Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Lessons From Using System Dynamics For Complex Decisions
Tom Mullen, PA Consulting Group Colonel Jim Baker, Joint Staff
25 July 2011
© PA Knowledge Limited 2011. Page 2
Context – lessons from work helping leaders on important challenges …
These are hard problems with many dynamics!
Analysis is only part of the challenge …
We face many big challenges with:• Overlapping objectives
• Complex histories and relationships
• Unintended consequences
• Long time delays
• Poor coordination across involved groups
• Difficulty understanding the full impacts of actions
Sophisticated non-state
extremist groups
Humanitarian disastersTransnational
criminal operations
Natural resource scarcity
Near peer trajectories & relationships
Rogue state aggression
Globalization & technology
Grievances & local
insurgenciesPopulation
growth
“Homegrown”terrorist threats
Partner relationships
& capacity
WMD proliferation
Economic dependencies
& trends
Formal alliances & agreements
Cyber attack
Susceptibility to radicalization
Introduction
© PA Knowledge Limited 2011. Page 3
There are many lessons from using System Dynamics to help address difficult problems in complex organizations
A few recurring challenges …
1. Match the method to the problem
2. Theories are different from reality
3. You will be a sound bite
4. Work in both the boardroom and the field
5. Use different formats to get your message out
6. Don’t rely only on System Dynamics
7. Don’t be discouraged by politics
Lessons Learned
These may be familiar to many of you …… we have some fresh examples to share!
© PA Knowledge Limited 2011. Page 4
Equipment
XXX
XXX
XXX
WORKING DRAFT
xxx
Retention xxx
Facilitiesxxx
xxx
Recruitingxxx
xxx
xxx
Training
xxx
Staffxxx
xxxxxx
Funding FY07FY08
FYXX
XXX
XXX
XXX
XXX
XXX
XXX
xxx
xxx
xxx
• Xxx• Xxx• Xxx
• Xxx• Xxx• Xxx
• Xxx• Xxx• Xxx
• Xxx• Xxx• Xxx
• Xxx• Xxx• Xxx
• Xxx• Xxx• Xxx
• Xxx• Xxx• Xxx
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXX
XXXXXX
XXX
XXX
XXX
XXX
XXX
XXX
WORKING DRAFT
FUNDING
EQUIPMENT
RECRUITING
TRAINING
RETENTION
STAFF
FACILITIES
Causal Dynamics Diagram
System Dynamics Model
WORKING DRAFT
Enterprise-Level Block Diagram
Benefit:• Create integrated view
of organization to elevate discussions
• Fill a gap – had no rigorous way to assess enterprise-wide impacts of key actions and decisions
Benefit:• Help identify
challenges and examine mitigation options qualitatively
• Identify potential unintended consequences to manage
Benefit:• Quantify impacts over
time of different options to manage growth
• Quantify factors ignored by more linear / narrower analyses
• Help those who want hard numbers
Lesson 1: Match the method to the problem
Lesson 1: Match the method to the problem
If you hear … “We want to model this” … consider what a pproach makes sense:
© PA Knowledge Limited 2011. Page 5
Lesson 2: Theories are different from reality
Lesson 2: Theories are different from reality
One expert’s theory for solving the challenges of getting disaster relief to victims of the 2005 tsun ami:
"People in [this country] are very musically oriented so to make a quick impact after a disaster it is important to use the popular music favored by the different classes…."
Field Manual 3-24 summarizes DoD approach to counterinsurgency
We developed a model to illustrate the dynamics of this theory
Reality has many complications …
Theory… … Reality
Horn of Africa (HOA)Field Manual 3-24
HOA Complexities:•Low literacy•High poverty •Poor border controls•Low government
‘legitimacy’, with military abuses against citizens
•Limited/no public services
•Ethnic, clan and regional conflicts and many insurgencies
• ...
