the application of success probabilities, success driven...
TRANSCRIPT
-
The Application of Success Probabilities, Success Driven Project Management, and Some Critical Chain ConceptsTo the Oil & Gas Industry in Brazil
Russell D. Archibald, USAVladimir Liberzon, Russia
Peter Berndt de Souza Mello, Brazil
PMI COS 5th Conference, May 4-7 2008, Chicago USA 1
-
Presentation Outline
The Need to Manage UncertaintiesSuccess Driven Project Management/SDPM MethodologyUnique Features of SDPMComparison with Critical ChainMonte Carlo versus using 3 point estimatesExperience to date in BrazilConclusions
PMI COS 5th Conference, May 4-7 2008, Chicago USA 2
-
THE NEED FOR MANAGING UNCERTAINTIES, RISKS, & RESOURCE
RESTRICTIONS IN PROJECTS
These occur in all complex projectsThe need is to identify, monitor and control risks and resourcesThey must be reflected in plans & schedulesMethods to accomplish this developed in Russia are now being applied in Brazil (and 21 other countries
PMI COS 5th Conference, May 4-7 2008, Chicago USA 3
-
Success Driven Project Management(SDPM)
Planning Stage:
Calculate finish dates & costs with the required
probabilities of their successful achievement
Determine the Resource Critical Path/RCP
Set target dates, costs & other restrictions
Calculate success probabilities
Determine contingency reservesPMI COS 5th Conference, May 4-7 2008, Chicago USA 4
-
Success Driven Project Management (SDPM)
Execution and Control:
Calculate current probabilities of achieving goals
Track success probability trends
Manage contingency reserves
PMI COS 5th Conference, May 4-7 2008, Chicago USA 5
-
Resource Critical Path/RCP
True (resource) critical path must reflect ALL schedule constraints: resource, finance, supply, calendar, & imposed dates
All constraints must be considered in both forward & backward passes
PMI COS 5th Conference, May 4-7 2008, Chicago USA 6
-
Success Probability
PMI COS 5th Conference, May 4-7 2008, Chicago USA 7
1. System forecasts resulting required resources & contingency reserves based on user defined acceptable probability of success to meet specific scope, schedule & cost targets
2. System calculates the probability of meeting imposed targets (success probabilities)
-
Eight Integration Methods Used
PMI COS 5th Conference, May 4-7 2008, Chicago USA 8
1. Systematic scope definition (several indentured structures/WBS)
2. Network planning (CPM/PDM)3. Resources:
• Consumable, renewable, utilized & produced
• Units, teams/crews, interchangeable units or crews
• Assigned to project activities• Constraints in both forward & backward
passes
-
Eight Integration Methods(Cont’d)
4. Activity durations calculated: scope or volume ÷ rate using range estimates with or without Monte Carlo
5. True (resource) critical path calculated:• Logical & schedule constraints• Resource, financial & supply limitations
in both the forward and backward passes
PMI COS 5th Conference, May 4-7 2008, Chicago USA 9
-
Eight Integration Methods (Cont’d)
6. Risk & uncertainties simulated: probability distribution for main project results (project & its main phases finish dates, costs, resource requirements)
7. Actuals reported & compared, contingency reserves tracked
PMI COS 5th Conference, May 4-7 2008, Chicago USA 10
-
Eight Integration Methods(Cont’d)
8. Current probabilities of success calculated and trends determined for:
Schedules
Costs
Resources
PMI COS 5th Conference, May 4-7 2008, Chicago USA 11
-
Methods Unique to SDPM
• Multiple breakdown structures• Resource information & analysis• Activity duration calculated or estimated• Resource critical path, resource floats, &
resource contingency reserves• Risk simulation & success probability analysis• Success probability trends predicted and
tracked
PMI COS 5th Conference, May 4-7 2008, Chicago USA 12
-
SDPM versus CCM
Resource Critical Path:All schedule constraintsActivity & resource assignment reservesTotal floats: forward & backward passesDifferent ‘drum resources’ for various project phases
Critical Chain:One ‘schedule drum resource’ defines the ‘schedule constraint’and one ‘critical chain’Resource-leveled schedule set first, then determine the ‘critical chain’
PMI COS 5th Conference, May 4-7 2008, Chicago USA 13
-
SDPM versus CCM (cont’d)Many potential RCPsand paths with small floats may exist‘Feeding buffers’ not necessary:
Don’t ‘protect’ any chain/pathProject time and resource buffers, yes: ‘contingency reserves’
CC is the only single sequence that shall not change during project execution
‘Feeding buffers’ may postpone planned dates, creating holes in CC
PMI COS 5th Conference, May 4-7 2008, Chicago USA 14
-
SDPM versus CCM (cont’d)Risk Evaluation & Prediction
Monte Carlo or range estimatesPredict user defined success probabilities to meet targetsProject buffersRecommend using optimistic schedule dates
CCPM recommends using 50% probability duration, all extra time at end of CC: chain bufferCC “shall never change”CC says nothing about cost!
