Failure Analysis Case Studies II Episode 8 ppt

Using relevant fatigue crack propagation data for the valve spindle material at 300°C it was demonstrated that fatigue failures occurred at spindle deflections of between 0.9 and 1.6 mm.

(a) 1/2 inch flange size

Axial coil Transverse coil

(b) 1 inch flange size

Fig 10 Frequency response of node 33 showing peak displacement for the transverse and axial coil con-

Table 2 Natural frequencies for 1 in flange connections, with axial coil, showing the effects of reducing tube height and flange size;

(A) infield configuration, (B) modified

REFERENCES

1 AI-Asmi, K., Seibi, A,, Samanta, B and Siddiqui, R., Investigation into the Failure of Pressure Transmitter Impulse Line

2 Callistcr, Jr, W D., Material Science and Engineering An Introduction, 3rd ed John Wiley & Sons, Inc., 1994

3 Hertzberg, R W., Deformation and Fracture Mechanics of Engineering Materials, 2nd ed John Wiley & Sons, Inc., 1983

4 Shigley, J S and Mischke, C R., Mechanical Engineering Des@, 5th ed McGraw-Hill, 1989

Sahma Booster Station, Internal Report ME/SQU/July, 1995

Failure Analysis Case Studies II

D.R.H Jones (Editor)

CONTROL SYSTEM

J H BULLOCH* and A G CALLAGY

Power Generation, ESB., Lower Fitzwilliam Street, Dublin 2, Ireland

on the spindle Using relevant fatigue crack propagation data for the valve spindle material at 300°C it was demonstrated that fatigue failures occurred at spindle deflections of between 0.9 and 1.6 mm Finally, it was demonstrated that the fatigue breakage problem could be significantly reduced, especially at the lower end of the valve spindle deflection range, by a combination of re-profiling the thread root and shot peening 0 1998 Published by Elsevier Science Ltd All rights reserved

Keywords: Fatigue, fatigue crack growth, fatigue markings, plastic deformation, power-plant failures

1 INTRODUCTION

During the past decade or so certain ESB stations that operate 270 MW units have encountered operational problems involving main steam turbine throttle valve spindles Basically the problems were identified as:

(i) valve sticking as a direct result of spindle undergoing permanent plastic deformation or bending; (ii) spindle fracturing during operation near the top of the valve spindle in the threaded section The present study considered four separate spindle failures which occurred over a three year period and involved:

(i) a detailed failure analysis or micromechanic assessment; and

(ii) a basic engineering stress analysis approach

or

2 MECHANICAL CONSIDERATIONS

A detailed view uf the valve spindle arrangement is shown in Fig 1 and in this instance the valve was in the closed position All failures occurred at position A at the root of a spindle thread while the spindle bending problem was observed at position B At these positions the working temperature was assessed at about 300°C Also, the forces acting on the valve spindle are illustrated in Fig 1

between piston C and the actuating lever where the vertical force is the result of steam pressure and

the angular force was caused by the angle of the actuating or lifting lever The resultant force is a sideways bending force on the valve spindle and it was calculated that the relationship between the

bending stress uB, and value of spindle deflection 1, at position A could be expressed as follows:

a Author to whom correspondence should be addressed

Reprinted from Engineering Failure Analysis 5 (3), 235-240 (1998)

I STEAM '

TO TURBINE

Fig 1 Valve spindle system

where uB was in MPa and 1 was in mm At the valve spindle bending location, Le position B, the value of uB was 1.37 times greater because it was 118 mm further away from the applied force In the stainless steel valve spindle, bending occurred when the material flow stress was attained, Le half the sum of the yield and tensile stress, viz., 830 MPa, and hence the maximum stress at position

A (see Fig I) was around 600 MPa or a spindle deflection value of 1.75 mm As this coincided with

a machined I S 0 thread, a local stress concentration factor (SCF) of around 4 was prevalent

At position B an area of wear, around 0.5 mm deep (see Fig 1) was observed on one side of the

valve spindle surface indicating the presence of a significant bending action or stress This bending stress could have been the result of:

(i) clearance between the piston and cylinder wall, position C (see Fig 1) or

(ii) misalignment of the value actuating housing

However, the facts that (i) clearance related bending stresses are negligible at unit loads approaching

200 MW and (ii) failure occurred at high loads, suggested that the failures observed at position A

were the result of spindle deflections or bending stresses caused by misalignment or possible thermal distortion in the valve spindle system

3 FAILURE CONSIDERATIONS

From a detailed examination of the broken spindle fracture surfaces it was assessed that failure was the result of a ductile fatigue crack extension process which was initiated at the thread root

Fig 3 Coarse fatigue striations Striation spacing around 15 pm

location (see Fig 2) and propagated by ductile striated crack growth (see Fig 3) In three of the

four failures surface crack initiation was helped by the presence of near surface titanium based non- metallic inclusions The coarse nature of the striation spacing (see Fig 3) suggested a high stress, low cycle fatigue situation Indeed the short service lives of the four broken spindles, viz., 33, 136,

