Improving Concurrency Assessment and Resolving Misconceptions About But For Delay Analysis

Scholarly Paper
Improving Concurrency Assessment and Resolving

Misconceptions about But-For Delay Analysis Technique
Moneer Bhih, S.M.ASCE 1; and Tarek Hegazy, M.ASCE 2
Downloaded from ascelibrary.org by NUST NAT'L UNIV OF SCI & TECH on 02/17/20. Copyright ASCE. For personal use only; all rights reserved.
Abstract: But-for delay analysis is a popular technique used in practice and accepted by arbitration boards and courts. However, miscon-
ceptions are common when the analysis results are interpreted from different parties’ viewpoints. In addition, the adoption of either the literal
or functional views on concurrent delays affects the results. This paper thus clarifies the misleading interpretations of but-for results and
introduces improvements and an explicit implementation procedure that matches the delay analysis requirements of professional bodies such
as the Association for the Advancement of Cost Engineering International (AACEI). To more accurately perform but-for analysis considering
all parties’ viewpoints, the paper uses Venn representation and suggests a simplified procedure to check for true concurrency. A case study
was used to show a detailed procedure for applying but-for with multiple analysis windows as a more accurate approach to assess concurrent
delays and to consider baseline updates. The applicability of the proposed improvements was then confirmed using a second practical case
study. The paper is expected to remove the existing but-for misconception and provide a procedure for more accurate and repeatable delay
analysis considering concurrent delays. DOI: 10.1061/(ASCE)LA.1943-4170.0000378. © 2020 American Society of Civil Engineers.
Author keywords: Construction; But-for; Collapsed as-built; Concurrent delay; Delay analysis; Forensic delay analysis.
Introduction Fawzy and El-adaway (2012), who reported the highly subjective
nature of the technique.
Traditional but-for analysis (also known in the literature as collapsed In the last two decades, several professional bodies have shown
as-built delay analysis) is used in the preparation and justification of their interest in forensic schedule analysis by establishing proto-
delay claims with less time and effort (Zack 1999). Although it is cols, recommended practices, and standards aimed at promoting
more reliable than several other delay analysis techniques, it is too clearer contract conditions regarding concurrent delays, and meth-
easy to manipulate (Zack 1999). Because its simplicity, however, it ods to determine compensation and time extension. There are cur-
has been accepted in courts (e.g., the case of Zurn Constructors v. rently three main professional guidance documents for schedule
Castaic Lake Water Agency), reported in Lifschitz et al. (2009). Dale delay analysis (Livengood 2017), briefly introduced as follows:
and D’Onofrio (2017) listed 16 court cases between 1979 and 2012 • Delay and Disruption Protocol, Society of Construction Law
in the United States, Canada, Australia, UK, and Hong Kong in (SCL), UK: The Society of Construction Law was the first to pub-
which the but-for analysis method was used. One of the reasons be- lish the Delay and Disruption Protocol in October 2002. In Feb-
hind this acceptance is that it uses the most realistic as-built schedule ruary 2017, the second edition of the protocol was published
in the analysis (Arditi and Pattanakitchamroon 2006; Yang and (SCL 2017). The document is intended to be a balanced document
Yin 2009; Dale and D’Onofrio 2017). The availability of enough that equally represents the interests of all parties and not a state-
progress data to construct an accurate as-built schedule including ment of the law. It introduces 22 core concepts, which are not
delay events caused by different parties is a minimum prerequisite geared toward the use of any specific method of delay analysis.
for but-for analysis (Zack 2000). One of the known drawbacks of • Forensic Schedule Analysis Recommended Practice 29R-03, As-
this technique is that it produces different results when adapting dif- sociation for the Advancement of Cost Engineering International
ferent parties’ viewpoints (owner versus contractor) as explained in (AACEI): AACEI published the latest version of its recom-
case studies by Braimah (2013) and Mbabazi et al. (2005). Because mended practice in April 2011 (AACEI 2011). The objective
of this discrepancy, either party can use it to support its argument is to “provide a unifying technical reference for the forensic ap-
(Stumpf 2000). Moreover, but-for analysis is not able to identify con- plication of critical path method (CPM) of scheduling and to re-
current delays, as clearly reported in Dale and D’Onofrio (2017) and duce the degree of subjectivity involved in the current state of the
SCL (2002). This is also emphasized in El-adaway et al. (2014) and art. The RP is an advisory document to be used with professional
judgment based on working experience and knowledge” AACEI
1
Ph.D. Candidate, Dept. of Civil and Enviromental Engineering, Univ. RP 29R-03 (AACEI 2011). AACE RP 29R-03 attempts to
of Waterloo, Waterloo, ON, Canada N2L 3G1; formerly, Lecturer, Dept. of categorize the delay analysis methods used in practice into nine
Civil Engineering, Univ. of Benghazi, Benghazi, Libya. ORCID: https:// categories called methods of implementation. It also discusses
orcid.org/0000-0002-8408-7785. Email: mbhih@uwaterloo.ca the factors affecting the selection of a particular method.
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of • Schedule Delay Analysis [ASCE 67 (ASCE 2017)], ASCE: In
Waterloo, Waterloo, ON, Canada N2L 3G1 (corresponding author). ORCID: 2017, ASCE published its new proposed standard in accordance
https://orcid.org/0000-0002-6093-0037. Email: tarek@uwaterloo.ca
with the American National Standards Institute (ANSI). The
Note. This manuscript was submitted on April 2, 2019; approved on
September 13, 2019; published online on February 11, 2020. Discussion
document introduces 35 guidelines of schedule delay analysis
period open until July 11, 2020; separate discussions must be submitted for principles that enable apportioning the responsibility for delays
individual papers. This paper is part of the Journal of Legal Affairs and in the project milestones and completion date, as well as calcu-
Dispute Resolution in Engineering and Construction, © ASCE, ISSN lating compensations or liquidated damages. This standard does
1943-4162. not emphasize any specific delay analysis method; rather, it
© ASCE 04520007-1 J. Leg. Aff. Dispute Resolut. Eng. Constr.
J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

