419 The Elastic Properties of Steel at High Temperatures .
By Bertram Hopkinson , M.A. , Professor of Mechanism and Applied Mechanics in the University of Cambridge , and F. Rogers , B.A. ( Cantab .
) , M.Sc .
, M.Eng .
( Communicated by Professor Ewing , F.R.S. Received April 13 , \#151 ; Received in Revised Form June 5 , \#151 ; Read June 8 , 1905 .
) Hitherto , investigations into the elastic properties of metals have been confined to comparatively low temperatures .
Cray , Dunlop , and Blithe have measured the modulus of rigidity and Young 's modulus for wires up to temperatures of 100 ' C. , and found that both these quantities decrease as the temperature rises.* Martens determined the influence of heat on the strength of iron up to temperatures of 600 ' C. , but his experiments were the ordinary tensile tests carried to rupture , and though he also found a substantial diminution of Young 's modulus with rise of temperature , he did not go into the point fully , being mainly concerned with breaking stress and elongation . !
In the experiments here described the elastic properties of steel and iron have been investigated at higher temperatures , ranging up to 800 ' C. , and for stresses greatly below that required to rupture the material .
We have found that as the temperature rises the stress-strain relations undergo a remarkable change , which may best be expressed by saying that what is variously called the " time-effect , " or " elastische nachwirkung , " or " creeping , " increases greatly with the temperature .
Steel , at high temperatures , behaves like indiarubber or glass ; if it is stressed for a time , and the stress removed , it does not at once recover , but after the immediate elastic recovery there is a slow contraction perceptible for many minutes .
Such " creeping " can be detected at ordinary temperatures , but at a red heat it attains a different order of magnitude , becoming ( in its total amount ) a substantial fraction of the whole deformation .
The test-piece was 4 inches long , about 02 inch diameter , and had enlarged ends which were screwed into two steel bars each 1-J- inch diameter and 10 inches long .
The whole was set up in a vertical electric resistance furnace , wound with three coils of nickel wire .
The currents in these coils could be separately controlled , and in this way the temperature along the test-piece could be made very approximately uniform .
The temperatures were measured by three thermo couples , placed one at each end and one in the * ' Proc. Inst. C. E. , ' vol. 104 , p. 209 ( 1891 ) .
t ' Roy .
Soc. Proc. , ' vol. 67 , p. 180 ( Oct. , 1900 ) .
VOL. LXXVI.\#151 ; A. 2 G Prof. B. Hopkinson and Mr. F. Rogers .
[ June 5 , middle of the test-piece .
Fig. 1 shows the apparatus with one side of the furnace removed .
Changes in length between the points A , B could be 1905 .
] Elastic Properties of Steel at High Temperatures .
421 measured correctly to 1/ 50000 of an inch , by means of an extensometer of Professor Ewing 's pattern .
The furnace was supported separately , and the test-bar , with the attached extensometer , was hung free within it .
Tension up to 112 lbs. ( 1^ tons per square inch ) could be very rapidly applied or removed by means of a foot lever at the lower end .
The interior of the furnace was closed from the atmosphere by means of mercury locks , and the test-piece was kept surrounded by an atmosphere of nitrogen so as to avoid oxidation .
It will be seen that the extension observed included the elongation of the end-pieces , as well as that of the test-piece .
The area of the latter being 1/ 30 of that of the ends , and its length one-fifth , it appears that of the total extension 87 per cent , is contributed by the test-piece and the remainder by the end-pieces , if the elastic properties of the two are the same .
At low temperatures this is approximately the case , but at higher temperatures the average temperature of the ends is less than that of the test-piece , and they , therefore , contribute a less proportion to the total extension .
In stating the results in this paper , the total extension is alone referred to , and it is stated in extensometer divisions , each of which is 1/ 5000 of an inch , or 1/ 20000 of the length of the test-piece .
It is probable that at high temperatures over 90 per cent , of this extension should be credited to the test-piece .
Two materials were tested , one being steel containing about 0'5 per cent , of carbon , and the other Low Moor Iron .
Eig .
