You are on page 1of 33

\

W

Division of Engineering

BROWN UNIVERSITY PROVIDENCE, R. I.
J ENGINEERING MATERIALS RESEARCH LABORATORY

CREEP OF POLYURETHANE "UNDER VARYING TEMPERATURE FOR NONLINEAR UNIAXIAL STRESS J. S. Y. [AI and W. N. FINDLEY

t

I/

Materials Research Program Brown University ; ARPA SD-86 : ARPA E-76 EMRL-44 February 1971 "

CREEP OF POLYURETHANE UNDER VARYING TEMPERATURE FOR NONLINEAR UNIAXIAL STRESS by by•* TrOHNIOAL LIBRARY

J. S. Y. Lai

and W. N. Findley

BLDG. 805
A0ERDEENRO-vIG GROUND, BTEAP-TL

0)

Summary Two methods are described to account for varying temperature during creep. Both employ the modified superposition principle. One uses a reduced time involv-

ing a shift factor which is a function of both stress and temperature history. The other considers the strain to be a function of the current values of stress and temperature. Experiments on polyurethane include constant stress creep and recovery at several temperatures in the nonlinear range and an experiment in which the stress was held constant while the temperature increased at a constant rate, then the stress was removed and the temperature decreased at a constant rate. The strain

in this experiment was predicted by means of the theories from the results of the constant temperature creep tests. The strains in the constant temperature creep and recovery tests were described by means of the multiple integral representation and the modified superposition principle. Most of the nonlinearity and temperature effect were found

in the coefficient of the time-dependent term.

Assistant Professor of Civil Engineering, University of Utah, Salt Lake City, Utah; formerly Research Associate, Brown University, Providence, Rhode Island. Professor of Engineering, Brown University, Providence, Rhode Island.

Introduction In a recent paper [1] the authors described the results of creep experi-

ments in the nonlinear range on polyurethane in torsion at various temperatures. The results showed that the time dependence could be described adequately by a power function whose exponent was independent of stress and temperature. In the

present paper tension creep experiments in the nonlinear range performed on the same specimen of polyurethane as part of the same series of experiments are reported for several temperatures. Constitutive equations describing these results

are employed to predict the behavior under a constant rate of temperature rise (and fall) during creep (and recovery) at constant stress. Previous investigations of nonlinear viscoelasticity have mostly been confined to room temperature. However, Bernstein, Kearsly and Zapas [2] and Lianis

[3] have extended the concept of a thermorheologically simple viscoelastic material to the nonlinear range. Earlier investigations by the authors [4 to 9] on

poly (vinyl chloride) and polyurethane at room temperature under combined tension and torsion showed that the creep strain could be separated into time-dependent and time-independent parts. The latter could be expressed as a power function The time-independent term was It was

whose exponent was independent of stress state.

nearly linear, but the time-dependent coefficient was strongly nonlinear.

found that the nonlinear behavior could be adequately described by the first three terms of a multiple integral series. For varying loads either the modified super-

position method [5,10] for the mixed time parameters in the kernel functions or the product form [6,10] gave satisfactory results.

N Numbers in brackets identify references listed at the end of this paper.

was cured at a temperature of 250 0 F for two hours.11. All experiments were performed on the same specimen.9971 in. Dead weights were used in the present work. Tension and torsion were produced by dead weights or servo control acting through levers and pulleys. was produced by CPR DiviThe sion..00 in. The average outside diameter was 0.5 per cent. It was a full density solid polyurethane having a specific This material. and Supplied by the Lawrence Radiation Laboratory.25.05928 in.0005 in. The carbon content was 59 per cent. with a maximum deviation of 0. The tensile strain measurement was accomplished by using pairs of rods attached to the upper and lower gage points of the test specimen to transmit the motion of the gage points below the furnace. torsion. The specimen was a tubular type with enlarged threaded ends machined from a solid rod. the Upjohn Company. with a maximum deviaThe material tion of 0. and internal pressure [13]. and the gage length was 4.5 per cent. material was prepared from the prepolymer and combined with toluene diisocyanate.9.12].05 per cent. identified as XR6-77.. hydrogen 6.. The loading and measuring devices are as follows. ash 0. and the balance oxygen. The lower ends of one pair of the rods were attached to the core and the ends .-2- Material and Specimen The material used in the experiments was the same as employed in previous work [8.0003 in. the average wall thickness was 0.. It has been used to investigate creep and stress relaxation behavior under various loading conditions for plastics [4-12] and metals [14-16] and plasticity [17].3 per cent. Experimental Apparatus The testing machine was designed to test tubular specimens under combined tension. nitrogen 7. gravity of 1. The polyester resin was derived from a dicarboxylic acid and a triol. chlorine less than 0.