provides general guidelines that are applicable in conjunction the standard described concurrent delay situations as “a situation
with any methodology. where two or more critical delays are occurring at the same time.”
Among the three mentioned guides, only AACEI RP 29R-03 The commentary on this definition explained that both events
provides guidelines for the delay analysis methods. The tradi- should not be starting or ending at the same time. The standard has
tional but-for analysis has a resemblance to two of those methods no clear position regarding the time interval between delay events;
[MIP 3.8 and MIP 3.9 (AACEI 2011)], which are descriptive guide- however, it could be understood that it requires both events literally
lines rather than step-by-step methods. The AACEI MIP 3.8 and occur at the same time.
MIP 3.9 methods represent guidelines for the subtractive analysis A variation of concurrent delays, called offsetting delay, was
techniques in practice such as the traditional but-for analysis in- presented by ASCE 67 (ASCE 2017). When the contractor is be-
cluding both the owner and contractor points of view. The subtrac- hind schedule, all of the remaining activities with zero or negative
tive techniques use the as-built schedule as the analysis base floats are considered critical. In this case, any owner-caused delay
and subtract delay events from it to assess the events’ effect. The may offset the contractor’s liquidated damages, even if it does not
AACEI MIP 3.8 and 3.9 methods allow subtracting delay events cause further delay to the project completion date. The debate about
altogether (global subtraction) or piecemeal in periods moving offsetting delay still exists among practitioners because it requires
backward on the project (stepped subtraction) (Dale and D’Onofrio the owner to grant the contractor a time extension without causing
2010). The difference between the two methods is that MIP 3.9 further project delay and to waive its liquidated damages (Nagata
considers multiple baseline updates. In their minimum implemen- 2018). This debate is due to the recent viewpoints on the definition
tation, both methods require identifying and quantifying concurrent of critical activities, in view of the complexities that result from us-
delays and respecting event chronology in the analysis. ing the advanced features of scheduling software such as schedule
Although the three industry guides have defined concurrent constraints and multiple resource calendars. Some practitioners and
delays, those definitions reflect the differing viewpoints that origi- industry standards adopt the longest path or critical path to identify
nated from the inconsistent application of concurrent delays in critical activities, while others define critical activities as those hav-
practice (Livengood 2017). AACEI RP 29R-03 (AACEI 2011) dis- ing zero or negative floats. AACEI RP 29R-03 (AACEI 2011)
tinguishes between two types of concurrency, literal versus func- adopts the more typical definition, while ASCE 67 (ASCE 2017)
tional concurrency; the discrepancy between the two definitions adopts the latter definition and uses it to justify offsetting delay.
has shed more heat than light in delay analysis. Under literal con- In this paper, literal concurrency is adopted assuming that the
currency, delay events need to be literally simultaneous in time, work progress is recorded accurately, and thus the as-built schedule
while under functional concurrency delay events can occur in dif- exactly describes the actual progress. Adopting literal theory gives
ferent but close time periods. AACEI RP 29R-03 defines concur- the most accurate and equitable analysis results. This assumption
rent delays as “two or more delays that take place or overlap during was made to present the ideas in a clear and consistent manner;
the same period, either of which occurring alone would have af- however, the introduced concepts can be extended easily to work
fected the ultimate completion date.” As such, AACEI RP 29R-03 under the functional theory.
accepts both the literal and functional concurrency definitions,
reflecting the American practice of concurrent delays.
SCL (2017) defines true concurrent delays as “the occurrence of Research Methodology
two or more delay events at the same time, one an Employer Risk
Event, the other a contractor Risk Event, and the effects of which The current state of the traditional but-for analysis was compared
are felt at the same time.” It requires that both events be critical and with the similar AACEI methods (MIP 3.8 and MIP 3.9) as high-
affect the critical path of the project. Thus, SCL (2017) adapts the lighted in Table 1. Based on this comparison, improvements to the
literal concurrency viewpoint and calls it “true concurrency.” It de- but-for analysis are proposed (last column of Table 1) to match
fines concurrent effects as “two or more delay events arise at differ- AACEI guidelines. The other two industry guidelines, SCL (2017)
ent times, but the effects of them are felt at the same time.” From and ASCE 67 (ASCE 2017), have no specific analysis methods; they
the protocol viewpoint, the concurrent effects are not concurrent introduce guidelines for schedule delay analysis to be applied with
delays. Fig. 1 illustrates the differences between true concurrency any schedule analysis technique. As shown, the key objective of
[Fig. 1(a)] and concurrent effects [Fig. 1(b)]. Under functional this paper is to clarify the misconceptions associated with the use
theory, both progress scenarios in Figs. 1(a and b) could be treated of the traditional but-for method, improve its concurrency assessment,
as concurrent delays if the events in Fig. 1(b) are considered close and translate the introduced enhancements into a clear step-by-step
enough in time. analysis methodology that matches the general practice guides.
ASCE 67 (ASCE 2017) defines concurrent delays as “delay to To achieve the targeted improvements to the but-for analysis,
the project critical path caused concurrently by multiple events not this research introduces two main enhancements to the technique:
exclusively within the control of one party.” In another definition, correcting the existing misinterpretations of results and improving
Fig. 1. Concurrency examples: (a) true concurrency; and (b) only concurrent effects (not true concurrency).

