Application of Time Buffers to Construction Project

Task Durations
Marion M. Russell 1; Gregory Howell, M.ASCE 2; Simon M. Hsiang 3; and Min Liu, A.M.ASCE 4

Abstract: For this research, a time buffer is defined as the extra time added during planning to individual task durations to compensate for
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

uncertainty and protect against workflow variation to assure a predictable hand-off to subsequent crews. Although previous research has
acknowledged this addition of time buffers, their use in practice has not been studied. This paper reports on what causes people to add and size
time buffers. A nationwide survey was administered to project managers, superintendents, and foremen to identify the most frequent and
severe reasons for adding time buffers to construction task durations. Forty-seven buffer factors were grouped into nine categories: project
characteristics, prerequisite work, detailed design/working method, labor force, tools and equipment, material and components, work/jobsite
conditions, management/supervision/information flow, and weather. Contributions to the body of knowledge include (1) identifying the 12
most frequent and severe causes of time buffer; (2) analyzing (understanding) how buffers are viewed differently by foremen, superintendents,
and project managers, between trades and between general contractors and subcontractors, and the perception among different levels of
experience; and (3) investigating how companies that do not regularly use the Last Planner System and those that do view those factors
differently. Additionally, the research quantitatively developed risk profiles of the buffer factors through an integrated risk assessment
approach. Understanding the application of time buffers and their associated frequency and severity will help construction managers address
potential problem areas and inefficiencies in a prioritized manner. DOI: 10.1061/(ASCE)CO.1943-7862.0000735. © 2013 American Society
of Civil Engineers.
CE Database subject headings: Construction management; Uncertainty principles; Time factors.
Author keywords: Construction; Time; Buffer; Uncertainty; Labor and personnel issues.

Introduction are used in construction, but the underlying causes of time buffers,
both the frequency and severity thereof, are unclear. Although we
There exists a natural tendency in construction to add a buffer probably cannot eliminate all uncertainty and associated time buff-
to tasks due to the inherent uncertainty and resulting variability. ers, understanding and addressing the root causes of time buffers
Buffers in production (such as excess inventory, added capacity, will help us allocate them where they are needed most and conse-
and deliberate delays) are commonly used to protect against uncer- quently reduce project durations and costs. This research serves to
tainty. Perhaps most importantly, buffers mask the sources of fill a gap in the body of knowledge pertaining to the causes of time
uncertainty that make them necessary. While there are several types buffers in construction project task durations. The three primary
of buffer, the focus of this research is the study of time buffers research objectives are as follows:
(hours, days) added to individual construction project task dura- 1. Determine which factors are the most prevalent and severe
tions by construction personnel to compensate for uncertainty causes of time buffers included in construction task durations.
and potential variation. Based on the research to date, there seems Goldratt (1997) and Ballard (2000) have acknowledged the
to be a definite and growing interest in the study of several aspects existence of time buffers in task durations and recommended
of buffers in construction (Howell et al. 1993; Ballard and Howell methods to manage buffers. Neither author examined the cause
1995; Goldratt 1997; Ballard 2000; Horman 2000; Horman et al. of time buffers. This research investigates the causes of uncer-
2003; Alves and Tommelein 2004; Horman and Thomas 2005; Lee tainty or potential problem areas that construction personnel
et al. 2006; Hopp and Spearman 2008). It seems evident that buffers consider when adding time buffers to work plans or task
durations and determine which are the most important from
1
Graduate Student, Dept. of Civil, Construction, and Environmental both a frequency and severity standpoint.
Engineering, North Carolina State Univ., Raleigh, NC 27695-7908. E-mail: 2. Using an integrated risk assessment approach, classify the 47
[email protected] individual risk factors as risk profiles.
2
President, Lean Construction Institute, 1400 N. 14th St., #400, Ideally, it would be desirable to eliminate all of the uncer-
Arlington, VA 22209. E-mail: [email protected] tainty and variation associated with the different factors; how-
3
Derr Professor, Dept. of Industrial Engineering, Texas Tech Univ., ever, in today’s construction industry environment of limited
Lubbock, TX 79409. E-mail: [email protected] resources, risk profiles help to categorize and prioritize the fac-
4
Assistant Professor, Dept. of Civil, Construction, and Environmental tors that management needs to address.
Engineering, North Carolina State Univ., Raleigh, NC 27695-7908 (corre-
3. Determine the differences in opinion and perception between
sponding author). E-mail: [email protected]
Note. This manuscript was submitted on March 11, 2012; approved on different levels of management, different trades, and different
May 13, 2013; published online on May 15, 2013. Discussion period open levels of experience, the difference between general contrac-
until December 24, 2013; separate discussions must be submitted for indi- tors and subcontractors, and the difference between contrac-
vidual papers. This paper is part of the Journal of Construction Engineer- tors using traditional management approaches and those
ing and Management, © ASCE, ISSN 0733-9364/04013008(10)/$25.00. using lean construction techniques.

© ASCE 04013008-1 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

 A time buffer is added in varying amounts and for varying Table 1. Types of Buffer Defined
reasons depending on whom one asks. The reason for these Type of
differences can be tied to the differences in the levels of under- buffer Definition
standing, control of the levels of management on a construc-
Inventory Buffers of physical material stockpiles (Horman and Thomas
tion jobsite, and other construction environment factors. 2005); large buffers of inventory can lead to congestion, which
impedes performance, but material stockpiles that are too low
can lead to stopped, slowed, or disrupted production.
Literature Review Capacity Buffers of additional manpower or equipment provided to an
operation beyond the anticipated need for completion (Horman
When addressing a topic such as buffers, it is very important to first and Thomas 2005); additional capacity gives an operation the
clearly define what they are. There exist not only several definitions ability to rapidly respond to situations caused by uncertainty
and functions of a buffer in the construction literature but also sev- and variability; too much capacity buffer can also result in
inefficient labor and equipment use.
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

