Transcript
Project Control Methodologies
Vladimir Liberzon
Project Control Methodologies
In this presentation we will discuss methods and tools of project planning and performance per formance analysis. Some of these methods are widely used and we will touch them very briefly, others are known and applied by advanced teams in complex projects. There are also very v ery efficient methods used in Russia and Eastern Europe but not known in the West.
Project Control Methodologies
In this presentation we will discuss methods and tools of project planning and performance per formance analysis. Some of these methods are widely used and we will touch them very briefly, others are known and applied by advanced teams in complex projects. There are also very v ery efficient methods used in Russia and Eastern Europe but not known in the West.
Project Control Methodologies
We will discuss project control methodologies evolution and advanced methods and tools of project control using simple sample project.
Critical Path Method
A start to modern project management development was done when Critical Path Method was created and applied in late 50-s. Basing on CPM it became possible to calculate project duration and estimate the consequences of management decisions before their application. CPM includes calculating forward and backward schedules, determining activity total floats (an interval between early and late dates), selecting activities with zero total floats (critical) and determining the sequence of linked critical activities from project start to project finish milestones.
Critical Path
Critical Path is the longest sequence of linked activities. In our example it contains activities 4-1, 4-2, 4-3 that are shown in red. Critical Path duration is the same as project duration (41 days in our example).
Critical Path
But if the schedule contains imposed dates like Start No Earlier Than, or use different activity and resource calendars, Critical Path (consisting of activities with zero total float) may include only one final activity as in the example below.
Considering uncertainties
It looks like we can decide that planned project duration of our sample project shall be set as 41 days. But for setting reliable targets it is necessary to consider project risks, uncertainties and restrictions. Potential risk events may change project scope, activity network, project budget and resource requirements. Estimates of project parameters like activity durations, costs, resource requirements are uncertain. Project targets shall be set taking into account all risks and uncertainties.
Three estimates
Let's collect optimistic, expected and pessimistic estimates of our sample project parameters.
CPM Duration Distribution
Risk simulation shows that expected duration (41 days) has less than 35% probability to be achieved.
Critical Indices
Most critical activities with Criticality Indices 0.39 (critical in 39% of all simulations) belong to the path with minimal duration and thus need most attention. Critical activities have the least Criticality Index.
Resource Constraints
But Critical Path Method does not consider resource limitations. With limited resources, supplies, financing it may create schedules that are not feasible. In our example sample project has limited resources A and B, two units of each. But the schedule created by CPM requires more resources than are available as shown in the next slide. In resource constrained schedule activities are delayed because resources needed for their execution are busy performing other activities.
CPM Schedule Resource Histograms
Resources are overloaded
Resource Constrained Schedule
Resources are leveled but project finish delayed.
Resource Critical Path
Resource Critical Path shows the longest sequence of project activities in the schedule (if the schedule has no imposed dates). Activities that belong to Resource Critical Path have zero resource constrained floats. Another name for RCP is Critical Chain. But RCP depends on the leveling algorithms used by the scheduling software. Resource constrained schedules created for the same projects by different software packages may be different and most software packages calculate wrong resource constrained floats and thus show wrong RCP, as shown in the next slide.
Schedule Comparison
In the left schedule RCP consists of activities 3-1, 1-2, 2-3, but the software shows activities 2-1, 2-2, 2-3 as critical though first two activities of this path have positive resource constrained floats.
Risk Simulation of Constrained Schedule
If to apply risk simulation to resource constrained schedule project duration probability distribution will depend on the software leveling algorithm. If to use one algorithm for Monte Carlo simulation and another for project resource management the results of MC simulation will be useless. So using external software (add-in) for MC simulation is feasible only for the schedules where resources are not limited. Next slide shows sample project duration probability curve if the software at each iteration searches for an optimal resource constrained project schedule.
Project Duration Distribution
It shows that with constrained resources project will certainly last much more than 41 work day and it is reasonable to set 50 days as target duration for reliable execution.
Project Duration Distribution
But if to use most common resource allocation rule (activities with least total float have highest priorities) a probability to finish in 50 days will become only 23.47% instead of 73.52% if resource allocation decisions were optimized.
Resource Constrained Criticality Indices
Criticality Indices show which activities require most attention and again some non-critical activities (4-2, 4-3) have larger CI than resource critical. Without risk analysis this could be missed.
Cost Planning
The same approach is used for planning costs and other project parameters. Entering three estimates of initial data permits to simulate uncertainties and set reliable targets.
Cost Probability Curve
Risk simulation shows that calculated project budget (3700) will be met with only 37.3% probability.