#1 most unstable nation in
world
#2 most unstable nation
© PA Knowledge Limited 2011. Page 6
© PA Knowledge Limited 2009 Page 8
• The “base case” represents likely behavior given best assumptions / current model structure
• We can run many types of “what if” tests and compare outcomes against the base case to assess impacts
• Given the many sources of uncertainty and recognition of “90-day” model structure maturity, we first show:
1. Base Case behavior
2. “Optimistic” variation of Base Case
3. “Pessimistic” variation of Base CaseTime/Effort
Con
fiden
ce
Diminishing returns –well executed effort can get to 80% solution rapidly
“Max potential confidence” depends on the nature of the system – situations with many ‘soft’drivers such as human perceptions and limited data will be less certain than more ‘tangible’systems
We are here: COIN is inherently soft topic, but there is a lot of Afghan. info and data to draw from
Base case variations are based on different assumptions for key uncertainties including: • Utility of xxx• Number of xxx doing xxx• Xxx impact on xxx• Time to distribute xxx• Normal time to develop xxx• Sensitivity to xxx and xxx • Impact of xxx and xxx on xxx• Impact of xx and xxx on xxx, xxx, and xxx
Model Confidence “Status”(Illustrative)
Strong forward progress on model development, refin ement & preliminary ‘what if’ tests, but critical to recognize ’90-day’ status & view outputs without false precision
We are here
© PA Knowledge Limited 2009 Page 8
• The “base case” represents likely behavior given best assumptions / current model structure
• We can run many types of “what if” tests and compare outcomes against the base case to assess impacts
• Given the many sources of uncertainty and recognition of “90-day” model structure maturity, we first show:
1. Base Case behavior
2. “Optimistic” variation of Base Case
3. “Pessimistic” variation of Base CaseTime/Effort
Con
fiden
ce
Diminishing returns –well executed effort can get to 80% solution rapidly
“Max potential confidence” depends on the nature of the system – situations with many ‘soft’drivers such as human perceptions and limited data will be less certain than more ‘tangible’systems
We are here: COIN is inherently soft topic, but there is a lot of Afghan. info and data to draw from
Base case variations are based on different assumptions for key uncertainties including: • Utility of xxx• Number of xxx doing xxx• Xxx impact on xxx• Time to distribute xxx• Normal time to develop xxx• Sensitivity to xxx and xxx • Impact of xxx and xxx on xxx• Impact of xx and xxx on xxx, xxx, and xxx
Model Confidence “Status”(Illustrative)
Strong forward progress on model development, refin ement & preliminary ‘what if’ tests, but critical to recognize ’90-day’ status & view outputs without false precision
We are here
Lesson 3: You will be a sound bite
Lesson 3: You will be a sound bite
… is not what they’ll remember:
What you say:
“Our new policy won’t
work.”
© PA Knowledge Limited 2009 Page 8
Must recognize sources and range of uncertainty – un der more optimistic assumptions the same inputs yield a more positive outlook vs base case
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0Green Amber Red
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0xxx xxx xxx
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
Base
Optimistic
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
TimeTime
Time
Time TimePRELIMINARY ANALYSIS- NOT PRECISE / PREDICTIVE
© PA Knowledge Limited 2009 Page 8
Must recognize sources and range of uncertainty – un der more optimistic assumptions the same inputs yield a more positive outlook vs base case
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0Green Amber Red
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0xxx xxx xxx
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
Base
Optimistic
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
TimeTime
Time
Time TimePRELIMINARY ANALYSIS- NOT PRECISE / PREDICTIVE
© PA Knowledge Limited 2009 Page 9
But just as likely - more pessimistic assumptions yi eld more negative outcomes than base case
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxxxxxxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0Green Amber Red
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0xxx xxx xxx.
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
Base
Pessimistic
Preliminary “what if” analyses in Appendix
Time Time
Time TimeTime
PRELIMINARY ANALYSIS- NOT PRECISE / PREDICTIVE
© PA Knowledge Limited 2009 Page 9
But just as likely - more pessimistic assumptions yi eld more negative outcomes than base case
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxxxxxxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0Green Amber Red
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0xxx xxx xxx.