These served as the heat Temperature in the copper bar was controlled by two Research Incorporated Thermac Model TC5192 temperature controllers operating on the two heaters. The specimen was heated in an electric The split-tube furnace which was maintained at constant temperature by a C. N. During each of the tests. In order to perform creep tests of polyurethane above room temperature. temperature was measured at six positions by means of chromel-alumel thermocouples with fused hot junctions. It was found that the smaller the spacing between the copper bar and One the inner surface of the specimen the better the temperature was controlled. but increased to 20 F on the outside and 70 F on the on the inside of the specimen at 150 0 F. Two small resistor heat- ing cores were attached at both ends of a copper bar and inserted inside the specimen. The temperature variation along the gage length was uniform at 75 0 F and 102 0 F. Two of these thermocouples were cemented on the inner surface of the specimen at positions slightly beyond the gage length./in. thermocouple was embedded in each end of the copper bar. Due to the poor conductivity of polyurethane a second heating source inside the specimen tube was used to achieve a more uniform temperature distribution. sensitivity of the strain measurement was 2 x 10-6 in. Instruments' Sirect Mark III proportional temperature controller. It was observed that the temperature could be maintained constant throughout 0 the testing period within ± 1/2 F. Four thermocouples were attached with spring clips on the outer surface at equal distances within the gage length of the specimen. Rela- tive motion of the upper and lower gage points during the test caused an output of the differential transformer which was directly related to the strain. . the following heating devices were added. S.-3- of the other pair were attached to the coil of a differential transformer. sensing device in the control system.

however. and it it shows up as an alters the gage length of The former was taken into account in the conThe stant rate of temperature rise test. Thus at 160OF the gage length increased about 0. Thermal expansion has two effects on tensile strain: apparent creep strain during temperature changes. c11 = 95 . heat was gradually turned down until it reached room tem- perature. . After unloading at the end of one hour./OF. 48. or appeared to have stabilized before proceeding to the next test.5 hr. 84. respectively./in. 46. Except for one experiment the temperature was maintained at a constant value throughout the testing period. In performing the constant stress creep tests. After reaching the test temperature the specimen was kept at this temperature for one hour before the load was applied. trollers were required.3 per cent.-4- Experimental Procedure and Results The rate of heating the specimen was controlled to about 20 F per minute. The zero strain for each test was taken to be the strain reading at temperature just prior to loading. After that. and 97 followed tests whose and £12 = 90. the strain measuring instrument. recovery stabilized at £11 = -84 . the specimen was loaded for a one-hour period for each test. Pilot tests indicated that a soaking period 0F from 0. £11 = 120 c11 = 63 x 10-6 in.0 hr. gage length was fixed at 75OF and no correction was made for change in gage length -6 with temperature./in. and Tests 69. but the latter was ignored in all tests. and the specimen was left to recover at room temperature until the strain had returned to less than 40 x 10-6 in. at the test temperature (75 to 160 0 F) did not show a significant difference in the creep response under the same loading conditions. The thermal expansion was found to be 35. 91.2 x 10 in. to 5. the Some adjustments of the set points of the con- specimen was left to recover at zero stress at the same test temperature for about one hour./in. £12 = 78 .. £12 = 102 .

In view of the scatter in ations the value n = 0.9.11] and elevated [1] temperature. The A tensile strain versus time curves from these tests are shown in Fig. it was found possible to describe the results of creep strain versus n . coi showed scatter and about the increased from same modest temperature effect observed in torsion [1]. it seems that (1) can be used to describe the The effect of tempera- strain versus time relationship of the present tests also.-5- The test program of the present investigation is shown in Table I. time under constant stress by a power law with a constant exponent o ii 0 + CEl + C+lt n + n() where c and £11 are functions of stress and n is a constant. £ . for the same reason [1]. n in (1) was determined by first obtaining the best fit of As in the companion tests in torn 0 each data set to (1) by a least squares method. the same as in torsion and it more than doubled over the same temperature range. . However. sion it was found that there was considerable scatter in 0 but no definite trend as a function of either stress or temperature. 1 and 2. ell showed a strong effect. Numbers in parentheses identify equations.19 to 0. together with the programmed and measured temperature. In previous work on the same material at room [8.23 from 75OF to 160 0 F.143 n resulting from the individual determin- used in earlier work [8. 1 and 2. £1 /a about 0. From the results shown in Fig. constant stress creep test was also conducted in which the temperature increased at a nearly constant rate. are shown in Fig. 3. then decreased at a nearly constant rate while the stress was removed when the temperature started to decrease. The strain versus time. ture on 0.11] with this material at room temperature and in the companion work in torsion at elevated temperature [1] was chosen as a common value for all stresses and temperatures.

