LIBRARY ~ENTRAL RESEARCH | ~ " DOCUMENT COLLECTION S 76 OAK RIDGE NATIONAL LABORATORY operated by UNION CARBIDE CORPORATION for the U.5. ATOMIC ENERGY COMMISSION ERGY RESEARCH \LIBRARIES (VEABA 3 445k 0513 "fa ORNL- TM- 2727 THE MECHANICAL BEHAVIOR OF ARTIFICIAL GRAPHITES - £3 AS PORTRAYED BY UNIAXIAL TESTS W. L. Greenstreet J. E. Smith G. T. Yahr R. S. Valachovic “ NOTICE This document contains information of a preliminary nature and was prepared primarily for internal use at the Ook Ridge National Laboratory. It is subject to revision or correction and therefore does not represent a final report. LEGAL NOTICE This report was prepared as an account of Government spfinsored work. Neither the United States, nor the Commission, nor any person octing on behalf of the Commission:’ A, Maokes any warranty or representotion, expressed or implied, with respect to the accuracy, completeness, or usefulness of the informotion contained in this report, or that the use of any information, up.poruru's, method, or process disclosed in this report mey not infringe privately owned rights; or . . B. Assumes any liabilities with respect to the use of, or for damoges resulting from the use of ony information, apparatus, method, or process disclosed in this report. As used in the above, "‘person acting on behalf of the Commission'’ includes any employee or contractor of the Commission, or employee of such contractor, to the extent that such employee or contractor of the Commission, or employee of such contractor prepares, disseminotes, or provides access to, any information pursuont to his employment or contract with the Commission, or his employment with such contractor, i ORNL-TM-2727 Contract No. W-T4O5-eng-26 Reactor Division THE MECHANICAL BEHAVIOR OF ARTIFICIAL GRAPHITES AS PORTRAYED BY UNIAXTAL TESTS . Greenstreet . Smith . Yahr . Valachovic Ty = e DECEMBER 1969 OAK RIDGE NATIONAL LABORATORY Ozk Ridge, Tennessee operated by - UNION CARBIDE CORPORATION for the U.S. ATOMIC ENERGY COMMISSION AIES LOCKHEED MARTIN ENERGY AESEARCH LIBRA (R 2 b \ 3 yusk 0533802 & T i1l }Contents Page N o T = Y o 1 TNETOAUCELON et s e oo vn st oe s ne e snneeneneseneensnnseneennsnnns 1 Mechanical Behavior ...veiiisiineiirersesscnnersesnsennans [ 2 Synopsis Of Reported RESULES «evrrerrennereerennneeenns 2 Additional Investigations ............. ettt 14 Conclusions .....ev.... e ettt et 41 ACKNOWI1EAEEMENTS « vt et ennerrsenneesosecnnassssennesesennnanas Lo RETFETEIICESE « v v et e aen s st aneansensesassnssenesnnssnsenennns L3 THE MECHANICAL BEHAVIOR OF ARTIFICTAL GRAPHITES AS PORTRAYED BY UNIAXTAL TESTS W. L. Greenstreet J. B. Smith G. T. Yahr R . S. Valachovic Abstract Stress-strain behaviors which are representative of nuclear- grade, or equivalent, graphites are described and discussed. oince c¢cyclic loading test results provide important insight into the characteristics of graphite behavior, emphasis is placed on results of this type. Monotonic loading curves are also consid- .ered, and both stress versus longitudinal strain and stress versus lateral strain curves are given for monotonic and cyclic loading. The new results given in this report are combined with results published in the literature to provide a unified descrip- “tion of the observed behavior. Keywords: graphite, materlals testlng, mechanical prop- erties, compression, tensile properties, stresses, deformation, heat treatment. - Introduction ' Many of the mechanical‘behavior charaoteristics of graphite have been studied. This is espeeielly true of'nuclear—grade, or equiyalent,_grsph— ites es a class of materials. These graphites are generally made from petroleum-coke and coal-tar pitch, but the same overali characteristics are exhibited by graphites made from other binder and filler_combinations | as Well It is the purpose of this report to bring out some of the salient _features of observed mechanlcal behav1or in & unified fashlon Important behavioral characteristics are descrlbed in the literature. These have been 1dent1f1ed from monotonic and cyclic loadlng tests, but there are certain aspects of cycllc loadlng behav1or that elther have not been studied or have not been studied in detail. Therefore, addlt;onal investigations were made to overoome this deficieney. Three types of cyclic loading were considered, including (1) cycling of specimens, which were preloaded in compression, between zero and a constant stress level less than the preload stress value, (2) cycling in compression wherein the specimens were only partially unloaded, and (3) cycling alternately in campression and tension. Studies of the first two types have not been reported previously, and studies of the third were used to experimentally examine the validity of an often stated premise regarding reloading be- havior. The results from these investigations when combined with avail- able data provide important background information for constitutive equa- tion development. Mechanical Behavior Despite material variations from grade to grade, from block to block of a given grade, and from piece to piece within a given block, there are definite characteristics of mechanical behavior associated with artificial graphite as a class of materiéls. Most graphites are anisotropic, and stress-strain data generally show that there is rotational symmetry of the anisotropy in an element. That is, the material may be classified as transversely isofropic. | The anisotropy is a result of the preferred orientation of the coke. particles used in the manufacture and the orientations of the‘cryétallites within the particles. Although the forming method, that is, molding or extruding, governs the preferred orientations of the particles, this method influences the character of the anisotropy only in terms of actual measures Of deformation resistance. Therefore, the characteristics to be discussed are independent