krm8 Ism B
krm8 Ism B
Supplement
B Simulation
PROBLEMS
The small sample size of just 10 batches may cause us some estimation errors. Another
approach is to work with the expected values of the five probability distributions. They
can be computed as:
Simulation l SUPPLEMENT B l 97
These numbers suggest that a much longer simulation would show that machine 2 is
the slightly better choice. Its shorter processing times more than compensate for the
longer setup times, given the sizes of batches that arrive. Smaller batches favor machine 1.
3. Because either machine has plenty of capacity, and continuing to assume equal operation
and maintenance costs, we should purchase the lower price machine. In other words, the
decision should not favor the machine with higher capacity. Capacity in excess of that
needed has no value. Greater capacity merely results in more idle time.
4. Omega University
a. Preliminary estimates or utilization and proportion of unanswered calls:
arrival rate: 90 calls per hour ´ 60% forwarded to office = 54 calls/hour
answering service rate: 60 minutes/hour/1 minute/call = 60 calls per hour
estimated utilization = 54/60 = 90%
Of 90 calls, we would expect 36 (or 40% ´ 90) to be answered by the professors. 54
would be forwarded, and because the clerk has some idle time, we might expect the
lion’s share of those calls to be answered as well. Surely only a few calls would go
unanswered.
b. Simulation. See table showing the simulation.
The first three random numbers in the first row of the table are from the first two
digits in the second column of Appendix 2, moving from top to bottom. The simulation
shows that during the 60 minutes, 82 calls were placed. Of those, 68 were answered.
Even though this simulation is for an hour when fewer than the expected average of 90
calls were received, 14 calls or 14/82 = 17% went unanswered by anyone.
c. Professors answered 34 calls (41%) and 48 (59%) were forwarded to the department
office. Of the 48 forwarded calls, only 34 calls (or 71%) were answered by the clerk.
The clerk was idle 26 of 60 minutes. Utilization was only 34/60 = 57%, not the
estimated 90%.
The simulation shows that even though the clerk has lots of idle time, calls were being
missed because they do not arrive at a steady pace.
98 l PART 2 l Managing Processes
10:52 36 1 45 Yes 0
Simulation l SUPPLEMENT B l 99
10:53 26 1 20 Yes 0
10:54 26 1 46 Yes 0
10:55 41 1 78 No 0 Ö
10:56 79 2 73 No 45 Yes 0
10:57 87 3 47 Yes 77 No 89 No 0
10:58 99 4 78 No 08 Yes 21 Yes 61 No 1
10:59 24 1 15 Yes 0
5. Voice mailboxes. The office assistant is currently spending 57% of his time answering the
telephone. See table showing the simulation. Assuming that time saved could be
productively used elsewhere,
Labor savings = $3,000/month ´ 57% ´ 60% = $1026/month.
Voice mail cost = $25/month ´ 32 telephones = $800/month.
Yes, order voice-mail system.
6. E-Z Mart
a. Random Number Sales
00–09 60
10–23 61
24–57 62
58–79 63
80–91 64
92–99 65
b. RN 28 83 73 7 4 63 37 38 50 92
Demand 11 13 13 10 10 12 11 11 12 14
8. A machine center
a. Two random numbers could be used for each client—one for demand and one for
processing time. Once this has been done for all four clients, it is possible to compute
the value of R for the year just simulated. The result is one observation for
constructing a frequency chart or probability distribution.
b. For the first year simulated:
RN Event
88 A’s demand is 4200 units (in 70–99 range)
24 A’s processing time is 10 hours/unit (in 0–34 range)
33 B’s demand is 800 units (in 30–79 range)
29 B’s processing time is 90 hours/unit (in 25–74 range)
52 C’s demand is 3000 units (in 10–59 range)
84 C’s processing time is 15 hours/unit (in 25–84 range)
37 D’s demand is 600 units (in 0–39 range)
92 D’s processing time is 80 hours/unit (in 95–99 range)
9. BestBuy sales activity. The spreadsheet is printed below, showing the averages sales at
4.75 cars per week and the average weekly revenue at $95,000. The frequency table in the
lower left corner shows a close correspondence with the original frequency distribution
with some small differences. For example, the simulation resulted in sales of 5 cars per
week 29 percent of the time, rather than 30 percent.
102 l PART 2 l Managing Processes
10. BestCar with price variability. Now there are two uncontrollable variables: weekly demand
and sales price. The spreadsheet shown below results in an average of 2.73 cars sold per
week (compare with the 2.88 car average in Figure B.2 when only 50 weeks were
simulated). The average revenue is $57,516 per week.