In the analysis of the effect of temperature the data from the first 20 experiments were ignored because of the changes which were occurring. E In the analysis in following sections equaand n in Table I were employed rather Discussion of Results Several experiments were repeated one or more times at intervals during the series of 118 creep tests under tension. Employing these restrictions values of 0 + E and El were re- calculated and are shown in Table I. in Table II. As described in [1] the effect of history was greater in torThe results of the repeated tests in tension are summarized Examination of all the data showed that there were changes which The greatest changes occurred during the Possible causes were generally resulted in increased strains. These repeated experiments were used to indicate how much the creep behavior was changing as a result of the history of stress and temperature on the single specimen employed. There was. however. an extensive period in which changes were tests 39 and 76 and tests 41 and 67 in Table II. l . Compare. torsion and combined tension and torsion.-60 Also consistent with the work on torsion [1] + e was taken to be independent of 0 temperature in view of the much smaller apparent effect of temperature on 0 + ci0 than on l. The experiments employed in the analysis were undoubtedly influenced to some extent by the prior history of stress and temperature. Also some of the last tests were omitted and a few others whose strains were so out of line as to suggest errors of observation. . discussed in [1]. sion than tension. tion (1) with the values of than the experimental data. These history effects were ignored in the analysis. small. early tests and again when the largest stresses were applied. for example.

representing the test data. where H(t) has the value 1 when t . a11 a F For uniaxial stress Eii the following equation can be obtained for the creep strain £ = £ .-7-- Analysis Nonlinear Constitutive Equation A nonlinear constitutive equation in the multiple integral form has been described by several investigators. N may have the following forms H(t) is the Heaviside unit function. see for example [4].T) = R(t.T)ao Determination of Kernel Functions Comparison of (3) kernel functions 2 + N(t. suggests that the R . (3) with (1). The kernel functions (under different isothermal conditions) and time arguments as indicated. constant stress 0 when t < 0 . T is temperature E an arbitrary time prior to t and 6(&) =aa(&)a In this equation only terms from the first three orders have R . M .T)W( 1 dE 1 +ffw~t-ý lt-E.T)d I% 2)d& 1dE W& 00 + 000 J 0 6 . .T)a + M(t. a(t) = aH(t) * .0 Thus under isothermal conditions and constant stress (2) becomes £(t.T)a 3 . . f R(t-& . M . N of (2) depend on temperature Under and been retained. (2) where t is time.

120OF and 140 0 F) was plotted as + In Fig. From (5b) + (aT 1 a where T1=T - = b (Tl) + b (T )a + b (T )a2 2T1 31 (7a) 1w1 750 F . 4. the time-independent and the time-dependent coefficients (5a) and (5b). a2 = a3 = 0 + . and are functions of temperature as suggested by the test results. . (6) The time-dependent part of the strain + from Table I versus temperature Then +/a versus a T was plotted for each stress level as shown in Fig. respectively. cz for each stress as given in Table I shows that o Therefore Eo The data of can be approximated as a linear function of stress.143 . (4) b2 a2 . The reference temperature was chosen as 750 F because . for each of several temperatures (75 shown in Fig. 4. a3 are independent of temperature.195 per cent/ksi.T) N(t. 120OF and 140OF at all stress levels were obtained from the lines of Fig. 5. = a1 + b (T)tn a 2 + b 2(T)tn a3 a + b3(T)tn b.T) where b3 n = 0. and comparing with (1) results in o c =a + Inserting (4) into (3) +aa 2 +a 3 3 2 3 (5a) (5b) b1(T)a + b2(T)a 2 +b(3 Ma From the results in Table I. were determined as follows. . 5 values of for temperatures 100 0 F. E 1000 F. e0 a= al 2 a1 0. 0 F.-8 - R(t. a1 .T) M(t.

the curve for different temperatures can be obtained by a vertical shift of the S+/a versus b2 a curve for b3 T1 = 0 . and F(0) = 0 . Therefore (7a) can be rewritten as Ee(a. bI . b2 . in Fig. sented by the time-dependent part of the strain can be repre- + £ (c.T) = (b1 + 01 T1 + 8 2 T1 )a + b 2 a 2 2 .0045 per cent/ksi(OF) 2 (7d) From (7a). constant temperature as characterized in this section may be summarized as follows by introducing (4) through (8) into (3) . per cent/(ksi)2 b3 0 . (7c). This implies that only b1 depends on tem- perature. and are independent of temperature.00038 per cent/ksi OF (7d) and 82 = 0. (7b).017.-9- T = 75 0 F was the lowest temperature used in this study. b 3 = 0 were determined graphically from versus a . at T bI b with the result 0. per cent/ksi = 0. b3 are independent of temperature. F(TI) = 81T + 8 2 11 2 where 81 = 0.TI)2 c += a1 b + b a + bCy2 + F(T) 2 3 1 F(T ) + (7b) (b is the vertical c /c where shift.004. curves of Thus at different temperatures are parallel straight lines. c+/a versus a In Fig. 5. bI . 5. 6.000. (8) Thus the constitutive relation of creep strain under constant stress. b2 . (7c) £+/ F(T ) was determined from the amount of vertical shift of Thus F(T ) was approximated by versus temper- ature as shown in Fig.

