ML20041G433
| ML20041G433 | |
| Person / Time | |
|---|---|
| Site: | Brunswick  |
| Issue date: | 04/30/1981 |
| From: | Wheadon W GENERAL ELECTRIC CO. |
| To: | |
| Shared Package | |
| ML20041G431 | List: |
| References | |
| NUDOCS 8203220293 | |
| Download: ML20041G433 (31) | |
Text
.
ATTACHME WT 3 E Nct.osu RE 1 HIGH DENSITY SPENT FUEL STORAGE SYSTEM FRICTION TESTS SLIDER PAD- SUPPORT BASE MATERIALS W.C. WHEADON April 1981 l
i l
l SPENT FUEL SERvlCES OPER ATION NUCLE AR FUEL AND SERVICES Olvi$lON e GENERAL ELECTRIC COMPANY SAN JOSE, CALIFORNI A 95125 GENER AL h ELECTRIC c oc gefj29,3 e,2pof,2,4, P
l - _ _ __
4 NOTICE AND DISCLAIMER This report was prepared by General Electric Company for use in relation to high density spent fuel storage applications. General Electric assumes no respons-ibility for liability or damage from any other use of the information disclosed in this report.
The information contained in this report is believed to be an accurate and true representation of the facts known, obtained or provided to General Electric at the time the report was prepared. General Electric Company and the individual contributors to this report make no express or implied warranty of accuracy, completeness, or usefulness of the information contained in this report with respect to any change of fact or law set forth therin, whether material or otherwise; and General ElectH c Company makes no warranty or representation, express or implied, with respect to the accuracy, completeness, or usefulness of the information contained in this report, other than for its use in high density fuel storage systems, or that the use of any information disclosed in this report may not infringe privately owned rights including patent rights.
Page 1 of 8
l ABSTRACT A program was developed and implemented during which materials were identified and tested for use as slider pads and support bases in a spent nuclear fuel storage system. Breakaway (static) coefficient of friction values were determined when various material combinations were caused to slide against each other in a controlled water-lubricated e nvironnent. It was required that the slider pad-support base system be capable c' operating, after being at rest under load in 100 3 1250 F water, for up to 40 years while subject to neutron and gamma radiation.
Various test data and specific test results are on file at the Spent Fuel Services Operation, General Electric Company, San Jose, Ca. These data are proprietary to the General Electric Company and are not contained in this report.
l Page 2 of 8
PRELIMINARY INVESTIGATION AND SCREENING During the design phase of the high density spent nuclear fuel storage system, it was determined that a free standing module approach was practical if materials could be found to serve as slider pads and support bases so as to allow the storage module to slide upon the support base if subjected to si tsmic related horizontal forces. In order to qualify for use in the fuel storage system, the material (s) had to have previously been used (or approved for use) in underwater applications while being subjected to neutron and gamma radiation for prolonged periods of time and at the same time not be subject to galvanic corrosion when in contact with each other. In addition, the coefficient of friction of the materal(s) had to satisfy the requirements established by design engineering.
Literature searches and personal contacts were made to determine the extent and availability of existing friction data for materials which could be applied to the fuel storage application. It was found that the United States Department of the Interior perfomed sliding friction tests on metals for application where the mating surfaces slide over one another under high nomal pressures after having been imersed in water for extended periods of time. In a report from the Canadian AECL, Chalk River, Ontario, a program is described which was undrtaken to test and develop materials with good resistance to galling, wear, and corrosinn fnr use as nuclear reactor components. The tests included measuring and recordin9 the starting and running friction values. In a paper by S. Kato and N. Sato, it is stated that when two surfaces are in stationary contact under boundary lubrication, the static (breakaway) frictional force gradually increases with an increase of contact time or stick period. The paper suggests that the static or breakaway frictional force will not continue to increase with time after about 10,000 seconds (2.78 hours9.027778e-4 days <br />0.0217 hours <br />1.289683e-4 weeks <br />2.9679e-5 months <br />).
As a result of the intial investigation, several mate ials were identified that seemed to have the necessary corrosion resistance .id prior use requirements, l
but breakaway coefficient of friction data were not available. A limited test t
program was begun to obtain " ball park" friction data so that only those materials which appeared to meet all of the pre-established acceptance criteria would undergo final quali fication tests. During this period the candidate materials were caused to slide against type 304 SST in a water lubricated environment using " standard" l equipment to measure the coefficients of friction. It was desirable to use type
! 304 SST as one part of the slider - support base system using standard machine i
l finish surfaces.
I page 3 of 8
\
OVALIFICATION TESTS During seismic activity, the fuel storage container, and hence the attached slider pad, can move in any direction relative to the stationary support base. Such movement could be cyclic and the direction of movement could change at any tirre.
This random type of movement cannot be easily simulated with the equipment used during the previous screening tests since the relative motion between the test materials was unidirectional. A test fixture was designed and built which would slide a test block (slider pad material) against test plates (support base material) in a back and forth cyclic motion. Each complete cycle, from top to bottom and back to top, provided two (2) breakaway friction values. This continuous cyclic motion may more closely duplicate seismic-related horizontal movement than would other test mthods. The test fixture also allowed the test plates to be positioned so as to evaluate various surface finishes. A hydraulically actuated test machine was used for all cyclic testing. One test plate was bolted to the test fixture and a second test plate was bolted to a movable mounting plate which connected to a ho ri zontal hydraulic cylinder such that a load normal to the sliding surfaces could oe applied. A test block was attached to the test machine vertical actuitor rod and caused to slide between the two (2) test plates. The test fixture fr.me, and thus the test plates, was attached to a load cell, which in turn was attached to the frame of the test machine. The force required to move the slider block between the test plates was measured by use of the load cell . Demineralized water was ci rculated through the test fixture. The configuration and surface finish of the slider block and the test plates were Closely controlled. The surface finish of the various test pieces was measured using a Brush Surfindicator Model f151300.01.
PROCEDURE and TEST SETUP The following test procedure was used:
- 1. The inside of the test fixture, the slider plates and the slider block were cleaned with alcohol, then with reagent grade acetone.
