ML20005D880

Rev 1 to Root Cause Analysis for Pressure Transmitter Failures at Trojan Nuclear Power Plant

ML20005D880
Person / Time
Site:	Trojan File:Portland General Electric icon.png
Issue date:	09/09/1988
From:	Chiu C FAILURE PREVENTION, INC.
To:
Shared Package
ML20005D872	List: ML20005D871 ML20005D880
References
88-011, 88-011-R01, 88-11, 88-11-R1, NUDOCS 9001020136
Download: ML20005D880 (80)

Text

A,l URf pocwa@tist. Gotratrel De8k December 20, 1989

-Enclosure 1 54 Pages REVENTlOhc w ~ ~ ~. ~

a.m.n~ awxs.,nimia fV-M Root Cause Analysis For Pressure Transmitter Failures At The TROJAX Nuclear Power Plant l

.\\

FAILURE PREVENTION INC. Report No.88-011 Revision 1 September 9, 1988 F'

OCF C 44 P

PDC

1 Problem Stat-nt Six failures of ITT Barton pressure transmitters have occurred while the plant was at power since 1985. These transmitters belong to ITT Barton 763A class and are environmentally qualified for harsh conditions resulting from accidents inside the containment. They are used to measure reactor coolant pressure (1700-2500 psia span, 0-3000 psi range). The failure mode for these transmitters has been detemined to result from longitudiaal and not through-wall cracks in the Bourdon tube inside the pressure transmitter (Reference 1).

Because these failures have been repetitive, suggesting that the failures may not be isolated cases, and because failure of these pressure transmitters requires significant maintenance work inside the containment while plant is at its full power, it is imperative to find out the root cause and corrective actions of these failures. As such, a comprehensive root cause analysis was initiated and the results are reported here.

Purnose of Root Cause Analysis L

The purpose of this root cause analysis includes:

1) identify the root causes and contributing factors that lead to the failure of the Bourdon tubes, 2) suggest methods to prevent recurrence, and l

3) identify alternative pressure measurement instrument that has a good 1

performance track record. '

l l

December 20, 1989 EVENT l Obc woaw<.mumu w a u.w <a cus.m. m m 1

i l

. Root Cause Analysis For Pressure Transmitter Failures At The 4

TROJAN Nuclear Power Plant l

1 l.

l l

l l

\\

FAILURE PREVENTION INC. Report No.88-011 Revision 1 September 9, 1988 1

L 44 P

FDC I

~

AILURE REVENTIObc kNWNDMbMk

.WI 7tWO Aldmtl * $p b (sigam (Jiioni GTj e #71J1.'493 1

ROOT CAUSE ANALYSIS POR ITT BARTON PRESSURE TRANSMITTER PAILURES AT THE TROJAN N0 CLEAR POWER PLANT I

3 g'

PRINCIPAL INVESTIGATOR l

C. Chiu Contributors and Reviewers M. W. Hoffmann J. Carter

\\

J. Mody PORTLAND GENERAL ELECTRIC CO.

e :n g..r

+

~~

Report No.88-011 September 9, 1988 Revision 1

• • P+,

=.we ** a

. - - - +

... +.....

e Problem Statamant l

Six failures of ITT Barton pressure transmitters have occurred while the plant was at power since 1985. These transmitters belong to ITT Barton 763A class and are environmentally qualified for harsh conditions resulting from accidents inside the containment. They are used to measure reactor coolant pressure (1700-2500 psia span, 0-3000 psi range). The failure mode for these transmitters has been determined to result from longitudinal and not through wall cracks in the Bourdon tube inside the pressure transmitter (Reference 1).

Because these failures have been repetitive, suggesting that the failures may not be isolated cases, and because failure of these pressure transmitters requires significant maintenance work inside the containment while plant is at its full power, it is imperative to find out the root cause and corrective actions of these failures. As such, a comprehensive root cause analysis was initiated and the results are reported here.

Purnosa of Root cause Analysis The purpose of this root cause analysis includes:

1) identify the root causes and contributing factors that lead to the failure of the Bourdon tubes, 2) suggest methods to prevent recurrence, and 3) identify alternative pressure measurement instrument that has a good performance track record.

1

Desian of the Bourdon TL's i

e A typical Bourdon tube is shown in Figure 1.

Its assembly configuration is shown in Figure 2.

A Bourdon tube consists of two sections; one is the round tube section and the other is the flat tube section. When pressure is applied, the r

'l Bourdon tube deflects.

The deflection is proportional to the applied pressure.

At the Bourdon tube rated pressure of 3750 psi the deflection is 0.055" (+0.005 to +0.010 inch uncertainty), as stated in Table 2 for various classes of 763 tubes.

The deflection is typically 0.028' at 2500 psi.

The material used for Bourdon tube is Haynes 25, which is a cobalt base material with the followis s voperties:

1)

Tensile stress - 146 Ksi (Reference 2).

2)

Yield stress 70 Ks1 (Reference 3) i 3)

Elongation - 60f,(Reference 3) l l

Relevant Data and Facts for Root Cause Analysis Based on (1) the data transmitted from Jaisen M9dy and Mark Hoffmann to C. Chiu, l

(2) the information obtained during a meeting between C. Chiu and ITT Barton personnel (Reference 4), and (3) the data transmitted from ITT Barton after the meeting (Reference 5), relevant data and facts regarding the failed Bourdon tubes are reviewed and summarized in this section.

1)

The initial offering of the 763A Bourdon tubes started sometime in 1985.

2-i

i 2)

Three utilities have 763A transmitters (22 mil Bourdon tube wall thickness) for use in pressurizer pressure measurementi they are Calvert i

Cliffs - 1 & 2 (BG&E) (not installed), Palo Verde 1, 2 & 3 (ANPP) and i

4 Trojan (PGE) (1700-2500 psia).

Model 763 transmitters have, however, I

]

been provided to other sites using Bourdon tubes from the same heat lot as the 763A transmitters in question.

i 3)

PGE has four 763A Bourdon tube transmitter locations (PT455, PT456, PT457 and PT458). Six transmitters have failed in a period of six months l

1 (4 for PT455 and 2 for PT456) (Note: Two more failures in PT456 location 8/7/88 and 8/14/88).

No failures, however, have been reported i

i by ANPP.

l l

l 4)

There were no failures of the 763 Bourdon tube transmitters which had l

been used at Trojan from its initial operation in 1981 up to 1986 at which time 763A Bourdon tube transmitters were installed. Also, there were no failures of the 763A Bourdon tube transmitters from August 1986 outage until September 1987 shortly after the startup following the 1987 refueling ouage.

5)

The original 761 Bourdon tube transmitters were installed sometime in 1981. They are made of Haynes-25 material from a different heat than that used for 763A Bourdon tube transmitters.

i l - -

d 6)

Sometime before the manufacturing of Trojan's 12 763A Bourdon tubes, a few changes to the design and assembly of the Bourdon tube assembly were instituted.

1 hey are described below:

a)

A higher rating Bourdon tube is required.

Before the change, the l

pressure rating requirement was 3,200 psi.

The change required this pressure rating be increased to 3,700 psi to solve the drifting problem experienced by 763 transmitters.

This change, requiring use of the 3750 psi rated tube, is only for the tubes of 1700 2500 psi range. Consequently, the thickness of the Bourdon tube was required to be increased to 22 mils.

Before this change, there was no requirement for the tube thickness.

Both 22 mils and 19 mils tubes were alleged to be used in " lot 7" 763 transmitters.

b)

The size of the small connecting rod that connects the open end of the Bourdon tube to the strain gauge is increased from 0.007" l

l diameter to 0.032" diameter.

Also, the connecting rod is spot l

l welded, rather than silver soldered, to the strain gauge step off pad.

7)

There were four Model 763 failures at Vogtle plant. All of these 763 transmitters were used to measure the pressure downstream of reactor coolant pump (- 2270 psig).

Three of them have been confirmed to have tube thickness of 0.022 inch.

The pressure fluctuations, caused by vortex shedding induced quarter-wave resonance as the RCS coolant flowing through the mouth of the instrument nozzle, are measured to be from 2040 to 2448 psi (peak-to peak).

l 4

4 s-

,..-.y..

r n-.

.nn,

i i

8)

During the interval of 9/13/87 to 2/22/88 (all but initial failure and no current installation) the time to failure for four out of six replacement l

763A Bourdon tubes installed at Trojan ranged from 24 28 days of power operation.

f 9)

The time-to failure for the four failed Bourdon tubes at Vogtle is l

typically 6 days. The failure mechanism has been determined to be related l

l t

to pressure pulses (1210 psi) generated by vortex shedding and amplified i

by accoustic resonance from the closed end instrument pipe.

l 10)

The manufacturer of the Bcurdon tubes is a subcontractor to ITT Barton.

Because the manufacturer has some products competing with ITT Barton, it does not want to release detailed information regarding the metal forming process. However, as a result of ITT Barton's audit, the subcontractor revealed a list of steps used for fabricating the Bourdon tube.

It also stated that these steps have not been changed in the past. Table 1 lists these fabrication steps, t

11)

Measurements were made to detemine the distance between two outside flat faces at several locations of the flat Bourdon tube.

The results of the measurement show that both 19 mils and 22 mils Bourdon tubes have identical flat face to flat face ' distance of 0.090 0.002 mils.

l l

L

s 12)

Based on the Figures documented in Reference 1 it appears that all the cracks initiated at locations with the highest tensile stress.

Specifically, all the cracks are located around the inside tip of the flat Bourdon tube.

It is apparent from Figure 3 and 4, and 5 that all the cracks initiated from some surface grooves around the inside tip of the bend.

13)

Figures 6 and 7 are the Scanning Electron Microscope 200X pictures (Reference 5) showing the surface roughness at the inside tip of the flat Bourdon tube, it appears that the 19 mils tube has a much smoother surface than that of the 22 mils tube. Also, it appears that the tip radius of the 22 mils tube is much smaller than that of the 19 mils tube, 14)

The gross fast pressure fluctuations suggested by ITT Barton were thought to be in excess of 6,000 psi. No such changes were observed.

Fluctuating level (LT461) results and pressurizer spray control dead bands indicate that during operation the maximus' fluctuation is i 2%

level or 40 50 psi.

Higher resolution transmitters are being installed.

15)

Based on the detailed isometric drawings for PT455, 456, 457 and 458, three simplified drawings were drawn to study the hydraulic effects, if any, on the failure mechanisms.

They are shown in Figures 8, 9 and 10.

6-

i 16)

Based on a conversation with Mc, Nick Regoli at the Trojan plant, l

the replacement procedures for the failed transmitter were l

reviewed.

In summary, the replacement procedure includes three l

{

steps. The first step is to close the isolation valve. The second step is to replace the failed transmitter. The third step is to open the isolation valve.

l i

17)

ITT Barton has used Hi-Rel laboratory to diagnose the failure. The results of the analysis (Report No, MR-028338) indicate that there is i

some increase in microhardness.

Hi Rel also stated that because this

~

microhardness increase is absent in a new tube, the pressure range of the sensor is exceeded during operation.