and t in hours.K _TEAP-TL A:BERDEB . Recovery at Constant Temperature The modified superposition method [5] for multi-step load histories can be expressed as follows for tensile loading €il(t) p I = i=0 [f(ailt-ti) -O a i-l1t-ti t > t (10) where for the present experiments f(a. tI = 1 hr.00038T + 0. (11) Thus the recovery at zero a0 is from (10) for iso- stress after a period thermal conditions Cl1(t) = f(aot) .000. sion [1]. RO'VING GROUND.tT) = R(t. Agreement During recovery the contribution of the nonlinear terms is very small. 1 4 3 where c is in per cent. 1 and 2.f(aot-t 1 ) where f has the form of (9) and the ala term vanishes.T)a + M(t. Creep strains were cal- culated from (9) for each stress and temperature and are shown in Fig. at constant stress . (12) The recovery strains following constant stress creep were predicted from (9) by means of (12).195a + [(0.T)a2 + N(t. 2 £ = 0. TE0O11IIOAL LIBRARY BLDG 305 MD.- 10 - Sa al 1 or + [(b 1 + 61T + 11 21 T2T)a + b a2]t. a in ksi. between the theory and test data for tension (Fig.O0a 2 ]t 0 . 7. 7) is not as good as for tor- With the exception of one test the actual strains during recovery were smaller than the strains predicted from the tension tests.017 + 0.0045T2)a 11 (9) + O.T)a 3 which becomes (9) when numerical values are supplied. . The results for some of the tests are shown in Fig.

is the reference temperature and t ture shift factor. Consider a general constitutive relation under constant stresses and temperatures. where J is the creep compliance. . the determination of aT as a function of temperature necessary information for determination of the reduced time. no similar method is available to describe nonlinear viscoelastic behavior under varying temperature during loading. T o aT is the temperais the real time. it is intended in this section to explore a possible way of describing the effect of nonlinear creep behavior under varying temperature by generalizing the constitutive relation obtained through creep tests under different but constant temperatures. the definition given by Morland and Lee [18] for reduced time is ft J f dý 0 (15) and t = 0 corresponds to where t is the current time.t) t/aT(T) . the effects of temperature and time for cer- tain classes of material can be combined into a single parameter through the "time-temperature superposition principle" which may be stated as follows: J(T. the first application of stress.- 11 - Creep under Continuously Varying Temperature Method A In the analysis of the creep behavior of linear viscoelastic material at different but constant temperatures. • is any prior time.C) = J(T o . Aside from the work reported in [2] and [3]. as follows: . For tests involving varying temperature during loading. T will provide the Thus. (13) S= (14) ý is the reduced time.

Therefore. (16) where J = F (T) + F 2(T)a + T3(T)a 2 (16a) (16b) 2 S[F(T) + F(T)a + F(T)a G(t) In (16) it is assumed that the total creep strain can be separated into a timeJ 0a and a time-dependent part Jt a . J t = [a 1 + a 2 a+a 3a2 Rearrange (16b) such that 2 + 8 1 (T1 ) + 8 2 (T1 )a + 83(T)a ]G(t) (17) Rearranging by factoring out the temperature-independent terms yields Jt = [al + a 2a + a 3 a 2 ] f(Tla)G(t) where f(Tl~a ) (18) 2 Fi =_ 81 (TI) + 8 (T 1 )a + 83(T )a2 2 ++ 2 and a1 ' a2 ' a3 are independent of temperature. c= 0o + [a 1 a + a2a2 + a3]G(C) • (20) Equation (20) can be interpreted as a time-temperature superposition principle [20] provided that the shift factor aT depends on both temperature and stress.- 12 S= Jo + JO C j0 a+jt . while T1 = T-To Ind a 1 =$ 2= 3 8l 82 83 are 1 functions of temperature f(T 1 . Equation (9) can be con- independent part sidered as a special case of (16). .O)G(t) time C 0atT=T 0 orT =0 in (18) can be combined into G(c) by defining a reduced a(TlO) aT t ' (19) where a T(T la) is a shift factor depending on both temperature and stress.

