An Introduction to Systems Thinking and Dynamic Modeling

Report
Introduction to
Systems Thinking and
Dynamic Modeling (ST&DM): Part I
For Tahoma School District on June 3, 2010
Tahoma contact: Dawn Wakeley
[email protected], 425-413-3424
Paul Newton (Boeing)
[email protected], 206-544-7641
Dr. Dexter Chapin (Seattle Academy of Arts and Sciences)
[email protected], 206-323-6600
Jim Ray (retired Boeing engineer)
[email protected], 425-865-9319 (home)
1
Agenda: Intro to ST&DM Part I
• Broad application areas of systems thinking
and dynamic modeling (ST&DM)
• Slinky
• What is ST&DM?
• Drug-related crime
• Modeling example: filling a water glass
• ST&DM at Boeing
• First feedback loops
2
Broad Application Areas of ST&DM
• To Technology Problems
– Control engineers do ST&DM all the time, although they might not call it that
– Examples: autopilot, thermostat, paper machine, electric blanket, cruise
control, steam engine, electric motor, computer, etc.
• To Social Problems
–
–
–
–
–
–
–
–
Business dynamics
Family dynamics
Community dynamics
Insurgency dynamics
Ecological dynamics
Organizational dynamics
Urban dynamics
Etc…..
3
What is the systems lens?
Hint: Structure and behavior
Introduction from Meadows, D. H., & Wright, D. (2008). Thinking in
systems: A primer. White River Junction, Vt: Chelsea Green Pub.
4
What is systems thinking?
• A perspective and a set of conceptual tools
that enable us to understand the structure and
behavior of dynamically complex problems
• A rigorous modeling method that enables us
to build computer simulations of dynamically
complex problems and use them to design more
effective policies and organizations
5
[slightly modified from Sterman, John (2000) Business Dynamics: Systems Thinking and Modeling for a Complex World. Irwin McGraw-Hill]
What is system dynamics? Some quotes…
System dynamics is the use of computer simulation for policy analysis in complex systems. Its big contribution is helping people to build
progressively richer understandings of some dynamic problem, and anticipate weaknesses in policy initiatives that would develop over time. It
gets a lot of its power from a 'feedback' perspective -- the realization that tough dynamic problems arise in situations with lots of pressures and
perceptions that interact to form loops of circular causality, rather than simple one-way causal chains. Humans are really good at thinking up
all that interconnected complexity and really weak at inferring its implications without the support of simulation models.
George Richardson
Rockefeller College of Public Affairs and Policy, State University of New York at Albany
System dynamics deals with how things change through time, which includes most of what most people find important. It uses computer
simulation to take the knowledge we already have about details in the world around us and to show why our social and physical systems
behave the way they do. System dynamics demonstrates how most of our own decision-making policies are the cause of the problems that we
usually blame on others, and how to identify policies we can follow to improve our situation.
Jay Forrester
Sloan School of Management, Massachusetts Institute of Technology
The what, why and how of system dynamics:
What:
A rigorous way to help thinking, visualizing, sharing, and communication of the future evolution of complex organizations and
issues over time,
Why:
for the purpose of solving problems and creating more robust designs, which minimizes the likelihood of unpleasant surprises
and unintended consequences,
How:
by creating operational maps and simulation models which externalize mental models and capture the interrelationships of
physical and behavioral processes, organizational boundaries, policies, information feedback and time delays; and by using
these architectures to test the holistic outcomes of alternative plans and ideas,
Within:
a framework which respects and fosters the needs and values of awareness, openness, responsibility and equality of individuals
and teams.
Eric Wolstenholme
School of Management, University of Stirling, Scotland.
System dynamics is a framework for thinking about how the operating policies of a company and its customers, competitors, and suppliers
interact to shape the company’s performance over time. System dynamics models are:
1)
Maps, diagrams, words, and friendly algebra to activate and capture team knowledge
2)
Frameworks to help organize, filter and structure the vast amount of knowledge that an experienced team shares, and
3)
Microworlds, microcosms of reality, learning environments that managers can use to test, challenge, and refine their own mental
models.
John Morecroft
6 School, U.K.
London Business
The real system…
The trouble with feedback
is that it is often invisible …
…vs. what we often see…
“System dynamics demonstrates how
most of our own decision-making
policies are the cause of the problems
that we usually blame on others, and
how to identify policies we can follow
to improve our situation.”
Jay Forrester
Sloan School of Management, MIT
Edited extract
7
Page 38 of Morecroft, John (2007) Strategic Modelling and Business
Dynamics. Wiley
From Events to Dynamics and Feedback:
Drug-related Crime
"Drugs are a big worry for me, not least
because of the crimes that people
commit to fund their dependency. We
want the police to bust these rings and
destroy the drugs. They say they're
doing it and they keep showing us sacks
of cocaine that they've seized, but the
crime problem seems to be getting
worse".
Typical description of the problem
by the victims of drug-related crime
8
[Morecroft (2007) p 46]
Unintended
Dynamics of Drug-Related Crime
What feedback
structure could
explain this
puzzling divergence?
DrugRelated
Crime
reported
puzzling
divergence
tolerable
Time in Years
9
[Morecroft (2007) p 47]
CLD for Drug-Related Crime
“What feedback structure could explain this puzzling divergence? Reported crime is growing and
we know that growth arises from reinforcing feedback. The persistence of unwanted growth in crime
suggests a feedback loop that weaves its way around society, and by doing so it goes unnoticed.”
