Lecture 2: Fundamentals of

Simulation
Outline
 Systems

 Simulation Model Components

 Performance Metrics

 Hand Simulation
 Additional Readings

 Kelton et al., Chapter 1, 2 (except 2.2, 2.7)

 Law et al., Chapter 1 (up to an including
1.4.2)
 Banks et al., Chapter 1
 these slides are mostly based on Kelton et
al. and Law et al. books
Simulation Model Components
 Entities Arena preview: all of
these are used in Arena!
 Attributes
 (Global) Variables
 Resources
 Queues
 Statistical Accumulators
 Events
 Exemple : ATM
Type
Bound/conditio
Events nal
Conditions Changes in states
 Exemple : ATM
Type
Bound/conditio
Events nal
Conditions Changes in states

Arrival of a Add 1 customer to

B
customer queue with "waiting"
status

Starting the "Free" counter Customer 1 switches

C to "using the
use of ATM and at least 1
customer in counter". Counter
line changes to "Busy"

Finishing The client goes to

B
the use of out. The counter
ATM changes to "Free"
 Example: bank branch

The branch remains open between 9 a.m. and 5 p.m. The arrival of customers is
random (20 customers per hour). Two attendants (one young and a second senior,
more experienced and therefore faster).
If the number of waiting customers reaches 3, a second (senior) agent leaves his
activities elsewhere in the branch to open a third counter.
The latter returns to its tasks when, by finishing with a customer, the number of
waiting customers is 1 or 0.
Average Junior Senior
service time Probability Probability
1 0,05 0,1
2 0,1 0,25
The bank wants to predict customer waiting times 3 0,2 0,15
4 0,15 0,2
and wonders about the relevance of other forms of 5 0,2 0,15
organization. 6 0,15
7 0,07
8 0,05
9 0,03
Solution
States:

Customer Employee Extra employee

waiting Idle At counter
Getting service Busy Out
Out

Entities:
Go to bank
• customers
• employees Out Client wait
Events:
• Arrival Service
• Start of service
employee
• Start service (+) If service
• end of service finished, or Idle If three
less than 3 customers
customers waiting
waiting Extra employee

out
Detailed description of events

Event Type Conditions Chenge in the status

 Event Type Conditions Chang. in states

Add 1 customer to queue with

Arival B
"waiting" status

Start of service j = free and at least 1 customer in Customer = in service

C
(with junior) line j= busy

Start of service s = free and at least 1 customer in

C Customer = in service and s = busy
(with senior) line

3 or more customers in line and+ =

elsewhere
Start service (+) C or Customer = in service and+ = busy
2 or more customers in line and + =
in decision

End of service (j) B Customer= out and j = Free

End service (s) B Customer= out and s = Free

End service (+) B Customer= out and + = in decision

<2 customers in line and+ = in

Departure + C + = out
decision
What else do we need to
define for a simulation?
 Starting and stopping

 You need to decide

 starting conditions (initial state)
 stopping conditions

 at a particular time (e.g., 20 minutes)?

 when a particular event occurs (e.g., 100 parts processed)?
Units of Time Arena preview: need
to set base units!
 “Base” time units
 pick whatever is most
appropriate/familiar/convenient
 Be sensible! E.g., don’t use seconds when
you need to simulate 20 years! Don’t use
days when simulating 2 minutes!
 Choose units that avoid extremely small
and extremely large values in the same
model (to avoid round-off errors!)
Performance Metrics
 Performance Metrics
Arena preview: will see in
 Counter statistics data-collecting modules
and in output reports!
 Accumulated sums

 Tally (discrete-time) statistics

 average, minimum, maximum of a list of #s

 Time-persistent statistics
 time-based average, minimum, maximum
 Example 1: Drill Press System
Objective: determine whether we should
run the drill press at a higher speed