be used to redefine the reduced time ft 0 (18) and (19) can C as follows: ds (21) aT[Tl(s). yields satisfactory agreement with experiment. (c) plus the strain that would have occurred had the current stress been applied at time t to the previously unstressed material.t-E. The modified superposition method considers that the strain following a change in stress at time t1 is the sum of: (a) the strain that would have occurred had the stress not been changed. (b) minus the strain that would have t to a previously occurred had the prior stress been applied at the time unstressed material.T] provides a means of account- ing for creep behavior under continuously varying stress and temperature as follows: C(t) Ja(t) + 0 [) +a + ) 3 ]G(•- )} aa( ) dý qSC(c (23) .t T dE(• dE da (22) Introducing (20) into (22) for Q[o(E). it has been found that the concept of a "modified superposition principle.. In integral form the modified superposition principle can be written as follows =tT t f 0 aQa(E).a(t)] where s is time prior to the current time t and t = 0 at the time stress is first applied." proposed by Findley and Lai [5] in summation formh and employed in integral form by Pipkin and Rogers [19]. For describing creep behavior under an arbitrary stress history.13 - For continuously varying temperature and/or stress.

and for varying stress a = a(t) may be described by (23) provided that Tt ds aTLT[T110•]" 0 t a T[LTl. when t < 0 It has the value one when t >0 and zero .0(0)] 0 ST For T = T0 .a) = 1 . and C . f(T. E the strain but constant.Oo] (30) ST and C is not involved. (28) a[Tl'CF(E)] aT 0 aT[TlO(Y)] For varying temperature but constant stress reduces to E(t) a a H(t) 0 .(t)] (27) 'ds • = . Thus (23) J a + [ala + a 2a2+aa 3 ]G()3 (29) where ft 0 ds aT[Tl(s). H(t) is the Heaviside function. a constant.- 14 - where Ct 0 ds (24) Sa aT[TI(s).(23) reduces to the form originally proposed for the modified superposition method as follows: C(t) J 0 (t) + f+{[ala(E) aa + 2 ) + a3 3G(tE)} aa(E) dE (26) 0 since as defined in For T > T0 (18) for T = T 0 .G(t)] = ds (25) aT[Tl(S). • = t .

and the tem- is time in hours. 3.- 15 - Equation (30) was used to predict the creep strain (total strain minus strain due to thermal expansion) for the test program shown in Fig.) Therefore G(C) = 0ITI + T2T] + b1i+ b 2 a n = at = f(Tl.75 0 F = 60t where t a = 4000 psi . + . perature In this program the stress was constant T (t) = T . n j 0 4 ksi (34) b I + b2 and a = the reduced in (30) as .8T' tn (31) 1 +b2a 21 1 f(T. + 02 TI aT[Tl.] For varying temperature in which time c is defined from (34) T + = 60t .a)G(t) = r + ýi T Hence (31) becomes c(t) = aa + (b1 a + b 2 a2 )On 1 (32) implies that (ln•] 1TI + 2T1] = + blI + b2 a 81T 1 + 82 1TI T e e b b b 1 +b2a T21 (32) (33) or 81OTI. as follows Equation (9) may be expressed in the form of (18) eT E(t) = a a + (b1 a + b2a2) 1 Comparing (31) with (18) shows that G(t) =tn . as follows.

- 16 - t 0 f 0 (60s) + e2 (60s) b + b 2(4) b 1 +b2(4 21 n ds . (36) where a = 35.)()T(t)])(•)} ./OF is the observed value of the coefficient of thermal expansion determined by measuring the thermally induced strain at zero load. (35) Equation (35) was evaluated numerically by using Simpson's rule. Employing the modified superposition method the creep behavior under varying load and temperature can be described from (16) and the modified superposition principle (22) e(t) where Jt() 2 (a1 + a 2 a + a 3or'f(Tl(t). For this calculation the temperature change was approximated as steps. the creep strain was calculated by using (35) b2 e. The strains were also evaluated numerically from the T = 60t actual temperature instead of as shown in Fig. 3. The total strain £T(t) = aT1 + C(t) cT(t) . . n given in (9). In (36) it was assumed that the thermal expansion coefficient was inde- pendent of the stress. Method B A more direct means of describing the effect of temperature (both constant temperature and varying temperature) on the time-dependent part of the strains (16b) F1 (T) is as follows. in (33) with values of is a1 Then b. 3. The values of e(t) and T(t) predicted from (33. . 1 2 .a)G(t-&) aJ(t) + 0 as follows do(. F2(T) and Consider that temperature only affects the coefficients F3(T) of (16b).2 x 10-6 in.)(37) 3{J[(t_)./in.35) were calculated and are shown in Fig.

