EDM: Earned Duration Management, a new approach to schedule performance management and measurement

Homayoun Khamooshi

^, , Hamed Golafshani^1,

Show more

http://dx.doi.org/10.1016/j.ijproman.2013.11.002

Abstract

The concept of schedule monitoring and control as one of the most important functions of project and program management has not been fully exploited. One possible explanation could be the dominance of the Earned Value Management System (EVMS, also known as EVM). EVM was originally developed as a cost management and control tool which was extended to track the schedule as well. EVM and its derivatives (e.g. Earned Schedule) use cost as a proxy to measure schedule performance to control the duration of the project. While there is a correlation between schedule, cost, quality, and scope of a project, using cost to control duration has proven to be misleading. In contrast to Earned Value and Earned Schedule, the authors have developed the Earned Duration Management (EDM) in which they have decoupled schedule and cost performance measures and developed a number of indices to measure progress and performance of schedule and cost, as well as the efficacy and efficiency of the plan at any level of the project. These new indices are easy to understand, have wider applications, and can be used by contractors, clients and the scheduling offices to assess and measure schedule performance. The newly developed duration performance measures are all schedule-based and can be used for forecasting the finish date of the project.

Keywords

• Earned Value; EVM; Earned Schedule; Earned Duration; Earned Duration Management; EDM; Schedule management; Monitoring and control; Efficiency; Performance measurement; Forecasting; Estimation index; Performance and progress index; Project management

1. Introduction

Project management is about being in control, making sure that the risk of failure is minimized by keeping the project in alignment with the developed project plan. It is also about making informed decisions for adjustments to the plan when required.

Project monitoring and control, on the other hand, is the process of observing the implementation of the project plan, collecting data on implementation, conducting analysis by comparing the planned values against the actual implementation records, and using this information to keep the project on track. One of the most widely accepted control systems used by practitioners in project management field is referred to as the Earned Value Analysis (EVA) or Earned Value Management System (EVMS) or in short EVM.

The evolution of EVM, for the most part, has been focused on cost management, control, and financial analysis for decades (Brandon, 1998, Fleming and Koppelman, 2004 and Kim et al., 2003). This analysis mechanism was developed based on a set of 35 criteria, called the Cost/Schedule Control Systems Criteria (C/SCSC). Later the C/SCSC method was modified to accommodate ease of use and pragmatism simply because the original version was too bureaucratic and complicated. Also, the organizations were reluctant to spend a lot of time and money on setup costs for its implementation (GAO, Significant Changes Underway in DOD's Earned Value Management Process, 1997). The U.S. government and the associated agencies opted for making the use of EVM a requirement of government contracting and bidding for projects (General Service Administration, Department of Defense, and National Aeronautics And Space Administration, 2006). There is abundance of literature on EVM and its basic principles and definitions (e.g. see Anbari, 2003 and Fleming and Koppelman, 2010).

In short, a clear project scope, a well-defined schedule, and detailed budget lay the foundation for implementing EVM and its derivatives for a project. The budget plan and schedule are to be set for the lowest level of Work Breakdown Structure (WBS) elements to be monitored and controlled. Such a plan will be considered a baseline for the project and the progress will be assessed against that baseline (Cioffi, 2006).

EVM defines parameters to enable monitoring and controlling of projects. EVM defines the Planned Value (PV) as the value that is planned to be spent for executing the work according to the original schedule at any point in time. Previously, PV was denoted as BCWS or Budgeted Cost of Work Scheduled. Earned Value (EV) is the monetary value of the progress made (work completed) at a certain point in time. Previously, EVwas denoted as BCWP or Budgeted Cost of Work Performed. Actual Cost (AC) represents the monetary value of what has been expensed to achieve the progress made at a certain point in time. Previously, ACwas denoted as ACWP or Actual Cost of Work Performed.

Prior to the dominance of EVM, PV and AC were mainly the determinant factors in measuring project progress. However, these measures were not enough for showing the entire picture and determining whether the project was getting the worth of what it was expending. EVM provides various performance measures that are designed to assist with monitoring and controlling the project depending on the type of information needed. For instance, Cost Variance (CV) is utilized to represent the difference between how much was actually spent and the value earned (EV) through the work performed. In that case, CV is defined as:

Hence, a negative value points out that the project has spent more for the executed activities than what it is worth. To the contrary, a positive value indicates that the project has spent less to gain the value of the executed work. In general, EVM and its derivatives use positive values to represent favorable signs of progress and negative values as unfavorable signs. The Schedule Variance (SV) per EVM has traditionally been used as an indicator which represents whether the project is on schedule or not:

Similar to CV, a negative SV means that the project is behind the planned schedule, whereas a positive SVrepresents a project that is ahead of schedule. It is to be noted that, this schedule performance measure is not very reliable when the cost of delayed critical activities is a fraction of the cost of other activities. Hence, the delay in completion of these critical activities does not present much of a variance! Additionally, when the activities finish, SV equals to zero as both EV and PV are the same upon completion of the activity. This outcome is irrespective of whether the actual duration of the activity happened before or after its planned completion date. There are other widely used performance measures that calculate project efficiency, or forecast a future state of the project, among other things. These include, Cost Performance Index:

