55

CHAPTER 4

FAILURE MODE AND EFFECTS ANALYSIS (FMEA) –

CASE STUDY

FMEA is a proactive analysis tool, allowing engineers to anticipate

failure modes even before they happen, or even before a new product or

process is released. It also helps the engineer to prevent the negative effects of

these failure modes from reaching the customer, primarily by eliminating

their causes and increasing the chances of detecting them before they can do

any damage. The actions generated by a good FMEA cycle will also translate

to better yield, quality, reliability and of course greater customer satisfaction.

FMEA was being used around for a very long time. Before any

documented format was developed, most of the inventors and process experts

would try to anticipate what could go wrong with a design or process before it

was developed. The trial and error alternative was both costly and time

consuming. FMEA was formally introduced in the late 1940’s with the

introduction of the military standard 1629. Being used for aerospace/rocket

development, the FMEA was helpful in avoiding errors on small sample sizes

of costly rocket technology.

FMEA was encouraged in the 1960’s for space product

development and served well on getting a man on the moon. Ford Motor

Company reintroduced FMEA in the late 1970’s for safety and regulatory

consideration. Ford Motor Company has used FMEA effectively for

production improvement as well as design improvement.

56

The output of an FMEA cycle is the FMEA table, which documents

how vulnerable a product or process is to its potential failure modes. The

FMEA table also shows the level of risk attached to each potential failure

mode, and the corrective actions needed (or already completed) to make the

product or process more robust. The FMEA table generally consists of 16 to

17 columns, with each column corresponding to a piece of information

required by FMEA.

4.1 PURPOSE OF FMEA

The purpose of FMEA is to identify the different failures and

modes of failure that can occur at the component, subsystem and system

levels and to evaluate the consequences of these failures. FMEA is not a

problem solver. It is used with problem solving tools. FMEA can be described

as a systematic group of activities intended

(i) to recognise and evaluate the potential failure of a product/

process and the effects of the failure.

(ii) to identify action that could eliminate or reduce the chance

of such potential failure occurring

(iii) to document the entire process.

4.2 FMEA PREREQUISITES

The prerequisites of FMEA are given below.

(i) Select proper team and organise members effectively.

(ii) Select team for each product/services, process/system

(iii) Create a ranking system

(iv) Agree on format for FMEA

(v) Define the customer need

57

(vi) Design/process requirement

(vii) Develop the process flow chart

4.3 UPDATING FMEA TABLE

FMEA table is to be updated when

(i) a new product or process is being designed or introduced.

(ii) a critical change in the operating conditions of the product or

process occurs.

(iii) the product or process itself undergoes a change

(iv) a new regulation that affects the product or process

(v) customer complaints about the product or process are

received

(vi) an error in the FMEA table is discovered or new information

that affects its contents comes to light.

4.4 BENEFITS OF FMEA

FMEA is designed to assist the engineer to improve the quality and

reliability of design. Properly used FMEA provides the engineer several

benefits and they are given below.

(i) Improves product/process reliability and quality

(ii) Increases customer satisfaction

(iii) Helps for early identification and elimination of potential

product/process failure

(iv) Prioritises product/process deficiencies

(v) Captures engineering/organisation knowledge

(vi) Emphasises problem prevention

58

(vii) Documents risk and actions taken to reduce risk

(viii) Provides focus for improved testing and development

(ix) Minimises late changes and associated cost

(x) Serves as a catalyst for teamwork and idea exchange

between functions.

4.5 KEY TERMS USED IN FMEA

(i) Criticality

Criticality rating is the mathematical product of severity and

occurrence ratings. This number is used to place priority on items that require

additional quality planning.

(ii) Critical characteristics

Critical characteristics are the special characteristics defined by

Ford Motor Company that affect customers’ safety and/or could result in non-

compliance with government regulations and thus require special controls to

ensure 100% compliance.

(iii) Causes

A particular element of the design or process results in a failure

mode, due to a cause.

(iv) Failure mode

Failure modes are sometimes described as categories of failure. A

potential failure mode describes the way in which a product or process could

fail to perform its function (design intent or performances requirement) as

described by the needs, wants and expectations of internal and external

customers.

59

(v) Severity

Severity (S) is an assessment of how serious the effect of the

potential failure mode is. A rating of 1 to 10 is chosen based on the severity.

The severity ratings are given in the Table A2.1 in Appendix 2.