Comments: xxx
XXX
2002 2006 2010 2014 2018 2022 20260.0
0.2
0.4
0.6
0.8
1.0
Comments: xxx
Base
Pessimistic
Preliminary “what if” analyses in Appendix
Time Time
Time TimeTime
PRELIMINARY ANALYSIS- NOT PRECISE / PREDICTIVE
© PA Knowledge Limited 2011. Page 7
Lesson 4: Work in both the boardroom and the field
Ineffective or excessive security measures cause resentment —leading to an oscillating dynamic where clear progress is hard to see
population feels resentful/
fearful
DELAY
Fixes that fail #2: Resentment or fear grows
faster than security
insurgent attacks
COIN security measures
feelings of security
popular support for insurgents
+
+
+
-
DELAY
+
feelings of resentment
feelings of security
time
Popular support for insurgents
“ Insurgents will lead to sabotage, terror spreading insurrection. It will be a remarkable government that will not be driven to stern repressive measures – curfews, the suspension of civil liberties, a ban on assemble, illegal acts that can only deepen popular opposition, creating a vicious circle of rebellion and repression until he economy is undermined, the social fabric torn beyond question, and the regime tottering on the verge of collapse.”
- Taber, War of the Flea
Occupying forces are perceived as temporary stop-gaps or occupying forces apply inconsistent ROE
Fixes that fail #3: Security measures do not
appear permanent
security dependent on
transient conditions
insurgent attacks
COIN security measures
feelings of security
popular support for insurgents
+
+
+
-
DELAY
+“If the Americans keep doing it, they can make a difference. But they have to stay.
Otherwise, it will never work.”
-- Ali Muhahmed-Ice Cream shop owner
-16 Feb 07
Lesson 4: Work in both the boardroom and the field
Insurgency study used many sources:• Officers with field experience• Retired officers with COIN experience
in multiple places• Review of publicly available sources• What’s missing – Iraqi perspective
Fort Wainwright, Alaska
Meeting those in field is important to:• Test insights with those who will use it
and ensure messages hold up• Add real-world insight / experience
We got the full experience when
modeling insurgency …
© PA Knowledge Limited 2011. Page 8
© PA Knowledge Limited 2003. All rights reserved.-xxx- 25/03/2002- 9
Simplified ‘block view’ of LUL’s business and
interactions with key stakeholders
Market
Employees
Govt
Assets
Finances
SuppliersCapitalSpend
OtherCivils
Track
Stations
Power
Lifts & Escalators
Communications
Signalling
Rolling stock
LUL’s Capital Assetsxxxxxxxxxxxxxxxxxx
Financesxxxxxxxxxxxxxxx
Governmentxxxxxxxxxxxx
Taxis and Other
Private CarTrain Operating Companies
Bus
LUL
Marketxxx
xxxxxxxxxxxxxxx
XXXXXXXXXX
XXXXXXXXXX
Suppliersxxxxxxxxx
xxx
Train DriversStation Staff
Operations Support Staff
Corporate/Administrative StaffMaintenance Technicians
Project Management Staff
Capital Program Technicians
Professional Engineers
LUL’s Workforcexxxxxxxxxxxxxxx
XXXXXXXXXX
XXXXXXXXXX
XXXXXXXXXX
XXXXXXXXXX
XXXXXXXXXX
XXXXXXXXXX
XXXXX
XXXXX
XXXXX
XXXXXXXXXX
XXXXX
XXXXXXXXXX
XXXXX
XXXXXXXXXX
“Insidious Decline”
Lesson 5: Use different formats to get your message out
“Workforce Dynamics”
“Dynamics of Partnership”
“Winning Under the Performance
Regime”
High level block diagram for executive discussions
A Dynamic simulation model was used for analyses and included a user guide and interface so that non-experts could test scenarios
Lesson 5: Use different formats to get your message out
Written issue papers for potential bidders and internal LUL staff distilled results of simulation model analyses to explain the key elements of a successful partnership
Detailed diagrams for working sessions & analyses
© PA Knowledge Limited 2003. All rights reserved.-xxx- 25/03/2002- 12
