Business Dynamics y and System y Modelingg
Chapter 8: Linking Feedback with Stockk & Flow l Structure Pard Teekasap Southern New Hampshire University
Outline 1. 2. 3. 4. 5 5.
First-order linear feedback systems Positive feedback and exponential growth Negative feedback and exponential decay Multiple-loop systems SS-Shaped Shaped growth
Quiz Take k an ordinary d sheet h off paper. Fold ld it in h half. lf Fold the sheet in half again. The paper is still less than a millimeter thick. • If yyou were to fold it 40 more times,, how thick would the paper be? times, how thick • If you folded it a total of 100 times would it be? ☺ Only O l iintuitive t iti estimate, ti t no need d ffor calculator l l t ☺ Give your 95% confidence interval
Paper Folding • 42 Folds = 440,000 kms thick More than the distance from the earth to the moon
• 100 Folds = 850 trillion times the distance from the earth to the sun
First order Linear Feedback System First-order • Order of a system of loop is the number of state variables • Linear systems are systems in which the rate equations are linear combination of the state variables and any exogenous inputs • dS/dt = Net Flow = a1S1+aa2S2+…+a … anSn+b b1U1+b b2U2+…+b … bmUm
Basic Structure and Behavior Goal
State of the System
State of the System
Time
Time + State of the System Net Increase Rate +
R
+ State of the S t System
B
Goall G (Desired State of System)
Discrepancy
Corrective Action +
+
Positive Feedback and Exponential Growth • First-order positive feedback loop • The state of the system accumulates its net inflow rate • The h new iinflow fl d depends d on the h state off the h system
Structure for first-order, linear positive feedback system
Solution for the linear first-order system Net inflow = gS = dS/dt
dS = gdt dt S dS = gdt ∫S ∫ ln(S ) = gt + C S(t) = S(0)exp(gt) S = State; g = fractional growth rate (1/time)
Phase plot diagram for the first-order, linear positive feedback dS/dt = Net Inflow Rate = gS
Net Inflow w Rate (units/ttime)
.
g 1
0 0
State of the System (units)
Unstable Equilibrium
Exponential growth: Phase plot VS Time plot • Fractional growth rate g = 0.7% St Structure t 10
Behavior
1024
t = 1000
6 t = 900
4 t = 800
2
t = 700
State of the System (Units)
8
768
0
128 256 384 512 640 768 896 1024 State of System (units)
7 68 7.68
640
6.4
512
5.12
384
3.84
256
2.56
128
Net Inflow (right scale)
0
0
8.96
State of the System (left scale)
0
200
400
Time
600
800
1.28
0 1000
Ne et Inflow (uniits/time)
Net Inflow w (units/time e)
896
10.24
Rule of 70 • Exponential growth has the property that the state of the system y doubles in a fixed period p of time • 2S(0) = S(0)exp(gtd) • td = ln(2)/g • td = 70/(100g) • E.g. E an iinvestment t t earning i 7%/ 7%/year d doubles bl iin value after 10 years
Misperception of Exponential Growth: it’s not linear Time Horizon = 0.1t d
2
State of the Sys stem (units)
State of the Sys stem (units)
2
Time Horizon = 1t d
0 0
2
4
6
8
Time Horizon = 10t d
0
20
40
60
80
100
8000
10000
Time Horizon = 100t d
30
1 10
State of the System (units)
State of the Sy ystem (units)
1000
0
10
0
0 0
200
400
600
800
1000
0
2000
4000
6000
Negative Feedback and Exponential Decay
Negative feedback • Net Inflow = - Net Outflow = -dS d = fractional decay rate (1/time). It is the average lifetime of units in the stock • S(t) S( ) = S(0) S(0)exp(-dt) (d) • This system y has a stable equilibrium. q Increasing the state of the system increases the decay rate rate, moving the system back toward zero
Phase plot for exponential decay Net Inflow Rate = - Net Outflow Rate = - dS
Net Inflo N ow Ratte (units s/time)
Stable Equilibrium 0
State of the System (units) 1 -d d
Exponential decay: Phase plot VS Time plot Structure
0 t = 40
Net Inflo ow (units/tim me)
t=30 t = 20
t = 10
Behavior
100 t=0
10
-5 0
20 40 60 80 State of System (units)
100
50
5
Fractional decay rate d = 5%
Net Inflow (right scale)
0 0
20
40
60 Time
80
0 100
Net Inflow (un nits/time)
State of the System (left scale)
Exponential decay with the goal not zero • In general, the goal of the system is not zero and should be made explicit p • Net Inflow = Discrepancy/AT = (S*- S)/AT • S* = d desired i d state off the h system, A AT = adjustment time or time constant • AT represents how quickly the firm tries to correct the shortfall
First-order linear negative feedback system with explicit goal General Structure dS/dt Net Inflow Rate +
S State of the System
S* Desired State of the System
B -
AT Adustment Time
Discrepancy (S* - S)
dS/dt = Net Inflow Rate dS/dt = Discrepancy/AT dS/dt = (S* - S)/AT
+
Examples Desired Inventory
Inventory Net Production Rate +
B -
AT Adustment Time
Inventory Shortfall
+
Desired Labor Force
Labor -
Net Production Rate = Inventory Shortfall/AT = (Desired Inventory - Inventory)/AT
Net Hiring Rate +
AT Adustment Time
B Labor Shortfall
+
Net Hiring Rate = Labor Shortfall/AT = (Desired Labor - Labor)/AT
Phase plot for first-order linear negative feedback system with explicit g goal Net Inflow Rate = - Net Outflow Rate = (S* - S)/AT
Net Inflo ow Rate (units//time)
1 -1/AT 0 S*
Stable Equilibrium State of the System (units)
Exponential approach to a goal Sta ate of the e System m (units))
200
100
0 0
20
40
60
80
100
Time constants and half half-lives lives • S(t) = S* - (S* - S(0))exp(-t/AT) • 0.5 = exp( exp(-tth/AT) = exp( exp(-dt) dt) • th = ATln(2) = ln(2)/d ≈ 0.70AT = 70/(100d)
Goal seeking behavior Goal-seeking 2000
Labor Force (peo ople)
1750
1500 1250
1000
Net Hiriing Rate (people e/week)
1. AT = 4 weeks 2. AT = 2 weeks
Desired Labor Force
0
2
4
6
8
10
12
14
16
18
20
22
24
0
2
4
6
8
10 12 14 16 Time (weeks)
18
20
22
24
0
Goal seeking behavior Goal-seeking 2000
Labo or Force (pe eople)
1500 1000 Desired Labor Force 500 0
Net Hiriing Rate (people e/week)
AT = 4 weeks Does the workforce equal the desired workforce?
0
2
4
6
8
10
12
14
16
18
20
22
24
0
2
4
6
8
10
12
14
16
18
20
22
24
0
Time (weeks)
Solution
Solution
Multiple loop Systems Multiple-loop • Assuming that we disaggregate the net birth rate into a birth rate BR and a death rate DR • • • •
Population = INTEGRAL(Net Birth Rate, Population (0) Net Birth Rate = BR - DR Net Birth Rate = bP – dP = (b-d)P b = fractional birth rate, d = fractional death rate
Phase plot for multiple linear firstorder loops Structure (phase plot)
1
Birth Rate
Net Birth Rate 1 b-d
0 Death Rate
E Exponential ti l Growth G th
b
1
Population
Birth and Death Rates
b>d
Behavior (time domain)
-d
0 0
Population
0 Equilibrium
Birth Rate Net Birth Rate 0
Popu ulation
Birth and Death Rates
b=d
Death Rate
0
Population
0
0
b
Time Exponential Decay
Birth Rate 0 Net Birth Rate
Popu ulation
Birth and Death Rates
Time
Death Rate 0 0
Population
0
Time
Nonlinear first-order systems: SShaped growth • No real quantity can grow forever. It will eventuallyy approach pp the carrying y g capacity p y of its environment • As the system approaches its limits to growth growth, it goes through a nonlinear transition from a regime where positive feedback f dominates to a regime where negative feedback dominates • It’s a S-Shaped growth
Diagram for population growth with a fixed environment • Net Birth Rate = BR – DR = b(P/C)P – d(P/C)P Population Birth Rate +
R
B
+
Death Rate
+
+
B Fractional Birth Rate -
+ Population Relative to Carrying Capacity Carrying Capacity
B Fractional Death Rate +
Nonlinear birth and death rate
Fractiona al Birth an nd Death R Rates (1/tim me)
• Sketch the graph showing the likely shape of the fractional birth and death rate
0 0
1
Population/Carrying Capacity (dimensionless)
Large
Fractiional Birrth and Death R Rates (1/time)
Nonlinear relationship between population density and the fractional ggrowth rate Fractional Birth Rate
Fractional Death Rate
0 0
1
Fractional Net Birth Rate Population/Carrying Capacity (dimensionless)
Phase plot for nonlinear population system Positive Feedback Dominant
Negative Feedback Dominant
Birrth and D Death Ra ates (individu uals/time e)
Death Rate Bi th Rate Birth R t 0
•
(P/C)inf
0
Unstable q Equilibrium
•1
Stable Equilibrium
Net Birth Rate Population/Carrying Capacity (dimensionless)