PMI COS 5th Conference, May 4-7 2008, Chicago USA 15
-
SDPM versus CCM (cont’d)Progress Evaluation & Decision Making
Project performance and buffer usage monitored by trends in success probabilitiesUp = goodNew risks are included as they occur
Buffer usage estimated qualitativelyApproach is linearNo quantificationNothing on cost
PMI COS 5th Conference, May 4-7 2008, Chicago USA 16
-
SDPM versus CCM (cont’d)Project Portfolio Management
Planned as one large projectProjects in Portfolio may have different priorities for resource allocationSimulate performance using portfolio success prob. trends
Projects in a portfolio shall be pipelined to avoid multi-taskingPriorities not discussed
PMI COS 5th Conference, May 4-7 2008, Chicago USA 17
-
SDPM versus CCM (cont’d)Periodic Schedule & Cost Performance Evaluation
SDPM approach based on regular recalculations of project and portfolio schedules, costs, and success probabilities and their trends
Recalculations shall be minimizedRegular adjustment of schedule is “firefighting”“People are free to finish the work as soon as possible”
PMI COS 5th Conference, May 4-7 2008, Chicago USA 18
-
MC vs Using 3 Estimates:Inputs
Using 3 Estimates:3 estimates made, durations and costsOpt. 10%Most prob.Pess. 90%
Monte Carlo:3 estimates made, prob. distributions assumedRisk events ID’d, prob. EstimatedConditional links may be used
PMI COS 5th Conference, May 4-7 2008, Chicago USA 19
-
MC vs Using 3 Estimates:Process Methods
Project execution simulated many times, activity durations and risk events via random number generatorProb. curve set by number of cases for each duration
3 scenarios calculated with res. leveling & budgetingProb. distr. set via predetermined curveAny reasonable shape may be used
PMI COS 5th Conference, May 4-7 2008, Chicago USA 20
-
MC vs Using 3 Estimates:Completion Date
Project finish date set to match desired probability under the defined distribution curve
The same
PMI COS 5th Conference, May 4-7 2008, Chicago USA 21
-
MC vs Using 3 Estimates:Progress Evaluation
Repeat process periodically, plot success prob. Trends
Rising trend=goodFalling trend=bad
The same
PMI COS 5th Conference, May 4-7 2008, Chicago USA 22
-
MC vs Using 3 Estimates:
Pro
More accurateCan show complex curves with more than one max
Fast and preciseDistribution may not be correct but results are repeatable, with consistent errorsSuccess prob. trend analysis is accurate
PMI COS 5th Conference, May 4-7 2008, Chicago USA 23
-
MC vs Using 3 Estimates: Cons
Time consumingIf activities are performed by same resources, results are not accurateCorrective actions are not reflected in the simulations; results not accurate
Estimations are less accurateMerge bias: see next slides
PMI COS 5th Conference, May 4-7 2008, Chicago USA 24
-
MC vs Using 3 Estimates:Using a Summary Model
For large projects people sometimes use a summary networkThis loses links between activities, cannot simulate resources properly, produces low accuracy
Summary networks not required
PMI COS 5th Conference, May 4-7 2008, Chicago USA 25
-
MC vs Using 3 Estimates:Merge Bias Question
Exists when using summary modelsMany paths merging at the endMerge bias within summary tasks not consideredCorrective actions cannot be modeled
Merge bias inside the project is ignoredMerge bias within a portfolio is consideredNot important: success probability trends are useful and accurate
PMI COS 5th Conference, May 4-7 2008, Chicago USA 26
-
The Russian Approach to Project Planning
Soviet Union rootsUniversal coding system“Scientific Work Organization”Use of norms and standardsEmphasis on resource planning and schedule optimization
PMI COS 5th Conference, May 4-7 2008, Chicago USA 27
-
Experience to Date:SDPM & RCP in Brazil
First known application of SDPM happened at Urucu/Manaus Pipeline in October 2006
The project had started in 2004 and was facing major delays due to a segmented view of the project, with several separated set of schedules for different regions (geographic division) and different areas (scope, costs, logistics and supplies).
PMI COS 5th Conference, May 4-7 2008, Chicago USA 28
-
Experience to Date:SDPM & RCP in Brazil
First known application of SDPM happened at Urucu/Manaus Pipeline in October 2006
An integrated schedule and the use of simulations with resource restrictions helped Petrobras to criticize and complement new contractors’ plans for project recovery, with significant gains in productivity.Subprojects started in 2007 included requests to new contractors to apply lessons learned with the use of RCP [PMI Global 2007, Cancún]
PMI COS 5th Conference, May 4-7 2008, Chicago USA 29
-
Slowing Down to Speed Up !!