of forming method. Synopsis of Reported Results* It is common knowledge that, when a graphite specimen is loaded in simple tension or compression, nonlinear stress-strain behavior is exhib- ited from essentially zero stress to the failure stress. What is the characteristic of this nonlinearity? By loading to a stress level less than that at failure and unloading, one obtains a loading curve, OA, and an unloading curve, AB, as shown schematically in Fig. 1. Since the *Information summarized here as well as summaries of other informa- . . . . - 1 ‘ tion on graphite are given in a survey report. ORNL IWNG. 69-L105 STRESS c B STRAIN Fig. 1. Schematic Drawing of a Stress-Strain Diagram for Graphite. unloading curve does not retrace any part of the initial loading curvé, the behavior cannot be classified as nonlinear elastic. The nonlinearity must be described in some other way. It may be seen that the behavior exhibited is reminiscent of time- independent elastic-plastic behavior ascribed to metals. When the speci- men is fully unloaded, there is a residual strain, OB, as for materials that undergo elastic and essentially time-independent plastic deformations. The unloading curve, AB, is also nonlinear. On loading the specimen a second time, the resultifig-curve is again nonlinear and a hystéresis loop is formed between the unloading and reloading cufves.' Loading beyond the stress corresponding to that at the unloading point, A, gives a curve which becomes asymptotic to the cuffie that would have existed. had unload- ing never occurred.¥ There are no abrupt changeé in the slopes of any of the segments of the entire curve. The tensile strengths and the initial slopes of the stress-strain curves (elastic moduli) are generally greater in the with-grain direction than in the across-grain direction. There are pronounced differences be- tween both stress and strain at fracture in tension and in compression: Complete stress-strain diagrams for simple tension and compression are plotted together in Fig. 2 (Ref. 2) to.illustrate these differences in the tensile and compressive behavior. The material is EGCR-type AGOT graph- ite,f and the data are for the with-grain (parallel tb the extrusion axis) direction. Typically, fracture strainé on the order of 0.1 to 0.2% and 1.0 to 2.0% are found in tension and compfession, respectively, for nuc- lear-grade, or equivalent, graphites. Arfagon and Berthier® performed compression test étudies on 216 spec- imens made from extruded, petroleum-coke, industrial graphite. Three types of tests were used: (1) simple compression; (2) cyclic tests (loading- unloading-reloading) in which the cycles were made from increasing Stfess levels spaced at equal intervals; and (5) cyclic tests befween zero stress *This is an often stated premise which is based on many observations. A direct experimental examination of this premise will be discussed later in this report. TEGCR—type AGOT graphite is a coarse-grained, nucleaf-grade, extruded graphite which was made by Carbon Products Division of Union Carbide Corp. ORNL DWG. 64-11422 2000 STRAIN (%) ‘ -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 TENSl}E 0.2 COMPRESSIVE ra -2000 = 2 / o} w . w / @ ot -4000 » — - -6000 Fig. 2. Complete Uniaxial Stress-Strain Curves Parallel to Extrusion Direction. and a fixed maximum. The slopes of the stress-strain curves at the origin for continued loading (type 1) were the same order of magnitude as given by sonic measurements. Seldin® also found good comparisons between sonic moduli and the slopes at zero stress as measured from tensile and compres- sive stress-strain curves for several grades of molded graphite. Arragon and Berthier found that cyclic compressive tests of type 2 cause the apparent density to increase. The hysteresis loops, which are formed by loading, unloading, and reloading, increase in size with in- creased maximum stress, and the slopes of straight lines connecting the unloading and reloading points decrease with increased stress. (The slope of a line connecting the two points of one cycle is termed the "paraelas- tic modulus.") | | A schematic diagram, which depicts the essential features of the be- havior observed from a test of fype 2, is shown in Fig. 3. The envelope curve corresponds to that for simple compression. Arragon and Berthier3 also discovered that straight lines drawn through the unlcading fioint and the point of zero stress for each ¢ycle converge at a single point as shown in the figure. The coordinates of this point are both negative (taking compression as pog}tive), and it was reasoned that the existence of this point is a manifestation of the history of the virgin specimen. In each case, the‘reloading curves asymptotically approach the enve- lope curve after each cycle of loading, unloading, and reloading. This and the other details of the behavior, as described above, are_genefally typical of the graphites being considered. Corroborations can be found by studying the results for extruded graphite reported by Losty and Orchard® and for molded graphites by Seldin.® In the case of test type 3, which was used in the study by Arragon 3 each specimen was subjected to 12.cycles. During the first and Berthier, cycles, the total deformations at the unlocading and reloading points in- creased with increased cycle number, but, after the sixth cycle, these deformations were essentially constant, afid the hysteresis loop was re- traced on each subsequent cycle. The form of the hysteresis lcop remained constant throughout the cyclic loading as did the paraelastic modulus. There was a slight increase in apparent density during cycling. ORNL DWG. 67-2981 STRESS STRAIN Fig. 3. Schematic of Compressive Stress-Strain Diagram. 