View
222
Download
2
Tags:
Embed Size (px)
Citation preview
Kven og Kva
System Dynamics: System Dynamics: Uses, Benefits, PerspectivesUses, Benefits, Perspectives
Prof. Jose J. GonzalezAgder College & Powersim ASGrimstad, Norway
Agder College
2
Content
•About System Dynamics (Presentation)•Modelling with Powersim Constructor (Demonstration)•Developing Internet Simulators with Powersim Metro (Demonstration)
3
What is System Dynamics
•Deals with time-dependent behaviour of managed systems•Origins in servomechanisms engineering and management•Perspective based on information feedback and mutual or recursive causality to understand the dynamics of complex technological, biological, and social systems.
4
Uses of System Dynamics
•Corporate planning and policy design•Public management and policy•Process development and optimisation•Dynamic decision making•Biological and medical modelling•Energy and the environment•Complex nonlinear dynamics•Theory development in the natural and social sciences
Kven og Kva 5
System dynamic modelsEffort Perceived Remaining, Hiring and Scheduling
Net_hiring_RATE
Time_remaining
Workforce
Workforce_adj_time
Perc_task_compl_rate
Effort_perc_remaining
Perc_productivity
Desired_workforce Indicated_workforce
Willingness_to_Chng_Workforce
Physical process (flow of workforce)
Physical process (progress of project)
Information flow (Desired_workforce affects recruitment)
Perception (how far the project has proceeded)
Pressure as consequence of perceptions (delays in project affect desired workforce)
Soft variable
Feedback
6
Example: Pollution Control(courtesy Prof. Graham Winch)
•Air pollution in Mexico City is amongst the worst in the world.•The authorities decided to limit vehicle use – every car has a colour-code, and for one workday a week is banished.•The expected result was a 20% reduction in car usage on weekdays.•There now seems more cars than ever, and they seem to be producing ever increasing pollution.
7
Example: Pollution Control (courtesy Prof. Graham Winch)
Rationing
PollutionTarget
Pollution
Car Usage
TargetPollution
TravelNeeds
Numberof Cars
Week-end Use&
Age Effect
8
Example: Project management
•Large (even) huge time and costs overruns in large projects•Consistent experiences in different branches and in all countries•In Norway e.g.:
– Overruns of hundreds of millions in software
– Overruns of billions in building projects– Overruns of tens of billions in offshore– …time overruns of many months and up
to several years, incl. never-completed projects
9
Example: (…Project management)
•Traditional methods for project management fail, because of
– Static approach– Failure to capture feedback– “Primacy of map over terrain”
10
Example: (…Project management)
•Mapping software project management: Sector diagram
Human Resource
Management
Controlling Planning
SoftwareProduction
ScheduleTasks
Completed
EffortRemaining
WorkforceNeeded
WorkforceAvailable
ProgressStatus
11
Example: (…Project management)
•Mapping software project management: Causal-loop diagram
Workforce[men]
Nethiring
[men/month]
+
-
(-)
Desired workforce
[men]
Task completion
rate[tasks/month]
+
Productivity[tasks/man/month]+
+
Tasks completed
[tasks]
+
Tasks remaining
[tasks]
-
Initialproject
size[tasks]+
Effortremaining
[man-months]
+
Productivity[tasks/man/
month]-
+
Scheduled completion time
[months]
-
(-)
Scheduledtime
remaining[months]
+Time
elapsed[months]
-
12
Example: (…Project management)
•Mapping software project management: Stock-and-flow diagram
Human Resource Management Software Production – Manpower Allocation Sector
Some useful definitions
Ave_Daily_MP_pr_Staff
Total_WF
Total_WF
Experienced_WFAve_Daily_MP_pr_Staff
Daily_MP_for_TrainingDaily_MP_for_Training
Rework_MP_Needed_pr_Error
Detected_Errors
Fract_Effort_for_System_Testing
Workforce_Needed
Ceiling_New_Hirees
Assim_Rate_New_Empl
Ave_Assim_Delay
Ceiling_Tot_WF
Cum_Train_Man_days
Daily_MP_for_TrainingFraction_WF_ExpFull_time_Equiv_WF
Full_time_Equiv_Exp_WF
Exp_Empl_Transfer_Rate
Hiring_Delay
Hiring_Rate
Most_New_Hirees_pr_Exp_Staff
New_Empl_Transfer_Rate
Exp_Empl_Quit_Rate
Transfer_Rate_Out_Project
Transfer_Delay_People_Out
Trainers_pr_New_Employee
Experienced_WF
Workforce_Gap
New_Workforce
Workforce_Sought
Rate_of_Perc_RW_MP
Adj_in_Plan_Fract_of_MP_for_QA
Ave_Daily_MP_pr_Staff
Act_Fract_of_MP_for_QA
Cum_Dev_Man_days
Rate_of_Dev_Man_days
Cum_QA_Man_days
Cum_Rework_Man_days
Cum_Man_days_Expended
Des_Error_Correct_Rate
Desired_Rework_Delay
Daily_MP_Avail_after_Training
Daily_MP_for_Dev_Test
Daily_MP_for_QA
Daily_MP_Alloc_for_Rework
Daily_MP_for_Softw_ProdPerc_Rework_MP_Needed_pr_Error
Time_to_Adj_Perc_Rework_MP
Total_Daily_Manpower
Pcent_Job_Act_Worked
Ave_Employment_Time
Sched_Pressure
Quality_Objective
Plan_Fract_of_MP_for_QA
Total_WF
Team_Size_at_Start
13
Example: (…Project management)
•Why are large projects difficult to manage?