Table 1. Comparison of the traditional and related but-for methods
Feature Traditional but-for AACEI RP 29R-03 MIP 3.8 and MIP 3.9 Targeted improvements
Clear steps to apply the method No Descriptive methods Detailed method
Same results regardless of viewpoint No Descriptive guidlines Yes
Can use multiple analysis windows No Yes Yes
Concurrency assessment No Descriptive guidlines Detailed
Respect event chronology No Partiallya Detailed
Multiple baseline updates No Descriptive guidlines Detailed
Analysis of three-party delays No Descriptive guidlines Detailed
a
From one window to another but not within each window.
its concurrency assessment. The research methodology involves Owner’s Point of View
four main steps: (1) use a small case study to describe the existing
Removing the owner’s events from the as-built schedule produces
misconception and introduce a proper interpretation of results; the schedule in Fig. 3, with 10-day duration (without owner’s
(2) use a modified case study to illustrate the inability of the tech- events). Thus, the as-built schedule was reduced by 1 day after re-
nique to distinguish true concurrency and propose improvements to moving the owner’s events, so in the owner’s point of view, the
resolve this shortcoming and to accommodate cases with baseline owner is responsible for a 1-day project delay, and the remaining
updates; (3) use a validation case study to highlight the appli- balance of the project delay (2 days) is due to contractor events. The
cability of the enhanced method; and (4) extend the step-by-step following steps summarize the owner’s point of view in but-for
analysis methodology to the general practical case of three-party analysis:
delays. 1. Total project delay = 3 days;
2. As-built duration = 11 days;
3. As-built without owner (O) events = 10 days;
Resolving the Misinterpretation of But-For Results 4. Resulting owner responsibility = Step 2–Step 3 = 1 day;
5. Contractor (C) responsibility = Step 1–Step 4 = 2 days; and
A small case study of four activities (Fig. 2) was used to explain 6. Final result: O ¼ 1 day, C ¼ 2 days.
the common misinterpretation of but-for results. The case study
was adopted from the literature (Mbabazi et al. 2005) with minor
changes to suit this paper. Both the as-planned and as-built Contractor’s Point of View
schedules are shown with work delay events by the owner and Removing the contractor’s events from the as-built schedule
the contractor shown on the as-built schedule. The project produces the schedule in Fig. 4, with 11-day duration. Thus, the as-
as-planned duration was 8 days, while the actual completion ex- built schedule did not collapse after removing the contractor’s
hibited a 3-day project delay, making the project as-built duration events, so from the contractor’s point of view, only the owner is
11 days. responsible for all 3 days of project delay. The following steps sum-
marize the contractor’s point of view in but-for analysis, and Table 2
gives the results of the contradicting points of view:
1. Total project delay = 3 days;
Fig. 3. As-built schedule (Scenario 1) without owner events.
Fig. 2. Case study schedules: (a) as-planned schedule; and (b) as-built
schedule (Scenario 1). Fig. 4. As-built schedule (Scenario 1) without contractor events.

Table 2. But-for results of Scenario 1 under traditional misinterpretation determined to be truly concurrent delays because the two delay
Responsibility (days) events occurred at the same time (simultaneous events).
Point of view Owner Contractor Both concurrently
Owner 1 2 0 Improving Concurrency Assessment in But-For
Contractor 3 0 0 Analysis
By using the correct interpretation of results, as done in the previous
2. As-built duration = 11 days; section, the technique still lacks the ability to distinguish between
3. As-built without C events = 11 days; true concurrency and the concurrent effects. To demonstrate this
4. Resulting C responsibility = Step 2–Step 3 = 0 days; issue, the as-built schedule of the case study has been slightly modi-
5. O responsibility = Step 1–Step 4 = 3 days; and
fied in Scenario 2, as shown in Fig. 6. The difference in this situation

