TA

- 8

YW02449

B 3 IDE MOD

EXCHANGE

<* McGILL UNIVERSITY

PAPERS FROM THE DEPARTMENT

ENGINEERING.

I£ TV )

y

— ON THE EFFECT OF TIME AND TEMPERATURE ON THE STRENGTH OF STEEL AND IRON.

BY

ERNEST G. COKER, D.Sc.

[Reprinted from the Trans, of the Canadian Institute of Civil Engineers, Vol. xv, Part I, Paper No. 158.]

MONTREAL, 1902.

W^r*X

I,1

ON THE EFFECT OF TIME AND TEMPERATURE ON THE STRENGTH OF STEEL AND IRON.

BY ERNEST GEO. COKE# B.A. (CANTAB.), D.Sc. Assistant Prof, of Civil Engineering, McGill University, Montreal.

INTRODUCTION

The behaviour of iron and steel under simple stresses of various kinds has long been a subject of investigation by the Engineer. and very complete data are available for use in designing structures in metal.

When iron or steel is subjected to tensional stress sufficient to cause a permanent stretch in the material, it is well known that the ductility of the material is reduced, and that ultimately the elastic limit rises.

As a first consequence, however, the elastic limit is first of all lowered, in some cases to zero, but if the bar be left to itself the elastic limit ultimately rises to a higher value than before.

In the laslt few years the phenomenon has received considerable at- tention at the hands of Bauschinger*, Ewing| and MuirJ.

The present paper relates to a few experiments, which have been made in the Testing Laboratory, McGill University, following the lines indicated by the authorities mentioned above.

I. MEASUREMENT OF SMALL STRAINS.

The measurement of the change in length of a bar, by such a small quantity as 1/50,000 inches has been rendered comparatively

* Unwin's Testing of materials of construction.

t Ewing "On the measurement of small strains in the Testing of materials and Struc tures" Proc. R. S 1895. j Muir "On the recovery of Iron from overstrain." Phil Trans 1899.

easy by aid of extensioaneters devised by Bauschinger, Unwin and Ewing. The instrument mainly used in these experiments was one constructed after Professor Swing's design, and has several novel features. It is, moreover, especially adapted for work where several settings o-f the instrument are necessary.

II. EWING EXTENSOMETER.

The instrument is shown in Fig 1, mounted upoin a test piece, and consists of two clips, A and B, fixed upon the bar at a standard dis- tance apart by set screws C, D. Upon the upper clip A, are two pins E, P., upon which latter the frame G- is mounted. The frame carries a reading misoros'cope H, and a micrometer screw I; this latter gears with a conical hole in the lower clip B, and is kept in engagement by a spring. The reading microscope is sighted upom a thick wire mounted horizontally in a plate, J and illuminated from* behind by a small mirror.

Hg I

The clips are made of gun-metal, and are formed so as to- give slight- ly when the set screws engage the test piece, and thereby if a slight diminution of section occurs in the bar, while the test is proceeding, the screws do not become slack, and one sourse of error is thereby avoided.

The clips A and B are each centred upon the bar, and each has one degree of freedom with respect to it, viz., that of rotation, about their respective set-screws, and by causing the micrometer screw I to gear with the lower clip, both degrees of freedom are removed and the apparatus becomes a rigid whole.

In setting a flat tension barf in the jaws of a testing machine, it is difficult to avoid giving it a slight twist, but by pivoting the frame G upon the upper clip in a transverse direction, a slight twist will not cause error in the readings.

The micrometer screw and observation wire are set equidistant from the set screws D upon the clip B, so that thei wire can be moved up or down by turning the micrometer screw, and this move- ment is measured by reference to a glass scale in thei eye piece of the reading microscope.

In this manner the scale is calibrated by reference to the: micro- meter screw, which latter has 50 threads to the inch, and the scale is so arranged that a movement of the wire through 1-50 inch cor- responds to a reading of 50 scale divisions. In actual woirking since the micrometer screw engages the clip B, any extension between the set screws will cause the relative movement between the reading microscope and the wire to be doubled, so that the value of one scale division =1-2 x 1-00 x 1-50 — 1-5,000 inch, and as 1-10 division can be estimated, a reading of 1-50,000 on an 8 inch length can be measured.