- 2. The dimensions of the slider block were taken and the contact surface area calculated to determine the hydraulic pressure setting for the nomal force actuator.
- 3. The slider block and the test plates were mounted into the test fixture.
4 Cover plates were attached to the test fixture, the water Connections were made, and demineralized water was circulated through the test fixture.
- 5. The nomal pressure was applied to the test pieces and held for a specified ti me .
Page 4 of S
i l
- j l
- 6. The normal pressure was increased to a specified level after which the slider I block was caused to move and the breakaway forces were recorded as a function of time on an X-Y recorder. Most tests were continued for 200 cycles which provided 400 data points.
In addition to the various material combinations, the following test variables tere closely moni tored.
- 1. Magnitude and time the normal contact pressure was held before sliding started.
- 2. Magnitude of ac rmal contact pressure during sliding.
- 3. Surface finishes of the test pieces.
- 4. Number of test cycles (sliding distance).
To determine if the breakaway coefficient of friction would be influenced due to time under load before first movement, the slider block was held stationary in contact with the test plates for specified times under pre-established mean contact pressures before sliding occurred. In general, when the test data indicated that specific material combinations may have the necessary frictional characteristics for use in the fuel storage system after the first hold time under load, additional tests were perfomed using di fferent hold times.
DATA EVALUATION Analyses were performed using data from each test to determine a realistic breakaway coefficient of friction % when the slider block slides against the slider plates subject to the physical conditions maintained during the tests. Analyses were also perfomed to determine if hold time, normal pressures and slider plate surface finishes had any effect on the coefficients of friction. All mathematical analyses and statistical comparisons were made at the 95% confidence level. The manner in which the test data were mathematically evaluated is signi ficant. Analyses indicated that data from none o'f the tests could be considered as coming from a normal population.
It was, therefore, not proper to use mathematical techniques intended for the eval-uation of a normal distribution such as I plus or minus some number of standard devi a tions . If one were interested in constructing a confidence interval for the population mean, the deviation from normality is not such a serious consideration, unless the deviation from normality is pronounced or unless the sample size is very small. This is a result of the central limit theorem of statistics. The problem at hand, however, was not to construct a confidence interval of the mean, but to (1) evaluate the di fferences which may exist between the data from each of the various tests, and (2) construct or determine a prediction interval defining a boundary which could be considered to reasonably contain a certain proportion of Page 5 of S
4 future coefficients of friction when sliding occurs, or to verify that future observations could be reasonably expected to come from the population defined by the test data. A procedure was used to construct the desired interval which assumes only that the data have no discontinuities. The limits using such methods would be wider (more conservative) than those where nomality can be assumed.
Data evaluations included the following: 1) Comparing first movement data with middle of test data and with end of test data, and 2) comparing data from first hold period with data from subsequent hold periods. Two types of statistical comparisons were made, the first compared average perfomance (mean) and the second compared the diversion or spread (standard deviation) of the data from the average performance.
If a statistically significant difference was noted between data sets, it could be assumed that the di fference may not have been due to chance alone, but that an assignable cause, such as hold time, or surface finish of the test ?lates may have resulted in the di fference. If no signi ficant di fference was noted, it could be assumed that the di fference may have been due to chance alone.
SUMMARY
Based upon the several evaluations and comparisons of the test data, it was concluded that five (5) materials, when sliding against type 304 SST and subject to the conditions established during the tests, met the criteria established by desi gn engineering and could be used as slider pads in the spent nuclear fuel system.
No statistical signi ficances were found which would indicate that a further increase in hold time would result in a change in the friction coefficient. The evaluations of data from these five (5) materials also indicated that the average coefficient of friction was consistently repeatable between tests and within tests.
The following is an alternate explanation of the experimental data which was provided by a consultant wno is an acknowledged expert on friction. All the reported sliding tests were in fact carried out under essentially identical conditions, and the differences in friction may have resulted in chance differences in surface cleanliness, etc. The non-normality of the test data is attributable to non-normality of other influences. Throwing all the friction coefficient data "into one pot",
one can state that for surface preparation techniques, speeds and pressure similar to those used in the tests, all the friction coefficient values seem to lie within the desired region, which is quite satisfactory.
Page 6 of 8
Using nonparametric techniques to establish a prediction interval, it can be asserted with 95% probability that at least 95% of all friction factors which may be encountered in the proposed application will lie within the desired range.
APPENDIX The following curve is based on the theory of non-parametric tolerance limits and was prepared from "Non-Parametric Tolerance Limits" by R.B. Murphy, Annals of Mathematical Statistics, Vol . 19, 1949. The user of the curve can make the following dete rmination: How sure can one be that the limits found by drawing inferences from the sample actually include the true limits? By increasing the sample size, the probability 6 , can be increased. More specifically, one can assert with a probability 6 that at least j@ percent of the population lies in a range estab-lished by the extreme values of the sample.
A special feature of this method is that it is distribution-free. The character-istic need not be distributed normally, or in any other specific manner as long as it is distributed continuously. When the characteristic actually does follow a normal distribution, more precise statements about the limits can be made using a table of normal tolerance limits. The above techniques for establishing population limits should not be confused with those for estimating parameters of the dist-ribution, such as the average or standard deviation. Statistical methods provide di fferent rationale for selecting the proper sample size for such problems.
Page 7 of 8 1
1
. l
' l
-j
-r i i ,
i i !
- i ,i .
t i l'***o eeoo
(
l i I I l t I i ,
! i '
I l e see ;
Minintin number of ra' ndom isamples (N): l I i
,,,, I requi red to assert wi th-T "-probabi44te--. j.---- { - t-- l --t -.
- uni ts havp '
lowest-b --, -"~' , ,',
i j
*- thTt it TH35tg % o f, ali ivMues b.e. tween ths hTaheit7nd i i i f - ~ ~ ~ '
. .yalues of the .samplenVAlid_for_any j i c'ontiriuous distribution. ' I I I
' i i I t ! i !
+ .
i
-i l
! i i
r -
1 ( l i
l t l '
' i i i
., i,
' I ' -
. _ . _ . . . _ . __ 2.