Turn MathJaxon

and Schedule Performance Index:

Turn MathJaxon

Despite EVM's numerous benefits and its widespread use, there are some concerns with utilizing EVM as the only tool for monitoring and controlling cost and schedule of a project. Paige in a lucid paper some 50 years ago proposed the separation of concerns with schedule from concerns with budgetary matters (Paige, 1963). It was not till a few years ago where researchers and practitioners brought up the issues with use of EVM for schedule management. The cost management component of EVM is considered to be very effective whereas its schedule aspect has been questioned conceptually in the last few years. Fleming and Koppelman (Fleming and Koppelman, 2004) recommend that Schedule Performance Index should be used just as a warning mechanism and not as a real tool to analyze how the project is performing with regard to schedule.

Lipke argued (Lipke, 2003) that “from the time of the development of the EVM indicators, it has been known that the schedule indicators are flawed and exhibit strange behavior over the final third of the project when performance is poor”. He introduced the concept of Earned Schedule, ES, as an extension to EVM. This research shows that SPI is not an accurate or reliable measure of schedule performance. The reason given is the fact that any finished activity has an SPI equal to one. This perfect score is irrespective of the actual performance of the completed activity. To improve the performance of SPI, the Earned Schedule method converts the Earned Value at a given point in time into its equivalent duration (on the PV or planned value graph) required to achieve that Planned Value. Fig. 1 demonstrates this conversion on a conceptual EVM graph.

Fig. 1.

Conceptual EVM graph.

Figure options

Using this approach, the method provides the Earned Schedule, ES(t), for the project. Therefore, Earned Schedule could mathematically be defined as adapted from ( Jacob, 2006) as:

Turn MathJaxon

where ES(t) is Earned Schedule at the status date, EV is the Earned Value at the status date, PV_t is the Planned Value at time instant t, and the calendar unit represents the unit in which duration t is measured. The corresponding duration from the beginning of the project till status date is generally defined as Actual Time (AT) or elapsed time. We use the term Actual Duration (AD) in place of Actual Time for consistency and accuracy in this paper.

The resulting ES(t) is then compared to Actual Duration (AD). The Earned Schedule defines Schedule Performance Index denoted by SPI(t) as:

While we agree that ES and consequently SPI(t) is a better schedule/duration performance measure of the progress compared to SPI, we argue that still there are some issues associated with the use of this method for schedule performance measurement as well.

In this paper, we first present a critical evaluation of Earned Value and Earned Schedule as a tool for assessing schedule performance. We then introduce a comprehensive set of new indices which are easy to measure, understand and calculate with wider applications and implications in managing projects. These indices can be used to measure progress and performance of the project. As an example, the Duration Performance Index introduced measures the duration/time-based progress of the project compared to the actual duration passed. We also discuss the potential advantages of these measures over Earned Value and Earned Schedule indices. Additionally, we develop duration and cost estimation performance measures which can be used by contractors, clients, and the scheduling offices to assess the efficiency and efficacy of the estimated and planned durations and costs. These new measures are then applied to a real life project and the results are compared with Earned Value and Earned Schedule.

The list of acronyms and symbols which are used in this paper is given at the end of the paper for ease of reference. In addition, we have provided a separate table listing all the formulas.

2. Review of existing schedule and cost progress and performance measures

The Earned Schedule performance measure which was introduced by Lipke (W. Lipke, 2003) and later reinforced by Lipke et al. (2009) uses Earned Value which is measured in monetary terms (cost-based) to capture duration-based level of progress made in the project. Please refer to Fig. 1 and the overview of Earned Schedule in the previous section.

Vandevoorde and Vanhoucke (2006) and Vandevoorde and Vanhoucke (2007) in their comparative analysis of schedule performance measures came to the conclusion that SPI(t) is a better measure of project schedule performance compared to all the other measures including the classical Schedule Performance Index (SPI) and Cost-Schedule Index defined as:

Thus, when ES is used as a measure to predict the remaining duration or completion of the project, it provides a more reliable forecast at different stages of the project. Lipke et al. (2009) also conducted a similar research comparing the accuracy and reliability of a number of indices which could be used for forecasting the duration of a project, at different stages of progress toward completion, presumably based on past performance. They claimed that SPI has shown to produce reasonably good predictions only for large projects like those of the United States Department of Defense projects where the duration of the project is long and the volume of data used in the forecasting process is large, whereas SPI(t) generally performs better.

Though SPI(t) is argued to be a preferred option to measure duration/time performance compared to SPI, it has some conceptual shortcomings and could not be validated as a general method which is applicable to all scenarios. The main drawback of SPI(t) is the fact that, similar to SPI, it measures schedule performance using monetary terms of Earned Value (EV) and Planned Value (PV). According to “Schedule is Different” where the idea of ES is introduced, “the cumulative value of ES(t) is found by using BCWP to identify in which time increment of BCWS the cost value occurs” ( Lipke, 2003). In other words, the Earned Schedule method uses EV as a proxy to get to corresponding duration. EV is not necessarily an accurate measure of schedule/duration progress made. It is, on the other hand, generally considered the best measure of “value earned” monetarily.