eral types of buffer addressed in that literature. This literature re-

Plan Buffers that are inventories or backlogs of workable
view will summarize those definitions and functions and includes a
assignments (Ballard and Howell 1995); a plan buffer provides
discussion of five main types of buffer in the construction literature: alternative tasks for crews to perform that keep them working
financial, inventory, capacity, plan, and time. Hornby (1974) wrote in the right sequence when the main tasks planned cannot be
that a buffer is an apparatus for lessening the effect of some impact. performed or when assignments are completed sooner than
Horman and Kenley (1998) defined a buffer as an allowance used to expected.
accommodate the impact of unexpected influences and other diffi- Time Buffer that takes the form of additional time added into a task
culties encountered in a construction project. Alves and Tommelein to protect against uncertainty and to absorb variation; the
(2004) define buffers concisely as resource cushions, e.g., money, concept of float is one such use of a time buffer and is seen in
time, materials, and space, used to protect processes against varia- the critical path method. Float provides some flexibility in
determining start dates for activities, without delaying the
tion and resource starvation. According to Ballard and Howell,
project’s completion (Alves and Tommelein 2004). A similar
buffers serve to provide a cushion or shield against the negative example is the use of a deliberate pause or time lag inserted
impact of disruptions and variability (Howell et al. 1993; Ballard between steps in an operation (Horman and Thomas 2005).
and Howell 1995). Production systems must be able to absorb Lee et al. (2006) point out that time buffers have been used
variation to avoid loss of throughput, wasted capacity, inflated mainly as a contingency such as adding a percentage of the
cycle times, larger inventory levels, long lead times, and poor cus- activity duration at the end of the activity to accommodate
tomer service (Hopp and Spearman 2008). Ballard (2005) has uncertain and variable conditions.
called construction one type of production system, albeit one of Financial Money in construction project budget reserved to pay for
greater complexity and uncertainty, that uses many types of buffer unforeseen design or construction costs (Risner 2010).
to absorb variation that occurs due to uncertainty in construction
projects. The construction literature focuses on five main types
of buffer. Hopp and Spearman (2008) list inventory, capacity,
and time as three types of buffer. Ballard and Howell (1995) intro-
duced a fourth type of buffer called the plan buffer. The literature
also discusses the use of cost contingency as a financial buffer.
These five types of buffer are described in Table 1.
The focus of this research is the within-activity time buffer added
to individual task durations to compensate for uncertainty and pro-
tect against variation. Fig. 1 is intended to help illustrate the time
buffer discussed in this research. A time buffer is the difference
between estimated/planned duration and the minimum duration the Fig. 1. Activity time buffer
task should take based on optimum or baseline productivity.
A time buffer in tasks is usually subject to either Parkinson’s law
or the student syndrome (Lechler et al. 2005). Parkinson’s law says planning techniques to develop schedules for each milestone-based
that people will always use this time buffer because the task will phase of work. Both Goldratt and Ballard acknowledge the exist-
grow to take as long as the time allotted for it to be done. The stu- ence of time buffers and suggest methods for managing them, but
dent syndrome is based on the tendency of people to waste time little research has been done to examine and identify the root causes
buffers by starting their tasks later. Both of these issues result in of time buffers in construction project task duration estimates. A
reduced productivity. Goldratt (1997) argues that instead of being literature review revealed a gap in the body of knowledge in regards
used to deal with uncertainties and variation, a time buffer is always to the causes and perceptions of time buffers in construction.
used as a part of the time to perform a task. Goldratt developed Factors that affect construction productivity are equally impor-
Critical Chain Project Management to manage this time buffer tant in developing a thorough list of factors for investigation. A
and calls for removing all buffers within activities and placing them significant amount of research has been done on factors affecting
at the end and allowing activity delays to be absorbed by the pooled construction productivity. Borcherding and Gardner (1981) identi-
buffer. Unfortunately, there are few if any construction projects fied material and tool availability, rework, overcrowded work areas,
published where this method has been used successfully. One rea- inspection delays, foreman incompetence, crew interference and
son may be that allowing a pooled buffer to absorb delays will have turnover, and foreman changes as the top factors affecting produc-
an immediate and problematic domino effect on the downstream tivity. Researchers in the 1990s identified scope, work content,
activities in regards to, for example, schedule, manpower, equip- work complexity, design features, specifications, rework, materials,
ment, and cost due to the complex interdependencies existing in tools, construction equipment, information, weather, site conges-
construction projects. As part of the Last Planner System, Ballard tion, crew size and skill, design accuracy, degree of repetition,
(2000) introduced Phase Scheduling, which uses pull and team working conditions, and site access as factors affecting productivity

© ASCE 04013008-2 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

 (Thomas and Sakarcan 1994; Portas and AbouRizk 1997; Somnez related to communication, trust, changes, and getting ques-
and Rowlings 1998). In the last decade, research has found addi- tions answered when they arise.
tional factors, including scheduling, manpower experience and 9. Weather: concerns/uncertainty about the climate at the loca-
motivation, scope changes, lack of detailed planning, inadequate tion of the project and the prevalent weather conditions, such
supervision, lack of information, lack of foreman planning and as temperatures, rain, and wind.
communication skills, poor communication between foremen and The 47 potential reasons for adding buffer to construction task
project managers, engineering drawing management, lack of craft- duration estimates were assigned to the appropriate aforementioned
level technical training, poor quality of plans and specifications, category. Table 2 displays the entire list of buffer factors and the
slow response to questions, and lack of qualified labor (Rojas and nine categories to which they belong.
Aramvareekul 2003; Liberda et al. 2003; Dai et al. 2009; Kimpland A general information survey for contractors and a time buffer
2009). Wambeke et al. (2011) completed a thorough literature re- survey were developed and used to answer the research questions.
view of productivity factors, including the factors previously men- The contractor general information survey was typically completed
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