A high level cause-effect view of key drivers of
Underground performance
Market
GovernmentPolicy
Workforce
Suppliers
Finance Assets
Natural Wastage
LUL Employees
Redund-ancies
In-House Experience/ Competence
Desired LUL Employees
Prevailing Wages/Labour
Practices
Wages/ Labour
Practices
Staff Performance
Fraction of Desired
Staff Available
Productivity
Comparable Services
Benchmarks
Planned Service Level
Accident/ Threat
IncidenceNon-LUL Accidents
LUL Marketshare
LUL Ridership
Economy
Greater London
Transport Need
Tourists
Employment Level in London Intermodal
Seamlessness
Ease of UsePerceived
Safety
Safety Standards
Network Seamlessness
Relative Attractiveness
Journey Time
Social Benefit
Ambience
Integrated Transport
PolicyGovernment
GrantPrivate Sources
Stability of Grant
Fare Policy/ Structure
Fare Level/ Structure
Margin
Passenger Revenue
Operating Costs
Cost Per Train
Kilometer
LUL Budget
Other Revenue
Interest Cost
Interest Rates
Performance Improvement Investment
InvestmentContractual Efficiency
Investment Efficiency
Utilization/ Crowding
Delivered Trains/Hour
Age/ Technology
Assets
Maintenance Spending
Condition
Facility Availability
Capacity
Available Capacity
Maintenance Backlog
Maintenance Required
Hires
MoraleLost Time
Calibre
HeadwayConsistency
Supplier Competition
Relative Supplier Cost
EfficiencyUse of Suppliers
Supplier ”Flux”
Supplier Experience/ Competence
Average Contract
Size/ Lifetimes
Market
GovernmentPolicy
Workforce
Suppliers
Finance Assets
Natural Wastage
LUL Employees
Redund-ancies
In-House Experience/ Competence
Desired LUL Employees
Prevailing Wages/Labour
Practices
Wages/ Labour
Practices
Staff Performance
Fraction of Desired
Staff Available
Productivity
Comparable Services
Benchmarks
Planned Service Level
Accident/ Threat
IncidenceNon-LUL Accidents
LUL Marketshare
LUL Ridership
Economy
Greater London
Transport Need
Tourists
Employment Level in London Intermodal
Seamlessness
Ease of UsePerceived
Safety
Safety Standards
Network Seamlessness
Relative Attractiveness
Journey Time
Social Benefit
Ambience
Integrated Transport
PolicyGovernment
GrantPrivate Sources
Stability of Grant
Fare Policy/ Structure
Fare Level/ Structure
Margin
Passenger Revenue
Operating Costs
Cost Per Train
Kilometer
LUL Budget
Other Revenue
Interest Cost
Interest Rates
Performance Improvement Investment
InvestmentContractual Efficiency
Investment Efficiency
Utilization/ Crowding
Delivered Trains/Hour
Age/ Technology
Assets
Maintenance Spending
Condition
Facility Availability
Capacity
Available Capacity
Maintenance Backlog
Maintenance Required
Hires
MoraleLost Time
Calibre
HeadwayConsistency
Supplier Competition
Relative Supplier Cost
EfficiencyUse of Suppliers
Supplier ”Flux”
Supplier Experience/ Competence
Average Contract
Size/ Lifetimes
Market
GovernmentPolicy
Workforce
Suppliers
Finance Assets
Natural Wastage
LUL Employees
Redund-ancies
In-House Experience/ Competence
Desired LUL Employees
Prevailing Wages/Labour
Practices
Wages/ Labour
Practices
Staff Performance
Fraction of Desired
Staff Available
Productivity
Comparable Services
Benchmarks
Planned Service Level
Accident/ Threat
IncidenceNon-LUL Accidents
LUL Marketshare
LUL Ridership
Economy
Greater London
Transport Need
Tourists
Employment Level in London Intermodal
Seamlessness
Ease of UsePerceived
Safety
Safety Standards
Network Seamlessness
Relative Attractiveness
Journey Time
Social Benefit
Ambience
Integrated Transport
PolicyGovernment
GrantPrivate
Sources
Stabil ity of Grant
Fare Policy/ Structure
Fare Level/ Structure
Margin
Passenger Revenue
Operating Costs
Cost Per Train
Kilometer
LUL Budget
Other Revenue
Interest Cost
Interest Rates
Performance Improvement Investment
InvestmentContractual Efficiency
Investment Efficiency
