ENGINEERING ESEARCH INSTITUTE UNIVERSITY OF MICHIGAN ANN ARBOR BIMONTHLY PROGRESS REPORT NO. V THEEMAL-SHOCK INVESTIGATION By P. F. CI3ENEA A. R. BOBROWSKY L. L. THOMAS Project M949 WRIGHT AIR DEVEOIPMENT CENTER, U S. AIR FORCE CONTRACT AF 33(038) —21254; EC.O NO. 605-227 SR-3a April, 1952

F}P - 4.t - I -- '.,J. p;,, 1 Ir I 9 -;~C, r,,~,, j 5_1 2 '6. ar~

I. -- ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN BIMONTHLY PROGRESS REPORT NO. V TERMAL-SHOCK INVEST GATION OBJECT The object of this research is to evaluate optimum design of test specimens and criteria which will permit correlation of thermal-shock data with performance of the material in the form of turbine buckets. SUMMARY Tests were run to: a) observe effect of specimen temperature on i) number of thermal-shock cycles to failure, and ii) scatter of experimental results, b) observe effect of prior application of alternating stress on number of thermal-shock cycles to failure, and c) check variations in setup variables. A tentative procedure has been devised for possible correlation of thermal-shock results on different specimens. Visits were made to four groups of persons interested in thermalshock tests. i

.- ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN INTRODUCTION The previous progress report indicated that reproducibility of test results might be a function of the severity of the thermal-shock cycle, with best reproducibility at greatest severity. This period of research was consequently spent primarily in: a) altering the experimental rig so that lower air temperatures may be used in future tests, b) determining the change in severity occasioned by changing specimen temperature, and c) checking out the altered setup. The period of research covered is from December 11, 1951, to February 11, 1952. APPARATUS The previous apparatus was altered by rotating the air-storage.tanks to a horizontal position in order to: a) increase mixing of incoming air and air in the storage tank, b) reduce pipe friction, and c) permit installation of suitable cooling coils. A two-dimensional nozzle with replaceable throat was constructed and adjusted in order to determine how best to obtain reproducibility in construction of duplicate nozzles. A sound-minimizing shelter was constructed around the test rig to reduce noise from the air jet. The second thermal-shock installation was completed. A by-pass runs between the two air tanks to increase plenum capacity. 2

ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN - A total-radiation pyrometer:has been procured and will be placed in use shortly, VISITS Visits were made to the General Electric Company at West Lynn, Massachusetts; the Pratt and Whitney Aircraft Division at East Hartford, C-onnecticut; the General Electric Company at Lockland, Ohio; and the Rational Advisory Committee for Aeronautics at Cleveland, Ohio, for the purpose of exchanging thoughts on thermal-shock research. The results of these visitsi that were of most immediate interest were: (1) At the West Lynn plant, some results of tests showed good correlation between impact strength and thermal-shock resistance. (2) At the East Hartford plant, failures in turbine buckets were seen that were apparently caused by mechanical instability of the bucket, due, at least in part, to unequal temperature distribution. (3) At the Lockland plant, a turbine bucket was seen that showed small regularly-spaced cracks on the leading edge. his bucket appears to present the most clear-cut evidence yet seen of the presence of thermal cracks in turbine buckets. (4) At the N.A.C.A. laboratory, references were available to German works that indicated a correlation between impact strength and time of loading under static stress at elevated temperature (reference 1). The general line of thought on this project appears to be in reasonable agreement with opinions given at the places visited. ANALYSIS General Problem Three- problems that appear to be among the most difficult to attack by theoretical means have been under investigation for some tiine on this project. They are: 3

