Lab 9

1. The Spuds-n-More model files are provided with the solutions as L9_6 Exercise 1 Spuds-n-More1.Mod,
L9_6 Exercise 1 Spuds-n-More2.Mod, and L9_6 Exercise 1 Spuds-n-More3.Mod.

1.a. The ProModel output results are summarized in the table below. Following the Bonferroni approach
with paired-t confidence intervals, three 98% confidence intervals were computed using the data and
are shown at the bottom of the table. The half-width (hw) of each of the confidence intervals below is
much larger than the half-width (hw) of the corresponding confidence intervals computed in Section
L9.4 where we used Common Random Numbers (CRN). This illustrates that the use of the CRN
technique on these models resulted in more precise estimates (smaller half-widths) of the performance
measures using the same number of replications. This is the expected result when good
synchronization of the random number streams is achieved across the three simulations. The benefit is
most dramatically illustrated when comparing Spuds-n-More2 with Spuds-n-More3. In column (G) in
the table below, the half-width (hw) of 4.18 customers produces a wider confidence interval that
includes a value of 0 customers, which suggests that the average number of customers processed by
Spuds-n-More2 and Spuds-n-More3 is not significantly different. In Section L9.4 using CRN, we
detected that the performance of the two systems, although small, is, statistically speaking,
significantly different. As in Section L9.4, we conclude that both of the new restaurant designs
(Spuds-n-More2 and Spuds-n-More3) process more customers per day on average than does the
current “as-is” Spuds-n-More1 restaurant. The overall confidence in the conclusions here and in
Section L9.4 is 94 percent.

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
 (A) (B) (C) (D) (E) (F) (G)
Spuds-n-More1 Spuds-n-More2 Spuds-n-More3
Rep. Customers Customers Customers Difference (B-C) Difference (B-D) Difference (C-D)
Processed Processed Processed
(j ) x 1j x 2j x 3j x (1-2)j x (1-3)j x (2-3)j
1 64 75 81 -11 -17 -6
2 67 73 69 -6 -2 4
3 61 78 83 -17 -22 -5
4 59 71 79 -12 -20 -8
5 60 79 64 -19 -4 15
6 66 77 76 -11 -10 1
7 57 70 78 -13 -21 -8
8 58 78 78 -20 -20 0
9 56 73 68 -17 -12 5
10 55 81 76 -26 -21 5
11 61 79 84 -18 -23 -5
12 59 67 80 -8 -21 -13
13 57 76 76 -19 -19 0
14 63 80 80 -17 -17 0
15 65 73 65 -8 0 8
16 57 85 85 -28 -28 0
17 53 69 74 -16 -21 -5
18 58 71 87 -13 -29 -16
19 65 86 81 -21 -16 5
20 58 80 81 -22 -23 -1
21 55 82 77 -27 -22 5
22 66 71 79 -5 -13 -8
23 63 79 59 -16 4 20
24 53 77 84 -24 -31 -7
25 57 78 90 -21 -33 -12
Sample Mean(i - i '), for all i and i' between 1 and 3, with i < i ' -16.60 -17.64 -1.04
Sample Standard Dev(i - i '), for all i and i' between 1 and 3, with i < i ' 6.39 9.42 8.42

t 24,0.01= 2.485
hw 3.18 4.68 4.18
98% C.I. Low -19.78 -22.32 -5.22
98% C.I. High -13.42 -12.96 3.14

1.b. The ProModel output results are summarized in the table below. Note that these are the same results
used in exercise 1.a.

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
 (A) (B) (C) (D)
Spuds-n-More1 Spuds-n-More2 Spuds-n-More3
Rep. Customers Customers Customers
Processed Processed Processed
(j ) x 1j x 2j x 3j
1 64 75 81
2 67 73 69
3 61 78 83
4 59 71 79
5 60 79 64
6 66 77 76
7 57 70 78
8 58 78 78
9 56 73 68
10 55 81 76
11 61 79 84
12 59 67 80
13 57 76 76
14 63 80 80
15 65 73 65
16 57 85 85
17 53 69 74
18 58 71 87
19 65 86 81
20 58 80 81
21 55 82 77
22 66 71 79
23 63 79 59
24 53 77 84
25 57 78 90
Sample Mean 59.72 76.32 77.36

Performing ANOVA using the data from the table above:

K= 3 restaurant designs
n= 25 replications of each design
N = nK = (25)(3) = 75

Degrees of
Source of Variation Freedom(df) Sum of Squares Mean Square FCALC
Total (Corrected) N-1 = 74 SSTC = 7232.67
Treatment (Restaurant Design) K-1 = 2 SST = 4898.43 MST = 2449.22 75.55
Error N-K = 72 SSE = 2334.24 MSE = 32.42

Given α = 0.05. FCRITICAL = F(df(Treatment), df(Error, α) = F(2, 72, 0.05) = 3.124.