167 and 3 16 days added credence to this suggestion

4 FATIGUE CRACK GROWTH ASSESSMENT

In this section relevant real material fatigue crack growth data at 300°C was used in an effort to assess the valve spindle deflection values required to cause the real “in field” short service time failures observed over the past few years Amzallag and Maillard [l] have reported fatigue crack growth results at 300°C for a similar martensitic stainless type bolting steel and the upper bound data was described as:

da

- = 1.05 x 10-6(AK)2,

where daldn = fatigue crack growth rate in mm per cycle and AK = stress intensity range = 1.12Ao

,/Z in MPa Jm

In an attempt to determine the number of fatigue cycles based on the unit power load a typical

load profile is shown in Fig 4 From this figure it can be seen that over a one day period there was

one large unit power change and about four transient power changes which were around half this value

Using a range of spindle deflection values (bending stresses), AKand thus the fatigue crack growth characteristics, over a range of defect sizes, could be assessed Also, taking the spindle failure criteria

as bein that fatigue crack length where K approaches the materials fracture toughness of - 250 MPa &, the number of fatigue cycles or time to spindle failure was obtained The relationship between valve spindle deflection and time to failure is illustrated in Fig 5 From Fig 5 it can be seen that the four fatigue failures occurred over a valve spindle deflection range of 0.9 to 1.6 mm and resided somewhat below the maximum spindle deflection possible at this location of 1.75 mm

It is clear that certain facts have emerged which strongly indicate that spindle breakages occurred due to a high stress, low cycle, ductile, fatigue process, viz.:

(i) high bending stress or deflections are needed to bend spindles;

(ii) significant surface indentations of about 0.5 mm were observed;

Fig 6 Schematic of spindle deflection versus unit power and calculated cycles to fatigue failure

(iii) the failures occurred over very short and unacceptable service durations; and large spindle deflections occurred at high unit loads as failures occurred in this regime and could only have come from out of alignment of the actuating mechanism

As a result Fig 6 is semi-schematic in nature and illustrates the spindle bending and spindle fatigue breakage regions together with the notional increase in valve spindle deflection with unit power and the number of fatigue cycles required to cause spindle failure It is clear that reducing the spindle deflection increased the working life of the spindle; indeed at deflections approaching 0.6 mm it was predicted that the spindle life was 21,200 cycles or almost 12 years

It has been reasonably demonstrated that the series of spindle failures were the result of a high stress, ductile fatigue process that was caused by significant out-of-alignment of the actuating mechanism during service During an outage, an exercise was conducted to determine the “cold” amount of misalignment of an actuating mechanism where a fatigue failure occurred The measured amount was 0.38 mm which was more than 40% of the lower end valve spindle deflection level of 0.9 mm required to cause fatigue failure As such, it is not difficult to envisage that thermal distortions during the hot “on-load” excursion could easily account for spindle deflections attaining the critical range necessary for fatigue failure

Such a recurring failure in a critical plant component needed to be urgently addressed in an effort

to obviate or at least mitigate the problem In the present instance it was suggested that the high stress situation at the spindle threaded location be reduced by two actions; firstly, changing the thread profile to a rounded thread (e.g NF000-032 type thread) which had an associated stress concentration factor which was about 25% lower than the present square I S 0 thread and secondly

by introducing significant compressive stresses (which need to be overcome before fatigue can occur)

by shot peening The influence of these actions on the valve spindle deflection-time to fatigue failure relationship is illustrated in Fig 7 From Fig 7 it can be seen that both shot peening and re-profiling the thread significantly increased the valve spindle life at spindle deflection levels of around 1 mm Indeed at 0.9 mm spindle deflection the spindle service life was increased from less than one year to around 8 years The service temperature in the region of valve spindle failure was estimated to be around 300°C and it is known that thermal relaxation of the compressive stresses can occur at high temperature However, recently Gauchet et al [2] have reported encouraging results where significant

Fig 7 Time to fatigue failure versus spindle deflection

compressive stresses (by shot peening) have remained in HP heater water chambers after 5 years at 270°C

Finally, it has been shown that thread re-profiling and shot peening can mitigate fatigue failure

in valve spindles Also regular checks of actuating system out-of-alignment should be carried out together with good insulation at key locations which should help minimise thermal distortions

Failure Analysis Case Studies /I

D.R.H Jones (Editor)

0 200 1 Elsevier Science Ltd All rights reserved 24 1

FATJGUE FAILURE OF HOLD-DOWN BOLTS FOR A

HYDRAULIC CYLINDER GLAND

C TAO,* N XI, H YAN and Y ZHANG

AVlC Failure Analysis Center, PO Box 81-4, Beijing, 100095, P.R China

(Receiced 22 September 1997)

