ESD.36 System Project Management + - Lecture 12 Strategic Project Management Instructor(s) Dr. James Lyneis October 18, 2012 + - Today’s Agenda • Strategic Project Management • Example 1: Project Preparation • Example 2: Project Planning • Example 3: Project Execution 2 + - System Project Management ESD.36 Framework Project Preparation Doing the Right Job Project Planning Project Monitoring Project Control Doing the Enterprise has chosen what product or system to develop Next Project Job Right Project Adaptation Project Completion Project Learning 3 + - What is corporate strategy as it applies to projects and the project portfolio, versus “strategy” as it applies to an individual project ? 4 + Corporate Strategy for the Project Determining the fit of the project to business objectives (the “mission” – doing the right job) features / scope of end product schedule milestones (time to market) delivered quality (defects) resources & budget (development cost) And the mix/timing of “projects” necessary to achieve corporate strategy Operationally, “projects” implement corporate strategy. 5 + Strategic Project Management Understanding how project “design” decisions affect project performance … Scope/schedule/ ... (i.e., mission feasibility) Organization, process, ... Buffers, phase overlap, ... Staffing strategies, schedule slip, ... ... … and how they affect other current projects (portfolio issues), and future projects. Learning from past projects. Operationally, “day-to-day project decisions” implement project strategy. 6 + - Example: Strategic/Tactical vs Operational Staffing Decisions Strategic/Tactical Hire experienced staff rather than inexperienced Start with all of staff you need or gradually build How much training for inexperienced staff …. Operational Who specifically and with what experience How many, and/or at what ramp up When, what programs, etc. 7 + - System Project Management ESD.36 Framework Project Preparation Doing the Right Job Project Planning Strategic Enterprise has Project chosen what product or system to Management develop Corporate Strategy Next Project Project Monitoring Project Control Doing the Job Right Project Adaptation Project Completion Project Learning 8 + - DISCUSSION? 9 + What is SD useful for? Conceptualization of project dynamics and the issues/tradeoffs involved in strategic management of projects Quantification of above … Heuristics Specific forecasts and decision guidance Project-to-project learning 10 + SD Qualitative Insights -1 - 1. A feasible plan is essential, including: Estimates of rework, undiscovered rework, and delays in discovering that rework Estimates of productivity loss dealing with rework Adequate buffers and reserves for rework [Rework increases with project uncertainty and complexity] 11 + - SD Qualitative Insights – 2 2. A feasible plan recognizes the “iron triangle”; there will be multiple “feasible” plans depending on priorities. 3. Tradeoffs in the plan can often be improved by changes in project structure and organization to reduce rework and delays in discovering rework. 12 + - SD Qualitative Insights – 3 4. Attempts to achieve an infeasible plan via project control actions lead to “vicious circle” side effects which increase project cost and duration. On complex projects, these costs usually exceed the “direct” costs of infeasibility 5. Project “changes,” and risks which materialize, are fundamentally the same as an infeasible plan. (Lecture 13) 13 + - SD Qualitative Insights – 4 6. Project managers need buffers and/or flexibility (e.g., slip schedule, cut scope, ship with “bugs”) to respond to changes and uncertainties. These have costs that need to be evaluated; the importance of different tradeoffs differs by project. (Lecture 13) 7. The costs of project control can be minimized by understanding the sources of the vicious circles. The timing, magnitude, and duration of different controls affects performance. 14 + - SD Perspective: Typical project dynamics result in schedule &/or budget overrun ... Project Staffing What can we do to avoid/minimize the dynamics ... … in project preparation and planning? … in project execution and adaptation? Typical Plan Time 15 + How Does It Get Started? - Other Risks Effort Applied Productivity + Scope Growth Original Work to Do Fraction Correct and Complete - Progress Rework Generation Progressx Rew ork Work Done Infeasible Plan Rework to Do These are characteristics of “complex” (vs. “simple”) projects + Uncertainty & Complexity Rework Discovery - Undiscovered Rework Changes + Time to Discover Rework 16 + Example Project Scope = 1000 Tasks Scheduled Completion Date = 30 (Month) Staff = 40 (Implied budget of 1200 person- months, including 200 tasks estimated rework) Normal Quality = 0.85 Productivity = 1 task/month/person Note: Infeasible Plan 17 + - Project Behavior Cost = 1570 person-months, Finish 39.25 Staff & Progress 2,000 Tasks 100 People Total Tasks = 1570 Work Done Staff 1,000 Tasks 50 People 0 Tasks 0 People 0 6 12 18 24 30 36 Time (Month) Work Done : Variable Fraction Correct Cumulative Work Done : Variable Fraction Correct Staff : Variable Fraction Correct 42 How do we change & manage the project60to 48 54 improve its Tasks performance? Tasks People 18 + - Today’s Agenda • Strategic Project Management • Example 1: Project Preparation Developing a Consistent Plan • Example 2: Project Planning • Example 3: Project Execution 19 + - A Consistent (Feasible) Project Avoids the Dynamics “SD Class 3” Model With: Scope = 1000 (tasks) Scheduled Completion Date = 35 (month) [versus 30 in Class 3 model] Delivered Quality > 99% Normal Fraction Correct = 0.85 Staff = 50 (people) [Versus 40 staff ; Implying a budget of 1750 person-months, versus 1200 person-months] Estimated Rework = 750 tasks [versus 200] 20 A Consistent Project Avoids the Dynamics + - Staff & Progress 60 People 1,000 Tasks Basic Behaviour Work Done Staff 1,000 Tasks 1,000 Tasks 1,000 Tasks To Do Done 500 Tasks 500 Tasks 500 Tasks 30 People 500 Tasks Undiscovered Rework 0 Tasks 0 Tasks 0 Tasks 0 People 0 Tasks 0 6 12 18 Staff for Output : SD4 Feasible Plan1 Work Done : SD4 Feasible Plan1 24 30 36 Time (Month) 42 48 54 60 People Tasks 0 6 12 18 24 30 36 Time 42 Work to Do : SD4 Feasible Plan1 Work Done : SD4 Feasible Plan1 Undiscovered Rework : SD4 Feasible Plan1 48 54 60 Tasks Tasks Tasks Plan fully accounts for rework tasks; Schedule and staffing plan reflect rework cycle 21 Normal design evolution accounted for in plan + - Productivity Fraction Correct and Complete 1 Effect of Experience 1.1 1 Productivity 0.9 Fraction Correct 0.75 Fraction Task/(Person*Month) 1.2 Effect of Experience 0.5 Effect of Undiscovered Rework 0.25 0 0.8 0 6 12 18 24 30 36 Time (Month) Productivity : SD4 Feasible Plan1 Effect of Experience on Productivity : SD4 Feasible Plan1 "Effect of Intensity/Hours on Productivity" : SD4 Feasible Plan1 42 48 54 60 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Fraction Correct and Complete : SD4 Feasible Plan1 Effect of Experience on Fraction Correct : SD4 Feasible Plan1 Effect of Undiscovered Rework on Fraction Correct : SD4 Feasible Plan1 "Effect of Intensity/Hours on Fraction Correct" : SD4 Feasible Plan1 22 - Infeasible projects initiate the dynamics when management responds … Trying to achieve inconsistent objectives can lead to disaster … Staff for Output 100 Infeasible, Control 75 People + 50 25 Feasible Plan Infeasible, No Control 0 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Staff for Output : SD4 Infeasible Plan Control Staff for Output : SD4 Infeasible Plan No Control Staff for Output : SD4 Feasible Plan1 23 + What do we expect? - Other Risks Effort Applied Productivity + Scope Growth Original Work to Do + Progress Project Progressx Staffing Uncertainty & Complexity Fraction Correct and Complete - Rework Generation Work Done Expectation: Rew ork Feasible Plan Infeasible Plan Rework to Do Rework Discovery - Undiscovered Rework Changes Infeasible plan + Time to Discover Rework Time 24 + But when management reacts - +!&! + Effort Applied Productivity + Scope Growth Original Work to Do Quality + Progress Progressx Infeasible Plan Rework to Do + Rework Discovery - People Resources: Overtime Work Intensisty + Experience Dilution - + Experience - Rework Generation Rew ork Changes Undiscovered Rework Overlap & Concurrence - Scope Growth + Productivity + + + Rework Generation Progress - Add People + Hiring Work Intensity Haste Makes Waste + + Work Done Progressx Rework Undiscovered Rework Rework to Do Rework Discovery + Errors Build Errors + Work More - Original Work to Do Errors Create More Work Unknown Errors in Prior Work + Deadline Overtime - + Quality + + Known Work + Remaining - + Time Remaining + -- Haste Creates Out-ofSequence Work + + Time to Discover Rework + + + Effort Needed + + Workforce + Effort Applied Hopelessness Add Resources + Burnout Morale - + Too Big to Manage + Fatigue - Work Done + Congestion & Communication Difficulties Increased Turnover Work Faster or "Slack Off" Effort Needed + - + + + Time Remaining Known Work Remaining + Deadline 25 + Effect of Experience on Fraction Correct 1 0.9 Dimensionless - Trying to achieve infeasible plan … Infeasible, Control Feasible Plan 0.8 Effect of Experience on Fraction Correct 0.7 0.6 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Effect of Experience on Fraction Correct : SD4 Infeasible Plan Control Effect of Experience on Fraction Correct : SD4 Infeasible Plan No Control Effect of Experience on Fraction Correct : SD4 Feasible Plan1 26 - Which snowballs via “errors on errors” feedback … Effect of Undiscovered Rework on Fraction Correct 