Among the variables below, construct a CLD that could create this puzzling divergence.
Event-oriented
thinking.
drug-related crime
Why does price not
influence demand as it
does in most markets?
+
+
call for police
action
demand
+
price
-
Crime Spiral
+
drug seizures
supply
-
Stakeholders
represented?
- Community
- Police
- Drug users
- Drug dealers
10
[Morecroft (2007) p 47-48]
Agenda
• Broad application areas of systems
thinking and dynamic modeling (ST&DM)
• Slinky
• What is ST&DM?
• Drug-related crime
• Modeling example: filling a water glass
• ST&DM at Boeing
• First feedback loops
11
Drug Related Crime: qualitative,
yet mathematical, thinking.
Combined Qualitative &
Quantitative Thinking Example:
Filling a water glass
(Go to Vensim)
12
WaterGlass1.mdl
initial water in glass
height of glass
0
4
Water In
Glass
spigot flow rate
initial water in pan
0
Water in Pan
overflow rate
1
Water in the Glass and Pan
in
5
0
0
2
4
6
8
Time (Second)
10
Water In Glass : Current
Water in Pan : Current
13
WaterGlass2.mdl
Water in the Glass & Spigot Flow Rate
initial water in glass
0
Water In
Glass
+
spigot flow rate
S
4 in
2 in/Second
3 in
1.5 in/Second
B
O
gap
desired water in
glass
3.5
2 in
1 in/Second
1 in
0.5 in/Second
0 in
0 in/Second
0
Water In Glass : Current
spigot flow rate : Current
1
2
3
4
Time (Second)
5
6
7
8
in
in/Second
14
WaterGlass3.mdl
Water in the Glass & Spigot Flow Rate
4 in
2 in/Second
initial spigot flow
rate
initial water in glass
1
0
3 in
1.5 in/Second
Water In
Glass
+
spigot flow rate
2 in
1 in/Second
B
effect of gap on
flow rate
O
gap
desired water in
glass
3.5
1 in
0.5 in/Second
0 in
0 in/Second
0
1
Water In Glass : Current
spigot flow rate : Current
2
3
4
5
Time (Second)
6
7
8
in
in/Second
15
Agenda
• Broad application areas of systems
thinking and dynamic modeling (ST&DM)
• Slinky
• What is ST&DM?
• Drug-related crime
• Modeling example: filling a water glass
• ST&DM at Boeing
• First feedback loops
16
ST&DM in Boeing: Where is it
done?
• The modeling and simulation group Paul Newton (one of the
two presenters of this slide show) belongs to…
– …is part of Boeing Research & Technology, Boeing’s R&D
organization
– …is like an internal consulting firm: fee for service to BCA & BDS
– …does other kinds of modeling and simulation as well
– …contains four people doing ST&DM, with several others
learning, and is hiring (we have growing demand)
– …has 1 PhD, 2 Master’s, 1 Bachelor’s
– …engages summer interns, from HS seniors, to PhD students.
• Elsewhere in Boeing
– Boeing Test & Evaluation: systems thinking to improve
organizational change and performance dynamics
– Scattered interest elsewhere, e.g. BCA, Information Technology
17
ST&DM in Boeing: Dynamic Business
Problems
• Boeing examples:
–
–
–
–
Learning curve dynamics
Aerospace industry dynamics
Future workforce dynamics (STEM)
New business strategy dynamics
• Boeing customer examples:
– Boeing Commercial Airplanes (BCA) customers example:
business strategy dynamics, like People Express shown below
– Boeing Defense Systems (BDS) customers example: better
understanding insurgency & irregular warfare dynamics
• Show two examples:
– Autopilot (airplane design – technical systems)
– People Express Vensim model (business design – social
systems)
18
Proportional “Altitude Hold” Autopilot
•
•
•
•
Desired state – 40,000 ft
Current state – variable
Actions – variable vertical winds
Feedback
wind drafts
changing altitude due t o
wind drafts
Alt it ude
changing altitude
desired altitude
changing altitude due t o
speed increase
desired less act ual
alt it ude adjustm ent
t ime
19
Proportional Autopilot for Holding Altitude During Vertical Wind Drafts
Wind Drafts and Altitude - No Control Loop
wind drafts
42,000 feet
40 miles/hour
changing altitude due
to wind drafts
Altitude
changing altitude
37,000 feet
0 miles/hour
desired altitude
No Control Loop
32,000 feet
-40 miles/hour
0
2
4
6
8
10 12
Time (Minute)
14
16
desired altitude : NoControlLoop
Altitude : NoControlLoop
wind drafts : NoControlLoop
18
20
feet
feet
miles/hour
Wind Drafts and Altitude - With Control Loop
wind drafts
42,000 feet
40 miles/hour
changing altitude due
to wind drafts
With
Control Loop
Altitude
changing altitude
37,000 feet
0 miles/hour
desired altitude
changing altitude due Control Loop
to speed increase
desired less actual
32,000 feet
-40 miles/hour
0
2
4
desired altitude : NoControlLoop
Altitude : NoControlLoop
wind drafts : NoControlLoop
6
8
10 12
Time (Minute)
14
16
18
40000
20
feet
feet
miles/hour
altitude
adjustment time
1
20
available
passenger
miles
S
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
growth rate
S
churn
increase of
potential passengers
S S
O
conversion
ratio
O
Service
Reputation
Burrs personal
growth target
S
Motivation
change in
motivation
S
O relative productivity of
time to change
motivation
loss of potential
passengers
Potential
Passengers
new to experienced staff
S
true service
capacity S
S
S
S O
service quality
S
normal
productivity
time to perceive
quality
effective
experienced staff
S
O O
hidden
coaching
