DEPARTMENT OF ENG INE RING RE9 EARCH UNJIIVERSITY OF MICHIGAN'ANN ARBOR Prelirninary Report on DETERMINATION OF CREEP CHARACTERISTICS BY BARR AND BARDGETT METHOD J/by C. L. lark G. L. AVerse J. Jamieson Project Number 491-48 Report Number 2 for The Detroit Edison Company February, 1932

6.J,

Progress Report for February, 1932 on Project Number 491-48 From: C. L. Clark to The Detroit Edison Company G. L. Verse J. Jamieson DETERMNINATION OF CREEP CHARACTERISTICS BY BARR AN4D BARDGETT METHOD Since the phenomenon of creep was first observed in metals at elevated temperatures, various attempts have been made to develop a test which would allow the creep characteristics to be determined in a relatively short period of time. It was first felt that the accurately determined prooortional limit value at any given temperature was a measure of the metal's ability to resist creep at that same given temperature. Results which have since been obtained indicate that such is not the case. It is true that for certain small ranges of temperature, which vary with different metals and alloys, such may be the case. It is our feeling that accurately determined proportional limit values are a true measure of the creep characteristics only at temperature ranges in the immediate neighborhood of the lowest

2 temperature of recrystallization or the equi-cohesive temperature of the given metal or alloy. The next short time test for determining the creep characteristics was advanced in 1928 by Dr. I. H. Hatfield and was designated by him as the "time-yield" test. This test has not received universal support and its use is largely limited to the laboratories of Dr. Hatfield. The procedure of this test has been changed from time to time and according to present day practice the time yield is defined as that stress which produces in a 48-hour period immediately following the first 24-hour period of the test, a total deformation not exceeding 48 millionths (0.000048) of an inch per inch. Dr. Hatfield further states that. the safe working stress will be twothirds of this time yield value. Certain German investigators have -also been active in their uttempt to develop a suitable short ti-me test. In 1927 Pomp and Dahmen developed a test which they felt had considerable merit. They defined the practical limiting cree stress as the stress below which the rate of creep does not exceed 0.001 per cent per hour in the period between the third and sixth hour after the apolication of the load. In 1930 this definition was somewhat modified by Pomp and Enderso

Accordin:, to the latest wcrlk of these investigators the limniting creep stress may be determined by one of the three following methods: Method 1: A rate of extension of 0.005 per cent. per hour between the third and sixth hour. Y:ethod 2: A rate of extension of 0.003 per cent. per hour between the fifth and tenth hour. Method 3: A rate of extension of 0.0015 oer cent. per hour between the twenty-fifth and thirty-fifth hour0,While there is no question that these tests as well as the time-yield tests of Hatfield's yield useful information, none of them are finding very extensive use. Their main limitation is that the time of test is so short that all creep values are determined in the so-called first stage of flow. In other words, for the first few hundred hours of a cree test, flow is continuing at a decreasing rate and the majority of experimental evidence to date does not show a relationshiP to exist between the flow in the first stare and in the second stage, that iPC, the stage in Thich flow proceeds at a constant rate. A new test has rcently been developed in England by w, Barr and E'. E. Bardgett for determining the creep characteristics of metals at elevated temperatures. This test differs from those nor in cowmmon use in that instead of

4. tho' load being kept constant and the rate of creep determined, the load is decreased until the rate of creep becomes very low. "Thile this test is also claimed to give -eliable creeD c'.:aracteristics in a relatively sport period of time it differs from the other short time metthods in thlat it is extended through the first stage and into the second s tage of creep. The work -iven in this re')ort was undertaken to deterwnine if the creep characteristics obtained by this method are comparable to those obtained by our usual method of creep tests. The work is, of course, not complete and:Timany more tests will have to be undertaken before any definite conclusions may be reached. It was felt, however, because of the large amount of interest which is being shown in all the attempts to develop a more rapid means for determininrg of creep characteristics, that a preliminary report would be advisable at this time.

