A Hypertext system is used to store and give access to in-
formation concerning previously experienced schedule risks.
The system is composed of schedule risk information (facts,
data, and heuristics) linked together using hypertext tools. The
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

information can be in the form of text, graphics, or pictures.

Thereby, typical project risks can be documented and made
available to assist new project teams in becoming aware of
general risk information and the possible inferences for a spe-
cific project.
FIG. 4. CDF: Performance Time of Project The HyperCard shell program was used to develop the risk
identification module. It is an object-oriented program and is
based on the concept of a series of cards making up stacks,
bility exists that the project will be finished during month 59
with information and linking devices located on each card. The
(Fig. 3), the expected performance time E; however, if things
HyperCard program can be used to create page formats con-
go badly, with a probability of 1 in 20, the project may not
taining objects with information and buttons, which can be
finish until month 67, the upper confidence level. Similarly, if
used for linking and browsing through information in the da-
things go well, again with a proability of 1 in 20, the project
tabase. Once developed, the system protocol requires little ad-
could finish as early as month 52, the lower confidence limit
ditional programming to add information to the database. An-
(Cowie and Carr 1984).
other feature offered by HyperCard is that the system enables
An alternative to a probability statement and a more de-
the user to backtrack. This is achieved by a screen display of
scriptive tool for conveying the uncertainty in project perfor-
minimaps giving a survey of documents visited. The structure
mance time is the cumulative density function (CDF). The
of the HyperCard knowledge base is shown in Fig. 5.
CDF displays the same information contained in the probabil-
The system’s links provide the means to access the docu-
ity statement. However, the CDF enables the direct calculation
ments within the database. The links, which guide the user
of a range of confidence and the direct comparison of uncer-
through the database, can be divided into two types: organi-
tainty in alternative project scenarios. Refer to Fig. 4, which
zational and navigational. Organizational links connect the
shows the CDF resulting from the assessment of schedule risk,
structure of the system. These links appear as buttons along
for the project under consideration, which has a planned du-
the bottom of every screen. Clicking the buttons executes the
ration of 58 months.
links and thus allows the user to move between the various
The more variance a critical-path activity contains, com-
documents in the database.
pared with other critical-path activities, the more important it
In the database the basic elements of information are con-
is in determining the amount of direct risk to the performance
tained in objects. Once an object has been defined, it is pos-
time of a project. Therefore, the variance of each critical-path
sible to define navigational links that lead to other documents
activity Vt can be used to determine the amount that an activity
in the database. These navigational links were used to provide
contributes to the total project performance time risk VT. Thus,
features such as importance factor criteria and confidence level
to reduce the project performance time risk VT, it would be
definitions.
necessary to reduce the variance Vt of the critical-path activi-
ties.
The standard deviation Vt1/2 and expected performance time MODELING UNCERTAINTY
te of the critical-path activities can be used to measure the Following an interactive session with the HyperCard infor-
relative risk between activities. The activity with the highest mation system, the effects of the risks identified in the four
relative risk is subject to a larger proportional change in du- dimensions of schedule uncertainty can be evaluated. For each
ration and therefore is considered at higher risk than other dimension of schedule uncertainty (engineering design, pro-
critical-path activities. The proportionate measure of variabil- curement, construction, and project management) it is likely
ity of an activity is determined by calculating its coefficient that some combination of risks will be encountered. To illus-
of variation, which is equal to its standard deviation divided trate the computational basis and the implementation of the
by its expected performance time. approach, a small sample of the EXCEL database used for
modeling uncertainty in the engineering design phase is shown
APPLICATION OF COMPUTERS
in Table 3, and a portion of the EXCEL spreadsheet for mod-
The computer-based system for the assessment of schedule eling the total project schedule risk is shown in Tables 4 and
risk consists of a HyperCard risk factor identification module, 5. The information contained in the table can be displayed
which contains information acquired from many experts and pictorially on the computer screen using the graphical func-
previous construction projects and then uses statistical tech- tions of the EXCEL program.
niques embedded in an Excel spreadsheet. The output of the The EXCEL spreadsheet model also provides the means for
system includes the schedule confidence limits and the risk sensitivity analyses for different outcomes. Sensitivity analyses
profiles of the critical-path activities. can be performed by varying one uncertain element at a time
The HyperCard knowledge base consists of four risk di- and examining the effect of the change in that element on the
mensions, namely, engineering design, procurement, construc- total project performance time. Sometimes changes in an ele-
tion, and project management and risk factors for each dimen- ment will not significantly affect the project schedule. On the
12 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999