8 A schematic diagram of the type behavior observed during the first few cycles is shown in Fig. 4. Additional illustrations of this behavior, as obtained by the present study, are given in Figs. 5 and 6. The latter figures show results obtained from EGCR-type AGOT and from RVD¥ graphite, respectively. The initial slopes of the stress-strain curves, or Young's moduli, for nuclear-grade, or equivalent, graphites are about the same in tension and in compression. This is generally true for both the with-grain direc- tion and the across-grain direction. However, exceptions may be fofindf%y comparing across-grain data for some graphites. Close comparisons of tensile and.compressive curves for a given direction also reveal that there is a tendency toward gfeater deformation resistance in bompreséion than in tension. This especially is true for the across-grain direction. The preceding discussion was limited to stress versus longitudinal strain behavior. ©Stress versus lateral strain curves for these graphites have different curvatures in tension and compression; the diagrams for tension are concave,toward the stress axis, while those for compression are convex. These observations were first reported by Seldin.* Schematic diagrams of the tensile and compressive behaviors are shown in Fig. 7. In uniaxial compression tests, the transverse—tOalongitudihal strain ratios are essentially independent of stress, yielding almost conétant values. However, the ratios in tension are functions of stress, decreasing as the stress is increased. Figure 8, which shows the strain ratios for EGCR-type AGOT graphite as functions of longitudinal strain,2 provides a clear illustration. In this figure, the subscript 3 refers to the paral- lel, or with-grain direction, while the subscripts 1 and 2 refer to two orthogonal (across-grain) directions in the plane of isotropy. (Trans- verse isotropy is assumed.) The first of the double subscripts indicates the difection of applied stress and the second indicates the direction of induced strain. The vertical bars in the figure represent standard devia- tions. ¥RVD is an extruded graphite manufactured by Carbon Products Division of Union Carbide Corp. CRNL IWG. 67-2980 STRESS STRAIN Fig. 4. Schematic Diagram Showing Cyclic Behavior. 10 QRRL IMG. 69-L106 4000 3500 ////// I 3000 N wn O o STRESS ( pst) N o O O 1 500 1000 500 0 o1 02 03 04 0.5 0.6 LONGITUDINAL STRAiN,EA(%) Fig. 5. Cyclic Stress-Strain Curves for a With-Grain EGCR-Type AGOT Specimen (12 Cycles). 11 ORNL TMC. 69-k104 e 8000 7 9000 7000 6000 5000 STRESS (psi) 4000 3000 2000 // 1 000 0 0.2 0.4 0.6 0.8 1.0 12 LONGITUDINAL STRAIN.EA(%) Fig. 6. Cyclic Stress-Strain Curves for a With-Grain RVD Specimen (S-Cycles). 12 - ORNL DWG. 67-3LT7 TENSION 4 4 3+ n e - a2 w , & % — l — 02 0 0 o A 2 3 4 LATERAL € LONGITUDINAL € STRAIN (PER CENT) COMPRESSION ‘ E‘ —"-" e 7 0 w x - " | Lt L)l oz . ol 4 8 12 16 20 LATERAL € LONGITUDINAL € STRAIN (PER CENT) Fig. 7. ©Schematic Drawings of Longitudinal and Lateral Stress-Strain Curves. STRAIN RATIO 0.20 0.15 0.10 0.05 0.15 0.10 0.05 0.15 0.0 0.05 ] 13 ORNL-DWG 64-10983 T ] ITT L o T - J l == | K ]y. + il { 1 ‘ 2 ‘| 2 Hi2 v F21 | H13 . 123 - ' s 1 [ K43, 123 LONGITUDINAL STRAIN IN COMPRESSION (%) Fig. 8. Il1 7 = _ I R HH | | HHHH | _— 1 | 1 ’ B30, H32 K31, 132 0.5 1.0 1.5 20 0 0.05 0.10 0.4% LONGITUDINAL STRAIN IN TENSION (%) Strain-Ratio Curves from EGCR—Type AGOT Specimens. 1L When a specimen is loaded and released, the transverse residual strain is positive regardless of whether the load is a tensile or a com- pressive one. Thus, the volume of a specimen pulled in tension and re- leased is increased since all linear dimensions are increased. Additional Investigations The investigations reported in this section were made using speci- mens with the design shown in Fig. 9. This particular design was selected so that the specimens could be tested in either tension or compression. The 0.100-in.-diam longitudinal hole was bored in each specimen to allow for making permeability measurements, and a smooth machine finish was used, without grinding or polishing. Each specimen was marked for gage location using a HB grade drafting pencil to avoid scratching the surface and to provide indications thatr would not be removed by heat treating the specimen af 3000°C (to be ais- cussed later). The latter aspect is important because the specimens were heat treated and reinstrumented‘fwice in ‘some cases, and it was necessary to mount the gages in the same location each time. They were instrumented on opposite sides at the midpoint of the gagé length with strain gages oriented in the axial and circumferential directions. Budd Metalfilm, type C6-121-A, strain gages with a 0.125-in. gage length were used. | Tensile tests were pérformed using steel clam-shell,fixturés attached to the ends and 0.20-in.-diam, stainless steel, stranded-wire cables for transmitting the. load to the specimen. A specimen, ready for test, is shown in Fig. 10. The campressive tests were carried out using a subpress with a miniature load celi placed in the subpress below the specimen. By placing the load cell at this location, inaccuracies in readings due to plunger friction were eliminated. The test setup for compression testing is shown in Fig. 11. For those cases in which the specimens were heat treated, a 100 kw, 3000 Hertz induction furnace was used for this purpose. The inside dimen- sions were a diameter of 9 in. and a length of 14 in., and the specimens were heated in Argon at atmospheric pfessure. The time to reach the 3000°C temperature level was 45 minutes. The furnace was cooled from 3000 to 15 ORNL-DWG 69-5079 3.50 in. 0.50 . ) e 2.500 i <:::1{32—1'n. R. (both ends) j Lo.10 L Dia. 0.625-in. Dia. ’ 2-in. R. j;;”' R. L—1.125-1n. Dia. 1.250 in. - _flZB.szs in. Fig. 9. Drawing of Test Specimen. 16 PHOTO TE4ET Fig. 10. Specimen Prepared for Tensile Testing. bd = o 3 0 L B @ = o ;n ] Q i = B o o & O B i) v B o oy )] ~ i = u B - o )] 2y o 1. 