KING SAUD UNIVERISTY, COLLEGE OF ENGINEERING, RIYADH SAUDI ARABIA

Computer Simulation of Engineering Systems

GE-501

Prof. Adel Abdennour

2011

King Saud University

College of Engineering
Electrical Engineering Department
GE501-Simulation of Engineering Systems
Course Contents
1. Introduction to System Modelling:
a. Principles of modelling
b. Types of systems models: continuous, discrete, deterministics and
stochastic models
c. Sub-models and integrated models
d. Examples of models for engineering systems
2. Principles of Computer Simulation
a. Simulation of system structure and functions
b. Experimental nature of simulation
c. Timing rules
d. Simulation vs analytical methods
e. Performance measures
f. Validation and evaluation of simulation results
g. Simulation examples
3. Random Number Generators
a. Generation of random number for various distribution functions
b. Testing of distribution
c. Monte Carlo simulation approach
4. Simulation Approaches
a. Simulation using high level languages
b. Simulation using general purpose packages
c. Case studies
5. Continuous System Simulation
a. Simulation of basic blocks, time delays
b. Function generation
c. Transient response
d. Output display
e. Case examples representing samples of engineering applications
Page 1 of 34

GE-501 Computer Simulation of Engineering Systems

6. Discrete System Simulation

a. Simulation of discrete events, transactions and facilities
b. Arrival patterns and service time
c. Queuing disciplines
d. Case examples representing samples of engineering systems

Page 2 of 34

GE-501 Computer Simulation of Engineering Systems

Hints on writing technical papers and making presentations

http://ieeexplore.ieee.org/search/freesrchabstract.jsp?tp=&arnumber=762947&queryText%3Dhints+on+writing
+technical+papers%26openedRefinements%3D*%26searchField%3DSearch+All

Page 3 of 34

GE-501 Computer Simulation of Engineering Systems

System
A system is a collection of objects (people, machine etc) joined in some regular
interaction or interdependence in order to serve a common goal.

State
The state of a system is the collection of variables necessary to describe a
system at a particular time, relative to the objectives of the study.

Discrete system
The state variables change instantaneously at separate points in time (e.g. bank).

Continuous system
The state variables change continuously with respect to time.

Physical models
Physical models are based on some analogy between such systems as
mechanical and electrical or electrical and hydraulic. In such models, the system
attributes can be voltage, shaft position etc.

Mathematical models
Mathematical models are based on symbolic notation and mathematical
equations (e.g. differential equations)

Static model
It shows only the values of the attributes when the system is in balance.

Dynamic model
It follows the changes over time.
Page 4 of 34

GE-501 Computer Simulation of Engineering Systems

Analytical model
It uses the deductive reasoning of mathematics to solve a model (only certain
forms of equations can be solved).

Numerical methods
It applies computational procedures to solve equations.

Page 5 of 34

GE-501 Computer Simulation of Engineering Systems

NO
describe the problem

use analytical method

define model

reasonable
?

simulate?

YES
plan the study

write a program

validate model
run model

reasonable

YES
more
runs?

NO
verify results

reasonable
?

The process of simulation

Page 6 of 34

NO

YES

redefine model

stop

GE-501 Computer Simulation of Engineering Systems

system

actual

model

physical

mathematical

dynamic

analytical

numerical

simulation

Advantages of simulation
Useful in absence of analytical models
Allows for estimating results for long time frame (projections).
Allows the change of operating conditions.

Disadvantages of simulation
If the model is not valid the results of simulation are useless.
Simulation models are usually expensive and time consuming.
For stochastic cases, many runs may be required to get good estimate.

Page 7 of 34

GE-501 Computer Simulation of Engineering Systems

System simulation
System simulation is the technique of solving problems by the observation of
performance over time of a dynamic model or system.

The Monte Carlo Method

The Monte Carlo Method is numerical computation method consisting of
experimental sampling with random numbers.
Suppose we have a positive function f(x) with lower and upper limits a and b.
suppose also that the function is bounded above the value.