The chief advantage of the instrument is that it can. be calibrated in position, that the reading is a mean of two sides of the bar, and that it can be quickly and accurately set on the specimen.

In order to set the clip accurately upon the specimen a distance piece is provided, having V shaped recesses to receive corresponding studs on the clips, and clamps to hold the clips in place, until these latter are set on the bar.

The apparatus for applying tensional stress was a single lever Wicks-tead testing machine, which was capable of exerting a pull of 200,000 pounds. As the stresses used were comparatively small, the weigh lever was) mounted to cause the jockey weight to travel the whole length of the weighbeam for a range of 50,000 pounds, and so render the measurement of applied stress as accurate as possible.

III.— RECOVERY OF STEEL FROM TENSIONAL, OVERSTRAIN.

When a bar of iron or siteel is subjected to a tensional stresfc

sufficient to oause a permanent stretch in the material, the physical

3

character of the bar changes, and it becomes semi-plastic, while its elastic limit generally reduces to zero, followed by recovery, wnich is comparatively rapid! in the case of wrought iron, but less so in the case of steel.

As an example of this, the case of a mild steel bar may be taken. This bar, and others referred to later, were cut from a piece of boiler plate, and afterwards- straightened and annealed. The bars were machined to a breadth of 2.07 inches, and the thickness was Iteft untouched, the mean thickness in the present case being 0.377 inches, and the length under test 8 inches.

The Extensometer used was of the EWing type, ome division cor- responding to 1-50,000 inches.

A test was) first made to give the bar a permanent stretch, with the following results: —

LOADS IN LBS.	READING OF EXTENSOMETER.	DIFFERENCE.
0	400
		—35
2,000	435
		—35
4,000	470
		—34
6,000	505
		—35
8,000	539
		—34
10 0	573
		—35
12,000	608
		—35
14,000	643
		—36
16,000	679
		—36
LS,000	715
		—38
20,000	753
23,000	Went off scale.
	Extended
	length.
24,000	8.10 inches.
24,500	8.12 '
25,000	8.14 <
26,000	8.15 '
16,500	8.L>5 '
27 000	8.165 <
0	8.15 <

The measuring instrument was immediately re-set to the standard distance of 8 inches, and a re-test was made, rising by increments of 2,000 pounds to 24,000 pounds, and then to zero again by decre- ments of 2,000 pounds. The following results were obtained: —

4

LOAD IN POUNDS.	READING OF EXTENSOMETER.	DIFFERENCE.
0	400
		—35
2,000	435
		—34
4,000	469
		—37
6,000	506
		—37
8,000	543
		-37
10,000	580
		—42
12,000	622
		—42
14,000-	662
		—42
16,000	704
		—40
18,000	744
		-54
20,000	798
		—48
22,000	846
		—51
24,000	897
		—35
22,000	8i2
		—35
20,000	827
		-37
18,000	790
		—42
16,000	7^8
		—39
14,000	709
		—40
12,000	669
		—40
10,000	629
		-41
8,000	588
		-42
6,000	546
		—42
4,000	504
		— 44
2,000	460
		—45
0	415

A comparison of the readings obtained at once shows how differ- ently the bar behaves after the tensile overetrain. There is now no linear relation between the stress and the strain, but Itlie latter

5

rises irregularly as the stress increases; the differences are irregular and do not agree except accidentally with thie differences obtained as the load is removed.

The bar was now laid on one side, and a gradual recovery of its elastic properties took place, which was shown by thei approach to the first readings made. At the end of three hoiurs there was a marked recovery, and the following reading® were taken: —

Load, Ibs.	Beading.	Difference .
0	400
		—34
2,000	434
		—34
4,000	468
		—37
6,000	505
		—37
8,000	542
		—38
10,000	580
		—39
12,000	619
		—41
14,000	660
		—40
16,000	700
		—42
18,000	742
		—45
20,000	787
		—42
22,000	829
		—46
24,000	875
		—34
22,000	841
		—35
20,000	800
		—37
18,000	769
		—38
16,000	731
		—38
14,000	693
		09
12,000	654
		—41
10,000	613	—40
8,000	573
		—41
6,000	532	—43
4,000	489
		—43
2,000	446
		—44
0	402