Drilling Center
Drill Press
Arriving Departing
Blank Finished
Parts Parts

Queue

Part in
Process
 Drill Press: Metric 1
 Total production (# of parts that
complete production and leave) during
20 minutes
 denote by N
 since it is an accumulated sum, this is an
example of a counter statistic
Arena preview: called
a counter statistic!
©
Drill Press: Metric 2
 Average waiting time in the queue
 Let Wqi be the waiting time of part i
 The average is
 Since averaging over data which is
countable, this is an example of discrete-
time or tally statistic (there is a natural fi
rst,second, . . . Observation)
Arena preview: called a tally
statistic!
34
Metric interpretation

The maximum waiting time in queue of

parts that enter service at the drill press
during the simulation is a worst-case
measure.
which might be of interest in giving service-
level guarantees to customers. Small is
good.
Drill Press: Metric 3
 Time-average number of parts waiting
in the queue
 Let Q(t) be the # of parts in the queue at
any time instant t

 Since averaging over time, this is a time-

persistent statistic Arena preview: called a
time-persistent statistic!
35
Metric interpretation

Such time-persistent statistics are

common in simulation. This one indicates
how long the queue is (on average),
which might be of interest for allocating
floor space.
 Drill Press: Metric 4
 Utilization (another time-persistent stat!)
 Proportion of time resource is busy
 Let B(t) = 1 if the drill press is busy at time t
 Let B(t) = 0 if the drill press is idle at time t
 Define utilization as the area under B(t) divided
by the length of the simulation run:
 Your turn!
 Let Ci be the total time part i
spends in the system (cycle time)
1. Expression for Ci in terms of Wqi?
(Wqi = waiting time of part i)
2. Average time in the system?
3. Define another metric and classify it
1. Ci= Wqi + Process time of item i
2. : C i
Hand simulation
 Single-Server Queue

 We want to simulate a teller in a bank

 Why?
 Is the teller efficient?
 Do we need more tellers?
 Single-Server Queue
 Step through time using lists of
interarrival and service times
 Display system, state variables, clock,
event list, statistical counters… all
after execution of each event
 Stop when number of delays hits n = 6,
compute output performance measure
 Hand Simulation
 We will try to simulate a system ”by
hand” (on paper or in Excel)

 Why? To understand the logic behind

simulation
 A. M. Law, Simulation Modeling and Analysis, Chapter 1

Hand Simulation
Interarrival times:
0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9

 Service times:2.0, 0.7, 0.2, 1.1, 3.7, 0.6,

…

 Simulate until 6 customers ”delayed” (i.e.,

until the 6th customer goes into service)

 Find avg. delay in queue, time-avg. number

in queue, server utilization
 Time :0, time of next event: 0.4

Status
shown is
Hand Simulation
after all
changes
have
been
made in
each
case …

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …

What is the first event to occur?

Q(t) = the # of parts in the queue at any time instant t
Arrival?
B(t) = 1 if the resource is busy at time
Departure?
A. M. Law, Simulation Modeling and Analysis, Chapter 1
Arrival of customer 1
 Time :0.4 Previous job arrival time + inter-arrival
Next arrival time: 1.6 time = 0.4 + 1.2 = 1.6

Arrival
time +
service
time
=
0.4 +
2.0

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …

Customer Arrive Served

at Number that
1 0.4 0.4 started service Total delay time for all
those who started service
A. M. Law, Simulation Modeling and Analysis, Chapter 1
Arrival of customer 2 at 1.6
 Arrival of customer 2 at 1.6

Location of the customer in the queue

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …

Q(t) = [the previous value of Q(t)] +[number in Queue in previous page]*[the time since the last event] =
0+0 *(1.6 - 0.4) = 0
Customer Arrive Served
B(t) = 1 * (1.6-0.4) = 1.2
1 0.4 0.4
2 1.6 ?