, . . i i
i ; i i -
. l i i ,
' : i i i ;
i . ! 4
. I ? I I I
- . i j
, ! i
! . a.
i ' ' - ! I , ,.
. . _ y . . .. e ___ .. , . . _ a .._ -4 . _ p _ _ ._,. . q
. . _ . . . _ . - - . . . _ . . _ - - sce
. 4 ..-L..4 . . . _ . - I . 4 .L.-.d. _ _ . . _
- 2 c
~ *
- __.._{
. .. _ __. a. . .. _. . . _ _ . _ . _ _
i i l co I
N
[: 99% . ._..._. - _ __ ..
i
. . . _ . ea i
. < 6 i
b 95%! '
. ..- aa
_ ~ . ._._ - .
.O o
o
. . . - . .o t - - - - -- - - - p - - + - p - - ; j -1
. ____. . . . __ ; . . . _ . . L. ._ _: 1 . .J
' i i I'O i ,
' I
- . ! l ,
- O I I i ! l l
- .i. . _ . . . . . _ _
i
-- L _ _ _ . . -_! '
l' i l l I j
' ' ' l ' ' l o
- ! i i i ! ;
I 1
, r
,i
' 6 l q
t j i i ,
l i
l
- i
,' I i l _
92 93 94 95 96 97 98 99 100 g % of Population
/"'
MT AcaMEwt 3 ERNESr RABINOWICZ 14 EXMOOR RO AD NEWTON Mass. 02150 November 23, 1977 RECEIVED CV 2 61977 David R. Sponseller Project Engineer General Electric Company 175 Curtner Avenue - M/C 859 San Jose, CA 95125
Dear Mr. Sponseller:
I herewith am sending you two copies of my report, in which I have incorporated the changes recor.nnended by your people and have added an appendix.
Sincerely, mi i ld 4 Ernest Rabinowicz Enclosure t- , O
$ /~ , t- . .c- '
'a *,' .. . ps;;y 1 "
Cyc.;:: w c. ..
i
.h 1. . k_ h.e..'.
F.e . *'..->t ". - :.'
l 3
F. . .
EMN30T MAllN!WlCZ to Exwrom noAD unwrom,wAss. omise gg gg C.E. P.O. 529-CC084X September 26, 1977 RECElVED NCV 2 81977 D. R. SPONSEtt[R Report to General Electric Nuclear Energy Programs Division Atta: Mr. D. R. Sponseller Friction coefficient value for a high density fuel storage system j Abstract Friction coefficient values obtained earlier in a study of a 304 steel -
304 steel water lubricated sliding systems have been used to estir. ate friction coefficient values likely to be encountered in a high density fuel storage system currently being designed by the General Electric Co=pany. The mean friction coefficient anticipated is .523, while the lowest friction coeffi-cient likely to be encountered, (1.69 sr.andard deviations below the mean) is
.349.
O q
'hy
- o l Dyl M . t l-l-l ma e
[
l Introduction This study was undertaken to derive a realistic minimum friction coeffi-cient value for a high density fuel storage system being designed at GE. In this system, it is desired to resist horizontal shear forces due to seismic action through friction to the spent fuel pool floor. The two sliding mater-ials are 304 stainless steel, the ' lubricant' is denineralized water at 70-150*F (21 - 66*C), and the nominal interfacial pressure is 700 - 2000 psi.*(1).
Some months ago the writer undutook an experimental study of the fric-tion to be anticipated in a high density fuel storage system being constructed by Boston Edison (2). Since the Boston Edison conditions closely resembled those of the GE application, it was decided to carry out no additional test-ing but to use the Boston Edison data; to apply correcting factors to make them applicable to the GE conditions; and then to use the reconputed friction values to estimate the lowest friction coefficient likely to be encountered by CE.
In the tests carried out for Boston Edison it was found that variables such as temperature and apparent pressure had little demonstrable effect on the friction coefficient. Surface roughness appeared to have some effect, in that very rough surfaces gave aslightlylower value for mean friction coeffi-cient and a distinctly lower standard deviation, indicating that the lowest value of friction coefficient (i.e. the mean friction coefficient minus one or two standard deviations) was larger for rough surfaces than for smooth ones.
All these results are in good agreement with what is known about the friction coefficient of very poorly lubricated metallic surfaces.
The only important systematic effect encountered in the Boston Edison tests was that the friction coefficient was a function of the sliding velocity, with the friction increasing as the velocity decreased (the typical negative characteristic, responsible for frictional oscillations). In the Boston Edi-1 _ -_.
l .
son case, the tests were carried out at speeds of .04 and 4 in/ sac, bracketing the sliding speeds believed to be of interest. In addition, a number of sea-tic friction coefficient values, af ter times of stick covering a rather wide range, were determined.
In the GE case considered here there are no sliding velocity requirements as such, but rather it is believed that the horizontal shear forces increase from zero to peak in a time interval which is in the range .09 to .18 seconds. (3).
This information must now be converted to equivalent sliding velocities. In a paper published some years ago (4), the writer suggested that the static friction coefficient f, corresponding to a time of stick ty, would be the same as the kinetic friction coefficient f gat a sliding velocity v ,y provided that 1
where d is the diameter of the junction formed between the two sliding sur-faces. .
Later work (5) has shown that d is given by the relationship 60,000 W ab P
where W ab is the surface energy of adhesion of the contacting materials (in this case a value of twice the intrinsic surface energy or 3400 erg /cm seems l reasonable), while p is the penetration hardness of the 304 stainless steel (in this case about 2.5 x 10 10 dyne /cm2 ). This yields a value for d of 8.2 x
~
10- cm, or 3.2 x 10 inches. The writer estimates that an error of + a factor of two is associated with this calculation, so that junction diameters of anywhere in the range 1.6 x 10 ~3 inches to 6.4 x 10- inches are possible.
