Académique Documents
Professionnel Documents
Culture Documents
Vladimir Liberzon, PMP Peter S. Mello, PMP, PMI-SP Spider Project Team
History
Success Driven Project Management (SDPM) methodology was developed in Russia in 90-s and since then was successfully used in many projects, programs, and organizations SDPM is supported by Russian PM software Spider Project but its basic approaches can be used with other PM software tools
History
Success Driven Project Management methodology has some common features with Critical Chain Project Management approaches but many differences too
Application of SDPM approaches showed very good results and the number of companies that implement SDPM is growing very fast
SDPM Ideas
Triple constraint and multiple project success criteria make project management too complicated.
The same project schedule and budget targets set for all project stakeholders leads to project failure.
There is a need for setting different targets for project workforce, for project management team, and for project sponsor.
SDPM Ideas
Target for Project Team Member: Project schedule and budget shall be optimistic (no reserves included);
Target for Project Management Team: Includes contingency reserves (scope, time, cost);
Target for Project Sponsor: Project schedule and budget shall include management reserves for unknown unknowns.
SDPM Ideas
Cost Time
SDPM Ideas
Cost Time
SDPM Ideas
Project management team shall have time and cost buffers for managing project risks and uncertainties. These buffers are not linked with any activity sequence.
Project buffer is a difference between target value and the value for the same parameter in the optimistic schedule.
Cost buffer
Cost
Management Team
Time buffer
Management Team
Time
SDPM Ideas
Targets shall be set using risk simulation. These targets shall have reasonable probabilities to be met. Risk simulation shall calculate necessary project cost and time buffers.
Cost buffer
Cost
Management Team
Time buffer
Management Team
Time
SDPM Ideas
Project status information is useful but not sufficient for decision making.
Decision making shall be based on the analysis of project trends.
SDPM Ideas
There is a need to have tools for measuring project buffer penetrations and project performance analysis. The best indicator of buffer penetration and project performance status is current probability to meet project target. If the probability to meet project target is rising then project buffer was consumed slower than we expected, in other case project buffer was consumed too fast and project success is endangered. Success probability trends are the best integrated performance indicators they take into consideration project risks, they depend not only on performance results but also on the project environment.
SDPM
SDPM methodology also includes approaches to creating project schedule models and organizing project data. In this presentation we will discuss: o Resource Constrained Scheduling and Resource Critical Path, o Risk Simulation methods and objectives, o Setting right project goals, o How to set and to manage Project Buffers, o Success Probabilities, o Management by Trends
Scheduling Tasks
Calculation of feasible activity floats and those activities that are critical;
The problem of project schedule development without allowing for resource constraints has a correct mathematical solution (Critical Path Method), which would be the same for all PM packages, provided that initial data are identical.
Other tasks are solved using different approaches and yielding different results.
Resource constrained schedules produced by different PM software are different. The software that calculates shortest resource constrained schedules may save a fortune to its users.
Traditional notion of Critical Path works only in case of unlimited resources availability. Let us consider a simple project consisting of five activities, presented at the next slide. Activities 2 and 5 are performed by the same resource.
Please pay attention to activities that became critical. Now delaying each of the activities 1, 2 and 5 will delay the project finish date. We call these activities Resource Critical and their sequence comprises Resource Critical Path.
For many projects it is necessary to simulate financing and production, and to calculate project schedules taking into account all limitations (including availability of renewable resources, material supply and financing schedules);
True critical path should account for all schedule constraints including resource and financial limitations;
We call it Resource Critical Path (RCP) to distinguish it from the traditional interpretation of the critical path definition.
The calculation of RCP is similar to the calculation of the traditional critical path with the exception that both early and late dates (and corresponding activity floats) are calculated during forward and backward resource (and material, and cost) leveling; This technique permits to determine feasible resource constrained floats; Activity resource constrained float shows the period for which activity execution may be delayed within the current schedule and with the set of resources available in this project without delaying project finish.