A. M. Law, Simulation Modeling and Analysis, Chapter 1

 Arrival of customer 3 at 2.1

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …
The delay counters (total delay) are unchanged, since this is not the end of anyone’s delay
in queue
Customer Arrive Served
Q(t) = 0 + 1(2.1-1.6) = 0.5
1 0.4 0.4
B(t) = 1.2 + (2.1 – 1.6) = 1.7 2 1.6 ?
3 2.1 ?
A. M. Law, Simulation Modeling and Analysis, Chapter 1
 Departure of customer 1 at 2.4
2.4+0.7=3.1

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …

Customer Arrive Served Delay Left

1 0.4 0.4 0 2.4

Q(t) = 0.5 + 2 * (2.4 - 2.1) = 1.1 2 1.6 2.4 0.8 ?

3 2.1 ? ? ?
B(t) = 1.7 + 1 * (2.4 - 2.1) = 2.0

A. M. Law, Simulation Modeling and Analysis, Chapter 1

 Departure of customer 2 at 3.1

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …

Customer Arrive Served Delay Left

1 0.4 0.4 0 2.4
Q(t) = 1.1 + 1 * (3.1 - 2.4) = 1.8 2 1.6 2.4 0.8 3.1
3 2.1 3.1 1 ?
B(t) = 2 + [1 * (3.1 - 2.4)] = 2.7

A. M. Law, Simulation Modeling and Analysis, Chapter 1

 Departure of customer 3 at 3.3

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …

Customer Arrive Served Delay Left

Q(t) = 1.8 + [0*(3.3-3.1)] = 1.8 1 0.4 0.4 0 2.4
2 1.6 2.4 0.8 3.1
B(t) = 2.7+ [1 * (3.3-3.1)] = 2.9
3 2.1 3.1 1 3.3

A. M. Law, Simulation Modeling and Analysis, Chapter 1

 Arrival of customer 4 at 3.8

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …

Customer Arrive Served Delay Left

Q(t) = 1.8 +[ 0 * (3.8-3.3)] = 1.8
1 0.4 0.4 0 2.4
2 1.6 2.4 0.8 3.1
B(t) = 2.9+ [0 * (3.8-3.3)] = 2.9
3 2.1 3.1 1 3.3
4 3.8 3.8 0 ?

A. M. Law, Simulation Modeling and Analysis, Chapter 1 40

 Arrival of customer 5 at 4.0

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …
Customer Arrive Served Delay Left
Q(t) = 1.8 +[ 0 * (4.0-3.8)] = 1.8 (based on 1 0.4 0.4 0 2.4
previous queue array) 2 1.6 2.4 0.8 3.1
3 2.1 3.1 1 3.3
B(t) = 2.9+ [1 * (4.0 – 3.8)] = 3.1
4 3.8 3.8 0 4.9
5 4.0 ? ? ?
6 5.6 ? ? ?
A. M. Law, Simulation Modeling and Analysis, Chapter 1 41
 Departure of customer 4 at 4.9

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …
Customer Arrive Served Delay Left
Q(t) = 1.8 +[ 1 * (4.9-4.0)] = 2.7 1 0.4 0.4 0 2.4
2 1.6 2.4 0.8 3.1
3 2.1 3.1 1 3.3
B(t) = 3.1+ [1 * (4.9-4.0)] = 4.0
4 3.8 3.8 0 4.9
5 4.0 4.9 0.9 8.6
6 5.6 ? ? ?
A. M. Law, Simulation Modeling and Analysis, Chapter 1 42
 Arrival of customer 6 at 5.6

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …
Customer Arrive Served Delay Left
1 0.4 0.4 0 2.4
Q(t) = 2.7 +[ 0 * (5.6-4.9)] = 2.7
2 1.6 2.4 0.8 3.1
3 2.1 3.1 1 3.3
B(t) = 4.0+ [1 * (5.6-4.9)] = 4.7
4 3.8 3.8 0 4.9
5 4.0 4.9 0.9 8.6
6 5.6 ? ? ?
A. M. Law, Simulation Modeling and Analysis, Chapter 1 43
7 5.8
 Arrival of customer 7 at 5.8