Even though we might see correlations between duration and cost profiles throughout the life of a project, these profiles are not generally the same. Hence, performance measures which use a cost profile to offer a schedule measure will not be accurate. Consequently, ES, similar to any method which uses EVM's cost-based measures, will inherently be deficient in dealing with duration performance measurement. In using such models, the higher the disparity between time and cost profiles of a project, the more inaccurate the schedule performance offered will be. In such instances, SPI(t) and SPI will both show inaccurate results, and at times, SPI(t) could even perform worse than SPI. Any section of a project which shows disproportionate duration and cost profiles can demonstrate such inaccuracy, as shown in the following example: A completed procurement activity with a high PV and short duration could dramatically changeSPI(t) by overshadowing large delays of inexpensive activities on the critical path. This deficiency is illustrated in the sample project below.

In this example, there are two parallel paths. One path is longer, therefore is the critical path for the project. The critical path activities include A, B, and E. The other path contains the remaining activities, including activity D which has the highest value. This activity resembles long lead procurement activities that happen during the initial stages of a project. In this example and for simplicity, we have assumed that all activities are on time and on budget, except activity B that is not started. The marker on Fig. 2 shows the status date.

Using the given information, we calculate:

Turn MathJaxon

Turn MathJaxon

Turn MathJaxon

Calculating API_i using this equation is completely duration based.

Considering that API_i measures the schedule progress of an activity, it will always have a value of less than or equal to one. Value of API_i starts from zero and as the activity nears completion, its value approaches one, and becomes one upon completion of the activity.

Earned Duration of scheduled activity i: ED_i, at any point in time, is the value of work performed expressed as proportion of the approved duration assigned to that work for activity i (e.g. days). This variable for EDM is the duration counterpart or equivalent to EV of an activity in EVM.ED_i can normally be calculated as:

Actual Duration of scheduled activity i: AD_i, at any point in time, is the time in calendar units between the actual start of the activity and either that point in time if the activity is in progress or the actual finish date if the activity is complete. This measure for EDM method is the duration counterpart or equivalent to AC of an activity in EVM.

Baseline Planned Value of scheduled activity i: BPV_i, is the authorized budget assigned to the work to be accomplished for activity i. BPV_i is independent of the status date. Some may refer to it the baseline cost for activity i.

Planned Value of scheduled activity i: PV_i, at any point in time, is the authorized budget assigned to the scheduled work to be accomplished for the activity up to that time as defined in EVM.

Earned Value of scheduled activity i: EV_i, at any point in time, is the value of work performed expressed in terms of the approved budget assigned to that work for a schedule activity as defined and used in EVM.

Actual Cost of scheduled activity i: AC_i, at any point in time, is the total costs actually incurred and recorded for accomplishing work performed during the AD_i for the activity as defined by EVM.

It is to be reiterated that PD_i, ED_i, and AD_i for EDM are counterparts or equivalent twins of PV_i, EV_i, and AC_ifor EVM and are all dependent on the status date. BPD_i and BPV_i are indepedant of the status date and are the authorized duration and cost of activity i per baseline schedule.

3.2. Notations used at the macro level

At the macro level we define and use the following terms:

Baseline Planned Duration: BPD, is the authorized duration assigned to the scheduled work to be accomplished for the entire project irrespective of the status date.

Total Planned Duration: TPD, for the project, at any particular point in time, is the sum of PD_i for all the planned activities at that time according to baseline plan. It needs to be said that for activities crossing over the status date, only the duration of the activity to the left of the vertical status date line counts toward TPD.

Turn MathJaxon

where n is the number of in-progress and completed activities up to that time.

Total Earned Duration: TED, for the project, at any particular point in time, is the sum of ED_i for all the in-progress and completed activities at that time. Or

Turn MathJaxon

where n is the number of in-progress and completed activities up to that time.

Earned Duration: ED(t), for the project, at any point in time, is the duration corresponding to Total Earned Duration (TED) on Total Planned Duration S-curve, which mathematically could be expressed as:

Turn MathJaxon

where ED(t) is Earned Duration at status date or Actual Duration (AD). TED is the Total Earned Duration atAD and TPD_t is the Total Planned Duration at time instant t, and the calendar unit represents the unit in which time instant t is measured. At the end of the project ED(t) is the same as BPD (Baseline Planned Duration). Fig. 4 presents a conceptual graph which depicts calculation of ED(t). Table 2 depicts ED(t) calculation for the example project.

Table 2.

EDM results for the example project.

Total Actual Duration: TAD, at any point in time, for the project is the sum of AD_i for all the in-progress and completed activities at that time. Or

Turn MathJaxon

where n is the number of in-progress and completed activities up to that time.

Planned Value: PV, for the project (as used in EVM terminology) at any point in time, is the authorized budget assigned to all the activities or scheduled work to be accomplished up to that point. It can be defined and calculated as:

Turn MathJaxon

where n is the number of planned activities up to that time.

Earned Value: EV, for the project (an EVM terminology) at any point in time, is the value of work performed expressed in terms of the approved budget assigned to all the in-progress and completed activities at that time. Earned Value for the project can be defined and calculated as:

Turn MathJaxon

where n is the number of in-progress and completed activities at that time.