SUMM;ARY OF CONCLUSIO NS The creep characteristics of five steels were determined by both the usual type of creep test and by the newly-developed Barr and Bardgett test in order to determine if these two methods of tests would yield comparable results. The results obtained are summarized in Table 1. Table 1 Creep Characteristics of Various Steels at Indicated Temperatures as Determined by the Usual Creep Test and by the Barr and Bardgett Methods *Results from Usual Temp. Creep Test Barr and Material ~F. Per Cent per 1000 Hours Bardgett Test 0001 0 10 1Q00 Stress Hours Grade A 850 7,000 10,500 15,700 ( 1 ) 20,500 400 9,700 13, 000 17,000(2) 11,250 14,000 17,000(3) KS 1000 3,200 6,800 14,500(1) 13,700 200 OS 1000 1,880 3,550 7, 00(1) 7,200 200 M_- 9 1000 4,200 6, 600 10, 700 ( 1 ) 17, 200 360 D-1 1000 11,250 15,000 20,000(1) 20,000 280 *Values expressed in pounds per square inch. (1) Up-step method of loading employed. (2) Single-step method of loading employed. (3) Down-step method of loading employed~

The results presented indicate the Barr and Bardgett method of testing to yield considerably higher creep values in all cases, espacially if the co-rp Drisn is made on the basis of the stress re ruired in the usual ty-pe of creep test to orolduce creep at the ratse of.01 per cent per 1000 hours. If the stress employed is that necessary to produce creep at the rate of 1.0 per cent per 1000 hours, then the agreements are especially good in the case of Steels KS, OS and J-1 at 1000~F., but an aporeciable d ifference still exists in the case of the other two st eels. Wlhile the agreement between the actul va lues obLained from the two methods of test is not good it should be noted that all of the steels arrange themselves in the same order, insofar as their creep characteristics are concerned, regardless of which method of testing is employed. The B'arr and Bardgett test may b_ verv valuable, therefore, in selectin_ the outstandin cree resistini steels for any i-iven series. The information collected to date is necessarily very limited, and before too general conclusions ray be drawn regarding the agreement between these two tyoes of test, or as to the relative value of the Barr and Bardgett test, much additional information will be required.

'7' 9-i M9I9LT 7? Th7 T7'Z This test was developed in 4ngland by,. Barr and'!. E. Bardgett and is described in British Paltent -oo. 339,890 which is entitled, "An Improved Y.letbod of and Apparatus for Determining the Limiting Creep,tress of'materials." The method consists essentially of applying a known stress to a system consisting of a ca'librated weigh bor which is maintained at room temoerature and of a specimen the creeo characteristics of which are to be determined. This specimen is held at the desired temoerature by means of an electrical resistance furnace. As the heated specimen elongates under the applied stress, the stress on the system will be reduced and the amount of reduction is measured by extensometers attached to the calibrated weigh bar. The test is continued until either the heated specimen no longer elongates, at least within the sens tivity of the extensometers employed, or until it continues to elongate at a very slow rate. It is claimed that the specimen attains very rapidly the stress at which no further cre:p occurs, or at which it occurs at a very slow rate. Also that this stress of no creep, or of very slow creep, corresponds to the limiting creep stress obtained by the usual cree) test.

DESCRIPTION OF APPARATUS One of the Barr and Bardgett apparatus was ordered from the manufacturers last September, but because of the delay necessary before it could be received in this country, we made two of these machines in our own laboratories. The machine has since been received and differs from ours mainly in that it has been designed to test specimens not greater than 0.345 inches in diameter, while those made by us will test specimens of 0.505 inches in diameter. A photograph of our machine is shown in Figure 1 while a cross-sectional view is given in Figure 2. The frame is made of comparatively large angle iron which was welded together to form a strong substantiai structure. As shown in the diagram, there are three connecting links to which the two specimens are attached. These were made 1 1/2 inches in diameter and are of a high-strength heatresisting steel, known commercially as Cyclops 1K2 special die steel. The two test specimens and their adapters when joined together extend through the center of the frame and in order to keep the loading as uniform as possible, the assembly is fastened to the lower prert of the machine by means of a spherical nut which rests on a curved seat. The assefnbly is held at the top by means of a large nut which

