December 11, 2006

Last month Part 1 of this article was published in the Construction Update newsletter.  To view Part 1 visit www.lorman.com/newsletters.

Tricks of the Trade

The theory behind critical path method scheduling is that the network of activities is designed to model the way in which the project will be constructed.  If the network closely models the project’s plan, the predictions calculated from the schedule will be reliable.  With this objective in mind, CPM software developers have worked to improve this modeling capability. 

The earliest such innovations involved the calendar and enabled the scheduler to define the anticipated project workdays.  This was followed by the ability to define multiple calendars and assign each activity to an appropriate calendar.  Now, the schedule could consider in its calculating process the fact that concrete would not be placed on a weekend day, but the concrete would cure over the weekend; this better model resulted in a more accurate forecast date for stripping the forms.

Current scheduling software allows the scheduler to define a variety of relationship types between activities.  The relationship between two activities can be defined so that one activity starts after the completion of the other, both start together, both finish together, or so that one finishes when the other starts.  All of these relationships can be defined with a lag so that some positive or negative amount of time is factored into the relationship. 

And, the dates, durations, and float of activities can be constrained in a variety of ways.  For example, activity early and late start and finish dates can be constrained to ensure that an activity will not be forecast to start or finish earlier or later than a specified date.  An activity can also be constrained to start on a particular day.  An expected finish date can also be imposed on an activity, causing the program to calculate the activity’s remaining duration.  Mandatory dates may also be imposed, overriding the network logic leading to and from the constrained activity.  Additionally, an activity’s float can be constrained so that its late dates are forced to equal its early dates or so that it will start as late as possible.

In addition to activity constraints, the way the schedule considers progress can also be controlled by the scheduler.  For example, the schedule can be calculated such that progress made on an activity out of sequence with the flow of the network will override the schedule logic preceding it, or the logic of the network can be retained despite such out-of-sequence progress.

All of these features have served to enhance the schedule’s ability to model the plan to construct the project.  Each of these constraints supplements the logic of the schedule—and each has its place.  For example, the contract may require that certain aspects of the work be completed by a certain date or that access to a particular area may not be given until a certain date.  In such cases, a constrained finish or start may be appropriate.  Or, the contract may require limited disruption to a particular area once the work in that area is started.  In such case, zero float constraints will help model this contractual access limitation.

As additional logic, these advanced scheduling features affect how the schedule calculates the forecast dates.  In some cases, the effect can be dramatic.  As a result, each must be used knowledgeably and with caution.  However, this is good news.  This superior modeling capability provides a powerful project management tool and, when properly analyzed, a more reliable determination of critical project delays.

Analyzing Delay

There are many types of analysis techniques being used by experts to prove delays.  Many of these fail simply because they express a one-sided view of the project.  Others fail because they are not grounded in the documents used by the participants to manage the project.  A properly performed delay analysis will have certain attributes regardless of the precise methodology used. 

To begin with, the analysis must be performed objectively.  One way to achieve an objective analysis is for the analyst to focus on determining the source and magnitude of all critical project delays without regard to the party responsible.  For example, an analysis of the schedule may reveal that the late start of foundation excavation caused a critical delay to the project.  This conclusion should be independent of what caused this excavation work to start late.  Determining the party responsible for this delay should be a separate task.

This leads to the next attribute.  The analysis should account for all project delays and savings throughout the duration of the project.  The total delay to the project will be known as the difference between the planned completion date and the actual completion date.  The analyst should identify all of the critical delays and savings that total up to the actual total project delay.

A delay analysis should rely on the contemporaneous project schedules as the basis of analysis to the maximum extent possible.  By using the management tools employed by the parties during the project, the analyst is able to adopt the perspective of the project manager at the time of a particular delay and avoid applying the “wisdom” of hindsight.  Reliance on the contemporaneous project schedules helps keep the analysis objective and guards against the analyst drawing erroneous conclusions.  For example, was the steel delivered late allowing more time to construct the foundations or was the steel delivery pushed back because the foundations were late.  Analyses that stray far from the contemporaneous schedules, relying on after-the-fact creations, are usually biased and unpersuasive. 

Assessing the reliability of the CPM schedules to form the basis of an analysis of delays may be one of the most important decisions that the analyst has to make.  While it is true that there may be sufficient cause to abandon the contemporaneous schedules as an analytical tool, this decision should be made carefully and with good reason. No schedule is perfect.  Most contain or omit logic that could be characterized as errors.  But many of these are minor and have no effect on the project’s critical path.  And most errors are self-correcting, meaning that, if the schedules are properly analyzed, the actual progress of the work will establish the correct answer. 

To better understand this principal of self-correction, consider the following example.  Presume that in a sequence of critical path work involving the installation of wall studs, rough-in wiring, and drywall, the scheduler omits the activity for rough-in wiring.  The schedule presumes that drywall installation will begin at the completion of the stud installation; no time has been allotted for roughing in wire.  One might argue that the schedule is flawed because the contractor cannot really complete all of the contract work on time.  But what really happens?  When the stud installation is actually completed, the drywall work fails to start.  Instead, the contractor first roughs in the wire.  The contractor is assigned a critical project delay equal to the late start of the drywall work.  And so, the error self-corrects.  Had the contractor included rough-in wiring in the original schedule, it would have done so in a way that did not increase the contract time.  This same choice is available when the omission is discovered.  Perhaps the contractor can perform the rough-in wiring concurrent with some of the wall stud and drywall work such that no delay occurs.  Either way, the error in the schedule is self-correcting.

So when is the schedule no longer good enough?  The contemporaneous CPM schedule should only be abandoned as an analytical tool when there is evidence that the schedule is not a reliable model of the contractor’s plan or when the schedule contains such gross errors as to render its predictions useless.  Often, a schedule that has been significantly revised to mitigate a large number of days of delay will begin to contain gross errors of the type that cause the schedule to model an unexecutable plan.  In such cases, the plan being modeled by the schedule and the actual execution of the project depart, and the schedule is no longer a reliable tool to measure project delay. 

Here, a word of caution is appropriate.  The proper use of advanced scheduling features should not be mistaken as scheduling errors.  The mere existence of an additional calendar, a lag, or an activity constraint is not, in and of itself, an error.  While the use of advanced scheduling features may make the task of analyzing the schedule to identify and measure delays more challenging, understanding how these features affect the calculation of the schedule allows the analyst to preserve the contemporaneous schedule as an objective analytical tool.

William A Manginelli is the President of Trauner Consulting Services, Inc. in Philadelphia, PA.  Learn more about Trauner Consulting Services, Inc. at http://www.traunerconsulting.com. He can be reached at (215) 814-6400 or at [email protected].

Don't miss Part 3 of this article in next month's Construction Update newsletter that will be published January 22, 2007.  You can view Part 1 at www.lorman.com/newsletters.