18 2500°C in approximately 15 minutes, from 2500 to 2000°C in about 1 hour, and from 2000°C to room temperature in 12 to 14 hours. In the first series of tests, specimens from EGCR-type AGOT graph- ite were subjected to the following program of loading. Each specimen was loaded to a given stress level in compression and unloaded. It was then cycled ten times between zero and a stress, 00,'which was in the range given by 1 5 0 < o, < o, where O is the maximum stress level, that is, the stress corresponding. to the first unloading point. In each case, the specimen was subsequently loaded to failure. Two with-grain and two across-grain specimens were tested, and both lateral, or transverse, and longitudinal, or axial, strain versus stress diagrams were obtained. Compressive behavior was chosen because relatively large stresses and strains can be induced with- out failure. Figure 12 shows the results for the with-grain direction. Here again it may be seen that, when the specimens were loaded to failure after being cycled, the stress-strain curves asymptotically approached exten- sions to the initial loading curves that would have been obtained by con- tinued loading beyond the unloading stress, o - The hysteresis loops exhibited by the stress-strain curves do not change in size and coincide rather than change position along the strain axis during the first few cycles, as was shown in Fig. 4. The cfirves for the across-grain specimens are shown in Fig. 13. - Except for very slight deviations in the initial portions of the first few reloading curves; the hysteresis loops are again retraced on each cycle. Asymptotic approach to a monotonic loading curve upon loading to failure may also be seen. | Partiasl unloading characteristics were examined using a second set of four with-grain and four across-grain EGCR-type AGOT graphite speci- mens. A specimen was loaded in compression to a given stress level, O and unloaded. It was again loaded to a higher compressive stress level, O, unloaded, reloaded to o, & second time, and unloaded. A companion 19 ORNL DWG. 69-2977 3500 ' / // _— | /fl/ 4 A/ // ol S /] /. 1Y ol [y n 14d Q o n o o o STRESS (pm) 1000 0 02 04 06 08 (O 12 0 -002 -004 -006 -008 -0.0 LONGITUDINAL STRAIN, €, (%) , LATERAL STRAIN,E (%) 4000 — | i Wi /| /, /// -/ N o O o STRESS (p=m1) 3 oF e 1500 1000 . 500 0 0 Q4 06 0.8 1.0 1.2 0O -002 -004 -006 -008 -0.0 LONGITUDINAL STRAIN, €, (%) LATERAL STRAIN,CI_(%) Fig. 12. Cyclic Curves for Two Preloaded With-Grain EGCR-Type AGOT Specimens. 4 20 ORNI, DWG. 69-8209 sor_ > - N 711 L7 o Y/ 7 00 /,//- - B // // - [/ ' [ p Y LY f 4 STRESS (psi1) 500 TT TT I.2. 1.4 0 -O.dZ -004 -Q06 -008 -010 -0I2 0 02 o LATERAL STRAIN, € (%) 06 08 1.0 LONGITUDINAL STRAIN.E, (%) 3500 g 2 Q 7 /I 7 1/ /1) / o 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 -002 -004 -006 -008 -010 -0J2 LONGITUDINAL STRESS €,(%) LATERAL STRAIN,€ [ (%) 8 o STRESS {ps!) 1500 ar ‘ FPig. 13. Cyclic Curves for Two Preloaded Across-Grain EGCR-Type AGOT Specimens. ' 21 specimen was then loaded in compression to O partially unloaded, re- loaded to the higher compressive stress level, O partially unloaded, reloaded to on,-and unloaded. | ‘ | The stress-strain diagrams for two companion with-grain specimens are shown in Fig. 14. The diagrams for a second set of with-grain speci- mens are shown in Fig. 15, and Figs. 16 and 17 show the results for sets of across-grain specimens. | The cyclic behav1ors of the fully unloaded speclmens follow the pat— terns described earller An examination of the stress versus longltudlnal strain curves‘revealS'several features. The hysteresis loops increase in size and the paraelastlc moduli decrease w1th increase in unloadlng stress level the total deformatlons increase upon loading a second tlme to the higher compressive stress level, o n’ and the reloading curves became asymptotic to extensions of the initial loading curves when the specimens are loaded beyond O The stress versus lateral strain curves are very similar in character to the stress versus longitudinal strain curves. In the case of partial unloading, hysteresis loops are again formed, and the same general features that were observed for full unloading may be seen. The hysteresis loops are smaller, but thé segments of each curve are entirely nonlinear. The asymptotic approach to the extension of the initial loadlng curve in each case corresponds to that for full unloading. Since the detalls of the asymptotic approach feature are important to the development of mathematlcal analogs to the materlal behavior, the qurves in Figs. 14 through 17 were carefully studied to provide guanti- tative data. When a specimen is loaded to a stress level such as g m’ unloaded either fully or partlally, and reloaded, the relcading curve undergoes an accelerated change in slope as the.stress approaches the stress, © ; at the point of unloading. This accelerated change persists until the reloadlng curve essentially begins to trace what would have been the extension to the initial loading curve provided unloading had not occurred. The quantitative data obtalned were the stress values at the extréemities of the accelerated change reglon The lower llmlt is denoted by o/ and the upper limit is referred to as o”. The values for o/ and ¢’/ were obtained by plotting the ratio of longitudinal strain to stress as a function of stress. The data on either a 22 ORNL IWG. 69-2981R 3500 § 8 \% =~ i / E Ay L 71/ / Y L ) o 01 02 ©03 04 03 06 - 0 -002 -004 -G06 -008 -0.0 -QI2 LONGITUDINAL STRAIN, €, (%) o LATERAL STRAIN, €_ (%} e 10/ 10/ (a) Full Unloading 2000 VI AT ] il / 1] § N N\ BN 500 A : o / / okt / ‘ 1010 o 0.1 02 03 ©04 05 08 . : 0 -002 -004 -006 -008 -QI0 =012 LONGITUDINAL STRAIN, €, (%) LATERAL STRAIN, € (%) (b) Partial Unloading Fig. 1. Cyclic Curves for Fully and Partially Unloaded With-Grain EGCR-Type AGOT Specimens. ' 23 ORNL DWG. 69-2982R 3500 3000 2500 é [ @ 2000 A ; / i ® 1500 /////// // 1 1000 //// e //fl ' 500 [— Nd/4 o Il ,,.. 