(vi) Occurrence

Occurrence (O) is an assessment of the likelihood that a particular

cause will happen and result in failure mode during the life and use of a

product. Occurrence rating is given from 1 to 10 which is to be chosen as

given in the Table A2.2.

(vii) Detection

Detection (D) is an assessment of the likelihood that the current

control (design and process) will detect the causes of failure mode or the

failure mode itself, thus preventing it from reaching the customer. Detection

rating of 1 to 10 is to be chosen as given in Table A2.3 in Appendix 2.

(viii) Current control

Current control (design and process) are the mechanisms that

prevent the causes of failure mode from occurring, or which detect the failure

before it reaches the customer.

(ix) Risk Priority Number (RPN)

The RPN is the mathematical product of the Severity (S),

Occurrence (O) and Detection (D).

RPN = S x O x D

60

4.6 CASE STUDY ON FMEA

FMEA was carried out at an industry which is a leading

manufacturer of Engine Valves. situated at Chennai, Tamilnadu, South India.

4.6.1 Process Flow Chart

The process flow chart of an engine exhaust valve is shown in

Figure A2.1. FMEA was carried out for the friction welding process of the

engine exhaust valves.

4.6.2 Friction Welding

Friction welding employs the heat produced due to friction welding

and pressure to accomplish the fusion of materials. The basic operation is

carried out by rotating one component and made to be in contact with the

secondary component which is held stationary. Axial pressure is applied

during the rotation which aids in generating heat. This process of pressurising

and heating creates a bond at the interface of the two mating parts. It is

significant that the metal at the interface does not melt, but rather becomes

plastic. Welding heat is obtained at the joint by rotating one part against the

other at a constant or varied RPM, with an axial force applied to the mating

components. Energy is provided to this joint from a continuously running

prime mover, directly connected to the machine spindle. This energy source is

infinite with respect to time, and is supplied to the interface until the proper

total heat is obtained. When this point is reached, the rotating member is

stopped and a forging load is applied to the parts to be joined. Figure 4.1

shows the various components of a friction welding machine. The photograph

of the friction welding machine that was used in this study is shown in

Figure 4.2.

61

Figure 4.1 Components of a Friction Welding Machine

Figure 4.2 Friction Welding Machine

62

4.6.2.1 Study of Key variables and process parameters

The variables in friction welding process can essentially be divided

into machine related variables and non-machine related variables. Non-

machine variables will include the material type to be welded and the part

configuration and size. These variables like in any other welding process will

determine the selection of welding parameters. The machine variables include

friction and forge pressures, speed of rotation. The rotational speed and the

pressure will control the ultimate quality of the weld.

The rotational speed and pressure affect both the width and the

shape of the heat-affected zone (HAZ). High pressures tend to compress the

HAZ, especially at the center, heavily work the interfacial material and cause

a notch effect at the junction. Higher speeds tend to increase the width of the

HAZ and also the grain size. Subsequent use of high pressure forging after

spindle stop is used to work the structure and refine the grain size. The

parameters that can be selected and controlled will optimise the metallurgical

condition.

The process parameters of the friction welding process are soft

force, soft force time, upset force, friction force, upset time, burn off,

permissible shrinkage limit, slide home position, clamp open position and

lube cycle.

4.6.3 Failure Report

Failure reports of the various types of exhaust valves in passenger

car line for two months have been collected and given in TableA2.4 and

shown in Figure 4.3.

63

Figure 4.3 Various failure modes

From the failure report it was observed that the majority of the

failures are due to runout. Part Number 40767 has got the highest failure

percentage. The prominent failure modes are weld crack and low tensile

strength and part number 40767 has a production volume of more than 98%

and hence part number 40767 was chosen for carrying out FMEA. Table 4.1

gives the specifications of the parts to be welded.

Table 4.1 Specifications of the parts to be welded

Compositionin %Structure

C Si Mn Cr Ni P

Dimensions inmm

Before FrictionWelding

FinalDimensions in mm

Head Austenite 0.4To

0.55

1To2

<0.6 7.5To9.5

<0.6 <0.03 Dia- 6.175Length-125

Stem Martensite 0.15To

0.25

0.75To

1.25

1.5 20To22

10.5To

12.5

<0.04 Dia- 6.175Length-63

Dia- 6.175±0.025

Length-181

4.6.4 Failure modes and detection

(i) Low tensile strength: The valve has to withstand a

minimum tensile strength of 700N/mm2. Tensile testing is

64

used to detect the tensile strength of the valves. For every lot

of 1000 numbers, 7 valves are subjected to tensile testing.