Utilization/ Crowding
Delivered Trains/Hour
Age/ Technology
Assets
Maintenance Spending
Condition
Facil ity Availability
Capacity
Available Capacity
Maintenance Backlog
Maintenance Required
Hires
MoraleLost Time
Calibre
HeadwayConsistency
Supplier Competition
Relative Supplier Cost
EfficiencyUse of Suppliers
Supplier ”Flux”
Supplier Experience/ Competence
Average Contract
Size/ Lifetimes
Market
GovernmentPolicy
Workforce
Suppliers
Finance Assets
Natural Wastage
LUL Employees
Redund-ancies
In-House Experience/ Competence
Desired LUL Employees
Prevailing Wages/Labour
Practices
Wages/ Labour
Practices
Staff Performance
Fraction of Desired
Staff Available
Productivity
Comparable Services
Benchmarks
Planned Service Level
Accident/ Threat
IncidenceNon-LUL Accidents
LUL Marketshare
LUL Ridership
Economy
Greater London
Transport Need
Tourists
Employment Level in London Intermodal
Seamlessness
Ease of UsePerceived
Safety
Safety Standards
Network Seamlessness
Relative Attractiveness
Journey Time
Social Benefit
Ambience
Integrated Transport
PolicyGovernment
GrantPrivate Sources
Stability of Grant
Fare Policy/ Structure
Fare Level/ Structure
Margin
Passenger Revenue
Operating Costs
Cost Per Train
Kilometer
LUL Budget
Other Revenue
Interest Cost
Interest Rates
Performance Improvement Investment
InvestmentContractual Efficiency
Investment Efficiency
Utilization/ Crowding
Delivered Trains/Hour
Age/ Technology
Assets
Maintenance Spending
Condition
Facility Availability
Capacity
Available Capacity
Maintenance Backlog
Maintenance Required
Hires
MoraleLost Time
Calibre
HeadwayConsistency
Supplier Competition
Relative Supplier Cost
EfficiencyUse of Suppliers
Supplier ”Flux”
Supplier Experience/ Competence
Average Contract
Size/ Lifetimes
Market
GovernmentPolicy
Workforce
Suppliers
Finance Assets
Natural Wastage
LUL Employees
Redund-ancies
In-House Experience/ Competence
Desired LUL Employees
Prevailing Wages/Labour
Practices
Wages/ Labour
Practices
Staff Performance
Fraction of Desired
Staff Available
Productivity
Comparable Services
Benchmarks
Planned Service Level
Accident/ Threat
IncidenceNon-LUL Accidents
LUL Marketshare
LUL Ridership
Economy
Greater London
Transport Need
Tourists
Employment Level in London Intermodal
Seamlessness
Ease of UsePerceived
Safety
Safety Standards
Network Seamlessness
Relative Attractiveness
Journey Time
Social Benefit
Ambience
Integrated Transport
PolicyGovernment
GrantPrivate Sources
Stability of Grant
Fare Policy/ Structure
Fare Level/ Structure
Margin
Passenger Revenue
Operating Costs
Cost Per Train
Kilometer
LUL Budget
Other Revenue
Interest Cost
Interest Rates
Performance Improvement Investment
InvestmentContractual Efficiency
Investment Efficiency
Utilization/ Crowding
Delivered Trains/Hour
Age/ Technology
Assets
Maintenance Spending
Condition
Facility Availability
Capacity
Available Capacity
Maintenance Backlog
Maintenance Required
Hires
MoraleLost Time
Calibre
HeadwayConsistency
Supplier Competition
Relative Supplier Cost
EfficiencyUse of Suppliers
Supplier ”Flux”
Supplier Experience/ Competence
Average Contract
Size/ Lifetimes
Market
GovernmentPolicy
Workforce
Suppliers
Finance Assets
Natural Wastage
LUL Employees
Redund-ancies
In-House Experience/ Competence
Desired LUL Employees
Prevailing Wages/Labour
Practices
Wages/ Labour
Practices
Staff Performance
Fraction of Desired
Staff Available
Productivity
Comparable Services
Benchmarks
Planned Service Level
Accident/ Threat
IncidenceNon-LUL Accidents
LUL Marketshare
LUL Ridership
Economy
Greater London
Transport Need
Tourists
Employment Level in London Intermodal
Seamlessness
Ease of UsePerceived
Safety
Safety Standards
Network Seamlessness
Relative Attractiveness
Journey Time
Social Benefit
Ambience