0.1 02 03 04 05 06 0 -002 -0D4 -006 -Q08 -0 ~-0.12 -0i4 LONGITUDINAL STRAIN, €, (%) . LATERAL STRAIN, €; (%) (a) Full Unloading 1500 3000 _ 2500 : 7 / w» 2000 / @ I : ) 1 @ 1500 / : ‘! 1000 // : // ; 500 / 0 1028 i 08¢ 6 o1 02 03 04 085 06 0 -002 -004 -Q06 -008 -OI1 -042 -Ol4 Lor‘:GlTUDINAL STRAIN, €, (%) LATERAL STRAIN, € (%) (v) Partial Unloading Fig. 15. Cyclic Curves for Fully and Partially Unloaded With-Grain EGCR-Type AGOT Specimens. ORNL DWG. 69-2983R 3500 3000 i L7 / L 7/ . s /1y T T ) LovarruomaL Seam. €, () Larcan, sTRan, €, () (a) Full Unloading f 2000 71 E 1500 /77 é/ /4 / T Y -~ Ay <] " 1 Y 1. 0 0.1 02 O3 04 05 06 07 08B 09 1.0 0 -002 -004. -006 -008 -0.10 LONGITUDINAL STRAIN, €, (%) LATERAL STRAIN, €, (%) (v) Partial Unloading Fig. 16. Cyclic Curves for Fully and Partially Unloaded Across-Grain EGCR-Type AGOT Specimens. ‘ - S TRESS {(pn) STRESS (ps:) o5 3500 3000 2500 p? 2000 ////;f”,/’/y' 1500 1 / / / vd CNANA 1000 v A /'/ | A 500 A /A/// N 104 oo 21 - o2 c3 pa ©05 06 07 08 09 10 LONGITUDINAL STRAIN, €, {%) (a) Full Unloading 3500 3000 2500 ] 2000 // 1500 1/ 4 AN 1000 L /' 300 '/ /// : 104e oo ol 02 03 04 05 06 07 0.8 Q9 1.0 LONGITUDINAL STRAIN, €, (%) (v) Partial Unloading Fig. 17. EGCR-Type AGOT Specimens. ORNL DWG. 69-2980R el 1/ "/ .7 1// 0 -a0z2 -Q04 -006 -008 -0 LATERAL STRAIN, €. (%) T / / va 0 -Q02 -004 -008 -0.08 -O! -or LATERAL STRAIN, €, (%) / 1044 Cyclic Curves for Fully and Partially Unloaded Across-Grain - 26 side of O are described by straight lines with greatly different slopes. The transition region then ccrresponds to the region where accelerated change in slope of the stress-strain diagram occurs, and the limits may be determined. A typical example of the plots used for determining o’ and o// is shown in Fig. 18. The two stress levels, denoted by o/ and o’/, and the value of o, for each stress versus longitudinal strain curve are listed in Table 1. Also included are the values for the ratios 0’/Gm and o’fl/om_. The dots on the stress~strain diagrams of Figs. 14 through 17 mark the limits of the transition regions. It may be seen from the table and the figures that there is consistency between the values. The value of 0, is the same for all cases. In five of the eight cases, the 0’/0m ra- tios are the same, and all ratios fall within the range from 0.93 to 0.97. The upperlextremities of the segments for full unloading are at higher stress levels than those for partial unloading, which shows that asymp- totic approach is achieved sooner in the latter case. For full unloading, the OJ//Gm . ratio ranged from 1.04 to 1.07, while this ratio ranged from 1.01 to 1.03 for partial unloading. The final test series reported herein was used in an attempt to ex- perimentally verify the asymptotic approach premise and to provide data Table 1. Data Obtained From Curves for / Partial Unloading Studies . / Specimen Om o o/’ . ‘ / r7 Number Orientation (psi) (psi) (psi) ¢ /Gm o’ o, 101 WG 2000 1850 2130 0.9% 1.07 101a WG 2000 1950 2040 0.97 1.02 102 WG 2000 1900 2130 0.95 1.07 1023 WG 2000 1900 2050 0.95 1.0% 103 AG 2000 1850 2080 0.93 1.0k 103a AG 2000 1900 2020 0.95 ° 1.01 104 AG 2000 1900 2100 0.95 1.05 10La AG 2000 1900 2025 0.95 1.01 27 ORNL DMG. 69-10965 2.8 _ : - — 2.6 2.4 : : ‘ (10 &Y 1b) \ 2.2 / z ¢ < 9 o '._. w| v |8 7 ZE20 . S| o« = o ~ T s —?f// o/ c/ 2 1.6 0 500 1000 1500 o' 0,0 2500 3000 STRESS (psi1) Fig. 18. Plot for Determination of the Extremities of the Accelerated Change in Slope Region for a Fully Unloaded Across-Grain EGCR-Type AGOT Specimen. 28 on cyclic behavior under alternate compressive and tensile loadings. Studies on the validity of the asymptotic approach premise were motivated by Seldin's? demonstrations that test specimens can be made to reproduce their original stress-strain responses in the longitudinal and lateral directions by heat treating them at their graphitization temperatures. Provided it is generally true, this is a revolutionary discovery in graph- ite testing, especially in the potential it provides for removing uncer- ~tainties in data analyses and interpretations that arise because of the almost inevitable variations in graphite properties. Pursuant to making the final studies regarding the cyclic behavior of graphite{ Seldin's method for removing deformation history was examined using specimens made from ATJ, RVD, and EGCR-type AGOT graphites. The maximum particle sizes are 0.006, 0.015, and 0.032 in., respectively. The first two are molded graphites. Each specimen was subjected to the following sequence of steps. 1. Heat treat at 3000°C for 60 minutes. 2. Load in tension to a predetermined stress level and unload. 3. Heat treat at 3000°C for 60 minutes. L. Load in compression to a stress level greater than in Step 2 and unload. 5. Heat treat at 3000°C for 60 minutes. 6. Load in tension to the stress level of Step 2 and unload. The first heat treatment was used to establish the graphitization tempera- ture in each case. | Stress-strain curves for two with-grain specimens of ATJ graphite are plotted in Fig. 19, where both the longitudinal and lateral strains are shown. The curves for two across-grain ATJ specimens are plotted in Fig. 20. All stresses, tensile and compressive, and the corresponding longi- tudinal strains are plotted as positive in all figures depicting results from the heat treatment studies, and the transverse data are plotted in an analogous manner. Note that the two strain scales for a given speci- men are not the same; hence, the differences between lateral strain curves are accentuated. Curve No. 1 is for the first loading in tension, No. 2 is for the compressive loading, and No. 3 is for the second loading in tension. 29 ORNL DWG. 69-4103 3200 e : : ‘ 2800 : // ’ : / 2400 ! ’ . . ’/,3 A / L g L/ STRESS '(“pnl g 'l N § N N o o) o AN /1/ A /711 | 1V 20! 