Simulation

Pseudo-Random Numbers
Simulation  Bec. computer program generates random numbers for mapping, numbers aren’t truly generated in
- representation of real system random fashion.
- employed management science techniques. - However, using standard statistical tests, numbers can show to appear drawn from random
- used to model random processes that are too complex to be solved by analytical methods. process. (Pseudo-random numbers)
- begins by mathematical statement of the problem.  In Excel, value can be obtained for probabilistic input that is normally distributed using:
 Some input values for a simulation model are constants and are referred to as the parameters of
the model. =NORMINV(RAND(),Mean, Standard Deviation)

Probalistic iputs TIME INCREMENTS

- inputs not known with certainty  Fixed- time simulation model
- base on estimate only - time period (incremented by fixed amount)
- for each time period a different set of data from input sequence is used to calculate effects on
2 TYPES OF PROBABILISTIC INPUTS: model
 Controllable inputs
- Capacity mgt dictate happens on input  Next- event simulation model
- time period (not fixed)
 Uncontrollable inputs - determined by data values from input sequence
- (eg. arrival per day)
- Base on estimate EXPERIMENTAL DESIGN
Experimental design
Risk analysis - important consideration (simulation process)
- process of predicting outcome of decision in face of uncertainty.  Issues such length of time (simulation) & treatment of initial data outputs from model must
 One approach to risk analysis is what-if analysis. addressed prior to collecting & analyzing output data
 Interested in result for steady state (long-run) operation of system being modeled
What-if analysis
- generating values for probabilistic inputs & computing the resulting value for output. Start-up period
- Initial data inputs to simulation
Worst-case scenario  (Important) data output for this start-up period be neglected for predicting this long-run period
- examined by (use) most pessimistic input values  For each policy under consideration by decision maker, simulation is run by considering a long
sequence of input data values (given by pseudo-random number generator).
Best-case scenario  Whenever possible, different policies should compare using same sequence of input data.
- studied using most optimistic input values.
SIMULATION APPLICATIONS
Monte Carlo simulation New Product Development
- random variable values are utilized in model through technique - OBJ: determine probability that new product will profitable
 Each random variable, mapped to a set of numbers so each time 1 # in that set is generated, the - A model- developed relating profit (output measure) to various probabilistic inputs (eg.
correspond value of random variable is given as input to the model. demand, parts cost, & labor cost)
 Mapping is done, (way) particular number is chosen is the same as the probability that correspond  Only controllable input- is whether to introduce the product
value of random variable occurs.  Variety of possible values will generated for probabilistic inputs, & resulting profit will be computed.

Arrive Overbooking
- OBJ: determine # of reservation an airline should accept for particular fight
- Simulation model- developed relating profit for flight to probabilistic input, # of passenger w/
reservation who show up, & controllable input, # of reservation accepted for flight
Inventory Policy
- OBJ: choose inventory policy that provide good customer service at a reasonable cost. They make a $.04 profit per postage stamp. The vending machine holds 230 stamps & it costs $4.00 in
- A Model- developed relating 2 output measures, total inventory cost & service level, to labor to fill the machine. They will fill the machine at the beginning of every 5th day. Conduct a 20-day
probabilistic inputs, (eg. product demand & delivery lead time from vendors), and simulation & determine expected profit per day. Assume machine must filled on the first day.
controllable inputs, (eg. order quantity and the reorder point)
- For each setting of the controllable inputs, possible values would be generated for the
probabilistic inputs, & resulting cost & service levels would be computed

Waiting lines
- OBJ: determine waiting times for customers at a bank’s automated teller machine (ATM).
- A model- developed relating customer waiting times to probabilistic inputs (eg. customer
arrivals & service times, & controllable input, # of ATM machines installed.

Example: Dynogen, Inc.

o The price change of shares of Dynogen, Inc.has been observed over the past 50 trades. The
frequency distribution is as follows:
If the current price per share of Dynogen is 23, use
random numbers to simulate the price per share over the
next 10 trades. Use the following stream of random
numbers:
Would filling every 4th day be more profitable?
- The simulation results suggest that they should fill the stamp machine every 5th day, rather
than every 4th day.

 First get RF by dividing total to each # of trade

Example: Mark Koff’s Process

o Mark Koff is a specialist at repairing large metal-cutting machines that use laser technology. His
repair territory consists of the cities of Austin, San Antonio, & Houston. His day-to-day repair
assignment locations can be modeled as a Markov process. The transition matrix is shown below

Assume Mark is currently in Houston. Simulate where Mark will be

over the next 16 days. What percentage of time will Mark be in
each of the three cities? Use the following random numbers:

Based on Probability distribution, expected price change per trade can calcu by: (Price change x RF)
Repeat the simulation w/
= (.08)(-3/8)+ (.04)(-1/4)+ (.16)(-1/8)+ (.40)(0)+ (.20)(1/8)+ (.06)(1/4)+ (.04)(3/8)+ (.02)(1/2) = 0.005 Mark currently in Austin.
Use FF:
Expected Price Range: 23 + 0.005 = 23.05 .13, .08, .60, .13, .68, .40, .40, .27,
. 23, .64, .36, .56, .25, .88, .18, .74
Inventory Example: Supermarket
o Supermarket installed postage stamp vending machine. Based 1 month operation, it estimates
number of postage stamps sold per day: Compare % w/ those found in
 HOUSTON.