S
O
change in quality
of service
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
turnover
S
Competitor
Fare
Morecroft’s Dynamic Hypothesis
for People Express Airlines
“People Express’ resource accumulation
processes…include a tangible resource system
departures
induction
hiring
that contains three reinforcing feedback loops, each a compelling
O
S
S
engine of growth in its own right...These three growth
S
engines…drive the kind of spectacular growth actually achieved by People
time to gain
max hiring
S
Express….But, the three engines of tangible resource growth are not well coordinated
experience
rate
because the underlying policies governing resource accumulation are so different. As
size of hiring
S
interviews S
S
fleet expansion and passenger growth begin to outstrip staff expansion, problems
team
become evident in the intangibles of perceived service level, customer satisfaction,
S
proportion
S to
priority
and employee motivation. No management action is taken to fix these problems,
accepted
interview rate
however, because: (1) the unmanaged intangible resources initially provide relatively
hiring
weak signals to the rest of the organization of latent growth stresses; and (2) the powerful logics underlying the policies governing tangible resource accumulation are
21
insensitive to such weak signals. This seeming lack of alignment of resource accumulation policies, leading to a virtual paralysis in the face of growing
problems, and, eventually as impending doom, is symptomatic of a loss of management coordination under conditions of dynamic complexity.”3,pages35&36
New
Staff
Experienced
Staff
People Express Airlines Performance
Resource Performance
200
20 B
2
4
plane
passenger*mile/year
fraction
dmnl
100
10 B
1
2
plane
passenger*mile/year
fraction
dmnl
0
0
0
0
plane
passenger*mile/year
fraction
dmnl
1980
1981
1982
1983
1984
Time (year)
1985
Planes : Current
actual passenger miles : Current
true service capacity : Current
Service Reputation : Current
Motivation : Current
Burrs personal
growth target
0.7
1986
1987
1988
plane
passenger*mile/year
passenger*mile/year
fraction
dmnl
priority to hiring
0.2
coaching load
0.3
time to gain
experience
2
time to change
costs
4
22
Agenda
• Broad application areas of systems
thinking and dynamic modeling (ST&DM)
• Slinky
• What is ST&DM?
• Drug-related crime
• Modeling example: filling a water glass
• ST&DM at Boeing
• First feedback loops
23
The following
is from:
published in
1988
24
Elementary
School
25
26
27
28
Intermediate
Grades
The Middle
(Junior High)
School Years.
The High School
Years.
29
Software
• Vensim PLE (free for educational use – to run models
Paul showed, & to create & save your own models)
– Download Vensim PLE (Personal Learning Edition) from
http://vensim.com/freedownload.html
– When installing,
• uncheck the default checkbox that reads "Install Vensim PLE for
evaluation purposes. Use limited to 60 days"
• check the checkbox for "Install Vensim PLE for academic, public
research or personal use. Commercial, proprietary, classified or
operational use not allowed.”
• Stella Trial Version (trial version is save-disabled, but will
run models Dexter showed)
– Download trial from http://www.iseesystems.com/
30
Learning More: ST&DM in K12 Education
•
•
Websites focused on ST&DM in K12 education:
– http://clexchange.org/
• Read this paper: http://sysdyn.clexchange.org/sdep/papers/D-4434-3.pdf
• Conference in June 2010, and not again until summer 2012:
http://clexchange.org/conference/cle_2010conference_registrationinfo.html
• Jay Forrester: founder of the field: http://sysdyn.clexchange.org/people/jayforrester.html
– http://www.watersfoundation.org/
• Among many other things, an online course for teachers
Books:
– Introduction to Systems Thinking with Stella, by Barry Richmond.
http://www.iseesystems.com/store/college_university/books.aspx
– Thinking in Systems – A Primer (2008), by Donella Meadows.
– Strategic Modelling and Business Dynamics: A Feedback Systems Approach (2007), by
John Morecroft
– Modeling the Environment , 2nd edition (2010), by Andrew Ford
– Many more books listed at:
• http://clexchange.org/lom/cle_books.htm
• http://pegasuscom.com/
31
Stuff We Didn’t Get To
(but maybe we’ll cover these another day…)
• People Express Airlines
Example detail
• Paper Folding Exercise
& Modeling
• Fishing in Bonavista,
Newfoundland game
• Standing & holding
hands in a circle loops
exercise
• High School economics
simulation
• Systems Thinking Skills
32
Modeling in HS Science
Examples shown by Dr. Dexter Chapin
Science teacher at Seattle Academy of Arts and Sciences
and
Author of the book, “Master Teachers: Making a Difference on the
Edge of Chaos”
http://www.amazon.com/Master-Teachers-Making-DifferenceChaos/dp/1578868637/ref=reader_auth_dp
Intern story – Dexter’s student Sarah
33
History of People Express Videos
http://blog.flightwisdom.com/2009/07/31/history-people-express/
In the presentation, we will only view one or two of the videos here.
The others are well worth watching to get the whole story.