A detailed discussion of this calculation is contained in Appendix I.
e
-3_
In the CE case, the highsst sliding velocity (and hence the lowest frie-tion coefficient) would be encountered when surfaces with junctions of the largest possible size (6.4 x 10 ~
inches) meet shear forces which increase in the lowest possible time (.09 seconds); the equivalent sliding velocity v g
~
would be 7.1 x 10 in/sec. The Boston Edison friction test data must there-fore be converted to this sliding velocity.
In the Boston Edison tests, for the first eight series of tests, half were carried out at the higher speed of 4 in/sec, and half at the lower speed, of .04 in/sec. The mean friction coefficient was found to be .37 at 4 in/
sec, and .55 at .04 in/sec. The ratio of friction coefficients is 1.49.
These experimental data agree quite closely with results obtained by the writer many years ago using an unlubricated plain steel on plain steel sli-ding system (figure 1).
This earlier series of tests suggests that the plot of friction coefficient against log sliding velocity tends to be a straight line; it has therefore been assumed that it is precisely a straight line.
Figure 2 represents the friction coefficient values for the two applica-tions.
It will be seen that the friction at .071 in/see is a . factor of 1.42 greater than at 4 in/sec, while being a factor of 1.05 smaller than at .04 1
in/sec. Accordingly, to convert the friction values of the Boston Edison tests to the CE case, multiply the high speed and low speed friction coeffi-cient values by the corresponding correction factor.
This takes care of the one hundred kinetic friction coefficient values, and leaves the question of what to do about the remaining 99 static friction values. They cover a variety of ti=es of stick, geometries, roughness and rates of application of the shear force, and it is suspected that the C2 fuel stor-age system may also encounter a variety of these variables since it is not a .
~
l l
l closely controlled laboratory situation. Secondly, the static and kinetic values have very similar means and standard deviations. It has therefore been decided simply to leave these values as they are.
Experimental friction values adjusted for CE case Test 1. 304 steel on 304 steel. Room temperature. Hemispherically ended slider. Roughness 29 microinches ras. Distilled water lubricant.
Sliding speed - 4 in/sec. Friction coefficients have been multiplied by 1.42.
f values:- 0.54, 0.64, 0.58, 0.50, 0.50, 0.53, 0.53, 0.48, 0.47,,0.47 i
Mean f riction ,,coef ficient = 0. 52.
Test 2. 304 steel on 304 steel. Room temperature. Bestspherically ended slider. Roughness 29 microinches ras. Distilled water lubricant.
Sliding spead = 0.04 in/sec. Friction coefficients have been divided by 1.05 f values :- 0.35, 0.44, 0.59, 0.62, 0.61, 0.61, 0.61, 0.58, 0 63, 0.59 Mean friction coefficient = 0.56.
Static f:- 0.62, 0.58, 0.57, 0.63, 0.74, mean = 0.62
( Test 3. 304 steel on 304 steel. 73C - 74C. Benispherically ended slider. Roughness 29 microinches rms. Distilled water lubricated. Sliding speed 4 in/sec. Friction coef ficient values have been multiplied by 1 A2.
f values:- 0.50, 0.60, 0.62, 0.55, 0.65, 0.71, 0.70, 0.70, 0.58, 0.55 Mean friction coefficient = 0.62 Test 4. 304 steel on 304 steel.72-76C. Henispherically ended slider.
Roughness 29 microinches rms. Distilled water lubricated. Sliding speed =
0.04 in/sec. Friction coefficient values have been divided by 1.05 f values:- 0.44, 0.50, 0.49, 0.61, 0.62, 0.68, 0.72, 0.57, 0.64, 0.78 Mean friction coefficient = 0.60
[ .
Static f:- 0.64, 0.46, 0.57, 0.56, 0.74 mean = 0.59 Test 5. 304 steel on 304 steel. Room temperature. Slider end diameter 0.09". Roughness 27 microinches ras. Distilled water lubricant. Sliding speed 3.9"/sec. Friction coefficient values have been multiplied by 1.42.
f values:- 0.51, 0.38, 0.45, 0.30, 0.36, 0.30, 0.44, 0.38, 0.47, 0.54 Mean friction coefficient = 0.41.
Test 6. 304 steel on 304 steel. Room temperature. Sliding speed =
4.3 x 10~ "/sec. Slider and diameter 0.09", distilled water, roughness 27 p" rms.
Friction coefficient values have been divided by 1.05 f values:- 0.39, 0.41, 0.47, 0.48, 0.43, 0.45, 0.48, 0.50, 0.50, 0.47 Mean friction coefficient = 0.46.
Static f:- 0.51, 0.49, 0.52, 0.47, 0.43, mean = 0.50 Test 7. 304 steel on 304 steel.71-79C. Slider end diameter 0.09".
Roughness 27 nicroinches ras. Distilled water lubricant. Sliding speed 3.8"/sec.
Friction coefficient values have been multiplied by 1.42. '
f values:- 0.72, 0.72, 0.55, 0.57, 0.57, 0.58, 0.51, 0.38, 0.43, 0.54 Mean friction coefficient = 0.56 Test 8. 304 steel on 304 steel. 72-76 C. Slider end diameter 0.09".
Roughness 27 microinches rus. Distilled water lubricant. Sliding speed 4.1 x 10- "/sec. Friction coefficient values have been divided by 1.05 f values:- 0.40, 0 36, 0.50, 0.42, 0.44, 0.42, 0.57, 0.45, 0.65, 0.41 Mean friction coefficient = 0.46 Static f:- 0.36, 0.36, 0.45, 0.70, 0.55, mean = 0.48
(.
I
. Test 9. 304 steel on 304 steel. Room temperature. Slider end diameter 0 0.09". Roughness 27 microinches rms. Distilled water lubricant with fine
~
iron oxide particles to a depth of .005". Sliding speed 4.2 x 10 in/sec.
Friction coefficient values have been divided by 1.05.
f values:- 0.44, 0.50, 0.49, 0.50, 0.50, 0.47, 0.53, 0.54, 0.58, 0.51 Hean friction coefficient = 0.51 Test 10. Same as test 9 but iron oxide particles to a depth of 0.25".