The tests of the bar were repeated at the end of one day, 3 days, 7 days and 28 days, and these showed a further recovery, whose rate diminished with the time, but at the end of the month the recovery was practically complete. The readings obtained are shown in the annexed table: —

LOAD LBS.	1 DAY AFTER.	3 DAYS AFTER.	7 DAYS AFTEB.	'28 DAYS AFTEB.
Reading. A	Reading. A	Reading. A	Reading. A
0 2.000	400 434 £?	40') 434	400 434	400 oo 433 "g
4.000	469 ...	469 "S	468 •*[	467 "^*
6.000	504 ~™	504 "22	503 ~^J	500 "OR
8.000	539 "r°	539 "5J	538 ~^	535 o?
10.000	576 "SA	574 ~!^	574 *g	570 S
I2.oeo 14.000 16.000	614 654 Jy 695	en ~j: 648 "JJ 685	609 -g 646 684	605 640 676
18.000	736 ~4J	TOK ~4" (ZO ,„	722 "2	712 ?
20.000	779 ~rl	7«r -40 A •)	761	747
22.000	822 ~4^	807	802	783 Xj
24.000	867 ~4£	851 ~44	847	818 S
22.000	832 "g	817 yf	813 ~:J*	784 oj
20.000 18.000 16.000 14.000	797 ~;j? 762 -* 724 686	782 -^ 746 710 ~:^ 672	779 ~34 34. 745 709 -*J 672 ~^	750 JJ 715 680 o? 645 "g
12.000	648	634 ~^	635	611 -?4
10.000	607	596	596 "g	576 "S
8.000	566	556 ~4"	558 ~'S	540 "2
6.000	525 ~*i	517 "5J	519	505 «
4.000 2.000 0	483 441 398	478 438 -4" 397	481 ~S 443 2 402	470 "iJ 435 399

In order to exhibit the changes in a manner which shows the recovery in a striking manner, the results were plotted in Fig. 2 after the manner first suggested by Ptrof. Swing.* A number pro- portional to the load was subtracted from each reading, and the new reading obtained was plotted instead; this accentuates the differences, and as each curve is plotted from a different initial point, the recovery is exhibited in a form which can be readily fol- lowed by the eye. In the present instance 25 divisions were sub- tracted for each increment of 2,000 pounds.

Referring to the diagram, curve I. denotes the original stress strain curve, only part of which is shown, as the total exension Table I. is 8,250 divisions at a load of 27,000 pounds. Curve II. and the succeeding curves show the result of the overstrain. The stress-strain curve becomes a loop, and the area enclosed is a

#Muir on the Recorery of Iron from Overstrain . Phi]. Trans. , 1899.

2.EOOO

Z400C

2CCOG

I600C-

Fig. 2

I- Original Stress Strain Curve E- Immediately afterwards

- 3 hours afterwards. TSL- 1 day afterwards. Y -3 days afterwards, afterwards. W-28 days

ISO 200

v » s s o N s sure of the work done on the bar by the loading and unloading, and therefore is a measure of the imperfectness of the elasticity.

The rest caused a gradual recovery of the metal, and showed itself in the gradual diminuition of the enclosed area until at the end! of 28 days the recovery was very nearly but not quite perfect.

SHEAR STRAIN.

The recovery of iron and steel from shear strain appears to follow the same general law as for tension. It will 'be sufficient to quote the case of a turned specimen of wrought irom, having a diameter of .0446 inch, and length under test of 4.00 inches. The specimen was subjected to shear by a twisting couple, and the distortion be- tween the sections was measured by a strain measuring instrument of special design.* The bar was first subjected to torque of gradually increasing amount until the elastic limit was passed. The readings obtained were as follows, the calibration of the instrument being sucTi that 12.8 divisions corresponded to 1 minute of arc.

*Coker «« on Instruments for Measuring Small Torsional Straics." Phil. Mag., 1899

8

Torque inch Ibs.	Reading . Diff .	Torque inch Ibs.	Reading. Diff.
0	0 —302
75	302		1489
	—303	3(30	—81
150	605		1568
	—303	375	—79
225	908		1666
	—304	390	—98
. 00	1212		1808
	—63	495	—142
315	1275		2038
	—64	420	—230
330	1339		2548
	—69 I 435	—510
345	1408		Specimen twisted
	—81	450	through.
360	1489		40

Immediately after (the load was released and a new series of read- ings obtained, the maximum torque applied being 375 inch pounds, and similar observations were recorded at the end of one hour, 3 hours, 1 day, 4 days and 8 days, the last experiment showing practically complete recovery.