34
The Rise & Fall of People Express1a
Background
de-regulation of US airline industry in early 1980s
charismatic founder Don Burr
passion for airlines and track record in the industry (credited with the
turnaround of Texas Air)
Spectacular Success
from startup in 1981 to fifth largest US airline in 1986
revenues in excess of $ 1 billion and 5000 employees by 1986
deep discount prices and innovative people management policies
Even More Spectacular Failure
“burned-out and bought out corporate carcass in only six months”2
35
People Express Problem Behavior1b
Passengers & Planes
What caused the success?
6 M pssngrs
200 plane
What caused the failure?
3 M pssngrs
100 plane
0 pssngrs
0 plane
1980
1981
passengers : Current
Planes : Current
1982
1983
1984 1985
Time (year)
1986
1987
1988
pssngrs
plane
John Morecroft, of London
Business School, analyses the
causes using:
1) Resource Based View (RBV)
2) The notion of “dominant
logic”
3) System dynamics (or more
commonly – systems thinking)
36
Toward a dynamic hypothesis:
Tangible and Intangible Resources3,p34
• Tangible
– Planes
– Staff
– Passengers
• Intangible
– Service Reputation
– Staff Morale
Note high degree of aggregation of resources!
Planes, staff and passengers could be greatly
disaggregated, but , for Morecroft, such disaggregation
is not necessary to explain the rise and fall of People
Express.
The rise and fall depends on “dynamic” complexity”
rather than “detail complexity.”
Dynamic complexity is present in business or social
systems whenever cause and effect are subtle or where
the effects over time of interventions are not obvious.
For example, when an action has dramatically different
effects in the short run and the long run, or when the
local consequences of an action differ from
consequences elsewhere in the system, then there is
dynamic complexity.4
37
Toward a Dynamic Hypothesis
Dominant Logic: FLEET EXPANSION
(See P-Ex Planes.ITM1b)
Planes
Pool of Readily Available Used
Planes, so no need to represent
‘Planes in Construction‘.
Planes have long lifetime, so no
need to model outflow.
Plane
Purchases
Burr's
Personal
Growth Target
vision
Target Increase
in Planes
Note: Some “role” selects
information to use for the
“plane purchases” policy!
(conceptual linkage to
VNA?)
38
Toward a Dynamic Hypothesis
Dominant Logic: STAFF EXPANSION
Recruitment Policy
(See P-Ex Staff.ITM1b)
staff
productivity
service
capacity
Experienced
Staff
New Staff
labour market
Induction
Hiring
rigour of
screening
interviews
size of the
hiring team
labour market
Departures
Note: Some “role” selects
information to use for the
“hiring” policy!
(conceptual linkage to
VNA?)
39
Toward a Dynamic Hypothesis
Dominant Logic: PASSENGER GROWTH
Marketing Policy: Word of Mouth
(See P-Ex Passengers.ITM1b)
Fliers with a favourable
impression of People Express
Pool of fliers in
region served by
People Express.
Size depends on
scope of
service and
convenience
(routes and
schedule)
marketing
spend
relative fare
low, low price
Fliers hearing
favourable
comments about
People Express
Increase of
Potential
Passengers
Potential
Passengers
Fliers losing
interest
in People
Express
Pool of fliers
in region
served by
People
Express
Loss of
Potential
Passengers
churn
conversion
ratio
service
reputation
Note: Some “role”
selects information to
use for the “increase of
potential passengers”
policy! (conceptual
linkage to VNA?)
40
Toward a dynamic hypothesis:
Summary of dominant feedback logic for tangible resources
Potential
passengers
Planes
Plane
Purchases
Increase of
Potential
passengers
R
dominant logic: Burr's vision of growth
New Staff
R
dominant logic: word-of-mouth
Experienced Staff
Induction
Hiring
R
dominant logic: selectivity and staff involvement
41
PEOPLE EXPRESS - THE SUCCESS STORY
Simulations of P-Ex Full Model.ITM
42
Toward a Dynamic Hypothesis
ANALYSIS OF INTANGIBLE RESOURCES
service reputation
motivation
43
Toward a Dynamic Hypothesis
SERVICE REPUTATION
An Invisible Intangible Resource – see P-Ex Full Model.ITM
service quality as perceived by the flying
public - based on accumulated experience
and hearsay
Service Reputation
Change of
Service Reputation
Potential
Passengers
dynamic complexity
masks link between
service quality and
potential passengers
gap
current
service quality
passenger
miles
service quality as
experienced
on day of flying
service
capacity
44
Toward a Dynamic Hypothesis
MOTIVATION AND PRODUCTIVITY
More Dynamic Complexity – see P-Ex Full Model.ITM
Staff
Motivation
performance
related factors
stock
options
profit
sharing
growth
profits
fleet size
Change of
Motivation
gap
indicated
motivation
staff
productivity
hard
work culture
work teams &
minimal hierarchy
job rotation &
simple work practices
participation &
responsibility
structural and
cultural factors
Burr's 'people'
precepts
quality of CSMs
& selective
recruiting
45
available
passenger
miles
S
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
growth rate
S
churn
Motivation
change in
motivation
S
O relative productivity of
new to experienced staff
S
true service
capacity S
S
S
S O
service quality
New
Staff
S
interviews S
S
interview rate
S
O
change in quality
of service
normal
productivity
time to perceive
quality
effective
experienced staff
S
O O
hidden
coaching
S
max hiring
rate
S
S
proportion
accepted
conversion
ratio
O
Service
Reputation
S
hiring
increase of
potential passengers
S S
O
Burrs personal
growth target
time to change
motivation
loss of potential
passengers
Potential
Passengers
induction
O
S
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
Competitor
Fare
turnover
Experienced
Staff
time to gain
S
experience
size of hiring
team
S to
priority
hiring
S
departures
S
Morecroft’s Dynamic Hypothesis
based on his reading of the People
Express Case – unfurled bit by bit
on the following slides.