To solve the integral using Monte Carlo method, we consider the rectangle
containing the function. The sides of the rectangle are c and b-a. We randomly
pick N points uniformly distributed within the rectangle and determine whether
these points are below or above the curve. If we find that n points are on or
below the curve, then we can say that n/N is equal to the area under the curve
divided by area of the rectangle.

Or

The precision of the method increases as the number N increases.

Page 8 of 34

GE-501 Computer Simulation of Engineering Systems

Figure. Monte Carlo Method of finding the area under the curve.

Page 9 of 34

GE-501 Computer Simulation of Engineering Systems

System Dynamics
System dynamics is a major concern in many fields. Understanding how the
response of a system can be changed is the key point to keep system under
control.
Studies of this nature with industrial systems used to be called industrial
dynamic studies. Urban studies led to urban dynamics and more recently, the
study of social and environmental problems led to what is called world
dynamics.
Since there is no major difference in the techniques used in these different areas,
the general name SYSTEM DYNAMICS is now being used.

What is system dynamics study?

System dynamics study is to understand the forces operating in a system in
order to determine their influence on the stability or growth of the system.
The use of the system studies is usually to suggest ways to solve existing
problems within the system or avoid potentially dangerous directions.

Exponential growth models

Growth rate of change-differential equations
Example

Using Laplace,

This growth model with growth rate k is a first order differential equation
whose solution is an exponential function.
Page 10 of 34

GE-501 Computer Simulation of Engineering Systems

Let

If at point A x(t)=a and at B, x(t)=2a, we will notice that the slope at B is twice
the slope at A. this is the nature of exponential growth. The growth rate is
directly proportional to current level.
If we take the logarithm of the variable or plot it in a semi-log paper, we should
obtain a straight line whose slope is the growth rate. Sometimes k is expressed
as,

Where T is the time constant.

Exponential Decay Models

This model is similar to the growth model but in this case, the variable decays
from some initial value x0 at a rate proportional to the current value. Such model
can be interpolated as a negative growth model.

Page 11 of 34

GE-501 Computer Simulation of Engineering Systems

Modified Exponential Growth Models

Let us go back to the simple market model. Suppose this model attempts to
describe how many items of a product will be sold. Obviously, there is a limit to
the market. The exponential growth is, therefore, not a good model because the
growth is not limited.
The exponential growth assumes that the rate is proportional to the existing
level. However, in reality, the rate is proportional to the number of people who
did not buy the product. Suppose that number of people who may buy the
product is X, and x is the number of people who bought the product. The
number of people who did not buy the product yet is X-x therefore the equation
of the model is,

The solution is,

Page 12 of 34

GE-501 Computer Simulation of Engineering Systems

Logistic Curves
The modified exponential curve still has an unrealistic feature. The maximum
slope occurs at the beginning. In the practice, the market starts with a small
slope and then increases as exponential growth and eventually the slope begins
to decline as in the case of the modified exponential growth. The result is an Sshaped curve.

The logistic function is a combination of exponential and modified exponential

functions. The differential equation that describes the logistic function is,

Initially x is small so X-x is almost equal to X

Page 13 of 34

GE-501 Computer Simulation of Engineering Systems

Eventually , X-x is almost equal to 0, X is almost equal to x,

Generalization of Growth Models

Sometimes even when there is a limit such limit may vary due to some changing
variables. In such case, the affecting variables must be included in the model.
As an example if we want to model the number of residential telephones in a
certain country, we may assume the logistic model. However, this market
depends on the number of houses H(t), the personal income P(t) and the service
revenue per telephone S(t). the model therefore can be,

If we have some initial data of the market, we can find the best values of the
constants that fit the data to the model.

Home Phones

year

Time Delays
If a variable has a D time units of delay, it should include the following decay
equation,

In general, the rule is,

Page 14 of 34

GE-501 Computer Simulation of Engineering Systems

This type of delay is called an exponential delay. This representation is over

simplified. Most systems have what is called finite delays.