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN - (1) What parameters based on material properties govern thermal cracking? How are these parameters (and properties) dependent on temperature? What are the local stresses and. temperatures in the specimen? (2) What is the variation in resistance to thermal-cracking due to differences in specimen shape and size? What is the variation due to testing in different rigs, by different observers? (3) How can test conditions in the actual gas turbine be described in a simple manner? What are the conditions of stress and temperature in a turbine?: Can these conditions be correlated with the "artificial" conditions in a thermal-shock test? Concept of Deterioration or Damae With-the dawning of the realization that fairly- exact answers to the above questions would be difficult or impractical to obtain by direct theoretical means, a new line of reasoning was set up in an attempt to obtain answers by side-stepping the major- difficulties without ignoring their existence. What is hoped is a start on a practical answer to these questions is tentatively set forth in admittedly oversimplified fashion below. It can be shown that no reasonable conditions of temperature distribution and restraint of specimen can cause cracking in the usual materials employed for gas-turbine blading because the strains set up by the temperature gradients cannot exceed the fracture strains (or strains at which necking begins) of the materials as determined by tensile tests. Exceptions to this statement can be found for the cases where: a) the tensile load is applied so rapidly that the strain at fracture is much less than for conventional static tests, b) a very high stress-concentration exists, as at the ends of cracks, and c) the specimen contains highly unequal crdss sections. These exceptions can be set aside for the following reasons: a) Qualitative observations of rates of cooling of thermalshock specimens in this project show that the rate of c.iolg appears to be less than that required to diminish the usual value of strain at fracture materially 4

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN b) The quality of turbine buckets is stich that no cracks (at least of any great depthi) are present in the new turbine bucket. c) The pe of t urbine -buckets is such that age hanges of cross section ae present in the usual zone of failure. The question was raised at all visits as to whether anyone had ever observed a thermal '(or qluezhing) crack in a-material of reasable ductility (say I0% elongation in 2 inches fr a test bar of 1/2-inch diameter). Mate;riaqls whicb transform during the shoek (or quench) to types definitely kno to possess very low ductility were excepted, as were specimens f very rapidly changing cross section. The replies indicated that no such cracking had been positively observed. The almost inescapable conclusion is that turbine-bucket materials crack during thermal-shock tests because the stress and/or temperature his — tory has altered the original. material so, that it no longer possesses the same ductility at fracture, and probably has altered o~her structure-sensitive properties also. Evaluatn of PDa It seems reasonable to assume that certain material parameters govern thermal crack..g and that these parameters are dependent on temperature. If these parameters, hwever, are dependent on the past history of stress and temperature experienced by the specimen,. it becomes necessary to.-evaluate the effect of this histoy on the properties of the materials under study. It is not necessary to speculate on the local mechanism by, which the' deterioration has ioccurred' if one is interested only in a measure of the _deterioration. A first thought on possible evaluation - such prperties as tensile strength and ductility might be that specimens could be run in the shock r-ig for a while and then tested in tension or in impact. The decrease. in tensile strength, duct ility, or impact strength would be a measure of the-d e done to the specimen drting the thermal-shock history. The use of the impact test to evaluate deterioration is not new, reference 1 shows the effe-ct of stress-rupture loading on impact strength; also, man investigators have worked with d~amage"line testing in fatigue. But the question arises as to how to improve the sens itivity of any process for evaluation of damage or deterioration. 5 I

-- ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN There is no reason to expect that the change in properties is the same throughout the entire specimen,. Thus, an attempt to detect such changes as altered ductility might well be predestined to failure if only a small volume of material had become brittle. A more effective test would be one that is more sensitive two the changes in small volumes of material. If the deterioration (as seems reasonable) is greatest on the outer surface of the specimen, a suitable test for deterioration should be sulch as to emphasize the properties of the surface material. Tests such as rotating-beam fatigue and thermal-shock itself might be suitable tests, since each emphasizes properties at the outer surface. An alternative procedure would be to cut smaller specimens for tensile testing from the original thermal-shock specimen. It is desirable to avoid additional specimen preparation, so the idea of cutting smaller specimens will be discarded, at least for the present. From another point of view, it can be assumed that thermal-shock failure is primarily a function of two simple failure mechanisms, namely, stress-rupture loading and fatigue loading. That is, a given thermal-shock history may be envisioned to proceed from a combination of fatigue cycles (equal in number to the number of thermal-shock cycles) and a simultaneous application of stress-rupture loading. A somewhat naive statement of the criterion of failure could be that for given thermal stress and temperature the amplitude of fatigue stress at other than a zero mean stress controls failure. But the very conditions of stress and temperature are functions of properties such as thermal conductivity, specimen size., speed.of air blast, and so forth. Thus,' this alternative point of view is too complicated to provide the complete answer concerning a thermal-.hock parameter, but it serves the purpose of indicating that rotating-beam fatigue and stress-rupture are suitable for causing deterioration as well as measuring prior deteriorat ion. For example, the deterioration due to thermal shock can be set equivalent to a certain amount of deterioration due to fatigue, the equivalence meaning that the number of thermal shock cycles to failure is the same for previous thermal-shock deterioration as for previous fatigue deterioration. Thus, fatigue (for example) can be used to damage a material, after which it is placed in a thermal-shock test to determine the change in nmber of thermal-shock cycles to failure; or thermal shock can be used to damage a specimen (short of cracking), after which the change in the number of fatigue cycles to failure may be determined. The damage done to the specimen in thermal-sholck test prior to cracking is a consequence of its individual history of stress and temperatureo The importance of evaluation of this history lies partly in the fact that the properties of the material itself —thermal expansivity, stressstrain relations, cold working properties, and so forth —serve to determine 6