Actual Cost: AC, (an EVM terminology) at any point in time, for the project is the total costs actually incurred and recorded in accomplishing work performed. Actual Cost for the project can be defined and calculated as:

Turn MathJaxon

where n is the number of in-progress and completed activities at that time.

Actual Duration: AD, for the project at any particular point in time, is the period of time from the start of the project till that time.

It is to be noted that TPD, TED and TAD for EDM are counterparts or equivalent twins of PV, EV and AC for EVM.

3.3. Progress and Performance Measures for Duration and Cost

The measures defined below, classified into micro and macro levels, will assist project managers in monitoring, evaluation, and control of duration/schedule and cost progress and performance and the ability to deliver the project on time and on budget.

3.3.1. Micro level progress and performance measures for duration and cost

Duration Performance Index, for activity i: DPI_i, at any point in time, is a measure of schedule performance in delivering the activity. In other words, it shows how well the activity is doing in achieving the target completion date. It can be calculated as:

Turn MathJaxon

DPI_i can have values of greater than one presenting a better performance than planned, lower than one indicating worse performance compared to plan, or one which suggests performance is according to plan. The higher or lower performance will result in the activity being seen as ahead or behind schedule respectively. Thus, DPI_i is an indicator of the activity being on, ahead, or behind schedule.

Upon completion of the activity: DPI_i represents the overall performance in delivering the activity compared to the plan. From a different perspective, this end value of DPI_i for activity i represents the accuracy of the original duration estimate. We call the value of DPI_i at completion of activity i the Duration Estimation Index for that activity (DEI_i) and define that index in the next section and explain its applications.

Earned Duration Index, for activity i: EDI_i, at any point in time, is a measure of duration earned (e.g. number of days) compared to what was planned to be done up to that point in time. We define:

Turn MathJaxon

At any point in time, an activity might have achieved more, less, or the same amount of work in comparison with the plan up to that point in time. Thus, this measure can have values of higher than one, lower than one, or equal to one, respectively.

Adapting directly from EVM, at the activity level, we define:

Schedule Performance Index, for activity i: SPI_i, at any point in time, could be formulated as:

Turn MathJaxon

The value of SPI_i, similar to API_i, is zero before the activity starts and reaches maximum value of one when the work is complete. However, despite the similarity in starting and finishing values of these two performance indices, SPI_i does not measure the overall duration focused progression of the activity. It, rather, measures the monetary value of the work done compared to the plan up to that time. Another difference between these two indices is in their range of possible values. Unlike API_i, SPI_i can take on values of greater than one when the activity is in progress and cost-wise has achieved more than what has been planned, or lower than one when the activity is underperforming. This characteristic might suggest SPI_i is similar to Duration Performance Index for activity i, DPI_i, due to the resemblance in the values possible. However, SPI_i is actually more similar to Earned Duration Index, for activity i, EDI_i. At the same time, we argue that SPI_i is not a direct schedule progress or performance measure. Hence, it may not necessarily provide the insight that either API_i, DPI_i, or EDI_i offer.

Adapting directly from EVM, we next define:

Cost Performance Index, for scheduled activity i: CPI_i (an EVM terminology), at any point in time, as:

Turn MathJaxon

3.3.2. Macro level progress and performance measures for duration and cost

At the macro, project, level we define:

Project Progress Index: PPI, at any point in time, is a measure of overall duration progress of the project. It measures ED(t) against the baseline planned duration as:

Turn MathJaxon

It is to be noted that because PPI represents the schedule progress, at macro level, its value is always less than or equal to one. PPI's value starts from zero and as the project nears completion, the value of ED(t) approaches to BPD which results in the value of PPI approaching one. At the end of the project, ED(t) is the same as BPD. Hence, PPI value will be one upon completion of the project.

Duration Performance Index, for project: DPI, at any point in time, represents the overall schedule progress performance toward the completion of the project. In other words, it shows how well the project is doing in achieving the target completion date in consideration of the critical path. Duration Performance Index is defined as:

Turn MathJaxon

The value of DPI will be less than one, if the project schedule is not performing at the same level as planned, hence it will be perceived as being behind schedule. DPI will be equal to one, if project is overall performing at the same level as planned. Thus, the project will be seen as on schedule. And DPI will be greater than one when the project schedule is overall performing better than planned which will result in project being ahead of the schedule. Similar to SPI(t), Duration Performance Index for project provides the measure of progress performance on the critical path and toward completion of project.

Earned Duration Index, for project: EDI at any point in time, is a duration-based measure of overall work performed in terms of Earned Duration, in comparison with the work planned up to that point in time. This measure does not consider activity dependencies and is irrespective of their interactions to achieve the end date of the project. We define:

Turn MathJaxon

At any point in time duration-wise a project might have achieved more, less, or the same amount of work in comparison with the work planned to be achieved by that time. Thus, this measure can have values of greater than one, lower than one, or equal to one respectively.

To demonstrate the difference between DPI and SPI(t) on one hand, and EDI and SPI on the other, we have prepared two similar conceptual graphs. Fig. 3 is the familiar EVM or ES graph whereas Fig. 4 is its EDM counterpart showing TPD, TAD, and TED in an EDM graph. Comparing these two graphs shows that DPI is a measure similar to SPI(t), and EDI is similar to SPI.

Fig. 3.

Conceptual EVM graph.

Figure options

Fig. 4.

Conceptual EDM graph.

Figure options

To demonstrate the use of these indices we calculate and present TED, TPD, and TAD for the example project above in Table 2 below:

From Table 2 one can see that TED at the status date or day 14 (i.e. AD = 14) is equal to 21 days. According to the TPD column, the project was planned to earn this duration by the end of day 11. Thus, ED(t) is roughly 11 days. Considering that the actual duration passed from the start of the project is 14 days, DPI will be calculated as:

Turn MathJaxon

For this sample project we can calculate the Earned Duration Index as:

Turn MathJaxon

Earlier, we found that SPI(t) = 0.93, and SPI = 0.92 for this example project. That is, schedule or duration performance suggested by ES or EVM is at 92 or 93 percent level. Comparing these values with DPI = 0.79 and EDI = 0.75 respectively and knowing the status of the project, it can be seen that in both cases, DPI andEDI present much more valid information about the schedule compared to SPI(t) and SPI for a project where a critical activity is severely behind schedule.

Cost Performance Index: for project, CPI, is the classical EVM index defined as:

Turn MathJaxon

The value of CPI will be less than one, if the project is over budget, equal to one, if on budget, and greater than one when under budget.

Project managers can use control charts to monitor DPI, EDI, and CPI over time with upper and lower control limits (UCL and LCL). If the index goes beyond these limits, then generally there is a need for investigation and possibly intervention, either to maximize and take advantage of an opportunity or to minimize the risks.

The progress and performance measures defined above, separate duration and cost performance measures to deliver more accurate information on the status of a project and its activities.

3.3.3. Forecasting and other uses

Similar to EVM, one other potential usage of the progress performance measures is forecasting the future state of the project. We believe that the performance of the project at each stage is a function of the nature of the project and the context and content of that stage (characteristics of the S curve) of project as supported by Bower (2007). We also believe that using any performance measure developed at one stage of a project as a proxy to forecast performance of the project many periods away in the future is questionable. As an example, in Engineering, Procurement, Construction (EPC) type contracts, assuming the entire contract will be behind schedule because the engineering stage is delayed, may not necessarily be an accurate prediction. The authors' experience shows that spending a little bit more time and resources on the engineering stage when deemed necessary could reduce the cost and duration of installation and commissioning activities (part of the construction stage) drastically. Such minimal delay, in those cases, could actually help achieve huge time and cost savings for the overall contract. Hence, if the stages of the project are different and heterogeneous, which normally is the case, there is no rationale for assuming past performance is a good predictor of the future.

In all these forecasts we assume the current performance continues and remains the same for the remainder of the project. However, we know that if the performance is not according to the plan, as a minimum the project manager normally tries to influence the project to bring it back on track, and consequently resulting in a change in performance. Thus, the performance of the project in the future is not necessarily constant and the same as the past but most probably is flexible and dynamic and responsive to past performance (Cooper, 1994, Cioffi, 2006 and Moslemi Naeni et al., 2011).

In spite of the difficulties and issues mentioned earlier, one could reasonably use any of the many simple forecasting models (e.g. moving average, exponential smoothing), and or more advance and complex approaches (fuzzy time series etcetera) to predict the value of the index for the next few periods. Forecasting project completion cost using CPI which uses EV has shown to be a relatively good choice ( Vandevoorde and Vanhoucke, 2006). On the other hand, to forecast the duration of the project, we suggest utilizing DPIwhich is based on EDM data and does not have the computational difficulties and potential errors of cost/value-based models (EVM, ES) as discussed earlier.

Despite the above, one can formulate a forecasting equation for EDM similar to the classical forecasting approaches for EVM (considering a constant performance for the remainder of the project). We define:

Estimated Duration at Completion, (EDAC), as:

Turn MathJaxon

Alternatively, one can use the following formula to estimate the duration at completion:

Turn MathJaxon

Or we can calculate the Estimated Duration To Complete, for project, EDTC as:

Turn MathJaxon

Additionally, similar to EVM, we can define a duration variances for EDM. At the activity level, Duration Variance for activity i, (DV_i), can be calculated as:

And at the project level Total Duration Variance for Project, (TDV), is defined as:

TDV shows the sum total of all the deviations from the plan which is a measure of the accuracy of the planning. In other words, it is a measure of total deviation of all the activities planned up to any point in time.

To measure the deviation from the plan on the critical path, we define Duration Variance for a project, DV, as:

Considering that we used cost parameters similar to EVM, there will be no changes to the cost variance performance measures. Hence, at the activity level:

And at the project level:

3.4. Duration and cost estimation performance measures

As discussed earlier, one aspect of control is to use the collected information for benchmarking, lessons learned, and assessing the performance of planning and scheduling departments. Therefore, we use the data collected as part of monitoring and controlling processes of project management to develop these measures. As this required data is currently available to project management teams for generating EVM reports, there will be no additional costs for data collection to produce EDM reports.

Once again, these measures can be defined at both micro, and macro levels.

3.4.1. Micro level duration and cost estimation performance measures

At the micro, e.g. activity, level we define:

Duration Estimation Index of scheduled activity i: DEI_i, as a measure to gain a better understanding of the accuracy of the original estimated duration of the activity. DEI_i compares the duration originally planned (authorized) for the activity versus what was the actual duration upon completion of the activity:

Turn MathJaxon

Similarly, we define:

Cost Estimation Index of scheduled activity i:CEI_i, as a measure to gain a better understanding of the accuracy of the original estimated cost of the activity:

Turn MathJaxon

DEI_i and CEI_i should be used to report duration and cost estimation accuracy or performance for activities only after they are completed.

Based on experience, these indices are usually less than one due to notional rules such as Parkinson's Law which states that “work expands so as to fill the time available for its completion”. These indices could occasionally be more than or equal to one as well.

3.4.2. Macro Level Duration and Cost Estimation Performance Measures

Now that estimation performance measures at the micro level have been established, we discuss macro or project level estimation performance measures. We define:

Duration Estimation Index for project: DEI, as the average DEI_i of all the completed activities (m) at any stage of the project. At the completion of the project, m is equal to M or total number of activities in the Project.

Turn MathJaxon

where m is the number of activities which are finished.

We use the average function at the project level, with some reservations. This averaging produces better performance measure for the activities within similar functions or departments such as engineering, design, development, etc. This index can be used for benchmarking or any kind of comparative analysis with the previously executed projects. One can use other methods such as weighted average (based on duration, cost, effort, etcetera, as most appropriate) to get a more accurate Duration Estimation Index at higher levels depending on the activity types. The weighted average version of Duration Estimation Index for project could be formulated as:

Turn MathJaxon

where w_i is the designated weight for the activity i and m is the number of activities which are finished.

Cost Estimation Index for project:CEI, is the cost counterpart of DEI and measures average cost estimation performance for the project based on all completed activities (m).

Turn MathJaxon

where m is the number of activities which are finished.

Similar to DEI, one can use weighted averages to get a more accurate Cost Estimation Index at higher levels depending on the element types, size, breakdown or based on duration, cost, effort, etc. In which case the following formula can be used:

Turn MathJaxon

where y_i is the designated weight for the activity i and m is the number of activities which are finished.

Knowing that the planned durations are not always one hundred percent reliable, we expect some variation in the values for DEI and CEI for different types of projects. For deterministic projects we expect higher reliability of duration and cost estimates and less risk of delays or over-runs. This is in contrast to probabilistic projects (Research, Engineering, and Development projects) in which the estimates are less reliable and performance variation is expected ( Khamooshi and Cioffi, 2013).

3.4.3. Possible values of defined estimation measures

The value of DEI_i, CEI_i, DEI, CEI could be higher or lower than one. For these indices, a value of more than one is a sign of better performance compared to what has been agreed on during development of plan. To the contrary, values less than one are indicative of worse than expected performance. One can use control charts to monitor DEI and C EI over time with upper and lower control limits (UCL and LCL) for evaluation of estimation. This process could provide us with the opportunity for taking advantage of any opportunity for improving the estimation process and minimizing the risks associated with estimates.

Using a project management software with built-in functionality to input and process performance management data, all the aforementioned measures can be developed and reported by filtering the relevant activities or work packages for each section, department, or part of project. This filtering operation, i.e. separating homogenous activities, could provide very valuable information for assessing the schedule and estimation performance of any particular section or department.

In addition to providing more accurate information about the duration status of the project and its activities, another benefit of EDM is its estimation performance measures. These performance measures provide us better tools to control the project, document lessons learned by comparing estimates against actuals, and improve the estimation for future projects. Consequently, EDM empowers the project manager to have a better handle on project duration control and higher chance of delivering a project successfully.

4. EDM applied to the United States Coast Guard (USCG) project

In this section we apply the suggested performance measures to a real life United States Coast Guard (USCG) project. This project upgrades the capacity and capability of an old storage garage boat management facility to that of a modernly equipped boat handling and maintenance facility. We have proportionally reduced the costs to maintain the confidentiality of project data while keeping the baseline schedule data intact.

We will first demonstrate EVM, ES, and EDM data when 18 out of 46 weeks of the planned project duration has passed. We show this information at the end of week 37 as well. Later, we will show the performance measures at project completion.

The calculations below are fairly straightforward and are accommodated by using a project management software package (Microsoft Project in this case).

Fig. 5 shows EVM graph at the end of week 18 for this project (Table 1.A in Appendix 1 presents the same data in a tabular format).

Fig. 5.

EVM graph at the end of week 18.

Figure options

EVM calculates SPI as:

Turn MathJaxon

Using Table 1.A containing the data extract of the Microsoft Project EVM report makes it easier to calculateES(t). EV at the end of week 18 is equal to $467,789. According to the PV column, the project was planned to earn this value roughly during week 11. Thus, ES(t) is approximately 11 weeks. Considering that AD is 18 weeks, SPI(t) will be calculated as follows:

Turn MathJaxon

Using USCG project data available for generating EVM graphs, we can produce the EDM graph as shown inFig. 6. This figure depicts a noticeably different profile for EDM compared with the one presented in Fig. 5 for ES.