(ii) Runout: Runout is the ‘out of roundness’ of valves. A

tolerance limit of +0.04 mm is set as threshold and the valve

which exceeds the threshold is scraped. Runout is detected

in the process by the operator by random checking. The

defective valves that pass through friction welding are

detected in centre less grinding process by 100% inspection.

Both the inspections are carried out by v-block and dial

gauge assembly.

(iii) Weld crack: Ultrasonic testing is used to detect the weld

crack in the weld zone. The rejected valves are kept

separately and examined again. The valves which show

discontinuation are scraped.

(iv) Overall bar length: The bar length for the valve should be

181 mm after friction welding. The length is randomly

checked by the operator. The defective valves which pass

through friction welding are detected during loading of valve

in upsetting machine.

4.6.5 Estimation of Severity Ranking

The severity of each failure can be estimated by interpreting the

failure mode and effect diagram with the suggested severity evaluation. The

failure mode and effect diagram is given in Figure 4.4. The estimation of

severity ranking was done using Table A2.1 and is given in Table 4.2.

65

Figure 4.4 Failure Mode and Effect Diagram

Table 4.2 Estimation of Severity Ranking

S.No. Failure mode Effect Severity Severityranking

1. Runout Scrap (<100%) Moderate 62. Low tensile strength Scrap Very high 83. Weld crack Scrap Very high 84. Overall bar length

low/highScrap (<100%) Moderate 6

4.6.6 Estimation of Occurrence Ranking

The occurrence of each failure can be estimated by interpreting the

Cause and effect diagram with the suggested PFMEA occurrence evaluation

from Table A2.2. The causes for the failures were arrived by Brain storming

LOWTENSILE

STRENGTHH

RUNOUT WELDCRACK

OVERALLLENGTH

LOW/HIGH

POORWELDING

REJECTIONIN C`LESS

UNFIT FOROTHER

OPERATION

REJECTIONIN

UPSETTING

SCRAP

66

the people concerned with the process and the probable causes are shown in

Figure 4.5 and Figure 4.6.

MISALIGNMENTOF JAWS

WORN OUT JAWS COLLETWEAROUT

RUNOUT

Figure 4.5 Causes for runout

LOW TENSILESTRENGTH

OVERALL BARLENGTH

WELD CRACK

IMPROPERBURN OFF

IMPROPERFRICTIONFORCE

IMPROPERUPSET FORCE

IMPROPERUPSET TIME

Figure 4.6 Causes for other failures

4.6.7 Occurrence of Failures

Occurrence is ranked by calculating the failure occurring per 1000

components and using the TableA2.2. The estimation of occurrence ranking is

given in Table 4.3.

67

Table 4.3 Estimation of Occurrence Ranking

S.No. Failure modeOccurrence/1000 valves

Occurrencecriteria

Occurrenceranking

1. Runout 314/138842= 2.26

OccasionalFailure

6

2. Low tensilestrength

186/138842= 1.34

OccasionalFailure

5

3. Weld crack 40/138842= 0.29

Relatively LowFailure

3

4. Overall barlengthlow/high

132/138842= 0.95

Relatively LowFailure

3

4.6.8 Estimation of Detection Ranking

Estimation of detection ranking was made using the detection

ranking table given in Table A2.3. Table 4.4 gives the detection ranking of the

failure mode based on the detection methods.

Table 4.4 Estimation of Detection Ranking

S.No. Failure mode Detectionmethod

Detectioncriteria

Detectionranking

1. Runout Error is detectedin multi stages

High 3

2. Low tensilestrength

Error is detectedin subsequentoperations

Low 4

3. Weld crack 100% inspectionat UT

Moderate 5

4. Overall barlength low/high

In station errordetection

Very high 2

68

4.6.9 Estimation of Risk Priority Number (RPN)

The priority of the problems is articulated via Risk Priority Number

(RPN). It is the product of occurrence, severity, detection.

Risk Priority Number = Occurrence(O) x Severity(S) x Detection(D)

The initial RPN for the failure modes have been calculated and are

shown in Table 4.5, the details of which are given in Table A2.5.

