Identification and Quantification of Concurrent Delay
A. Identification & Quantification of Concurrency
This is the most contentious technical subject in forensic schedule analysis. Because of this, it is important that both sides, if possible, agree on the theory employed in the identification and quantification of concurrency. Failing that, the analyst should be aware of the theory adopted by the adversary.
There is no consensus on many of these factors affecting the identification and quantification of concurrency. The one thing that seems to be universally accepted is that reliable identification and quantification of concurrency must be based on CPM concepts. Gross concurrency, or the method of counting concurrent delay events based purely on contemporaneous occurrence without regard to CPM principles, is not adequate basis in negating compensability.
Contractual definition is one major factor having significant impact on the determination of concurrency。Contracting parties are free to mutually agree on any method or procedure as long as those agreements do not violate public policy. Therefore the general rules, exceptions and considerations in this practice are applicable to the extent that they do not directly contradict contractual definitions and specifications.
Some contracts include in the definition of concurrent delay that it be a critical path delay. The requirement that the concurrent delay be critical, in effect, excludes other delay events with float values greater than the critical path from being evaluated for offsets against compensable delays. Absent such contract definition, non-critical delays can be used to offset compensable delay on a day-for-day basis after the expenditure of relative float against the critical path.
In addition to the contractual variable, there are at least five factors that influence the identification and quantification of concurrency:
1. Whether concurrency is determined literally or functionally.
2. Whether concurrency is determined on the cause or the effect of delay.
3. The frequency, duration and placement of the analysis interval
4. The order of delay insertion or extraction in a stepped implementation.
5. Whether the analysis is done using full hindsight or based on knowledge-at-the-time.
Literal Concurrency vs. Functional Concurrency
The difference here is whether delays have to be literally concurrent in time, as in "happening at the same time", or they need to be functionally concurrent so that only the separate network paths on which the delays reside be concurrently impacting the completion date.
Note that absolute, literal concurrency is an unachievable goal since time is infinitely divisible. It is more a function of the planning unit used by the schedule or the verification unit used in the review of the as-built data. For example, upon further examination, a pair of events that were determined to have occurred concurrently on a given day may not be literally concurrent because one occurred in the morning and the other in the afternoon.
Implicit in this difference is the conflicting views on whether float value is an attribute possessed by each individual activity (literal theory) or an attribute that each activity inherits from the network path on which it resides (functional theory). While both theories have their merits, the functional theory is more attuned to the workings of the critical path method.
Of the two, the functional theory is more liberal in identifying and quantifying concurrency. The assumption made by the functional theory practitioner is that most delays have the potential of becoming co-critical, once float on the path it resides have been consumed. In other words, delays are assumed guilty of concurrency until proven innocent by float analysis.
Whereas the practice based on the literal theory will result in far fewer identification of concurrent delays, since delays are dropped from the list of suspects if they do not share real-time concurrency. If the literal theory practice is combined with the contractual definition of concurrent delay as a critical path delay, the finding of concurrency becomes exceedingly rare.
The difference in outcome is significant. Given the same network model, the literal theory practitioner will find many more compensable delays for both parties. Because of the potential for more compensability for both parties, the resolution process tends to be more emotionally charged.
Whereas the functional theory practitioner will find many of those delays to be concurrent and hence be excusable but non-compensable for both parties. But note that the ultimate outcome may be similar, since when the compensation due for both sides under the literal theory model are combined for a net calculation, they may also cancel each other out.
The only significant difference despite the fact that the canceling effect operates under both theories is the timing of the canceling effect and its implication concerning damage calculation. Under the literal theory, an owner-delay and a contractor-delay of equal duration, occurring at different times are calculated as a period of compensable delay for the owner and a separate period of compensable delay of equal length for the contractor.
The two periods will cancel each other out in time, but not necessarily money, since more likely that not the owner's liquidated damages rate will not be equal to the contractor's extended project rate. So despite the canceling effect, there is still potential of award of compensability to one side or the other. In contrast, under the functional theory, the canceling effect is realized before calculation of damages; hence there will be no offsetting calculation for damages.
The practical effect is that the use of literal theory will benefit the owner if the liquidated damages or actual delay damages rate is greater than the delay damages rate used by the contractor. Conversely, if the contractor's rate is greater than the owner's the literal theory will benefit the contractor. In contrast, the functional theory tends to minimize compensable delay to both sides by concentrating on the detection of functionally concurrent periods and removing them from consideration of delay damages.
Cause of Delay vs. Effect of Delay
Another philosophical dichotomy that complicates the evaluation of concurrency is the difference between the proximate cause of the delay and effect of the delay.
For example a schedule activity with a planned duration of five days experiences work suspensions on the second day and the fifth day, thereby extending the duration by two days. The delay-cause is on the second and the fifth day, but the delay-effect is on the sixth and the seventh day. The differences become much larger on activities with longer planned duration that experiences extended delays. A good example would be delayed approval of a submittal that stretches for weeks and months.
The philosophical difference rests on the observation by the delay-effect adherents that there is no ‘delay' until the planned duration has been exhausted. In contract the delay-cause adherents maintain that the identification of delay should be independent of planned or allowed duration, and instead should be driven by the nature of the event.
The disadvantage of the delay-cause theory is that if there are no discrete events that cause a scheduled activity to exceed its planned duration, it would have to fall back to the delay-effect method of identifying the delay.