46
available
passenger
miles
S
Planes
plane
purchases
S
target increase in
planes
S
Burrs personal
growth target
Fleet (planes) tangible resource
management policy
S
47
available
passenger
miles
S
Planes
plane
purchases
S
S
actual passenger S
miles
potential
passenger miles
S
Potential
Passengers
increase of
potential passengers
S S
target increase in
planes
S
conversion
ratio
O
Burrs personal
growth target
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
Competitor
Fare
Potential Passenger tangible
resource management policy.
S
48
available
passenger
miles
S
Planes
plane
purchases
S
target increase in S
planes
S
S
load factor
S
actual passenger S
miles
potential
passenger miles
S
Potential
Passengers
increase of
potential passengers
S S
conversion
ratio
O
Burrs personal
growth target
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
Competitor
Fare
More policy for fleet management.
S
49
available
passenger
miles
S
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
churn
loss of potential
passengers
Potential
Passengers
increase of
potential passengers
S S
O
conversion
ratio
O
Service
Reputation
Burrs personal
growth target
service quality
S
O
change in quality
of service
time to perceive
quality
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
Competitor
Fare
Service reputation intangible
resource management policy
S
50
available
passenger
miles
S
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
growth rate
S
churn
change in
motivation
O
time to change
motivation
Motivation
increase of
potential passengers
S S
O
conversion
ratio
O
Service
Reputation
Burrs personal
growth target
S
loss of potential
passengers
Potential
Passengers
S
true service
capacity
S
S
service quality
effective
experienced staff
S
O
change in quality
of service
time to perceive
quality
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
Competitor
Fare
Staff motivation intangible
resource management policy
S
51
available
passenger
miles
S
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
growth rate
S
churn
change in
motivation
O
time to change
motivation
S
Motivation
increase of
potential passengers
S S
O
conversion
ratio
O
Service
Reputation
Burrs personal
growth target
S
loss of potential
passengers
Potential
Passengers
S
true service
capacity
S
S O
service quality
effective
experienced staff
S
O
change in quality
of service
time to perceive
quality
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
Competitor
Fare
Passenger effect on service quality,
hence on service reputation and
potential passenger loss rate.
52
available
passenger
miles
S
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
growth rate
S
churn
Motivation
change in
motivation
S
O relative productivity of
new to experienced staff
S
true service
capacity S
S
S
S O
service quality
New
Staff
S
interviews S
S
interview rate
S
O
change in quality
of service
normal
productivity
time to perceive
quality
effective
experienced staff
S
O O
hidden
coaching
S
max hiring
rate
S
S
proportion
accepted
conversion
ratio
O
Service
Reputation
S
hiring
increase of
potential passengers
S S
O
Burrs personal
growth target
time to change
motivation
loss of potential
passengers
Potential
Passengers
induction
O
S
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
Competitor
Fare
turnover
Experienced
Staff
time to gain
S
experience
size of hiring
team
S to
priority
hiring
S
departures
S
Staff resource tangible resource
management policy
53
available
passenger
miles
S
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
growth rate
S
churn
increase of
potential passengers
S S
O
conversion
ratio
O
Service
Reputation
Burrs personal
growth target
S
Motivation
change in
motivation
S
O relative productivity of
time to change
motivation
loss of potential
passengers
Potential
Passengers
new to experienced staff
S
true service
capacity S
S
S
S O
service quality
S
normal
productivity
time to perceive
quality
effective
experienced staff
S
O O
hidden
coaching
S
O
change in quality
of service
Peoples fare
time to change
O
costs
O
S
relative fare
O
S
change in
competitor fare
turnover
S
Competitor
Fare
Morecroft’s Dynamic Hypothesis
“People Express’ resource accumulation
processes…include a tangible resource system
departures
induction
hiring
that contains three reinforcing feedback loops, each a compelling
O
S
S
engine of growth in its own right...These three growth
S
engines…drive the kind of spectacular growth actually achieved by People
time to gain
max hiring
S
Express….But, the three engines of tangible resource growth are not well coordinated
experience
rate
because the underlying policies governing resource accumulation are so different. As
size of hiring
S
interviews S
S
fleet expansion and passenger growth begin to outstrip staff expansion, problems
team
become evident in the intangibles of perceived service level, customer satisfaction,
S
proportion
S to
priority
and employee motivation. No management action is taken to fix these problems,
accepted
interview rate
however, because: (1) the unmanaged intangible resources initially provide relatively
hiring
weak signals to the rest of the organization of latent growth stresses; and (2) the powerful logics underlying the policies governing tangible resource accumulation are
54
insensitive to such weak signals. This seeming lack of alignment of resource accumulation policies, leading to a virtual paralysis in the face of growing
problems, and, eventually as impending doom, is symptomatic of a loss of management coordination under conditions of dynamic complexity.”3,pages35&36
New
Staff
Experienced