L. -- ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN - this history. These properties vary from material to material and consequently the difference in history between 2 specimens in a thermal-shock rig may not be the same as the difference in the respective histories in the turbine. It is necessary then, to evaluate this artificial history, This line of reasoning leads to the following tentatively proposed technique for describing the history placed on a specimen during the thermalshock test. The specimen is -subjected to a given number of fatigue cycles at a given maximum stress (or a given stress-rupture loading for a given time, or some combination of the two), It is then evaluated in thermal shock. This procedure is repeated either with different stresses, different numbers of fatigue cycles, or different times of stress-rupture loading, The drop in number of thermal-shock cycles until failure would be set equivalent to fatigue deterioration for the number of cycles at the stress given (or other damage-producing effect ). No. of thermal- Fig. 1. shock Hypothetical graph of cycles to measurements of thermalfailure. shock damage. No. of fatigue cycles at constant stress. or Stress at constant number of fatigue cycles. or Time at load at given temperature. or Load for given time at given temperature. The stress-rupture type of deterioration has been evaluated by impact tests in reference i as already noted, It should be mentioned that the reference showed some materials were more resistant to rapid deterioration than others, and that some deterioration could be removed by subsequent heating. These factors may alter the shape:of the above curve for different materials. Application to General Problem (1) What parameters based on material properties govern thermal cracking? How are these parameters (and properties) dependent on temperaturez? What are the local stresses and temperatures in the specimen?

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN The analysis just given may be used to determine a procedure, not for answering the above questions, bit for obtaining the ultimate results towards which these questions were leading. Let the steady-state temperature and stress-rptue Loading the failure zone be determined for the particular turbine under consideration (experimentally, Or,by calculation based on experience). These values will vary for different turbine-bucket materials and conditions "bf operation. Determine deterioration at that stress-rupture condition for various times of loading by using thermal-shock as a criterion of damage. A scatter band| as in Pige. 1 may be expected to result. A proposed parameter might thenbe: number of themal-shock cycles to failure after a known ior history simulating turbine history hs been imposed on the speci men (2) What is the variation in resistance to thermal cracking. due to differences in specimen shape and size and differences in test rigs? A suitable method of evaluating the history given by the thermal-shbocktest may be to determine the equivalence of histoy in the different specimens in terms of a deterninable quantity such as fatigue damage..Each type of, specimen or rig would be used in thermal shock after several pairs of idenitical fatigue histories had been placed in the respective spec:imens. The decrease in the number of thermal-shock cycles for given identical histories would measure the relative histories of the two specimens. It is known that differences in fatigue life exist due to- variation of specimen size and shape,- but the differences are small if the' specimens are not too small in depth of cross section (say, below 1/4 inch in depth) and if stress concentration factors are minor. Stress-rupture. loading could be used similarly. (3) How can conditions in the gas turbine be taken into account? The problem here is now identical with the problem in (1), and 1a been -answered. C oncusionS It is hoped that the concept of deterioration may be used in. evaluating the: (1) equivalence of the history placed on the shocke:speciimen by thermal shock and an easily reproducible history such as fatigue and/or stress-rupture history, i 8

ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN (2) equivalence of specimens of different shapes and sizes, tested in different rigs, and (3) equivalence of thermal-shock test conditions and turbine conditions. BXPEEIMENt RESEWMJTS The check tests on the altered thermal-shock rig indicate that thermal-shock failures are occurring at approximately the same numbers of cycles as previously, or somewhat fewer. S-816 alloy specimens were' tested at maximum cycle temperatures of 1700, 1600,, and 1500~F. There is a fairly significant trend to indicate that decreasing the specimen temperature decreased the severity of siock or the susceptibility of the specimen to failure, since the number of cycles required for failure was increased. It is too early to discuss reproducibility at the different temperatures, but very tentatively, the scatter of results increased as specimen temperature was decreased. It is not obvious that variation of cycles to failure should necessarily have been as shown. The type 347 stainless-steel specimens that were tested after having been:-subjected to fatigue loading showed, n-the average, reduced numbers of thermal-shock cycles to. failure. It is too early to hypothesize on the shape of a curve relating fatigue cycles to thermal-shock cycles for failure. 1. Thermal c'racking required more thermal: shock cycles for S-816 alloy as temperature was lowered fro 1700 to 1500~F. 2. Prior fatigue damage short of failure appeared to reduce the number of thermal-shock cycles to produce cracking in type 347 stainless steel. 3. A concept of damage has been-deveped far possible ceorrela- | tion of results from different specimens and teest-rigs Te. equivalence of results may be expressed in terms of known prior daage. iThis correlation may be suitable for comparing results in turbines and in the shock rig, 9

--- ENGINEERING RESEARCH INSTITUTE * UNIVERSITY OF MICHIGAN PEEIPENCE, Versprdun ung Schadigung warmfesten Stable bei Dauerstandbeanepruchung by A. Thum and K. Richard. Arch fur das Eisenhuttenween. 15, No. 1, 33-45 (July, 1941)O

-- ENGINEERING RESEARCH INSTITUTE ~ UNIVERSITY OF MICHIGAN KEY TO LOG Column (2) Arrow indicates direction and location of cooling jet; cooling medium is air unless otherwise stated W Cooling medium is water.045 Width of cooled edge, inch P.F. Previously subjected to rotating-beam fatigue as shown in column (6) X "Failed during pre-fatigue Column (3) M Thermal-shock cycle manually controlled 1500/5 Automatic cycle control; maximum temperature F,0 and lengt;h of cooling period, seconds P1800 Dead load, 1800 lbs +10/100 Starting with stated maximum temperature, maximum'temperature -was: increased 10~F after each 100 cycles. 40.5K Reversed-bending (rotating-beam) fatigue tests; maximum stress, 4o0 500 psi to i80o Maximum temperatu're held constant after 1800F-was reached Column (4) A Air cooling for stated number of cycles W Water cooling for stated number of cycles no symbol Air cooling- for stated number of cycles Column. (5) 0 No failure visible 'F Fracture C Cracks G Grooves Column (6) -B -Specime warped due to thermal strains AO14 Area of cross section, square inch T300/1600 Heat treated'before testing 300 hr at 1600~F G1500 Grooves first appeared at 1500 cycles OH Stated maximum temperature was exceeded due to malfunction of control unit BT Br oke throug b to thermocouple hole 409.5K Previously subjected to 82,000 cycles at 40,500 psi 82000oo R! Reproducibility test N Specimen formed a- neck +100/5108 Temperature increased 100~F at 5108 cycles Check II Second test run to check operation of test rig after alteration 11

TEST LOG Specimen Cross Number Type of Number Section Cycle of Cycles Failure Remarks (1) (2) (3(4) (5) (6) Type.304 Stainless Steel i; '1~M 0 B 04-5i 2 1600/10 4400 A C B 500 W ' 1600/4 1783 C '~//~~~~~~~18 +10/100 4a Fatigue 40.5K 35oo00 F 4b Specimens 40.5K 2600 F, 17o00/4 1100 0 1800/4 675 C 6 1600/4 6240 0 G6500 9lgoo/4 124o0 c 7 1500/5 4130 F A 0.16 P60oo 8 1600/5 5082 0 T300/1600 1800/4 517 C.1' 12

LEST LOG (cont) Specimen Cross Number Type of Number Section Cycle of Cycles Failure Remarks (1) (2) (3) (4) (5) (6) 9 1500/5 5753 o 1600oo/14 1000 0 10 1700/4 100ooo o _Vk 1800/4 80 c:11 " o1500/5 1000 F A 0.132 P18oo/0 1500/ 12 P6oo00 5000 0 A 0.135 P900 12o00 0 P1800 205 F 13 16oo00/4 1284 c G 1115 14 1500oo/4 1000 F OH 15 1600/5 1900 C T300/1600 16 1600/5 409 C 13