2 shows the result of a series of tests carried out on a steel bar at 750 ' C. The bar was at no time heated much beyond that temperature .
It was loaded with 85 lbs. for one minute , then unloaded for two minutes and so on , and the curve shows the resulting changes of length in terms of the time .
It will be seen that even at this low stress ( about 1\#163 ; tons per square inch ) the metal flows fairly rapidly , and that the overstraining has a considerable hardening effect , as shown by the diminishing amount of the permanent set produced by successive loadings .
We found that this hardening disappeared with rest ; that is , if the bar were left unstressed at 750 ' C. for a couple of hours after having been hardened by successive loadings , it was restored to its original soft state .
With a slightly less load ( about 79 lbs. ) the flow of metal was very much slower , the permanent set produced by load applied for one minute amounting to only about 05 extensometer division , against 1*5 divisions for the load of 85 lbs. In respect of all the features hitherto mentioned the properties of the material differ only quantitatively from those of the cold bar .
Fig. 2 might , but for one remarkable difference , apply to a cold bar stressed to its yield-point .
The difference lies in the behaviour of the bar after the removal of the load .
The cold bar does not contract appreciably ; there is the instan Prof. B. Hopkinson and Mr. F. Bogers .
[ June 5 , taneous elastic contraction , then it stops.* The hot bar , on the other hand goes on shortening for two minutes or more after the load is off , as shown by the dotted line on the diagram , and the total amount of such shortening amounts to roughly one-third of the instantaneous contraction , or one-quarter Minubes Fig. 2.\#151 ; Temperature of bar , 750 ' .
Load , 85 lbs. 1 division of extensometer = = 2oioo Part 'f length of test-piece .
During the times covered by the dotted lines the bar was unstressed .
of the total contraction , or one-fifth of the average total extension after hardening has taken effect .
The iron bar behaved in much the same way , but the metal flowed at a lower stress .
There was considerable flow with a stress of but half a ton per square inch .
The shortening after removal of load was also perceptible at that stress .
* Professor Fwing , * Roy .
Soc. Proc. , ' vol. 58 , p. 123 , found a certain amount of creeping after the removal of the load from a cold bar which had previously been stressed beyond its yield-point ; but the effect was extremely small , the total creep never amounting to more than one-seventieth of the total extension of the bar when loaded .
In a bai which had not previously been overstrained no such effect was observed .
1905 .
] Elastic Properties of Steel at High Temperatures .
423 At 600 ' C. both bars exhibited greater tenacity .
A load of 112 lbs. ( 1*6 tons per square inch ) applied to the steel bar for one minute produced an immediate extension of 3'8 divisions , followed by a slow drawing out , which amounted in one minute to about 09 division .
On removal of load there was an immediate shortening of 3-8 divisions , followed by a slow contraction amounting in two minutes to 07 division .
The permanent extension produced was very small , if , indeed , there was any at all .
The iron bar behaved in a similar wa}^ , but as at 750 ' C. it yielded appreciably at a stress which was not sufficient to permanently deform the steel bar .
The experiment on the steel bar at 600 ' shows pretty conclusively that this slow recovery after release from stress is not solely , or even mainly , dependent on overstrain .
It seems to exist to a large amount with stresses which leave practically no permanent effect ; the strain develops slowly under application of stress and disappears slowly after it is removed .
This phenomenon is , of course , analogous to residual charge in glass and other dielectrics ; the stress corresponding to the electric force , and the strain to the electric displacement .
Whether the law of linear superposition of the effects of stresses\#151 ; closely followed in the electrical analogy\#151 ; is true for hot steel or iron , is an interesting question which our apparatus was hardly sufficiently delicate to answer .
The magnitude of this effect in steel may best be gauged by comparing it with other cases of the same kind , e.g. , with the slow recovery of a glass fibre after twisting ; if such a fibre be twisted through a considerable angle for several hours , it will recover all but one-fiftieth of the twist within two or three seconds of the removal of the stress.* The remaining slow " creep , " amounting to one-fiftieth of the whole deformation , corresponds to the slow return of the steel .