As you may notice in our example, Resource Critical Path may include activities that are not linked with logical dependencies. Resource Critical Path is actually not the path but the longest sequence of activities in the current schedule. One activity may depend on another because these activities are performed by the same resources. We call these dependencies as Resource dependencies. Resource dependencies may be shown in the project schedule with the dotted arrows but they are the result of the project leveling and not initial information like logical dependencies.
2 Success Criteria
If project success criteria are set as finishing project on time and under budget then proper decision making will be complicated;
For example, project managers will not be able to estimate if their decisions to spend more money and finish the project earlier are reasonable;
Construction of roads, power plants, bridges, ports, telecommunication networks, new product development and production etc. brings economic results and generate future profits; Implementation of the corporate information system will improve organization processes, etc;
In any case the delay of project finish date usually increases project indirect cost, and acceleration means saving some money.
So each day of project delay means some money losses and finishing project earlier means additional profit;
We can define cost of a project day (maybe separate and different for acceleration and delay) estimating these profits and losses;
This way we define the rules of the game that is called Project Management
Another option to set the profit that should be achieved at some point in time basing on the forecast of the revenues that will be obtained after the project will deliver its results;
Such success criteria will permit to weight time and money making managerial decisions.
At the next slide you may see the project schedule that is calculated without allowing for project financing and supply restrictions. There is a period when project has no money to proceed.
If project manager finds enough money then project total profit to the target date will be close to $245,695.
If we calculate project resource, supply and cost constrained schedule then it become clear that the project will loose more than $17,000 due to necessary delays.
To be able to weight options and to select the best it is necessary to consider not only expenses but also future profits.
Proper project (program, portfolio) schedule model is the powerful tool that helps to select the best decisions.
Our experience in project planning shows that the probability of successful implementation of deterministic project schedules and budgets is very low. Therefore project and portfolio planning technology should always include risk simulation to produce reliable results.
Risk Simulation
Risk simulation may be based on Monte Carlo simulation or use three scenarios approach. We prefer 3 scenario approach for the reasons explained further.
Accuracy: Precision:
Monte Carlo means Accuracy but lack of Precision. 3 Scenarios means Precision but lack of Accuracy. The choice depends on management approach. Our approach may be called Management by Trends. Applying Trend Analysis we rely on data precision. We think that trends supply management with most valuable information on project performance. We think that trend analysis helps to discover performance problems ASAP and to apply corrective actions if necessary.
It is the main reason why in our software 3 scenarios approach was selected. We think that the quality of initial data for project risk simulation is never good enough but Monte Carlo risk simulation creates an impression of accuracy that is actually dangerous for project managers. In any case we need Optimistic schedule and budget for project performance management. And we need to understand what happens with success probability during project execution and so we need data precision.
Risk Simulation
three scenarios approach
A project planner obtains three estimates (optimistic, most probable and pessimistic) for all initial project data (duration, volumes, productivity, calendars, costs, etc.) and creates optimistic, most probable and pessimistic scenarios of project performance Risk events are selected and ranked using the usual approach to risk qualitative analysis Usually we recommend to include risk events with the probability exceeding 90% in the optimistic scenario, exceeding 50% in the most probable scenario, and all selected risks in the pessimistic scenario
Risk Simulation
three scenarios approach
Most probable and pessimistic project scenarios may contain additional activities and costs due to corresponding risk events and may employ additional resources and different calendars. As the result project planner obtains three expected finish dates, costs and material consumptions for all project phases and the project as a whole. They are used to rebuild probability curves for the dates, costs and material requirements.
Risk Simulation
three scenarios approach
If probability curve is known the required probability to meet project target defines the target that shall be set. The area under the probability curve to the left of the target value determines the probability to meet the target.
P=S(blue)/S(whole)
Project/Program Targets
Target dates of most projects usually are predefined. They may be set not only for the whole program/project but also for its major phases. Project planning includes determining how to organize project/program execution to be able to meet required target dates with the reasonable probability.
Success Probabilities
Probabilities to meet approved project targets we call Success Probabilities. These targets may be set for all project parameters that will be controlled (profit, expenses, duration, material consumption). Target dates do not belong to any schedule. Usually they are between most probable and pessimistic dates. A set of target dates and costs for project phases (analogue of milestone schedule) is the real project baseline. But baseline schedule does not exist!