9. has ten threads to the inch, and which rests on a ballbearing socket. Surrounding the upper specimen, that is, the one the creep characteristics of which are to be determined, is an electrical resistance furnace, a sketch of which is given in Figure 3o This furnace is so wound that the temperature variation over the two-inch gauge length of the test piece is not greater than 100F. The temnnerature of the furnace is maintained to within 20F. by mteans of a Wilson-M.aeulen temperature control. The control couple of this instrument is placed in direct contact with the winding of the furnace. A drawing of the extensometer which is attached to the weigh bar and which is used to determine the applied stress as well as the change in stress occurring in the system due to the creeping of the speci.men under test is riven in Figure 4, while a detailed sketch of this instrument is shown in Figure 5. The principle of this instrument is very simple and can be readily seen from the sketch. A reading of one degree on the scale is equivalent to a stress of 24.2 pounds per square inch. In order that the temperature of the weigh bar, to which the extensometer is attached, may be kept as constant as possible, a water jacket is placed on the specimen adapter

AS-tEST04 A MU CO/ PLC S~~~I~ L-0-Cf- ULL 3

C, — c// /J 5CQ/' C',!/T/6/// (h< S_ tR k; It// Niur-c ~S~ -el / Conn Cjioun /l'rks #o S di L/ rL teC ensone' er - 5% TLYSO/e/~ TER F~a. 4.

EDcE -- Po i r4 -T F- F*KR LrVERR ~ n'~H T~r Fut-cC A urq 14 To bmr-r~ ~ ~ ~ ~ ~~' S(f,,~~i~ ~r y LONG GrTUDD IOI qL 4EC Ti Orl

1I immediately above the weigh bar. The entire lower part of the app ratus is also enclosed by transite board to keep the temperature as constant as possible. TEST P!TOCD 17DRE After the apparatus has been properly assemnbled and the specimen to be tested has been maintained at the correct temperature for at least 24 hours, a zero reading of the extensometers is taken. By means of levers attached to the upper large nut, the proper stress is applied by turning this nut until the extensometers on the weigh bar give the correct reading for the stress desired. Readings are then taken on the extensometer at hourly intervals for the first day, and twice daily thereafter, and the reduction of stress on the system is thus determined. Stress is then plotted against time and the test is continued until the stress-time curve appears to be approaching a definite asymptote; or until creep has stopped completely, at least witihin the sensitivity of the measuring a'paratus employed.

ll. 2?SULTS Because of the length of ti:ne required to construct these apparatuses and to get them in proper working order, a large number of tests have not as yet been run. Certain have been completed, however, and these are listed in the following table. Table 2 Steels Subjected to-Barr and Bardgett Tests Temperature of Test Steel Designation Tpe g. ahr Grade A 0.18 C. 8 50 KS Killed 0.10 C. 1000 OS Open 0.10 C. 1000 mA-9 Mn. -Mo. 1000 D-l Cr.-W.-Si. 1000 The results obtained from these tests are given in the following figures and tables. Grade A Steel at 8500F. The results obtained from this test are given in Figure 6. The test was continued for a period of 400 hours and during this time the stress was reduced from an orig=lnal value of 24, 750 pounds to one of 20,500 poundso Also, from the appearance of this curve during the last 100 hours of the test, it is evident that the stress would not have been appreciably reduced even though the test

12. had been continued for a much longer period of ti'.e. Our regular type of creeo test was also conducted on this same steel at 85OcF. and in Table 3 the results'tained are como.red with those from the Barr and rP rdrott test. Table 3 Creep Characteristics of Grade A Steel at 8500F. as Obtained from the Us al Creep Test and from the Barr and Parde t t Test *Results from Usual Stress from Barr Temp Creep Test and 3ardgett Test Material F.per 100 Hours at 4 —0 I'curE 0.01 0.10 1.00 Grade A 850 7,000 10,500 15,700(1) 20,500 9,700 13,000 17,000(2) 11,250 14,000 17,000(3) *Values expressed in pounds per square inch. (1) Up-step method of loading. (2) Single-step method of loading. (3) Down-step method of loading. From the above values it is evident that the result obtained by the Barr and Bardgett method is considerahly greater than those obtained by our usual creep testing method. Killed Carbon Steel at 1000~Fe The results obtained from the Barr and Bardgett test on killed carbon steel, rteel KS, at 1000~F. are shown in Figure 7. The initial stress was