Abstract-A hydraulic-cylinder gland system used in aircrah failed by leaking because the hold-down bolts broke in the course of a trial run The metallographic examination of the fracture surface and the stress calculations for the bolts are described in this article The investigation showed that the failure was caused by fatigue and the reason for failure was considered in relation to the processing, surface condition and assembly

of the bolts Measures to increase the fatigue strength of the bolts are proposed CJ 1998 Elsevier Science Ltd

All rights reserved

1 INTRODUCTION

A hydraulic-cylinder gland system used in aircraft failed by leaking in the course of a trial run The gland was fixed with one hold-down and four hold-down bolts Three of the four bolts broke in service The bolts were manufactured by turning and threading from 17-4PH steel The nominal composition of 17-4PH is OCr-17NiltCu-4Nb and typical mechanical properties are yield strength

= 1200 MPa, tensile strength ab = 1300 MPa after solution heat treatment at 1040"C, then water quenching and tempering for 4 h at 495°C This paper describes an analysis of the nature and the causes of fracture as well as preventive measures for avoiding fatigue failure of the hold-down bolts

2 METALLOGRAPHIC EXAMINATION

A schematic drawing of the hold-down and bolts is shown in Fig 1 The positions of the hold-

down bolts are indicated by 1#, 2#, 3# and 4# Each bolt head was cross drilled with two assembly holes at right angles to one another The 3# bolt broke away in the middle of the threaded portion

The 1# and 2# bolts broke away in the head between the assembly holes and the shoulder transition radius General views of the fracture surfaces taken in the scanning electron microscope are shown

in Fig 2

The fracture surface (Fig 2(a)) of the 3# bolt was characteristic of a typical fatigue fracture, i.e there was a crack initiation zone, a fatigue crack propagation zone and a final ductile fracture zone The fatigue crack initiated at one position in the thread root at a machining mark The crack propagated towards the far edge of the thread The origin mne was rough (Fig 3) and had many radial lines The propagation zone was smooth and there were distinct fatigue striations (Fig 4) In comparison with the fatigue surface, the final ductile zone was smaller and was around 20% of the total cross sectional area According to the above-mentioned features, the fracture surface is characteristic of fatigue The final ductile fracture zone was typically dimpled No material defects were found in the fatigue origin zone

The fracture surfaces of the 1# and 2# bolts initiated at the edges of the assembly holes, as shown

in Fig 2(b) and Fig 2(c) A schematic of the fatigue fracture sites is shown in Fig 5 The cracks obviously propagated towards the root of the bolt until the remaining cross section became unable

to support the load and failed by fast fracture In comparison with the macro-fracture surface of 3#

*Author to whom correspondence should be addressed

Reprinted from Engineering Failure Analysis 5 (3), 24 1-246 ( I 998)

Fig 1 Schematic of hold-down and bolts (numbered 1 4 )

Fig 2 Fracture surfaces of bolts (a) l#; (b) 2#; (c) 3# Fracture origins are marked with an arrow

bolt, the fracture surfaces of bolts 1# and 2# were rather rough with propagation radial lines, but

no fatigue beach marks However, the photographs of the fracture surfaces taken in the SEM at

high magnification showed that there were fatigue striations (Fig 6) Mixed zones between the dimples and the fatigue striations existed in the fracture surfaces of bolts 1# and 2#

243

, Fracture Surface

Fig 3 Fatigue origin zone of 3# bolt

Fig 4 Fatigue striations in fatigue propagation zone of 3# bolt

Fig 6 Photographs of the fracture surfaces taken in the SEM (a) 1#, (b) 2#

3 STRESS ACTING ON THE BOLTS Consider that the platform and the shell of the hydraulic cylinder are rigid because of sufficient thickness The forces acting on the bolts (Fig 7) are tensile stresses produced by the changes of oil pressure in the working condition

i

Fig 8 Stress acting on the fracture surface for I # and 2# bolts

3.1 Stress (ath) acting on the thread root

First, the effect of stress concentration was not considered The stress acting on the minimum

section of the bolt, i.e the minor diameter, d, = 4.134 mm, was the maximum The oil pressure

transmitted by the hold-down platform (d2 = 44.17 mm) was borne by four bolts The maximum

oil pressure P,,, was 3.42 MPa Take the static estimation as follows:

The result of tensile tests for the same group o f bolts showed that ab was around 1260 MPa This showed that the maximum stress was about 1/13 of tensile strength of material even in the minimum section

The fatigue strength of the alloy is much greater than

Fig 9 Comparison of fracture surfaces for (a) fracture, and (b) 3# bolt fatigue fracture

3.2 Stress acting between assembly hole and bolt root

It is known from Fig 5 that I# and 2# bolts fractured at the location between the assembly holes

and the bolt root Assume that the fracture surfaces were circular cones and were acted on by a

uniform tensile stress Q (Fig 8)

