373 On Speed Effect and Recovery in Slow-Speed Alternating Stress Tests .
By W. Mason , D.Sc .
( Communicated by Prof. B. Hopkinson , F.R.S. Received December 14 , 1915 .
) ( From the Engineering Laboratory , University of Liverpool .
) ( Abstract .
) In a series of repeated torsion tests of steel the writer noticed that when the steel had become fatigued the angular strain became considerably greater when the frequency of the cycles of stress was decreased , and vice versA although the maximum and minimum stresses of the cycle were unchanged The frequencies used were between 2 and 200 cycles per minute .
This effect was absent when the cyclic strains were purely elastic , while it became increasingly apparent as non-elastic strain developed .
Discussion of Results .
The results may be stated thus:\#151 ; If a piece of the steel used in the experiments has undergone a considerable number of cycles of torsional stress at 2 cycles per minute , then if the speed is changed to 200 per minute , the cyclic non-elastic strain immediately decreases about 50 per cent. ; if the piece has endured a large number of cycles at 200 per minute , then on changing to 2 per minute the cyclic non-elastic strain is increased by 50 to 75 per cent. The reduction of range of non-elastic strain found for the above change of speed appears to be the counterpart , for slow speeds , of the comparatively small range of extra-elastic strain discovered by Prof. B. Hopkinson* for very high frequencies .
After a change of speed from 200 to 2 cycles per minute , or vice versed , the condition of the steel is not a stable one .
After the former change there is a shrinkage in the extra-elastic strain , and after the latter change the strain increases at a more rapid rate than normally would be the case .
The shrinkage may be regarded as recovery\#151 ; though not necessarily as recovery of elasticity\#151 ; because , if the speed is put back to 200 per minute after an intermediate run at 2 per minute , the new range of the strain is found to be equal , approximately , to the original range at 200 less the shrinkage during the intermediate run .
The development * The range of extra-elastic strain for Hopkinson 's very high speed tests appears to have been about 1/ 15 of the range of the elastic strain ( \#163 ; Roy .
Soc. Proc./ A , vol. 86 ( 1912 ) ) .
VOL. XCII.\#151 ; A. 2 F 374 Dr. W. Mason .
On Speed Effect and Recovery in of cyclic strain appears to be definitely set back by the " recovery , " though it is by no means certain that the progress of fatigue , whatever that may be , has received a corresponding check .
The rate of variation of this recovery , as illustrated by the diagram , is rather significant .
The rate is very big just after the change from 200 to 2 per minute , and subsequently diminishes quickly .
It is not easy to believe that recovery of elasticity can take place rapidly at a time when the cyclic strain has been largely augmented and the repetitions of the stress are continued .
And it is still less easy to reconcile with the supposition that such contractions of range of strain are to be regarded as recovery of elasticity the fact that a period of rest , interpolated between two runs at 200 per minute , results in less contraction of range of strain than a change of speed to 2 per minute , a subsequent run at this speed of the same length of time as the period of rest , and a change back to 200 per minute .
It seems more reasonable to suppose that the shrinkage of range of strain is due mainly , if not wholly , to hardening or loss of mobility in the material .
Considering the form of the stress-strain curve as found by Dr. L. Bairstow* for a cycle of stress ( tension and compression ) , there are curved lines corresponding to imposition of stress , and , what is rather remarkable , straight lines inclined at a gradient equal to the modulus of elasticity for relief of stress .
There would seem to be some change in the nature of the resistance to stress between imposition and relief of stress in a cycle , such as would result from hardening ; otherwise the lines for relief of stress would not be straight lines whose inclination is determined by Young 's modulus .
Under this view , the main factors concerned in the alteration of non-elastic strain , both in a cycle and in a series of cycles of stress , would be variation in mobility in the steel and speed of cycle .
It seems to the author that an explanation of the phenomenon of strain during alternating stress may be obtained on the basis of Prof. G. T. Beilby'sf hypothesis of formation of mobile material during slipping on crystalline gliding planes , and of the tendency to rapid hardening of such mobile material .
If it be assumed also that when crystalline slipping has once begun the rate of formation of mobile material depends largely on the rate of such slipping , and that loss of mobility ( or hardening ) requires a time not too small in comparison with the intervals of time afforded in a cycle for hardening , then an explanation of the foregoing experimental facts may be obtained** Thus , suppose a test proceeding at 200 cycles per minute , that * 'Phil .
Trans.,5 A , vol. 210 .
t 'Institute of Metals Journ.,5 vol. 6 ( 1911 ) .
Slon'-Speed Alternating Stress Tests .
* I S * / TOJOO-ZO pUCOJ U9LU/ C 9(/ s 9 \#163 ; 10-^6* SJ/ UJ/ / 9S91/ J Utl/ J/ M pi jjDjS ' / C/ goqojd j/ oojj 7'V Js\gt ; d S9/ oXb \#163 ; // p99Gfr JO T S/ JJZXJaddo JOJJ/ LU jo 9/ cog C/ O S9-/ J9LL/ /JC/ 9Q __ ' C/ /C-/ /g JO odl/ D^J / cyoj 2 f 2 s ts c ?
\gt ; 1$ $ e : I CO V. o 2*b 5 a* f. If S* 10 M til lS o\gt ; ^ v 5 0 b c : \lt ; ; n x2 fc\gt ; S I* !
tl s* I \#171 ; I 376 Speed Effect in Slow-Speed Alternating Stress Tests .
a considerable range of extra-elastic strain has appeared , and that the assumed opposing processes of production of mobility and of hardening are approximately balanced .
Let the speed be changed to 2 per minute ; there is now 100 times the interval for crystalline sliding to operate , and the range of strain is much increased , but the rate of production of mobile material will be less because the average rate of slipping is only something like one-sixtieth to one-eightieth of the previous rate , while the intervals afforded per cycle for hardening are 100 times as great .
The increased range of strain will thus tend to decrease until an approximate balance between production of mobile material and the hardening thereof is attained in a cycle .
Suppose now that the speed is increased to 200 .
The range of stress immediately becomes less , because of the diminished time for slipping ; the rate of slipping , and therefore the production of mobile material , becomes more , while the time for hardening per cycle is less , the resultant being an increase of range of strain until again an approximate balance per cycle is approached .
The cost of the machine and apparatus used in the experiments has been defrayed by grants from the Eoyal Society , the Institution of Mechanical Engineers , and the University of Liverpool .
The author is indebted also to Prof. W. H. Watkinson for facilities for carrying on the work in the engineering laboratories of the University of Liverpool .