Integrated Transport
PolicyGovernment
GrantPrivate Sources
Stability of Grant
Fare Policy/ Structure
Fare Level/ Structure
Margin
Passenger Revenue
Operating Costs
Cost Per Train
Kilometer
LUL Budget
Other Revenue
Interest Cost
Interest Rates
Performance Improvement Investment
InvestmentContractual Efficiency
Investment Efficiency
Utilization/ Crowding
Delivered Trains/Hour
Age/ Technology
Assets
Maintenance Spending
Condition
Facility Availability
Capacity
Available Capacity
Maintenance Backlog
Maintenance Required
Hires
MoraleLost Time
Calibre
HeadwayConsistency
Supplier Competition
Relative Supplier Cost
EfficiencyUse of Suppliers
Supplier ”Flux”
Supplier Experience/ Competence
Average Contract
Size/ Lifetimes
Market
GovernmentPolicy
Workforce
Finance Assets
Suppliers
Market
GovernmentPolicy
Workforce
Finance Assets
Suppliers
© PA Knowledge Limited 2003. All rights reserved.-xxx- 25/03/2002- 8
User’s Guide has detailed description of equation structures and lets users drill down to specific assumptions
Table of Contents:
LUL Dynamic Simulation Model Documentation
SECTION 1: A HIGH LEVEL VIEW OF THE MODEL .......... .......................... 1
SECTION 2: THE SECTORS OF THE MODEL ................ .............................. 2
SECTION 3: MARKET SECTOR ........................... ......................................... 3
SECTION 4: SERVICE DELIVERY SECTOR ................. ................................ 4
SECTION 5: LUL STAFF SECTOR ........................ ........................................ 5
SECTION 6: LUL FINANCE SECTOR...................... ...................................... 6
SECTION 7: GOVERNMENT POLICY SECTOR................ ............................ 7
SECTION 8: INFRACO STAFF SECTOR.................... ................................... 8
SECTION 9: INFRACO FINANCE SECTOR .................. ................................ 9
SECTION 10: INFRACO ASSETS SECTOR.................. .............................. 10
SECTION 11: PERFORMANCE REGIME SECTOR.............. ....................... 11
SECTION 12: MODEL DEVELOPMENT AND VALIDATION PROCES S .... 12
London UndergroundLondon Underground
DynamicSimulation ModelDocumentation
October 1999 PA-Consulting Group
© PA Knowledge Limited 2011. Page 9
Lesson 6: Don’t rely only on System Dynamics…
Lesson 6: Don’t rely only on System Dynamics
SD analysis identified a big delay in this large development program
Managers of the troubled area rejected the analysis, claiming they were on track
They attacked the SD analysis
To help reach agreement, we led a conventional analysis of the areas in question -- using their data -- and highlighted the hugely optimistic assumptions that were implicit
Change in Overall Mission Systems Productivity
0%
20%
40%
60%
80%
100%
120%
140%
2002 2003 2004 2005 2006 2007 2008 2009 2010
Change in Overall Mission Systems Productivity
0%
20%
40%
60%
80%
100%
120%
140%
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Cha
nge
in A
vera
ge M
S P
rodu
ctiv
ity
Base Red AggressiveRed Risk Blue
?
?
Red Base = Ave 30% Improvement
Blue = Ave 60% Improvement
Red Aggressive Red Base Red RiskBlue Outlook
(Average ∆Efficiency = 42%) (Ave ∆Efficiency = 30%) (Average ∆Efficiency = 25%) (Average ∆Efficiency = 60%)
Red Aggressive Red Base Red RiskBlue Outlook
(Average ∆Efficiency = 42%) (Ave ∆Efficiency = 30%) (Average ∆Efficiency = 25%) (Average ∆Efficiency = 60%)
~In the boxChange in Overall Mission Systems Productivity
0%
20%
40%
60%
80%
100%
120%
140%
2002 2003 2004 2005 2006 2007 2008 2009 2010
Change in Overall Mission Systems Productivity
0%
20%
40%
60%
80%
100%
120%
140%
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Cha
nge
in A
vera
ge M
S P
rodu
ctiv
ity
Base Red AggressiveRed Risk Blue
?