Staff
<normal
productivity>
<Motivation>
potential service
capacity
<New Staff>
<Service
Reputation>
service days per
year
vision of service
time to perceive
service reputation
S
average carriers
per route
passenger miles per
plane day
maximum market
size multiple
route share limit
<flights per year>
maximum
passenger miles
available
passenger miles
S
O
S
growth rate
O
potential
passenger miles
S
S
O
load factor
growth reduction
from load
O
S
Motivation
change of
motivation
capacity
S
S
S
S
S
normal
productivity
O
S
New Staff
induction
hiring
O
S
total staff S
<Time>
date
Peoples fare
relative fare
S
O
change in quality of
service
time to change
costs
O
O
S
O
Competitor
Fare
change in
competitor fare
time to perceive
quality
initial competitor
fare
<total staff>
<Planes>
cost of staff
turnover
Experienced
Staff
S
departures
S
S
initial experienced
staff
size of hiring team
interviews
priority to hiring
interview rate
conversion ratio
O
unit operating cost
time to gain
experience
S
increase of potential
passengers
S S
initial potential
passengers
Service
Reputation
O
effective
experienced
staff
O
S
coaching load
initial new staff
maximum staff per
plane
max hiring rate
churn
O
S
S
hidden coaching
proportion
accepted
initial servvice
reputation
service quality
S true service
Potential
Passengers
loss of potential
passengers
S
initial motivation
time to change
motivation
limit on staff
O
relative productivity of
new to experienced staff
time to perceive
growth
S
S
S
O
initial growth rate
effect of route
saturation
miles per flight
Burrs personal
growth target
target increase in
planes
O
indicated
motivation
S
actual passenger S
miles
load factor
influence time
plane purchases
passengers
route saturation
flights per year
S
Planes
initial planes
<actual passenger
miles>
O
<Experienced
Staff>
rookie fraction
<miles per flight>
operating cost of
planes
staff per plane
<actual passenger
miles>
<Peoples fare>
revenue
cost of service
fraction service
cost
gross profit
Sketch of Morecroft’s
Full Simulation Model
cost multiple
55
Resource Performance
200
20 B
2
4
plane
passenger*mile/year
fraction
dmnl
100
10 B
1
2
plane
passenger*mile/year
fraction
dmnl
0
0
0
0
Simulation of Morecroft’s Dynamic Hypothesis…
…shows that his hypothesis
CAN create
the behavior-over-time of interest!
plane
passenger*mile/year
fraction
dmnl
1980
Planes : BaseSim
actual passenger miles : BaseSim
true service capacity : BaseSim
Service Reputation : BaseSim
Motivation : BaseSim
1981
1982
1983
1984
1985
Time (year)
1986
1987
1988
plane
passenger*mile/year
passenger*mile/year
fraction
dmnl
Does that Morecroft’s hypothesis CAN create the behavior-over-time of
interest mean that his hypothesis is correct?
56
Peter Senge’s Dynamic Hypothesis
for the rise and fall of People’s Express2(Fig1)
Which dynamic hypothesis, Senge’s or
Morecroft’s, is correct?
How do we answer this question?
How does this systems stuff help us then,
if it can’t definitively answer this question?
Having seen this case study, how do you
think we use ST&DM at Boeing?
57
We model interesting behaviors
(problems?), not whole systems
We don’t model the whole system.
Instead we use the systems lens to model interesting (often
problematic) behaviors-over-time.
The system we model contains only those elements of the
whole system that are deemed necessary to give rise to the
behaviors-over-time of interest.
The systems lens is not about answers, but about LEARNING to
ask better questions.
58
System dynamics’
iterative modeling process
1. Problem Articulation
(Boundary Selection)
5. Policy
Formulation
& Evaluation
4. Testing
2. Dynamic
Hypothesis
Monitoring and
Evaluation
3. Formulation
Figure 3-1 Results of any step can yield insights
that lead to revisions in any earlier step (indicated
by the links in the center of the diagram).
From Chapter 3, pages 83 – 105 in Sterman, John (2000) Business Dynamics: Systems Thinking
and Modeling for a Complex World. Irwin McGraw-Hill
59
The Iceberg
Events
Increasing
Leverage
Behavior Patterns,
Change over time
Systemic Structure
60
A leverage point –
where small action yields large results
“Maybe we should write that spot down.”
61
The Iceberg
Events
Increasing
Leverage
Behavior Patterns,
Change over time
Systemic Structure
Mental Models
62
Sometimes we get stuck in our mental models following
rules that don't really exist
“Hey! They’re lighting their arrows!...Can they do that?”
63
available
S passenger miles
S
actual passenger S
miles
potential
passenger miles
S
Planes
plane
purchases
S
S
load factor
target increase in S
planes
S
S
growth rate
churn
Motivation
change in
motivation
S
O relative productivity of
new to experienced staff
true service
S capacity S
S
S
S O
service quality
New
Staff
S
interviews S
S
interview rate
S
O
change in quality
of service
normal
productivity
time to perceive
quality
effective
experienced staff
S
O O
hidden
coaching
S
max hiring
rate
S
S
proportion
accepted
increase of
potential passengers
S S
conversion
ratio
O
Service
Reputation
S
hiring
Potential
Passengers
O
Burrs personal
growth target
time to change
motivation
loss of potential
S
passengers
induction
O
S
Peoples fare
time to change
O
costs
O
S
Three types of
diagrams that
systems thinkers
draw
relative fare
O
S
change in
competitor fare
Competitor
Fare
turnover
Experienced
Staff
S
departures
S
time to gain
S
experience
size of hiring
team
S to
priority
hiring
- Stock and flow diagrams (SFDs), or simply flow diagrams
- Causal loop diagrams (CLDs)
- Hybrid diagrams (add boxes to the stocks in a CLD)
Neither representation is better than the other.