Table 4.5 Initial RPN

Sl.No. Failure Mode RPN

1 Low Tensile strength 160

2 Weld crack 120

3 Run out 108

4 Bar length low/high 36

From the above failure modes, since the RPN for the failure modes

of low tensile strength, weld crack and run out were high, they were given

priority and the cause and effect diagrams were prepared which are shown in

Figures 4.7 to 4.9.

69

MAN MACHINE

MATERIAL METHOD

NOTFOLLOWINGSOP

OPERATORSKILL

TIMEVARIATION

LOAD CELLMALFUNCTIONING

SPINDLE SPEEDVARIATION

LOWCLAMPING

HARDNESSVARIATION

MICROSTRUCTUREVARIATION

BAR MIX UP

OIL

VARIATION

IMPROPERUPSETFORCE

IMPROPERFRICTION FORCE

IMPROPERBURN OFF

IMPROPERSPINDLE SPEED

WELDCRACK

Figure 4.7 Cause and Effect Diagram for weld crack

70

MAN MACHINE

MATERIAL METHOD

OPERATORSKILL

NOTFOLLOWINGSOP

LOAD CELLMALFUNCTIONING

SPINDLE SPEEDVARIATION

HARDNESSVARIATION

MICROSTRUCTUREVARIATION

BAR MIX UP

IMPROPERUPSETFORCE

IMPROPERFRICTION FORCE

IMPROPERBURN OFF

LOWTENSILESTRENGTH

Figure 4.8 Cause and Effect Diagram for low tensile strength

71

MATERIAL METHOD

VARIATION INUPSET FORCE

POOR AXESALIGNMENT

MULTI BORECOLLET WEAR

PRECISION

COLLET WEAR

DEFLECTIONUNIT SETLENGTH HIGH

STEM AND HEADHARDNESSVARIATION

RUNOUT

V BLOCK WEAR

MAN MACHINE

FC VALVE INDEFLASH

BAD FACEPLATECONDITION

END STOPPERROD WEAR ANDTEAR

IMPROPER GRIP OFMULTIBORECOLLET BY THE

Figure 4.9 Cause and Effect Diagram for runout

4.7 REDUCTION OF RPN

4.7.1 Tensile Strength and Weld Crack

The reduction of RPN with respect to low tensile strength and weld

crack can be achieved by optimising the process parameters. In order to

optimise the process parameters, Design of Experiments (DOE) was used as

explained below.

72

Upset force, friction force, burn off and spindle speed are the major

causes of weld crack and low tensile strength. These parameters were

optimised using 2 level DOE involving multiple factors.

The experimental factors of the DOE are upset force, friction force,

burn off length and spindle speed. The factors and the levels are shown in

Table 4.6.

Table 4.6 Factors and levels of DOE

Upset force(Tons)

Friction force(Tons)

Burn off(mm)

Spindle speed(rpm)Factors/Levels

A B C D1 0.712 0.3516 4 21002 1.03 0.415 4.5 2300

4.7.1.1 Experimental results

Crack free micro structure, Tensile strength of minimum

700N/sq.mm and HAZ on head and stem are the response variables of the

DOE. A full factorial DOE was carried out and the experimental results are

shown in Table A2.6.

4.7.1.2 Column effects method

This approach was suggested by Taguchi as a simplified ANOVA

to subjectively point out factors which have large influence on the response.

The sum of the data associated with the first level is subtracted from the sum

of the data associated with the second level for each column of the array. The

magnitudes of the differences are compared to each other to find out the

relatively large effects. The relative magnitudes indicate the relative power of

73

the factors in affecting the results. The strongest factors or interaction will

have the largest differences. Table 4.7 shows the column effects analysis

method summary. Looking at the difference row, factor B has the largest

effect and considered as the most critical factor while the other factors have a

moderate effect on the response. Considering tensile strength, weld crack and

HAZ the selection of process parameters by Columns effects method is shown

in Table 4.8.

Table 4.7 Number of defects

Trial No. A B C D No. of defects1 1 1 1 1 02 1 1 1 2 13 1 1 2 1 04 1 1 2 2 05 1 2 1 1 06 1 2 1 2 07 1 2 2 1 08 1 2 2 2 09 2 1 1 1 010 2 1 1 2 211 2 1 2 1 112 2 1 2 2 013 2 2 1 1 014 2 2 1 2 015 2 2 2 1 016 2 2 2 2 0

SUM 1 1 4 3 1SUM 2 3 0 1 3

Difference -2 4 2 -2

74

Table 4.8 Selection of process parameters by Column effects method

Factors Level Parameter

Upset force 2 1.03 T

Friction force 2 0.4156 T

Burn off 2 4.5 mm

Spindle speed 1 2100 rpm

4.7.1.3 Anaysis of Variance (ANOVA)

ANOVA is a statistically based, objective decision making tool for

detecting any differences in average performance of group of items tested. A

sample of 30 valves was taken and the four way ANOVA was carried out for

four factors A, B, C and D at two levels. The number of defects with respect

to the factors and their levels are given in Table 4.9 and the summary of the

four way ANOVA is given in Table 4.10.