6. Final result: O ¼ 3 days, C ¼ 0 days. is that the owner and contractor events are not occurring simulta-
neously; thus, true concurrency did not exist. For the reasons ex-
Correct But-For Result Interpretation plained previously, literal concurrency was adopted in the analysis.
Applying the correct but-for analysis to this new scenario is as
To easily explain the common misinterpretation of results, the follows:
Venn representation proposed by Mbabazi et al. (2005) is utilized. 1. Total project delay ðO ∪ CÞ ¼ 3 days;
Fig. 5(a) shows the graphic representation of total project delay 2. As-built duration = 11 days;
(3 days) caused by only the owner (OO), only the contractor (OC), 3. As-built without O events = 10 days;
and both concurrently (O ∩ C). Fig. 5(b) shows the case of remov- 4. OO responsibility = Step 2–Step 3 = 1 day;
ing O events as was done in Fig. 3, which results in a responsibility 5. As-built without C events = 11 days;
of 1 day assigned to OO [indicated in Fig. 5(b)]. The misinterpre- 6. OC responsibility = Step 2–Step 5 = 0 days;
tation is clear that the owner assumes the rest of the delay is due to 7. O ∩ C responsibility = Step 1–Step 4–Step 6 = 2 days; and
only the contractor OC, while it is in fact a combination of OC and 8. Final result: OO ¼ 1 day, OC ¼ 0 days, O ∩ C ¼ 2 days.
O ∩ C. Thus, in the common interpretation of but-for results, each These results are identical to those of Scenario 1 (OO ¼ 1 day,
party ignores its contribution to concurrent delays (O ∩ C). O ∩ C ¼ 2 days), despite the difference in the timing of each
Similarly, removing the C events, as was done in Fig. 4, results party’s events. If the concept of the offsetting delay was not con-
in an OC responsibility of 0, as indicated in Fig. 5(c). Because both sidered in the analysis, the 2 days of O ∩ C concurrent delay result
parties ignore the O ∩ C component, it is possible to calculate is clearly wrong because the as-built schedule in this scenario shows
it easily from the results (i.e., total Venn area of O ∪ C ¼ OO þ that the two parties’ events are not simultaneous. As such, the but-
OC þ O ∩ C). In the present case, 3 ¼ 1 þ 0 þ O ∩ C, thus for method in this case identified a concurrent effect. Such a re-
O ∩ C ¼ 2 days, as indicated in Fig. 5(d). Based on this discussion, sult was obtained because of the common single-window analysis,
the following step-by-step calculations represent the correct and which makes the but-for technique unable to respect the chronologi-
repeatable but-for analysis when owner and contractor events exist cal order of events. This is a common drawback of but-for analysis
on the schedule: (Dale and D’Onofrio 2017). To determine the correct results con-
1. Total project delay ðO ∪ CÞ ¼ 3 days; sidering the chronological order of events, let us first look at the as-
2. As-built duration = 11 days; planned critical paths (ABD and ACD) [Fig. 2(a)]. The contractor
3. As-built without O events = 10 days; events on Days 4 and 5 caused 2 days of delay to the project com-
4. OO responsibility = Step 2–Step 3 = 1 day; pletion date (Days 9 and 10), and this delay has made ACD the
5. As-built without C events = 11 days; longest path and created 2 days of free float for Activity B before
6. OC responsibility = Step 2–Step 5 = 0 days; it may hit its successor (Activity D); however, because the contrac-
7. O ∩ C responsibility = Step 1–Step 4–Step 6 = 2 days; and tor delay is nonexcusable by definition, no time extension will be
8. Final result: OO ¼ 1 day, OC ¼ 0 days, O ∩ C ¼ 2 days. granted for the contractor and Activity B will still have a zero total
By visually examining the as-built schedule of Fig. 2(b), float. The analysis now considers the subsequent 3 days of owner
the concurrent part of the project delays (O ∩ C ¼ 2 days) is events on Activity B, which will first consume the 2 days free float
created earlier by the contractor event before it hits Activity D and
causes a further 1-day project delay. Based on this analysis, the
responsibility is OO ¼ 1 day, OC ¼ 2 days, and O ∩ C ¼ 0 days.
Considering the offsetting delay according to the viewpoint
of ASCE 67 (ASCE 2017), because contractor delays are nonex-
cusable, by definition, no time extension will be granted, and
Activity B will remain critical with a total float value of 0, although
it has 2 days of free float. This viewpoint will produce different
Fig. 5. Venn representation of project delays: (a) Venn representa-

tion; (b) removing owner events; (c) removing contractor events; and
(d) calculating concurrent delay. Fig. 6. As-built schedule (Scenario 2).

responsibility of OO ¼ 1 day, OC ¼ 0 days, and O ∩ C ¼ 2 days
because the owner-caused delay will first offset the earlier 2 days of
contractor-caused delay and then will cause a further 1-day project
delay that is the owner’s responsibility. Although this is a con-
siderable viewpoint, offsetting delay was not considered in this
paper because AACEI RP 29R-03 (AACEI 2011), and in particular,
MIP 3.8 and 3.9, were used as the benchmark to improve the tradi-
tional but-for analysis. Thus the results without considering offset-
ting delays are selected in this case.
Improved Concurrency Assessment Using

Multiple-Window But-For Analysis
To overcome the previously mentioned drawback, an additional