tioned, to identify 166 factors. Tommelein et al. (1999) stated that by a project manager, collecting information such as the company
successful project managers take proactive steps to establish buffers type (subcontractor or general contractor), company size, annual
to shield crews from variability in construction. They noted that revenue, average project size, backlog or pending work, and
common causes included change orders, late replies to requests whether or not they use the Last Planner System. The time buffer
for information, lack of materials, physical interference between survey included three sections. The first section asked participants
materials, work-space congestion, and others but did not develop to provide background information such as trade, position, and
a thorough list of factors or explore them further. experience. Sections 2 and 3 involved obtaining feedback pertain-
ing to the previously discussed categories and individual reasons
for adding buffers in task durations. In Section 2 the participants
Methodology were asked to rank order the nine overall categories based on which
categories they felt contained the most critical and prevalent causes
A survey was developed to study which causes of uncertainty or of uncertainty when they are budgeting time into task durations.
concerns about potential for variation result in the most frequent The participants were also asked whether or not they felt the
and severe addition of a time buffer in construction task durations. uncertainty associated with the category was their responsibility
Through a combination of a literature review and research team to control or mitigate. The third section expanded the nine catego-
discussions with construction project managers, superintendents, ries into the 47 factors or reasons that affect task duration estimates.
and foremen, 47 individual factors related to time buffers in con- The purpose of the third section was to find out which factors most
struction tasks were identified and included in the survey. The frequently and severely affect task duration estimates. For each
seven categories that impact productivity established by Koskela factor, the respondents were asked how frequently the factor influ-
(2000) to include connected work, detailed construction design, enced their duration estimate by circling one of the following seven
components and materials, workers, equipment and tools, space, frequency responses: never, rarely, occasionally, sometimes, fre-
and external conditions, along with Wambeke et al. (2011) added quently, usually, or always. Next, for each factor, the respondents
eighth category of management/supervision/information flow, were were asked to consider a 2-week (10-day) activity and estimate how
used as a framework to separate the 47 buffer factors. Based on much time (days) they would include or allocate for the given factor
the nature of some of the identified buffer factors, one additional in their duration estimate to protect against the effects of uncer-
category was included: project characteristics. The nine categories tainty. The respondents chose one of the following seven severity
are listed and described below: responses: 0, 0.5, 1, 2, 3, 5, or 7. The decision to use seven choices
1. Project characteristics: concerns/uncertainty about characteris- for both frequency and severity was made to balance having too
tics specific to a project and a given trade. many choices with still being able to capture just noticeable dif-
2. Prerequisite work: items that must be completed before one ferences. Fig. 2 provides an example of a question from the third
can start a task. If there is concern or uncertainty that the item section of the contingency survey. The question in the example is
(permits, prerequisite work, or rework on a prerequisite task) about factors related to labor.
will not be completed on time, then consider the extent to A pilot study was conducted with a construction company in
which one’s duration estimate is affected. This can also be Colorado and a local construction company to help avoid potential
thought of as confidence in the schedule or work plan. problems associated with the length and clarity of the questions and
3. Detailed design/working method: concerns/uncertainty about the instructions provided for completion of the survey. Addition-
having an accurate and available design and a feasible working ally, the pilot study was used to help validate the relevance of
method to complete the required task. the factors chosen and allow suggestions for additional factors.
4. Labor force: concerns/uncertainty about availability, reliabil- Overcommitment was modified as liability pressure and included
ity, and capability of the labor force to complete the required concerns about overcommitting due to a tight schedule, liquidated
task. damages, and contractual deadlines. The factors of positive com-
5. Equipment and tools: concerns/uncertainty about the availabil- pany recognition, preparing for duration negotiation, concerns
ity, reliability, and capability of the required equipment and about being pushed into using more manpower, and poor perfor-
tools to complete the required task. mance due to unfamiliarity (with the scope) were added. The sur-
6. Material and components: concerns/uncertainty about receiv- vey was distributed (online and hard-copy versions) to general
ing the correct and necessary materials from suppliers when contractors and subcontractors throughout the United States. All
and where you need them. This can be thought of as trust or participating companies and respondents completed the surveys
confidence in one’s suppliers. voluntarily. Four local contractors were provided surveys for dis-
7. Work/jobsite conditions: concerns/uncertainty about the phy- tribution, and another nine contractors from six different states were
sical space available to perform one’s job. contacted via a construction consulting company. Two professional
8. Management/supervision/information flow: concerns/ civil engineering organizations also provided contact information
uncertainty about the management system regarding issues for multiple construction companies throughout the United States.

© ASCE 04013008-3 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

 Table 2. Time Buffer Factors A total of approximately 175 different companies across 37 states
Category Individual cause of time buffer received the survey and a request for completion during the summer
of 2011, with the survey period running approximately 10 weeks
Project Contract delivery method
from August 1 until October 14.
characteristics Contract period
Size of project
Complexity of project (interdependency of activities)
Complexity of task (degree of difficulty/inherent Survey Response
nature of work)
Size of company The final count of useable surveys was 180 surveys from 36 different
Prerequisite work Delays in obtaining permits companies including both general contractors and subcontractors.
Completion of prerequisite work (work before you is General contractors made up 28% of the participating companies
not done yet) and subcontractors the other 72%. Project managers accounted
Rework required due to quality of prerequisite work for 51% of the responses, superintendents completed 27% of the
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