Table 4.9 ANOVA Table indicating the defects

A1 A2Factors with their levels

B1 B2 B1 B2

D1 0 0 0 0C1

D2 1 0 2 0

D1 0 0 1 0C2

D2 0 0 0 0

75

Table 4.10 Four way ANOVA Table

FactorsSum ofSquares

Degrees ofFreedom

Mean Sum ofSquares

FPercent

ContributionA 0.0083333 1 0.0083333 1 --

B 0.0333333 1 0.0333333 4 15

C 0.0083333 1 0.0083333 1 --

D 0.0083333 1 0.0083333 1 --

A*B 0.0083333 1 0.0083333 1 --

A*C 0 1 0 -- --

A*D 0 1 0 -- --

B*C 0.0083333 1 0.0083333 1 --

B*D 0.0083333 1 0.0083333 1 --

C*D 0.0333333 1 0.0333333 4 15

A*B*C 0 1 0 -- --

A*B*D 0 1 0 -- --

A*C*D 0.0083333 1 0.0083333 1 --

B*C*D 0.0333333 1 0.0333333 4 15

Error 0.0083336 1 0.0083336 -- 55

Total 0.1666666 15 100

4.7.1.4 Pooling up estimates of error values

The pooling up strategy entails F-testing the smallest column effect

against the next larger one to see if significance exists. If no significance

exists, then these two effects are pooled together to test the next larger column

effect until some significant F ratio exists.

76

Table 4.11 Pooling error variance ANOVA summary table

Factors Sum ofSquares

Degrees ofFreedom

Mean Sum ofSquares F Percent

contributionB 0.0333333 1 0.0333333 6 16.67

C*D 0.0333333 1 0.0333333 6 16.67B*C*D 0.0333333 1 0.0333333 6 16.67Error 0.0666666 12 0.005555 -- 50Total 0.1666666 15 --- -- 100

From Table 4.11, it is found that the factor A does not affect the

final table and hence HAZ effect at level 2 is selected. B is the most critical

factor of all and the interactions of C, D and B, C and D are also considered to

be critical. The mean values of the occurrence of the defect for 30 valves

under the combination of influential factors (B, C and D) are given in

Table 4.12.

Table 4.12 Mean value of defects

D1 D2

C1 0.5 0.75B1

C2 0.25 0.5

C1 0 0.25B2 C2 -0.25 0

The levels B2, C2 and D1 give less mean value and hence these are

selected. The optimised parameters are given in Table 4.13.

77

Table 4.13 Selection of process parameters by ANOVA

Factors Level ParameterUpset force 2 1.03 T

Friction force 2 0.4156 T

Burn off 2 4.5 mm

Spindle speed 1 2100 rpm

4.7.2 Runout

4.7.2.1 Causes for runout

The various causes for runout were analysed by brain storming and

the causes and the remarks are given in Table 4.14.

Table 4.14 Causes for runout

Sl.No. Factor

Is it a significantcause for the

defect?Remarks

1 Poor axesalignment in m/c

No Machine was tested by themanufacturer to ensure

2 V block wear No SOP says clearly about thefrequency of change of VBlock unit

3 Multi bore colletwear

No SOP says clearly thefrequency of change ofCrawford Collet

4 Precision colletwear

No SOP says clearly thefrequency of change ofprecision collet

5 End stopper rodwear and tear

No Stopper rod is regroundproperly in case of wear andtear

78

Table 4.14 (Continued)

Sl.No. Factor

Is it a significantcause for the

defect?Remarks

6 Bad face platecondition

Yes To be studied further

7 Improper grip ofmultibore colletby the adaptor

Yes To be studied further

8 Variation in upsetforce

No Machine is built with closedloop system

9 Flow controlvalve at deflashunit

No There is a provision toavoid misuse. Periodicalmaintenance is carried out

10 Deflection unittraverse high

No There is matching markwith which the limit switchis set to control the traverselength

11 Stem and headhardness variation

No Receiving inspection isclear at the raw materialstage by metallurgical lab

12 Operator Yes To be studied further

From Table 4.14, it was identified that face plate wear, multi bore

collet and operator may be the causes to be analysed and studied further.