The readings were as follows: —

Torque	Immediately	One hour	ihree hours	One day	* our days	Eight days
inch	after	a ter	after	after	after	after
Ibs.	I	I	I	I	I	I
	Read-	Read-	Read-	Head-	Read-	Read-
	in?. A	ing. A	ing. A	ing. A	T'g. A	ing. A
0	0	0 0	0	0	0
	-305	—304	—301	—301	301	—302
75	305	304	301	301	301	302
	—324	—306	—303	—302	303	—303
150	629	610	604	603	604	605
	—324	—312	—307	—305	304	—304
225	953	922	911	903	908	909
	—339	—319	—317	—314	307	—303
300	1292	1241	1228	1222	1215	1212
	-361	—349	—322	—316	310	—305
375	1653	1590	1550	1546	1525	1517
	-^3 3j -301	—301	—302	302	—301
300	1350	1289	1249	1244	1223	1216
	—305	—307	—304	—303	303	—302
225	1045	982	945	941	920	14
	—315	—312	— 109 —307	306	303
150	730	70	636	634	614	611
	—325	—324	—316	—314	307	304
75	405	346	320	320	307	307
	—338	—332	321	319	308	306
0	67	14	I	+1	1	1

The readings obtained are given in the above table, and are plotlted on Fig. 3, after deducting 275 divisions for each increment or decrement of 75 inch pounds of torque. The rapid recovery of wrought iron is well shown by the looped stress-strain curves II. to VII, and the rapid diminution off hysteresis in one hour after the overstrain is very marked.

Fig. 3

I - Original Stress Strain Curve

II - Immediately afterwards. Ill- / hour affer I.

IV - 3 hours after I. - I day affer I. affer I. VH - 8 days affer I.

100 200 300 400 500 600

DIVISIONS OF ARC

of instrument - I minute of arc - 12-35 divisions)

THE INFLUENCE OF TEMPERATURE UPON THE RECOVERY FROM OVERSTRAIN.

It is well known that if a bar of iron or steel be stressed beyond the elastic limit in tension, it will regain its elastic properties if raised to a dull red heat and allowed to cool slowly in air or in lime, and this process may be repeated again and again without

10

Fig. 4

I -Initial Stress Strain Curve

ARer boiling in water for 10 minutes ID- ....... Second boiling in inter for 10 minutes

BT ....... Hun/ » ...«

Y .._... fourth _

firth

600 800 1000 1200 1400

1 V I S I O N

(€00

11

seriously affecting the physical properties of the bar. Recently our knowledge of annealing has been greatly increased by the discovery of Muir* acting on a suggestion of Prof. Ewinig, that a compara- tively low temperature, such as that of boiling water wild restore a strained bar to its elastic condition. As an example of this a test made in -the McGill Laboratory shows the effect of boiling water upon an overstrained bar. The specimen was a flat 'bar of boiler plate, width 2.07 inches, thickness 0.375 inch, and length under test 8.00 inches. This specimen was similar to the one used in the first experiment, and was annealed at the same time. The Mstory of the bar is shown by aid of figure IV., in which Curve I. shows the relation of stress to strain when the bar was subjected to a gradually increasing stress. The first notable deviation from perfect elasticity began w.hen a total stress of 20,000 was applied, and a-ta 25,000 the extension amounted to 8.13 inches, an increase of stress to 27,000 pounds, caused a further extension to 8.15 inches. The load was then removed, and the bar was exposed for 10 minutes to the temperature of 'boiling water and again re-tested. Curve II. shows the resulting stress-strain curve, with a yield point at about 31,000 pounds, considerably above the first. The boiling was re- peajted in the same manner, and a new yield point (Curve III.) was obtained at 36,500 pounds, while further tests after annealing in boiling water gave new yield points at 41,500, 45,600 and finality 48,100, when the bar broke, with a total extension of about 15 per cent. This extension is less by about 6 per cent, than would have been obtained if the bar had been stressed to breaking with- out any intermediate treatment. Even a temperature of 50 degrees C, effects a considerable physical change in a bar of overstrained ma- terial, restoring its elasticity and raising the yield point.