(Based on the analysis documented in this report, Hi-Rel's conclusion is unfounded and erroneous.)-

18)

ITT Barton has performed an overpressurization and fatigue test on a 763A

{

pressure transmitter.

ITT Barton stated that the tube does not crack i

after an overpressurization of 6,000 psia and 120 X 10' cycles of 1100 psi cyclic stress.

Observation Based on Facts and Data Based on the facts and data collected above, a few observations related to the-failure mechanism can be made.

These observations are stated below:

1)

The majority of the failed Bourdon tubes have wall thickness of 22 mils (six at Trojan and three at Vogtle, out of a total of ten failures).

7-

r 2)

The failure mechanism is related to high tensile stress since all the observed cracks occurred in the highest tensile stress region.

3)

The inside surface around the tip of the flattened Bourdon tube with thickness of 22 mils shows definite sign of bendina induced bucklina.

4)

The tip inside radius for a 22 mil tube is signifS.antly less than that of a 19 mil tube.

This fact, together with the fact that the distance between two outside flat faces for these two classes of Bourdon tubes is approximately identical (87 89 mils based on ITT Barton's measurement on two randomly selected tubes), suggests that the radius for the 22 mil tu'be is at least on the order of 3 mils less than that of a 19 mil tube.

This is a 105 reduction (3 mils divided by (90 mils - 44 mils) /2) in tip radius. Moreover, because of the rigidity increases as the tube thickness increases, the tube tends to maintain a more straight shape and resist curving in the transition region between the tip and the flat portion of the tube. As a result, the tip radius will be reduced by more than IM.

Please compare the tube radius in figures 6 and 7.

The actual reduction in the tip radius is about 2 M.

5)

There is no obvious difference between the instrument pipe layout for PT455 and PT456 versus PT457 and 458.

The only difference, from the viewpoint of accelerating the failure or of lowering the failure threshold, worth mentioning, is that the horizontal pipes immediately adjacent to PT455 and PT456 are sloped in such a way that PT455 and PT456 are located at the lowest elevation.

The equivalent pipe for PT457 is not sloped and for PT458 is sloped in such a way that PT458 is at a higher elevation.

8

l 6)

The time-to-failure for the failed Bourdon tubes varied significantly.

This characteristic is different than the fact that all the Vogtle transmitters failed in about one week of operation. This varying l

l time-to failure characteristic is sometimes a sign of varying failure threshold from tube to tube.

It suggests that for a consistent operation and maintenance practice at Trojan, this variation has to derive either l

from variation in material property or from variation in localized j

j

~

aggressive condition, i.e., hydrogen concentration.

i 7)

Based on the failure analysis for the Vogtle Bourdon tubes, the tubes j

failed as a result of a i 210 psi pressure pulsations.

Even though the i

instrument pipe should not be designed so that its quarter wave acoustic l

resonance (its frequency equals to nC/4L, where n=1, 3, 5

, C'is the i

sonic velocity in water, and L is the length of the pipe) equals the I

vortex shedding frequency (its frequency equals to 0.35 U/0, where U is 1

the coolant velocity and D is the pipe diameter), the resultant pressure l

pulsations should not be large enough to crack a tube with a rated pressure of 3,700 psia. This fact suggests that there is a inherent deficiency in these Bourdon tubes used at Vogtle.

Note that more information about vortex shedding induced acoustic pressure pulsations 1

can be obtained in Reference 6.

8)

The overpressurization test performed by ITT Barton indicates that the tube does not crack during a 6,000 psi overpressurization.

l l

.g.

-l

Causes for Bucklino On Inside Surface Since the buckling on the inside surface of the flattened Bourdon tube is a significant factor in the failure mechanism, it is necessary to understand the l

possible factors that lead to thirs surface buckling phenomenon.

i 1

l It has been well known in the metal forming industry (Reference 7, 8 and 9) that during bending of.sneet metal, the inside surface may buckle due to compressive stress. This buckling roughens the inside surface of the bend and, at an extreme, will lead to incipient cracks.

Buckling is a condition to be avoided in any metal bending process.

Buckling depends primary on two parameters.

One is the slenderness and the other is the bend severity.

The slenderness is defined as the ratio of linear length of defomed metal (h) divided by the thickness of the material (t). The bend sensitivity is the ratio of linear length of deformed metal divided by the inside radius of the bend (r).

In general, h/r is a dominant factor over (h/t) when r is relatively small.

The above described phenomenon can be summarized in Figure 11.

In this figure, it shows that increasing either h/t or h/r beyond a threshold will result in inside surface buckling.

The possible locations of the 22 mils,19 mils and 15 mils relative to the buckling threshold are also indicated in this Figure.

As can be shown in this Figure, the 19 mils Bourdon tube is the only tube that is slightly inside the threshold.

As a result, it will not buckle whereas the 22 mils and 15 mils tubes will buckle.

This fact is consistent with both the measured data and the theories established for metal foming.

Failure History of Cobalt-Base Material i

A review of the past failure history of cobalt base material has been performed r

to identify possible metallurgical factors that could lead to premature failure of Haynes 25 material. The results of the review are summarized below; 1)

The material has a very strong resistance to corrosion, stress corrosion cracking as induced by chloride, and to high temperature creep, i

2)

The material is susceptible to hydrogen embrittlement if it is not heat treated properly. According to the past failure history of the cobalt l

base material reviewed by the author, this material is susceptible to a

hydrogen embrittlement if it experiences low temperature tempering (400'F 1200'F with the effect most pronounced between 1000'F - 1200'F together with cold working) during heat treatment, stress relief, or l

fabrication. As a result, premature cracks will develop even when l

l exposed to a slight amount of hydrogsa gas. Current Barton's procedures require heat treatment at 900'F after cold working in the presence of hydrogen for dimensional stabilization.

l l

3)

The fractography of the fractured surface of a hydrogen embrittled cobalt base material, MP35N, is shown in Figures 12 and 13 (Reference 10). As can be seen in these two figures, the cracks follow the grain and twin boundaries.. -

.- __. - =..

i Some Failure Scenarios that are Imoossible In the process of this root cause analysis, many failure scenarios were hypothesized and then rejected because they either contradict established engineering principles or contradict observed data and facts. These scenarios that are regarded to be impossible are briefly summarized below.

1)

Laree Pressure Fluctuation in Pressurizar Result in Faticue Failure This scenario is impossible based on the fact that the safety relief i

valve on the pressurizer has'not opened for the past two years during operations. Also, it is impossible to identify a source that can induce j

large pressure fluctuation inside the pressurizer.

2)

Water H r inside RCS Results in Overload Failures This scenario is impossible because pressure waves generated in the RCS, if any, will not be able to propagate through the steam space of the l

l pressurizer and reach the Bourdon tubes.

l 1

3)

Inadeauate Maintenance Procedure for Transmitter Reclacement The replacement procedure is judged to be adequate and is no different than the replacement procedure commonly used in the industry.

. 1 i

i 4)

Overnressur' ration of Bourdon Tube Caused By Heatina Un the Added Demin i

Water with

solation Valve Closed This scenario is only possible if cold domin water is added to the tube l

downstream of the pipe isolation valve.

Later, heating up the cold water with a closed isolation valve may result in excessive water expansion and high stress.

This scenario is discounted because of the maintenance procedure used.

In addition, the fractograph examination does not support a ductile "one-time" pressure overload.

Likelv Failure Scenarios Based on all the collected evidence, the most likely failure scenario can usually be hypothesized.

However, due to lack of heat treatment information for the material used for the 763A tubes, the contribution of hydrogen embrittlement to the final failure is uncertain. As such, two hypothesized scenarios are regarded equally likely; one with the contribution of hydrogen embrittlement and the other excluding any contribution from hydrogen embrittlement. These two scenarios, in tems of chronological events, are described below.

Failure Scenario #1 The inside surface of the bend is buckled, generating many surface corrugations which are stress risers.

The tube is subjected to a rated pressure test, as called for by the test requirement set by 'ITT Barton for the tube manufacturer.

The rated pressure test may initiate incipient cracks in the deep and sharp grooves located at the inside tip of the bend...

Failure Scenario #1 (Continued) d The Bourdon tube is put into service and the stress risers help to raise the localized stress level to a level at or beyond the ultimate strength.

Cracks initiate from the stress risers or the incipient cracks and propagate, probably with the aid of small cyclic stress induced by normal pressure fluctuations.

1 The cracks propagate up to 3 M to 5 3 of the tube thickness and then slow

~

down because of a reduced stress field.

Because the cracked tube is more compliant, it deflects more than what it l

1s designed for. As a result, the indicated reading becomes erroneously high and the transmitter fails, i

l Failure Scenario #2 The material of the tube has been annealed or stress relieved between l

400'F and 1200*F for a few hours.

The material is embrittled via the low temperature tempering process.

The tube is subjected to a rated pressure testing as called for by the l

test requirement set by ITT Barton for the tube manufacturer.

The rated pressure test may initiate incipient cracks in the deep and l

sharp grooves located at the inside tip of the bend, l

l 14 -

4

.=

)

i Failure Scenario #2 (Continued) l The Bourdon tube is put into service and the stress risers help to raise f

the localized stress to a level beyond the yield strength, but below the I

ultimate strength, i

Hz, present in the water, is in contact with the Bourdon tube.

e i

He decreases significantly the fracture toughness of the material. As a e

result, cracks start to propagate from the stress risers or incipient l

~

cracks.

The propagation may be aided by the small cyclic stress resulting from normal pressure fluctuations.

The cracks propagate up to 30% to 50% of the tube thickness and then slow down because of a reduced stress field, Because the cracked tube it more compliant, it deflects more than what it e

is designed for.

As a result, the indicated reading becomes erroneously high, and the transmitter fails.

5 -

a Sunnertina Evidence for Scenarios #1 and #2 There are no facts or evidence that contradict failure scenarios #1 and #2.

However, failure scenario 82 can explain two unique conditions, whereas failure scenario #1 has to attribute the occurrence of these two conditions to pure probability and variations in tube's resistance to cracking.

These two conditions are:

1)

All the failures occurred at PT455 and PT456, whereas no failures occurred at PT457 and PT458.

2)

The overpressurization test performed by ITT Barton demonstrated that failure did not occur until the pressure had reached somewhere beyond 6,000 psia.

The hydrogen embrittlement assumed in scenario #2 explains why these two conditions could exist:

s 1)

Preferential Failure of PT455 and PT456 As discussed previously, the horizontal pipes adjacent to PT455 and PT456 are sloped in such a way that the pressure transmitters are located at the. lowest elevation.

This type of sloping is good for self-venting the trapped air by allowing the air bubble to escape with the aid of its own buoyancy force.

However, this self venting would allow condensed water to flow into an area in close proximity to the Bourdon tube.

As a result, the released He from the cooled water will be in contact with the Bourdon tube, causing it to crack prematurely.

16 -

l i

The horizontal pipe next to PT457 is not sloped.

The horizontal pipe next to PT458 is sloped in such a way that PT458 is at a higher l

elevation. As such, tha air bubble trapped in these horizontal pipes will not be vented, thus preventing the hydrogen laden water from coming l

in close contact with the Bourdon tube. Consequently, the effect of l

i hydrogen attack is much less pronounced for PT457 and PT458 than for i

PT455 and PT456.