However, the load- bearing area is reduced by the presence of the assembly holes, and there are appreciable stress concentration factors at the various changes in cross-section However, a tensile test to fracture of the bolt showed that the threaded portion was still the weakest element with respect to static loading (Fig 9)

On this basis, a , is comparable to the value obtained above for

4 REMEDIAL MEASURES

( I ) Increase the distance between the shoulder and the assembly holes (dimension h in Fig 5)

(2) Increase the radius of the filler between the head and shank of the bolt

(3) Use thread rolling instead of threading for manufacturing the thread form

(4) Consider the use of forging for producing the basic bolt shape

Failure Analysis Case Studies I1

D.R.H Jones (Editor)

0 200 1 Elsevier Science Ltd AI1 rights reserved 247

ANALYSIS OF A VEHICLE WHEEL SHAFT FAILURE

J VOGWELL

Department of Mechanical Engineering, University of Bath, Bath BA2 7AY, U.K

(Received 24 April 1998)

Abstract-This paper describes an investigation which was carried out on a failed wheel/drive shaft component

used on an unmanned, remotely operated vehicle for manoeuvring military targets As many vehicles had been

manufactured and delivered to customers it was necessary to establish whether it was thought likely that more

failures might occur A study of the broken shaft shows how vulnerable such a rotating component can be to failure by fatigue, even when operating under steady conditions, if basic preventative design actions are not taken The analysis considers the effects of both transmission torque and weight (thus bending) upon stress levels and assesses their individual affect on the breakage and upon any subsequent modifications needed to improve the design The drive shaft arrangement is compared with the feasible alternative of using a driven wheel arrangement rotating on a stationary axle Findings confirm the importance of recognizing in advance the salient factors leading to fatigue and the necessity in paying adequate attention to detail during design and manufacture if long service life is to be achieved 0 1998 Elsevier Science Ltd All rights reserved

Keywords: Fatigue, fatigue design, machinery failures, vehicle failures

by the fact that they carry heavy batteries and must be protected by armour plating which adds considerably to the weight carried and thus the potential for high bending moments along the shaft The design requirement is further complicated by additional factors, many of which are of a variable nature These include the effect of changing torque transmitted during acceleration of the vehicle from start up to full speed and also abrupt breaking (these cffects will result in torsion and thus changing shear stresses), travelling over uneven terrains without suspension will contribute shock loading (further adding to fluctuating bending stresses) Consequently, the vulnerability to fatigue damage is clearly very real and so it is essential to realise this and identify weak locations and take preventative steps at the design and manufacturing stages

2 THE WHEEL SHAFT DESIGN

The wheel shaft has been made from a stainless steel bar (grade 316) and has been turned down

to a central diameter of 20 mm with an overall length of 725 mm A keyway slot is machined near

Reprinted from Engineering Failure Analysis 5 (4), 271-277 (1998)

one end to secure the driven timing belt pulley wheel The ends of the shaft are reduced in diameter

(to nominally 12 mm diameter along a 76 mm length with 3 mm fillet radius at the shoulder) and

act as hubs for mounting the 150 mm diameter wheels (which are shrink-fitted onto 76 mm long sleeves) as shown in Fig 2 A keyway slot runs 73 mm along the length of the 12 mm hub diameter

and a M6 screw fits in the sleeve to secure the wheel from moving axially along the shaft as more clearly illustrated in Fig 3

3 ASSESSING THE FAILURE The failed wheel shaft had broken in two, having separated close to the end of the keyway slot

on one of the 12 mm diameter hubs-approximately 73 mm from one end, as shown in Fig 2 A fatigue crack is clearly evident for about half of the broken sectional area (that nearest to the keyway groove) as the characteristic circular lines radiating outwards from the corner of the keyway are clearly visible The remaining section failed through static fracture-being insufficient to support

the loads From observation of the failed shaft it is not evident whether shaft bending or torsion is

the primary cause or whether it is a combination of the two and so analysis is necessary

4 SHAFT ANALYSIS

Because it is not immediately obvious whether bending or torsion has been the major cause of failure, both effects are considered independently

4.1 Bending under static loading

The trolley weight (with annour plate), W = 320 kg (3.2 kN) and is assumed to be evenly shared among the four wheels The bending moment occurring along the shaft may simply be determined from taking the product of the wheel reaction force and the moment arm (from wheel centre to location of interest)

At end of keyway, bending moment,

4.2 Torsion due to drive transmission

The electric motor power rating, P = 0.25 kW with an output speed, N = 1500 rpm

The maximum shaft driven speed, N = 1500 x 16/48 (pulley teeth ratio) = 500 rpm

Shear stress due to drive, z = ~ = = 14MN/m2

Consider effect of sudden braking: