11/09/2021

Mathematical Part 2: Dynamic Model

Modelling

1 2

Content of part 2
Outline
• Eigenvalue method
Chapter 1: Introduction to dynamic models

Chapter 2: Analysis of dynamic models • Eigenvalue methods for discrete systems

Chapter 3: Simulation of dynamic models

3 4

3 4

1
 11/09/2021

Eigenvalue method Step 1: Ask a question

• Reconsider the tree problem example in last lesson
In an unmanaged tract of forest area, hardwood and softwood trees
compete for the available land and water. The more desirable
hardwood trees grow more slowly, but are more durable and produce
more valuable timber. Softwood trees compete with the hardwoods by
growing rapidly and consuming the available water and soil nutrients.
Hardwoods compete by growing taller than the softwoods can and
shading new seedlings. They are also more resistant to disease.
Assume that hardwoods grow at a rate of 10% per year and softwoods
at a rate of 25% per year. An acre of forest land can support about
10,000 tons of hardwoods or 6,000 tons of softwoods. The extent of
competition has not been numerically determined. Can both types of
trees coexist in stable equilibrium? • Note that, we assume additionally that 𝑟 = 0.1, 𝑟 = 0.25, 𝑎 =
. .
,𝑎 =

5 6

5 6

Step 2: Select a modelling approach Step 2: Select a modelling approach

• We will model this problem as dynamical system and
apply eigenvalue method to do sensitivity analysis • If any part of eigenvalue has a positive real
• Suppose we are given a dynamical system 𝑥 = 𝐹(𝑥) part, then the equilibrium is unstable.
where 𝑥 = 𝑥 , … , 𝑥 is an element of state space 𝑆 ⊂
𝑅 and 𝐹 = 𝑓 , … , 𝑓 . A point 𝑥 ∈ 𝑆 is an equilibrium • The eigenvalue method is based on a linear
or steady state if only if 𝐹 𝑥 = 0. approximation. Even if 𝑥 = 𝐹(𝑥) is not linear,
• There is a theorem that states that an equilibrium point we will have
𝑥 is asymptotically stable if the matrix have
eigenvalues with all negative real parts.
in the neighborhood of the equilibrium point.

7 8

7 8

2
 11/09/2021

Step 3: Formulate the problem Step 4: Solve the problem

• Let 𝑥 = 𝐻 and 𝑥 = 𝑆 denote our two state
variables, defined on the state space
{ 𝑥 , 𝑥 : 𝑥 ≥ 0, 𝑥 ≥ 0} Solving this system, we have three points
• The steady-state equations are

and a point is intersection of two following lines

9 10

9 10

Step 4: Solve the problem Step 4: Solve the problem

• The dynamical system equations 𝑥 = 𝐹(𝑥) where 𝐹 =
(𝑓 , 𝑓 ) and

where 𝐷 = 𝑎 𝑎 − 𝑏 𝑏 . Now we have specified values

for 𝑎 , 𝑎 , 𝑟 , 𝑟 but not for 𝑏 , 𝑏
• We continue to assume that 𝑏 < 𝑎 and 𝑏 = 𝑎 . Then
the coordinates of the equilibrium point are 𝑥 =
𝑥 , 𝑥 , where

11 12

11 12

3
 11/09/2021

Step 4: Solve the problem Step 4: Solve the problem

• The partial derivatives • We obtain the matrix

• The eigenvalue of this 2x2 matrix can be computed as

the roots of the equation

13 14

13 14

Step 4: Solve the problem Step 5: Answer the question

We have found that hardwoods and softwoods can
coexist in stable equilibrium. There will be approximately
9,300 tons per acre of hardwoods and 1,300 tons per
acre of softwoods in a mature, stable forest. These
conclusions are based on certain plausible assumptions
about the degree of competition between the two types
of trees. A sensitivity analysis will be conducted to
determine the effect of these assumptions on our broad
conclusions.

15 16

15 16

4
 11/09/2021

Sensitivity analysis Sensitivity analysis

For the sensitivity analysis, we will still assume that 𝑏 = • The differential equations of this system are
𝑡𝑎 but we will relax the assumption that 𝑡 = . The 𝑥 = 𝑓 𝑥 , 𝑥 where
conditions

imply that 0 < 𝑡 < 0.6. The coordinates of the equilibrium

point (𝑥 , 𝑥 ) are

17 18

17 18

Sensitivity analysis Sensitivity analysis

And the partial derivatives Substituting back we get

The characteristic equation we must solve to find

eigenvalue is

19 20

19 20

5
 11/09/2021

Sensitivity analysis Sensitivity analysis

• Solving above equation we get

Graph shows a graph of 𝜆1 and 𝜆2 versus t over the

interval 0 < 𝑡 < 0.6. From this graph we can see that
𝜆1, 𝜆2 are always negative, so that the equilibrium is stable
regardless of the strength of competition.