In indiarubber , under certain circumstances , 10 per cent , of the strain disappears in time after the removal of the stress.f But in steel , at 600 ' C. , the proportion is about 15 per cent. The apparatus used was not entirely satisfactory , having been designed for the purpose of measuring larger strains than have been dealt with in this paper .
The principal difficulty lay in the slow variations of temperature in the bar and end pieces , which could not be completely controlled , and which produced changes of length masking to some extent the changes due to stress , especially when the latter were spread over considerable times .
We cannot do more at this stage , therefore , than assert the existence of a large time-lag between the stress and the strain in steel and iron at temperatures of 600 ' C. and over ; and the figures which we have given must be taken as indications of its order of magnitude only .
* Dr. J. Hopkinson , ' Original Papers , ' vol. 2 , p. 350 .
t Phillips , 'Phil .
Mag. , ' April , 1905 , p. 513 .
424 Prof. B. Hopkinson and Mr. F. Rogers .
[ June 5 , One effect of such a time-lag will be to cause dissipation of energy if the material be subjected to alternating stress , for it will lead to a difference of phase between the stress and the strain ; and the amount of the dissipation will depend on the period of the oscillations .
Gray , Dunlop and Blythe found an increase in the rate of decay of the torsional oscillations of an iron wire as its temperature was increased to 100 ' C. On the other hand , Horton* has found a decrease in the rate of decay under circumstances which were apparently the same , except that the period of the oscillations was very much less than in Gray 's experiments .
These results might be reconciled and explained by the existence of such a time-lag as we have observed at higher temperatures .
With our apparatus we could detect no time-lag at temperatures lower than 400 ' C. ; but it is quite possible that it exists to the small extent necessary to account for the decay of oscillations .
Young 's Modulus .
Another effect of " creeping , " such as we have observed , is to make the determination of Young 's modulus a matter of some uncertainty .
Thus the extension of the bar at 600 ' C. produced by a given load varies 15 per cent , or more , according to the time of application of the load .
When , however , the load is applied for a very short time , say of the order of one or two seconds , the strain produced seems to approach to a definite limiting value , wdiich is the instantaneous extension or contraction of the bar observed in our experiments when the load is applied or removed .
It seems reasonable to define Young 's modulus for a metal in this state , as the stress divided by this limiting instantaneous strain .
It is then independent of the manner of loading , and is a definite physical constant ; otherwise not .
We have shown in Fig. 3 the relation between Young 's modulus , so defined , and the temperature .
The ordinate is the reciprocal of the instantaneous extension produced by the load of 112 lbs. Owing to the effect of the ends , the reciprocal of the extension is not quite proportional to Young 's modulus , their ratio being somewhat greater at high temperatures than at low , as already explained .
With regard to this Fig. it should be noted that the error in the determinations on the cold bar and at 400 ' C. is probably not more than about 2 pei cent. At higher temperatures the error is greater , as , owing to the rapid drawing out of the bar , it was difficult to be sure of the instantaneous extension .
It is , however , fairly certain that the ratio of Young s modulus in the cold bar and at 750 ' C. ( as shown by the point Dr ) is not more than 10 per cent , in error .
The points were observed in alphabetical order , and it was * ' Phil. Trans. , ' A , vol. 204 , p. 1 .
1905 .
] Elastic Properties of Steel at High Temperatures .
425 Fig. 3.\#151 ; Variation of Young 's modulus with temperature .
Load , 112 lbs. found that , in spite of our efforts to secure a neutral atmosphere , the bar had scaled somewhat after heating to 750 ' C. This accounts for the larger extension shown by the point E , which was taken in the cold , after heating .
After taking this point the bar was taken out , cleaned and gauged , when its area was found to be reduced by about 6 per cent. Allowing for the reduced area , the points E and A are in good agreement ; but , of course , there is some little uncertainty from this cause as to the position of the point D. If the full reduction of 6 per cent , in the area be allowed for , the corrected position is at D ' , and this is probably not far from the truth .
In the iron bar the change of Young 's modulus with temperature was of the same character but greater .
The value in the cold being taken as unity , that at 600 ' C. was about 0-6 , while at 750 ' it was about 0'5 .