4.7.2.2 Operator

Five operators were studied to find out whether there is any

significant difference in their ability to cause the runout and it was found that

operator was not a major cause.

79

4.7.2.3 Face plate wear

In order to find whether the wear of face plate is a significant cause

for the failure, trials have been made with the worn out and new face plate.

Table 4.15 and Table 4.16 give the details of the occurrence of the runout

with a worn out face plate and with a new face plate respectively and it was

found that face plate wear is not a major cause for runout.

Table 4.15 Runout defects with a worn out face plate

Trial No. Day Shift No. of defects No. of components produced1 A 11 23002 B 12 23003

1

C 15 24174 A 14 24115 B 13 23126

2

C 12 23007 A 14 24138 B 11 23009

3

C 13 2345

Table 4.16 Runout defects with a new face plate

Trial No. Day Shift No. of defects No. of components produced1 A 11 23172 B 10 23113

1

C 14 24174 A 13 24115 B 11 23126

2

C 12 24447 A 12 24428 B 10 23009

3

C 13 2445

80

4.7.2.4 Multi bore collet adaptor

The multi bore collet and face plate assembly is shown in

Figure 4.10. To find whether the wear of the collet adaptor is a major cause

for the runout, trials were made with a new adaptor and the runout defects are

given in Table 4.17. It was found that there is a considerable decrease in the

number of defects by using a new adaptor.

(a) (b) (c)

Figure 4.10 Multi bore collet and face plate assembly(a) Multi Bore collet Adaptor (b) Multi Bore collet with face plate

(c) Assembly of (a) and (b)

Table 4.17 Runout defects with a new adaptor

Sl. No. Day Shift Defects No. of components produced1 A 3 23172 B 2 23223

1

C 1 23004 A 2 24125 B 2 23006

2

C 2 2326

81

Table 4.17 (Continued)

Sl. No. Day Shift Defects No. of components produced7 A 2 2415

8 B 3 2352

9

3

C 2 2300

10 A 3 2314

11 B 4 2316

12

4

C 3 2418

13 A 3 2320

14 B 4 2412

15

5

C 2 2300

16 A 2 2314

17 B 3 2410

18

6

C 2 2306

Total 45 42154

4.8 IMPLEMENTATION AND REVIEW OF THE

RECOMMENDED ACTIONS

4.8.1 Tensile Strength and Weld Crack

The recommended actions were incorporated and trials have been

taken. A trial batch of 57000 valves of passenger car exhaust valve line was

examined and the results of the implemented parameters are given in

Table 4.18. It was found that none of the valves was rejected due to micro

crack in ultrasonic testing. The tensile strength was well over the limit of

700N/mm2.

82

Table 4.18 Results of the implemented parameters

Day Head HAZ Stem HAZ Tensile strength Weld crack

1 1.00 1.20 914 Nil

2 1.12 1.30 876 Nil

3 1.00 1.12 962 Nil

4 1.03 1.01 916 Nil

5 1.06 1.15 920 Nil

6 0.99 1.20 893 Nil

7 1.30 1.10 897 Nil

8 1.10 1.10 904 Nil

9 1.05 1.20 912 Nil

10 1.10 1.15 916 Nil

11 1.00 1.15 892 Nil

4.8.2 Runout

The wornout collet adaptor was replaced with a new one and the

spindle runout is below 100 micron.

4.9 IMPROVED RPN

The improved occurrence ranking is shown in Table 4.19. The

details of improved RPN are given in Table A2.7. The comparison of the

initial RPN and the improved RPN for the three failure modes is shown in

Table 4.20.

83

Table 4.19 Improved Occurrence Ranking

Occurrence per 1000valves

OccurrenceRankingSl.

No. Failure ModeBefore After Before After

1 Runout 2.26 1.068 6 42 Low Tensile

Strength1.34 Nil 5 1

3 Weld crack 0.29 Nil 3 1

Table 4.20 Initial RPN and Improved RPN

Sl.No. Failure Mode Initial RPN ImprovedRPN

1 Low Tensile strength 160 402 Weld crack 120 403 Run out 108 72