1 Feasible Plan 0.85 Dimensionless + Infeasible, Control 0.7 Infeasible, No Control 0.55 Effect of “Errors on Errors” 0.4 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Effect of Undiscovered Rework on Fraction Correct : SD4 Infeasible Plan Control Effect of Undiscovered Rework on Fraction Correct : SD4 Infeasible Plan No Control Effect of Undiscovered Rework on Fraction Correct : SD4 Feasible Plan1 27 + - With end result worse (schedule/cost) than if project budgeted higher at start! Test Finish Cost(person-mos) Infeasible Plan Targets 30 1200 Infeasible, No Control Infeasible, with control) Feasible Plan 1 Feasible Plan 2 39.25 36.25 1570 2148 33.75 30.125 1615 1650 Best choice depends on corporate strategy. Note: Feasible Plan 1 (Initial Staff 50, Schedule 35, Budget 1750); Feasible Plan 2 (Initial Staff 60, Schedule 30, Budget 1800) 28 + - The “Iron Triangle” Scope Cost Project Schedule There are alternative feasible plans that reflect project priorities 29 + - Survey Question 1 Does your organization plan for rework in establishing project budgets and baselines? 1. Yes, we explicitly try to estimate the expected amount of rework 2. Yes, but only by adding a “management reserve” 3. No 30 + - Survey Question Do you feel that on the typical project in your organization, budget and schedule are … 1. More than is needed __________ 2. Tight, but manageable __________ 3. Insufficient enough that the vicious circles are significant __________ 31 + - Why Won’t We Develop a Realistic Plan? Then why add resources when situation realized? 32 + Getting a Feasible Plan Use a model Use data from prior projects (learning!), and calibration, to estimate: Normal Productivity Normal Fraction Correct and Complete Time to Discover Rework Total rework and undiscovered rework profile Strength of effects … Include buffers and have a sound project control plan (see example 3) 33 + SD Qualitative Insights Review 1. A feasible plan is essential, including: Estimates of rework, undiscovered rework, and delays in discovering that rework Estimates of productivity loss dealing with rework Adequate buffers and reserves for rework [Rework increases with project uncertainty and complexity] 2. A feasible plan recognizes the “iron triangle”; there will be multiple “feasible” plans depending on priorities. 4. Attempts to achieve an infeasible plan via project control actions lead to “vicious circle” side effects which increase project cost and duration. 34 + - SD Qualitative Insights – 2 2. A feasible plan recognizes the “iron triangle”; there will be multiple “feasible” plans depending on priorities. 3. Tradeoffs in the plan can often be improved by changes in project structure and organization to reduce rework and delays in discovering rework. 35 + - Today’s Agenda • Strategic Project Management • Example 1: Project Preparation • Example 2: Project Planning Deciding on the Process Model • Example 3: Project Execution 36 + What Increases Cost & Schedule? - Uncertainty that reduces fraction complete and correct. Technical complexity Uncertainty about customer requirements 37 + - Strategic Project Planning What changes in process, organization, etc. might help deal with technical or customer uncertainties? Increase planned design iterations? Autonomous (dedicated) integrated product team vs. functional? Waterfall vs. d/b/t iterative vs. spiral vs. …? More phase overlap and concurrency? How do we assess what process model is right for out project? 38 How do we assess what process model is right for our project? + - Determining Impact on Dynamics: 1. Model project with current processes, policies, … 2. Specify direct impacts of alternatives on - Scope (added tasks) Productivity Fraction correct and complete Rework discovery Strength of productivity and FCC effects ... [Secondary impacts assessed via simulation] 3. Simulate and compare performance 4. Test sensitivity to uncertain assumptions 39 Example: Three-Phase Model (from Lecture 7) + - Requirements Design Start Design Rampup Requirements Planned Staff Trigger for CDR Requirements Fraction of Effort to Rework Requirements Staff on Original Work Requirements Indicated Staff Requirements Maximum Work Rate Based on Rework Tasks Available <Initial Requirements Work to Do> Requirements Fraction Correct and Complete Design Startup Requirements Relative Effort Required for Rework Requirements Staff on Rework Requirements Priority to Original Work Requirements Original Work Being Accomplishmed Requirements Productivity Requirements Rework Generation on Rework <Initial Requirements Work to Do> Requirements Rework Being Accomplishmed Design Staff on Original Work Design Indicated Staff Design Maximum Work Rate Based on Rework Tasks Available Design Productivity <Initial Design Work to Do> Design Fraction Correct and Complete Design