21 22

21 22

Outline Eigenvalue methods for discrete systems

• Eigenvalue method

• Eigenvalue methods for discrete systems

23 24

23 24

6
 11/09/2021

Illustrated example: Docking problem Illustrated example: Docking problem

Astronauts in training are required to practice a Move the acceleration control so that it is opposite to
docking maneuver under manual control. As a part the closing velocity (i.e., if closing velocity is positive,
of this maneuver, it is required to bring an orbiting we slow down, and we speed up if it is negative) and
spacecraft to rest relative to another orbiting craft. proportional in magnitude (i.e., we brake twice as hard
The hand controls provide for variable acceleration
and deceleration, and there is a device on board if we find ourselves closing twice as fast). After a time,
that measures the rate of closing between the two look at the closing velocity again and repeat the
vehicles. The following strategy has been proposed procedure. Assume that it takes 5 seconds to make
for bringing the craft to rest. First, look at the closing the control adjustment, and another 10 seconds until
velocity. If it is zero, we are done. Otherwise, we can return from other tasks to observer the
remember the closing velocity and look at the velocity indicator once again. Under these conditions,
acceleration control. will our strategy for matching velocities be successful?

25 26

25 26

Step 1: Ask a question Step 2: Select a modelling approach

• We will use model this problem as discrete time

dynamical mode, and apply eigenvalue method to
analyze the sensitivity.

where 𝑐 = 5, 𝑤 = 10, k = 0.02 (additional assumptions)

27 28

27 28

7
 11/09/2021

Step 2: Select a modelling approach Step 2: Select a modelling approach

29 30

29 30

Step 2: Select a modelling approach Step 2: Select a modelling approach

As in the continuous case, the eigenvalue method for discrete Therefore, the behavior of our original nonlinear system
time dynamical systems is based on a linear approximation. in the neighborhood of 𝑥 is approximately the same as
Even if the iteration function G(x) is not linear, we will have the behavior of the linear discrete time dynamical system
defined by the iteration function

in the neighborhood of the origin. The linear

in the neighborhood of the equilibrium point x0. In other words approximation is good enough so that if the origin is a
the behavior of the iteration function G in the neighborhood of stable equilibrium of the linear system, then 𝑥 is a
the equilibrium point x0 is approximately the same as the stable equilibrium of the original nonlinear system. So it
behavior of the linear function Ax near the origin. only remains to discuss the conditions for stability of the
linear system.

31 32

31 32

8
 11/09/2021

Step 2: Select a modelling approach Step 3: Formulate the problem

• In this case we already have a linear system with
equilibrium 𝑥 = (0,0). The iteration function is

33 34

33 34

Step 4: Solve the problem Step 5: Answer the question

• We calculate We assumed 15 seconds between control
adjustments; 5 seconds to make the adjustment
and 10 seconds slack time. Using a correction
factor of 1 : 50, we can guarantee success for
our proportional method of control. In practical
terms, a correction factor of 1 : 50 means that if
the velocity indicator reads 50 m/sec we will set
the acceleration controls for -1 𝑚/sec ; if the
reading is 25 𝑚/sec , we set the controls at -0.5
𝑚/sec , and so on.

35 36

35 36

9
 11/09/2021

Sensitivity analysis Sensitivity analysis

• Graph of 𝜆 and 𝜆 over the interval 0 < 𝑘 < 𝑘

• We see that both eigenvalues have absolute value less than 1, so

that the equilibrium (0,0) is stable over this entire range of 𝑘

37 38

37 38

Sensitivity analysis Sensitivity analysis

• For k < 𝑘 < 𝑘 both eigenvalues are complex, and the • Graph of 𝜆 and 𝜆 over for 𝑘 ≤ 𝑘
condition for stability is that

• Which reduces to 𝑘 < .

• It is easy to see that the smaller eigenvalue λ2 has

absolute value greater than one for all such 𝑘

39 40

39 40

10