Since FCALC > FCRITICAL, we conclude that at least one restaurant design processes significantly more
customers.

Performing the LSD analysis using the data in the table above yields the following:

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
 2(MSE )
LSD()  t (df ( Error ), / 2)
n
Given t ( df ( Error ),  / 2) t ( 72,0.025) 1.993
2(32.42)
LSD(0.05) 1.993 3.21
25

From the table above, X 1 59.72, X 2 76.32, and X 3 77.36. Therefore,

| X 1  X 2 | 16.6
| X 1  X 3 | 17.64
| X 2  X 3 | 1.04

If | X i  X i ' |  LSD(  ), then  i and  i' are significantly different at the  level of significance.

Comparing Spuds-n-More1 and Spuds-n-More2

16.6 > LSD(0.05). Thus, there is a significant difference in the performance of Spuds-n-More1 vs.
Spuds-n-More2.

Comparing Spuds-n-More1 and Spuds-n-More3

17.64 > LSD(0.05). Thus, there is a significant difference in the performance of Spuds-n-More1 vs.
Spuds-n-More3.

Comparing Spuds-n-More2 and Spuds-n-More3

1.04 < LSD(0.05). Thus, there is not a significant difference in the performance of Spuds-n-More2 vs.
Spuds-n-More3.

Here, we conclude that Spuds-n-More2 and Spuds-n-More3 process the same number of customers on
average (their performance is not significantly different). In Section L9.4 using the Bonferroni
approach with paired-t confidence intervals in conjunction with the CRN technique, we detected that
the performance of Spuds-n-More2 vs. Spuds-n-More3, although small, is, statistically speaking,
significantly different. Note that the CRN technique cannot be used in conjunction with ANOVA.

2. The ProModel output results are summarized in the table below. The results for Spuds-n-More4 were
generated using the L9_6 Exercise 1 Spuds-n-More3.MOD file, which is included with the solutions
manual, with the capacity of the Pickup_Q location set to six and with the common random numbers
option activated. The Spuds-n-More3 results were taken from Table L9.1. The 98% confidence interval
using the paired-t method is shown at the bottom of the table. Given that the confidence interval
excludes zero, we conclude that there is a significant difference in the average number of customers
processed by the two systems. The confidence interval suggests that Spuds-n-More4 produces an
estimated 0.63 to 2.65 more customers per day on average than does Spuds-n-More3.

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
 (A) (B) (C) (D)
Spuds-n-More4 Spuds-n-More3
Rep. Customers Customers Difference
Processed Processed (B-C)
(j ) x 1j x 2j x (1-2)j
1 79 75 4
2 84 81 3
3 90 88 2
4 73 72 1
5 70 69 1
6 74 74 0
7 71 71 0
8 65 65 0
9 84 84 0
10 76 76 0
11 78 70 8
12 91 90 1
13 79 79 0
14 66 66 0
15 89 89 0
16 86 83 3
17 72 72 0
18 81 77 4
19 78 78 0
20 72 70 2
21 80 77 3
22 76 76 0
23 80 76 4
24 73 72 1
25 73 69 4
Sample Mean(i - i '), for all i and i' between 1 and 2, with i < i ' 1.64
Sample Standard Dev(i - i '), for all i and i' between 1 and 2, with i < i ' 2.04

t 24,0.01= 2.485
hw 1.01
98% C.I. Low 0.63
98% C.I. High 2.65

3. The ProModel output results from running Lab 8_4 Exercise 1 Picayune System 1.MOD and Lab 8_4
Exercise 1 Picayune System 2.MOD, which are included with the solutions manual, for 15 replications
are summarized in the table below. The computation of the 90% confidence interval using the Welch
method is shown below the table.