NOTE:

The transmitter elevations and the slope of the associated instrwnent piping is based on as built drawingst a field verification of the as built drawings has not been completed.

l l

I 2)

ITT Rarton's overnressurization Test Because the test condition did not include the adverse effect of hydrogen on crack initiation and propagation, it is reasonable to expect that the failure threshold for crack initiation is much higher than that under the plant environment.

l --

Fractonranhv Examination As described before, hydrogen aided crack propagation will generate intragranular cleavage planes. However, if pure fatigue is the crack propagation mode, the crack path will show striation marks. The distance between two striation marks is proportional to the difference between the maximum and the minimum stress intensity factor, AK. This ralation makes the distance between two striation marks proportional to the difference between the maximum cyclic stress and the minimum cyclic stress for a given crack configuration.

Under the corrosion fatigue situation, i.e., combining the hydrogen aided crack propagation and the fatigue crack propagation, a small stress fluctuation can lead to a large distance between two striation marks.

The fracture mode of the Bourdon tube can be any one of the following four categories:

(1) pure fatigue mode - striation marks of short distance (2) pure brittle (hydrogen embrittlement) overload mode - cleavage planes and slip bands (also called L0 der bands)

(3) corrosion fatigue mode striation marks of large distance (4) a mixed of brittle overload and corrosion fatigue - cleavage planes and striation marks of large distance 18 -

\\

i To pinpoint which mode is really responsible for the fracture, fractograph examination with high magnification (4,000X to 10,000X) has been performed. The 1

results of this examination are summarized below:

i i

(1)

The majority of the fracture surface (>50%) is covered by cleavage planes and slip bands, (2)

There are a few small areas which contain striation marks of large distance, as shown Figures 14, 15. and 16.

l Based on these observations, it appears that the fracture mode is a mixed mode of brittle overload and corrosion fatigue. This fracture mode will be further confirmed by the fracture mechanics analysis perforined in the following section, j

i f

Fracture Mechanics Analysis Based on the striation marks shown in Figure 16, the average distance between two striation marks is 1 x 10** inch.

This distance is too large to be produced by a i 20 psia (typical pressure fluctuations) pressure fluctuation in a pure fatigue mode.

This fact is confirmed by the analysis as follows:

l Based on the elastic stress analysis perforined in the following section, the ratio of the tip tensile stress to the internal pressure for a 22 mils Bourdon

]

tube is' l

l 160 Ksi Stress to pressure Ratio -

• II*I III 2,250 psi 19 m

---

r

~

Using this ratio, we can determine that as a result of i 20 psia pressure change, a i 1422 psi cyclic stress exists at the tip of the Bourdon tube inside bend.

This value will translate into stress concentration factor AK as below:

AK = aa / na

= 2.84 / n x 0.011 = 0.52 Ksi(in)%

(2)

Now, let us determine the AKth, the threshold AK, below which the Haynes 25 material will not crack regardless of the number of cycles it experiences.

Because of lack of fatigue data of Haynes-25, we have to use a very conservative method to extrapolate data from other weaker materials. One way is to use the fatigue data of stainless steel as a function of the stress ratio R, which is defined as the ratio of minimum stress to the maximum stress in one stress cycle.

In the case of interest, because of a very high residual stress that is much greater than the amplitude of the cyclic stress, R is very close to 1.0.

For I

stainless steel, which has a fatigue strength at least a factor of two lower than l

Haynes-25, the AKth can be expressed by the following formula:

AKth = 6.4 (1 0.85R) Ksi /3 (Reference 11)

(3)

Assuming R = 1.0, AKth for stainless steel is 0.96 Ksi/In. The equivalent AKth value for Haynes 25 can be approximated by the following formula:

l AKth (Haynes 25)

Ultimate Strength (Haynes 25)

AKth (Stainless Steel)

Ultimate Strength (Stainless Steel)

(*)

The above formula is valid because the ratio of endurance limit to the ultimate strength is almost a constant. As such, AKth (Haynes-25) is calculated to be 1.4 Ksi/3.

~

t

$1nce1.4Kst/Inismuchgreaterthan0.52KsijIn,itappearsthatcorrosion (i.e., hydrogen embrittlement) fatigue is responsible for lowering tKth from 1.4to0.52KsijIn. As a result, the observed striation marks are caused by corrosion fatigue, not pure fatigue process.

This conclusion supports the hypothesis that the low temperature tempering of the Haynes 25 did occur in the process of heat treat, stress relief, or fabrication.

Finita-Element Strent Analysis The purpos6s of the stress analysis are to (a) determine the stress to pressure ratio as described in Equation (1), (b) determine if the inside tip of the bend experiences plasticity, and (c) determine if the elastic tensile stress at the tip of the 19 mils tube is higher than that for the 22 mils tube.

A finite-element computer code, which is able to calculate elastic stress, is chosen (Reference 12) to perform the calculation. This code is selected among many other options because it has been verified and validated by the developer, and it takes much less effort to create the model and much less computer time to perform the calculation. The major limitation of this code is that it cannot deal with plasticity and will yield very conservative stress results when the stress is greater than the yield strength. However, for the purpose stated above, this code is sufficient to yield good results.

4,

j i

A 500 node finite element model is prepared to calculate the stress field in the area of the inside tip of the bend. The nodal configuration of the analyzed tube is shown in Figure 17.

Three cases were run.

The first case was a 22 mil tube with a distance between 2

two outside flat faces of 90 mils.

The second was a case of 22-mil tube with a i

flat face to flat-face distance of 95 mils.

The th,ird case was a 19 mil tube with a flat-face to flat face distance of 90 mils.

The second case yields a tip stress about 30 Ksi lower than the first case.

This is primary due to a more generous radius of the inside tip.

The third case is about 30 Ksi greater than the first case, even though the tip radius for this case is greater than that for the first case.

For all three cases, the tip stress is greater than the yield stress (70 Ksi). As a result, it indicates that the stress at the tip is always very high and is high enough to result in plasticity regardless of the small variation in tip radius and tube thickness.

Moreover, based on the results of the third case versus those of the second case, it appears that a thinner tube (19 mil) will have a higher tip stress than that of a thicker tube (22 mil) when the surface conditions are smooth and free of surface buckling. However, based on the stress concentration factor of 3.9 accounting for the stress concentration effect of large and deep grooves, the stress field around the tip for a 22-mil Bourdon tube is much greater than that l

l t. --

I of a 19 mil tube.

The absolute value of this stress level is estimated in the following section with the following considerations:

1) plasticity 2) residual stress from metal forming 3) stress concentration at the roots of bucklird grooves l

Estimation of Plastic Stress at the Tio of the Band As mentioned in the previous section, the finite element analysis calculates elastic stress, not plastic stress. However, we can use the calculated elastic l

stress to estimate the plastic stress that actually exists at the inside tip of the band. This estimate, with an accuracy of 20%, can be performed by first calculating the elastic stress.

Then, using the stress strain curve for Haynes-l 25, we can estimate the plastic stress that yields the same strain as does the elastic stress.

j i

Assuming tha't the residual stress after a spring back of the flattening process is equal to 2/3 of the yield stress (Reference 13), the total tensile stress without the stress concentration effect of the grooves is l

l

\\

l

• total

' total (No Stress Concentration) = 2/3 'y + 130 Ksi i

=

2/3 x 70 Ksi + 130 Ksi j

l

= 176.7 Ksi (5) l l

With the stress concentration effect, the total elastic tensile stress is:

' total

' total (No stress concentration) x Stress contentration factor (6) 23 -

l

\\

The stress concentration factor for a sharp groove ranges from 5.3 to 6.5 (References 14 and 15) and the stress concentration for,a round groove j

(approximated by a half round circle) is 2.5.

By examining the failure pictures, the configuration of some typical grooves have a tip peometry somewhere between a sharp groove and a round groove. As such, we use an average stress concentration factor of 2.5 and 5.3.

That is 3.9.

By inserting this value into Equation (6), we can estimate the total elastic j

stress to be 689.1 Ksi. Using a modulus of elasticity of 32.6 x 10' psi (Reference 3), an elastic stress of 689.1 Ksi is translated to 21% of strain.

Assuming the stress strain curve for Haynes-25 has characteristics of a typical strain hardening material, such as Inconel, Haynes 818, etc., we can produce a stress-strain curve for Haynes-25 as shown in Figur's 18. Using this figure, we estimate the plastic stress at the bend tip to be about 120 Ksi.

The same calculation process can be applied to the 19 mil Bourdon tube without i

tha 3.9 stress concentration factor. As a result, the tip stress is estimated to be 80 Ksi.

This stress level is much lower than the 120 Ksi stress for the 22 mil Bourdon tube.

Because it experiences about 40 Ksi lower tensile stress, tt is expected to have a higher resistance to both hydrogen embrittlement and j

I cyclic stress at normal operation condition. This difference may explain why all 763 tubes (19 mils tube) previously used at Trojan did not fail.

24 -

O Results of the Root Cause Analysis j

Based on the above root cause analysis, it appears that the root cause of the Bourdon tube failures at Trojan, probably as well as at Vogtle, are a result of poor fabrication process, which allows bending induced buckling (References 7, 8, and 9) occur at the inside tip of the bend. The buckling produces many stress risers around the tip and results in a high plastic stress during operation.

As such, cracks initiate at the stress risers during normal operation and propagate by a mixed mode of overload and fatigue fracture process, probably aided by the hydrogen gas existing in the pressurizer water and steam space.

Discussion Based on the calculations documented in previous sections, it is concluded that the plastic tensile stress at the inside tip of the bend is about 120 Ksi.

~

However, it should be noted that this value is determined based on a stress concentration factor of 3.9, a mean value between th'ose for sharp grooves and round grooves.

For a given tube under bending induced buckling condition, the shape of the groove at the tip of the bend can be anything between a sharp groove and a round groove. Therefore, the associated stress concentration factor can range from 2.5 and 5.3.

This range of stress concentration factor will translate into a range of tensile stress from 74 Ksi to 140 Ksi. This large range of tensile stress means that even under a bending induced buckling situation, not every tube has a tensile stress that is so high that small cyclic forces or hydrogen gas during operation will result in tube cracking. This fact offers a possible explanation as to why not every 763A tubes (22 mil thickness) f ails.

Of course, another possible scenario (as discussed in previous sections) is that the transmitters that do not fail are installed in pipe layouts different than those for the failed transmitters. As such, the failed Bourdon, tubes may experience a higher concentration of hydrogen, causing premature cracks due to hydrogen embrittlement.

Note that one of the key bases for this hydrogen embrittlement scenario to be valid is that the Haynes 25 material has gone through the low temperature tempering process during heat treatment, stress relief, or fabrication.

Corrective Actions and Recomandations The corrective actions to prevent recurrence could be divided into two categories.

One is to improve the current fabrication process through which the l

Bourdon tube is manufactured.

The other is to use a more reliable pressure gauge, with a different design and a good track record.