Conversely, in cases where the delay was a result of a series of discrete events, the delay-effect method of chronological placement of delay would often be at odds with contemporaneous documentation of such discrete events.
The difference in outcome is pronounced under the literal theory, since it affects whether or not a delay is identified as concurrent. Under the functional theory the significance to the outcome depends on whether the analyst is using a static method or a dynamic method.
Using a static method, the cause-effect dichotomy makes no difference because the entire project is one networked continuum. But using a dynamic method, it does make a difference because the chronological difference between the cause and effect may determine the analysis interval in which the delay is analyzed.
At the individual activity level, there are logical bases for the application of both theories of thinking. But at the overall project level, the delay-effect theory makes very little sense because it simplifies the entire network into one summary bar and evaluates the net effect of various delay scenarios by comparing the length of the bars. In effect, it reduces concurrency analysis to a measuring exercise requiring only a ruler or an accurate eye.
The best practice that incorporates the best features of both theories is to use the cause theory where discrete delay events exist and to use the effect theory where there are no discrete events that led to the delay. But note that in many cases the identification of discrete causes is a function of diligence in factual research, which is in turn dictated by time and budget allowed for the analysis.
Frequency, Duration and Placement of Analysis Intervals
Analysis interval refers to the individual time periods used in analyzing the
schedule under the various dynamic methods. The frequency, duration and the placement of the analysis intervals are the most significant technical factors that influence the determination of concurrency.
The significance of the analysis interval concept is also underscored by the fact that it creates the distinction in the taxonomy between the static versus the dynamic methods. The static method has just one analysis interval, namely the entire project, whereas the dynamic model segments the project into multiple analysis intervals.
Frequency & Duration
The variables of frequency and duration are related to each other and are dependent on the overall duration of the project. A thousand-day project can be segmented into ten equal analysis intervals of a hundred days each; and a seven-day project can be segmented into two analysis intervals consisting of two days and five days. While prevailing conventional wisdom states that the accuracy of the analysis is enhanced by increasing the frequency of analysis intervals, the number of intervals must be considered in relation to the duration of each of the intervals.
The caveat is applicable in evaluating any dynamic method， but would also apply when evaluating static methods. For example, a periodic implementation of the static observational method that evaluated the as-built in relation to the as-planned in daily increments may be a much better analysis than an implementation of the dynamic observational method using reconstructed updates where there are only three ‘windows', each containing several months.
Concurrency is evaluated discretely for each analysis interval. That is, at the end of each period, accounting of concurrency is closed, and a new one opened for the next period. This is especially significant when analysis proceeds under the functional theory of concurrency in cases where two functionally concurrent delay events, one owner-delay and the other a
contractor-delay, are separated into separate periods.
If those delay events were contained in one period, they would be accounted together and offset each other. When they are separated, they would each become compensable to the owner and the contractor respectively, thereby, in essence, forcing the functional theory to behave like a literal theory.
However, the distinction between the functional and the literal theories do not disappear automatically with the use of multiple analysis intervals. Two delay events separated by time within one analysis interval will still be treated differently depending on which theory is used. The distinction becomes virtually irrelevant only when the duration of the analysis interval is reduced to one day.
When multiple analysis intervals are used an additional dimension is added to the canceling effect that was discussed in the comparison of the literal theory to the functional theory. As stated above, the separation of two potential concurrent delay events into different analysis intervals causes the functional theory to behave like the literal theory.
Because the change from one period to another closes analysis for that period and mandates the identification and quantification of excusable, compensable and non-excusable delays for that period, it is only after all the analysis intervals, covering the entire duration of the project, are evaluated that reliable results can be obtained by performing a ‘grand total' calculation. In other words, the ultimate conclusion cannot be reached by selective evaluation of some, but not all, analysis intervals.
The general rule that all the intervals must be evaluated will assure the reliability of the net result. But the analyst can still influence the characterization of the delays by determining the chronological placement of the boundaries of the intervals, or the cut-off dates.
There are two main ways that the analysis intervals are placed. The first method is to adopt the update periods used during the project by using the data dates of the updates, which are usually monthly or some other regular periods dictated by reporting or payment requirement.
The other is the event-based method in which the cut-off dates are determined by key project events such as the attainment of a project milestone, occurrence of a major delay event, change in the project critical path based on progress (or lack thereof), or a major revision of the schedule. Event-based cut-off dates may not necessarily coincide with any update period.
The most distinguishing feature of the event-based placement of cut-off dates is that there is significant judgment exercised by the forensic analyst. Because the cut-off date is equivalent to the data date used for CPM calculation, it heavily influences the determination of criticality and float, and hence the identification and quantification of concurrent delays. Also, as stated above, the placement of cut-off date plays a major role in how the canceling effect operates.
Order of Insertion or Extraction in Stepped Implementation
In a stepped insertion or extraction (3.8) implementation, the order of the insertion or extraction of the delay will affect the identity of potentially concurrent delays and the quantification of such concurrency.
As a general rule, for additive modeling methods where results are obtained by the forward pass calculation, the order of insertion should be from the earliest in time to the latest in time. For subtractive modeling methods the order is reversed so that the stepped extraction starts with the latest delay event and proceeds in reverse chronological order.
There are other systems, such as inserting delays in the order that the change orders were processed, or extracting delays grouped by subcontractors responsible for the delays. In all these seemingly logical schemes if chronological order of the delay events is ignored, the resulting float calculation for each step may not yield the data necessary for reliable determination of concurrent delays.