APPLIED STATISTICS IN BUSINESS
CASE STUDY 11
THE CLARKSON COMPANY: A DIVISION OF TYCO INTERNATIONAL
In 1950, J. R. Clarkson founded a family-owned industrial valve design and manufacturing
company in Sparks, Nevada. For almost a half century, the company, known as the Clarkson
Company, worked on advancing metal and mineral processing. The Clarkson Company became
known for its knife-gate and control valves, introduced in the 1970s, that are able to halt and
isolate sections of slurry flow. By the late 1990s, the company had become a key supplier of
knife-gate valves, helping to control the flow in many of the piping systems around the world in
different industries, including mining, energy, and wastewater treatment.
The knife-gate valve uses a steel gate like a blade that lowers into a slurry flow to create a
bubble-tight seal. While conventional metal gates fill with hardened slurry and fail easily thereby
requiring high maintenance, Clarkson's design introduced an easily replaceable snap-in elastomer
sleeve that is durable, versatile, and handles both high pressure and temperature variation.
Pipeline operators value Clarkson's elastomer sleeve because traditional seals have cost between
$75 and $500 to replace, and considerable revenue is lost when a slurry system is stopped for
maintenance repairs. Clarkson's product lasts longer and is easier to replace.
In the late 1990s, the Clarkson Company was acquired by Tyco Valves & Controls, a division of
Tyco International, Ltd. In 2012, Pentair acquired the valve and flow control operations of Tyco
International in a deal worth around $5 billion. This acquisition fit quite well with Pentair's main
business of providing products and systems that control the filtration, treatment and storage of
water and other fluids. The Clarkson wafer style slurry knife gate valve is still an important
product for Pentair in this effort.
DISCUSSION:
1. The successful Clarkson knife-gate valve contains a wafer that is thin and light. Yet, the
wafer is so strong it can operate with up to 150 pounds-per-square-inch (psi) of pressure on
it, making it much stronger than those of competing brands. Suppose Tyco engineers have
developed a new wafer that is even stronger. They want to set up an experimental design to
test the strength of the wafer but they want to conduct the tests under three different
temperature conditions, 708, 110°, and 150°. In addition, suppose Tyco uses two different
suppliers (company A and company B) of the synthetic materials that are used to
manufacture the wafers. Some wafers are made primarily of raw materials supplied by
company A, and some are made primarily of raw materials from company B. Thus, the
engineers have set up a 2 × 3 factorial design with temperature and supplier as the
independent variables and pressure (measured in psi) as the dependent variable. Data are
gathered and are shown here. Analyze the data and discuss the business implications of the
findings. If you were conducting the study, what would you report to the engineers?
ANSWER
Using two-way ANOVA, we applied a two-by-three factorial design. Temperature and supplier
are the only two independent factors. 70°, 110°, and 150° are the three temperature treatment
levels. Suppliers are divided into two categories: A and B. The valve strength, expressed in psi,
is the dependent variable. The results of this analysis are displayed in data format below.
Anova: Two-Factor With Replication
SUMMARY
Supplier A
Count
Sum
Average
Variance
700
1100
1500
Total
3
483
161
4
3
474
158
13
3
423
141
21
9
1380
153.3333
96.75
3
476
158.6667
25.33333
3
476
158.6667
2.333333
3
447
149
43
9
1399
155.4444
41.02778
6
959
159.8333
13.36667
6
950
158.3333
6.266667
6
870
145
44.8
Supplier B
Count
Sum
Average
Variance
Total
Count
Sum
Average
Variance
ANOVA
Source of Variation
df
Sample
Columns
Interaction
Within
S
S
20.05556
800.1111
84.77778
217.3333
1
2
2
12
Total
1122.278
17
M
S
20.05556
400.0556
42.38889
18.11111
F
1.107362
22.08896
2.340491
P-value
0.313385
9.5E-05
0.138598
F crit
4.747225
3.885294
3.885294
RESULT
We begin by looking at the observed F for interaction, which is 2.340491 and has a p-value of
0.138598. We proceed to main effects analysis because interaction is not significant at any
widely used alpha (0.05 in this example). The difference between the two suppliers is not
significant (F = 1.107362, p-value = 0.313385).