The second category of corrective actions is somewhat appealing because the fabricator of the Bourdon i

tube, a subcontractor of ITT Barton, has decided not to release any detailed fabrication process information for the raason of proprietariness.

This position I

taken by the subcontractor may make the necessary improvement hard to implement and be quality assured.

Nevertheless, to ensure no more failures of ITT Barton pressure transmitters, following two corrective actions are reconnended:

i (1)

Establishment of an acceptance criteria for all future pressurizer ITT Barton pressure transmitters that no internal buckling defects should exist at the inside tip of the Bourdon tube.

This acceptance criteria can be enforced through the PGE's QA program.

(2)

Replacement of all existing pressurizer ITT Barton transmitters, whenever convenient, with the transmitters that are free of buckling defects as provided and quality assured by ITT Barton.. -. - - -

l L

l

-4 i

Regarding alternate pressure transmitters, I have reviewed the past maintenance record for both Foxboro and Rosemount transmitters.

It appears that the a Rosemount transmitter requires much less maintenance effort (roughly 20%) than j

that for a Foxboro transmitter. As a result, I recommend PGE consider Rosemount transmitters (specifically Model No.1153 GD9) as possible replacements in case l'

more failures occurred before ITT Barton can effectively improve its fabrication i

s process for Bourdon tubes.

l 1

1 l

l l

l l

l l.

t

s References 1.

Jeff Carter to Jaisen Mody, " Examination of Failed Bourdon Tubes from Barton Pressure Transmitters," JWC-008 8817, PGE Memo, February 25, 1988 2.

Tube Methods Inc., test report for customer's order no. 1588 February 13, 1973.

3.

Personal Communication with Haynes Engineering Manager, April 10, 1988, also in Superalloys:

a Technical Guide, ASM,1988 4.

Arnold Preiser to C. Chiu, " Minutes of April 5,1988 Meeting," ITT Barton-letter 3877b/de,' April 11, 1988 5.

Arnold Preiser to C. Chiu, "ITT Barton letter 3925b/de, April 19, 1988 6.

C. Chiu and M. A. Hersehthal, " Root Cause Analysis of Safety Valve Vibration Problems," ASME Pressure Vessel and Piping Conference, Pittsburg, June 21, 1988.

7.

W. F. Hosford and R. M. Caddell, Metal Formina Mechanics and Metallurov, Prentice-Hall Inc.,1983 8.

W. W. Wood, E1Dal Reoort on Sheet Metal Formina Technoloav, Vol II, ASD-TDR-63-7-871, July 1963 9.

Handbook of Metal Formina, Edited by Kurt Lange ASM,1985.

j j

References (Continue)

^

10.

R. D. Kane and B. I. Berkowitz, "Effect of Heat Treatment and Impurities on the Hydrogen Embrittlement of a Nickel Cobalt Base Alloy", Corrosion, l

i Vol 36 (1), January 1980.

11.

S. T. Rolfe and J. M. Barson, Fracture and Fatiaue Control in Structures, Prentice-Hall, Inc.1977 j

12.

" LIBRA Finite Element Analysis Software, Serial #1095", INTERCEPT

' Software, Campbell, CA 95008 13.

W. Johnson and P. B. Mellor, Enaineerina Plasticity', Ellis Horwood Limited, 1983 l

14.

R. E. Peterson, Stress Concentration Factors, John Wiley & Sons Inc.,

i 1974 15.

J. E. Shigley and Larry D. Mitchell, Mechanical Enaineerina Desion, McGraw-Hill, Inc., 1983 4

J PRSTBFLR.CC..

. xygan :-n I

1 4

t

(

i i

)

[ This page is intentionally blank. )

t f

(

I t

Y%

AILURE REVENTION me I'6t.m fW;r Lu Oa %)W w me ?ou."imec,

. M + & w kr;w,.i w,n m

h-

h

.. + t.*

t-y>

L

_G

/*..

N, h,,

6'"

E,

,~

?1.

T.

l

..,+,.. '

l

...;,, c

'~i l

~.

1 ts; 1s.; 1 s.

Figure 1 Typical Configuration of a Bourdon Tube

( Courtesy of Mr. Jeff Carter)

A 4

Amn

.,as am, a--,

,w=,.,_,,

s w_------w-i m

[,.

j i

4L E :

!_I ua

! q

,)munalliitidillitidl/h llY;

{3;j g,

p1m...............m..m i loi

e o

g

,r 7

w a

i i

anumumn.titanat i.t i

i t a t i f a

l.

'I gdg as, =.sso.

=

?l!!!!!!!!'t!I? !bil!II E li 4 inia

{

j l

,5 g

j

.t 4

j

1. o o

p.

i s

2 4

E e

o l

,,h e

l 2

I u

u s

~

s l-e g

r J

E g

i g

lj

- ~ ~

i E

.I 5-t nn*a 3

ee i

4

- a v 6( gr z

4 5 g!-

g r ::

['<~! ~ 4

)

"l!

e i di E

$fo q

1 lh I

8

-~)

j ss a

e a

l!!g e

E e

(

s t

!d1 W'

i

\\$

h sa u g !

1

-~

i 4-i u

3

=v i e llh E l

5 l

C3 R

+

i

.re C@ 55%$I-ea 5

4 it 4

I l

h

$$ kk$ $ 5 l

s !)-~,,s : 9

~

~

l l

Ik5hf'" A ME L

m -

~

AILURE

^

REVENTION mc Ihe le.Kiin.'IW! in ROOI Cxt %It 9.i Iol the fou et l%lusin ygg p_,gjg, gyn (y,4,,,,y, (3,,.,y,y,, ;,.. ;;..,,.,,;

1 l

l-i q

.6

' ': [3,,'A s'...

M..,'?' e', "

c';, :..t. e 9;'

s.. -

...',.C'&, &;,.

G.L....

.v c -

64 l

e

..i i t.

s Figure 13 268X i

Figure 3 Stress Concentration Caused By Surface Grooves of a Failed Bourdon Tube. The Culprit Groove is Pointed By The Arrow. This Picture is Quoted from Reference 1.

~

~

o l..

1

All.URE

\\

REVEhTIONx Ih (tin? [wt:in Rott Ghe Mah %y the PCutt hdu%

3n gpyg g y3,9nq,3n( g 3,(gy,,,.p y ;,,9 ;;,.u...:

t

'~

hkkbh!!?[l' \\',.,

hklc.**.:ifb ili 1

R.;'.

L,.

w. %.. j. -t*;. y A t..... ". ;- - - - o '.

p.

7

' Figure 14 Figure 4 Stress Concentration Caused by Surface Grooves of a Failed Bourdon Tube. This Picture is Quoted frea Figure 14 of Reference 1.

s AILURE REVENTION mc Iiv ir.s1. i\\;d dxt DN AMh % Ior be Pou et hdu%

gg,;p y g g,9,, 9 ( g a,,,(;,,,,,,,.,;,y, 9 _..

,,,,;

L 11.

g,

. g.

^

y: '

.Jf.'

g.,s

^3$

• 1

\\

y,.

?-

s.

s.3 m

,f gi,, *. <* e y...

g,g: *. r 3..

~

f 1

Figure 15 130X 1

Figure 5 Stress Concentration Caused By Surface Grooves of a Failed Bourdon Tube. This Picture is Quoted from Figure 15 of Reference 1.

~

AILURE REVENTIOLc.

M ra:: r '^r'O N e n'y'f % r, W m

_p. p

,,g,;.

-., y,,

.. y k'l,.;

.. p.'

\\

\\.,

...y l

y

~.

\\

l l

i.

a

\\

\\

\\,

d ff # fa6t h

.C/j Mc =_

V e-t.w:t

• 2 R O u g - p-cf x,/

Figure 6 Inside Surface Contour for a 19 Mils Bourdon Tube. Note That the surface is Practically Free of Large and Deep Grooves. Courtesy l

of ITT Barton (Reference 5) for Figures 6 and 7.

L-_- _ __

.~

AILURE i

REVENTION mc.

'el.y !w Vai.Jw W O'"~%.r"%n

~

e.

_;; ;,,,,9 y,97 a..,

l l

l s.....

,,. t....../.......

• h ':l,. ';.

,. ~

l r-l

• a l

ry-

' e? '

d

f. ;

l l

C

/.

e -.'

,' {, ' l

..t.,. ;.:...s.

f.

2,.*

.- ),f.l.'. '.*.' *' * ?.

}

i Fe ar secry 3,o c ps i 07%

39 ; %

Figure 7 Grooves and Ridges Observed On the Surface of Inside Bending Tip. Tube Thickness is 22 Mils.

A

+e,-

O e

S c

"9h N

L s

i

\\

in

1.

s t

to u.

i t

=

'[

$M

(($5 i

p b d-E

/

Q.

H l

-m i

8

a 4.,,

n,4 4

9 Mm

-a' m-A<

m

=

4

--me--

• AM sl4 M-ee A

6-m 4

4' N

6

\\\\

x.,.h.

i

~'

N

)

A t

m gB L

b8 1

m=

M i

Q k"

R h

N

.I M

P h

C I

~

L N

E i

9i 8

s s#

-s

.2a_ _ -.

4

.s 4.

O e

l l

i-t Aa.

a

~

p 1a (d

.1-x ei g

'T D7>

' gl g

s x

.: v 3

i Si e

t-Mm 8

.['

i i

H 22 MILS i

E e

s O

'19 MILS

.X 15 MILS THRESHOLD FOR c

c INSIDE SURFACE-BUCKLING h

SLENDERNESS I

G O X - HYPOTHESIZED. LOCATIONS FOR 22.

19 AND 15 MIL BOURDON' TUBES 1:.

l L

FIGURE 11 l

THRESHOLD OF INSIDE DURFACE BUCKLING BASED l

ON THEORIES DOCUMENTED IN REFERENCE 6 AND 7, L

~

AILURE REVEhTIO%c Thele]dingbpellinRoolQ:HeAnalui:Iolthefouethduk

.gl pm.\\ldonn e !Jn lu3n Qui 4Jnu Gliv"ta 4.'Ir*\\ * 'll.'l W

. l l

i l

l i,-

./

t,.,,

5

- g. r j.

j- ~

^

l'f'

'Y.

• gf.. '.,

t.

- a gi h f. [

~.

l %*'~

r M

7

~

,e.c \\;;'.

e

_f.L n>

...),

y j

s

[

- jj

;qi gn

.. : i \\ -:

.'j['I;b' l-1 Figure 12 Fractoghaphy of MP35N (cobalt-base) with Low Temperature Tempering (1100 F and~4 Hours). Fractured by Hydrogen Embrittlement.

+

+

1 I

s-j

~

l-AILURE REVE\\TIO%c

~

. The leadity bpett in Root Quie.4uh% lot the PatetIndustiv

.'% Pa% e kn bn 66'n GWarna 0.'t,*3 * 'W.'h4 M l

I t-l l

l a

r l

Y

'l.

lr

_hyt S A. l,.

~

' q u' N fj wv3 yK gyygg.