Delays in inspections for previously completed surveys, and foremen the remaining 22%. The trades were sepa-
work rated into four trade groups for comparison. The utilities category
Detailed design Design constructability included the mechanical, electrical, plumbing, and fire protection
and Quality of documents trades and accounted for 44% of the responses. The structural cat-
working method Poor performance due to unfamiliarity with scope of egory included the steel, concrete, masonry, roofing, and earthwork
work
trades and accounted for 22% of the responses. The finishing/
Strict specification requirements
Quality control requirements surface category included carpentry, drywall, ceiling, painting, and
Low degree of repetition in tasks glazing and accounted for 17% of the responses. The last category
Labor force Reliability of labor force was for general contractors responsible for multiple trades. This
Availability of labor force category accounted for the remaining 17% of responses.
Inefficiencies in crew due to lack of experience/skills
Concerns about being pushed into using more
manpower and creating inefficiencies Survey Analysis
Low morale or lack of motivation
Language barrier among workers/supervisors
Equipment and Reliability of equipment or tools Research Objective 1
tools Availability of equipment or tools The first objective of the research was to determine which factors
Capability (productivity) of equipment and tools are the most prevalent and severe causes for time buffers to be
Time required to repair equipment if breakdown
added to construction task duration estimates.
occurs
Table 3 summarizes the results of Section 2, which asked the
Time required to replace equipment if breakdown
occurs participants to rank order the uncertainty in the nine categories
Materials and Receiving incorrect quantity of materials and then answer whether or not they felt the uncertainty related
components Receiving incorrect material type or damaged to that category was their responsibility to manage. It is interesting
materials to note that as the category rank went up (i.e., less uncertainty
Receiving materials for task later than expected/ associated with the category), the percentages of participants who
planned felt responsible for managing the category were higher. This means
Work and jobsite Overcrowded or cluttered work area/jobsite the categories such as management/supervision/information flow,
conditions congestion labor force, equipment and tools, and materials and components
Difficult access to work area had less uncertainty as a result of responsibility to control or
Method of material transfer required from receiving
mitigate that uncertainty. On the other hand, the detailed design/
area to task location
Distance of material transfer required from receiving working method, project characteristics, work/jobsite conditions,
area to task location and prerequisite work had the most uncertainty associated with
Management, Confidence in request for information (RFI) process them, yet less than half of the participants felt the uncertainty was
supervision, and Liability pressure (e.g., liquidated damages, their responsibility to mitigate.
information flow contractual deadlines) To determine the most prevalent and severe causes for time
Preparing for duration negotiation buffers to be added to construction task durations, the 47 factors
Positive company recognition contributing to uncertainty in task duration planning were rank
Trust in superintendent (based on reputation, ordered based on their average frequency and severity responses.
experience, and knowledge) The frequency responses were first assigned a quantitative value.
Trust in project manager (based on reputation,
The responses of never, rarely, occasionally, sometimes, frequently,
experience, and knowledge)
Trust in owner (based on reputation, experience, and usually, and always were assigned values of 0, 1, 2, 3, 4, 5, and 6,
knowledge) respectively. The top 12 (i.e., the top 25%) frequency factors and
Required coordination with other trades severity factors are summarized in Table 4. The most frequent and
Changes in scope of work (tendency of owner to severe factors coincide strongly with the overall category ranking
make changes) completed in Section 2 of the survey and previously discussed. A
Communication between owner/engineer and project few of the top ranked factors are consistent with previous research.
manager Lee et al. (2006) found that design errors (quality of documents)
Communication between project manager/ and changes (tendency for scope changes) are two of the main
superintendent and foreman factors that cause uncertainty in construction. Chan and Au (2009)
Communication between foreman and workers found project complexity to be one of the most important factors
Weather Climate—potential weather conditions associated
that contractors consider when they are pricing time-related
with project location
contract risks. Additionally, one factor just outside the top 12,

© ASCE 04013008-4 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Contingency survey example question

at 13, the request for information process, is consistent with that were not included in the top 12 frequency factors. This finding
Kimpland’s (2009) finding that slow responses to questions was leads to the second research objective of determining the risk pro-
one of the top external factors that impacted productivity, which file to which the different factors belong.
led to uncertainty. After speaking to a few of the respondents, it
became apparent that contract period was closely tied to the liability
pressure factor in its interpretation as contract periods result in pres- Research Objective 2
sure to finish on time and can lead to overcommitment. Similarly, The second objective of the research was to classify the 47 indi-
prerequisite work was closely associated with project complexity or vidual risk factors into risk profiles using an integrated risk assess-
interdependency of tasks. The top three factors, project complexity, ment approach.
complexity of trade task, and quality of documents, were the same Risk assessment involves identifying risks in the area of concern
for both frequency and severity, but in a slightly different order. The and then classifying the risk by assigning an appropriate risk level
factors of required coordination with other trades, contract period, (Wheeler 2002). The first objective of this research is equivalent
material transfer distance, material transfer method, and work area to risk identification because it identifies the root causes of time
access were highly ranked frequency factors that did not show up buffers in construction project tasks. As a complement, the second
in the top 12 severity factors. Conversely, the factors of strict speci- objective creates risk profiles through an integrated approach of
fication requirements, quality control requirements, low degree of histogram data analysis and a risk assessment matrix, commonly
repetition, and late materials were highly ranked severity factors used in military planning, to analyze which factors belong to each
risk profile. An integrated approach was used to create risk profiles
for the 47 factors and determine which factors had a high-frequency
Table 3. Summary of Section Two Responses and high-severity time buffer occurrence, a high-frequency and
Overall Responsible low-severity time buffer occurrence, a low-frequency and high-
category (Percentage severity time buffer occurrence, and a low-frequency and low-
Category rank “yes”) severity time buffer occurrence. First, a histogram data analysis was
Project characteristics 2 22
run on the frequency and severity responses for the 180 completed
Prerequisite work 4 48 surveys. Histograms use tabular frequencies over discrete intervals
Detailed design/working method 1 46 or bins to show the proportion of cases that fall into each of several
Labor force 7 85 categories (Howitt and Cramer 2005). The bins for this analysis
Equipment and tools 9 94 were the seven different response options for both frequency
Materials and components 6 86 and severity. An example of the histogram output for the quality-
Work/jobsite conditions 3 44 of-documents factor can be seen in Table 5.
Management/supervision/information flow 5 76 Next, four separate counts were calculated for each factor’s fre-
Weather 8 2
quency and severity. Each count was based on the sum of the upper