step of testing for concurrency needs to be performed on the analy-
sis results. In this step, the timing of critical events and the critical
paths were investigated to ensure that event simultaneity exists;
otherwise, the analysis results should be modified to indicate non-
concurrent delays (i.e., contractor delay in this case). True concur-
rency exists if the delay events happened simultaneously and its
effects (project delays) arose concurrently. Thus, if the analysis
results show concurrent delays, the events causing delay should be
investigated to ensure that they are simultaneous, and accordingly
correct the analysis results if necessary. This concurrency test aims
to differentiate between truly concurrent delays and concurrent
effects. Because the concurrency test is not a simple task for large
projects, a more accurate delay analysis method becomes necessary
[e.g., the daily windows method of Hegazy and Zhang (2005)].
However, it is possible to improve the but-for concurrency assess-
ment by applying the correct analysis with multiple windows to in-
crease the analysis resolution. The window size must be selected so
that both party’s events are on separate windows, unless the events
are simultaneous. For example, in Scenario 2, using two windows of
analysis (Days 1 to 5 and Days 6 to 8), as shown in Fig. 7, will ensure Fig. 7. Applying but-for analysis with multiple windows (Scenario 2):
that the delay events of the two parties are on separate windows. (a) Window 1 contemporaneous schedule (Scenario 2); (b) Window 1
Updating the as-planned schedule up to Day 5 produces the contemporaneous schedule (Scenario 2) without contractor events;
contemporaneous schedule in Fig. 7(a) with a duration of 10 days. (c) Window 2 contemporaneous schedule (Scenario 2); and (d) Window
Applying the correct but-for analysis to this window is as follows: 2 contemporaneous schedule (Scenario 2) without owner events.
1. Total project delay (O ∪ C) of this window = 2 days;
2. As-built duration [Fig. 7(a)] of this window = 10 days;
3. As-built without O events of this window = 10 days;
4. OO responsibility = Step 2–Step 3 = 0 days; Table 3. Delay analysis results of Scenario 2
5. As-built without C events [Fig. 7(b)] = 8 days; Responsibility (days)
6. OC responsibility = Step 2–Step 5 = 2 days;
7. O ∩ C responsibility = Step 1–Step 4–Step 6 = 0 days; and Delay analysis technique Owner Contractor Concurrent
8. Final result: OO ¼ 0 days, OC ¼ 2 days, O ∩ C ¼ 0 days. But-for, single window, 1 0 2
Afterward, updating the as-planned schedule up to Day 8 pro- before concurrency correction
duces the contemporaneous schedule in Fig. 7(c) with a duration of But-for, single window, 1 2 0
11 days. Applying the correct but-for analysis to this window is as after concurrency correction
But-for, two windows 1 2 0
follows:
1. Total project delay beyond previous window = 1 day;
2. As-built duration [Fig. 7(c)] of this window = 11 days;
3. As-built without O events [Fig. 7(d)] = 10 days; But-For Analysis Considering Multiple Baseline
4. OO responsibility = Step 2–Step 3 = 1 day; Updates
5. As-built without C events = 11 days;
6. OC responsibility = Step 2–Step 5 = 0 days; The as-planned schedule represents the contractor’s best plan for
7. O ∩ C responsibility = Step 1–Step 4–Step 6 = 0 days; and the work execution based on past experience and the informa-
8. Final result: OO ¼ 1 day, OC ¼ 0 days, O ∩ C ¼ 0 days. tion available in the planning stage (Menesi 2007). However, it is
The final analysis results are then determined by summing all possible that the contractor will revise the as-planned schedule
windows’ results. Accordingly, the final analysis results are OO ¼ (including changes in logic) during construction to recover un-
1 day, OC ¼ 2 days, and O ∩ C ¼ 0 days. Table 3 summarizes expected project delays or to accommodate additional work or
the Scenario 2 results under the corrected but-for interpretation for even to expedite project execution based on the owner’s request.
comparison. During the project execution, when project parties agree on a new

