doe project

BUTTERFLY PLATE THREADED INSERT IMPROVEMENT EFFORT

A SUBMISSION FOR DESIGN OF EXPERIMENT

Prepared By: NISARG SHAH

GURKIRAN KAUR MILAN PATEL

QUAL 54513 – Design of Experiments Page | 2

1.0 Objective

The objective of this experiment is to determine which factors affect the loosening or movement of the threaded inserts in the butterfly plates. Once these factors are identified, an optimal combination of factors and level settings will be used to minimize the insert movement. Additional experimentation beyond this first experiment may be required.

2.0 Procedure

In this experiment half factorial method is used. Minitab is used in order to simulate all the results.

We have taken 5 factors in consideration which are as follows:

Tooling

Cati-coat

Minor diameter

Pitch diameter

Insert type

In this experiment, we are using the generator: I=ABCD.

Lastly we have used Minitab software for analysing the significance of the main effects and interaction effects.

2.1 Pre-Experimental Planning

2.1.1 Recognition and Statements of the Problems

Threaded inserts are helically formed coils of diamond shaped steel wire that re threaded into a drilled and pre tapped hole. They provide a controlled level of friction to the installed screw or setscrew, which keeps the screw or setscrew in place even in a high vibration environment.

Threaded inserts in particular pneumatic valve butterfly plates have a tendency to come loose when the setscrew is being installed. When this happens, the valve usually must be disassembled and the butterfly plate removed in order to rework the butterfly plate to reinstall a new insert and setscrew.

There are different versions of this butterfly plate with different kinds of inserts and plate base materials. It is unknown which combinations of materials, inserts and other factors are causing the problem.


2.1.2 Choice of Factors and Level

Table 1: Design Factors Table

Factors Level

Tool 2(New or Existing)

Cati-coat 2(No or Yes)

Minor Diameter 2(.171”dia(tight) or.178”dia(loose))

Pitch Diameter 2(H2 or H5)

Insert 2( Helicoil and 10 digit)

Tool

A tool currently exists to install the inserts, this will be one factor level. A new tool was purchased as a second factor level. One of the insert manufacturers believes the tooling used can make a difference.

Cati-coat

First-hand experience leads the technicians involved in the installation of inserts to believe that cati-coat corrosion preventative applied to the threaded hole prior to insert assembly helps to keep the insert in place.

Minor Diameter and Pitch Diameter

There is an allowable tolerance on the threaded –hole size that insert is installed in. High and low settings were chosen for these two features that determine threaded hole size. It is believed these two features have an effect on the amount of friction between the threaded hole and the insert.

Insert

The types of inserts are known as helicoil and 10 digit.

2.1.3 Selection of response variable

In this experiment the value of responsible variables were chosen randomly and we did not consider any specific method of generating this random variables.


2.2 Choice of Experiment

Half factorial design (2k-p)

Number of replicates =1

Experiments =16

Number of blocks=0

Number of factors=5

2.3 Performing the Experiment

The experimental design is called for 32 runs.it was unreasonable to obtain 32 butterfly plates to use for this experiment. To resolve this problem, two rectangular test beds were manufactured of the same material as the two butterfly plates. The thickness of the beds were the same as in the butterfly plate application. The holes in the plate were drilled and tapped to include the factor settings of the minor diameter and pitch diameter factors.

2.3.1 Experimental setup and conduct of runs

In this experiment, we are using 2k-p fractional factorial design. The value of k is 5 and p is 1, hence we get a total of 16 runs.

2.3.2 Experimental Procedures

The response variable is the movement of the insert. The movement will be measured by angular displacement of the insert from its installed position. This angular measurement was done usually by two people. The figure below gives an example of insert movement. Any deviations more than 10 degrees was made compatible by reviewing disputed measurements.


Table 2: Response Table

Std Order

Run Order

Tooling Cati-Coat

Minor Dia

Pitch Dia

Insert Type

Response

1 5 Old No 0.171 H2 helicoil 8

2 10 New No 0.171 H2 10 digit 9

3 7 Old Yes 0.171 H2 10 digit 34

4 9 New Yes 0.171 H2 helicoil 52

5 14 Old No 0.178 H2 10 digit 16

6 1 New No 0.178 H2 helicoil 22

7 3 Old Yes 0.178 H2 helicoil 45

8 13 New Yes 0.178 H2 10 digit 60

9 4 Old No 0.171 H5 10 digit 6

10 11 New No 0.171 H5 helicoil 10

11 2 Old Yes 0.171 H5 helicoil 30

12 12 New Yes 0.171 H5 10 digit 50

13 8 Old No 0.178 H5 helicoil 15

14 16 New No 0.178 H5 10 digit 21

15 15 Old Yes 0.178 H5 10 digit 44

16 6 New Yes 0.178 H5 helicoil 63

2.4 Statistical Analysis of data

First of all, all the response variables are analysed with Yate’s Algorithm.