© ASCE 04013008-5 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

 Table 4. Overall Top 12 Most Frequent and Severe Causes of Time Buffers
Average
Cause frequency Category
Project complexity 3.90 Project characteristics
Complexity of trade task 3.79 Project characteristics
Quality of documents 3.54 Detailed design/working
method
Size of project 3.32 Project characteristics
Required coordination with 3.23 Management/supervision/
trades information flow
Contract period 3.11 Project characteristics Fig. 3. Risk assessment matrix (Department of the Army 1998)
Design constructability 3.01 Detailed design/working
method
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

Tendency for scope changes 2.93 Management/supervision/

An attempt was made to sum the upper and lower three
information flow
Material transfer distance 2.90 Work/jobsite conditions
bins, which confirmed that the use of two bins was adequate. This
Material transfer method 2.88 Work/jobsite conditions research then integrated the use of a risk assessment matrix, com-
Work area access 2.84 Work/jobsite conditions monly used in military planning, to analyze which factors belonged
Weather/climate 2.84 Weather to each risk profile. According to Wheeler (2002), risk has two
Average
components: a probability or likelihood of occurrence and a con-
Cause severity Category sequence of occurrence. Consequence is typically defined as some-
thing impacting cost, schedule, or performance parameters. These
Quality of documents 2.14 Detailed design/working two components are represented well by the survey questions in
method
this research, with likelihood being the frequency with which each
Project complexity 1.93 Project characteristics
Complexity of trade task 1.91 Project characteristics factor causes uncertainty affecting task duration estimates and con-
Tendency for scope changes 1.61 Management/supervision/ sequence being the severity in days of time buffer each factor
information flow caused in those task durations. The resulting risk, categorized as
Weather/climate 1.56 Weather low, moderate, high, and extremely high in the risk matrix, can then
Design constructability 1.44 Detailed design/working be prioritized for mitigation or elimination (Department of the
method Army 1998). This risk assessment matrix can be seen in Fig. 3.
Size of project 1.41 Project characteristics The normalized counts of low and high frequency were plotted
Work area access 1.35 Work/jobsite conditions on the x-axis and the normalized counts of low and high severity
Strict specification 1.34 Detailed design/working
were plotted on the y-axis, resulting in four separate risk plots. The
requirements method
Quality control 1.24 Detailed design/working risk assessment matrix was essentially overlaid onto the risk plots
requirements method to identify those factors in the upper right (high and extremely high)
Low degree of repetition 1.24 Detailed design/working regions. The researchers calculated the risk level or expected value
method of risk (product of frequency and severity) for the factors in each
Late materials 1.24 Materials and components profile to determine the top 12 important factors for each profile.
These factors are summarized in Fig. 4.
Although some factors tended to have moderate frequency and
Table 5. Histogram Output for Quality-of-Documents Factor severity throughout the four graphs and did not end up in a specific
profile, this method accounted for the profiling of 77% (36=47) of
Factor Count of Factor Count of
frequency responses severity responses the factors. Additionally, some of the most prevalent factors overall
showed up in more than one risk profile. The low-frequency/
0 18 0 51 high-severity factors were the most difficult to profile as very few
1 7 0.5 4
of the factors were obviously located in the high or extremely high
2 22 1 33
3 28 2 33 portion of the risk matrix. Taleb’s black swan theory provides some
4 51 3 24 explanation for this result in the low-frequency/high-severity pro-
5 25 5 16 file. Taleb says that planners fail to buffer or protect against the rare
6 29 7 19 yet severe factors of uncertainty (those he calls the black swans)
(Taleb 2007). Wambeke’s (2011) case study research on the causes
of variation shows the existence of black swans in the construction
or lower two values. The high-frequency sum was based on the industry as material delivery (late materials), overcommitment
total count of “usually” (5) and “always” (6) responses. The high- (liability pressure), and request for information processes were
severity sum was based on the total count of “5 day” and “7 day” three causes of low-frequency/high-severity variation. Another pos-
responses. The low-frequency sum was based on the total count of sible explanation is that there is not a lot of frequency and severity
“never” (0) and “rarely” (1) responses. The low-severity sum was tradeoff in the construction industry and it is more about ranking or
based on the total count of “0 day” and “0.5 day” responses. The prioritizing the uncertainty. In other words, construction personnel
decision to use only the sum of the upper and lower two bins was tend to add more time buffers to those factors about which they are
made to emphasize the most important factors for each profile, and more frequently concerned. This correlation is apparent in the
the counts of the three bins in the middle were not included. The graph of the overall average frequency and severity in Fig. 5.
analogy of using a highlighter when reviewing an article applies. The graph in Fig. 5 was further examined by first partitioning
Typically one only highlights the major or most important points. If the data points into four risk levels using SAS software to conduct
someone were to highlight over half or the entire article (i.e., sum the cluster analysis. Cluster analysis provided a method for inves-
upper and lower three or four bins), there would be no highlights. tigators to decide where to draw lines between similar groups or

© ASCE 04013008-6 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Factor risk profile summary grid

clusters. Clustering hierarchically groups or clusters objects so that are the causes of time buffers that construction personnel are most
objects in the same cluster are more similar to each other than concerned about due to uncertainty and potential for variation. As
objects in the other clusters (SAS Institute 2012). The causes in one moves down the steps, the confidence in the causes increases
the upper steps represent those requiring attention first as they and there is less uncertainty and need for a time buffer in the task