Original Work to Do Design Rework Generation on Rework Rqmts Fraction of Work Done Correct and Complete Assumptions: <Initial Build/Test Work to Do> Build/Test Original Work to Do Design Work Done Design Undiscovered Rework Design Time to Discover Rework Maximum Effect of Undiscovered Rework on Fraction Correct Scope = 100 Tasks Staff = 6 Productivity = 2 tasks/month/person Duration = 8.33 months (no rework) NFCC = 0.75 <Fraction Design Rework Discovered> <Initial Build/Test Work to Do> Build/Test Fraction Correct and Complete Build/Test Build/Test Rework Work Done Done Correctly Design Fraction of Work Done Correct and Complete <Maximum Effect of Undiscovered Rework on Fraction Correct> Sensitivity of Build/Test Fraction Complete to Rmnts Undiscovered Rework Scope = 1000 tasks Staff = 25 Productivity = 4 tasks/month/person Duration = 10 months (no rework) NFCC=0.7 Build/Test Fraction Reported Complete Project Finished Switch Build/Test Build/Test Rework Undiscovered Generation on Rework Rework Build/Test Rework Discovery Design Work Believed to Be Done Build/Test Rework Being Accomplishmed Build/Test Rework Generation on Original Work Build/Test Rework to Do Build/Test Effect of Design Undiscovered Rework On Fraction Correct Build/Test Productivity Build/Test Original Work Done Correctly Design Rework Discovery <Fraction Rqmts Rework Discovered> Sensitivity of Design Fraction Complete to Rmnts Undiscovered Rework Normal Build/Test Fraction Correct and Complete Design Rework Done Correctly Design Rework to Do Build/Test Productivity on Rework Build/Test Original Work Being Accomplishmed Build/Test Minimum Time to Finish a Task Design Rework Generation on Original Work Design Effect of Rqmnts Undiscovered Rework On Fraction Correct Rqmnts Work Believed to Be Done Design Rework Being Accomplishmed Build/Test Priority to Original Work Build/Test Maximum Work Rate Based on Original Work Tasks Available Design Fraction Reported Complete Design Original Work Done Correctly Normal Design Fraction Correct and Complete Requirements Rework Discovery Build/Test Maximum Work Rate Based on Rework Tasks Available <Initial Design Work to Do> Staff on Rework Build/Test Relative Effort Required for Rework Build/Test Staff on Original Work Build/Test Indicated Staff Design Priority to Original Work Design Productivity on Rework Design Original Work Being Accomplishmed Requirements Work Done Requirements Undiscovered Rework Design Relative Effort Required for Rework Design Maximum Work Rate Based on Original Work Tasks Available Design Minimum Time to Finish a Task Reqmts Time to Discover Rework Build/Test Fraction of Effort to Rework Design Staff on Rework Requirements Fraction Reported Complete Requirements Rework Done Correctly Build/Test Staff Design Fraction of Effort to Rework Requirements Rework Generation on Original Work Requirements Rework to Do Build/Test Planned Staff Build/Test Startup Design Staff Design Planned Staff Requirements Productivity on Rework Requirements Original Work Done Correctly Requirements Original Work to Do <Switch for Phased Startup> Switch for Phased Startup Trigger for SRR Requirements Minimum Time to Finish a Task End Build Rampup Start Build Rampup End Design Rampup Project Staff Requirements Staff Requirements Maximum Work Rate Based on Original Work Tasks Available Build/Test Build/Test Work Believed to Be Done Build/Test Time to Discover Rework <Fraction Build/Test Rework Discovered> Scope = 1000 tasks Staff = 40 Productivity = 1 tasks/month/person Duration = 25 months (no rework) NFCC= 0.95 40 + Rework Discovery Assumptions (similar to CityCar HW#3) 60% of rework discoverable in design One design planned iteration & limited design review Fraction of Rework Discovered in First Iteration = 30% Fraction of Rework Discovered in Later Design Iterations = 70% two iterations, 95% three iterations (note: derivable via DSM and signal flow graph simulation?) Tasks repeated per iteration = 25% Build starts when design is 70% reported complete 41 + Simulation results for current processes … - Work Done No rework finish Project Staff 1,000 80 Design 500 Build/Test 250 12 40 Design 20 0 0 6 Build/Test Requirements Requirements 0 Total 60 People Task 750 18 24 30 36 Time (Month) Requirements Work Done : Three P Four Stock V5 Base Design Work Done : Three P Four Stock V5 Base "Build/Test Work Done" : Three P Four Stock V5 Base 42 48 54 60 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Requirements Staff : Three P Four Stock V5 Base Design Staff : Three P Four Stock V5 Base "Build/Test Staff" : Three P Four Stock V5 Base Project Staff : Three P Four Stock V5 Base Design “done” 42 + - Can we improve performance by shifting more rework discovery to design? Original Work Design Stocks 1,000 0.75 Undiscovered Rework Rework 500 250 -0.0002 