3 are the strains calculated from (38) the actual temperature instead of Figure 3 shows that (38) data than (33-36) (Method A). and (39) using Also shown in Fig. Recovery under Continuously Varying Temperature After the rising temperature reached the programmed value of 1350F at one hour the temperature was caused to decrease gradually. e2 n given in (9) are also plotted on Fig. 3.. farther below the test data for the increasing temperature test. Fig. the load was re- The temperature and strain during this period of recovery at . and T . integration of (37) by means of the Heaviside function and making use of (9) yielded the following prediction of strains during this part of the experiment: E(t) = a1 a + {[b 1 + e1 (60t) + e 2 (60t) 2]o + b 2a • (38) The corresponding total strain is CT(t) = i(60t) + E(t) The results calculated from (38. bI .17 - For the present material the temperature and stress function in given by the function in square brackets in (9).75 = T1 = 60t . moved at one hour. but they lie below and nearly parallel to the increasing temperature test instead of above as would be expected. 3 a = 4000 psi . (39) a1 . This fact may result from differences in temperature disThe prediction of Method A lies still tribution between the two types of tests. Jt is In the test program shown in Since the stress was constant. as would be expected from the nature of Method B. and (39) (Method B) are closer to the experimental Also shown by square points in Fig. el. Also.39) . using the values of b2 . T1 = 60t . 3 are the strains at the indicated time from creep tests at a constant stress of 4000 psi and the constant temperature indicated. stress but increasing temperature it Comparing these with the data for constant is observed that the four points nearly coin- cide with Method B.

the temperature variation was approximated by a series of appropriate steps of constant temperature. 8. where a0 4 ksi . The results for the rising portion T1 = 60t as shown in Fig. 3. at decreasing temperature are shown in Fig. The strain corresponding to the recovery at decreasing temperature was computed for both Methods A and B by using the modified superposition principle to account for the abrupt removal of the stress at one hour. G() = and G(C .C1.C2) were evaluated by Simpson's rule. minus (b) the creep at the same stress applied at one hour with the decreasing temperature that actually occurred from one hour on. stant rate of temperature rise shown in Fig. The calculated values of strain by Methods A and B for the recovers. and (41. by the step method were nearly the same as for Computation of the recovery strain by Method A u3ing the modified superposition principle was as follows: E(t) = (bla + b2 a )[G() -G( - (40) t > 1 hr.18 - decreasing temperature were as shown in Fig. 8. especially during falling temperature. the actual history was considered to be equivalent to (a) creep at constant stress through the entire period but at the actual variation in temperature. This method accounted for the small variations from con3. Since the tempera- ture variation did not occur at a constant rate.= n and where t =0 was taken to be the time at which the first stress was applied. That is.. Figure 8 shows that in the early .

The time-independent part of the strain was found to be independent of The time-dependent part of temperature and linearly proportional to the stress. (1) Results of the constant stress creep tests at different temperatures indicate that creep strains are separable into time-independent and time-dependent parts. Reed for their care in performing the experiments and to Mrs. One method employed a reduced time involving a shift factor which was a function of both stress and temperature history. It was found that the time-dependent strain was a strongly nonlinear function of stress and temperature. the shape is predicted more accurately by Method B. Tracy and P. R. Conclusions The following conclusions may be drawn from the experiments and analyses presented in this paper. The authors are greatly indebted to Messrs. . The latter method agrees somewhat better than It the former with the trend of experiments on varying temperature and recovery. A. F. the strain was represented by a power law of time with a constant exponent independent of temperature. Gingrich for typing the manuscript. M. is also the easier to employ. modified superposition principle. However.- 19 - stage of recovery the test data lie between predictions of the two methods but closest to Method A. The other considered the strain to be a Both employ the function only of the current temperature not its past history. (2) Two methods were considered to account for varying temperature by using the constitutive relation derived from different constant temperatures. Acknowledgments This investigation was performed under the auspices of the Advanced Pesearch Projects Agency of the Defense Department [Materials Science Research Program (SD-86)]. J.