Page 15 of 34

GE-501 Computer Simulation of Engineering Systems

Probability concepts in Simulation

When the output of a system can be described completely in terms of its input,
the system is said to be deterministic.
When the output is random, the system is said to be stochastic.
We can have discrete or continuous deterministic systems as well as discrete or
continuous stochastic systems.

Discrete probability functions

If a stochastic variable can take I different values xi (i=1,2,3) with the
probability p(xi)then,

Example 1
Suppose we collect data for the number of problems being solved correctly for a
small sample of a class of 30 students.
Number of
problems xi
1
2
3
4
5

Number of
students ni
6
5
3
10
6
Total students=30

Probability p(xi)
0.2
0.167
0.1
0.333
0.2

Cumulative
distribution P(xi)
0.2
0.367
0.467
0.8
1

To estimate the probability of getting xi problems right, we divide the number of

students for xi by the total number of observations N,

The cumulative distribution function is defined as a function that gives the

probability of random variable being less than or equal to a given value. For

Page 16 of 34

GE-501 Computer Simulation of Engineering Systems

discrete data, the CDF is denoted by P(xi). P(xi) by definition increases to a

maximum of 1 with increasing the values of random variable.

Continuous probability function

The probability density is denoted f(x). the probability that x falls in the range
x1 to x2 is given by.

And by definition,

The cumulative distribution function is the probability that the random variable
is less than or equal to x and it is denoted by F(x).

From definition F(x) is a positive number ranging from 0 to 1 and the

probability of x falling in the range x1 to x2 is F(x2)-F(x1)
Example 2
Let the probability density function is given by.

After integration by parts we get,

F(x) is always increasing to a maximum of 1.

Measures of probability functions
Page 17 of 34

GE-501 Computer Simulation of Engineering Systems

There are certain key parameters used to describe probability functions

1- Mean value (m)-Expected value

Using the previous table,

m=(0.2)(1)+(0.167)(2)+(0.1)(3)+(0.333)(4)+(0.2)(5)=3.166
it is the same as.

If the random variable is continuous.

Note: in discrete case, the mean (m) can be a number that cannot actually occur
as in the above example. For this case use the nearest feasible number to mean
value.

2- The mode
The mode is defined as the most frequently occurring value of a random
variable. When the probability density function has a peak, the value of x at
which the peak occurs is called the mode. If the function has more than one
peaks, then the mode is where the highest peak occurs.
In the previous discrete example mode is 4. In previous continuous example.

From here mode is equal to 1.

3- Median
The median is the value of x where the CDF is equal to the half
Page 18 of 34

GE-501 Computer Simulation of Engineering Systems

F(x)=0.5
For discrete example median =3 and for continuous example.

x=1.68
Example number
1
2

Mean
3.166
2

Mode
4
1

Median
3
1.68

4- Variance and standard deviation

The standard deviation is given by.

For continuous random variable.

The standard deviation is a measure of how the data is dispersed.

It is more useful since it is independent of the scale of the data.

Note: The above formula cannot be applied if the random are auto correlated
Note: The data is assumed to be collected independently from time.

Page 19 of 34

GE-501 Computer Simulation of Engineering Systems

Continuous Uniform Random Numbers

Uniform distribution within a certain range means that every point in the range
is equally likely to be chosen.

Computer generation of random numbers

Most of the computer programs generate uniformly distributed random number
from 0 to 1 using the residue method.

= previous random number

P = largest prime number that the machine can process
= positive primitive root of P.

Usually it is difficult to avoid repetition of numbers. However, if variables are

carefully selected, then we can obtain a very long sequence. Because of this
term, pseudo random number generation is used to describe the procedure.
Equally good results can be obtained using.

For 36bit machine

P=335-31
=55
For 32bit machine
P=231-1
=75

Generation of non uniform numbers

There are many ways to generate non uniform random numbers from uniform
random numbers. The easiest of which is called the rejection method; which is
very similar to Monte Carlo method in computing integrals. The only difference
is that f(x) is a probability density function.
Page 20 of 34

GE-501 Computer Simulation of Engineering Systems

Exponential distribution
The CDF of exponential distribution is given by.