0 6 12 18 0 24 30 36 Time (Month) 42 48 Original Work to Do : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 Undiscovered Rework : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 Rework to Do : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 Work Done : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 Design “done” 54 Fraction Task 1 Work Done 750 Design Design Design Design Fraction Design Rework Discovered 0.5 0.25 60 0 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Fraction Design Rework Discovered : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 Discovery by design Discovery by build 43 + - 1. Sources of Rework – Categories (from Lecture 7) Classical “Quality” or design misexecution from people or technical coupling. Discoverable by further design work such as iteration, review. 2. Technical complexity/novelty; customer uncertainty. Discoverable by build/test work, including d/b/t iterations. 3. Knock-on Rework Work done “correctly” but ultimately needing rework. Discoverable by both. 44 + Example: Planned Design Iterations - 1. Add iteration tasks Design Staff Build/Test Staff Design Staff on Original Work Build/Test Staff on Original Work Design Staff on Rework Design Productivity Build/Test Staff on Rework Build/Test Productivity Design Original Work Being Accomplishmed <Build/Test Initial Work to Do> Design Rework Being Accomplishmed Build/test Original Work Being Accomplishmed Build/Test Rework Being Accomplishmed <Build/Test Initial Work to Do> Design Original Work Done Correctly Build/Test Original Work Done Correctly Design Original Work to Do Design Fraction Correct and Complete Build/Test Original Work to Do Design Rework Generation on Original Work Design Rework to Do Design Rework Generation on Rework Design Rework Discovery Build/Test Rework Generation on Original Work Design Work Done Design Undiscovered Rework Build/Test Fraction Correct and Complete Design Fraction Reported Complete Design Fraction of Work Done Correct and Complete Build/Test Work Done Build/Test Rework to Do Build/Test Rework Generation on Rework Build/Test Rework Discovery Build/Test Undiscovered Rework Build/Test Fraction Reported Complete Rework discovered in Design 2. Which discover more rework in design Rework discovered in Build/Test 45 + Increasing design iterations … - … increases design original work, but reduces downstream rework. Design Cumulative Original Work Done Three Two 1,500 1,000 40 One 500 One 30 People 2,000 Tasks Design Staff Two Three 20 10 0 0 0 6 12 18 24 30 36 Time (Month) 42 48 54 Design Cumulative Original Work Done : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Design Cumulative Original Work Done : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 Design Cumulative Original Work Done : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 60 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Design Staff : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Design Staff : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 Design Staff : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 46 - … pushes more rework discovery into design Fraction of Design Selected ReworkVariables Discovered Over Time 2 1.5 Fraction + Three 1 0.5 One Two 0 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Fraction Design Rework Discovered : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Fraction Design Rework Discovered : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 Fraction Design Rework Discovered : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 47 + - Three iterations discovers all the “discoverable” rework Fraction Rework Discovered by Design as Fraction of Max 1 Three Fraction 0.75 Two 0.5 0.25 One 0 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Fraction Rework Discovered by Design as Fraction of Max : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Fraction Rework Discovered by Design as Fraction of Max : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 Fraction Rework Discovered by Design as Fraction of Max : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 Derivable via DSM and signal flow graph simulation? 