Friction coefficient values have been divided by 1.05 f values:- 0.58, 0.55, 0.54, 0.60, 0.51, 0.50, 0.51, 0.54, 0.53, 0.55 Mean friction coefficient = 0.54 Static friction test I. 304 steel on 304 steel. Room temperature.
Slider end diameter 0.09". Roughness 27 microinches rms. Distilled water Initiating sliding speed = 6 x 10 ' in/sec.
lubricant.
Time of stick Friction coefficient values
! l sec .23, .33, .44, .23, .70 .62, .56, .61, .68, .68, .64 10 .39, .39, .42, .32, .50 .61, .54, .61, .68, .66, .67 100 .36, .51, .37, .47, .50 .59, .54, .62, .67, .69, .68 l 1,000 '. 33 .46, .35, .48, .60 .61, .63, .69, .60 10,000 .29 .58, .72 50,000 .59 Static friction test II. Same as static friction test I, but surface roughness is 310 microinches ras. -
Tine of stick Friction coefficient values 1 see .44, .46, .48, .45, .49, .49, .54, .49 10 .52 .46 .49 .48 .51 .53 .52 .60 100 .45 .45 .43 .49 .50 .51 .45 .60 1,000 .52 .51
( .43 .49 10,000 .50 .49 46 .54 l 50,000 .43 t
l , .
( All the adjusted friction. coefficient. values obtained' arei ta' bulated in Table 1, and a histogram and normal distribution which matches the distribution are given in figure 3. The X value between the histogram and the curve (for 9 degrees of freedom) is 10.3 giving a P value of .25, which suggests that the data are consistent with a normal distribution, and thus it is appropriate to use the properties of the normal distribution (e.g. the proportion of the curve lying outside f i 1.690) in analysing the data.
Discussion The question in the GE case is that of arriving at a realistic value for the least st.atic coefficient of friction which might be encountered. Nor-mally, this would be represented by the mean friction coefficient minus three or four standard deviations. As stated in the writer's Boston Edison report, there was a tremendous amount of averaging in the actual application as opposed to the experimental test conditions,.since the total normal load and the total area of contact was of the order of 1000 times greater. Because the low experimental friction valuesall seemed attributable to very small patches on the 304 steel surfaces, a mean friction coefficient minus two standard deviations seemed quite adequate. However, in the Boston Edison situation, both maximum and minimum friction coefficients were of interest, and the (f i 20) criteria gave 95.5% assurance of identifying these values, as against a 4.5% probability of failing to do so. In the GE case, only the lower tail of the normal distribution is of interest, and to impose the same requirement of a 95.5% assurance of having identified the minimum value, the (f - 1.69 c) point must be found. In this case, this gives a minimum fric-tion value of .349.
It should be mentioned that in this analysis all the friction data were
--- - - - ~ ~ ~ ~" * ' ' ~ ~ ~~~ ~~~~
. c=obined into one population, whereas in the Boston Edison study the high and low friction values were considered separately. This was done because the earlier study contained two essentially dissimilar sets of friction coeffi-cients, obtained at two widely different sliding speeds, whereas in the pre-sent study the results have been adjusted to the same sliding speed.
Ibv reliable is the friction coefficient value we have derived? It should be clear that essentially random errors of about .005 or .010 produced by instrumental factors like recorder reading error and zero drif t uncertain-ties produce negligible effect when combined with the statistical fluctuation of more than ten times the magnitude. Systematic errors due to having measured or computed the wrong quantities have all been chosen so as to give a conservative result. For example, an exceptionally large value of sliding velocity v was used in the calculations, since the maximum possible junction size and the minimum possible tLse of application of the shear force was con-sidered. If a value of v smaller by a factor of two had been used, kinetic friction coefficient values larger by about .04 would have resulted, which would in the end have led to overall ('E - 1.69 a) values larger by about .02.
In the case of the static friction coefficient values, if there is a systematic error it is that the friction coefficient was measured af ter short periods of sticking (16 hours1.851852e-4 days <br />0.00444 hours <br />2.645503e-5 weeks <br />6.088e-6 months <br />, or less), whereas in the application the times are likely to be in the months range. Although it is hard to estimate the effect of this, the writer's opinion is that this factor also will increase the over-all ('E - 1.69 c) value by about .01 or .02. As indicated in the Boston Edi-son report, this may by now be a minority view. The majority would argue that no substantial correction because of the difference in the times of stick is warranted.
w . . . - - - . . . . . . . . - . . . . .
_9_
The only possible source of error that remains is that due to contamina-tion. Depending on how the 304 surfaces were made there may be some contami-nants lef t on the surfaces which will give initially low friction coefficient values, perhaps as low as .05 below those given in the tests. (For example, when running the tests for Boston Edison the author carried out some prelimi-nary runs using 304 steel specimensjust as they were found in the laboratory, without any preliminary cleaning, and this gave friction coefficient values up to a value of .05 lower than the values obtained later after the specimens were cleaned by abrasion ) However, this is a situation which improves with time, i.e. at the high temperatures, the contaminants will tend to move to the water-air interface at the top of the fuel storage system.
In any case, this source of error can be greatly reduced if care is taken that the 304 steel pads of the fuel storage racks are thoroughly cleaned, preferably using coarse, clean abrasive paper, before they are introduced into the fuel storagexsystem. (The advantage of using rough surfaces is that the lowest value of friction coefficient is increased even though the mean value is somewhat reduced. Refer to static friction tests 1 and 2. In test 1, using a smooth surface, the mean friction is .53, but the lowest friction is .23. In test 2 using a rough surface, the mean friction is .49 l
and the lowest friction is .43).
i 1
i Conclusions The recommended minimum coefficient of friction to be used for design in the GE application is 0.349.
l Acknowledgements I wish to thank Mr. David R. Sponseller and Mr. Wallace Wheadon of GE, l as well as Dr. John Reed of EDAC, for helpful inputs. Dr. Mukti Das of the Boston Edison Company provided much useful information with regards to the experimental work.
~.
k.