J. Constr. Eng. Manage. 1999.125:8-15.

Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

FIG. 5. Structure of HyperCard Knowledge Base

TABLE 4. Sample of EXCEL Spreadsheet Used for Modeling Schedule Risk Design Phase
Activity Time Estimate (months) Relative
Engineering importance
risks Importance Confidence Optimistic Likely Pessimistic Expected Variance (%)
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Productivity High Medium 33 37 45 37.7 14 20
Design errors High Low 36 37 40 37.3 1.6 5
Design criteria High Medium 35 37 41 37.3 3.5 12
Note: Expected performance time of the engineering design phase is 37.8 months.

TABLE 5. Sample of EXCEL Spreadsheet Used for Modeling Schedule Risk: Project Performance

Selected Project Duration (months)

40 45 50 55 60 65 70 75 80
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Probability project
duration (%) 0 0 1 12 54 91 96 99 100
Note: Expected project performance time is 59.6 months.

other hand, sometimes even small changes will have a dispro- cus and common understanding and perception of the
portionately large effect. project risks and their potential consequences.
2. Modeling uncertainty in a project forces a reexamination
BENEFITS OF USING MODEL of the assumptions and identifies which factors drive
schedule performances.
The scheduling of industrial construction projects can ben- 3. The risk identification process using the HyperCard da-
efit greatly through the extension of traditional project sched- tabase can reduce pessimism and lead to the recognition
uling processes to include an assessment of schedule risk pro- of hidden assumptions.
cess. The following are the significant benefits: 4. The system attempts to achieve a balance between the
precise choice of a project performance time and the im-
1. Use of the HyperCard information system provides a fo- precision in the available information.
JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999 / 13

J. Constr. Eng. Manage. 1999.125:8-15.

 5. Risk assessment provides support for early planning de- the prototype system discussed in this paper, should enhance
cisions and can lead to choices giving more valuable the effectiveness of scheduling processes. The following four
prospects for the owner and project manager. recommendations are proposed, which could move the devel-
6. The process allows the owner and the project manager opment process forward:
to act quickly and decisively when a risk event occurs.
7. The system provides a specific instrument and commu- 1. The assessment of schedule risk should be considered
nication structure (i.e., the HyperCard database) for the one of the owners responsibilities for the overall project.
transfer of learning experiences between projects. 2. The identification and measurement of risk should be
included as an essential element of the conceptual plan-
RESPONSIBILITY AND TIMING OF ASSESSMENT OF ning process.
SCHEDULE RISK 3. Some form of an electronic knowledge-based system
should be developed and used within the construction
All risks have a price and risk taker; there are many risk industry to document and transfer project experience and
takers involved with a major construction project: the owner, institutional knowledge to new projects.
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