TEST LOG (Cont) Spec imen Cross Number Type of Number Sect ion Cycle of Cycles Failure Remarks (1) (2) (3) (4) (5) (6) 17 1500/5 300 F A 0.14o pl8oo, 18 18oo/4 1950 C G 1500 -9./1700/3 550 W Q 20 1500/3 1000 0 BT Type 347 Stainless Steel 600oo/4 866 C ~1 +10/.~~+10[0 ~~~2 ~ 1600/4 1147 C +10 /100 -71.020 1-500/4 575 C BT ~10/100 4a Fatigue 54K 5200 F 4b Speciiens 54K 10400 F 40.5K 82000

TEST LOG (cont) Specimen Cross Number Type of Number Section Cycle of Cycles Failure Remarks (i) (2) (3) (4) (5) (6) 5 7 1500/4 1326 c 6 1500/4 1990 C +10/100 ____> 1600/35 7 +10/100 2700 G to 180 8 (Defective) 9 F71600/4 2863 C B.03s10 1600/4 3787 C Check II, —~" ~ 02o 11 1600/4 2580 C - ~' 050 12 1600/4 3162 C G 736 15 1600/4 2204 C G 2072 - 020 15

TEST LOG (cont) Specimen Cross Number Type of Number Sect ion Cycle of Cycles Failure Remarks (1) (2) (5) (4) (5) (6) 14 1600/4 2707 C G 2604 15 1600//4 3003 C G 2820 V R ~o3& 16 1600/4 2518 C R,~ 02_0 17 1600/4 4850 0 Check I ~04S ~18~~ ~ Fat igue F 54K 64K 7200 103300 19 1600/4 1825 C ~ ]-9 zdoo/~ z~~~~~~~~ep c R~~o 37K/2 17100 42K/1l0o00 20 Fatigue 48K/35600 64K 4300 F 54K/l0oo 000o 59K/10400 21 \ 1600 o/4 4430 C

TEST LOG (cont) Specimen Cross Number Type of Number Section Cycle of Cycles Failure Remarks (1) (2) (3) (4) (5) (6) 22 (Defective) 23 1600/5 2962 C 24 \ Fatigue V.oo ~59K 52900 F 25 \7, 1600/5 1562 C 54K/5000 /oO P.F. 53K/52000 59K/12000 26 1600/5 1960 C 64K/0oo0 70K/1000 75K/500O 53K/52000 27 x F 59K/11500 P.F. 53K/52000 59K/12000 28 771600/5 1594 C 64K/loo V~ O/b P.F. 70K/1000 75K/500 53K/52000 59K/12000 29 x C 64/1000ooo V.o0, P.F. 70K/1000 75K/300 1:7

TEST LOG (cont).Specimen Cross Number Type of Number Sect ion Cycle of Cycles Failure Remarks (1) (2) (3) (4) (5) (6) 30 1600/5 1973 C 0/0 31 1600/5 2764 C H.S. 21 (Vitalliux) ECast] 1 15o/00/3 1000 C BT + 10/100 Inconel +10/100 1 _ ~ 1500/3 1750 CG15 2 2i/ 0ioo 2730 C.-cj 030 +!0/100'' ~~~~]3 1570/3.5 428 c _.4< ~ +10/100 4 18

TEST LOG (cont) Specimen Cross Number Type of Number Section Cycle of Cycles Failure Remarks (1) (2) () (4) (5) (6) 6 16oo/ 7C49 7 8 9 10 11 12 19

TEST LOG (cont) Specimen Cross Number Type of Number Section Cycle of Cycles Failure Remarks (1) (2) (3) (4) (5) (6) 13 S-816 Alloy (Wrought) A 0.08 1500/4 1788 A oo8 1788 o P700 18391 C +100/5108 No load +100/10000 A 0.08 1500/4 2657 F N 2 P1100 to P700 ~j3 \7 1700/4 2256 C ~4 \V/ ~1700/4 2550 C 5 1600/4 3870 C 6 1500/4 2630 C 20

TEST LOG (cont) Spec imen Cross Number Type of Number Section Cycle of Cycles Failure Remarks (1) (2) (3) (4) (5) (6) 7 X.15 _00/4 132o80 8 \/ 600oo/4 7497 C......__~ 21

UNIVERSITY OF MICHIGAN 3 9015 02514 8324