as-planned schedule to replace the old one, it becomes the baseline The contemporaneous schedule up to Day 8 is similar to the
for measuring work progress after the update, while the earlier part schedule of Fig. 7(c) with a duration of 11 days. Applying the
of project work is to be measured against the original baseline. correct but-for analysis to this window is as follows:
In such cases, delay analysis must consider the baseline updates if 1. Total project delay beyond previous window = 1 day;
the project has two or more baselines. 2. As-built duration [Fig. 7(c)] of this window = 11 days;
Multiple-window but-for analysis can be extended to handle 3. As-built without O events (Fig. 8) (baseline update) = 8 days;
cases with multiple baseline updates. To account for such cases, in 4. OO responsibility = Step 2–Step 3 = 3 days;
addition to the analysis windows selected previously to respect 5. As-built without C events = 11 days;
event chronology and account for the adopted concurrency theory, 6. OC responsibility = Step 2–Step 5 = 0 days;
a new analysis window is to be added every time the baseline is 7. O ∩ C responsibility = Step 1–Step 4–Step 6 = 0 days; and
updated. During the analysis, if the current analysis window cor- 8. Final result: OO ¼ 3 day, OC ¼ 0 days, O ∩ C ¼ 0 days.
responds with a baseline update, then the responsibility of the delay The final analysis results are then determined by summing all
or acceleration due to this update should be assigned. First, the windows’ results including the results accumulated in the base-
contemporaneous schedule (updated schedule) duration before line update calculations. Accordingly, the final analysis results are
the baseline update (CSd ) and the new baseline update duration OO ¼ 3 day, OC ¼ 0 days, and O ∩ C ¼ 0 days.
(Bd ) should be calculated. If CSd < Bd , then the situation is accel-
eration; accelerations are assigned as negative delay (Zhang and
Hegazy 2005); If CSd > Bd , then the situation is a delay. Delays Second Case Study
and accelerations should be apportioned based on causations to the
owner and/or the contractor. To highlight the correct but-for analysis and the improvements in
To explain multiple baseline analysis procedure using but-for concurrency assessment, a larger practical case study obtained from
analysis, the case study (Scenario 2) was slightly altered in Sce- the literature (Stumpf 2000) was analyzed. The case study involves
nario 3 by adding a second baseline at the end of Day 5. After the 12 activities for the construction of a house and a garage. The as-
contractor caused 2 days of nonexcusable delay on Days 4 and 5, planned and as-built schedules (Fig. 9) show the activities’ dura-
the owner did not accept the new project duration of 10 days and tions and relationships, in addition to the delay events of the owner
asked the contractor to accelerate the work and finish it within the
and the contractor. The project as-planned duration was originally
original duration (8 days). The new baseline (Fig. 8) was approved
16 weeks, with the as-built duration being 24 weeks (a total project
at the end of Day 5 by introducing a negative lag of 2 days to the
delay of 8 weeks). In this case study, the results of the traditional
finish-to-start relationship between Activity C and Activity D as a
corrective action by the contractor. The as-built schedule remains but-for under different perspectives are as follows:
as is for Scenario 2 including the logic change introduced in the • But-for with owner’s perspective: OO ¼ 3 weeks, OC ¼ 5 weeks
second updated baseline so that the project was completed 3 days (Fig. 10); and
behind schedule. • But-for with contractor’s perspective: OO ¼ 6 weeks, OC ¼
The correct but-for analysis considering the multiple baseline 2 weeks (Fig. 11).
The correct but-for analysis with a single window was applied to
updates was applied to this scenario. The contemporaneous sched-
ule up to Day 5 before the baseline update is the same as Scenario 2 this case study in the following steps:
given in Fig. 7(a) with a duration of 10 days. Applying the correct 1. Total project delay (O ∪ C) = 8 weeks;
but-for analysis to this window is as follows: 2. As-built duration = 24 weeks;
1. Total project delay (O ∪ C) of this window = 2 days; 3. As-built without O events (Fig. 10) = 21 weeks;
2. As-built duration [Fig. 7(a)] of this window = 10 days; 4. OO responsibility = Step 2–Step 3 = 3 weeks;
3. As-built without O events of this window = 10 days; 5. As-built without C events (Fig. 11) = 22 weeks;
4. OO responsibility = Step 2–Step 3 = 0 days; 6. OC responsibility = Step 2–Step 5 = 2 weeks;
5. As-built without C events [Fig. 7(b)] = 8 days; 7. O ∩ C responsibility = Step 1–Step 4–Step 6 = 3 weeks; and
6. OC responsibility = Step 2–Step 5 = 2 days; 8. Final result: OO ¼ 3 weeks, OC ¼ 2 weeks, O ∩ C ¼ 3 weeks.
7. O ∩ C responsibility = Step 1–Step 4–Step 6 = 0 days; and Accordingly, the overall result of the proposed correct but-for
8. Final result: OO ¼ 0 days, OC ¼ 2 days, O ∩ C ¼ 0 days. method is OO ¼ 3 weeks, OC ¼ 2 weeks, and O ∩ C ¼ 3 weeks.
At the end of this window, the contractor updated the baseline To improve the but-for results, the multiple-window but-for analy-
by introducing logic changes to accelerate the work and maintain sis was applied by first looking at the as-built schedule of Fig. 9(b)
the same project duration. This acceleration of 2 days is then attrib- and identifying seven distinct windows as follows to satisfy the
uted to the contractor, and thus a responsibility of OC ¼ −2 days is window-size rule described previously so that different parties’
accumulated. events are on separate windows, unless the events are simultaneous:
1. Week 1 only: 1 week of noncritical owner and contractor
simultaneous events. Responsibility (weeks): OO ¼ 0, OC ¼ 0,
O ∩ C ¼ 0.
2. Weeks 2–4: 3 weeks of critical owner event simultaneous
with noncritical events (owner and contractor). Responsibility
(weeks): OO ¼ 3, OC ¼ 0, O ∩ C ¼ 0.
3. Weeks 5–10: 5 weeks of noncritical owner and contractor si-
multaneous events. Responsibility (weeks): OO ¼ 0, OC ¼ 0,
O ∩ C ¼ 0.
4. Weeks 11–13: 2 weeks of critical owner and contractor events
and 1 week of critical contractor event, simultaneous with
noncritical owner and contractor events. Reasonability (weeks):
Fig. 8. Updated baseline schedule (Scenario 3).
OO ¼ 0, OC ¼ 1, O ∩ C ¼ 2.

Fig. 9. Second case study schedules: (a) as-planned schedule; and (b) as-built schedule.
Fig. 10. As-built schedule without owner’s events (second case study).
5. Weeks 14–15: 1 week of noncritical owner event and another It can be seen from Table 4 that the proposed correct but-for
week of noncritical simultaneous owner and contractor events. analysis combines the two viewpoints of the owner and the contrac-
Reasonability (weeks): OO ¼ 0, OC ¼ 0, O ∩ C ¼ 0. tor to produce repeatable results. Applying multiple-window but-for
6. Weeks 16–21: 2 weeks of noncritical owner events followed by analysis increased the analysis resolution and arrived at more accu-
4 weeks of owner events on another activity, of which only rate results and represents the best that but-for can do. The results
2 weeks of events are critical (Weeks 20 and 21). Responsibility obtained, in this case, using the proposed multiple-window but-for
(weeks): OO ¼ 2, OC ¼ 0, O ∩ C ¼ 0. analysis are identical to those obtained by Hegazy and Zhang (2005)
7. Weeks 22–24: 2 weeks of noncritical contractor events. Respon- using the more advanced daily windows analysis. The analysis
sibility (weeks): OO ¼ 0, OC ¼ 0, O ∩ C ¼ 0. windows were selected, as required by the proposed procedures,
Summing the results of these windows represents the final to separate different parties’ events; this made the multiple-window
analysis results, as given in Table 4, which summarizes the results but-for analysis able to capture the critical path fluctuations and
of the different methods. arrive at the correct results.