Cost Target
Let's set 4000 as the target budget that has 80.1% probability to be met.
Setting Multiple Targets
Setting multiple project targets requires special attention. Let's assume that set duration target has 70% probability to be achieved and the same probability has budget target. If to suppose that activity duration and cost estimates do not depend on each other the probability to meet both targets is 49% and it is not sufficient. Setting multiple targets requires multidimensional risk analysis. Cost-Schedule Scatter Diagram shown in the next slide is a tool for setting complex target.
Scatter Diagram
Probability to meet both targets (50 days project duration and 4000 budget) is 58.89% though probabilities to meet each of these targets separately exceeded 70%.
Integrated Target
Setting multiple targets makes management decisions complicated. It does not answer to the question if project execution is successful if the project is late but saves money, if it makes sense to pay additional money for acceleration, etc. That is why we advice to estimate the cost of one day of project finish delay and acceleration and set one integrated target – project budget that includes cost of the time. Next slide shows expected integrated budget distribution if to set that one day of project finish delay costs 300 and one day of finish acceleration saves 50.
Integrated Target Budget Distribution
Meeting 4000 budget and finishing in less than 50 days means the same as setting 4000 cost target in the following cost probability distribution that takes into account potential penalties and awards.
Working Schedule
Setting reasonable project targets means to create project time and cost buffers. There are two options: to create and to use the schedule that finishes on target date and to assign costs that constitute target budget or to use tight schedules and budgets but to add and to manage project time and cost buffers. The usual approach is to create and manage two schedules: Contract schedule that has target duration and budget, and Working schedule that is tight, with project finish earlier than the target date and the budget that is lower than target budget.
Parkinson Law Work expands so as to fill the time available for its completion
To avoid Parkinson Law problems it is necessary to set tight targets for project workforce and to keep contingency reserves for management to deal with risks and uncertainty. Using tight schedule for project management and managing project buffers is an approach common to Critical Chain and Success Driven Project Management (CCPM and SDPM) methodologies. CCPM suggests to use most probable estimates and SDPM suggests to use optimistic estimates of activity durations in the working schedule and analyze project performance by estimating buffer penetration.
Working Schedule
This slide shows optimistic schedule that is selected as Working compared with most probable schedule that can be used as the baseline, though its duration and budget are not the same as the target duration and budget. Besides, there is different Critical Path.
Some Conclusions
Critical Path Method is used for creating project schedules when resources are not limited or these limitations are not known.
Feasible schedules shall take into account resource, supply and financing constraints and usually are longer than produced by CPM. Different software packages use different leveling algorithms and may produce different resource constrained schedules for the same project. Collecting three point estimates and analyzing project risks is necessary to set reliable project targets and understanding which activities require most attention.
Some Conclusions
Monte Carlo risk analysis shall use the same leveling rules as the software that will be used for project management.
Setting multiple targets makes management complicated and achieving them all less probable. It makes sense to set integrated project success criterion that helps to justify project management decisions. Working schedule shall be tight and time and cost contingency reserves created and managed.
Performance Analysis Methods
Project Performance Analysis (further PPA) is necessary for:
1) Estimating past project performance, 2) Forecasting future project performance, 3) Making decisions on necessary corrective or preventive actions
Methods of PPA include^
1) Variance Analysis 2) Earned Value Analysis 3) Trend Analysis 4) Success Probability Trend Analysis
Variance Analysis
Variance Analysis compares current schedule with the baseline or some previous project version and finds those parameters that were changed and need attention. It helps to find the origins of delays and cost overruns for analysis and decision making. In the next slide current schedule was compared with the Baseline (most probable schedule) and initial Working schedule (optimistic). The difference between current and baseline values of activity durations and costs are shown in corresponding columns.
Variance Analysis
Duration and Cost variances with the baseline are shown also as signals (red – if duration is longer than in the baseline, cost is larger than in the baseline, yellow if they are the same).
Earned Value Analysis
Earned Value Analysis is widely used in large construction programs. It is based on comparison of the following three parameters: EV (Earned Value) is the budgeted (baseline) cost of the work actually performed,
PV (Planned Value) is the baseline cost to the current moment,
AC (Actual Cost) is the actual cost of the work performed.
Earned Value Analysis
Comparing EV and PV we can estimate time performance – if we have done more or less work than was planned. Schedule Variance SV=EV-PV is positive when the budgeted cost of the work done exceeds budgeted cost of the work scheduled in the baseline. Schedule Performance Index SPI=EV/PV shows the ratio of the Earned Value and Planned Value and exceeds 1 when project time performance is successful.