THE EFFECT OF TEMPERATURE ON OVERSTRAIN PRODUCED

BY SHEAR.

The annealing effect of boiling water is equally marked upon material overstrained by shear. As an example, a test was made upon a turned steel bar of 0.425 inch diameter, and having a length of 4 inches under test. On each occasion the stress was carried well beyond the yield point, and was followed by immersion in boiling water for 10 minutes. The curves obtained are plotted in Fig. 5, the unit of strain being such that 12.85 units correspond to an angular distortion of 1 minute of arc. Curve I. shows the initial stress-strain curve, and curve II. represents the behaviour of the bar after 10 minutes' immersion in water at 100 degrees: C. The remaining curves III. to IX. show the relation of stress to

* Muirloc. cit.

12

1200

»050

900

750

U

Z

- 600

-u 450

»

car a

o 300!

I-

150

L

7"

200 400 600 800 100 1200

DIVISIONS

1 12 85 Divisions =• I minufe)

1400

1600

strain after each subsequent annealing in boiling water. After each stress and subsequent heating, the recovery was practi- cally complete, and the yield point rose after each annealing. At the end of 9 separate twists the bar exhibited no change externally. A test of a steel bar was also carried out for me by Messrs. Elanchard, Clement and G-agnon, Fourth Year students in the Civil Engineering Department. A turned steel bar of similar quality to the previous one was chosen, having a diameter of 0.757 inch, and a length under Itest of 8.00 inches. The strain measuring instru- ment was a simple graduated circle divided to 15 minutes, and readings could be estimated to 1 minute of arc. Each separate appli- cation of twisting moment was carried beyond the yield point,

13

and was followed by immersion in water at 50 degrees C. for 15 minutes. Even' this low temperature was sufficient to cause a very effective recovery to the elastic condition, the recovery being in all cases accompanied by a rise in the yield point. The data ob-

Fig. 6

O" 4 8* 12* l«f V? Ut Z?

tained are exhibited in Fig. 6, and the history of the treatment of the bar is told by the curves marked I. to XV., I. being the initial and XV. the final curve of the series. Even at the end of the fifteenth twist, when the total permanent twist had reached .270 degrees, there seemed no probability of the bar fracturing.

The remarkable effect of a slight increase in temperature in re- storing an overstrained bar to approximately its primitive condition may ultimately prove to be of considerable practical importance, as it affoirds a means of annealing, which, although not as perfect as that accomplished by the blacksmith, has the advantage that polished surfaces of iron and steel retain their polish and are not distorted by the process. In bridge structures and the like, exposed to repeated stresses and corresponding fatigue of the metal, the suggestion that this fatigue may be partially counterbalanced by the heat of a summer's day may not be so improbable as it at first sight seems to be.

THE EFFECT OF LOW TEMPERATURE! UPON THE RECOVERY FROM OVERSTRAIN.

It is interesting to determine what action low temperature, such as that afforded by a Canadian winter, has upon the behaviour of steel after overstrain, and in order to contrast this with recovery at

the temperature of the Laboratory, viz., about 68 degrees Fahr., a specimen was taken similar to the one described with, reference to Fig. II. and the accompanying tables. The bar was 2.07 inches in breadth, and had a mean thickness of 0.378 inch, the length under test being as before, 8 inches. The specimen was first per- manently stretched in exactly the same manner as in. the previous case, and a comparison of the appended table with that on page 4 shows a very close agreement.

December 15th, 1900.

Load, pounds.	Reading.	A
0	400
		- 34
2,000	434
		-36
4,000	470
		—36
6,000	506
		—35
8,000	541
		—33
10,000	574
		—34
12,000	608
		—34
14,000	642
		—35
16,000	677
		-34
18,000	711
		—33
20?000	744
	—42
22,000	786
	—82
24,000	868
26.000		8.08"
27,000		8.12"
27,500		8.16"
0		b.155"