Fig. 6.

EDM graph at the end of week 18, the total duration (y axis is in days) whereas the project duration/progress is in weeks.

Figure options

The same data can be shown in a tabular format presenting TED, TPD, and TAD ( Table 2.A in Appendix 2).

TED at the status date of week 18 is equal to 180 days. According to the TPD column, the project was planned to earn this duration during week 16. Thus, ED(t) is roughly 16 weeks. Considering that the actual number of weeks passed from the start of the project is 18 weeks, DPI and EDI will be calculated as:

Turn MathJaxon

The Project Progress Index measures the time-based (duration) progress of the project compared with the baseline planned duration and is calculated as follows:

Turn MathJaxon

The status of the project according to EVM at the end of week 37 is presented in Fig. 7. Table 3.A inAppendix 3 presents the same data in a tabular format.

Fig. 7.

EVM graph at the end of week 37.

Figure options

EVM calculates SPI as:

Turn MathJaxon

Using Table 3.A containing the data extract of the Microsoft Project EVM report makes it easier to calculateES(t). EV at the end of week 37 is equal to $870,915. According to the PV column, the project was planned to earn this value roughly during week 29. Thus, ES(t) is approximately 29 weeks. Considering that AD is 37 weeks, SPI(t) will be calculated as follows:

Turn MathJaxon

EDM graph at the end of week 37 is shown in Fig. 8. This figure depicts a noticeably different profile for EDM compared with the one presented in Fig. 7 for ES. The same data is shown in a tabular format presentingTED, TPD, and TAD Table 4.A, Appendix 4.

Fig. 8.

EDM graph at the end of week 37, the total duration (y axis is in days) whereas the project duration/progress is in weeks.

Figure options

TED at the status date of week 37 is equal to 282 days. According to the TPD column, the project was planned to earn this duration during week 30. Thus, ED(t) is roughly 30 weeks. Considering that the actual number of weeks passed from the start of the project is 37 weeks, DPI and EDI will be calculated as:

Turn MathJaxon

The Project Progress Index is calculated as follows:

Turn MathJaxon

This time around, SPI(t) is much closer to the performance suggested by DPI. As explained in the previous sections, this is mainly due to cost and schedule profiles getting closer to each other at this stage of the project. And is not a sign of SPI(t) being a good schedule performance measure, or that SPI(t) could be used in place of DPI as a good schedule measure. Had the cost and schedule profiles not been close at this stage, similar to the results at the end of week 18, the performance suggested by SPI(t) would have been different and inaccurate compared to the value suggested by DPI.

The result from both the earlier example project and the USCG project shows that EDM provides a more realistic schedule assessment. In case of USCG project SPI(t) is a much better measure than SPI but as it was discussed above, the duration-based index is a more rational alternative and natural choice when dealing with duration or time.

Using the USCG project data upon completion, the calculated results for micro (activity) level measures are presented in Table 5.A, Appendix 5.

As expected, the activity progress index for each activity is at 100% showing the activity is completed. As explained earlier and can be seen in Table 5.A, upon completion of the activities, the duration performance and estimation indices will be the same.

Using this data, and the calculated DEI_i and CEI_i we can calculate Duration Estimation Index for the USCG project. DEI is given by:

Turn MathJaxon

This means that the overall duration estimation accuracy was about 82%.

And CEI is calculated as follows:

Turn MathJaxon

This means that the estimating team had an overall accuracy of about 89% in estimating the cost for the activities of this project.

These findings could be used for review and assessment of estimation methods and improving the accuracy of planning data for the future projects.

Considering that the USCG project was completed in week 57 as opposed to the scheduled week 46, hence, we calculate the Duration Performance Index, for project as:

Turn MathJaxon

As mentioned earlier, EDI at the end of the project will always be equal to one. We can calculate this index as:

Turn MathJaxon

And, of course, we expect the Project Progress Index at the end of the project to be 100% or equal to one as follows:

Turn MathJaxon

Table options

Appendix 1.

Table 1.A below presents EVM/ES results for the USCG project at the end of week 18.

Table 1.A.

EVM results, USCG project, at the end of week 18.

Appendix 2.

Table 2.A below presents EDM results for the USCG project at the end of week 18.

Table 2.A.

EDM results, USCG project, at the end of week 18.

Appendix 3.

Table 3.A below presents EVM/ES results for the USCG project at the end of week 37.

Table 3.A.

EVM results, USCG project, at the end of week 37.

Appendix 4.

Table 4.A below presents EDM results for the USCG project at the end of week 37.

Table 4.A.

EDM results, USCG project, at the end of week 37.

Appendix 5.

Table 5.A below presents activity actual duration and cost upon completion of the USCG project along with the performance measures for each activity.

Table 5.A.

Activity actual duration and cost along with performance measures for the USCG project, upon completion of the project.