1 i R,.i J ? f '/ % l M ! l Q '

M-

.,/. i-

...+

p ffg gy '; P N; l';..

veg j_.:

g,-

~.

+ _

I*fy.Q,. 's.

q

'2,.# :

o

.f, p,'. f. L 4 Y

/

.N-p.

f g;,'g, V..+ v 3 4

r.

3

..;3<.

, l

- ) ~ q ; 'i ' l l",.

E*f, l.

<+;

L

.... a x... _.

2808 3

Figure 13 500X of RP35M Fractography. Note That the Fracture 160de is Transgranular and the Luder Bands are Present.

\\

AILURE REVENTION mc The leading f wen in Roof QR laJk & the POW Nthin 3y,J pag gnya, jy gy gggy, Qipn.py:... x.

~, ;.

-. c '

' ^

_. m-g.

. : {..

N,;,

m; Ws.

p,?..

n ny s &

r

. :W

[ ~

l

...N,% y i

g

...e y7 3

a,. p r p -....,.. -

.cc

1. w *.-.-

l

. - n..M

\\*

07wr,e+.

? "

g - t. y - 4'39-

. c W [ };.

25 k.*b e :~" h

y

.% [.(.I.

I

{'[

me f J 'l y ' w q,

'W.

t ?+- 3] M

/

$4k w 15 6 4 ' ~

i e

l l

l Figure 14 One Area That Bas Fine Striation Marks l

on the Fracture Surface ( located 11 l

ails from the edge of the crack) l a

- l 'AILURE REVENTION mc Iht ltadJnt {\\pellin ROOl DUR kb Si!I0r the 90tt er Indu5h 2%l PJMNka 0 kn luan DDim blitwne 0]V; e 'la lstMGni l

L T.

l 1

l 1

6 l

l'

' 1, E

e, i

l Figure 15 Another Area That contains Striation Marks ( about 13 mila from the edge of the crack)

AILURE REVENTION x.

IMtJdme !wn m Rool Ovie AndiuMolllle f0w l%n 39 pg gy,,, _qgy 9,,,,,,, Q,,y.,;,;,,y:,..,,

s p

i...< &*

• , ar.rm 4'

o,

~'

t er l

l Figure 16 Striation Marks Under 9700X High Magnification. Distance Between Marks

~0 is About 10 inch.

l l

l l

1 4 '

Figuro 17-Detailed Modal Configuration for the Stress Analysis j

Y s41 s

5

\\

y VIEWINGPOINT X:

8,8 V:

0,8 Z:

1,8 LOCALORIGIN X:

0,8 Y:

00,8 2:

,8 Y

I d

m i s

g

\\

+.,

\\

f.

.1

,1 l

l

I I

J

~

I i

-1 8

I-01

-5

,,1E I

-ll8 e

i::

g.

2

_gB B

E

-l

I

,,r IIIIIggaaes*59995 4 Gen.

l

j,.

i a

l dE g

Z-u-

CD o

e g d

z M

g 8

~

~

m

.e.

O M a a

s ik g

g o

1 a;

Z l O

g 3l_________

E I

j l

ai w

~1 I

i 1

i 1

I 1

I 8

8 8

8 (Ts>0 NIVtdd.S 8

.. ~

Tablo 1 Fabrication Procana for ths Bourdon Tubao x

FACILITY AUDIT OF NOVEMBER'1987 Sequence-of Manufacturing 1.

A.

Cut tube to length.

Flatten'(rolls and fixed tooling) "C" portion of tube.

B.

C.

Form first major bend, roll upper section into completed pressure element.

o

[

Verify all dimensions and visually inspect for tool D.

damage or material rupture.

L E.

Place formed elements in heat treatment facility at 900 Deg. F for five (5) hours, hydrogen atmosphere, air cool to ambient overnight.

Route tube to welder for attachment of adapter and plus end 2.

of tube.

~

A.

Base weld of tube to, adapter is fusion welded, no filler, 1.A.W. DWP-W-10763 G.T.A. weld.

NAS "W" filler, per Endoftubeisplgweldusinf0763.

I B.

5786A, G.T.A. weld per DWF-W-l C.

Process involves using heat sink / clamp on tip of tube for flow control of inert _ gas during welding.

Welders are qualified / certified to MIL-T-5021-D.

D.

Facilities have been surveyed and approved by ITT i

Barton on 4-10-86.

Supplier - G.R. Babcock.

First/ completed / testing of completed elements, Custos Tube.

3.

A.

Linkwire tabs are silver soldered to tubes using fixtures and torch braze I.A.W. 9999-3016.2.

Finish tubes are bead blasted on tips to remove weld B.

discoloration, tested 100% at pressure indicated, inspected and shipped to ITT Barton.

2909Q:73 A

s

Tablek(Continusd)

~ '

PAGE 2 Base-adapters are welded in position at ITT Barton and 4.

subject tubes placed in stock as completed assembly upon acceptance by inspection and leak testing per 9999-1014.2.

tance of product is controlled in Receiving Final accep/Special, Department l

5.

017.

Inspection Note:

As of 1-19-87, Custos Tube was placed on "Self Release" (BAV) 1.A.W. Barton QA1 04-04.

Vendor L

Discrepancy hav,e been less than 14 for the period L

of time of years 1985 thru 1986.

Subject quality L

indicates vendor qualities for BAV Self Release.

E a

i I

i-l-

(

t-l 2909Q: 73 L

e i

r

mg,mpw -

c Fr j

et g

a pi

j a

N-6iw 1)

,b

}

-4 s

b-i t

i

.i c) 4

[ This page is intentionally plank. ]

t s

t s

F t

b

'i

~d e.Ww wvnL

~

Document Control Desk' Pontana G:n:ral Euctnc December 20, 1989 Nuccar Dmsion

. Enclosure 2 26 Pages MEMORANDUM i

TO:

C. M. Dieterle J40-0003-89M yRON:

J. W. Carte DATE:

January 19, 1989

SUBJECT:

Metallurgical Examination of Haynes 25 (L-605) Bourdon Tubes The initial evaluation of tuo f ailed Bourdon tubes made early in 1988 (JWC-008-88M, dated February 25, 1988) suggested that the probable cause of failure was high cycle low overstress fatigue cracking. This was based primarily on a comparison of the metallographic aspects of the crack morphology compared to literature examples.

An evaluation using the same samples combined with an audit of the vendor's fabrication practices was performed by Mr. C. Chiu of Failure Prevention, Inc. He constructed an alternative explanation that placed major emphasis on environmental effects, specifically hydrogen embrittlement. This analysis is contained in his Report 88-011 of May 2, 1988.

One of the factors identified as important was the existence of bending-induced buckling on the inside flank of the 22-mil wall tubes known to be failure-prone. The earlier 19-mil wall tubes were relatively free of this phenomer.on because of the geometry dif ferences (ie, bend radius / thickness ratio is greater).. Another f actor was the dimensional stabilizing heat treatment performed at 900*F.

Literature sources indicate that heat treatment in this temperature range greatly degrades the resistance of the cold-formed material to hydrogen embrittlement.

It also provides almost no relief of the very high residual stresses attendant to the forming process. This heat treatment is supposedly common to all tubes regardless of wall thickness.

In order to investigate the matter further, a metallurgical examination was made of a number of f ailed and unf alled Bourdon tubes. The results of this examination are summarized below and support the phenomenon of hydrogen embrittlement as being the primary mode of failure.

The tubes examined have the following histories:

a.

S/N 444, a 19-mil wall tube f rom a Trojan pressure transmitter that had seen several years of successful service prior to replacement with a 763A model transmitter. Not failed, b.

S/N 781, a 22-mil wall tube f rom the Vogtle plant that f ailed in an RCS pressure transmitter due to fatigue induced by cyclic pressure fluctuations of 1210 psi caused by vortex shedding. The pressure fluctuations of 1210 psi cause a cyclic stress fluctuation of nearly 115,000 psi at the inside sharp radius flank of the tube.

Failed by 6

fatigue after over 12 x 10 cycles.

Trojan Excellence-Our Way of Doing Business

Portand G:nwral Electnc Nuc!2ar Division I

C. M. Dieterle January 19, 1989 page 2 c.

S/N 813, a 19-mil wall tube, purchased from Wolf Creek, which was exposed to static hydrogen at 2,300 psig for 30 days. No indication of failure by instrument output.

d.

S/W 1823, a 22-mil wall tube, purchased from Wolf Creek, which was exposed to static hydrogen at 2,300 psig. Failed in Al hours, S/N 1955, another tube from the Vogtle plant with a fatiguo failure e.

due to pressure fluctuations, f.

S/W 1981, a 22-mil wall tube failed in Trojan pressurizer pressure transmitter service.

g.

S/N 1982, a 22-mil wall tube failed in Trojan pressurizer pressure l

transmitter service..

l I

h. ' S/W 1993, a 22-mil wall tube f ailed in Trojan pressurizer pressure transmitter service.

1.

S/N 1999, a 22-mil wall tube failed in Trojan pressurizer pressure transmitter service.

l

-}

The dimensions of the tubes are tabulated in Figure 1.

Also shown are the values for maximum fiber strain on the inside or outside surface as j

calculated by an expression for fully plastic bending:

I o=-et=in[Roh1/2 s

o q

tRi) i Where RO and Ri are the outer and inner radii of curvature, respectively. Note that this strain is somewhat less for the 19-mil wall l

l tubes, especially S/W 813.

S/N 813 has an unusually large "a" dimension, I

which decreases the radius of curvature ratio.

A typical tube is shown in Figures 2 and 3.

Since none of the Trojan or laboratory fractures penetrated the tube wall, no evidence of failure is detectable by visual inspection. The two fatigue failures from Vogtle i

had through-wall cracks, which were located at the points indicated in 5

Figures 4 and 5.

j The first tube to be considered is S/N 444.

A transverse view of the j

tight-radius bend inside diameter is shown in Figure 6.

Note the j

relatively smooth contour. A closer view of this area reveals what appears to be very slight penetrations, which turn out to be at the grain boundaries, although this cannot be determined in Figure 7.

Etching of the unfalled 19-mil wall tube produced a response different from the other tubes (Figure 8).

Ditching of a number of grain boundaries is present, which, coupled with the slight intergranular attack, suggests a different heat treatment from that of later tubes.

Trojan Excellence - Our Way of Doing Business

PomanC General Elemne Nuc M r D m pon I

C. M. Dieterle January 19, 1989 Page 3 Figure 9 is an optical microscope view of the crack in S/W 781. The opened fracture surface is in Figure 10, with the crack front indicated by a dashed line. The nature of the fatigue region is shown in Figures 11-13.

Figures 12 and 13 are taken right near the crack tip, and show the striations or arrest marks which are rather widely spaced at this late stage of the fracture.

Figure 12 is a bright field view, while Figure 13 is a dark field view.

'.5e same dineral area was also examined by scanning electron microscopy

( S EM). The boundary between the fatigue crack and the laboratory tear is located near the left-hand third of Figure 14 A closer view of this boundary is found in Figure 15, with some relatively coarse striations visible in Figuta 16.

There appear to be finer striations present that have been obscuesd by rubbing during crack closure.