The specimen was then exposed to the outside temperature for 3 days, during which the temperature varied from a minimum of — 2.5 degrees Fahr. to a maximum of 11.2 degrees Fahr., ami it was then subjected to a cycle of stress ranging from zero to 24,000 pounds. The results are recorded in Fig. VII., Curve I., and a comparison with Fig. II shows it to have a fair agreement with Curve III. ol that figure. The bar was again exposed to the outside temperature, which had a mean value during the nexft four days of 18.52 degrees Fahr., a minimum value of — 2.5 Fahr., and a maximum value of 31.72

15

Fig. 7

29000

250

50 100 150 200

SCALE DIVISIONS

degrees Fahr. A second test, on December 18th, showed1 very little change in the condition of the bar, as will be seen by com- paring the figures in Table III. with those previously obtained, and the bar was therefore given a long rest through the re- mainder of December and the whole of January. The mean tem- perature for the remaining days of December was 25.64 degrees Fahr., the highest 39 degrees Fahr., and lowest 0.6 degree

16

Fahr. During January the mean temperature was 12.75 degrees Fahr., the lowest temperature recorded being — 16.7 degrees Fahr., and the highest 39.6 degrees Fahr. The bar was tested again on February 1st, see Table III. and Curve III., Fig. VII., and, as will be seen, it exhibited some slight indications of recovery, but very small in comparison with the recovery which would 'have been effected by the ordinary indoor temperature. A final test was made on Fe-b. 26th, the mean temperature during the interval being 12.46 degrees Fahr., with a minimum of — 4.0 degrees Fahr., and a maximum of 28.4 de- grees Fahr. The recovery during this period was practically nil. It ap- pears, therefore, that low temperatures tend to retard the recovery of steel, which has been subjected to overstrain in quite a marked degree.

TABLE III.

LOAD LBS.	DEC. 18, 1900. I.	DEC. 22ND. 11.	FEB'YIST. III.	FEB'Y 26TH. IV.
Reading. A	Reading. A	Reading. A	Reading. A
0	400 o.	400 _u	400	400
2,000	434 X7	4.) 4 qq	432 -11	431 ~H
4,000	469 22	467 £	464 ~*t	463 -;"
6.000	502 ~f£	502 ~q7	498 ~™	496 "g
8,000	538 _™	536 ""JJ	531	530 "J*
10,000 12,000	611 "I**	571 SJ2 607 3°	562 ~3J 599 "'H	564 "tj 598 ~?J
14,000	648 "ii	645	633 ~t*	633 a!
16,000	685 3<	683 -;j°	670	671 •**
18,000	724	722 -JJ	708 3*	706 ~S
20,000	763	762 ~f	746	744 ~q®
22,000	805 44	804 :«	785	783
24,000	849 -44	847 j£	828 "|f	825 "SJ
22,000	816	815 ~QO	797 •«*	794 -**
20,000	782 -?£	783	765 -^	762 "g
18,000	749	750 -£	733 -g	729 „
16,000	713 *JJ	716 "g	699 -f*	696 qq
14,000	679 "5|	682 ^	667 ~»J	663 qj
12,000	642 "Si	646	632 "^	629 ~qV
10,000	605 £'	608 ~^	599	594 '2
8,000	567 "2	573 g	561 -g	558 ^
6,000	529 -;£	535 ,"	527	522 "^
4,000	491 3°	496	491 ~f	475 "2f
2,000 0	453 OQ 414	456 412	454 "J* 413	440 q? 403

In conclusion, the author desires to thank Professor Bovey, Past- President of the Institution, who placed the Testing Laboratory of McGill University at his disposal; Prof. McLeod, Secretary of the Society, who kindly supplied the information regarding the tempera- ture variations from the McGil Observatory records, and Messrs. Blanchard. Clement and Gagnon. who made all the observations re- corded on Fig. 6.

17

RETURN TO the circulation desk of any University of California Library

or to the

NORTHERN REGIONAL LIBRARY FACILITY Bldg. 400, Richmond Field Station University of California Richmond, CA 94804-4698

ALL BOOKS MAY BE RECALLED AFTER 7 DAYS

• 2-month loans may be renewed by calling (510)642-6753

• 1-year loans may be recharged by bringing books to NRLF

• Renewals and recharges may be made 4 days prior to due date.

DUE AS STAMPED BELOW

KlftR^t

12,000(11/95)

LD 21-100m-9,'47(A5702sl6)476

02449

662552

v

UNIVERSITY OF CALIFORNIA LIBRARY