i I, i I: j 1 ~.........t..............'-'"t......... -','...........-.; —..........~..................' i i t ~' i' i i................'~. — — "7.................... d......,,:, ~, ~ ~ t' Si'ECLjN. 5. T-7'~ ~ ~oo~ i 2ot'}' —..................................................-,-.......t _ ~.,....,.....t............................_.q.__. _ IT-........,.................., —', —....!...........~ -..............' -: — T' —'?' —i —— ~ ~...... — ~!, I -.l! i i ~ ~ i',' i: }:! i',, I.................. t..........J -. - [.....,,: ~ I I I' { i;:......:': "' ~'!''~':......[ -'-t.............. —' — p... 7 ~:,' s, ] -".............:...... t } ~' e,; I I: I ~ I tO, ~ i i:: i i ~ ~!' ~ t'' t, --. -,,,. 4.. _,.......:..............., j i { ~ ~,. i i i I' i: ~:. i,.. 5. ~'!'.............; - "-~- ~..............J.....:''' ~'' } ~-i, i,' ~.................. i................................... i i i. I' i i i: I -,, I 7 --......'...... I' k;' i i. I,.::"' ~':' t ~ i' i i I ~. t.,!..... i I i., ",', t i, I' l'' i i, } i t: ~ "' {, i i i i -, { t I, to,:, ~,, ~' ~' j i I _ i i-' 0!....1, i,;... i ~..... ~. _ 20.., c 4,0 80,, oo,',Y.,O /~.'oo.' ~,3,Z'oO T, ME — H ours. —

12,400 ounds per square inch and at th nd of the test, 22,400 pounds per square inch and at the end of the test) which was 200 hours, this stress had been reduced to 15, 700 pounds. When the test was stopDed the specimen was still continuing to elongete, that l:, the stress was continuing to decrease at a relatively rapid rate. Our usual type of creep test was also conducted on this steel at 1000F. and a comparison of the results obtained by the Iwo methods is shown in Table 4. Table 4 Creep Characteristics of Killed Carbon Steel at 1DOO0F. as Obtained from the Usual Creep Test and from the 3arr and T3ardgett Test *Results from Usual Creep Stress From Barr Temp. Test and Bardgett Test Material ~F. Per Cent per 1000 Hours at 200 M'ours 0.01 0.10 1.00 KS 1000 3,200 6,800 14,500(1) 13,700 *Values expressed in nounds per square inch. (1) Up-step method of loading. As in the case of Grade A steel at 850OF., the result obtained from the Barr and }Bardgett test is considerably above those obtained from the usual creep test, with the exception of the value expressing a flow rate of 1.0 per cent per 1000 hourso

Open tte e1 o t 1OO'". Te re sults obt',ined from the Barr and Ta rdCe tt t est o n or steel., that i s'teel cO, at 1000Ov., nre iven in 1"'ixure P.:)urin&' the period of 210 hours this test was conducted the stress was reduced from ani initial value of 21, 7.0 oounds per square inch to one of 7,200 pounds. Also, frxn its a'pearance the curve a.vears to be an ro f chinr.- s, stress of about 7,000 oounds as its a sy.nptote. A comparison of the vw.1u o:tl in~d fro',. thi-s, test and those from the usual tyTpe of creep test i.:i vei i-n'a b 1 e 5G Xlble f-_ Creep Characteristics oP Ozn Carbon rteel at l0 1', as Obtoined froTr the UTsuI.. Creep Test and from the Barr and Bardgett Test *Results from Usual Creep Stress frol, >" rr'femp Test and ir3dget T e st: aterial 0F. Per Cent atr 100:iours at 200:u'rs 0.)1 C.10 1.00 OS 1000 1,800 3,S550 7,000(1) 7,.00 Va lues exOressed in pounds per square inch. (1) U-)-step method of loading. It is evident from the above vtwlues t l;:t agai. th^e 3arr and 3ardgett test yields a higher value than thVose oltained from the usual type of creep test. The value for 1.0 per cent creep in 1000 hours, however, is in very rood atreoment with the limiting creep stress obtained by the B3arr a nd Bardgett test.