S/N 813 survived a 30-day exposure on pure hydrogen at 2,300 psi with no damage indicated on the instru>ent readout. Post-test examination showed the inside flank to be relatively smooth, with no otvious evidence of cracking (Figure 17). Under closer examination by opticci microscope, one small cracklike feature was found, as seen in Figures 18 and 19.

An SEM view of the area confirmed a typical hydrogen crack about 0.3 mLis long. It is not known whether this crack would have propagated to the point where it would have affected the operation of the instrument.

However, its presence in a tube with relatively good contour and low residual stresses indicates that this alloy in the condition provided should always be considered vulnerable to hydrogen.

S/W 1823 is a typical 22-mL1 wall tube in all respects. It failed after 41 hours4.74537e-4 days <br />0.0114 hours <br />6.779101e-5 weeks <br />1.56005e-5 months <br /> in 2,300-psi static hydrogen, gxamination of a cross-section near the middle of the tube shows a small crack on one flank (Figure 21) and two larger cracks on the opposite flank (Figures 22 and 23). Close views of these latter cracks (Figures 24 and 25) again reveal the transgranular mode of following slip planes that was observed in the earlier Trojan plant failures.

An SEM examination of the crack tip in Figure 25 reinforces the picture of the crack following slip and cross-slip through the microstructure (Figure 26).

An SEM examination of the S/N 1823 f racture surf ace (af ter completion of the break) showed a typical cleavage failure with no evidence of the striation formation seen in the fatigue failure of S/W 781 (Figures 27 and 28).

This topographical dif ference will later be compared with the Trojan plant failures.

S/N 1955 was another fatigue failure from Vogtle.

An optical micrograph of the through-wall crack is seen as Figure 29.

When mounted, polished, and etched in the transverse direction, the through-wall crack is found to be typical of f atigue (Figure 30). That is, it is transgranular in nature with a few small side 'oranches that do not deviate much from the main direction of the crack.

This can also be seen on the other flank, where two-part wall cracks are found (Figures 31 and 32). Note the Trojan Excellence - Our Way of Doing Business

I

+

P;m:nc General Electne Nuc0:rDrw on C. M. Dieterle i

January 19, 1989 Page 4 difference in crack morphology at high magnification when compared to the hydregen failure, S/N 1823.

This difference may also be observed in S/W 1981. Figure 33 is an example of bending-induced buckling with two hydrogen cracks in an unetched saraple.

In another region, an etched view of two adjacent cracks (Figure 34) is seen, and an unetched view is found in Figure 35.

Note the extremely jagged nature of the cracking.

An SEM examination of the cracking in S/W 1981 found a typical brittle

[

cleavage fracture surface as found for the laboratory fracture, S/N 1823. Figure 36 consists of a cleavage fracture from the inside flank of the tube (right-hand side of the photomicrograph) to the point marked by the dashed line.

past this point, the fracture was completed in the laboratory.

A closer view of this is seen in Figure 37, where the dividing line is near the middle of the photomicrograph.

t 3/N 1982 also exhibits bending-induced buckling and two cracks (Figure 38).

In Figure 39, the jagged discontinuous cracks clearly j

depart from two buckles on the inside flank surface.

An etched view of l

S/N 1993 shows a similar jagged crack morphology in Figure 40.

An SEM examination of the opened crack again revealed the typical cleavage l

fracture topography (Figure 41).

1 Unetched (Figure 41) and etched (Figure 42) views of the flank cracks found in S/W 1999 complete the series of tubes examined. All failures except the Vogtle tubes appear to have cracked in the same fundamental way, io, by hydrogen embrittlement. While small cyclic stresses may have accelerated the process, the differences in the detailed crack and fracture surf ace morphology from the Vogtle tubes clearly point to this conclusion.

It is interesting to note, however, that literature reports indicate that hydrogen embrittlement in other cobalt-base alloys exhibits mixed-mode cracking (ie, transgranular and intergranular). Why the cracking should be wholly confined to transgranular slip planes in this inst-nce is not known.

Possibly, the specific he.at treatment / cold work coma'. nation may have some effect.

JWC/53640 Trojan Excellence - Our Way of Doing Business

i,

.y I

Li i

t I(h I

)

/

i b

141_

a.. in,

b. in.

8 444 0.0865 0.2475 0.289 781 0.0905 0.2455 0.332 813 0,1020 0.2430 0.233 1823 0.0890 0.2470 0.341 1955 0.0895 0.2405 0.341 1981 0.0904 0.2470 0.334 1982 0.0890 0.2460 0.341 1993 0.0890 0.2465 0.341 1999 0.0890 0.2460 0.341 FIGUR8 1 JWC/53640

6 " 4:

jif{ll {1{

li:

nb.

lwllg%(

n.

g ny:q;;, ;~g;j g;9 -

y 4s -

.~.

v-

.c y

g$;'e cc, e y::q %

e g;

'%p g fj, q,A "'

g v

r iss; N'

if[k ey[Nk;$h@J;%

g 5$ ).ih j!

fQ$

f

}f je,Q. 4.

- j.gg [.

m

y fvt.4 On W1 4

. tyQ;QG#yy:;

yH[w(kh.)jjfY

,n agus f4 E h

-sGn w ;c van O!

- <;b l

N-,

/

.;)

f*1gure 2 S/N 1982 g.4 "j$!

[

.-{[ Qjf fdihki*h

>4 i

e

'aQ[$y)MR(j?

l6- @ [

N!i hiSng y

r t

fLf ; f

[',

% )e

. f;\\)Iky&ll \\;'ed ~

L

";Q

,[e

.v,

'=

J qf i

ify$p,q"g l'

7.

Q.:S l

't ggg l

'"';$f'tpfT h ".a/r. e

,y' 7)s g.

" hj

+p, 44'ey w

.h yf$x

%2kh)Ld.f@%@%p%%n$

p2 -

,y L Q;F^f l

}@,

% h >% @r% af e: G i j!

% )mn%g}

a;o

.)Q n a %(tfljg.m n

~

v O6 l

. nua h

b l

1 I

scg

y. r,.

ja

. d;.3 j

• 4 1

o..n3 a }-n..e' 4

9

• 4 y r

.t w

4m

-it. S;ig.J;e i

< ;m., y[.fa wy;[f$yj$

- PlI /k ;i; t

r

.n-

{ Q;;1[f.)S h

,#,f fi h,. c W%

'V fyfry M

,N

[1;; g;.].dfg! ne,/ly[ 3 q

p w' ' %gf%;;Ay.

a s&gr +

) _ x;;*.

y s

y, -

.w.

l, _ ve p % hQij. m.-

n, t.:S s.

gp,

y-:up x+

' f

(.' ;QgYg/*

q' egU,

'2,v,

,\\

lfn 81 }

"n ww,;,' en

~~ 7 Q( > {j

< ?f:pTjc:'

/

A S.ud.. -

Figure 4 S/N 781

y.,

fr '~

,i.9

'~

js

-r y,

1. 1

'(

,P0, y -

.,# /

3

'Jf,

', N S' /

![

j",(

' (,'

'p w A

Ll 4,

g h.O

'O q:

s t

. i.

w rG f 7..

'?

kh i

.;p-... nhQ$@6.jW 1

J 4

, y ;c gd.. ;4'.

edh q p &v,.p; t r*;f y

hQQ;b;g tk. -

e in.. -:

jf(6 # ?. TQ )?t?

O l]? yhjGy!

M ? ';;s '

.( 4 9 4

y;i t

glfhf;l?[M

$Q/;Oj D[(dihhb[jP Ty ;t p

i fhth

-Nff

&a y${kl ng@}f$,Q_p3,,@

jbRh Mt@Q$

fQ M

p,.g.

-s ab l

i

.) p e a}?biQib,y-14lsiIOgi 4

'?l w-;.

fN 9:,

i rigure 5 g s 1955

Ingure 6 SN 444 1 n,y t

4 b

t 4

e sisb

(

5 g

h g,h I

,g

S O

63'fS* ' i'* *an h&jY(d s

M 9 N @hh l

< '.mJ$k, vi t~e?..,,;

• y. 7*f*$ I_ e y

nw y

4&gll

'4 t

i38gsl);;Qp.

.GTip.f.n w,v'.,

y($r%;1'm4*b.,.'Ik {n. N-c ej.

^* ~

u,

• Td'd ; g i

9l%n* t, *M. [.'.

• 'ecc

'.q% V. :. ? ls l'.'

JI".3hA s}*?.9k',

9 vi,<.-. y,'..

%e%' j': *Jfyg g

Y,

• *i t

v

., r6'.

' ~ b < j, % *)g 4

s

[ !J, 4 )

1. g ' ~.

- +

> p. ?r,.

4

.:.ec..yp ha,a s..:. -::,w :

m

• J

e i

e e

e G

t h

'S

. g~

g >"a, y_,-

m 9

. 37.

l

.?

i 2)

'=-

!h hh M

e i

h

....I

-l 4

l'i g u t t 10 srN 781

} g g y, 4

-[.f _I

' q}

f f-

,. s -

. 3

., 3_ ;

[> f } -

,.t, 3

PA M :. -

. p.. b, f 's.

_.2 1

1 ffl/jp;g, dh d Qwkk ww gv A 3.gs y yr$

m<

s s -

.i 4

0

[

,5f ' % ;

1 3

L F

, T+:#

t.;

.,.4.

... a.

. a +/;:..-

.g, e.-

s

?

.I 5%i

_yggp 9

q

,1

_ ' iy.l, '

ill4f d y(

~i

[, d I b

\\ [hf lt b

l 6) l l

l j,

g..

,s

%f s-4 yy J f,. ', }n.

c

\\

_h g

.? *

+

4 9

,s j.,, '

%\\g n\\, $* '-

l 4

y g y ( R.1 ;: t s

re

's ;-

4

.:n O

l' N

, w?

,, '. 3 p.,

~.'

s

. f' wQ%.a,$ ~' l%_

' Q,.

+

l, w g

.$?

a

.g n

a..,.

! ! s* u ! L' j

'y[

a

_,,, ~. _ _. ',,

.,9% : ' " '

f 4. -. **. :. _...-

y Wj' N4 A;

W

%..'.ts]

f y Q )

.."?yp - f } n -

w 9

p 3,, fr~yy

,f +

c 4

,c y-j YN-

~,./'. -

' N. h ' 'j; ' [ ' '.,.,

d c,; ;s.-

2'

.w t.

' [ [.. -

, a,g

'3n..

1

~.

1 R __

4 's <

p 5:

e

~.,4 y. ; _. -

/

w -

mc4 s

'(

1,y..

..s-g.

[

q v-s,f l J

. ~ ';_l Y' f

,b

..9, % A I ' '-

u ky M

..A.

p

, c.. m.g: *<

s

.. ; 4 _ '...g g-j

.e Ngure 14 SfS 781 500K

.,o

,l A'

,.T f' g o

! gi>.-

+

' g,.:g, *.

P n,

w v

+-

[

2

.- ' f_ /..

, s ']

t.

r.

% -up g')I

{g\\

M "

'x

.s

' h g 4 g.. '

S f

]

f:,

a-

p. j -.3,.