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …
Customer Arrive Served Delay Left
1 0.4 0.4 0 2.4
2 1.6 2.4 0.8 3.1
Q(t) = 2.7 +[ 1 * (5.8-5.6)] = 2.9 3 2.1 3.1 1 3.3
4 3.8 3.8 0 4.9
B(t) = 4.7+ [1 * (5.8-5.6))] = 4.9 5 4.0 4.9 0.9 8.6
6 5.6 ? ? ?
A. M. Law, Simulation Modeling and Analysis, Chapter 1 44
7 5.8 ? ? ?
8 7.2
 Arrival of customer 8 at 7.2

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …
Customer Arrive Served Delay Left
1 0.4 0.4 0 2.4
2 1.6 2.4 0.8 3.1
Q(t) = 2.9 +[ 2 * (7.2-5.8)] = 5.7 3 2.1 3.1 1 3.3
4 3.8 3.8 0 4.9
B(t) = 4.9+ [1 * (7.2-5.8))] = 6.3 5 4.0 4.9 0.9 8.6
6 5.6 ? ? ?

A. M. Law, Simulation Modeling and Analysis, Chapter 1 7 5.8 ? ? 45?

8 7.2 ? ? ?
9 9.1 ? ? ?
 Departure of customer 5 at 8.6

Hand Simulation

Interarrival times: 0.4, 1.2, 0.5, 1.7, 0.2, 1.6, 0.2, 1.4, 1.9, …
Service times: 2.0, 0.7, 0.2, 1.1, 3.7, 0.6, …
Customer Arrive Served Delay Left
1 0.4 0.4 0 2.4
2 1.6 2.4 0.8 3.1
Q(t) = 5.7 +[ 3 * (8.6-7.2)] = 9.9 3 2.1 3.1 1 3.3
4 3.8 3.8 0 4.9
5 4.0 4.9 0.9 8.6
B(t) = 6.3+ [1 * (8.6-7.2))] = 7.7
6 5.6 8.6 3 ?
7 5.8 ? ? ?
A. M. Law, Simulation Modeling and Analysis, Chapter 1 8 7.2 ? ? 46
?
9 9.1 ? ? ?
 Final output performance
measures:

Average delay in queue= 5.7/6 = 0.95 minutes/customers

Time-average number in queue = 9.9/8.6 = 1.15 customers

Server utilization = 7.7/8.6 = 0.90 (dimensionless)

Questions?
Randomness in Simulation
Readings

 Rossetti – Section 1.5

 Kelton et al. – Section 2.6
 Now we know the main model
components! To simulate, we need
Example: Drill Press
to know the timing of events…

Drilling Center
Drill Press
Arriving Departing
Blank Finished
Parts Parts

Queue

Part in
Process
 One-Day Input

Part # Arrival Time Inter-arrival Service Time

Time
1 0.00 1.73 2.90
2 1.73 1.35 1.76
3 3.08 0.71 3.39
4 3.79 0.62 4.52
5 4.41 14.28 4.46
6 18.69 0.70 4.36
7 19.39 15.52 2.07
 What conclusion
can we make
One-Day Output about the system?

Performance Measure Value

Total production 5 parts
Avg. waiting time in queue 2.53 minutes per part
Avg. total time in system 6.44 minutes per part
Time-avg. #of parts in queue 0.79 parts
Drill-press utilization 0.92 (dimensionless
proportion)

Is this conclusion valid?

Cannot make valid
conclusion!!!
We evaluated only
one (very short)
scenario!!!!
Scenarios

 Different days will have different inter-

arrival and service times

 Different input = different output!!!