Table I. . The f riction coef ficient values 20 30 40 50 60 70 0 2 1 15 6 4 1 2 10 8 1 2 1 3 4 6 4 3 2 2 6 5 3 4 6 10 4 2
( 5 2 8 6 2 6 5 5 2 1 7 1 8 6 2 8 3 6 8 5 1 9 1 3 11 4 2 l
Note: This table indicated that there were 11 friction coefficient values of
.49 (the intersection of the 9 row and the 40 column), as well as 15 values of .50 (the intersection of the 0 row and the 50 column).
(-
r~~
l References (1) Telex, D. R. Sponseller of GE to Dr. Ernest Rabinowicz, 8/8/77.
(2) E. Rabinowicz, " Friction coefficients of water-lubricated stainless steels for a spent fuel rack assembly," unpublished report to the Boston Edison Company dated Nov. 5, 1976.
(3) Letter, Subj ect "G.E. P.O. 529-CC084X," D. R. Sponseller of G.E. to Dr.
Ernest Rabinowicz, 8/12/77, with attached preliminary drawing, C5442E-103, dated 8/18/77.
(4) E. Rabinowicz, "The intrinsic variables affecting the stick-slip pro-cess ," Proc. Phys. Soc. , 71,, 668-675, 1958.
(5) E. Rabinowicz, " Friction and Wear of Materials," John Wiley & Sons, N. Y. ,1965, pp 151-158.
I r
4 9
i
12
.{' .
STEEL ON STEEL 6~ N Com .
p *co 30tt sitel f .-
, 30tg gggj j fs 37
- 4- ,
ga.g$ %'
= 4rm G
- ~ *.
. ~. .
m
~
2- ~~., ,
~
PERFECTLY LUBRICATED O , , , a e i 0 01 01 1 1 CM/SEC 10 10 0 V
Figure 1. Plot dating to 1958 (reference 4) showing the friction-velocity functier for 1020 steel sliding on 1020 steel at various stages of lubrication.
The data obtained with water-lubricated 304 steel fits the line denoting unlubricated sliding very well.
.: t.
l
- t
p ,
,-p- .m -a = p :- ==-=_. __;.-4==t===t 1. .=- l . - :r--- l = - t 4 =- F = i _:- -. L . _ . e.4==b_=t.-
, . i .: . : c:- =M-.- - -
.--) :p-- t=- l --- --pe t=it p. c l -. t :.:t__: = . p t- -t:=le r .
, l= -
-5@t "=tg_=f = =l ~Ei*i =l- 1 ,f ~WFR ! =- l i@l I -- = F-- t - k =i e -- I 5=l==l A==!=_j50 M-~' - !"I*iEl =d .5tl=ELEl=l =l -MW EP: ".PEEPE
(- ,
=r m=W i _ , _ _ _. r___
l =_:r, _.
I-- t _--f =_ . . . e __-t N tnz.EfE =t--i=i = 3 =t=f=__r,___.
, , , , _ 4.___,__. y __ _
7_
,___,_ ,__.f__ l g_i
_j_. m fy_, .
1 _a ___4--- -I__ I__-d___.t--t i -t
. _ ___ p..-_p_ _
, _{. ___. t , ,
1
!= ~=t: -i.U-1--i --s:f!=-t-t-i---I = t---l---l - :I= t =k= 1 - -l#!!=-t==I:[ -W= ~EE-!--
(. 3
'+i~-E5k=_ T._ F _i=l#sli-i MEME=ii=j=fM=#li i -i-5f-' /= @E ==E
= 1 -
- l - r f-' --+ --
=ur r'm - 1 1 p.: g_2i m ._ =4.=.. PT_ '.1_ =L t _._ = rt4.--- t =.__.5_ p== t=/[/_x-- -t---
~__d
._ _j
_ /_y. ._
. _i ._. 7 i _ ..
L_. __ % _. c--= t. . _ i __ !
i s .
j .___ _ _ _ _ _ .
, 1 i l I !
,. F -
, ,, . m. . .
- l . . .
[~~ ,,. , 0-o i '
3 9
t ..
- -} } .c j ;_ l _- : - 3L .::- : =. b -- : t : -- - i- i-.-' .i: / l u- ---~
1 ._----h-l [-9{ = J _ --- ld p" ,F -:-i ' ;;I) : _ f = p_.< j __4.. " } :_ u) .- ; (: .:- i/ l .i : J.p- ; . -i ==- . =j fC: :E ,h 7 l 4 = t==-1:E-' "T- 5=l=M'=l i3-i ==h:= t:=-l :=21 =- f=h = t = t -~~ =l#=ri =: . -- i -. -t:E
, l M i- E l W -i T ~ - [ -
, . . _ _ _ _ . .p_.,_ 1_L__l: ElGEi E!E lsi?-f .__..__ :l ~ ' P-Wi =W#= =:i-6 :4 7 ....
_.r__-
j [:=. m_: %__7_ .
- . . p _ p__ t . r. _. . ..
__ ,p__._12_:.
___ . =_ __1 - p. :j. _
3, n__.. ._r_ .! =_ . l =. . . p._.. =._ _ _ g, _____ g _.
__w4 _ $_]_--/
i . _ . r' - T ' ! -
__ . 6_ p_
r ~ 12 2 zLr[z ; p z'.o r
_ ,/ri[ri.__: T _!_ [ __
, l ZJ-.x_4.,_p .p '-._ .
4_._ _
I W =f G-_-tZ i =3 # f = f: F W / - l -i "
. U=- l -I -t=Tl " i _ P: T
( 3 YH-i l :l _N 5;P=5tE _ M f-E [ N i! I T- u -
1 4 .:
__ Ji ==F=:-,
~
i -t- =1-- - p /-
t - -
- ;I l-
__ s_.. -
-r_m /_ _r_ _- _: ri r +:.c-i -t ri . I.
l* ..
.l.- ^-
v-, . _
I.
r -. _
.p..!_ ___
1
_t._ _ J-%_.- T,- 4 _. . . . . - p _.
_- . 5
- - ~ " - " -
_' : - ~~ . " ~ ~ :.---
o, _ _ 'Z_._
. _ - - .J.
l
_.__- ._ ._. _ _ . . _ ._D __ .