the financial agency, the project manager, the designer, the 4. A formalized contingency approach should be used when
constructor, the manufacturers and suppliers, labor, the insur- establishing the performance time estimate for a project.
ance agency, the legal advisor, and the public and regulatory A project’s schedule contingency should provide a rea-
agencies. However, by the time a project moves into the im- sonable degree of confidence that the planned project
plementation phase, the risk takers are forced to manage within performance time can be accomplished successfully.
the strategies and tactics of the owner already in place. There-
fore, it is only from the broad context of the owner’s total Interest in risk assessment is growing. With an increasingly
project perspective that all of the risks associated with a proj- complex and rapidly changing business environment, owners
ect can be identified adequately and solutions proposed. and their contractors are being challenged to manage risk
It is not acceptable to wait for risk to emerge before taking while maintaining control and improving performance. How-
risk management action. A more proactive approach is re- ever, some owners are not familiar with the concepts of risk
quired; it is in the owner’s best interest to undertake a consid- assessment. Therefore, the onus must fall on the construction
eration of potential risks early in the project life cycle (Diek- industry to market the concepts of risk assessment so that all
mann et al. 1988). Unless a project is founded on a realistic owners recognize that the analysis of schedule risk in not just
basis the long-term opportunity for the project to be profitable a cost but rather an investment, in terms of actual money
will be reduced greatly. Difficulties also will be revealed in saved, when resources are used more efficiently and the con-
the following three areas of project control: (1) Project coor- sequence of a delay to the project can be avoided.
dination; (2) budget and time management; and (3) general
administration and supervision.
APPENDIX I. SOURCES OF RISKS
SUMMARY, CONCLUSIONS, AFFECTING SCHEDULE
AND RECOMMENDATIONS
1. Engineering design
Since 1960, the construction industry has used quantitative • Regulatory, code, and safety
network – based modeling processes to schedule construction • Requirements
projects; the industry has acknowledged that the use of net- • Drawing control process
work scheduling processes has paid handsome results. Not- • Environment impact assessment
withstanding the successes traditional scheduling processes • Location and number of engineering centers
have enjoyed, on occasion they are deficient in meeting the • Engineering resource qualifications and pool depth
demands of construction practice. • Engineering estimate
Traditional planning and scheduling methods such as CPM • Project scope definition
are limited by the assumption that each activity has only one • Early engineering deliverables
possible outcome and will be completed successfully. In re- • Technology
ality, there are multiple possible outcomes for an activity with • Design criteria
different consequences; and this is something that a schedule • Engineering and procurement interfacing
risk analysis can model. Therefore, to produce an effective • Engineering productivity
schedule, a more thorough scheduling process is required with • Engineering resources requirements
a formal recognition and assessment of project uncertainty, • Material substitution procedure
principally, during the conceptual stage of the project. • Site investigation
The computer-based system discussed in this paper is in- 2. Procurement
volved with construction of a time-based model for the as- • Vendor bid greater than estimate
sessment of schedule risk. The system provides a structured • Long lead items equipment and bulk material
approach to identify the sources of risk in a project and based • Identification of equipment and material
on these risks determine the range of schedule outcomes. The • Management techniques and systems
system can assist in improving the effectiveness of the tradi- • Specification changes affecting manufacturing
tional project scheduling processes by (1) ensuring project par- • Vendor quality control
ticipants formally evaluate uncertainty in the internal and ex- • Vendor drawing control
ternal project environment, (2) opening channels of • Warranties
communications, (3) overcoming the deterministic limitations • Procurement document control process
of traditional scheduling processes, and (4) providing an ap- • Manufacturing process
proach to document institutional knowledge and foster orga- • Material management organization
nizational learning. Thus, the use of the system should ensure • Tender evaluation and purchase order cycle
that all projects are planned in a more realistic manner. • Vendor performance
Microcomputer-based scheduling systems are evolving con- • Transportation concerns
tinuously. Extensions to network modeling concepts, such as • Vendor labor problems
14 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999