What is expected with the adoption of Success Driven Project Management is the ability to identify what phases of the project should be delayed to make critical resources available to other critical phases, thus expanding the productivity in areas of greater working dependencies with other teams.
PMI COS 5th Conference, May 4-7 2008, Chicago USA 30
-
Fragnets
Fragnets are modeled phases of the project that are carefully planned to be applied to the whole project.One example is the modeling of “one km of pipeline”
PMI COS 5th Conference, May 4-7 2008, Chicago USA 31
-
Fragnets
By using automated scheduling optimization and simulation, activities in each fragnet are reorganized to the best sequence to increase global production.
PMI COS 5th Conference, May 4-7 2008, Chicago USA 32
-
Fragnets
Schedule optimization
PMI COS 5th Conference, May 4-7 2008, Chicago USA 33
-
Fragnets
PMI COS 5th Conference, May 4-7 2008, Chicago USA 34
800 meters per day800 meters per day
Slow down
Speed up
1150 meters per day1150 meters per day
-
Scheduling optimizationin a few words
Almost any network of activities can be improved if we simply try the alternatives.
Jane takes 4 days to write a quality plan and 2 days to write a communication plan for the project.
If both plans are reviewed by her boss, who spends 3 days for each document, how long does this project take ?
PMI COS 5th Conference, May 4-7 2008, Chicago USA 35
-
PMI COS 5th Conference, May 4-7 2008, Chicago USA 36
-
9 DAYS
PMI COS 5th Conference, May 4-7 2008, Chicago USA 37
10 DAYS
-
SDPM is expanding in Brazil
STD is an engineering company developing several projects around Brazil for oil & gas. Starting October 2007, their project managers have been receiving training in PM based on PMI best practices and SDPM application concepts.
A successful five day planning project for a hot-swap in a gas pressure reduction city-gate proved the importance of simulating schedules with corrective actions (optimistic x probable x pessimistic). After that training, all their new projects undergo a new planning phase with detailed scheduling never applied in their projects before.
PMI COS 5th Conference, May 4-7 2008, Chicago USA 38
-
SDPM is expanding in Brazil
Three new projects have started in January 2008. The average size of each of these projects is twice the size of the projects they had during 2006/2007.
There are a total of 22 on-going projects with a total value of R$ 35 million (US$20.7 million) (61% of the company’s portfolio)
The three new projects surpass a total value ofR$ 22 million (US$13 million) (39% of the company’s portfolio)
PMI COS 5th Conference, May 4-7 2008, Chicago USA 39
-
SDPM is expanding in Brazil
Several shared resources are still the same as the ones available for previous projects, but scheduling based in the Eight Integration Steps of SDPM has permitted portfolio resource planning with significant cost reduction.
All new projects are planned under SDPM recommendation
All on-going projects are being re-evaluated to guarantee portfolio resource sharing
PMI COS 5th Conference, May 4-7 2008, Chicago USA 40
-
SDPM is expanding in Brazil
Before SDPM: Schedules had an average of 5 levels of detailing in the WBS and less than 2,000 tasks.
After SDPM: Schedules have multiple WBS, sometimes with over 9 levels of detailing and accurate resource planning.
A new “project interface phase” was introduced to reflect activities under the responsibility of the customer and subcontractors (to guarantee immediate analysis of project deviations due to external dependencies).
PMI COS 5th Conference, May 4-7 2008, Chicago USA 41
-
SDPM is expanding in Brazil
After SDPM:Acquisition of parts are planned using fragnets with estimates for early and late deliveries.
A “top-management summary schedule” with an average of 30-50 control points is used by the PMO and customer to follow projectprogress.
Detailed planning now surpasses 15,000 activities
Activities from many functional areas are now included in the planning for resource critical path analysis.
PMI COS 5th Conference, May 4-7 2008, Chicago USA 42
-
Management x Detailed Scheduling (use of hammocks)
PMI COS 5th Conference, May 4-7 2008, Chicago USA 43
-
Three point estimates
PMI COS 5th Conference, May 4-7 2008, Chicago USA 44
Optimistic applied Pessimistic applied
Optimistic Planning
Pessimistic Planning
-
Most probable with Critical Schedule Calculated
PMI COS 5th Conference, May 4-7 2008, Chicago USA 45
-
Success Probability
PMI COS 5th Conference, May 4-7 2008, Chicago USA
TipicalSchedule The real world
Bufferprotection
46
-
Success Probability – The paradox
As Project Managers, we want all the time and resources made available (pessimistic readiness)
Over-protected projects aren’t cost effective and we cannot afford having results behind our competitors.Unprotected projects (single estimate) will simply take longer and be more expensive than planned, leading us to unpredictable results.