One’s choice depends on the nature of the problem, and one’s
objective for producing the diagram (beyond the scope of today’s
talk).
64
People Express References
1) Morecroft, John (2007) Strategic Modelling and Business Dynamics, Wiley, including
book CD contents for Chapter 6, especially
a) Notes for using P-Ex model components.pdf
b) People Express simulation models
2) Morecroft, John (2009) System Dynamics, RBV, and Behavioural Theories of Firm
Performance: Lessons from People Express. Proceedings of the System Dynamics
Society Conference, Albuquerque, NM 2009. http://www.systemdynamics.org
3) Morecroft, John, Ron Sanchez, Aime Heene (2002) Resource Management Under
Dynamic Complexity, Chapter 2 in Systems Perspectives on Resources, Capabilities,
and Management Processes.
4) Senge, Peter (1990 & 2006) The Fifth Discipline: The Art and Science of the Learning
Organization. Doubleday
5) Morecroft, John, Ron Sanchez, Aime Heene (2002) Integrating Systems Thinking
and Competence Concepts in a New View of Resources, Capabilities, and
Management Processes; Chapter 1 in Systems Perspectives on Resources,
Capabilities, and Management Processes.
65
Stuff We Didn’t Get To
(but could on another day, if you want to)
• People Express
Airlines Example
• Paper Folding
Exercise & Modeling
• Fishing in Bonavista,
Newfoundland game
• Standing & holding
hands in a circle
loops exercise
• High School
economics simulation
• Systems Thinking
Skills
66
Paper Folding Exercise
From
The Systems Thinking Playbook: Exercises to
Stretch and Build Learning and Systems Thinking
Capabilities
By Linda Booth Sweeney and Dennis Meadows
http://www.amazon.com/Systems-ThinkingPlaybook-ExercisesCapabilities/dp/1603582584/ref=sr_1_1?ie
=UTF8&s=books&qid=1267883826&sr=1-1
67
Stuff We Didn’t Get To
(but could on another day, if you want to)
• People Express
Airlines Example
• Paper Folding
Exercise & Modeling
• Fishing in Bonavista,
Newfoundland game
• Standing & holding
hands in a circle
loops exercise
• High School
economics simulation
• Systems Thinking
Skills
68
Ships in
Harbor
Bonavista, Newfoundland
Fishing Simulation Game
initial fish stock
ships moved to
harbor this year
Net Growth Loop
S
Fish in the
Sea
net fish increase
rate
Ships at Sea
catch rate
S
O
S
purchase of new
ships this year
Crowding Loop density Fishing Loop
O
sea carrying
capacity
Natural Fishery Behavior Over Time
4,000 fish
1,000 fish/year
80 ship
4,000 fish
Harvested Fishery Behavior Over Time
2,000 fish
500 fish/year
40 ship
2,000 fish
0 fish
0 fish/year
0 ship
0
5
10
0 fish
0
5
10
Fish in the Sea : natural fishery
15
20
25
Time (year)
30
35
40
fish
Fish in the Sea : harvested fishery
catch rate : harvested fishery
Ships at Sea : harvested fishery
15
20
25
Time (year)
30
35
fish
fish/year
ship
69
Adapted from Morecroft, John (2007) Strategic Modelling and Business Dynamics. Wiley
40
Stuff We Didn’t Get To
(but could on another day, if you want to)
• People Express
Airlines Example
• Paper Folding
Exercise & Modeling
• Fishing in Bonavista,
Newfoundland game
• Standing & holding
hands in a circle
loops exercise
• High School
economics simulation
• Systems Thinking
Skills
70
Standing and Holding Hands in a Circle Feedback
Loops Exercise
From
The Systems Thinking Playbook: Exercises to
Stretch and Build Learning and Systems Thinking
Capabilities
By Linda Booth Sweeney and Dennis Meadows
http://www.amazon.com/Systems-ThinkingPlaybook-ExercisesCapabilities/dp/1603582584/ref=sr_1_1?ie
=UTF8&s=books&qid=1267883826&sr=1-1
71
Stuff We Didn’t Get To
(but could on another day, if you want to)
• People Express
Airlines Example
• Paper Folding
Exercise & Modeling
• Fishing in Bonavista,
Newfoundland game
• Standing & holding
hands in a circle
loops exercise
• High School
economics simulation
• Systems Thinking
Skills
72
Four
Equivalent
Representations
Hydraulic Metaphor:
Stock and Flow Diagram:
Stock
inflow
outflow
Integral Equation:
Stock (t) = Stock (to) +
Differential Equation:
t
t [ Inflow(s) – Outflow(s) ] dt
0
d(Stock)/dt (t) = Net change in stock = Inflow (t) – Outflow (t)
Figure 6-2 Four equivalent representations of stock and flow structure.