400 201 / ° ° Qaos QI0 0I5 02 025 03 Q35 0.4 043 001 0 -00I -002 -Q03 -004 -005 -006 -Q07 LONGITUDINAL STRAIN, €A (%) LATERAL STRAIN, ET('/n) - z / T | IRy AV, /// | IS A/ M s00 A/ | / /i/ 470 /) 400 STRESS ~ 1200 202 202 045 ‘ ool C -001 -+002 -003 -004 -005 -006 -QO7 o 005 0.1 005 02 025 030 035 040 LATERAL STRAIN, €; (%) LONGITUDINAL STRAIN,E, (%) Fig., 19. ©Stress-Strain Curves for Two With-Grain ATJ Specimens. 30 ORNL DWG. 69-2975 3200 2400 = - . -7 " 7y / STRESS {pa+ ) g N v - ,/ «00 / — - // oy o o Q0% at s 0.2 25 03 03% 04 0ad 003 002 00 0 -001 -002 -0035 -D0O4 -005 -006 - LONGITUDINAL STRAIN , €, (%)} LATERAL STRAIN.€ (m} 004 -005 -006 -QO07 . T 3200 - 2800 / 2400 Ve ' .‘Z - "2 - /) 22000 Y Pt 3 Y a W EIGDO ’A / 1200 / 4 M/ 800 ,/ /// / a4 J 400 // / // . ros // . o 005 01 Q5 02 02% 03 Q3% 04 045 003 002 00 0 -001 -002 -003 -004 -GO5 -006 -007 LONGITUDINAL STRAIN, €, (W LATERAL STRAIN, €4 (%) Fig. 20. ©Stress-Strain Curves for Two Across-Grain ATJ Specimens. 51 The success of the heat treatment method for removing deformation history is determined by comparing curves 1 and 3 for each specimen. Since the second with-grain specimen failed during the second tensile loadlng, comparisons are more difficult to make for the curves shown in the lower part of Fig. 19 Figures 19 and 20 show that the hlstory was almost entirely removed by heat treatment. Although there is a small decrease in deformatlon resistance between the first and second tens1le loading, the agreement between the curves is very good. ans is true for both the with-grain and across-grain directions. Some of the.beheyiora;flaSPects which were discussed in tne previous section may be seen from these curves. The'deformation resistances are essentially the same in tension'and in compression, with a very slight _tendency toward greater deformation res1stance in compression than in tension for the across- grain direction. The durvatures of the stress versus lateral strain curves in tension are different from those in com- pression, but the initial slopes are nearly equal. Finally, positive residual lateral strains are incurred as a result of loading in tension. The stress-strain curves shown in Figs. 21 and 22 are for two with- grain RVD specimens and one across-grain specimen. Curves 1 and 3 for the with-grain specimen are in good agreement. Again, slightly less deformatlon resistance is found for the second loading in tension. Although the differences between curves 1 and 3 are greater for the across-grain specimen than in the above cases, the‘agreenent is gocd. Greater deformation resistance in compression than in tension may also be seen in Fig. 22. | | " | Results obtained from EGCR;type AGOT specimens are given in;Figs. 23 and 24. The curves for the with-grain ‘specimen again show that the heat treatment was. réasonably effective. It was less:effective for the across-grain specimens, however, as may be,seen by examining Fig. 24. The second across- -grain specnnen was not loaded to the same stress level in tension during the first and third loadings, but thls does not detract significantly from the comparisons. Again, the across-grain specimens exhibit greater deformation resistance in compression than in tension. The above results for the three graphites show that, in general, heat treating does almost entirely remove the prior deformation history 3200 2 2800 2400 §2000 & 1600 1200 800 / 400 / o/ o 0 3200 2800 STRESS (pm) nN n g 8 3 © o g 800 400 o 003 0. 015 02 025 0.3 LONGITUDINAL STRAIN, €, (%) 033 04 0.43 7 7 =3 I ) /| / = NN Jor o 0.03 Fig. 21. 0.1 045 02 025 0.3 LONGITUDINAL STRAIN €, (%} 0.3 04 0.48 Stress-Strain Curves ORNL IWG. 69-8210 AL //,-, } N /) 1Ly /// -0.03 -0.04 -00% -006 A 004 003 002 OO0l 0 -0.01 -2.& LATERAL STRAIN, €, (W frr /8 1 /L) A 0.04 003 002 00/ O -00 -002 -003 -0.04 -0.03 -0.6 LATERAL STRAIN, € (%) 302 for Two With-Grain RVD Specimens. ct 2000 2000 o 'O o 11200 STRESS (psi ) 800 400 L 7 / - Z4 / z AN N\ 304 005 ol 0i5 0.2 025 03 035 LONGITUDINAL STRAIN, €A (%) Q.49 045 ORNL, DWG. 69-2972 / ¢t /| 1Y 304 003 002 oo . O -QO0l -002 -003 LATERAL STRAIN,€ (%) Fig. 22. Stress-Strain Curves for an Across-Grain RVD Specimen. 3k ORNI, DWG. 69-2973 1200 1000 [— . " ‘3 AEEy/a N/ "’ / /4 w LS /I /;/ W 200 P // P o V4 0 002 004 006 008 010 0004 O -0004 -0.012 LATERAL STRAIN, EI (%) LONGITUDINAL STRAIN,E, (%) 1200 / | / 1000 ’ / - S/ T Y/ A/ /) 7T /] 0 o 0.02 004 0.06 008 010 0004 0 -0.004 -0.012 LATERAL STF\’AIN,ET (%) LONGITUDINAL STRAIN, €, (%) o © © N S ~. srm-fs (pst) N SO\ H Q O Fig. 23. Stress-Strain Curves for Two With-Grain EGCR-Type AGOT Specimens. 600 2 500 V4 = / 3 g. 400 " v & / & 300 A/ 7 [ o) 200 / W/ T 0 0.02 004 006 008 LONGITUDINAL STRAIN, €, (%) 600 - s00 ? // ; 3 200 A 8 | / 7 W 300| - /.// _ & / / 200 ,//// o A/ ) / 2T o . 0O 002 004 006 008 O LONGITUDINAL STRAIN, €, (%) O.1 35 ORNL DWG. 69-2971 T 000 ] . -0004 -0.008 -00I2 LATERAL STRAIN, ET (%) -00lI6 %—_ L™ s .\ = P — / 2T 0.004 0 -0004 -0008 -00I12 -0.016 LATERAL STRAIN, € _(%) Fig. 24, Stress-Strain Curves for Two Across-Grain EGCR-Type AGOT Specimens. 