'.._f~~~-~~~~~__
uy 1 l ,, ...
.. /
/ --
==-p_- p== ==b= -_= k= y =-f =y=p=-
d: $ Yr.
- :p 4 - --
l
, Eg , t. ,: -i= - r . . b = %dm - : F .E () . -_v=-
g3 ,
. . _ p: =-b =-t=~ Y-U-5:"= F/- 5=b =it=t*= :==-t : = :-i :-l --P--- : =-I - = f=Ef=- il ==b=- dM:r i =iP "
W , ==-t---H t-f- :tE iMW=F=! I E 1 ? = " '-t=l~ =- :=W-134;d:-E L = 4 2: =-- iz. -
l H ,
[- =4=-E=g_Fg===p-Kr-- -t='
r- tie!--+ =-t =-L:=={ =-r_=- ===gef=4 :-q-5:-t= r--
m 4 ,
M J_syv, 6 -
- _i___.ja._p__.
_._ _ 4+ __
- w . ;_ e_
,_ _ r-
.._,___4
- 1 C . . _ f_.__ __ H !
_ _ _ r --: +- ,
, ._ ! ; __ . . f; P--- i_ _. !
_ _4._.__ _i ___- _rf t-- . t:x -- s crtrnh_ :,- .
t--
+ W :=-t -I =l'EI-iF==E-t :=h==l==l=il ~= f-Ef E:f ~=T3 - --!=- =
%=Mli=-t : -~= !=== - ::---
, =l=_ _:= L- j = F - h-1 i i =- WEFi=i i=#f_-~1. p -1 = i :-u _.. _ _ _ _ _ j =- _.p _T_- =j:=ip Eg
(- ..;
- - }_ r.= p, =__.;_ ; M u__;_
t= _g =_ _r.__.r t
th :;---- r ._g-_- n_.t. _=c_ ty tr ._u-l i.;,
.- . n.I .r. :[ _ ['.
g; t i a} =x t _l.2__;. gg -[ - hu_:r-7,..-_
, t ~~ f
~-
.- b d h - -h- :$$
_ _g_ _ _ _ _ _
4 ? gg_ _ -
7 3p
_ 7_unip g -_~_. _.
_p-.-g. ..__._
(x
- ~ ~ ~ -~
_.__ t :_- ___. fl.- L_- -
._.-___}__ -
. q _ _ [_ _.q _ _ ___l- _ . f _.2_ .
_ _. h} __ ___- . 9 J_[_ l .__. __
__ _ p . _ g . ._ ____
- . _ } _ _.4 . _. . _ _1_ _. p__4 __ . .___ t . . _. [_ ; _ . _ [ __ t - _. . i __ i . l g ; i .;.. q i. i .i__
. __Figure . . j ._2..g__ Schematic _ q __ } . _. p_ . l ..
_L11ustra,q__-. tion o,f the-fricl
.i i 7 tion-velocity--relationships t ._
.l
- f o,r-s,tainle{ss ,s teel, waterplubricated.[surfacesq_._ l _ . , _ . _j _ . _ . 3 __ . _ q._
i .
_14_
i .
t ,
i g- ,
i w- , , .
y
. M ,
' f i I 6 .
' i *
, . l ;
I i i t , , . : ,
i s I
. r, 7 .
. i :
. j .
o j \ ., ,
A i
/ \ ! -
- i
! : i ,
p \
a -
+ a . . . ,.
l '\ .
r< . F \
J \
I \
4
. I
/ \
+' 1 i / k
~
4 1 \
$ e
/ \
rs g .
,/ i1 3 / \
2 ; .
/ T
- i
- I
/ Tg n I i ,.
/ .
~
! , j !
l / i
\
i ! i i i I/
i A
, , i i - A ,
i
\
i ! / i <
l l I
. L/ ,
l i
\x M l I f f
' f !
j -n i ; j . 31 - i . it t ft .n 1 7R - 3 i
, i .
i .
L J , r, ra j g t r, r i c t . us (Oljictint-
~
Figure 3. Histogram of the 199 friction coef ficient values, also the 1
' I corresponding normal distribution. The mean friction coefficient is
~
.523, while the standard deviation is .103. '-
c t.- .
w =m. --+w e
e ve e-@ hwe
@ -me
,,,.,,wpee,_ ' = & 9e m
m'- s s,
-Al--
Appendix I. The size of iunctions
~
on page 3 of the report, a value of 6.4 x 10 inches is adopted for the size of junctions to be anticipated in the IOSS. In this appendix a more extensive discussion of the reason for adopting this particular value is given.
Methods for determining the diameters of junctions were first developed in the 1950's. A summary is contained in Table 3.1 of reference 5, reproduced below.
Table 3.1 Estimates of Junction Diameter Combi. Junction f
i nation Load Lubricant 3fethod Diameter Reference Copper on 1 kg None /, - distance 7 Rabinowicz, steel 1951 Steel on 1 kg None f, - distance 6g Rabinowicz, copper 1951 Copper on 1 kg Cetane f, - distance 8x Rabinowicz, copper 1951 Copper on 2 kg None Particle size 31g Rabinowicz,
- steel 1953
- Copper on 0.1 kg None fa autocorrelation 10g Rabinowicz,
., copper 1956
, Copper on 0.1 kg None fa f!uctuations Sg Rabinowicz,
- copper 1956 i Steel on 50 kg Contami. f. - t vs.fi - e 10g Rabinowicz, i steel nsted 1958
, Steel on
.) steel Any None 24000y/p 13g Eq. 3.15
,4
, I It will be seen that the values shown range from 0.2 x 10- inches to l' 3
\-
1.2 x 1G inches.
! A little later, the surface energy theory of wear particle formation was developed. According to this theory, during sliding at moderate loads, (1 kg or less), junctions are formed, at some of them adherent wear particles L
are generated with diameters equal to the junction diameter, then these
e -A2-adherent wear particles tend to grow in size, until their diameter reaches a value given by the equation diameter = 60,000 W ab
/P Al At this point the particle leaves the sliding surface.