-o

4..,,-;...,.,,

o w

y -

. l _

g.

y, W

jI

? -', g j. ~ f jh.. p

7..

.' t x

9N Figute 15 5

751 IN

' ~ ' - '

-s 5

a g'

+.

e s N _'

w

'3 w :

...(.

. f :,"

I w.

y

+

om w

-a

\\

+

a hs

.;. 4  :.'.

v s) s.

l.'%..

A g'~

j 'i '

~

.de h x '

s.

h

. :,.4

.s P-Figure 16 S5 781 5000X l

% ws

y. %, e a.

W@se&.+fy :...

d I

f

.?

e.

w p

..g

~ +,avy,

h,t:

m 9.-

s 1

- 6.y

. T' l

2 Vgal}Q \\

/.gf, ".

\\

%'(

q..

,. 92% h y

,\\

~,h ' C',S',::T :(* ' / r s){*;". yf-'+~Q c:.k . :' t,t ga[~ i.a.

y' i Y'p'.:.

j O?p. gl,)),) 'j N, w

" 6,s y ??

na c a

S 0

6 G

e R. 4;w A

'. g.y <,

M',e J

4-s.~

[

9'<*-

,4 i

g h

s! ", e

..% ey g>,

.s a.,

E Y,y@' *, +

a

,4[

'

• j,f_A y+

V r

a..w

'3 et

.:.,,,d W

g[fA(.,>

e

-Qff t',l *'Y'.

? Q g.

%,p\\.1: l n o.

4.

u

-:y:

m.g, v gw, vu =.,h.

.W-

[4 e < #

I

. i

• 4 a

9 6

Figure 16 S.N 813 3 <,g x a".<,'94.:k1%-.

-.. v.~

g, s,:

n..

... 4

. -w,s 4

1.e hs :'

f w w>.,.n,

-., - 4 if 3

S.

<- ?

.irt >.- ';

l-:;

q..,

. g e

6

.g' y

T

's

.u

'y x.

s,l ' ~

y g +, '. '

t }g h

i i

l ngm to sxys

4 9

9 9

'W e

h (I [ b EfNt%3#D6M#hE Sf*

I v.g a' ),

y wm nwsv6 e

m 's:. f. h41AfN.*

k V(,:lp

.,qq% Mt v

i Mk"J'tj,g

,.c:w.c r

'u %;

' y i

3 i

pkj' tf';V""ff lbDhhf

. "Y$

D j

5%w$p$?al$58 a s e W p %e ~

~

igg a u p4 3 g g @ i

n lSild '

,ugg t

O}]R-^tp'JEgS*g*tkNh1-~_.

n g

cc E

1;d).

T YMN 4 / +

4

), 's4I l_'

6 ib

9 4

D e

I i

M

/

.,.t A..,

.{;.,L. : y -'

'I

,e o g(.

~.'

\\

s

, l P,, j f

,',s.

g -

,9 g

k~

8, -A_,

L'

. p,

\\

.l;1l k ', / k M \\[,, D } /4,

. & N;.; l, s

7, f.-

i.

,)

N

, a '['I"g

.,#,'.,Gt

'1 s

4,-

4 f a(L.

gx K4 11

3. g

. g,.

i Figure 22 SLN 1623 S M.

R.

e T!I8Ig .' " l.

. rg-(* 14 4

'?( ' :

ea y e, n ;g -(t- (;e

. w..

l

~v

~gg l tt, c.,

I.

[( ' ' '.

f' ' " E '-s

, ' ' g;. - r -

% tq,

p,.- g,

,' ' / tj ?

'J4l. ;h,',0 $-

{'f i..i

- +

{ ($ M ' 5 p.k:' ' h ' ** /

~

l

.iq 2p 3 cp. 3 3-1 l

7,I, -.mA[S' " ki s.

s n. -

l W 1 n.-

,3' ge. 3 t

6 m.

q, F.i g u r e 23 5N

'52'3

i

s e

l e

0 e

l

' !i.h 2'd 'Af l',',.hf' 44 'h 8

.j

, s *,.g kl l

~.9) ' '

l 1

l'

,.. Af, y..'.g % ?

• 4{, f, '.,.9,l.. f 'Q,*Q*

r l

+

_ s FT * *

• 5e l E

I g

5,..,.' ', 4.F M Q[d s..

- [

di j e

' t. t.:,, Q. r c.a.,,y s;y '.. p,.

,, g

\\.

• 1

-}. y 4

.,"su'l

~V 48 4

1 it i,,.

y

, ;hk.. i,. -...' bh, ) ' :,if.

. :. 1 '-

• A+W w

.5

,f'pt d,, '

)

i ad-g i

i

,A <

p%.

t,, %

.r.' $

g

~,,

f ' _.L y.

~

s

),, v. -, -..'

.r i -

q.

6,.

• N ;- ?,'

s+

4taQrn t,

a,

.I'

.f )m., +

5

,;;g.

J-f Y '

l yg, 1. k

• '.,A-,, '

s

% PEN gs

'J.. $

$F %'j%i-j6'h),

' ' s

^'g i

3

,.1

m -..@

g'.1

. J +$_hn. ;'

p.

?

f y.J J'l*4 *

, : e

. $gk< l%'

• R' 4,..'. W.V

.~.

j' h.'.'.e

,,.,...s. :

g

-..N.

J.;

. j r.

(,

'_,;j. N I'

O-g,.j'. '

. j -,

. q _. 7.,,

.. y

+

-r

_y

'E 9.'.

Figure 24

- N 1873

'iWA 4

b l. '. ' " 9.I ' 'i.'. '.t - s.

D )

l' i.. % ;,' g...e

~

~

r.. :

2 *

' : #., - s

+

1

.g e

.s j..,

5 e

9'.

4 *..'

M 4

g

^

4 f

9 Iv

'q f.

.y g

e o

t,%

.5

'i fQ p

r:

...1. s '.

4

. j.

a 9

e

's a.

fA

,. k',,*

, %.S O - *-

,- '," d.,:

i[

.. W('

p.;,,

4,

".., +. ;.

.... i. o.w.

y0' i

b3.

,,,p., a..i y g.,j 4

fhg p,'

- ;4

,4, wla

, "{ }

g [ *.,

,[ g

1. 2

' *I

5. 'I F.,

i 4

f p,,

j,g 9

4,

-s ].- +,;.. '

t

.....}t..y.g pc.' r 1f h. 5 pd

+

., 2 i,,,

w.

% i.

s,

. ; S.l-f

-v, t

.g.

,q a.'

s d. a '..

Ay,[.

.y

.,j,,

+,

.., g.,

di 2 -

it

.'m ?. j;f.. :>.nk..

.>. w,

e[%}H,j f f-W

,[& fy.

},).

w% ;,

c,a #,m % o

,y s.-

g b

3

I i

e l

l

/

+j!

,.gw- _ _ws,.y,,.,,-

a 4%y, a

m

.j

(

,. ; p l

l Pigure 26 S 'N 1823 2200x

,ca.

,1 p x.~. -

f; ~-

l

'h

~.

,[

V

0

~f

_ _i 7%.<

g. %. (

g W,. - ".

[ (o.

~

- L. ' _ ' ';

,. _ e ~.x

,n h.g. _ -

4 g

,r -

r

.: 4 a,

k!, k ' ; -

,)

l > p3.KM {,7: -

• '1

..~.

n J

N,.4 k

[

,.8 I

,d..

1

~

w L,

4 a.

r y

w t

t s.

t._...

4; '

• s.

.4 4; ' ' jym

^

sm s

n gui c 27

9 9

h 6

l I-WY f.

~

.s

'I

. =

'A }

.t

\\

.l, _.

..}.

,o 3

.t e

S

. : n.

5-

'f._

2

-<-p.-

~..

g.

h4.

}

j ':

' :. ' j}' I, Q._ t l.

yNr.

f m

/.

}-l,[ _ [ ; } \\h Figure 28 s/N 1823 20001 i

3 p.

.. m'

.f

.g

L:

f,,

.'l W & w, a.. m e.% A,

W.s

1. MM A.,:#

v

.. x.

u

.x% n

,3 :;,n g

' ~ =

** ' M' 'i'; ;,l y' ;n f '.j.l;,? $. e

't 'W-

.w.

c 4 :&

J.

' ~

h..

.~.

! c 4,,:

.p

• -t c

3.,..

l g,..,

e

'f '.

f,) - ll ',-?

. _4l.ll l

. k.

l _,;*'

l J ' ~ ~

'f

.c.lw :

$$$$$$dk@$.g,db

. +,,.

\\

l Figuie

.Y

,.,c.

1 J

w-.--.~.--.--.-.-.s-----s----

h 9

=

S e

_pyh 4

o

,t

-.f..

,/

-,)s / '

(~'r:n,

j

).,,

w

'/

'M s[

$fk

.s.

I'M S

-. t

' a... w,. ~,

n m/. T,, 'il i pg,;. w,.w

. Qf: ;.

i 1 * * :.Q:'t/

i*

x

.. #,.:. +

c~

8 W "y fs k, im

.... ~.

v Ah..[ 't

'A il -

' -i O h

.I

.J

+:

/

n'

,4 i

/

+

.s

.y'

".. t j, ' ', '. '

b._.

., ',l

.. Q $:_,'@lY.-

g, f

3

.,t or

t

.'l ','fl,\\. ~,'

• A

, 'i

m. ; 7,.g m s..

,. (. q,.. e. y g

.,e.t 4

.n ; j. eg, '

}'.

', f

.,e

- Vo i.

g-

..,..... y?

...v. :.

s

4. -

,1......,

- k ' f f k:,,

. c,

...,,' ^ '

~'

4p c c...

. 'i;f l '..s.

h<

.6 \\' [;.;%,'!{;j.

• ) b,

4

'%,e'.T:/ J. M.,g.G h 3t") e.'.

g e s tw F 13 m e 31

t 9

G 4

4

-)

s s.

F

+.-

o.

e

~

~*

.q; 6

I

'1,

'{

q 6

Wk-

...*c:,. 4 - 4 J

'M

'g

' ' j

.e.

o 3

g- ' ' ^

..t

• ~

q r

's,i w'

' t' ' ' ' '

a

'^ *- '.'

• k'n i d

-@g.gs 4.,

J Q. N. [..

t $.,,-

~ *.

. i,l;Il'3., '

4 di;[ '_ i-s.

<,hsi

. nJ

<k.,

4 r ei w

4 =;/g

< -, g -- ..

4 s-

• W

,,3.

4 1

s.

y' 4 y

,y cE-i,,. <

J ::. *

"',,P' t

, ', ' * -(,

.,;,7}*. def,*

'

• h.j.'

4:.,',

.',, t, y,.. Y 4,k) t'

' ',_k '

[ ' j.

,s, r, '

-:.f, i $, :.

f, g

4 4

4'I,

[,

+-

.'*"4

, 4 g.,,

f-( Y,

,. - : :i -

g- -

9, g. 3'h v-

.f 7,.,,.[4 [). (, 'l -i 4

' ? 'S, #

. f,<

( '(d " '

h

,4

.,,. ~.'4, o,>,p.,. bl,..

