PRT-57, 21 Nov 2010
Measuring Schedule Performance NASA Project Management Challenge 2011
8 - 10 February 2011
Tony Harvey
Measuring Schedule Performance NASA Project Management Challenge 2011
8 - 10 February 2011
Tony Harvey
Los Angeles Washington, D.C. Boston Chantilly Huntsville Dayton Santa Barbara
Albuquerque Colorado Springs Ft. Meade Ft. Monmouth Goddard Space Flight Center Ogden Patuxent River Silver Spring Washington Navy Yard
Cleveland Dahlgren Denver Johnson Space Center Montgomery New Orleans Oklahoma City San Antonio San Diego Tampa Tacoma Vandenberg AFB
Approved For Public Release
PRT-57, 21 Nov 2010
AcknowledgementsAcknowledgements
NASA JSC, Constellation Mission Operations Project funded the concept development and tool development
Terri Blatt for her support in applying the technique to the MOP PP&C environment
Greg Hay for keeping monthly revisions of the MOP schedule and providing them for use in testing the schedule comparison tool
Mike Stelly for his help with the presentation content
2Approved For Public Release
PRT-57, 21 Nov 2010
OverviewOverview What is Schedule Performance?
Purpose: develop methods/techniques to analyze schedule performance over time
Development to date consists of two pieces: Performance Metrics and Toolset for analyzing data
Toolset takes two schedules and extracts appropriate data MS Project-based schedules Prototype stage of development
Batch program that created an Excel style tab-delimited text output Excel macros to format the spreadsheet for easy viewing Prototype Desktop program for immediate display and optional saving to an Excel
file
3Approved For Public Release
PRT-57, 21 Nov 2010 4
What is Schedule Performance? and Why do we need more metrics?
What is Schedule Performance? and Why do we need more metrics?
The collection of project cost performance measures based on actual resource cost is often made difficult by inaccurate or missing resource cost data
Yet even basic schedules includes quantitative data, which can be used in measuring schedule performance. This includes: Task start and end dates Task durations Task completion assessment
The VALUE expressed in all schedules is in the TIME COST (duration) of the tasks
Using two schedules in a comparison process provides A statement of planned activity, in the earlier schedule A statement of performed activity, in the later schedule
Schedule Performance is simply “measuring a project’s ability to complete its planned activities in a given timeframe”
Approved For Public Release
PRT-57, 21 Nov 2010
Schedule Performance,Definition of Terms
Schedule Performance,Definition of Terms
Paralleling the traditional earned value approach to performance measurements we can define terms for measuring our schedule performance against the plan as follows: Planned Duration Of Work Scheduled (PDWS)
Original planned duration of activities
Planned Duration of Work Performed (PDWP) Earned duration of completed activities
Actual Duration of Work Performed (ADWP) Actual duration of completed activities
5 Approved For Public Release
Schedule 1
Schedule 2
Original plan is for 10 days, therefore
PDWS=10
Completed 100% of the 10 day planned activity, therefore PDWP=10
Activity took 12 days to complete, therefore
ADWP=12
PRT-57, 21 Nov 2010 6
Schedule Earned ValueMetrics
Schedule Earned ValueMetrics
Based on the performance measures a number of metrics can be calculated Schedule Variance for Duration (SVd): PDWP - PDWS
The difference between earned duration and planned duration Negative values imply a schedule slip
Schedule Performance Index for Duration (SPId): PDWP/PDWS Schedule efficiency factor representing the relationship between the earned
duration and the planned duration Values less than 1.0 indicate a performance shortfall
Schedule Cost Performance Index (SCPI) : PDWP/ADWP A Schedule cost efficiency factor representing the relationship between the
earned duration and the actual duration Values less than 1.0 indicate a cost (duration) overrun
Approved For Public Release
If PDWS=10, PDWP=10, and ADWP=12, thenSchedule Variance (SVd): PDWP – PDWS = 10 - 10 = 0 - Interpretation: The scheduled task is earning value on schedule
Schedule Performance Index (SPId): PDWP/PDWS = 10 / 10 = 1.0- Interpretation: The scheduled task has earned value perfectly against its planned value
Schedule Cost Performance Index (SCPI): PDWP/ADWP = 10/12 = .833- Interpretation: The scheduled task took longer (cost more) to complete than originally planned
PRT-57, 21 Nov 2010
Project Level Measures,A Simple Example
Project Level Measures,A Simple Example
7Approved For Public Release
Schedule Variance (SVd): PDWP – PDWS = 25-30 = -5 - Interpretation: The cumulative effect of all schedule tasks analyzed are 5 days behind schedule (not to be interpreted as the overall project is 5 days behind schedule)
Schedule Performance Index (SPId): PDWP/PDWS = 25/30 = .83- Interpretation: The project has currently earned 83% of the duration that it had planned to-date
Schedule Cost Performance Index (SCPI): PDWP/ADWP = 25/32 = .78- Interpretation: Tasks are taking longer to complete than originally planned
Schedule 1
Schedule 2
Task 1PDWS: 10PDWP: 10ADWP: 12
Task 2PDWS: 10PDWP: 8ADWP: 10
Task 3PDWS: 8PDWP: 7ADWP: 8
Task 4PDWS: 2PDWP: 0ADWP: 2
Overall ProjectPDWS: 30PDWP: 25ADWP: 32
PRT-57, 21 Nov 2010 8
Why Create a Schedule Comparison Tool from Scratch?
Why Create a Schedule Comparison Tool from Scratch?
There are no existing tools that create these performance measures
We were motivated to build a tool that could be used for comparing ANY two schedule instances from the same project
Performance measures can be output in a user friendly format (Excel) or directly to a database
Comparing the current schedule against the original schedule provides performance measures for the project up to the current schedule’s status date (Data Date)
Using schedules with monthly Data Date intervals will provide performance measures for that month interval
Monthly performance measures can be used for performance trending
Performance Indices can be used for duration projections and schedule confidence level analyses
Approved For Public Release
PRT-57, 21 Nov 2010 9
Creating aSchedule Comparison
Tool
Creating aSchedule Comparison
Tool
PRT-57, 21 Nov 2010
Key Requirements for our Schedule Comparison ToolKey Requirements for our
Schedule Comparison Tool Use any two revisions of a project’s schedule
Create output that “aligns” the two schedules at the task level Tasks are aligned by Task Name
Create Performance Measures for each task
Create Performance Measures for the project
Retain the schedule hierarchical structure Schedules are by their nature organized in a hierarchical structure of
summary tasks and regular tasks
Create Performance Measures at each Summary task level Allows performance measures to be used to reflect the task
organization, as modeled in the schedule hierarchy (project teams?, project phases?)
Create data capable of being stored in a database
10Approved For Public Release
PRT-57, 21 Nov 2010 11
Real World DataReal World Data
Schedule 1 Structure
Schedule 1 Task Data
Task Performance Metrics
Project Performance Indexes
Interpretation of Project Metrics• All currently analyzed tasks are cumulatively behind
schedule by 61.32 days. • Analyzed tasks have earned 92.1% of the duration they
should have earned• Tasks are being accomplished at a 65.5% efficiency rate
Schedule 2 Task Data
Summary Performance Metrics
Projected Durations using SCPI
Approved For Public Release
PRT-57, 21 Nov 2010 12
Current Issues & Future Development
Current Issues & Future Development
Current Issues Metrics and process
SPId tends toward 1 as task nears completion May exceed 1 if task finishes early
SPId subject to how scheduler determines “% complete” field
SCPId is used for new duration estimate SCPI (PDWP/ADWP) can be zero when task is in progress
Tool Uses task names as the identifier to compare tasks
Renamed tasks are not comparable unless new name “contains” old name Changes in schedule hierarchy makes analysis difficult
Tool option allows looking one level deeper Added tasks are ignored Deleted tasks are excluded from performance metrics
Future Development Data capture for trending analysis Extraction of task “confidence measures” (low, mode, high duration)
Approved For Public Release
PRT-57, 21 Nov 2010
ConclusionsConclusions Strengths
Does not require resource cost data for performance measures Allows ANY two revisions of a schedule to be compared Creates Task, Summary Task and Project-level performance measures Summary Task performance indices identify items needing attention
Weaknesses Relies on project percent complete measures from the schedules New (projected) duration estimation function needs refining
Conclusion Useful performance measures can be obtained from basic schedule data Performance indices are useful focusing functions Schedule comparison provides an earned value (duration) perspective
without an EVMS
13Approved For Public Release