It means that application of usual project performance measurement approaches (like Earned Value Analysis) is complicated. Without certain schedule and cost baselines it is impossible to calculate Planned and Earned Value. If we select some schedule (Optimistic or Most Probable) as the project management baseline the values of Performance Indices that are lower than 1 do not mean that project performance is worse than expected.
Buffers
We recommend to use optimistic schedule for setting tasks for project work force and manage project reserves. The schedule that is calculated backward from the target dates with most probable estimates of activity durations we call Critical schedule. The difference between start and finish dates in current and critical schedules we call start and finish time buffers (contingency reserves). The difference between activity (phase) cost that has defined probability to be met and optimistic cost of the same activity (phase) we call cost buffer.
There are time, cost and material buffers that show contingency reserves not only for a project as a whole (analogue of Critical Chain project buffer) but also for any activity in the optimistic project schedule.
Performance Measurement
Performance measurement routine shall be set for all projects belonging to the portfolio/program. Portfolio/Program schedule is revised regularly. For most large programs it is done weekly. To be able to reschedule the portfolio/program it is necessary that all projects belonging to the portfolio/program have the same data date. So the portfolio/program management team requires from all project management teams to enter actual data of their projects at specified dates and time (for an example: each week on Tuesday before 12:00 the actual status on Tuesday 08:00 shall be entered).
Performance Measurement
If different projects have different data dates then program scheduling became impossible and most reports will not be reliable. So setting the rules for entering actual data is mandatory for program/portfolio management. Project/Program/Portfolio planners shall keep performance archives to be able to get trends of project/program/portfolio parameters.
Management by Trends
We recommend to manage projects and portfolios basing on the analysis of performance trends. If some project is 5 days ahead of the baseline but one week ago it was 8 days and one month ago 20 days ahead of the baseline then some corrective action shall be considered. If the project is behind the schedule but the distance become smaller then project team improved project performance process and corrective actions are not necessary.
Management by Trends
So trend analysis shows short term performance results and helps to make timely management decisions. Usually project management team analyses trends of main project parameters like duration, cost, profit.
Earned Value Analysis is another method that is used for estimating program/project performance. But this method shall be used very carefully and only in combination with other methods because:
the real situation may be distorted, project managers are motivated to do expensive jobs ASAP and cheap jobs ALAP, it does not consider if activities that were performed were critical or not, it does not consider project risks.
We consider success probability trends as the really integrated project performance measurement tool. Success probabilities may change due to:
Performance results Scope changes Cost changes Risk changes Resource changes
Thus success probability trends reflect not only project performance results but also what is going on around the
Success probability is a measure of buffer penetration. If in the middle of the project half of project buffer was consumed it does not mean that the project is performed as expected. If most risks were behind then success probability will become higher and it will tell us that project buffer consumption was lower than expected, if success probability went down then buffer consumption is too high and it is necessary to consider corrective actions.
Success probability trends may be used as the only information about project performance at the top management level because this information is sufficient for performance estimation and decision making. We call Management by Trends methodology Success Driven Project Management.
5 Conclusions
Multiple project/portfolio success criteria make management decisions complicated and subjective.
1.
We recommend to set one integrated project success criterion for decision making and performance analysis.
1.
We recommend to set reliable target dates and costs basing on risk analysis and simulation, but to use optimistic project schedule for setting tasks for project work force.
Time and Cost contingency buffer penetrations shall be regularly re-estimated. If these buffers are consumed too fast there is a need for corrective actions;
5.
5.
If trends are negative corrective actions shall be considered even if the status is good;
Earned Value Analysis supplies management with the useful information on project status. But it shall be used carefully and only as the supplement to other methods of project performance measurement. In any case it is useful to analyze EV parameters trends;
8.
Success Probability trends are the best integrated indicators of project health.
Positive trends show that recent project performance is better than expected. Negative trends show that buffers are consumed faster than expected and corrective actions may be necessary.