48 Increasing rework discovered in design reduces rework left for build … + - Fraction Rework Discovered by Design 1 0.8 Three 0.5 Two One 0.25 0 0 6 12 18 24 30 36 Time (Month) 42 48 One 0.6 Fraction 0.75 Fraction Fraction Rework Discovered By Build Two 0.4 Three 0.2 0 54 Fraction Rework Discovered by Design : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Fraction Rework Discovered by Design : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 Fraction Rework Discovered by Design : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 60 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Fraction Rework Discovered By Build : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Fraction Rework Discovered By Build : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 Fraction Rework Discovered By Build : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 49 Improving build “quality” and reducing build rework + - Build/Test Fraction Correct and Complete Cumulative Build Rework 1 600 Three 0.5 Two 0.25 0 6 12 18 24 30 36 Time (Month) Two -13.1% 300 Three 150 -26.6% One 0 One 450 Tasks Fraction 0.75 0 42 48 54 60 "Build/Test Fraction Correct and Complete" : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 "Build/Test Fraction Correct and Complete" : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 "Build/Test Fraction Correct and Complete" : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 Cumulative Build Rework : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Cumulative Build Rework : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Two New5 Cumulative Build Rework : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 50 + - With the “Base Case” Assumptions … While build effort is reduced with more design iterations … … the increasing design cost indicates two iterations are “optimal”. What assumptions impact this tradeoff? 51 + Assumptions Fraction of design tasks that need to be repeated per iteration Relative cost of build/test versus design When build starts (overlap with design) 52 + - The benefits of design iteration increase the higher build cost Cumulative Effort (Person-Months) 0.5 One Iteration Two Iterationss Three Iterations 1 1187 1903 1199 1.01% 1887 -0.84% 1243.5 4.76% 1904 0.05% 1.25 Build Cost Multiplier 1.5 1.75 2261 2619 2231 -1.33% 2575 -1.68% 2234 -1.18% 2565 -2.08% 2977 2919 -1.95% 2895 -2.76% 2 3 3335 4767 3263 -2.16% 4639 -2.69% 3225 -3.30% 4546 -4.64% Increasing Build Cost 53 Build is starting before design rework is fully discovered + - One Iteration Three Iterations 1 1 0.75 0.75 Dimensionless Dimensionless Effect of Design Undiscovered Rework on Fraction Correct Build FCC from Design Build FCC from Design 0.5 Build/Test Ramp-up 0.25 0.5 Build/Test Ramp-up 0.25 0 0 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 0 6 12 18 24 30 36 Time (Month) 42 48 54 60 "Build/Test Effect of Design Undiscovered Rework On Fraction Correct" : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5Effect of Design Undiscovered Rework On Fraction Correct" : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Thre "Build/Test Fraction of Released Design Work Correct and Complete : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 Fraction of Released Design Work Correct and Complete : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 "Build/Test Startup" : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Iter New5 "Build/Test Startup" : Three P Four S V5 BNFCC 0pt95 Sens 0pt75 Middle One Three New5 Delaying build with one iteration will have less benefit because build needed to discover rework. Iterations 2 & 3 occurring months 18-24 54 + - Benefits of delaying build start "Middle" Project Test "New5 Results" Cum Build Rework One Iteration Two Iterations Three Iterations, Start 70% 425.16 369.38 311.86 Three Iterations, Start 60% Three Iterations, Start 70% Three Iterations, Start 80% Three Iterations, Start 90% Two Iteration, Start 60% Two Iteration, Start 70% Two Iteration, Start 80% Two Iteration, Start 90% Design Effort Build Effort Total Effort Finish -13.1% -26.6% 404.4 444.45 516 1432 1376 1321 1903 1887 -0.84% 1904 0.05% 51.6875 52.875 54.8125 337.67 311.86 285.49 271.99 -20.6% -26.6% -32.9% -36.0% 516 516 516 516 1353 1321 1291 1275 1935 1.68% 1904 0.05% 1874 -1.52% 1857 -2.42% 53.5 54.8125 55.4375 56 386.26 369.4 359 348.72 -9.1% -13.1% -15.6% -18.0% 444.45 444.45 444.45 444.45 1396 1376 1364 1353 1907 0.21% 1887 -0.84% 1875 -1.47% 1864 -2.05% 51.125 52.875 53.4375 54.0625 Three iterations, start at 90% “optimal” cost, but finish is later. 55 + Other Factors Affection Desirability of More Planned Iterations Normal amount of rework Amount of rework discoverable in design (vs in build/test) Additional rework discovered per iteration … 56 + Developing Heuristics by Project Type Parameter Normal FCC Frac Discoverable in Design Frac Discoverable First Iteration Frac Discoverable Later Iterations Tasks Repeated # Iterations Build Start “Novel” 0.6 0.3 “Repeat” 0.7 (examples) 0.6 “Mature” 0.8 0.9 Depends on product & organization: analyze projects, use DSM & signal flow graph simulation to estimate. 1 3 2 When planned iterations done. Use simulation to develop heuristics by project type. 57 + - Summary 1. Under almost all situations, two design iterations are most cost effective. The benefits of multiple iterations increases the more design rework that can be discovered by design. Hence, multiple iterations makes more sense for “Repeat” and “Middle” projects than for “Novel” projects. 