– Complexity (structural and dynamical)– Feedback– Delayed perceptions– Nonlinearities
•E.g. In what proportion should one allocate available software engineers to coding, testing and rework?
– Answer depends on productivity parameters, project status, etc.
14
Example: (…Project management)
•Often behaviour is counter-intuitive•E.g.:Trying to get project on schedule
Experienced Workforce
NewWorkforce
+
Productivity
+
+
Hiring
+
In Training
+
–
Experienced Workforce
NewWorkforce
+
Productivity
+
+
Hiring
+
+
15
Example: (…Project management)
•Software Project Management model provides insights and lets you test robust strategies
16
Evolving Uses of System Dynamics
•In recent times system dynamics has emerged as a powerful tool for organisational learning:
– Knowledge capture– Better utilisation of knowledge– “Double-loop learning”
Kven og Kva 17
‘Model’ of Knowledge in a Reality Domain
Partially erroneous knowledge
Networks utilising fragmented knowledge
Fragmented, individual knowledge Ideal
knowledge
Hayek
Kven og Kva 22
Growth of Knowledge Recruitment
Increased individual knowledge
Feedback (correction of misperceptions)
Larger and stronger networks
23
Single-loop Learning and Acting in Dynamic Complex Domains
Reality domain
Decisions
PolicyMental model
of realitydomain
Information feedback
24
Double-loop Learning and Acting in Dynamic Complex Domains
Reality domain
Decisions
PolicyMental model
of realitydomain
Information feedback
25
The Logic of Failure
• Research by Dörner et al. about thinking, decision-making and acting in complex domains
• Most people fail and the behaviour patterns are (quite) ‘universal’
• … but a few master complexity
29
The Logic of Failure
• Typical (“linear”) reasoning: Killing of apes will lead to greater harvests
Apes
Harvests-
-
Hunting
30
The Logic of Failure
• Untypical thinking in causal networks
Apes
Harvests-
-
Hunting
-Insects
-
Hungryleopards
Cattle- -
31
The Logic of Failure
• “The Logic of Failure” based on 20+ years research:– Dörner, Dietrich: Die Logik des Mißlingens.
Reinbek: Rowohlt, 1989.
– Dörner, Dietrich: (1996). The logic of failure: Why things go wrong and what we can do to make them right. 1st American ed. New York: Metropolitan Books, 1996.
40
Thinking and Acting in a Reality Domain
Reality domain
Decisions Information feedback
Policy Mental models
Real world•“One shot policy”
Unknown structureMany componentsHigh degree of couplingDynamicsFeedbackDelaysNot reproducible
Decisions in the real world•Implementation problems•Inconsistency•Group psychology (game playing)•Short-time gain decisive
Feedback in the real world•Incomplete or lacking•Ambiguous•Delayed•Biased, masked or erroneous
Real world•Late updating of mental models
41
Enhanced Insight via Virtual Worlds
Reality domain
Decisions Information feedback
Policy Mental models
Virtual world
Feedback in the real world•Incomplete or lacking•Ambiguous•Delayed•Biased, masked or erroneous
Decisions in the virtual world•Perfect implementation•Consistency•No group psychological processes•Insight and learning aremain goals•Long-time aspects
Feedback in the virtual world•Concentrated, complete•Unambiguous information•Immediate response•Accurate and correct
Decisions in the real world•Implementation problems•Inconsistency•Group psychology (game playing)•Short-time gain decisive
Real + virtual world•Strategy•Structure•“What-if” analysis•Long-time consequences via simulation•Sensitivity analysis
Real + virtual world•“Causal loop analysis”•Mapping of feedback•Disciplined reasoning•Group learning
•Unknown structure•Many components•High degree of coupling•Dynamics•Feedback•Delays•Not reproducible
•Known structure•Open for insight•GUI with graduated complexity•Reproducible experiments
Kven og Kva 44
Benefits of System Dynamics
SD Characteristic Organisation concerns
System viewpoint Organisation challenge
Feedback analysis Consequence of actions
Dynamic modelling Knowledge gain, concern withfuture
Simulation Testing ideas
Optimisation Robustness against uncertainty
Transparency of causal-loop diagram& simulation model
Organisational learning
Fast simulation What-if analysis