Since y is the CDF then it has the values between 0 and 1. If we use uniformly
distributed random numbers for y, then (1-y) will also be uniformly distributed.

The above formula can be used by a computer program to generate

exponentially distributed random numbers from uniform random number
generator.
Note: the exponential distribution is only described by its mean value

. For

exponential distribution mean, standard deviation are equal to . The coefficient

of variance will be equal to 1. If a set of data has coefficient of variance nearly
equal to 1, it is exponentially distributed,

Normal Distribution
The CDF cannot be expressed in terms of a closed form function. However,
many different approximations can be used.
Generate U1 and U2
Let V1=2U1 -1 and V2=2U2 -1 and w= V12+ V22
If w>1 for to the step 1, otherwise Y=(-2ln(w))/w)1/2
X1=V1Y and X2=V2Y
Both Xs are normal random numbers

Page 21 of 34

GE-501 Computer Simulation of Engineering Systems

Chi-Square Tests
It is the oldest goodness of fit hypothesis test.
Divide the entire range of fitted distribution into k adjacent intervals
Let Nj is the number of random variables in jth interval
Let mPj be the expected number of n random variables that would fall in
the jth interval.

Try to have k greater than or equal to 3.

Example
0
1
2
3
4
5
6
7
9
11

59
26
24
18
12
5
4
3
3
2
N=156

Interval

Nj

mPj

1
2
3

0
1,2
3,4,5,6,7,9,11

59
50
47

53.96
58.38
43.66

0.471
1.203
0.256
1.93

From the table of chi square distribution for (n-1)=2 we get test value=4.6 for
10% confidence level. The value of our result is less so it passes the test.

Page 22 of 34

GE-501 Computer Simulation of Engineering Systems

Table: Critical poins for chi-square distribution with v df.

v
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

0.250
0.10153
0.57536
1.21253
1.92256
2.67460
3.45460
4.25485
5.07064
5.89883
6.73720
7.58414
8.43842
9.29907
10.16531
11.03654
11.91222
12.79193
13.67529
14.56200
15.45177
16.34438
17.23962
18.13730
19.03725
19.93934
20.84343
21.74940
22.65716
23.56659
24.47761

.500
0.45494
1.38629
2.36597
3.35669
4.35146
5.34812
6.34581
7.34412
8.34283
9.34182
10.34100
11.34032
12.33976
13.33927
14.33886
15.33850
16.33818
17.33790
18.33765
19.33743
20.33723
21.33704
22.33688
23.33673
24.33659
25.33646
26.33634
27.33623
28.33613
29.33603

0.750
1.32330
2.77259
4.10834
5.38527
6.62568
7.84080
9.03715
10.21885
11.38875
12.54886
13.70069
14.84540
15.98391
17.11693
18.24509
19.36886
20.48868
21.60489
22.71781
23.82769
24.93478
26.03927
27.14134
28.24115
29.33885
30.43457
31.52841
32.62049
33.71091
34.79974

0.900
2.70554
4.60517
6.25139
7.77944
9.23636
10.64464
12.01704
13.36157
14.68366
15.98718
17.27501
18.54935
19.81193
21.06414
22.30713
23.54183
24.76904
25.98942
27.20357
28.41198
29.61509
30.81328
32.00690
33.19624
34.38159
35.56317
36.74122
37.91592
39.08747
40.25602

0.950
3.84146
5.99146
7.81473
9.48773
11.07050
12.59159
14.06714
15.50731
16.91898
18.30704
19.67514
21.02607
22.36203
23.68479
24.99579
26.29623
27.58711
28.86930
30.14353
31.41043
32.67057
33.92444
35.17246
36.41503
37.65248
38.88514
40.11327
41.33714
42.55697
43.77297

