Concept: Schedule adherence is adversely affected by tasks requiring rework (R-tasks) and by tasks that are impeded or constrained (I/C-tasks). As explained in last month’s post, R-tasks and I/C-tasks comprise distinct and separate sets. From the difference, new metrics emerge for schedule adherence.
Practice: R-tasks exclusively earn value after the ES time (ES) and as of the Actual Time (AT). Because the number of R-tasks inevitably declines, the value of R ultimately drops to zero.
In other words, the value of R depends on P. [1] As a project proceeds, P is increasingly unresponsive to improvement efforts. Regardless of management action (or inaction), P ends up at a perfect 1.0. So, R will consistently improve toward the end of a project, regardless of the steps taken.
That means R has limited usefulness for managing schedule performance: you don’t know whether improvement in R is due to management action or to the inescapable rise of P to 1.0. [2]
To avoid the problem and salvage a full life-cycle metric, R is normalized to the remaining work.
Normalization distributes the estimated amount of rework over the remainder of the project. It scales rework to the amount of work left, canceling the effect of dwindling opportunities for noncompliance.
As explained previously, normalization relates the amount of rework (R) to the baseline project budget (BAC) less the value already earned (EV). The resulting indicator is the Schedule Adherence Index (SAI):
SAI is a reliable indicator of schedule adherence throughout the life of a project. It offers a good fix on the impact of R-tasks, but what about the I/C-tasks?
Part of their story is expressed by the complement of P, i.e., 1 – P. But, because the complement covers all noncompliant tasks (i.e., both R- and I/C-tasks), it does not isolate the effect of I/C.
To isolate the effect of impediments and constraints, the approach is different from that for rework.
Recall that R is an estimate of the amount of rework. It is normalized to the remaining work on the project to set the SAI. Periodic readings of SAI are then used in a trapezoidal approximation to determine the value impact on the project.
Like R, the size of I/C tasks is a discrete quantity. Unlike R, the quantity is not units of work, and its value impact does not need to be approximated. The value of I/C can be measured directly. It is the current shortfall in value earned versus the value scheduled to be earned as of ES. In other words, it is visible where EV@ES – PV@ES is negative. [3]
Here are the details. The total amount of I/C (I/CTot) equals the value of tasks, i, where the difference between EVi@ES and PVi@ES is negative. That difference is multiplied by -1 to get the absolute value. All other differences (i.e., positive or 0) are ignored in the calculation. Expressed as an Excel-style formula, this is:
Still, the I/C-quantity is part of what is expressed by 1 – P. So, any metric based on the I/C-quantity will be affected by P’s inevitable rise to 1.0 at the end of the project. Intuitively, think of it this way: as the amount of work decreases, the opportunity for EV to lag scheduled PV decreases simply because there are fewer tasks.
The decline does not necessarily follow the same downward curve as R. Within ES, Rework originates from work done prematurely. [4] The resulting knowledge gaps are filled by assumptions that allow work to proceed. The assumptions generally turn out to be wrong, and rework follows.
Certainly, I/C occurs for the same reason: knowledge gaps. Indeed, one motivation for halting work is to wait for such gaps to be filled in, precisely to avoid rework. But, work is impeded or constrained for many other reasons: for instance, due to unplanned vacation, absence, and training. The downward path for I/C ends at 0, but it does not necessarily track the one for rework.
Because the amount of I/C can be measured directly, it avoids the necessity of a mathematical model to determine its size. Still, a mathematical model helps practically, as it is often difficult to deterermine the relevant quantities empirically, especially in large schedules. [5]
To avoid the effect of a decline in the amount of I/C (a decline that inevitably ends at 0), the amount of I/C can be normalized to the remaining work. That expresses the impact of I/C-tasks accurately across the whole project timeline.
Let’s call this new indicator the I/C Index (ICI). It is expressed by this formula:
Subsequent posts explain how the ICI is used to manage schedule adherence. The next post describes how the ICI can be used to unlock the cost of impediments and constraints. The current post mines the ICI from the hidden gems of ES theory.
Notes:
[1] More precisely, R depends on P’s complement: 1 – P. Recall the formula for R, noting particularly the term in bold: 
The first term
is the mathematical model that isolates the R-tasks from the total noncompliant EV.
[2] See Lipke (2011a), p. 11.
[3] The formula given in the texts is: EV@AT – PV@ES. But, the EV earned after ES and as of AT is not counted toward P—it is only the EV earned as of ES that is counted. Look at it this way—if value is earned as of ES, it is earned as of AT. As regards P, EV@AT and EV@ES amount to the same value.
[4] Walt notes several causes of rework (Lipke, 2011a, p 9) but focuses on out-of-sequence tasks where work is done prematurely, i.e., after ES and as
of AT.
[5] A brute-force method would be to, first, track all individual tasks across the project timeline, calculating the ES at the end of each period and, then, to assess the resulting performance task-by-task against ES and AT. Adding to the complexity, the results would have to be stored historically or re-calculated each time the analysis was done.
References:
Lipke, W. (2012). Schedule Adherence and Rework. CrossTalk, November-December.
Lipke, W. (2011b) Schedule Adherence and Rework. PM World Today, July.
Lipke, W. (2011a) Schedule Adherence and Rework. The Measurable News, Issue 1 (corrected version).
Revised 06 Aug 2020
|