W.. Y." accepted for publication in the March 1971 issue of Transactions. W. W.. Journal of Basic Engineering. and Findley. 12. Findley.. W.. 68B." Transactions of the Society of Rheology. E. and Lai.. 2. Bernstein. 361-380.. L. Nonlinear Creep of Polyurethane Predicted from Constant Stress Creep by Three Integral Representations. G. K. B... 12. pp. 22-27. Series E. Part 2.. "Prediction of Uniaxial Stress Relaxation from Creep of Nonlinear Viscoelastic Material." Transactions of the Society of Rheology. pp.. Findley. American Society of Mechanical Engineers. 1965. 10. Lai. W. 4. W. and Findley. Journal of Applied Mechanics. K. T. 6. Y. N. Vol.. Part 3." J. N. Issue 2. Nolte. 36. W. Lai. Lai. K.." Purdue University Report AA & ES. G. N. W.B. 8.20 - References 1. J. ARPA SD-86. of Research. N. "Combined Stress Creep Experiments on Rigid Polyurethane Foam in the Nonlinear Region with Application to Multiple Integral and Modified Superposition Theory. Nolte. Vol. Issue 2. pp. Vol. J. S.. 1. . W. N. American Society of Mechanical Engineers. p. Vol. 1968. "Multiple Step. "Thermodynamics of Perfect Elastic Fluids." Transactions. N. 299-327. Series D. J. 3." ASTM Journal of Materials. E-74 on Materials Research Program. and Findley. No. 243-257. 916... and Onaran. 103-113. "Experimental Determination of Some Kernel Functions in the Multiple Integral Method for Nonlinear Creep of Polyvinyl Chloride. J. Vol. Onaran. "A Modified Superposition Principle Applied to Creep of Nonlinear Viscoelastic Material under Abrupt Changes in State of Combined Stress. S. Lianis. "Behavior of Nonlinear Viscoelastic Material under Simultaneous Stress Relaxation in Tension and Creep in Torsion." Transactions of the Society of Rheology (in press)." Technical Report No. K. N. N. C.. March 1969. 65-1. "Combined Stress Creep Experiments on a Nonlinear Viscoelastic Material to Determine the Kernel Functions for a Multiple Integral Representation of Creep. "Product Form of Kernel Functions for Nonlinear Viscoelasticity of PVC Plastic under Constant Rate Stressing. 9. pp. G. Findley. 1967. No. Vol. Y. K. 12. N. Kearsly. pp. 1968. "Relationship between the Creep of Solid and Foam Polyurethane Resulting from Combined Stresses. ASME. and Stanley.S.. Onaran... 5. A. 11. pp. 1965. N.. 9.. S. Vol. pp. W. and Findley. 105114. 4. Y. 11... 1. 7. A. and Zapas. and Findley. S.. "Elevated Temperature Creep of Polyurethane under Nonlinear Torsional Stress with Step Changes in Torque. Journal of Applied Mechanics. 217-242." Transactions." Transactions of the Society of Rheology. and Findley." Transactions of the Society of Rheology.. "Integral Constitutive Equation of Nonlinear Thermo-Viscoelasticity. December 1968. and Findley. 1964. June 1970. March 1970. N.. 92.

1960. and Gjelsvik. "An Experimental Study of Subsequent Yield Surfaces: Corners. Journal of Applied Mechanics (in press). 62. S. G." Transactions. p.. J. Aug. "Stress Analysis for Linear Viscoelastic Materials with Temperature Variation.. W. D. Contract No. Findley. "A Nonlinear Integral Representation for Viscoelastic Behavior. "Concerning a Creep Surface for an Aluminum Alloy. Vol..D. and Lee. W. L. 72.. Brown University. National Congress of Applied Mechanics. "An Experimental and Theoretical Analysis of the Inelastic Behavior of a Structural Aluminum Alloy at Elevated Temperature. American Society of Mechanical Engineers. "Mechanicai Properties of Substances of High Molecular Weight Dispersion in Concentrated Polymer Solutions and its Dependence on Temperature and Concentration. 19. and Rogers. Pipkin." Journal of Mechanics and Physics of Solids." Ph. and Findley. E. "Short-Time Biaxial Creep of an Aluminum Alloy with Abrupt Changes of Temperature and State of Stress. Ferry. 16. pp. "A Biaxial Testing Machine for Plasticity.. G.- 21 - 13. C. 233-263. Normality. pp. P.. and Nolte. Findley.. N. and Findley." Transactions of the Society of Rheology.. Am. N. K. Bauschinger and Allied Effects. Soc. 893-907. S." Technical Report No. K. or Relaxation under Variable Principal-Stress Ratios. MCG . 1962. 17. Blass.. 1103-1118. W. J. pp.2. J. Bertsch. 16. H.. pp. 18. N. Lai. A. American Society for Testing and Materials. Chem. 59-72." 4th U.. Dissertation. Brown. 20. Office of Naval Research.. Y.. W. Morland. 14." J. 7. Vol." Proceedings. 1962. December 1969. T.. W.. N. IV. G.. 1950. 1968. J. N00014-A-01910003. A. Creep. 15. 1970. 3746. Vol.