Fig. 5. Step graph based on average frequency and severity

© ASCE 04013008-7 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

 durations. Clustering helped reveal something interesting about the work area access was more important to project managers (8th)
step graph. There is a greater tendency to tolerate or manage risk than superintendents (12th) and foremen (14th), perhaps due to
when the risk is lower (higher level of confidence). That is, there is their responsibility for the entire construction site rather than
much less tradeoff between severity and frequency in the bottom smaller work areas. Liability pressure was a much greater concern
step or lowest risk level. There is a greater tendency to trade fre- for superintendents (10th) and project managers (18th) than
quency for severity and vice versa when there is a high level of risk foremen (36th).
such as in the last three steps. This was determined by calculating The severity or magnitude of buffers due to uncertainty related
the change in frequency and severity between any two factors to scope changes was the most severe factor for foremen and ranked
within each step or cluster and comparing the percentage of positive sixth for superintendents and eighth for project managers. Also,
values (no tradeoff between frequency and severity) to the percent- required coordination with other trades was again a high emphasis
age of negative values (tradeoff between frequency and severity). factor for foremen coming in at 4th, while superintendents ranked it
The percentage of negative values for each step is included in Fig. 5. 24th and project managers ranked it 15th. Several of these differ-
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

This resulting reverse in correlation present in the partitioned ences are evident in both of the aforementioned tables. Overall the
groups is evidence of an occurrence called Simpson’s Paradox. results indicated a perception of larger frequency and severity of
Simpson’s paradox says that a correlation present when a data uncertainty as one moves from project manager to superintendent
set is analyzed becomes reversed when the set is broken into similar to foreman. One area not highlighted by the top 12 most frequent
groups (Pearl 2009). or severe causes is the much greater emphasis put on materials,
equipment, and labor by foremen than superintendents or project
managers. For example, foremen ranked the frequency of crew in-
Research Objective 3 efficiencies, labor reliability, and labor availability 14th, 15th, and
The second objective of the research was to determine the differ- 17th, respectively. Superintendents ranked those same three factors
ences in opinion and perception between the various levels of 32nd, 31st, and 27th, while project managers ranked them 36th,
management, various trades, and various levels of experience, 40th, and 23rd. Here, those actually doing the work are most con-
the difference between general contractors and subcontractors, cerned about the resources it will take to complete the work.
and the difference between contractors using traditional manage- The next four comparisons are summarized in the following
ment approaches and those using lean construction techniques. discussions. Upon examining the survey results by trade group,
The survey results do not allow us to determine a specific fre- the utility trades were found to have the largest frequency and
quency or buffer amount for each factor as participants were asked severity for the time buffer factors. The structural trade groups were
to consider the factors one by one and independent of each other. second and the finishing trades third.
Construction personnel likely consider multiple factors at once The researchers also hypothesized that the experience level of
when estimating task durations. However, the survey responses the participants might result in different perceptions. Specifically,
do allow for a comparison of the average frequency and average the least experienced (grouped by 5 years or less) participants
severity among different groups. The first comparison made was would be concerned about more of the factors more frequently
between the different levels of management—project managers, and buffers with greater magnitude. The least experienced group
superintendents, and foremen. The survey responses were sorted was compared to a group having 5–25 years of experience and a
by respondents’ job position and then the average frequency and group having more than 25 years of experience. The survey results
severity were taken for each factor. The frequency of the factors found that although there was not a large percentage difference
as causes of uncertainty and the magnitude of buffer added to pro- between the least experienced and the more experienced groups
tect against that uncertainty were then rank ordered for each job in regards to frequency, the least experienced group buffered with
position. A total of 17 factors were required to capture the top nearly 30% more magnitude (severity) than the more experienced
12 most frequent causes, and a total of 18 factors were required groups. General contractors as well as subcontractors participated
to capture the top 12 most severe causes for each level of manage- in the survey. The scopes and areas of responsibility are different
ment. The top 12 (approximately 25%) were included to highlight for these two entities in the completion of construction projects.
the differences in perception about the causes outside of the top 5. Overall, the frequency for general contractors and subcontractors
The top five most frequent and the top three most severe causes only differed by approximately 8% with the subcontractors buffer-
were nearly identical, aside from occurring in a slightly different ing slightly more frequently. The general contractors buffered more
order. Beyond those, there were some noticeable differences in frequently than the subcontractors for all project characteristic
the top 12 among the levels of management as well as some notable factors. This finding was expected because general contractors
differences outside of the top 12. In regards to frequency, foremen are responsible for the entire project scope and all of the involved
ranked required coordination with other trades as the third most subcontractors. The subcontractors, on the other hand, were more
important; however, superintendents and project managers ranked frequently concerned about the materials, equipment, labor, and
this factor fourth and sixth, respectively. This difference is possibly work/jobsite condition factors than the general contractors. When
a result of the foremen being more involved in those coordination looking at the amount subcontractors buffer compared to general
efforts than higher management levels. Perhaps for a similar reason, contractors, the results are staggering. Subcontractors perceive
material transfer distance and method were ranked higher by the the need to buffer against the uncertainty of factors with 30%
foremen than superintendents or project managers. Another big dif- greater magnitude than the general contractors.
ference between foremen and the superintendents and project man- The participants of the survey were sorted based on their re-
agers was in the factors of overcrowded jobsite and request for sponses in an effort to make a comparison of construction person-
information process. Foremen ranked these 10th and 11th, respec- nel who use lean construction techniques such as LPS with those
tively, while overcrowded jobsite came in at 16th for superintend- who use traditional construction management techniques. One of the
ents and project managers and request for information process goals of the Last Planner System is to reduce the amount of uncer-
came in at 25th for superintendents and 22nd for project managers. tainty that exists in a construction project through the use of col-
The tendency for scope changes ranked 7th among foremen and laborative and pull planning techniques. In theory, more reliable
superintendents, but only 15th for project managers. Conversely, planning and the reduction of uncertainty will reduce variation