36 so far as stress-strain response 1s concerned. The greater differences between the first and second tensile loadings for across-grain RVD and AGOT specimens may be associated with opening of microcracks too large to be healed by heat treatment. This conjecture is the most plausible for - the coarse-grained AGOT material. In addition to the investigations just described, studies were made to examine the influence of hold time at 3000°C. EGCR-type AGOT graphite specimens were used, and the hold times were 20, 30, and 40 minutes (in addition to the 60 minutes hold time used in cbtaining the results already discussed). The results were inconclusive because no definite trend with hold time was established. However, it was decided that the longer hold time is preferred to assure greater possibility for reproducible results. Returning to the dual purpose series for examining the asymptotic approach feature and for obtaining data on cyclic behavior under alternate compressive and tensile loadings, three (two with-grain and one across- grain) EGCR-type AGOT specimens were tested. After an initial heat treatmgnt at 3000°C for 60 minutes, the first loading for each specimen was used to establish a one-cycle envelope curve. FEach was loaded in compression near that required for failure, unloaded, loaded in tension to a given stress level, again near that required for failure, and un- loaded. During preliminary tests it was found that a specimen will fail in tension when the tensile strain, as measured from the zero stress point, corresponds to that at failure for a virgin specimen loaded in tension only, regardless of the position of the zero stress point. The envelope curves are shown as short-dashed lines in Figs. 25, 26, and 27. The apparent discontinuities in the slopes of the curves between the tensile and compressive portions were caused by changing fixtures to apply loads of different signs. Each specimen was heat treated a second time under the same conditions as above and retested. It was loaded in compression to a given stress level, unloaded, loaded in tension, unloaded, loaded in compression to a higher stress level than the first, and unloaded. The corresponding curves are shown by the long-dashed lines in the figures. Finally, the heat treatment and loading sequence was repeated using lower compressive . stress levels. The solid curves in the figures were obtained in this way. 37 ORNL DWG. 69-29T70 -3000 -2500 7 ¥ /:/ % /’ ; ‘ ) \\\\ -2000 7 q — // / ’I . \ \ : 1 W 1500 J | N ' /17 \ -1000 /}, /, ’/, !\‘\\\ // / Il | \\\ \ -500 //" II \\\\ \ STRESS ( N, ~ 1 0 ”4” \ /1) 7 ‘ /i \ [y 5P \ \ 500/ —L \ o -0.1 -2 -03 -04 -05 -06 o0B 006 004 002 0 LONGITUDINAL STRAIN, €A {%) LATERAL STRAIN, €. (%) Fig. 25. Multi-Cycle Curves for a Heat-Treated EGCR-Type AGOT Speci- men (With-Grain). 38 -3000 - 2500 —— T - / d’ , -2000 i ~ /// ,' v / ’ = e J %1500 4 , T ,1 ,I 3 /%// / A x-1000 7r) L/ - - / / ,l m 717, /// / /’ ~-500 A 2 // / /,’ C/ /) 0 / /: ,/l /7 £ s // ,I,/ ¥ 500 £ 6P 1000 o -01 =-02 -03 -04 -05 LONGITUDINAL STRAIN,GA (%) -06 ORNL IWG. 69-29T4 006 004 002 0 LATERAL STRAIN, € (%) Fig. 26. Multi-Cycle Curves for a Heat-Treated EGCR-Type AGOT Speci- men (With-Grain). 39 STRESS (ps¢) -3000 -2500 s . ] ) ”’II‘ -2000 ”/ II’ th ’l . /‘l ’ 4 / -1500 # - . 1 . A /f / /'I fl/ / ’ -1000 f 2 /7/ /s [/ - 500 [—HptF—rt iAo l/ / S 0 ll / / Yoour l . "I’ /i 6T 500 ' 0 -02 -04 -06 -08 ~-1.0 -2 LONGITUDINAL STRAIN, €, (%) Fig. 27. Specimen (Across-Grain). ORNL DWG. 69-2976 Q06 004 Q02 LATERAL STRAIN, ET (%) Multi-Cycle Curves for a Heat-Treated EGCR-Type AGOT Lo The maximum tensile stress level was essgsentially constant for all load- ingé on a given specimen. In all cases, the stress level for reloading approaches that at the unloading point before a transition in slope occurs. The curvature of a stress versus longitudinal strain curve generally changes rapidly in the immediate vicinity of the unloading point, while the change for a stress versus lateral strain curve occurs over a much wider stress interval. Behavioral details for the stress versus longitudinal strain curves shown as solid lines are - difficult to detect because of the small scale. In fact, the initial loading and relocading curves were separated in the fig- ures for clarity. The discontinuities in slope attributable to changing of loading fixtures are apparent in almost all cases. The hysteresis loops and other aspects of the behavior are in accord with the results from other cyclic tests. Consider the stress versus longitudinal strain curves. The‘asymp- totic approach feature at the lower stress levels is shown, but there is some deviation.as the unloading stress level is increased. However, Figs. 25, 26, and 27 show that the upper portions of the dashed curves lie above the envelope curve in two caseg and below this curve in one case. On the whole, asymptotic approach to an initial loading curve is demonstrated by these data. | Additional support is given by the stress versus lateral strain curves of Fig. 25, where the agreement with the envelope curve 1s excel- lent. However, the stress versus lateral strain curves in Figs. 26 and 27 indicate that asymptotic approach is at most closely approximated. Conversely, because the restoration of the original stress-strain response by heat treating was shown to be less than perfect by the studies made to examine this technique, fully satisfactory agreement would be difficult to achieve. As in the cases of the longitudinal strain curves for the two speci- mens which show greater differences, the deviations from the envelope curves for the lateral strains are in one direction. The two subsequent loading curves in each case appear to define an envelope curve although it is different from the one established during the first loading. al A better selection for studies of this kind would have been ATJ graphite specimens since heat treatment was shown to bé more effective for this than the other two materials examined. As demonstrated here, discovery of the heat treatment method allows for detailed comparisons where none could be made formerly. Conelusions This report describes the stress-strain responses of nuclear-grade, or equivalent, graphites under monotonic and cyclic loading conditions. Both stress versus longitudinal and stress versus . lateral strain data are considered, and it .is shown that consistent patterns of behavior are identifiable from the results. These include differences in deformation resistance in tension and compression and differences in strain-ratio data for the two types of loading. When a specimen that