~

..' f Q.,%.i:.;)3,l'.hi g M

p y, gc

. r.

. ;w, <. $.

,y

w.

,.:...n.

a. m.., y Pigute s.'

N 1 o '; ',

} 1 i.11 0

O 4

e e

',s ' '. i.

>e l *

.'j ly&y )14 ep.

.S -

.WJ w

L,;

3

~%

l Xu} Q %t

', g&CpyRQ:4jk..

4 Q MANDbh l

.., gsq.,re., e,m, e o. n ! N. t,+n.

3

.3.

g ga x;w A m u - (

y... < 1 Hg f, ;,.!gl k.'

N i

aptatn[y;y; yy-y k

9 !-

v. s y v

. ;IN "l

4. &.:i';.h, i d E d q_, gM-

-/

i m w:.

.[i"i Th:[. m[k,hlf[b

... x s.

E f1 hk5 ' ? >

k, ' '

,. r gy 9 %m, w. a WhQ I4D il $:s.1 :

rp~y t.jh >

l Nd,s '.

Agv Qb t h,sj e.

w' @ j, f, / m t w.jf.

,y

. g

,7p Ff*

!nyjf

.,n

'*l.

'a u;1.

Qp

.a s;.

'GhuubusDddLMhewag.

,e m:-

v dENG 5 G n.. _..

6 1trure 34 t N 1%1 lOOX kk,fl&ffN', ll?f$flfff-fff? ff

1. p 9; W ^ g ck Q $

wuan

]p H ng n% g y.

,,m.,

v', 4.,s,.c >!*p;+)b' y.:

I

,, '%:-b$,rtv,g:sf (i,,p,;ylgN.,? g< e 9 }i ~c 1, M

,, (> lg,_Th*m,4 w&p(yer, rye if

,k.

i e

n

..y n,n, N.ds;g;gg:';k N ; v.o i

it 5

3l:k{){e.l

....q b. lw

+2c W

'( *

] 3,i.t. >k,. %

i r f rP,,, J:

h', ; DplM:b !;M. 2 9'lt. e,s '9 %v. a.. j1.. "., s

1. ?

g< c 4

,4

,c i 4 j lJ:r R.-[bf; u' ;

v.

.n f yJN p 3

' 3 f O' sip'{;Sf.c,.X O.:hp.

p

. f. h.n!..

,$N' 4i g.%%.w,.% ' %cM:i a lu..A ' a.t,.

9-i43

<^

W t

g?Q;3.;1 e..; W<

MS, s

,n

'1ti i i 1 -5 S-k '

gg&g'y :yc W....:

(.-'

'- k : '

ig a..-;,.;..,

y i g

% _,.n; >&, r;$r.

My.,

~;

,vg, yun sje r.

x

,, Q f N

?'fI

.$ h

=:

,li f ": ' % ;L,Q

.J! e'p VQr: j b, y%. $,

u,,,

w' y a.f.-

f-jy w* [Q; '

v i

t..Ei ;

g g

1.,

3 i

s s.

i: J g,

y m.,- p' t

y

,l4'

,4

?

y

. ['. (i}

• y i:\\,.

i.'

y (g1 N

s.

.7s. N 5 j

a y

a iom s

0 o

4

. e4

'l t

.t,-

5:

0 6, y s

S ypl

.v5

'.;f.~>f.k.g,y

R'Q s.<Q,$Q

$~%%2 M 3 Q

.w.

'a[' '

..e ep 5

o 9

'3,;

j kg 3

-,..gN f n 4L$

s\\ 9

-e

.,e.

...3 g

Firute 36 s

's 1 (> 81 1rvs i

,) / / * ',..,

. p..

• p'

.e b

~

$. f.,

j f k R :l'( Q :, j

[-__

m w

z

.,z j _

y.

n.

v w

r

...,.{

'i i. V%r}.

' y % Q-. L Q f ;;, y 6 ww cw" '

n

'[.'

• A' ra x A~.me x

%,y s vg.

gt.

. m, w.._

4 p

r

'f

)

[I

,,h, ) -

5..

h;. i.

t g

wtgr

' del

, x' g,(._,[+.i.-[.[,,.'sq(f.,(, [,"'

d y f,,

',s.,...

7

...!.,p i

41

-ps

.e

)

sa k'

'\\

\\0

t h

1 i

8 ygifi'9-pg f j

-4 A. _lqtw@p%pq p 9

,up, qn. f k )(

ygy,+~

4e 4

... gm

[h h' ' -

,[?

(.,.]M[.M 1 (

W,,p },, ; MS W' Qidi:

1

+

Sf p Qs,Jf j

Agg d e,Me m ;

>.,l p 4

-, s.t o W

J j

w o s&w{ g+;Q?

%%Ny y r

J h

3 c, il'A$p 1.,#.l -?d(.!M.. agy;,q.h, %

kf f

i N%yhmhy y

'@\\'E x,%jk y *P.'

M My'l u;p' u.

t i

.' f;W a3.

ed d

i'pf '.' ' s g,NbMs 9

% ]bh'.g~ 4 ~,Je o

y h,

• 6 ':

5 %,/

R&SQ &,-])k h.lW. @g39 t-g%

f.,yQ y r, ';w l

V

?

t.,

y 4M. y MM UL &_" t c.y+p m

...,m

, s I

p m$, m41 ).

M+ p&,v e';i, 30 i

^

s i

$ ~ np y& @rmpnQ Ng 3 d

V h w' 7

J d

~~

,.Rigf:qq.L;$?:

h.

D l

m s q$.If c { pd..d, ; +3y, ;

,aH d

% :.aj e

,,. N. a.'.:t b Kj 5f #,p t

.c gj h

r%/. r y, y l ;.g.;t 3,)nh M,Icg';gg f p 4ht u d %c:ir %s3 eev o

v v

u wom.o, nw.&c m. mw n w,m, vs

. ma q 4 m p v > t M Q PQn n:y w p

,,Qw m,fM ; Jy: m q f nu LV hQ ysC y Qth ;3Q%,

q% %q d,h hkf:;w"wa:m:y$

.,p.

l t

n W,,

f.kI y-Y'g.hVf

%m ly&.a$?am%m ;,L Jy&gf$myn@y p nJpp ghg MW:nwypmana M

Mw Wa v

s 4Q 5Y&

h iMM f.fMfj MM Maulnam,wn %a%n: x sk M g aduk n%wnMM n ' n Figure 38 sN 1960 1 a3 f

l l

l n

f k hk!.f n n. n g,Ypm 4.

h A_b a..p p y a<m,r a-

,%u s ne %s??.0 N

M@emm:w e m n gy p w %g.gff.g;pgp 9A %jypq p

p mMww;mmq 4 Mh)kh[h$hh 3,)Nwb $f[4 l

W

!M$ w$ ? & &Nb M N Q h @b'nDWQM$$$d M. 7Nhpm Y$$

w@y}hh,gggAigy.pqyQ3%pd AAmuw 4W%%g%y m

y W-O Y

h h h k k h g$N e

@ 4 h k & g d f $ f,M D E h

y J

. MWMMd M. m'Wr Ho,@y,%+ng% %.4A;+Ny3..,

l vNm g%W';.

1 hm g %y % mv h whj mg y %. q 2 u

y

' $~/[.s:pTN QM 6;dd0 @Mgj %,d v i A

• M c m. a'm'Myt W tet

< m 'm y N $ JW N.

WW 6

I Y

) "

pW/

ip '

\\MWpyQnny;p9f;mjig Aam epuw4 My m p /g 4-g p4:ww&ym,J c,e g r M aa w % ypw q

^ qq

@; k h,y,Q

-vA@MygfJK DsQMc NOf @,%INgfQU Q f

egg w e f 6-lp 1w%' h lh

,..a-my w&m u%m %.i.

W ii M N e m 8 4,

g a da. -

a -dig

- s Mu v

g~

m[M cM M' p w %o m/.hy(d') [(O

[$' Ird h b O

$.p 4

1. ^k' b.INp%NN;-[

Ji

~

V' il $m%y $;D'tk P

na ww&;wyRyha%&Wy:;f~&p% MMn{ypngMt,9 y]nky WWfW f

W M%R M QA9

$u&pG WWij!k;JQ NZ%g y y-M)wWh$q sn a' SQAlq 5

?

wee w w; % x-y+g m.p.. y tm 4 yL p %g

-4 sMq %,

...q;m wv/;--

p mg ws p; &byd. m i ps W m.74 n&We e y..p 4ep.m5 40m a.w o WhwnWWh x, p#Ws a bM,4 3c. py b w&bw$iM.6 m py m, t eNv 99 i,

M 6&(a.d 5:

f i r, u r e 30 108) l

0 e

0 h

hi, "s hhue 4

fA.-p Eg,4

• 3

4. icd py ' %

f VQl%, Nyt

.. Mf1@4$7' "

YS' ',;

(t.

b%

.f},gn 3 Mar,

'9 J;)?[

'y ';~;,...

n

.c e

h.

r sh,\\

k$:'

[i W

1FW u uL n gure 40

c. N I w$

1

• a

._,._.(.

j'?' L _ _ - ' l'.

, N ' _ lY

^ '

^

N,;il, q r; r.

W d'.

s

~>

3. &/, x

\\

Yr

. l.

. 3. d[.

j'

?

y

'7g?w

.+ 7 ;.

'^

^ll]..?

, ". y

[ ~f, ;,. ; ; f.;;

~ ^

l G

q r.

.f ? g n pl f.

l h

ov y.

, ~..

> t.. '.. - g;.

I'

- Y. f'),

g f

f I 1 p' L. I O

, [4u3

++.

i

a o

e e4

.b e

t

,3.

4,'.

, kuj4 )

a st Q\\

.. +.,

i K.

1,

't

+

_.mpq[,;;

.,..*",t

,,,3

'. [f[h, h h Nb *: g 4 a:

'?,',je i. < ;, J ;

N Q y

-.jp

'.t s

e h-,sa '

g.,

p.

,--'J t

sf

.?

,x.

.+

n

'j:',

3

'pl4;q

• y ;

'i@%

X; s.. -

r 5M-i.y.hi@g. '

' -}f,y.

1 y

W wy,;9,-g4;.t :i c.

" tj(!j}.-

.:.f l l y'[In 1;'

ng;. %

l1..

<, ;p.F 4 i "g;.g41, i l

s l

Figure 42 S.N 1949 iggy l

l f

~nn,gwa. n. u.

lR%;m,b.. %+MM: % i gji.he

\\

.k; t e.+.er;qN.,, %c 3 4,,.

M p3 m e

g, pm,.:,

gg %pec,%

A5; p

g s-4 d.

1 m.x v. -

e DE:!. Y' q

3+'

iptst, q,v y Ng u

e

.**/

QO;h,f'

$; %l N,'

w.e-

.;.p f &y,<

b:

1 9

u.. : OE.g.

'4 e\\.

a.

l 1

' f

/,

f'

'h k

Iigure 43 s, 10 %

i,

-