Fig. 11. As-built schedule without contractor’s events (second case study).
Table 4. Delay analysis results of the second case study

Responsibility (weeks)
But-for method Condition Owner Contractor Concurrent
Traditional Owner perspective 3 5 0
Contractor perspective 6 2 0
Correct Single Window 3 2 3
Correct Multiple Windows
Week 1 0 0 0
Weeks 2–4 3 0 0
Weeks 5–10 0 0 0
Weeks 11–13 0 1 2 Fig. 12. Venn representation of project delays: case of three project
Weeks 14–15 0 0 0 parties.
Weeks 16–21 2 0 0
Weeks 22–24 0 0 0
Total 5 1 2
Note: Bold numbers represent the final results of the different methods third-party delays and all combinations of concurrent delays
compared; light or non-bold numbers are the sub results of analysis (ON, O ∩ C, C ∩ N, O ∩ N, and O ∩ C ∩ N) are excusable–not
cycles to be aggregated to arrive to the final results for the methods that compensable delays, while the OO delay is an excusable–
have more than one analysis cycle. compensable delay, and the OC delay is a nonexcusable delay.
Based on the preceding, the step-by-step calculations of the
correct but-for analysis that incorporate three-party events are as
Table 5. Delay responsibility and delay type correlation matrix follows:
Delay responsibility Delay type 1. Total project delay = w;
2. As-built duration = x;
OO Excusable–compensable 3. As-built duration without O events = y;
OC Nonexcusable
4. As-built duration without C events = z; and
ON Excusable–not compensable
5. Responsibility:
Concurrent delays
O∩C Excusable–not compensable Excusable-compensable delays ðECÞ ¼ x − y ¼ OO
O∩N Excusable–not compensable
N∩C Excusable–not compensable Nonexcusable delays ðNEÞ ¼ x − z ¼ OC
O∩C∩N Excusable–not compensable Excusable-not compensable delays ðENCÞ ¼ w − OO − OC
¼ ðON þ O ∩ C þ C ∩ N þ O ∩ N þ O ∩ C ∩ NÞ
If a more detailed analysis is needed for different compensation
Case of Three-Party Delays rules that require calculating the project delays for the seven project
delay combinations (OO, OC, ON, O ∩ C, C ∩ N, O ∩ N, and
To extend the ideas presented in this paper to the general practical O ∩ C ∩ N) separately, the modified but-for method (MBF) pro-
case where three project parties are involved (owner O, contractor posed by Mbabazi et al. (2005) can be used. MBF utilizes sets theo-
C, and third party N), the seven possible combinations of project rem to calculate the seven possible combinations of project delays.
delay responsibilities [OO, OC, only third party (ON), O ∩ C, With this ability to handle the three-party delay, the targeted im-
C ∩ N, O ∩ N, and O ∩ C ∩ N] were correlated with the three provements shown in Table 1 have been achieved; this includes
common types of project delays (excusable–compensable, nonex- correcting the misconceptions involved in the original technique,
cusable, and excusable–not compensable) as shown in Table 5. This improving its concurrency assessment, and interpreting these im-
correlation matrix was established based on the AACEI RP 29R-03 provements into a detailed step-by-step methodology, as shown
(AACEI 2011) and ASCE 67 (ASCE 2017) requirements. The in Fig. 13. But-for analysis with multiple windows respects the
Venn representation of project delays of Fig. 12 shows that event chronology only between the windows but can violate the

correct but-for analysis with multiple properly selected windows.
Alternatively, for larger projects that cannot be analyzed manually,
concurrency analysis is more accurate when using a more rigorous
analysis method than but-for. The developments in this paper
provide explicit and structured details to correctly implement the
but-for analysis in a manner that matches the general delay analysis
guidelines of international organizations such as AACEI.
Data Availability Statement