has been preloaded in compression is cycled in compression between zero and given maximum stress levels, three behav- ioral patterns are observed. For the case in which the maximum stress levels are continually increased and all are greater than the preload value on each subsequent cycle, the hysteresis loops become larger and the paraelastic moduli decrease with increasing maximum stress. Cycling between zero and the preload stress level produces hysteresis loops that do not change with cycle number and the paraelastic moduli remain the same. However, the loops translate along the strain axis for.the first few cycles, after which the loops coincide. Fioally, cycling between zero and a fixed maximum value which is less than the preload stress pro- duces hysteresis loops’that are essentially invariant both in size and location with cycle number. In the case of a specimen which is loaded, partially unloaded, and reloaded, hysteresis loops are formed by the stress versus longitudinal strain and stress versus lateral strain curves. Also, the reloading curves approach what would have been extensions to the initial loading curves more rapidly than for the case‘of full unload- ing. The approach to the extension of a continued loading curve on re- loading is often assumed. Here, this postulate is shown to be supported by data obtained from tests designed to examine its validity. Lo The use of heat treatment at the graphitization temperature essen- tially removes the effects of prior deformation, so far as stress-strain response is concerned. However, these heat treatments were more success- ful in rembving the prior deformation histories of ATJ and RVD graphites than of EGCR-type AGOT graphite. Careful use of this technique is shown “ to be of unique benefit for making comparisons of the stress-strain be- haviors of a single graphite specimen under different loading histories. Acknowledgements The authors gratefully acknowledge the assistance of J. M. Napier of the Oak Ridge Y-12 Plant in heat treating the specimens; J. M. Chapman and J. P. Rudd for their help in instrumenting and testing the specimens; and H. A. MacColl for preparation of the figures. 1. 2. L3 References W. L. Greenstreet, "Mechanical Properties of Artificial Graphites — A Survey Report," USAEC Report ORNL- M527, Oak Ridge National Labo- ratory, December 1968. W. L. Greenstreet, J. E. Smith, and G. T. Yahr, "Mechanical Propertles “of EGCR-Type AGOT Graphlte," Carbon, T7(1): 15-45 (1969). P. P. Arragon and R. M. Berthier, "Caractérisation mécanique du graphite artificiel,"” pp. 565-578 in Industrial Carbon and Graphite, Society of Chemical Industry, London, 1953. E. J. Seldin, "Stress-Strain Properties of Polycrystalline Graphites in Tension and Compression at Room Temperature,” Carbon, W(2): 177- 191 (1966) . H. H. W. Losty and J. S. Orchard, "The Strength of Graphite," pp. 519-532 in Proceedings of the Fifth Conference on Carbon, held at -Pennsylvania State University, Vol. 1, MacMillan, New York, 1962. 45 ORNL-TM-2727 Internal Distribution 1. S. E. Beall ' 34, G. B. Marrow, Y-12 2. H. W. Behrman, RDT 35. J. G. Merkle 3, 5. E. Bolt | 36. A. J. Miller 4. J. W. Bryson 37. S. E. Moore 5. J. P. Callahan 38. J. M. Napier, Y-12 6. 8. J. Chang : 39. A. M. Perry 7. W. H. Cook o - k0. C. E. Pugh 8. J. M. Corum | | 41. J. N. Robinson 9, W. B. Cottrell 42, M. W. Rosenthal 10. F. L. Culler 43, A. W. Savolainen 11. W. G. Dodge 4h, M. J. Skinner 12. W. P. Eatherly Ls-s4, J. E. Smith 13. A. P. Fraas 55. I. Spiewak 14-23. B. L. Greenstreet 56. J. L. Spoormaker 24. R. C. Gwaltney 57. D. A. Sundberg 25. P. N. Haubenreich 58. D. B. Trauger 26. P. R. Kasten 59-63. R. S. Valachovic 27. C. R. Kennedy 64. M. S. Wechsler 28. K. C. Liu 65. G. D. Whitman 29, M. I. Lundin 66-75. G. T. Yahr 30. R. N. Lyon T6-77. Central Research lerary 31. H. G. MacPherson 78-79. Y¥-12 Document Reference Section 32, R. E. MacPherson 80-84. Laboratory Records Department 3%, H. C. McCurdy 85. Laboratory Records, ORNL R.C. External Distribution 86. S. A. Bortz, IIT Research Institute, Chicago 87. H. L. Brammer, SNPO-C, NASA Lewis Research Center, Cleveland 88. L. C. Corrington, SNPO-C, NASA Lewis Research Center, Cleveland 89. J. F. Cully, SNPO-A, c/o USAEC, P. O. Box 5400, Albuquerque 90. R. J. Dietz, Los Alamos Scientific Laboratory 91. D. M. Forney, Air Force Materials Laboratory (MAC) Wright- Patterson Air Force Base, Ohio 92. C. W. Funk, Aerojet-General Corp., Sacramento 95. J. dJ. Gangler, Materials Engineering Branch, RRM, NASA, Washington,D.C. o4. Harold Hessing, SNPO-A, CMB Division, Los Alamos Scientific Lab. 95. A. N. Holden, Westinghouse Astronuclear Laboratory, Pittsburgh 96. Gary Kaveny, Aerojet-General Corp., Sacramento . 97. J. J. Lombardo, SNPO-C, NASA Lewis Research Center, Cleveland 98. L. L. Lyon, Los Alamos Scientific Laboratory 99. D. P. MacMillan, Los Alamos Scientific Laboratory 100. M. M. Manjoine, Westinghouse Astronuclear Laboratory, Pittsburgh 101. J. E. Morrissey, USAEC, Washington 162. R. E. Nightingale, Pac1f1c-Northwest Laboratory, Richland 105. W. G. Ramke, Air Force Materials Laboratory, Wright-Patterson Ai ir Force Base, Ohio 1042105, 106. 107. 108. 109. 110. 111. 112. 113. 11k, 115, 116-130. 131. ZQOE';UM"&:!';UEQ L . Rowley, Los Alamos Scientific Laboratory . Scheib, SNPO, USAEC, Washington . Schroeder, SNPO-C, NASA Lewis Research Center, Cleveland . Schwenk, SNPO, USAEC, Washington . Seldin, Parma Research Center, Cleveland . Singleton, Westinghouse Astronuclear Laboratory, Pittsburgh . Smith, Los Alamos Scientific Laboratory . Spence, Parma Research Center, Cleveland . Swanson, Westinghouse Astronuclear Laboratory, Pittsburgh Thielke, SNPO, NASA Iewis Research Center, Cleveland Tu Lung Weng, Parma Research Center, Cleveland Division of Technical Information Extension (DTIE) HgHOWQOHGOR 0N - Laboratory and University Division, ORO Yy '