2. The start of build should be delayed until the design effort has executed all of the planned iterations. 3. The benefits of additional design iteration increases the higher build/test costs are relative to design costs. 58 Revised Network/Gantt showing planned design iterations + - 2010 Apr May June July Aug Sept Oct Nov Dec 2011 Jan Feb Electronics & Software -- Original f Electronics Design h Software Dev p q Electronics Delivery Electronics Software Int 90 5/7/10 150 9/10/10 72 30 9/9/10 b,c,d 4/7/11 9/10/10 12/20/10 4/8/11 5/19/11 Mar Apr May June July Aug Sept Oct Nov Dec Design Build/Test Coupled g,f g,f h,p Electronics & Software -- Rework Discovered in Build/Test f h Electronics Design Software Dev p q Electronics Delivery Electronics Software Int 90 5/7/10 150 9/10/10 72 30 9/9/10 b,c,d 4/7/11 9/10/10 12/20/10 4/8/11 5/19/11 g,f g,f h,p Electronics & Software -- Planned Iterations f h Electronics Design Software Dev p q Electronics Delivery Electronics Software Int 90 5/7/10 150 9/10/10 72 30 9/9/10 b,c,d 4/7/11 9/10/10 12/20/10 4/8/11 5/19/11 g,f g,f h,p Added design iteration tasks … … to reduce unplanned iterations 59 + - SD Qualitative Insights – 2 3. Tradeoffs in the plan can often be improved by changes in project structure and organization to reduce rework and delays in discovering rework. See textbook Chapter SD4 for other examples. 60 + Today’s Agenda - • • • • Strategic Project Management Example 1: Project Preparation Example 2: Project Planning Example 3: Project Execution Deciding on Project Controls 61 + - SD Qualitative Insights – 4 6. Project managers need buffers and/or flexibility (e.g., slip schedule, cut scope, ship with “bugs”) to respond to changes and uncertainties. These have costs that need to be evaluated; the importance of different tradeoffs differs by project. (Lecture 13) 7. The costs of project control can be minimized by understanding the sources of the vicious circles. The timing, magnitude, and duration of different controls affects performance. 62 + Strategic Control Issues Incorporating rework estimates in planning and progress monitoring (see Chapter SD4.4). How much to rely on “work intensity” vs. overtime vs. adding staff? Should you slip the schedule? Early or late? Should you pay extra for experience when adding staff? How much training (delay in adding staff, but higher productivity and quality)? A Strategic View – Deciding in advance the best way to handle problems if they arise 63 + Project Resource Control You’ve misplanned, either because you don’t include rework estimates or because this particular project has unusually high levels …. Or Scope growth occurred on the project Other risks/problems materialized What do you do? (note – these are “permanent” impacts, not temporary delays on isolated parts) 64 Project Control + - “So the best thing to do is to do nothing, right?” + Experience Dilution - + Experience Congestion & Communication Difficulties Increased Turnover - + Hopelessness Overlap & Concurrence + Scope Growth + Workforce + Overtime + Effort Applied + -- + - - + + Work Intensity Rework Generation Progress Add People - + + Work More - + + Haste Creates Out-ofSequence Work Original Work to Do Haste Makes Waste Hiring + Work Done Progressx Rework + + Too Big to Manage No – the costs of project control Quality can be minimized Productivity + by understanding +the sources of the vicious Infesaible circles. The timing, Plan Changes magnitude, and duration + of different controls +affects performance. + Errors Create More Work Unknown Errors in Prior Work + + Burnout +!&! Morale + + Fatigue - + Undiscovered Rework Rework Discovery Rework to Do Work Faster or "Slack Off" Effort Needed + Errors Build Errors Time Remaining + Known Work Remaining Deadline 65 + - What do you do? 2012 What You Do at 30% Add People Longer Hours Intensity Slip Cut Scope Other Total First 10.6% 31.9% 25.5% 17.0% 14.9% 0.0% 100.0% Second 52.2% 23.9% 13.0% 8.7% 2.2% 0.0% 100.0% Third 17.1% 26.8% 19.5% 19.5% 17.1% 0.0% 100.0% Fourth 11.6% 16.3% 23.3% 23.3% 25.6% 0.0% 100.0% Fifth 14.3% 7.1% 21.4% 26.2% 31.0% 0.0% 100.0% Sixth 25.0% 0.0% 0.0% 25.0% 50.0% 0.0% 100.0% Fourth 9.5% 9.5% 21.4% 38.1% 21.4% 0.0% 100.0% Fifth 8.9% 13.3% 22.2% 24.4% 31.1% 0.0% 100.0% Sixth 25.0% 0.0% 0.0% 50.0% 25.0% 0.0% 100.0% What You Do at 65% Add People Longer Hours Intensity Slip Cut Scope Other Total First 16.7% 35.4% 16.7% 8.3% 22.9% 0.0% 100.0% Second 50.0% 29.2% 8.3% 10.4% 2.1% 0.0% 100.0% Third 31.1% 17.8% 26.7% 15.6% 8.9% 0.0% 100.0% 66 + - % Specifying 1st or 2nd Choice 2011 What You Do? Add People Longer Hours Intensity Slip Cut Scope Other Total 2012 At 30% 40.8% 24.3% 21.4% 5.8% 7.8% 0.0% 100.0% At 65% 34.7% 23.5% 19.4% 11.2% 11.2% 0.0% 100.0% What You Do? Add People Longer Hours Intensity