73
[Sterman (2000), Chapter 6]
An Intimidating (to most people) Mathematical Model
of the Structural Causes of Business Cycles
1) E = A + cY
4) Y = E* + μ(S* - S)
E is national expenditure ($/time)
A is autonomous expenditure ($/time)
c is the marginal propensity to consume, 0 < c < 1
Y is national output ($/time)
μ is a constant (μ > 0) of proportionality
(1/time)
5) dE/dt = λ(E – E*)
2) dS/dt = Y - E
dS/dt is the net change in the level of inventories
per unit time ($/time)
S is the level of inventories ($)
dE/dt is the change in expected sales per unit time
($/time/time)
λ is a constant (λ > 0) of proportionality
(1/time)
3) S* = hE*
S* is the desired level of inventories ($)
h is a constant (h > 0), called the desired
inventory-sales ratio (time)
E* is expected (or planned) sales ($/time)
Richmond, Barry (1997) Sophisticated Dynamics Without Complex Mathematics, Stella Applications Guide, pp. 49-67
and
Scarfe, Brian L. (1977) Cycles, Growth Inflation: A Survey of Contemporary Macrodynamics.
74
An identical sketch illustrating the same model
for the structural causes of business cycles
fractional correction
desired inventory
coverage
induced spending
income
marginal propensity
to consume
Inventory
national production
national sales
autonomous spending
(= the aggregate
expenditure that
occurs within a
macroeconomy,
independent of the
income of consumers
within the economy)
inventory correction
desired inventory
Expected
Sales
change in
expected sales
fractional adjustment
75
induced spending
income
marginal propensity
to consume
Inventory
national production
fractional correction
desired inventory
coverage
Which representation
would be more meaningful
to most people?
national sales
autonomous spending
inventory correction
desired inventory
Expected
Sales
change in
expected sales
fractional adjustment
76
Sophisticated Dynamics
Without Complex
Mathematics! (Really?)
marginal propensity
to consume
Inventory
national production
fractional correction
Do you see anything
“wrong” with this
model?
induced spending
income
national sales
autonomous spending
inventory correction
desired inventory
coverage
desired inventory
Expected
Sales
Inventory, Production and Sales
change in
expected sales
fractional adjustment
250 $
1,000 $/Month
Structure
150 $
800 $/Month
produces
50 $
600 $/Month
0
6
12
Inventory : Current
national production : Current
national sales : Current
18
24
30
36
Time (Month)
42
48
54
60
$
$/Month
$/Month
Behavior
In response to autonomous
spending increase of 25% at
the fourth month.
77
Four
Equivalent
Representations
Hydraulic Metaphor:
Stock and Flow Diagram:
Stock
inflow
outflow
Integral Equation:
Stock (t) = Stock (to) +
Differential Equation:
t
t [ Inflow (s) – Outflow (s) ] ds
0
d(Stock)/dt (t) = Net change in stock = Inflow (t) – Outflow (t)
Figure 6-2 Four equivalent representations of stock and flow structure.
78
[Sterman (2000), Chapter 6]
Stuff We Didn’t Get To
(but could on another day, if you want to)
• People Express
Airlines Example
• Paper Folding
Exercise & Modeling
• Fishing in Bonavista,
Newfoundland game
• Standing & holding
hands in a circle
loops exercise
• High School
economics simulation
• Systems Thinking
Skills
79
Systems Thinking Skills
• 10,000 meter thinking1
• System as cause
thinking1
• Dynamic thinking1,2
• Operational thinking1,2
• Closed-loop thinking1,2
• Non-linear thinking1
1)
2)
3)
•
•
•
•
•
•
Scientific thinking1,2
Empathic thinking1
Continuum thinking2
Generic thinking2
Structural thinking2
Quantitative Thinking3
Richmond, Barry (2001) Systems Thinking and the Stella Software: Thinking, Communicating, Learning, and Acting More Effectively in the
New Millenium, Chapter 1 in Richmond, Barry (2001) An Introduction to Systems Thinking. Available in print from High Performance
Systems, Inc. Hanover, NH. www.iseesystems.com
Richmond, Barry (1993) Systems thinking: critical thinking skills for the 1990s and beyond. System Dynamics Review Vol. 9 No 2. (Summer
1993) 113-133. Downloadable from www.clexchange.org.
Richmond, Barry (2002) In Search of a Clear Picture for Unifying our Community of Practice. Creative Learning Exchange’s 2002 Systems
Thinking and Dynamic Modeling Conference in Durham, NH. Downloadable from
http://www.clexchange.org/conference/cle_2002conference.htm
80
What distinguishes/defines Systems Thinking is
a unique collection of thinking skills1
10,000 Meters Thinking
System as Cause Thinking
Dynamic Thinking
Operational Thinking
Closed-loop Thinking
Continuum Thinking
Nonlinear Thinking
Quantitative Thinking
Scientific Thinking
Filtering Skills
(what to include, what to omit;
and at what level of aggregation?)
Representing Skills
(stocks, flows, converters,
feedback loops)
Simulating Skills
(internally-consistent numbers;
controlled experiments)
81
1)
This slide copied from Richmond, Barry (2002) In Search of a Clear Picture for Unifying our Community of Practice. Creative Learning Exchange’s
2002 Systems Thinking and Dynamic Modeling Conference in Durham, NH. Downloadable from http://www.clexchange.org/conference/cle_2002conference.htm
Thank you!
Paul Newton
[email protected]
206-544-7641
82

similar documents