Page 23 of 34

0.975
5.02389
7.37776
9.34840
11.14329
12.83250
14.44938
16.01276
17.53455
19.02277
20.48318
21.92005
23.33666
24.73560
26.11895
27.48839
28.84535
30.19101
31.52638
32.85233
34.16961
35.47888
36.78071
38.07563
39.36408
40.64647
41.92317
43.19451
44.46079
45.72229
46.97924

0.990
6.63490
9.21034
11.34487
13.27670
15.08627
16.81189
18.47531
20.09024
21.66599
23.20925
24.72497
26.21697
27.68825
29.14124
30.57791
31.99993
33.40866
34.80531
36.19087
37.56623
38.93217
40.28936
41.63840
42.97982
44.31410
45.64168
46.96294
48.27824
49.58788
50.89218

0.995
7.87944
10.59663
12.83816
14.86026
16.74960
18.54758
20.27774
21.95495
23.58935
25.18818
26.75685
28.29952
29.81947
31.31935
32.80132
34.26719
35.71847
37.15645
38.58226
39.99685
41.40106
42.79565
44.18128
45.55851
46.92789
48.28988
49.64492
50.99338
52.33562
53.67196

GE-501 Computer Simulation of Engineering Systems

Arrival Patterns and Service times

Most systems whether in life or simulation contain processes that end up with
congestion or sometimes called bottle-neck. Examples of these systems are
traffic flow, messages in computer network, customers at bank etc.
The entities waiting to be served are called queue. The area that focuses on such
problems is called queuing theory.
Congestions may be described in terms of three main characteristics:
Arrival Pattern
o Statistical properties of arrivals
Service process
o Service time (time needed for the service)
o Availability (when the service is available)
o Capacity (number that can be served simultaneously)
Queuing discipline
o How to select the next entity to be served.
In simple cases arrivals are independent of time (stationary). In more realistic
cases the arrivals are time dependent (e.g. traffic) (time variant). The problem
is formulated in a stochastic environment where measurements are
approximated to produce probability distributions.

Arrival patterns
The common way of describing arrivals is in terms of inter-arrival time (the
interval between the two successive arrivals). If the arrival pattern is not
variable, then the inter-arrival time is constant. If the arrival pattern is
stochastic, then it is necessary to define the probability function of the interarrival times.
If n arrivals are simultaneous, then n-1 of them will have zero inter-arrival
times.
Let Ta= mean inter-arrival time
Page 24 of 34

GE-501 Computer Simulation of Engineering Systems

= mean arrival rate

We have =1/Ta
Example
An office works 8hours/day 5days/week and it receives 800 phone calls/week.

Ta = 1/3= 0.333 calls/min.

Definition
Let A0(t) = arrival distribution = the probability that inter-arrival time is
GREATER than t.
Let F(t)=cumulative distribution = the probability that inter-arrival time is LESS
than t.
A0(t) = 1 F(t)
Poisson Arrival Patterns (discrete)
Arrivals are modeled to be completely random and they can occur at any time
subject to the restriction that the mean arrival rate be some given value.
In other words, the time of next arrival is independent of the previous arrival
and the probability of an arrival in an interval t is proportional to t. Since is
the arrival rate, the probability of an arrival in t is t. With these
assumptions, the distribution of the inter-arrival times is exponential. The
probability density function of the inter-arrival time is given by.

The probability of n arrivals occurring in a period of length t is.

This distribution is called Poisson distribution. Poisson distribution is discrete

while exponential distribution is continuous.