Table options

References

• Frank T. Anbari
• Earned value project management method and extensions
• Proj. Manag. J., 12–23 (2003)
• Douglas Bower
• New directions in project performance and progress evaluation
• School of Construction, Property and Project ManagementRMIT University (2007)
• Daniel M. Brandon Jr.
• Implementing earned value easily and effectively
• Proj. Manag. J., 80–85 (1998)
• D.F. Cioffi
• Completing projects according to plans: an earned value improvement index
• J. Oper. Res. Soc. (57) (2006), pp. 290–295
• View Record in Scopus | Full Text via CrossRef | Cited By in Scopus (7)
• K.G. Cooper
• The $2000 hour: how managers influence project performance through the rework cycle
• Proj. Manag. J., 25 (1994), pp. 11–24
• Fleming, Koppelman
• If EVM is so good… why isn't it used on all projects?
• The Measurable News (2004), pp. 1–5
• View Record in Scopus | Full Text via CrossRef | Cited By in Scopus (35)
• Quentin W. Fleming, Joel M. Koppelman
• Earned value project management
• (third edition)Project Management Institute, Newtown Square, PA (2010)
• GAO
• Significant changes underway in DOD's earned value management process
• GAO, Washington, D.C. (1997)
• GAO
• GAO cost estimating and assessment guide
• (2009) (Washington)
• General Service Administration, Department of Defense, and National Aeronautics and Space Administration
• Federal Acquisition Regulation
• U.S. General Services Administration (2006)
• D. Jacob
• Is “Earned Schedule” an unreliable indicator? No, but It's not necessarily the premier indicator for assessing schedule performance
• The Measurable News (2006)
• D.S. Jacob, M. Kane
• Forecasting schedule completion using earned value metrics…revisited
• The Measurable News (2004), pp. 10–17
• H. Khamooshi, Cioffi Denis
• Uncertainty in task duration and cost estimates: fusion of probabilistic forecasts and deterministic scheduling
• J. Constr. Eng. Manage., 139 (5) (2013), pp. 488–497
• Full Text via CrossRef
  • EunHong Kim, William G. Wells Jr., Michael R. Duffey
  • A model for effective implementation of Earned Value Management
  • Int. J. Proj. Manag. (2003), pp. 375–382
  • Article | PDF (215 K) | View Record in Scopus | Cited By in Scopus (28)
  • Walt Lipke
  • Schedule is different
  • Proj. Manage. Inst. Coll. Perform. Manage. (2003, 5 March)
  • W. Lipke, Ofer Zwikael, Kym Henderson, Frank Anbari
  • Prediction of project outcome: the application of statistical methods to earned value management and earned schedule performance indexes
  • Int. J. Proj. Manag., 27 (4) (2009), pp. 400–407
  • Article | PDF (164 K) | View Record in Scopus | Cited By in Scopus (33)
  • Leila Moslemi Naeni, Shahram Shadrokh, Amir Salehipour
  • A fuzzy approach for the earned value management
  • Int. J. Proj. Manag., 29 (6) (2011), pp. 764–772
  • Hilliard W.W. Paige
  • How PERT-cost helps the general manager
  • Harv. Bus. Rev., 41 (6) (1963), pp. 87–95
  • Project Management Institute
  • Practice standard for Earned Value Management
  • (second edition) (2005)
  • Project Management Institute
  • Practice standard for Earned Value Management
  • (2005)
  • Project Management Institute
  • A guide to the project management body of knowledge (PMBOK® guide)
  • (fifth edition)Project Management Institute (2013)
  • Standish Group
  • CHAOS manifesto 2011
  • (2011)
  • J.R. Turner
  • Evolution of project management as evidenced by papers published in the International Journal of Project Management
  • Int. J. Proj. Manag. (2010), pp. 1–6
  • Article | PDF (99 K) | View Record in Scopus | Cited By in Scopus (9)
  • Stephan Vandevoorde, Mario Vanhoucke
  • A comparison of different project duration forecasting methods
  • Int. J. Proj. Manag., 24 (4) (2006), pp. 289–302
  • Article | PDF (380 K) | View Record in Scopus | Cited By in Scopus (62)
  • S. Vandevoorde, M. Vanhoucke
  • A simulation and evaluation of earned value metrics to forecast project duration
  • J. Oper. Res. Soc., 58 (10) (2007), pp. 1361–1374
  • Mario Vanhoucke
  • Measuring time: improving project performance using earned value management
  • (2012)

Corresponding author at: Department of Decision Sciences, Funger Hall, Suite 415, The George Washington University, Washington, D.C. 20052, United States. Tel.: + 1 202 994 4862; fax: + 1 202 994 2736.

Federal Aviation Administration.

Copyright © 2013 Elsevier Ltd. APM and IPMA. Published by Elsevier Ltd. All rights reserved.

Note to users: Corrected proofs are Articles in Press that contain the authors' corrections. Final citation details, e.g., volume/issue number, publication year and page numbers, still need to be added and the text might change before final publication.

Although corrected proofs do not have all bibliographic details available yet, they can already be cited using the year of online publication and the DOI , as follows: author(s), article title, journal (year), DOI. Please consult the journal's reference style for the exact appearance of these elements, abbreviation of journal names and use of punctuation.

When the final article is assigned to an issue of the journal, the Article in Press version will be removed and the final version will appear in the associated published issue of the journal. The date the article was first made available online will be carried over.