At p-value = 9.5E-05, there is a substantial variation in the strength of the valves as a function of
temperature. 70 o has a mean psi of 159.83335, 110 o has a mean psi of 158.33335, and 150 o
has a mean psi of 145 psi. As a result, it looks that the valves are weaker at 150 degrees.
2. Pipeline operators estimate that it costs between $75 and $500 in U.S. currency to replace
each seal, thus making the Clarkson longer-lasting valves more attractive. Tyco does
business with pipeline companies around the world. Suppose in an attempt to develop
marketing materials, Tyco marketers are interested in determining whether there is a
significant difference in the cost of replacing pipeline seals in different countries. Four
countries—Canada, Colombia, China, and the United States—are chosen for the study.
Pipeline operators from equivalent operations are selected from companies in each
country. The operators keep a cost log of seal replacements. A random sample of the data
follows. Use these data to help Tyco determine whether there is a difference in the cost of
seal replacements in the various countries. Explain your answer and tell how Tyco might
use the information in their marketing materials.
ANSWER
The Single Factor ANOVA was utilized. Country is the only independent variable. Canada,
Colombia, Taiwan, and the United States are the four tiers of categorization. Suppliers are
divided into two categories: A and B. The cost of a seal, expressed in dollars, is the dependent
variable. Here's how it works:
Anova: Single Factor
SUMMARY
Groups
Canada
Colombia
Taiwan
United States
Count
7
7
7
7
Sum
1710
2265
1365
1555
Average
244.2857
323.5714
195
222.1429
Variance
961.9048
939.2857
433.3333
290.4762
ANOVA
Source of Variation
SS
df
M
S
F
Between Groups
Within Groups
64331.25
15750
3
24
21443.75
656.25
32.67619
Total
80081.25
27
P-value
1.22E08
F crit
3.008787
RESULT
The data demonstrate that the cost of replacing seals varies significantly among countries
(F = 32.67619, p-value = 1.22E-08). The means reveal that there may be large disparities
between countries (Canada $244.2857, Columbia $323.5714, Taiwan $195.00, and the United
States $222.1429).
3. In the late 1980s, the Clarkson Company installed a manufacturing resource planning
system. Using this and other quality improvement approaches, the company was able to
reduce lead- time from six to eight weeks to less than two weeks. Suppose that Tyco now
uses a similar system and wants to test to determine whether lead-times differ
significantly according to the type of valve it is manufacturing. As a control of the
experiment, they are including in the study, as a blocking variable, the day of the week
the valve was ordered. One lead-time was selected per valve per day of the week. The
data are given here in weeks. Analyze the data and discuss your findings.
ANSWER
This is a block design that has been randomly generated. The type of valve is the key
independent variable. Day of the week is the blocking variable. A two-way ANOVA without
replication is shown below in Excel. The data and explanation are provided below.
Anova: Two-Factor Without Replication
SUMMARY
Monday
Tuesday
Wednesday
Thursday
Friday
Count
6
6
6
6
6
Sum
10.2
10.1
8.2
9.4
11.1
Average
1.7
1.683333
1.366667
1.566667
1.85
Safety
Butterfly
Clack
Slide
Poppet
Needle
5
5
5
5
5
5
8.2
10.6
6.6
8.3
10.9
4.4
1.64
2.12
1.32
1.66
2.18
0.88
MS
F
0.194167 4.982
891
1.202133 30.8
503
0.038967
Variance
0.376
0.241667
0.246667
0.294667
0.199
0.148
0.052
0.037
0.018
0.082
0.052
ANOVA
Source of Variation
Rows
SS
0.776667
df
4
Columns
6.010667
5
Error
0.779333
20
Total
7.566667
29
P-value
0.005952
F crit
2.866081
9.54E-09
2.71089
RESULT
For type of valve, a very significant observed F value was determined (F = 30.8503 with a pvalue of.9.54E-09). The following are the various valve types and their average lead times:
Safety – 1.64
Butterfly – 2.12
Clack – 1.32
Slide – 1.66
Poppet – 2.18