Page 25 of 34

GE-501 Computer Simulation of Engineering Systems

Example
Consider the following measurements of e-mail messages arrival.
Arrival number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

Arrival time (min)

12.5
15.1
44.1
62.6
65.3
67.6
71.0
92.5
106.5
115.0
136.4
142.7
151.2
162.5
167.2
172.9
179.8
181.6
185.0
194.9

Inter-arrival time
12.5
2.6
29.0
18.5
2.7
2.3
3.4
21.5
14.0
8.5
21.4
6.3
8.5
11.3
4.7
5.7
6.9
1.8
3.4
9.9

1. Is it an exponential growth?
To test this, we arrange the inter-arrival time and plot it on a semilog paper. If
the resulting plot is linear, it is an exponential growth function and exponential
distribution.
2. Mean inter-arrival time

Mean arrival rate == 1/Ta =0.103

Page 26 of 34

GE-501 Computer Simulation of Engineering Systems

Figure. Semilog graph of the above data showing that the above distribution is
exponential.
3. It is the Poisson distribution: t [0 , 200] minutes. Let us take 10
intervals 20 minutes each.
Time period
0-20
20-40
40-60
60-80
80-100
100-120
120-140
140-160
160-180
180-200

Number of arrivals
2
0
1
4
1
2
1
2
4
3

From the above data =0.103 and t=20min.

Page 27 of 34

GE-501 Computer Simulation of Engineering Systems

Using the above formula and computing the probabilities for n=0,1,2,3,
Number of
arrivals in 20
min
0
1
2
3
4
5
A

Actual from
previous table

p(n)

Expected
arrivals

1
3
3
1
2
0

0.128
0.266
0.272
0.187
0.096
0.04

1.3
2.7
2.7
1.9
1.0
0.4

If we compare columns B and D, we notice that they are close. It means that
Poisson distribution is practical in representing arrivals.

Page 28 of 34

GE-501 Computer Simulation of Engineering Systems

Simulation of a Single Server Queuing System

Discrete event simulation

It deals with the systems that change at separate and countable number of points
in time. These points of time mark the occurrence of an event. An event is any
change that alters the state of the system. From the concept point of view,
discrete event simulations could be done by hand calculations. However for
most practical problems, the amount of data that needs to be stored and
manipulated is prohibitive which requires the use of digital computers.

Time advance mechanisms

Discrete event simulations are dynamic in nature. We must keep track of the
current value of the simulated time.
Fixed increment time advance
The clock is advanced at exactly t time. If an event occurred during the
post interval, it will be considered at the end of interval. Note that fixed
interval time advance does not skip inactive periods and it is the waste of
computer time. This mechanism is not popular.
Next event time advance
In this mechanism, the simulation clock is initialized to zero and the
times of future events are determined. In other words, the simulation
clock is advanced from one event time to another. Periods of inactivity
are skipped.

Organization of discrete event simulation

The following components are usually encountered when dealing with discrete
event simulation using the next event time advance mechanism.

Page 29 of 34

GE-501 Computer Simulation of Engineering Systems

System state
Simulation clock
Event list
Initialization
Timing routine
Event routine
Report generation
Main program

The collection of the variables necessary to describe the

system at time t
The current value of simulated time
A list of next time event(s) will occur
A sub-program to initialize the simulation at time 0
A sub-program that determined the next event from the
event list and advances the simulation clock
A sub-program that updates the system state.
A sub-program that measures the performance and
produces a report at the end of the simulation
Program that controls the sub-program, starting from
initialization to report generation.

Problem Statement
Consider a single server queuing system for which the inter-arrival times are
independent, identically distributed (IID) random variables (A1, A2, A3).
The service times S1, S2, S3 are IID and independent of A1, A2, A3
Customers are chosen as FIFO (First In First Out). The simulation will end
when the nth customer enters service.
Definitions
d(n)= average delay in the queue of n customers completing their delays

q(n)= average number of customers in the queue but not being served=
continuous.
Q(t)=number of customers in the queue at the time t.
Ti= Total time in the queue of length i.
T(n)= T0+ T1+ T2

U(n)= Expected utilization (how much server is busy) of the server (0<U(n)<1)
B(t)= 1; if server is busy at time t
0; if server is idle at time t
Page 30 of 34

GE-501 Computer Simulation of Engineering Systems

U(n)>0.9 is a good indicator of congested system. if U(n)=0.3, it means that

server of 70% of the time free.
Evaluation and Analysis of Simulation Results
Consider the following 10 different runs (replications) of a bank queuing
system.