Success probability trends depend not only on project performance but also on risks that may change during project life cycle.
Success Probability Trends may be considered as the best integrated project performance indicators necessary for proper decision making.
APPENDIX 1
59
RCP and CC
Resource Critical Path is the same as Critical Chain except suggested approach to determining Critical Chain does not include taking into account supply and financial constraints. We never met the clear description of CC calculation algorithms.
60
Drum Resource
CCPM suggests to find Drum resource and create the schedule and determine Critical Chain scheduling activities of this Drum resource. SDPM does not look for drum resource. At different project phases different resources may be critical, Resource Critical Path is calculated applying traditional resource leveling.
61
SDPM suggests to use Optimistic estimates for project work force management. CCPM suggests to use most probable estimates.
Using most probable estimates still means that there are some reserves for usual problems and these reserves will be lost in any case (Parkinson Law): WORK EXPANDS SO AS TO FILL THE TIME AVAILABLE FOR ITS COMPLETION
62
Project Buffers
CCPM suggests to estimate excessive contingency reserves that people added to most probable activity duration estimates, take them away, summarize and put in a dummy activity that is called Project Buffer and follows the last activity of Critical Chain. SDPM uses risk simulation for setting reliable targets and project time buffer is the difference between accepted project schedule finish and target finish. Project time buffer does not belong to any chain. Besides, SDPM also works with Project Cost Buffers and Material Buffers.
63
Feeding Buffers
CCPM suggests to create feeding buffers on activities that precede Critical Chain activities to protect Critical Chain. CCPM proposes that Critical Chain shall never change. SDPM does not protect any chain project schedule is regularly recalculated and risks analyzed. Besides Resource Critical Paths in optimistic, most probable, and pessimistic schedules may be different. Change of RCP during project execution is usual.
64
Multitasking
CCPM suggests to pipeline projects in the portfolio (to perform them one after another) to avoid multitasking. SDPM suggests almost the same always apply priorities to the portfolio projects calculating portfolio schedule. But if resources are available they may be used in the projects with lower priorities. Besides there are special cases when multitasking is useful. An example is in the next slide.
65
Multitasking
With multitasking:
Without multitasking:
66
CCPM does not suggest reliable quantitative methods for analyzing buffer penetration. Dividing buffer into three zones (green, yellow, red) is one of them. Entering yellow zone means an alert, red requires corrective actions. SDPM estimates buffer penetration by success probability trends. If the trend is negative then project buffer is consumed faster than expected. If in the middle of the project performance project buffer is half consumed it may mean excellent performance if most risks are behind and poor performance if most risks are ahead.
67
SDPM is fully supported by Russian PM software Spider Project but it does not mean that SDPM cannot be applied with other PM software. We will show that it is easy with any PM software though more complicated than with Spider Project.
We will start with discussing simplified calculation of the probability to meet project target. Lets suppose that we got optimistic, most probable and pessimistic values of some project parameter. It does not matter much what shape of probability curve will be selected. Remember that our initial data are not perfect and we plan to manage trends. So lets suppose that the distribution is triangular.
Lets restore the triangle distribution and calculate the probability to meet some target T. It is an area to the left of T divided by the whole area of the triangle. It is not hard to get the formula shown below where X is the probability to meet target T.
Now lets look at the simple sample project where we defined optimistic, most probable and pessimistic activity durations using PERT option in MS Project 2007. Applying PERT weights 6, 0, 0 we will make expected schedule the same as optimistic, with weights 0, 0, 6 expected schedule will become pessimistic.
Lets enter total project durations and formula for calculating target probability into Excel. You may see how Excel calculated probabilities to meet different target dates.
Optimistic schedule shall be used for setting tasks for project workforce.
During project performance planned project duration in optimistic, probable, and pessimistic scenarios will change. Entering these data in Excel we will be able to calculate current values of success probability.
When probability value reach 1 we may be almost sure that target date will be met. Probability trend shows when we had problems and when they were solved. Probability
12 10 8 6 4 2 0