© ASCE 04013008-8 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

 and the need for large time buffers in task durations. The survey tendency for scope changes, weather/climate, design constructabil-
results showed that those using LPS on their construction projects ity, project size, work area access, strict specification requirements,
had a lower frequency on 72% of the factors and a smaller amount quality control requirements, low degree of repetition, and late
of buffer on 85% of the factors. Further, those using traditional con- materials. Most of the top overall factors are associated with such
struction management methods buffered with 14% more frequency intangibles as information and communication. In fact only the late
and 41% more severity than those using lean construction methods materials factor directly involves a resource (materials, equipment,
such as LPS. This result is interesting and merits further investiga- labor) concern. Based on the survey results, there is a need for
tion into the effects and benefits of using LPS to achieve reliable emphasis on improving this communication and information flow
planning and reduce buffer amounts in task duration estimates. in construction.
The results of the research provide construction managers with
two tools to help target those factors related to uncertainty and
Validation resulting in the most frequent and severe time buffers in their con-
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

struction projects. Depending on the available resources, construc-

Discussions were held with a project manager, a superintendent, tion managers should target the high-frequency/high-severity
and a foreman in regards to the survey results. The project manager factors first, followed by the low-frequency/high-severity factors,
worked for a general contractor, while the superintendent and fore- and finally the high-frequency/low-severity factors. The low-
man worked for a trade subcontractor. All three had over 15 years frequency/low-severity factors are least likely to require attention.
of construction industry experience. None had seen the results of Since the risk profiles do not account for all of the causes, the step
the survey. The project manager mentioned that his biggest con- graph can be used as a secondary tool to prioritize those causes
cerns in task planning related to how complex the project was management and the construction team should focus on first.
(how many subs, activities, and their interactions), having a com- Regardless of which tool is used as a guide, it is very important
plete and accurate set of design drawings, and making sure his subs that management address these factors and develop effective strat-
got where they needed to be when they needed to be there. The egies to reduce the need for buffers because reducing the amount of
superintendent said he was most worried about getting the job done time buffers in construction task durations should improve produc-
right and on time (liability pressure) while ensuring coordination tivity and might also reduce both project durations and costs.
between the trades on the job in regards to both resources and The third objective involved comparing the differences in opin-
timing (prerequisite work/interdependency of tasks). The foreman ion and perception among the different survey groups. Overall, the
said he worried about change orders, material, labor, and equip- frequency and severity increase as one moves from project manager
ment. These responses are fairly consistent with the research find- to superintendent to foreman. The survey analysis also highlighted
ings and provide some validation, but additional research beyond larger frequency and severity of time buffers perceived by trades
the scope of this paper is being conducted by the authors. A series with more complex tasks and greater interdependency such as
of case studies is being conducted to gather additional survey data mechanical, electrical, and plumbing. Experience was also shown
as well as empirical data pertaining to time buffers in construction to affect how much time buffer is included in construction task
task durations. durations. No or limited experience (5 years or less) resulted in add-
ing a larger amount of time buffer. Due to the differences in scope
and responsibilities, general contractors and subcontractors were
Conclusions compared to reveal the differences in their perceptions. A final
A couple of research limitations must be acknowledged. First of all, comparison between construction personnel who use lean construc-
the subcontractors and general contractors who participated in the tion techniques such as the LPS and those using traditional
survey were all from the commercial sector of the construction construction planning techniques was made. The results of our
industry. Although this is a research limitation, the commercial survey corroborated at a minimum that the perception of those
sector is an important sector, accounting for approximately 48% using LPS was one of a reduced need for the frequency and mag-
of construction employees (Department of Labor 2012) and nitude of time buffers and led to the need for further empirical in-
25% of the total construction value put in place each year in the vestigation into the amount of time buffer in projects using LPS.
United States (National Research Council 2009). A second limita- The contributions to the body of knowledge include providing
tion of the survey is that the likely difference between union and an understanding of why time buffers are added and quantifying
nonunion perceptions of risk was not captured. The survey origi- the influence of various causes on the way construction personnel
nally included a question asking whether the respondents worked at various levels of management buffer and plan for uncertainty.
on union, nonunion, or both types of projects. The pilot study re- Filling this gap of knowledge will help construction companies take
vealed a reluctance to answer this question and concerned union proactive steps toward addressing those causes, managing uncer-
workers would not complete the survey. As a result, the question tainty, and reducing the associated time buffers in construction
was removed. The survey data input are also limited to those per- projects. Understanding how the members of a team think and plan
sonnel who volunteered to complete the survey from each of the is critical to the success of a project. Further, Hopp and Spearman
participating companies. (2008) asserted that understanding the underlying causes of vari-
This research addressed three main objectives, and the results ability and the buffers it begets is essential to the design and man-
have led to further questions for investigation. First, the overall agement of efficient production systems. This research is an effort
most frequent and severe causes of time buffers were identified. to understand and examine those buffers, specifically time buffers
The top 12 most frequent causes of time buffers were project com- as they relate to construction tasks.
plexity, complexity of the trade task, quality of documents, project
size, required coordination with other trades, contract period, de-
References
sign constructability, tendency of scope changes, material transfer
distance, material transfer method, work area access, and weather/ Alves, T. C. L., and Tommelein, I. D. (2004). “Simulation of buffering and
climate. The top 12 most severe causes of time buffers were quality batching practices.” Proc., 12th Conf. of the International Group for
of documents, project complexity, complexity of trade task, Lean Construction (IGLC 12), Lean Construction Institute, 277–290.

© ASCE 04013008-9 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.