?
Red Base = Ave 30% Improvement
Blue = Ave 60% Improvement
Red Aggressive Red Base Red RiskBlue Outlook
(Average ∆Efficiency = 42%) (Ave ∆Efficiency = 30%) (Average ∆Efficiency = 25%) (Average ∆Efficiency = 60%)
Red Aggressive Red Base Red RiskBlue Outlook
(Average ∆Efficiency = 42%) (Ave ∆Efficiency = 30%) (Average ∆Efficiency = 25%) (Average ∆Efficiency = 60%)
~In the box
Change in Overall Software Productivity
Time
Cha
nge
in A
vera
ge P
rodu
ctiv
ity
© PA Knowledge Limited 2010.
Hours Expended & Progress Achieved - Spent ~34% of t otal hours and achieved ~27% complete
Hours Expended Progress Achieved
% of EAC Spent as of XXX Roll-up of detailed data fr om other reports and plans
At the time of April Study:
~20% = Overall Progress Achieved
Latest Study:
~27% = Overall Progress Achieved
At the time of April Study:
Total Hours Expended = 29%
Latest Study:
Total Hours Expended = 34%
Spent 5% budget and made 7% progress – progress in this block is catching up, but overall need to average
10% progress per 5% budget from here forward to meet plan … with no scope growth or unexpected rework!
2
© PA Knowledge Limited 2011. Page 10
Lesson 6: Don’t rely only on System Dynamics
We modeled the drivers of
regional stability
But some people find
expert opinion more credible
…
Use both approaches!
Lesson 6: Don’t rely only on System Dynamics
Disaster Relief Study
XxxXxxXxxXxxXxxXxxxxx
xxx xxx xxx xxx xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
xxx
Question 6: xxxxxxx
Extent, Type of involvement
© PA Knowledge Limited 2011. Page 11
Lesson 7: Don’t be discouraged by politics … find the art of the possible
Health System challenges are complex and high stakes with no easy answers
The highest-leverage actions are “non-starters” in this highly politicized arena
Having the “right” analysis is not enough to “fix the system”
Instead, find and implement improvements that arepossible given the politics
© PA Knowledge Limited 2011. Page 29
Healthcare supply, demand, and costs must be analyzed along
many dimensions (type, quantity, location, etc…)
Costs
• XXX
• XXX
• XXX
• XXX
Enterprise Impacts/Risks
• XXX
• XXX
• XXX
Health System Supply
XXX XXX
XXX XXX
Health System Demand
XXX XXX
XXX XXX
HEALTH
CARE
DELIVERED
Important to understand specific cost drivers, but must understand the broader “system” in order to rigorously evaluate options and avoid “local optimization”
Legal, regulatory, political, procedural, and cultural issues must also be reflected in the
framework.
© PA Knowledge Limited 2011. Page 29
Healthcare supply, demand, and costs must be analyzed along
many dimensions (type, quantity, location, etc…)
Costs
• XXX
• XXX
• XXX
• XXX
Enterprise Impacts/Risks
• XXX
• XXX
• XXX
Health System Supply
XXX XXX
XXX XXX
Health System Demand
XXX XXX
XXX XXX
HEALTH
CARE
DELIVERED
Important to understand specific cost drivers, but must understand the broader “system” in order to rigorously evaluate options and avoid “local optimization”
Legal, regulatory, political, procedural, and cultural issues must also be reflected in the
framework.
Lesson 7: Don’t be discouraged by politics
Persistence can also pay off … many important changes begin as ‘non-starters’ …
© PA Knowledge Limited 2011. Page 12
Moving Forward -- Our challenges are not getting any simpler!
• Sustainable Budgets / Reducing Long-Term Debts
• Energy Options for the Mid and Long Term
• Understanding Economic Interdependence
• National / International Security
• Healthcare
• Demographic Changes
• Rising and Falling Powers
• ….
System Dynamics can be a powerful approach to help analyze many of these issues – we hope these ‘lessons’ are of some h elp to those of you addressing these and similarly complex problems in the private sector!
Moving Forward