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
 (A) (B) (C)
Picayune Sys. #1 Picayune Sys. #2
Rep. Customer's Average Customer's Average
Minutes in System Minutes in System
(j ) x 1j x 2j
1 4.125 5.072
2 4.788 3.852
3 3.885 3.903
4 4.06 3.581
5 5.054 3.442
6 8.056 3.554
7 6.832 3.434
8 5.031 4.828
9 3.729 2.715
10 3.457 2.973
11 4.432 3.869
12 4.211 4.718
13 4.289 3.257
14 5.335 6.847
15 6.561 3.307
Sample Meani 4.923 3.957
Sample Standard Dev.i 1.297 1.043

 0.10
n 1 15
n 2 15
X 1 4.923
X 2 3.957
S1 1.297
S 2 1.043
df 26.8
t df , / 2 t 26.8,0.05 1.7056
hw 0.733

The 90% confidence interval is

(X 1  X 2 )  hw  1   2 ( X 1  X 2 )  hw
0.233  1   2 1.699

With approximately 90 percent confidence, we conclude that there is a significant difference in the mean
time that customers spend in the two banks. The confidence interval suggests that customers spend on
average 0.23 to 1.70 minutes more in bank system #1 than they do in bank system #2. Therefore, system
#2 is the better system with respect to minimizing the amount of time that customers spend in the bank.
This is consistent with the guesstimate made in Section L8.4.

4. The model of the DumpOnMe facility introduced in a Lab 4 exercise and used in the third exercise in
Lab 8 is configured for the 6-truck system. The name of the model provided with the solutions for Lab

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
 8 is Lab 8_4 Exercise 3 DumpOnMe.Mod. To simulate the 8-truck system, change the Qty Each field
for the “trucks” entity in this model from 6 to 8 in the Arrivals table. This places 8 trucks at the location
called Loader_q at the beginning of the simulation. See the figure below with the Qty Each field
encircled. Likewise, to simulate the 10-truck system, set the Qty Each field to 10 in the Arrivals table of
the model.

Running Lab 8_4 Exercise 3 DumpOnMe.Mod with the specified number of trucks produces the output
summarized in the table below. The values under columns (B), (C), and (D) in the table are from the
locations section of the ProModel output report under the Total Entries column for the scale location.

(A) (B) (C) (D)

DumpOnMe 6 Trucks DumpOnMe 8 Trucks DumpOnMe 10 Trucks
Rep. Truckloads Processed Truckloads Processed Truckloads Processed
at Scale at Scale at Scale
(j ) x 1j x 2j x 3j
1 114 125 125
2 114 128 128
3 114 130 126
4 114 126 127
5 113 129 133
Meani 113.8 127.6 127.8
Standard Dev.i 0.447 2.074 3.114

Using the Bonferroni approach with Welch confidence intervals, three 98% confidence intervals were
computed using the data above and are shown below.

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
  1  2  3  / 3 0.06 / 3 0.02
n 1 5
n 2 5
n 3 5
X 1 113.8
X 2 127.6
X 3 127.8
S1 0.447
S 2 2.074
S 3 3.114

Comparing 6-trucks with 8-trucks

The 98% confidence interval is
df 4.4
t df , / 2 t 4.4,0.01 3.747
hw 3.555
(X 1  X 2 )  hw  1   2 (X 1  X 2 )  hw
 17.355  1   2  10.245
There is a significant difference in the performance of the 6-truck system vs. the 8-truck system.

Comparing 6-trucks with 10-trucks

The 98% confidence interval is
df 4.2
t df , / 2 t 4.2,0.01 3.747
hw 5.272
( X 1  X 3 )  hw  1   3 ( X 1  X 3 )  hw
 19.272  1   3  8.728
There is a significant difference in the performance of the 6-truck system vs. the 10-truck system.

Comparing 8-trucks with 10-trucks

The 98% confidence interval is
df 7.0
t df , / 2 t 7.0,0.01 2.998
hw 5.016
(X 2  X 3 )  hw  2   3 (X 2  X 3 )  hw
 5.216  2   3 4.816
There is not a significant difference in the performance of the 8-truck system vs. 10-truck system.

With an overall confidence of approximately 94 percent, we conclude that the 8-truck and 10-truck
systems process significantly more truck loads of coal on average than does the 6-truck system, and that
the performance of the 8-truck and 10-truck systems is not significantly different. So, the number of
trucks recommended by the scale operators and the loader operators provide essentially the same result.
Therefore, the 8-truck system is recommended. Looking at the output reports below, we see that the 8-
truck system and the 10-truck system result in a Loader 1 and Loader 2 average utilization greater than
99%. This means that any increase in the number of trucks beyond 8 trucks will not increase the
throughput of the systems unless another loader is added.

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
 ProModel output summary for the 6-truck system simulated for five replications appears below.

ProModel output summary for the 8-truck system simulated for five replications appears below.

ProModel output summary for the 10-truck system simulated for five replications appears below.

This study source was downloaded by 100000885119503 from CourseHero.com on 05-19-2024 08:08:51 GMT -05:00

https://www.coursehero.com/file/49722957/Lab-9-3rd-Ed-Solutionsdoc/
Powered by TCPDF (www.tcpdf.org)