 Ballard, G. (2000). Phase scheduling, White paper-7, Lean Construction Koskela, L. J. (2000). “An exploration towards a production theory and
Institute, Arlington, VA, 〈http://www.leanconstruction.dk/media/18435/ its application to construction.” Ph.D. thesis, Helsinki University of
Phase_Scheduling_.pdf〉 (Feb. 1, 2013). Technology, Helsinki, Finland.
Ballard, G. (2005). “Construction: One type of project production system.” Lechler, T. G., Ronen, B., and Stohr, E. A. (2005). “Critical chain: A new
Proc., 13th Annual Conf. on Lean Construction, Lean Construction project management paradigm or old wine in new bottles?” Eng.
Institute, 29–35. Manage. J., 17(4), 45–58.
Ballard, G., and Howell, G. (1995). “Toward construction JIT.” Proc., 1995 Lee, S., Pena-Mora, F., and Park, M. (2006). “Reliability and stability
Conf., Association of Researchers in Construction Management, buffering approach: Focusing on the issues of errors and changes in
Sheffield, UK. concurrent design and construction projects.” J. Constr. Eng. Manage.,
Borcherding, J., and Garner, D. (1981). “Work force motivation and 132(5), 452–464.
productivity on large jobs.” J. Constr. Div., 107(3), 443–453. Liberda, M., Ruwanpura, J., and Jergeas, G. (2003). “Construction produc-
Chan, E. H. W., and Au, M. C. Y. (2009). “Factors influencing building tivity improvement: A study of human, management and external
contractors’ pricing for time-related risks in tenders.” J. Construct. issues.” Proc., Construction Research Congress, ASCE, Reston, VA.
Eng. Manag., 135(3), 135–145.
Downloaded from ascelibrary.org by University of Nebraska-Lincoln on 09/17/13. Copyright ASCE. For personal use only; all rights reserved.

National Research Council. (2009). Advancing the competitiveness and

Dai, J., Goodrum, P. M., and Maloney, W. F. (2009). “Construction efficiency of the U.S. construction industry, National Academies Press,
craft workers’ perceptions of the factors affecting their productivity.” Washington, DC.
J. Constr. Eng. Manage., 135(3), 217–226. Pearl, J. (2009). Causality: Models, reasoning, and inference, Cambridge
Dept. of Labor. (2012). Industries at a glance, Construction: NAICS 23, University Press.
Bureau of Labor Statistics, Washington, DC, 〈http://www.bls.gov/ Portas, J., and AbouRizk, S. (1997). “Neural network model for estimat-
iag/tgs/iag23.htm〉 (Jun. 20 2012). ing construction productivity.” J Construct. Eng. Manag., 123(4),
Dept. of the Army. (1998). Risk management, Field manual (FM) 100-14, 399–410.
Washington, DC.
Risner, R. (2010). Auditing construction contingency, Association of
Goldratt, E. M. (1997). Critical chain, North River Press, Great Barrington,
Healthcare Internal Auditors, Wheat Ridge, CO, 37–38.
MA.
Rojas, M., and Aramvareekul, P. (2003). “Labor productivity drivers
Hopp, W. J., and Spearman, M. L. (2008). Factory physics, 3rd Ed.,
and opportunities in the construction industry.” J. Manage. Eng., 19(2),
McGraw-Hill, New York.
78–82.
Horman, M. J. (2000). Process dynamics: Buffer management in building
SAS Institute. (2012). “The cluster procedure.” SAS/STAT 9.3 user’s guide,
project operations, Ph.D. dissertation, Univ. of Melbourne, Melbourne,
〈http://support.sas.com〉 (Feb. 2, 2012).
Victoria, Australia.
Horman, M., and Kenley, R. (1998). “Process dynamics: Identifying a strat- Somnez, R., and Rowlings, J. E. (1998). “Construction Labor productivity
egy for the deployment of buffers in building projects.” Int. J. Logist. modeling with neural networks.” J Constr. Eng. Manage., 124(6),
Res. Appl., 1(3), 221–237. 498–504.
Horman, M., and Thomas, H. R. (2005). “Role of inventory buffers in Taleb, N. N. (2007). The black swan: The impact of the highly improbable,
construction labor performance.” J. Constr. Eng. Manage., 131(7), Random House, New York.
834–843. Thomas, H. R., and Sakarcan, A. S. (1994). “Forecasting labor productivity
Horman, M. J., Messner, J. I., Riley, D. R., and Pulaski, M. H. (2003). using factor model.” J Construct. Eng. Manag., 120(1), 228–239.
“Using buffers to manage production: A case study of the Pentagon Tommelein, I. D., Riley, D. R., and Howell, G. A. (1999). “Parade game:
renovation project.” Proc., 11th Annual Conf. of the International Impact of work flow variability on trade performance.” J. Constr. Eng.
Group for Lean Construction (IGLC-11), Lean Construction Institute. Manage., 125(5), 304–310.
Hornby, A. S., ed. (1974). Oxford advanced learner’s dictionary of current Wambeke, B. W. (2011). “Identifying, prioritizing, and reducing variation
English, Oxford University Press, Oxford. of construction related tasks.” Ph.D. dissertation, Dept. of Civil Engi-
Howell, G. A., Laufer, A., and Ballard, G. (1993). “Interaction between neering, North Carolina State Univ., Raleigh, NC.
sub-cycles: One key to improved methods.” J. Constr. Eng. Manage., Wambeke, B. W., Hsiang, S., and Liu, M. (2011). “Causes of variation in
119(4), 714–728. construction project task starting times and duration.” J Construct. Eng.
Howitt, D., and Cramer, D. (2005). Introduction to statistics in psychology, Manag., 137(9), 663–677.
Pearson Education, Essex, UK, 516. Wheeler, M. (2002). “The evolution and application of technical risk man-
Kimpland, S. L. (2009). 2009 Contractor productivity survey results, FMI, agement within the United States Navy.” Masters’ thesis, Naval Post
Raleigh, NC. Graduate School, Monterey, CA.

© ASCE 04013008-10 J. Constr. Eng. Manage.

J. Constr. Eng. Manage. 2013.139.