Slip Cut Scope Other At 30% At 65% 31.2% 33.3% 28.0% 32.3% 19.4% 12.5% 12.9% 9.4% 8.6% 12.5% 0.0% 0.0% 67 + Brooks’ Law "Adding manpower to a late software project makes it later." Brooks, Frederick P. Jr. The Mythical ManMonth. Reading, MA, Addison Wesley, 1995. Homework 5 Analysis: Under what conditions is this true. 68 + - Qualitative model representation Effort Applied Productivity + Scope Growth Original Work to Do + Resources: Fraction Correct and Complete Rework Generation Progressx Rew ork Rework Discovery - + People Overtime Work Intensisty - Progress Infeasible Plan Rework to Do + Work Done Changes Undiscovered Rework Effort Needed Add Resources + - + Time to Discover Rework Time Remaining + + Known Work Remaining + Deadline 69 + - Project Control 1. Project control is driven by estimates of how much effort is left ... Work to Do (Tasks) 2. Estimates are based on work to do and productivity (undiscovered rework?) Estimated Effort Remaining + (Person-Months) - Average Productivity (Tasks/Month/Person) 70 + - Project Control -- Staffing Staff + Staff Required to Complete on Schedule + - Work to Do (Tasks) Estimated Effort Remaining + (Person-Months) - How many people do I need to get the job done on time? Time Remaining + Staff Required = Estimated Effort Remaining / Average Productivity (Tasks/Month/Person) Scheduled Completion Date Time Remaining [People] 71 + - Project Control – Schedule How many people do I need to get the job done on time? Staff + When Can I finish with the current staff? Indicated Completion Date Staff Required to Complete on Schedule + - + Estimated Effort Remaining + (Person-Months) Work to Do (Tasks) Indicated Completion Date = Time + (Estimated Effort Remaining/Staff) [Month] Time Remaining + - Average Productivity (Tasks/Month/Person) Scheduled Completion Date 72 + - Project Control Based on Staff Required and Indicated Completion Date, three options: 1. Add Staff 2. Explicitly Slip Schedule 3. Exert “Schedule Pressure” (Work Intensity and Extra Hours) 73 + - Actions Determined By … + Work/Schedule Pressure “Willingness to Use Intensity & Extra Hours (0-1) “Willingness to Hire (0 -1)” Staff + Staff Required to Complete on Schedule + - Indicated Completion Date + Estimated Effort Remaining + (Person-Months) Work to Do (Tasks) “Willingness to Slip (0-1)” Time Remaining + Average Productivity (Tasks/Month/Person) Scheduled Completion Date + 74 + - Testing Brook’s Law? Effect of Experience on Productivity and Quality Effect = (New Staff * Relative Experience + Experienced Staff) / Staff Level Relative Experience of New Staff 0.0 New Staff Staff Hired Willingness to Hire What uncertainties would you test sensitivity to? 0 Staff Gaining Experience Experienced Staff 6 months Staff Leaving 40 Staff Level Required 75 + Options Add Staff Work OT Increase “intensity” Slip Schedule Some Combination 76 + Discussion – Resource Controls Relative impact on fraction correct (and productivity) Relative delays Can work intensity be sustained? Limits – greater for OT than WI? 77 + - Step Change in Overtime – Impact on … FCC/PDY Equivalent Staff 1 Net Output 78 + - Step Change in Staff– Impact on … FCC/PDY Equivalent Staff 1 Net Output 79 + - Change in Work Intensity – Impact on … FCC/PDY Equivalent Staff 1 Net Output 80 + - Project Control – Discussion Points What should you do when a project gets behind schedule? • • • • • • • When in the project should you use overtime (and/or for how long)? When do you? When in the project should you hire? When do you? Does it ever pay to work more “intensely” (cut corners, etc.)? Do you? When should you use buffers & slack? Slip Schedule? (as soon as recognized, or try to make up schedule?) 81 + Lessons -- Control - 7. The costs of project control can be minimized by understanding the sources of the vicious circles. The timing, magnitude, and duration of different controls affects performance. Lowest direct cost strategy – slip schedule If need to meet schedule, lowest cost strategy depends on … When during project problem recognized Limits of different resources Size and timing of secondary impacts of control May not always be able to achieve the schedule by adding more resources, but it will always cost you more. 82 + Next SD Class: - Case Examples of … Change management & disputes Risk management Project-to-Project Learning Multi-project dynamics 83 + RW Resources Needed OT WF RW RW WI 84 + - Step Change in Overtime – Impact on … FCC/PDY Equivalent Staff 1 FCC & PDY Net Output 1 85 + - Step Change in Staff– Impact on … FCC/PDY Equivalent Staff 1 FCC & PDY Net Output 1 86 + - Change in Work Intensity – Impact on … FCC/PDY Equivalent Staff PDY 1 FCC Net Output 1 87 MIT OpenCourseWare http://ocw.mit.edu ESD. 6\VWHP3URMHFW0DQDJHPHQW Fall 2012 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.