All data generated or analyzed during the study are included in the
published paper.
References
Works Cited
AACEI (Association for the Advancement of Cost Engineering
International). 2011. Forensic schedule analysis. AACEI RP 29R-
03. Morgantown, WV: AACEI.
Arditi, D., and T. Pattanakitchamroon. 2006. “Selecting a delay analysis
method in resolving construction claims.” Int. J. Project Manage.
24 (2): 145–155. https://doi.org/10.1016/j.ijproman.2005.08.005.
ASCE. 2017. Schedule delay analysis. ASCE 67. Reston, VA: ASCE.
Braimah, N. 2013. “Construction delay analysis techniques—A review of
application issues and improvement needs.” Buildings 3 (3): 506–531.
Fig. 13. Improved but-for analysis methodology. https://doi.org/10.3390/buildings3030506.
Dale, W. S., and R. M. D’Onofrio. 2010. “Reconciling concurrency in
schedule delay and constructive acceleration.” Public Contract Law
J. 39 (Jan): 161–230.
chronology within each window. Window size, therefore, has an Dale, W. S., and R. M. D’Onofrio. 2017. Construction schedule delays.
Toronto: Thomson Reuters.
impact on the analysis results; the smaller the window size, the
El-adaway, I. H., S. A. Fawzy, R. Bingham, P. Clark, and T. Tidwell. 2014.
more accurate the results (i.e., higher analysis resolution). How- “Different delay analysis techniques applied to the American Institute
ever, this is not always true; if the functional theory is adopted, of Architects A201-2007 standard form of contract.” J. Leg. Aff. Dispute
selecting too small a window size may separate events close in time Resolut. Eng. Constr. 6 (3): 02514001. https://doi.org/10.1061/(ASCE)
in different windows and disqualify them for concurrency. Analysis LA.1943-4170.0000145.
window size should be selected carefully to enable the technique to Fawzy, S. A., and I. H. El-adaway. 2012. “Contract administration guide-
distinguish between true concurrency and concurrent effects, but lines for effectively and efficiently applying different delay analysis
also to account for the adopted concurrency theory. Because the techniques under World Bank–funded projects.” J. Leg. Aff. Dispute
literal theory was adopted in this paper, the window size should Resolut. Eng. Constr. 5 (Feb): 120915032631000. https://doi.org
be selected to separate both parties’ events in different windows. /10.1061/(ASCE)LA.1943-4170.0000104.
Hegazy, T., and K. Zhang. 2005. “Daily windows delay analysis.” J. Constr.
If the functional theory is adopted in the analysis, analysis windows
Eng. Manage. 131 (5): 505–512. https://doi.org/10.1061/(ASCE)0733
should be arranged to qualify critical events close in time for con- -9364(2005)131:5(505).
currency. Furthermore, to consider multiple baselines in the analy- Lifschitz, B. J., E. M. Barba, A. M. Lockshin, and R. R. Compa. 2009.
sis, a new analysis window is to be added every time the baseline is “A critical review of the AACEI recommended practice for forensic
updated. As mentioned previously, in large projects, with many schedule analysis.” Constr. Law. 29 (4): 1–10.
events and more complex interactions, a concurrency test is not Livengood, J. 2017. “Knowns and unknowns of concurrent delay.” J. Leg.
a simple task and identifying the proper window size that could Aff. Dispute Resolut. Eng. Constr. 9 (3): 1–11. https://doi.org/10.1061
arrive at the correct results is not clear. In such situations, a more /(ASCE)LA.1943-4170.0000224.
accurate delay analysis method becomes necessary. Mbabazi, A., T. Hegazy, and F. Saccomanno. 2005. “Modified but-
for method for delay analysis.” J. Constr. Eng. Manage. 131 (10):
1142–1144. https://doi.org/10.1061/(ASCE)0733-9364(2005)131:10
Concluding Remarks (1142).
Menesi, W. 2007. Construction delay analysis under multiple baseline
The delay analysis results of the but-for method have traditionally updates. Waterloo, ON, Canada: Univ. of Waterloo.
been misinterpreted. Correct application of but-for analysis has Nagata, M. 2018. “Criticism of the ASCE schedule delay analysis
been presented in this paper using Venn representation to identify offsetting delay concept.” In Proc., AACE Int. 2018 Conf. & Expo,
Claims and Dispute Resolution, 1–20. Morgantown, WV: AACE
concurrent delays. However, even with this correct analysis, using
International.
but-for with a single analysis window produces results that do not SCL (Society of Construction Law). 2002. Delay and disruption protocol.
differentiate between truly concurrent delays and concurrent ef- Wantage, UK: SCL.
fects, thus requiring adjustments to the results. To improve the SCL (Society of Construction Law). 2017. Delay and disruption protocol.
but-for analysis, the paper suggested a process of investigating Wantage, UK: SCL.
the as-built schedule to examine the true concurrency and apply Stumpf, G. R. 2000. “Schedule delay analysis.” Cost Eng. 42 (7): 32.

Yang, J.-B., and P.-C. Yin. 2009. “Isolated collapsed but-for delay analysis Zack, J. 2000. “Pacing delays-the practical effect.” Cost Eng. 42 (7): 23–28.
methodology.” J. Constr. Eng. Manage. 135 (7): 570–578. https://doi Zhang, K., and T. Hegazy. 2005. “Apportioning concurrent delays and
.org/10.1061/(ASCE)CO.1943-7862.0000016. accelerations using daily windows.” In Proc., Construction Research
Zack, J. 1999. “But-for schedules—Analysis and defense.” AACE Int. Congress 2005: Broadening Perspectives–Proceedings of the Congress,
Trans. 43 (8): 13–17. 787–796. Reston, VA: ASCE.

Improving Concurrency Assessment and Resolving Misconceptions About But For Delay Analysis

Uploaded by

Improving Concurrency Assessment and Resolving Misconceptions About But For Delay Analysis

Uploaded by

Scholarly Paper

Improving Concurrency Assessment and Resolving

© ASCE 04520007-1 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

© ASCE 04520007-2 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

Fig. 3. As-built schedule (Scenario 1) without owner events.

© ASCE 04520007-3 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

fied in Scenario 2, as shown in Fig. 6. The difference in this situation

Fig. 5. Venn representation of project delays: (a) Venn representa-

© ASCE 04520007-4 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

Improved Concurrency Assessment Using

To overcome the previously mentioned drawback, an additional

© ASCE 04520007-5 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

© ASCE 04520007-6 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

© ASCE 04520007-7 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

Table 4. Delay analysis results of the second case study

© ASCE 04520007-8 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

Data Availability Statement

© ASCE 04520007-9 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

© ASCE 04520007-10 J. Leg. Aff. Dispute Resolut. Eng. Constr.

J. Leg. Aff. Dispute Resolut. Eng. Constr., 2020, 12(2): 04520007

You might also like