 S1 S2 S3 S4 S5

Metric
Total
Scenarios
5 3 6 2 3
production
Avg. waiting 2.53 1.19 1.03 1.62 0
time in queue
Avg. total time 6.44 5.10 4.16 6.71 4.26
in system
Time-avg. # in 0.79 0.18 0.36 0.16 0.05
queue
Utilization 0.92 0.59 0.90 0.51 0.70

Now we have values for five scenarios… what do

we tell our boss?
 S1 S2 S3 S4 S5 Avg. Sampl
e Std.
Metric Scenarios Dev.
Total 5 3 6 2 3 3.80 1.64
production
Avg. waiting 2.53 1.19 1.03 1.62 0 1.27 0.92
time in
queue
Avg. total 6.44 5.10 4.16 6.71 4.26 5.33 1.19
time in
system
Time-avg. # 0.79 0.18 0.36 0.16 0.05 0.31 0.29
in queue
Utilization 0.92 0.59 0.90 0.51 0.70 0.72 0.18
Scenarios  Replications
 For valid simulation results, need
multiple replications, i.e., multiple runs
of the simulation representing different
scenarios
IMPORTANT: (almost) always need multiple
replications for a dynamic simulation!!!

Arena preview: you need to

specify the # of replications
69
 Uses separate
Replications random numbers
to generate the
 Need to be scenarios

 independent, and
IID

 statistically-identical
Starts/stops
according to the
same rules and
uses the same
input settings
Questions?

Random numbers?
Why “statistical”?
What are “input
settings”?

How many
replications
should we
have?
Modelling Randomness
 We need a structured way to create
multiple scenarios, i.e., to run multiple
replications
 Randomness is captured by
 Random variables
 Distributions
 Generation of
Random Numbers
 Given a distribution, can generate a
sequence of random #s (a particular
scenario!)

 In our drill press example

 Inter-arrival times are exponentially-
distributed with a mean of 5 minutes
 Service times follow a triangular distribution
with a min of 1, mode of 3, max of 6
 Generation of
Random Numbers in Software
 You specify the distribution, it generates
random numbers
 E.g., select “Expo(5)” and it generates the
sequence 1.34, 1.5, 4.3, 0.2, 3.5…
 Different replications will result in different
random numbers from the same distribution
So what do we need to define
for a simulation?
So, you need to define…
 “base” time units
 performance metrics
 entities, attributes, resources, variables,
events, queues, variables…
 input distributions
 start/end conditions
 # of replications
 And what you get are…
 Values for performance metrics over
multiple replications, which you need to
analyze Avg. Std.
Dev.
S1 S2 S3 S4 S5 Metric

Total 3.80 1.64

Metric production
Total 5 3 6 2 3
Avg. 1.27 0.92
production
waiting
Avg. waiting 2.53 1.19 1.03 1.62 0 time in
time in queue queue
Avg. total 5.33 1.19
Avg. total time 6.44 5.10 4.16 6.71 4.26
time in
in system
system

Time-avg. # 0.79 0.18 0.36 0.16 0.05 Time-avg. 0.31 0.29

in queue # in queue

Utilization 0.92 0.59 0.90 0.51 0.70 Utilization 0.72 0.18

 Rough Schema
• “base” time units
• performance metrics
• entities, attributes,
Values for
resources, variables,
performance metrics
events, queues, variables
over multiple
• input distributions
replications, which
• start/end conditions
you need to analyze
• # of replications
Simulation
Model
About distributions
 Random Experiment:

Random Experiment: Random experiment is an experiment in which the

outcome is not known with certainty.
Sample Space: Sample space is the universal set that consists of all possible
outcomes of an experiment. Sample space is usually represented using the letter
‘S’.

Experiment: Flipping a coin

Sample space: {head, tail}

Experiment: rolling a die
Sample space: {1,2,3,4,5,6}
 Random Variable:

Consider drawing two cards from a bag of black and

red cards.
S: {(BB), (BR), (RB), (RR)}

Random variable is a function that maps every outcome in the

sample space to a real number.