0 2.118 0.216 0.780 " " " " " 0.0 1.0 2.125 0.0 4.0 2.298 0.133 0.0 1.195 " " 0.143 + + 0 % 38 40 51 52 53 54 63 65 41 46 72 48 77 47 50 84 92 91 78 81 97 1.092 0.0 2.0 3.0 1.100 0. n = 0.0 3.028 0.068 0.328 .0 4.185 0.0 4.017 0.0 1.0 75 100 130 140 150 160 75 102 115 130 135 140 75 130 140 75 102 115 126 135 145 0.390 " " 0.083 0.0 2.081 0.052 0.0 4.053 0. OF = c0 + tn.0 4.0 4.0 3.268 0.585 " " 0.0 1.TABLE I Tension Creep Experiments TEST TENSILE STRESS. ksi TEMPERATURE.267 0.0 2.142 0.062 0.178 0.

6439 0. psi TEMPERATURE.6672+ 1. OF 75 "l i STRAIN AT 1 hr. 2.5894 75 " 115 " 0.9957+ + Used in Fig.4368 39 63 76 2000 if " 41 67 49 " 115 0.TABLE II Repeated Experiments in Tension TEST STRESS.4925 " 150 0.4341 0.4419+ 0.4902+ 0. 1.0843 0. . Per Cent 0..6368 114 2 77 87 91 " 3000 " 4000 " 0.

96 C 0)0 =0.24 s tesa ifrn !lze osattmeaue - .92 CL 0/~ o 0 Experiment rheory 75L 0 0.6 Time hr 0.4 0.2 0.08 9 0 /45 L 1. Creep curve fo7 n 0.00 0.10 1.Iio 2 1..10 2 Fig.

1 0 Experiment S0.66 75F 09.74 77 0.86 1 0.78 00 *4 0. Fig.40O 2.5 C ee 2 ur e f r 2 d 3k si a stress t di fe e n c o s a t t m e a u e .82 LT z_ 3_ S _ _ _ _ _ _ _ 0.7 140L 0.0.62 0.

0 Time .05 30 11 1.0 _ _.05-d' 0 0 x) 7 30" EI- x 0. * a a 70 '-=4ksi Actual Temp.95 020 0 0.8 1. 1.I__5_ _ 40 . Creep at a constant stress of 4 ksi and increasing temperature . r/ 75OF + 60t o Total Strain I .2 0...36.1.6 0. IA Fig.Ti-60t I /0 50 1. 39.45 Test 88.T/=60t MethodA Actual T/ 0) 0)0 0.35 o x x x o Creep Strain _ _ A S9 1 Constant Temp.hr. 3.25 -Method 8 ActualT7 1 Me/hodA Eq.4 0. L) 1. Creep Tests Method 8 Eq.

0.24 // x S0.36 0. Coefficient of time-dependent strain versus temperature for several stresses .2 0.28 /x x x 0.12 0. Fig. 4.08 xi 0' .20 / 0.000 0.04 0 60 80 100 120 'F 140 160 Temperature.

04 + Ib o. 7a) . ksi Fig.08 0.0. 5.06 4-- 100O ° 0.02 0 0 2 3 4 5 o-. £÷/d versus d (see Eq.o 0.

6. Vertical shift of /dl with temperature (see Eq.' > 0 0 20 40 T= 60 T-75 0 F 80 Temperature.000..04 0 4-0 !t 0.06 a) I " I2 0. + 0.000387. Fig.02 . 0045Tr " 0."n 0. 7d) / .

2ksi.. Test 52. 135 . 3ksi.4 Time .04 •• L. 4ksi.10 Test5o.04 ca x . 06 m". 4ksio.2 .12 I Test 81. 140°F Test 72. Iksi. 160°F Test 53. 150°F r *0 CA.0 Fig.126°F e Theory 0F . 140°F -. Iksi. . 2ksi.hr. 140 Test 48. 2ksi.0 0* 0 . Iksi. 115 OF Test 54. 7. 135 0 F Test 78.8 1. Recovery at zero stress and at the indicated temperature following creep for one hour at the same temperature and at the stress shown __ _ ____ ___ .08 _ 0 F C Test 41.6 .

2 1. Recovery at zero stress during decresing temperature following creep at a constant stress of 4 ksi during increasing temperature (see Fig.03 9 0.01 0 1 1. hr.0 Time.0. 2.5 x x 140 0.2 2. 3) .4 x130 0.3 0'0 x /'120 0I0 ~0.4 1.6 1.6 Fig.8 2.4 2.05 C) CL 70t1oo~-7 0 0. 8.02 -40 0.