The aliasing of the main factors and interactions are shown below in the table and the generator is mentioned above.

From the Yate’s Algorithm, we obtain the contrast (Column name: “4”), from which we calculated the Estimated Effect and Sum of Square


Table 3 Estimation of Effect and Sum of Squares (Yate’s Algorithm)

From the above analysis and from the values of sum of squares we can say that the value of A, B, C and AB appears to be large. Further, the other main effects and interaction effects does not seems large hence adding them to error. From the above assumptions the following ANOVA table is created.

The model sum of squares is SSModel = SSA + SSB + SSC + SSAB = 5747.25, and this accounts for over 99 percent of the total variability.

Treamtment

Combination

Standard

Order

Run

No.

Respons

e1 2 3 4 Effect

Estimation

of Effect

Sum of

Sqaure

abcde 1 14 8 17 103 246 485 -- -- --

a 2 7 9 86 143 239 89 A + BDCE 11.125 495.0625

b 3 6 34 38 96 40 271 B + ACDE 33.875 4590.063

cde 4 4 52 105 143 49 55 AB + CDE 6.875 189.0625

c 5 5 16 16 19 136 87 C + ABDE 10.875 473.0625

bde 6 10 22 80 21 135 3 AC + BDE 0.375 0.5625

ade 7 12 45 36 24 26 5 BC + ADE 0.625 1.5625

abc 8 3 60 107 25 29 -11 DE + ABC -1.375 7.5625

d 9 15 6 1 69 40 -7 D + ABCE -0.875 3.0625

bce 10 8 10 18 67 47 9 AD + BCE 1.125 5.0625

ace 11 16 30 6 64 2 -1 BD + ACE -0.125 0.0625

abd 12 13 50 15 71 1 3 CE + ABD 0.375 0.5625

abe 13 2 15 4 17 -2 7 CD + ABE 0.875 3.0625

acd 14 1 21 20 9 7 -1 BE + ACD -0.125 0.0625

bcd 15 11 44 6 16 -8 9 AE + BCD 1.125 5.0625

e 16 9 63 19 13 -3 5 E + ABCD 0.625 1.5625


Normal Probability Plot:

From the normal probability plot of the effects, also we can see that effects A, B, C and AB are significantly high. Hence adding other main effects and interaction effects to error.

Table 3 Analysis Of Variance

Source of variation

Sum of squares Degree of freedom

Mean square F0

A 495.0625 1 495.0625 193.20

B 4590.0625 1 4590.0625 1791.24

C 473.0625 1 473.0625 184.61

AB 189.0625 1 189.0625 73.78

Error 28.1875 11 2.5625

TOTAL 5775.4375 15


As we can see here that 3 out of 5 main effects are significantly high at α = 0.05 and even at α = 0.01 level, and one Two factor interaction is also significant at the same level.

From the Fishers table we obtain the value of F0.05,1,11= 4.84 and F0.01,1,11= 9.65

Hence from ANOVA Table, we can conclude that the effects of A,B,C and AB are significantly large. This supports our earlier assumption that effect A, B, C and interaction effect AB are significant and proves our assumption.

The main factors A, B and C and the interaction AB have large positive effects. From the interaction graph it is clear that when A is high and B is high, the yield is large.


3.0 Conclusion

Main Effects:

The 3 main effects out of 5, Tooling, Cati-Coat and Minor dia are significantly high, as F0 < F0.05,1,11= 4.84, also F0 < F0.01,1,11= 9.65, hence we can say that main effects Tooling, Cati-Coat and Minor dia are significant at even 1% level of significance.

In order to minimize loosening or movement of threaded parts in butterfly plates, Tooling should be used old with no Cati-Coat and 0.171 Minor Dia.

The Response plot for main effects is shown in the plot below.


Interaction Effect:

Only the interaction between Tooling and Cati-Coat is significant, as F0 < F0.05,1,11= 4.84, also F0 < F0.01,1,11= 9.65, hence we can say that interaction between Tooling and Cati-Coat is significant at 1% level of significance.

From the graph, we can say that, in order to minimize the loosening or movement of threaded parts in butterfly plates, it should be used at low level of Tooling and low level of Cati-Coat.

The plot showing interaction of Tooling and Cati-Coat is shown below.

Based on this experiment, the factor setting to settings to reduce the insert movement

are:

Old Tooling

No Cati-Coat

0.171” Minor Dia

Any of the pitch dia

Any of the insert

doe project

Documents