Simulation Examples by Hand Using Excel
Simulation Examples by Hand Using Excel
Chapter 2
Process
Determine
Characteristics
Key Components
Random
Numbers
Number: between 0 & 1 Variable: some quantity; perhaps from a known distribution
Descriptive
Statistics
Values used for describing a systems and making predictions about its behavior
Note:
Random Variable
A
Randomness
True
Random
Where?
Generates real values: 0 <= val < 1 =RANDBETWEEN (low, high) Generates integers: low <= val <= high To use in Excel IF (RAND ( ) < 0.5, 0, 1) IF (A2 <= 0.33, 0, (IF A2 <= 0.66, 1, 2)
Problem
with Excel.
Random
is a mathematical function
values generated before sequence repeats Seed: Initialization value for a RNG
Carlo Simulation
10
Why not? IF (RAND() <= 0.3, 3, (IF RAND <= 0.75, 6, 10)) 11
Arrival Times
Arrival
Arrival
arrival Inter-Arrival Time: time between successive arrivals Inter-Arrival Time Arrival Time (Clock) Example: Initialize: Clock = 0
3 3 7 2 10 12
12
Infinite vs. Finite population Arrival Rate vs. Effective Arrival Rate
Service Service
(random) Arrivals
Service
defined by a distribution
Time between arrivals inter-arrival time Service times Arrival rate must be less than the service rate. What if it is not? Unstable, explosive
14
Queue Basics
System Event
State
Number & status of entities (units) Circumstance that causes a change in system state
Clock
Relative time
15
Arrive
Queue
Server
Depart
What are the state variables? What are the events? Refer to flow diagrams Pg. 42 +
16
Generate Events
up-front
Before simulation begins OK for small/short simulations
on-the-fly
As needed Used for professional/complex simulations
Generate
Brief Example
Cust # 1 2 3 4 5 6 IAT 0 2 4 3 2 6 A-time *Clock 0 2 6 7 9 15 S-begin *Clock 0 2 6 9 11 15 S-time 2 1 3 2 1 4
18
S-end *Clock 2 3 9 11 12 19
else can we keep track of during the simulation? Wait time in queue Time in system Server idle time Calculate these for previous example.
19
can we calculate at the end of simulation? Average inter-arrival time Average service time Server utilization (% busy) *Average queue length Calculate for previous example.
20
Customer
Server
Current queue length Server status (busy, idle) Customer wait time
1.
Average, max, min queue length Average, min, max service time; utilization Average wait time, max, min
22
2.
2.
3.
3.
Simulation Statistics
Numerous
standard statistics
of interest Some results calculated from parameters Used to verify the simulation Most calculated by program
23
24
More Statistics
Proportion of server busy time = number of time units server busy total time units of simulation Average service Time = total service time number of customers serviced
25
More Statistics
Average time customer spends in system = total time customers spend in system total number of customers Probability a customer has to wait in queue = number of customers who wait
26
Traffic Intensity
A
measure of the ability of the server to keep up with the number of the arrivals TI= (service mean)/(inter-arrival mean) If TI > 1 then system is unstable & queue grows without bound
27
Server Utilization
%
of time the server is busy serving customers If there is 1 server SU = TI = (service mean)/(interarrival mean) If there are N servers SU = 1/N * (service mean)/ (inter-arrival mean)
28
Weighted Averages
Necessary
when unequal probability of values. Example: Service times: 20% take 5 minutes, 38% take 8 minutes, 42% take 11 minutes. What is the average service time? Is it (5 + 8 + 11) / 3 = 8 ???
29
Correct Answer
20% take 5 minutes, 38% take 8 minutes, 42% take 11 minutes. AST = .2 * 5 + .38 * 8 + .42 * 11 = 1 + 3.04 + 4.62 = 8.66
30
Spreadsheet Homework
DUE: Page 78+ (Show all work & document)
# 2: Calculate expected number of customers per day & number of bagels needed: Based in these values, what is expected cost, income, & profit. Dont simulate. # 4: Calculate expected # of calls 9 am 5 pm. & avg. service time. What is utilization of taxi? What is utilization if 2 taxis? Complete an Excel Simulation for 9 to 5 day with 1 taxi. (Print off values & formulas version. Document well.) #51: Calculate best case, worst case, and
31