Simulation

Number Finish
served
(time)

Average
Delay (min)

Average
queue
length

1
2
3
4
5
6
7
8
9
10

484
475
484
483
455
461
451
486
502
475

1.53
1.66
1.24
2.34
2.00
1.69
2.69
2.86
1.70
2.60

1.52
1.62
1.23
2.34
1.89
1.56
2.50
2.83
1.74
2.50

8.12
8.14
8.19
8.03
8.03
8.32
8.09
8.19
8.15
8.24

Proportion of
customers
delayed <
5min
0.917
0.916
0.952
0.822
0.840
0.866
0.783
0.782
0.873
0.779

If the only one simulation run is used to draw conclusion, the results may be too
far from the reality due to the random nature of the output. One must know how
to correctly analyze the simulation results.

Confidence interval and estimating means

Let
where n= number of replications

A confidence interval is given by,

Page 31 of 34

GE-501 Computer Simulation of Engineering Systems

A 90% confidence interval can be calculated using

where n=10. In other words, when we take 10 replications of the simulation

results, a confidence interval for the mean is,

Example
Consider the simulation results for the bank queuing system. We would like to
find the approximate 90% confidence interval for the expected average delay of
a customer over a day.

Example
We would like to obtain a point estimate and an approximate 90% confidence
interval for the expected proportion of customers with a delay less than 5min.

Page 32 of 34

GE-501 Computer Simulation of Engineering Systems

Table : t- distribution table.

p
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
inf

0.600
0.324920
0.288675
0.276671
0.270722
0.267181
0.264835
0.263167
0.261921
0.260955
0.260185
0.259556
0.259033
0.258591
0.258213
0.257885
0.257599
0.257347
0.257123
0.256923
0.256743
0.256580
0.256432
0.256297
0.256173
0.256060
0.255955
0.255858
0.255768
0.255684
0.255605
0.253347

0.750
1.000000
0.816497
0.764892
0.740697
0.726687
0.717558
0.711142
0.706387
0.702722
0.699812
0.697445
0.695483
0.693829
0.692417
0.691197
0.690132
0.689195
0.688364
0.687621
0.686954
0.686352
0.685805
0.685306
0.684850
0.684430
0.684043
0.683685
0.683353
0.683044
0.682756
0.674490

0.900
3.077684
1.885618
1.637744
1.533206
1.475884
1.439756
1.414924
1.396815
1.383029
1.372184
1.363430
1.356217
1.350171
1.345030
1.340606
1.336757
1.333379
1.330391
1.327728
1.325341
1.323188
1.321237
1.319460
1.317836
1.316345
1.314972
1.313703
1.312527
1.311434
1.310415
1.281552

0.950
6.313752
2.919986
2.353363
2.131847
2.015048
1.943180
1.894579
1.859548
1.833113
1.812461
1.795885
1.782288
1.770933
1.761310
1.753050
1.745884
1.739607
1.734064
1.729133
1.724718
1.720743
1.717144
1.713872
1.710882
1.708141
1.705618
1.703288
1.701131
1.699127
1.697261
1.644854

Page 33 of 34

0.975
12.70620
4.30265
3.18245
2.77645
2.57058
2.44691
2.36462
2.30600
2.26216
2.22814
2.20099
2.17881
2.16037
2.14479
2.13145
2.11991
2.10982
2.10092
2.09302
2.08596
2.07961
2.07387
2.06866
2.06390
2.05954
2.05553
2.05183
2.04841
2.04523
2.04227
1.95996

0.990
31.82052
6.96456
4.54070
3.74695
3.36493
3.14267
2.99795
2.89646
2.82144
2.76377
2.71808
2.68100
2.65031
2.62449
2.60248
2.58349
2.56693
2.55238
2.53948
2.52798
2.51765
2.50832
2.49987
2.49216
2.48511
2.47863
2.47266
2.46714
2.46202
2.45726
2.32635