The current status of the adhesive wear theory in general is discussed in a recent paper (A1).
Unfortunately, this theory has not yet given good values for the sizes of adherent particles and hence of junctions. But empirical data suggests that the adherent particles, and hence the junctions, have about 1/3 the diameter of loose wear particles. Accordingly, an appropriate value for the junction diameter d is given by the equation d = 20,000 W /P A2 ab This equation is given on p 4.11 of reference A2.
Now Wab, the energy of adhesion, may be written in the form W =
Y, + Yb -Y ab A ab 3 where y, and yb are the surface energies of the two sliding metals a and b respectively, and y is the interfacial energv. In this case, a and b are ab both stainless steel, and hence Y ab is equal to zero, and y, as well .ts yb may be taken to have the same surface energy as iron, the main constituent of stainless steel, namely 1700 erg /cm (see Table 2.1 on p 2.3 of reference A2). If p is made equal o 250 kg/c:m , (and this value is based on experi-mental hardness measurements of various samples of 304-grade stainless steel in the Surface Laboratory at MIT) then by substitution in egn. A2 it is .
~ . . . ----- -- - - - - - - - -
-A3-
~
found that d is 3.2 x 10 inches.
Now this value of d) applies primarily to lightly loaded surfaces, typically with apparent pressures of about 100 psi or less, and total loads of but a few pounds. When the pressures are higher, and especially when a large total load is used, then the junction size is subject to change.
According to one school of thought the average junction size increases pro-portionately to the tenda root of the apparent pressure (ref. A3), while according to another view there is no change in average junction size, at any rate until pressures exceeding 10,000 psi are reached (ref. A4). This second view, based on more recent and sophisticated experiments, is more likely to be correct.
In this case, it seems safest to allow for an inerense in junction size by a factor of two over and above the value given by equation A3, which
~
brings it up to a value of 6.4 x 10 inches. All in all, this seems to be a realistic value, and at the same time it is, to the extent that it is in error, likely to be an overestimate. In the present situation that makes it a conservative assumption.
Note that the other possible source of error in the calculation for junction diameter lies :in the fact that the influence of the water has been l
ignored. This might reduce the surface energy of the steel by as much as a factor of two from the value of 1700 erg /cm , thus reducing the junction size by a factor of two. Considering both this factor and the influence of pressure considered above, the possible junction size values may vary by as l
~
much as a function of two about the value of 3.2 x 10 inches computed
, above by the use of egn. A2, as is pointed out on page 2 of the main report.
I I
e I
-A4-References to Annendix I A1) E. Rabinowicz, "The dependence of the adhesive wear coefficient on the surface energy of adhesion," pp 36-40 of " Wear of Materials - 1977" ASME, N.Y., 1977.
A2) E. Rabinowicz, " Study Guide to Friction, Wear and Lubrication, a Self-Study Subj ect," Center for Advanced Engineering Study, M.I.T. , Cambridge, Massachusetts, 1974.
A3) E. Rabinowicz, "A quantitative study of the wear process," Proc. Phys.
Soc. (London), 66B, 929-936,1953.
A4) J.B.P. Williamson, " Topography of solid surfaces," pp 85-113 of " Inter-disciplinary Approach to Friction and Wear," NASA SP-181, Washington, D.C., 1968. ,
9 I
-AS-Appendix II
' Technical Biography of Ernes: Rabinowicz -
Education Cambridge University, England 1944-1947 B.A. in Physics Cambridge University, England 1947-1950 Ph.D. in Physical Chemistry Thesis title "Autoradiographic study of frictional damage" Positions 1950-1954 Research staff member, M.I.T.
1954-1961 Assistant Professor of Mechanical Engineering, H.I.T.
1961-1967 Associate Professor o'f Mechanical Engineering, M.I.T.
1967- Professor of Mechanical Engineering, M.I.T. ,
1961 Summer Consultant, IBM 1969 Spring Visiting Professor, Haifa Technion 1970 Sum =er Consultant, IBM Professional Organizations Member, American Physical Society Me=ber, American Society of Lubrication Engineers Member, American Society of Mechanical Engineers Fellow, Physical Society of London Group
- Subscriber, Institution of Mechanical Engineers, London Registered Professional Engineer, Cocmonwealth of Massachusetts Awards Eodson Award of the American Society of Lubrication Engineers for 1957.
Research Experience His research has been in the fields of friction and wear, mechanical relia-bility, electric contacts, the mechanisms of polishing and comminution, and the use of radioisotopes.
4
==
-A6-Teaching Experience d( Courses taught have been in the fields of Friction and Wear, Applied Mech-anics, Materials, Experimentation, Electroplating, and Materials Processing.
Publications Books
' Friction and Wear of Materials,' Wiley, New York,1965.
'An Introduction to Experimentation,' Addison-Wesley, Reading, Mass., 1970.
(with N. H. Cook) ' Physical Measurement and Analysis,' Addison-Wesley, Reading, Mass., 1963.
(edited) ' Friction - Selected Reprints,' Amer. Inst. Phys. , New York,1964.
~
(with six other authofs) ' Mechanical Behavior of Materials,' ed. F. A.
McClintock and A. S. Argon, Addison-Wesley, Reading, Mass. ,1966.
(with six other authors) 'An Introduction to the Mechanics of Solids,'
ed. 'S. H. Crandall and N. C. Dahl, McGraw Hill, N.Y.,1959.
[%_ Videotape Lecture Series "An Introduction to Experimentation, A Self-Study Subject," Center for Advanced Engineering Study, M.I.T., Cambridge, Mass., 1972. ,
" Friction, Lubrication, and Wear, A Self-Study Subject," Center for Advanced Engineering Study, M.I.T. , Cambridge, Mass.,1974 Other technical writing One patent and sighty articles, includin3
' Wear,' Encyclopaedia Brittanica
' Friction,' Encyclopedia Americana
'Tribological Phenomena,' Encyclopaedia Brittanica
' Friction' and ' Wear,' McGraw-Hill Encyclopedia of Science and Technology Reference Work Listings Who's Who in America American Men and Women in Science #
(