Waiting Line Example: Wayne Airport

o Incoming passengers must 1st have passports & visas checked. This is handled by 1 inspector.  Next-Event Simulation Records
The time required to check passenger's passports & visas can described by probability distribution - For each passenger the following information must be recorded:
below.  When his service begins at the passport control inspection
 The length of time for this service
 When his service begins at the baggage inspection
 The length of time for this service

 Time Relationships
o After having passports & visas checked, passengers next proceed to 2 nd customs official who does - Time a passenger begins service by the passport inspector
baggage inspections. Passengers form a single waiting line w/ official inspecting baggage on a 1 st
come, 1st served basis. The time required for baggage inspection follows probability distribution = (Time the previous passenger started passport service)
shown below. + (Time of previous passenger's passport service)

- Time a passenger begins service by the baggage inspector

(If passenger does not wait in line for baggage inspection)

= (Time passenger completes service
A chartered plane from abroad land at Wayne Airport w/ 80 passengers. Simulate processing of the first w/ passport control inspector)
10 passengers through customs. Use ff random #:
(If the passenger does wait in line for baggage inspection)
For passport control: .93, .63, .26, .16, .21, .26, .70, .55, .72, .89 = (Time previous passenger completes
For baggage inspection: .13, .08, .60, .13, .68, .40, .40, .27, .23, .64 service w/ baggage inspector)

- Time a customer completes service at the baggage inspector

= (Time customer begins service with baggage inspector) + (Time required for baggage
inspection)

Other Simulation Issues

 Computer Implementation
 Verification and Validation
 Advantages and Disadvantages of Using Simulation

Computer Implementation
Spreadsheet Add-Ins
- The use of spreadsheets for simulation has grown rapidly in recent years.
- Two popular spreadsheet add-in packages are Crystal Ball and @Risk.
- Spreadsheets are generally limited to smaller, less complex systems.

Special-Purpose Simulation Packages

 - A variety of special-purpose simulation packages are available (eg. GPSS, SIMSCRIPT, opened at a seminar.
SLAM, & Arena)
- These packages have built-in simulation clocks, simplified methods for generating probabilistic b. Using 1st 10 random numbers, simulate number
inputs, & procedures for collecting & summarizing simulation output. of new accounts opened for 10 seminars.
- These packages enable analysts to simplify process of developing & implementing the
simulation model. c. Based on 10 seminars from part b, would you recommend that Gustin continue running the
seminars?
General-Purpose Computer Programming Languages
- Simulation models- can developed using programming languages (eg.BASIC, FORTRAN,
PASCAL, C, and C++)
- Disadvantage (using these languages), special simulation procedures are not built in.
- Usually, it takes many lines of computer code written in, say, BASIC, to perform the tasks
performed by a single command in a special-purpose package.

Model Verification and Validation

 Verification & validation of both model & method used by computer to carry out the calculations is
extremely important.
 Models which do not reflect real world behavior cannot be expected to generate meaningful results.
 Likewise, errors in programming can result in nonsensical results.
 Validation- done by having an expert review the model &computer code for errors.
 Ideally, the simulation should be run using actual past data.
 Predictions from the simulation model should be compared with historical results.

Advantages of Using Simulation o The management of Manufacturing Company is considered in introduction of a new product. The
 Ability to gain insights into model solution w/c may be impossible to attain through other fixed cost to begin production of the product is $30,000. The variable cost for the product is
techniques. expected to be between $16 and $24, w/ most likely value of $20 per unit. The product will sell for
 Once simulation has developed, it provides convenient experimental laboratory to perform "what if" $50 per unit. Demand for the product is expected to range from 300 to 2100 units, w/ 1200 units the
and sensitivity analysis. most likely demand.

Disadvantages of Using Simulation a. Develop the profit model for this product.
 A large amount of time may be required to develop the simulation. b. Provide the best-case, worst-case, & best-case analysis
c. Discuss why simulation would be desirable
 There is no guarantee that the solution obtained will actually be optimal.
 Simulation is, in effect, a trial-and-error method of comparing different policy inputs.
 It does not determine if some input which was not considered could have provided a better solution
for the model.

ANOTHER EXAMPLE
o To generate leads for new business, Investment Services offers free financial planning seminars at
major hotels in Florida. Attendance is limited to 25 individuals per seminar. Each seminar costs
Gustin $3500, & average 1st year commission for each new account is $5000. Historical data
collected over the past 4 years show that the # of new accounts opened at a seminar varies from
no accounts opened to a maximum of 6 accounts opened according to the ff probability distribution:

a. Set up in intervals of random numbers that

can used to simulate the number of accounts