J. Constr. Eng. Manage. 1999.125:8-15.

 3. Site construction • Committment to the schedule
• Temporary facilities • Records management
• Approved for construction drawings • Regulatory reporting
• Codes and standards • Change order control
• Standards of contract documents • Owner driven
• Contractor selection process
• Existing facilities APPENDIX II. REFERENCES
• Government permits Archibald, R. D., and Lichtenberg, S. (1994). ‘‘Uncertainty and change
• Labor relations require a new management model.’’ pmNetwork, VIII(5), 6 – 10.
• Labor resource planning Birrell, G. S. (1980). ‘‘Construction planning — beyond the critical path.’’
• Quality control J. Constr. Div., ASCE, 106, 389 – 407.
Cooper, K. G. (1994). ‘‘The $2,000 hour: How managers influence project
• Safety performance through the rework cycle.’’ PM J., XXV(1), 11 – 24.
• Site management process Cowie, G. F., and Carr, J. P. (1984). ‘‘Discussion of ‘integrated project
• Trades productivity and process management’ by Cowie and Carr.’’ J. Constr. Engrg. and
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 09/02/13. Copyright ASCE. For personal use only; all rights reserved.

• Construction turnover coordination Mgmt, ASCE, 110, 121 – 122.

• Design errors Diekmann, J. E., Sewester, E. E., and Taher, K. (1988). ‘‘Risk manage-
• Scope-related quantity increases ment in capital projects.’’ Rep. to the CII, Source Document 41, Uni-
versity of Colorado at Boulder, Colo.
• Percent complete of engineering at start site work Finkel, A. D. (1990). Confronting uncertainty in risk management. Center
• Design changes during construction for Risk Management Resources for the Future, Washington, D.C.
• Constructability reviews Halligan, D. W., et al. (1987). ‘‘Managing unforeseen site conditions.’’
• Contract strategy J. Constr. Engrg. and Mgmt., ASCE, 113, 273 – 287.
• Differing site conditions Ibbs, C. W. (1986). ‘‘Future directions for computerized construction re-
search.’’ J. Constr. Engrg. and Mgmt., ASCE, 112(3), 326 – 345.
• Force mais, oui
Keller, G. et al. (1994). Statistics for management and economics. Wads-
• Availability of equipment and material worth Publishing, Belmont, Calif.
• Loss control program Laufer, A., and Howell, G. A. (1993). ‘‘Construction planning: Revising
• Regulatory delays the paradigm.’’ PM J., XXIV(3), 23 – 33.
• Location Laufer, A., and Stukhart, G. (1992). ‘‘Incentive programs in construction
• Site management staffing projects: The contingency approach.’’ PM J., XXII(2), 23 – 30.
Lichtenberg, S. (1988). ‘‘New project management principles for the con-
• Weather effects ceptual stage: Outline of a new generation.’’ Int. J. of Proj. Mgmt.,
• Construction mistakes 7(1), 46 – 51.
• Site stores management Mohan, S. (1990). ‘‘Expert systems applications in construction manage-
• Defective materials ment and engineering.’’ J. Constr. Engrg. and Mgmt., ASCE, 116(1),
• Site access 87 – 99.
4. Project management Oglesby, C. H., Parker, H. W., and Howell, G. A. (1989). ‘‘Productivity
improvement in construction.’’ McGraw-Hill, New York.
• Third-party overview Project control for engineering. (1986). Publication 6-1, The Business
• Start-up plan Roundtable, New York.
• Major equipment plan Sanvido, V. E., and Merdeiros, D. J. (1990). ‘‘Applying computer-inte-
• Management experience grated manufacturing concepts to construction.’’ J. Constr. Engrg. and
• Project management budget Mgmt., ASCE, 116, 365 – 379.
Thompson, P., and Perry, J. (1992). Engineering construction risks: A
• Project organizational model and implementation guide to project risk analysis and risk management. Thomas Telford,
• Project control process London.
• Project procedures
• Definition of authority and responsibility APPENDIX III. NOTATION
• Scope definition and estimate
The following symbols are used in this paper:
• Financial/funding
• External actions E = expected project performance time;
• Management resource pool te = expected performance time of activity;
• Owner quality assurance VT = variance of project’s performance time distribution;
• Project complexity Vt = variance of activity’s performance time distribution; and
• Project duration Vt1/2 = standard deviation of performance time distribution of ac-
• Project schedule tivity.

JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JANUARY/FEBRUARY 1999 / 15

J. Constr. Eng. Manage. 1999.125:8-15.