Random variables are usually denoted using capital letters, such as X, Y, and
Z, whereas small letters, such as x, y, z, a, b, c, and so on, are used to denote
particular values of random variables written as P(X = x).
Random Variable:
S

P(D=0) = 0.5
P(D=2) = 0.25
P(D=-2) = 0.25
P(D>=0) = 0.75
 Discrete Random Variable:

If the random variable X can assume only a finite or countably

infinite set of values, then it is called a discrete random variable.

Number of orders received at an e-commerce retailer which can

be countably infinite.
Customer churn [the random variables take binary values:
(a) Churn and (b) Do not churn]
Any experiment that involves counting
 Continuous Random Variable:

A random variable X which can take a value from an infinite set of values
is called a continuous random variable.

 Market share of a company (which take any value from an

infinite set of values between 0 and 100%).
 Time to complete an order

In DES we can have both discrete and continues random variable.

Discrete: Number are customers arrive in a bank
Continues: Time of giving service
 Probability Mass Function:

For a discrete random variable, the probability that a random

variable X taking a specific value xi , P(X= xi), is called the
probability mass function P(xi).

That is, a probability mass function is a function that maps each

outcome of a random experiment to a probability.
The probability mass function, P(xi) satisfies the following
conditions:
 Cumulative Distribution
Function of a Discrete
Random Variable
Cumulative distribution function, F(xi), is the probability that the
random variable X takes values less than or equal xi.
F(xi) = P(X ≤ xi)

x f(x) F(x)2
0 0.18 0.18
1 0.39 0.57
2 0.24 0.81
3 0.14 0.95
4 0.04 0.99
5 0.01 1
 Expected Value, Variance, and
Standard Deviation of a Discrete
Random Variable
Expected value (or mean) of a discrete random variable is
given by

Variance of a discrete random variable is given by

Calculate E(x) and var(X)

E(x) =
(0*0.18) + (1*0.39)+(2*0.24)+
(3*014)+(4*0.04)+(5*0.01) = 1.5

Var(x) = [(0-1.5)2 * 0.18] + … +[ (5-1.5)2 * 0.01] = 1.25

 Cumulative Distribution
Function of a Continous
Random Variable
Cumulative distribution function F(a) is the area under the
probability density function up to X = a.

Probability density function

Cumulative distribution function
Expected Value, Variance, and
Standard Deviation of a
Continues Random Variable
 Stochastic process

A stochastic process is a collection of “similar”

random variables ordered over time, which are all
defined on a common sample space.

 If the collection is X1, X2, . . . , then we have a

discrete-time stochastic process.

 If the collection is {X(t), t 0}, then we have a

continuous-time stochastic process.
 Poisson Processes

The stochastic process {N(t), t 0} is said to be a Poisson

process if:

 Customers arrive one at a time

 The number of arrivals in the interval (t, t + s] is independent of

the number of arrivals in the earlier time interval [0, t] and also of
the times at which these arrivals occur.

 The arrival rate of customers does not depend on the time of day,
etc. (stationary)
 Poisson distribution

Probability mass function of a Poisson random variable ( = 4)

Poisson distribution example
 Poisson Processes

If we are interested in predicting the number of events over

a period of time, say between 0 and t, then the probability
mass function is given by

 N(t) is the number of events that occur over a period of time t or in [0, t].
 The parameter is the mean number of occurrences per unit time
 t is the mean number of occurrences in an interval [0, t].
 We also assume that the events are independent.

Example:
Poisson Processes Example
Exponential distribution
 Poisson Processes: relation between
exponential and Poisson distribution

The interarrival times for a Poisson process are IID exponential

random variables;
see Çinlar (1975, pp. 79–80).

THEOREM. If {N(t), t 0} is a Poisson process with

rate , then its corresponding interarrival times A1, A2, .
. . are IID exponential random variables with mean 1/ .
 Example: relation between exponential and
Poisson distribution

The demand for a product follows a Poisson process.

Sample data (50 cases) on time between demands (measured in number of days) are
shown in this table:

Calculate the expected number of demand for next two years

Example: relation between exponential and
Poisson distribution