PMI COS 5th Conference, May 4-7 2008, Chicago USA 47
-
Success Probability – The paradox
PMI COS 5th Conference, May 4-7 2008, Chicago USA 48
As Project Managers, we want all the time and resources made available (pessimistic readiness)
Over protected projects aren’t cost effective and we cannot afford having results after our competitors.Unprotected projects (single estimate) will simply take longer and be more expensive than planned.
SDPM will create calculated critical schedules that willestablish the necessary balance between
cost & time effectiveness and the expected successprobability, according to goals agreed with stakeholders.
-
Conclusions
1. Effective project planning and scheduling requires the identification of the Resource Critical Path (as defined in this paper)
2. Risk management and success probability trend analysis must be integrated into daily work of project planning, scheduling, and controlling
PMI COS 5th Conference, May 4-7 2008, Chicago USA 49
-
Conclusions (Cont’d)
3. PM maturity must be measured for PM processes and also available data banks
4. Progress evaluation through success prob. trend analysis is effective using 3 estimate approach; merge bias is not important
PMI COS 5th Conference, May 4-7 2008, Chicago USA 50
-
Conclusions (Cont’d)
5. Monte Carlo approach has practical limitations and unknown inaccuracies when applied to summary networks
6. RCP and SDPM are proving to be practical and effective in the field in Brazil
7. Western PM software vendors should include resource constraints in both forward and backward passes for network analysis
PMI COS 5th Conference, May 4-7 2008, Chicago USA 51
-
A Few Useful Links
http://www2.petrobras.com.br/ingles/index.asphttp://www.spiderproject.com.br [go to ‘seminars’]http://www.spiderproject.ru/aboutus_e.phphttp://www.russarchibald.com/ [go to ‘author>recent papers’] to download this present paper plus:
Liberzon, Vladimir, and Russell D. Archibald, “From Russia with Love: Truly Integrated Project Scope, Schedule, Resource and Risk Information,” PMI World Congress- The Hague, May 24-26, 2003;Archibald, Russell D., Peter Berndt de Souza Mello, & Jefferson Guimarães, “The Application of SDPM, Critical Chain and Portfolio Project Management Principles to the Construction of the 670 km Urucu/Manaus (Petrobras) Pipeline,” PMI World Congress-Latin America, Cancun, MX, Nov. 12-14 2007
PMI COS 5th Conference, May 4-7 2008, Chicago USA 52
http://www2.petrobras.com.br/ingles/index.asphttp://www.spiderproject.com.br/http://www.spiderproject.ru/aboutus_e.phphttp://www.russarchibald.com/
-
Russell ArchibaldPrincipal - Archibald Associates, USA/[email protected]
Vladimir LiberzonGeneral Director, Spider Project Team, Russia"Vladimir Liberzon"
Peter Berndt de Souza MelloDirector - X25 Treinamento e Consultoria, [email protected]
-
PMI COS 5th Conference, May 4-7 2008, Chicago USA 54
The Application of Success Probabilities, Success Driven Project Management, and Some Critical Chain ConceptsTo the Oil & GasPresentation OutlineTHE NEED FOR MANAGING UNCERTAINTIES, RISKS, & RESOURCE RESTRICTIONS IN PROJECTSSuccess Driven Project Management (SDPM)Success Driven Project Management (SDPM)Resource Critical Path/RCPSuccess ProbabilityEight Integration Methods UsedEight Integration Methods (Cont’d)Eight Integration Methods (Cont’d)Eight Integration Methods (Cont’d)Methods Unique to SDPMSDPM versus CCMSDPM versus CCM (cont’d)SDPM versus CCM (cont’d) Risk Evaluation & PredictionSDPM versus CCM (cont’d) Progress Evaluation & Decision MakingSDPM versus CCM (cont’d) Project Portfolio ManagementSDPM versus CCM (cont’d) Periodic Schedule & Cost Performance EvaluationMC vs Using 3 Estimates: InputsMC vs Using 3 Estimates: Process MethodsMC vs Using 3 Estimates: Completion DateMC vs Using 3 Estimates: Progress EvaluationMC vs Using 3 Estimates: ProMC vs Using 3 Estimates: ConsMC vs Using 3 Estimates: Using a Summary ModelMC vs Using 3 Estimates: Merge Bias QuestionThe Russian Approach to Project PlanningExperience to Date:SDPM & RCP in BrazilExperience to Date:SDPM & RCP in BrazilSlowing Down to Speed Up !!FragnetsFragnetsFragnetsFragnetsScheduling optimizationin a few wordsSDPM is expanding in BrazilSDPM is expanding in BrazilSDPM is expanding in BrazilSDPM is expanding in BrazilSDPM is expanding in BrazilConclusionsConclusions (Cont’d)Conclusions (Cont’d)A Few Useful Links