It is interesting to note that in the case of the open and killed carbon steels, the two steels arrange themselves insofar as their creep characteristics are concerned, in the same order by the Barr and Bardgett test as with the usual type of creep test. Steel WM-9 at 10000F. The results of the Barr and Bardgett tests on Steel NM-9 which is a Mn.-.t'o. steel at 1000~F. are given in Figure 9. The initial stress was 23, 500 pounds per square inch, and at the end of 340 hours, this had been reduced to a value of 17,200 pounds. These results are compared to those obtained from the usual type of creep test in Table 6. Table 6 Creep Characteristics of Steel MM-9 at 10000F. as Obtained from the Usual Type of Creep Test from the Barr and 3Bardgett Test *Results from Usual Creep Stress from Barr Temp. Test and Bardgett Test Material F. Per Cent per 1000 FHours at 360'ours 0.01 0.10 1.00 MM-9 1000 4,200 6,600 10,700 17,200 *Values expressed in pounds per square inch. As in all the previous cases, the value obtained from the Barr and Bardgett test is considerably above those obtained from the usual type of creep test. Again, however,

16. this steel and the open and killed carbon steel arrange themselves in the same relative order by the two methods of testing, Steel D-1 at 1000 F. The results of the sBarr and Bardgett tests on Steel D-l, which is a chromium-tungstensilicon steel, at 1000~F are given in F'igure 10. An original stress of about 26,000 pounds was applied and at the end of 280 hours this had been reduced to 20,000 pounds. The sharp initial decrease in stress was not nearly as marked with this steel as with those considered previously. These results are compared to those obtained from the usual type of creep test in Table 7. Table 7 Creep Characteristics of Steel D-1 at 10000F. as Obtained from the Usual Type of Creep Test from the Barr and Bardgett Test *Results from Usual Creep Stress From Barr Temp. Test aend Bardgett Test..a.te rial F. Per Cent per 1000 Hours at 280 Hours 0.01 0o10 1.00 D-1 1000 11,250 15,000' 20,000 20,000 *Values expressed in pounds per square inch. Again the Barr and Bardgett test result is considerably above those of the usual creep testing metho d with the exception of the stress which will produce creep at the rate of 1.00 per cent per 1000 hours.

17. The comparative creep strenth of this stee:l, horever, is superior to the reomininF steels considered regardless of which type of testing is eriployed. A constant difference, however, is not obtained with the two nethods of test. CO NTCLUS ITSOi Creep tests were conducted on Grade A steel at 85 0 F., and in four steels at 10006F. by both the usual tmethod for conductint creep characteristics a-nd by the nerwly-developed Earr and 3ardgett test. The four steels tested at 1000'F. were an open cerbon steel, a killed carbon steel, a manganese-molybdenutn steel, and a chromium-tungsten-silicon ste,.l. The reslults obtained allow the followini, conclusions. The results obtained indicate the Barr and')t rdrett -method to yield considerably higher creep v:-1ules in l cc sec, especially when they are comp'lred to the stresses rer:uired to produce creep at the rate of 0.l.,er cenrt )er 1000 hours (1.0 per cent per 100,000 hours). If the valu.e ch-,osacn ror comparison be the stress required for:a rite of creei of 1.0 per cent p-7.r 1-000 hours, then the agrreement is Imuch better. In fact, with the killed steel, the op:en steel, ArA the chromium-tungsten-silicon steel at 1000CF., t7e t:,reement between

the two tests is almost perfect. For the remaining two steels, the discrepancy in the results, even when this larger creep rate is used for comparison, is rather large. For all the steels, except Steel D-l, the ratio between the stress required for 0.01 pDr cent creep per 1000 hours and the liniting creep value for the Barr andl Bardgett test appears to be from 3:1 to 4:1. With Steel D-1, the ratio is somewhat'less then 2:1. Before any general statement may be made regarding this ratio, a considerably larger number of tests will have to be conducted. Even though the agreement between the values obtained from the two types of tests is not good, it is interesting to note that the steels arrange themselves in the same relative order by the two tects. That is, Steel D-1 is found to be the outstanding steel at 10000F. by both methods while the open carbon steel, Steel OS, is found to be the least resistant to creep~ On the basis of the results obtained to date, therefore, it would ap ear that the chief value of -he Barr and liardett test would be to pick out from any Liven rTrouD the outstanding steels insofar as their creep characteristics are concerned. The regular creep test should then be conducted on these outstanding steels to deteTr.ine their actual cree) chara cteristi cs.

UNIVERSITY OF MICHIGAN 3 1190111115 02827 2774111111 3 9015 02827 2774