Earned Value Analysis
Comparing EV and AC we can estimate cost performance – if we have spend more or less than was planned. Cost Variance CV=EV-AC is positive when the budgeted cost of the work done exceeds actual cost of the same work. Cost Performance Index CPI=EV/AC shows the ratio of the Earned Value and Actual Cost and exceeds 1 when project cost performance is successful.
Looking at SV, SV, SPI and CPI it is easy to understand if project performance up to date was successful or not.
Earned Value Analysis
EVA also forecasts future performance basing on past experience. Estimate At Completion (EAC) is expected project cost that is calculated by formula: −
EAC=AC+PF*(BAC-EV) where
−
BAC is Budget At Completion – Baseline Budget,
−
PF is Performance Factor
Default PF=1/CPI but other values may be applied. In particular it is recommended to apply PF=1/(CPI*SPI)
Earned Value Analysis
EVA shall be applied carefully for the following reasons: It does not distinguish between the works done on critical activities and activities with sufficient floats. A project could be late but EVA will not notice this problem if Earned Value exceeds Planned Value. It motivates project managers to do expensive tasks first delaying cheaper activities that could have higher priorities, It suggests to forecast future performance basing on past experience that may be wrong if resources that planned to be used in the future are not the same as in the past, It does not allow for risks and uncertainties
Earned Value Analysis
In the projects with expensive materials and equipment it is reasonable to apply EVA to the cost of work only. So EVA is usually applied to cost components and cost centers, man-hours, and other parameters that may be used for measuring work scheduled and work done. Application of EVA will be further discussed at the special Earned Value Management track and round table.
Trend Analysis
Management decisions shall be based on trend analysis. If the project schedule is 10 days ahead of the baseline it could be good if one month ago it was 5 days behind the baseline, or bad if one month ago it was 25 days ahead of the baseline. Project managers shall fight poor trends before they will cause serious problems. Analyzing trends of major project parameters and applying corrective actions to improve negative trends is the best way to prevent small problems to become big.
Risk Simulation and Traditional Performance Analysis
Risk simulation and setting targets that do not belong to the current schedule create problems with the application of traditional methods of performance analysis. First of all: project targets do not belong to any schedule and thus there are the problems with setting some baseline schedule. Creating project time and cost buffers means that PV at any moment is uncertain. It is necessary to estimate project performance by analyzing buffer penetrations. Probabilities to achieve project targets (Success Probabilities) show if buffers are utilized faster of slower than expected. If the probability to achieve some target is growing then project buffer is penetrated slower than expected that is fine.
Success Probability Trend Analysis
Looking at success probability trends we not only analyze past performance but also look to the future. The performance could be perfect but if new risks were identified and require additional contingency reserves that were not included in the initial project buffers then success probability will go down. Success probability trends are integrated project performance indicators that can be used for high level reporting and management decisions. They integrate scope, schedule, cost, and risk performance analysis and are the best project performance analysis tools.
Example of Performance Analysis Diagrams 12
10
8 Column 1 Column 2 Column 3
6
4
2
0 Row 1
Row 2
Row 3
Row 4
Some Conclusions
There are four main methods of performance analysis: Variance Analysis, Earned Value Analysis, Trend Analysis and Success Probability Trend Analysis. VA и EVA analyze project status comparing current data with the baseline or some other project version. TA and SPTA analyze not only statuses but also tendencies. VA, EVA and TA were developed and are used for analyzing deterministic schedules. SPTA is used with probabilistic analysis.
Some Conclusions
EVA is widely used but shall be applied very carefully and together with other methods because it may provide wrong motivation of project teams and does not consider activity network dependencies. Trends shall be used for timely management decisions because they show problems ASAP. SPTA is the best performance analysis method integrating scope, cost, schedule and risk information.
Program Management Office
Corporate Project Management System requires the usage of the common approaches, tools, corporate norms, coding system, cost components and cost centers, risk register and risk management approaches, compliance with the corporate requirements to schedule models, to contract conditions, to reporting and change management, etc. Without common approach to project management within the organization program and portfolio management cannot be effective. Implementation, development and maintaining of the corporate Project Management System is main function of the Program Management Office (PMO).
Program Management Office
Main functions of the PMO include:
1) development, implementation and management of common PM methodology, processes, knowledge bases and archives, 2) Development and maintaining of the corporate databases (like unit costs, production rates, material requirements per work volume unit, etc.) 3) Computer based support of project management processes, 4) Portfolio (program) management 5) Audits of project management process, consulting and training of project personnel.