Effect of thermal shock on clay bodies : a report

HI L L I N NO I SUNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN

PRODUCTION NOTE

University of Illinois atUrbana-Champaign Library

Large-scale Digitization Project, 2007.

UNIVERSITY OF ILLINOIS BULLETINISSUED WEEKLY

VoL XXVIII June 23, 1931 No. 43

[Entered as second-class matter December 11, 1912, at the post ofice at Urbana, Illinois, underthe Act of August 24, 1912. Acceptance for mailing at the special rate of postage provided

for in section 1108, Act of October 3, 1917, authorized July 81, 1918.1

THE EFFECT OF THERMAL SHOCK ONCLAY BODIES

A REPORT OF AN INVESTIGATION

CONDUCTED BY

THE ENGINEERING EXPERIMENT STATIONUNIVERSITY OF ILLINOIS

IN COOPERATION WITH

THE CLAY PRODUCTS ASSOCIATION

BYWILLIA R MORGAN

WILLIAM R. MORGAN

BULLETIN No. 229ENGINEERING EXPERIMENT STATION

PomusHID BT Ta UNIVoBImTY or ILImOIm, UmmsA

Parcz: Twrt CNwts

HE Engineering Experiment Station was established by actof the Board of Trustees of the University of Illinois on De-cember 8, 1903. It is the purpose of the Station to conduct

investigations and make studies of importance to the engineering,manufacturing, railway, mining, and other industrial interests of theState.

The management of the Engineering Experiment Station is vestedin an Executive Staff composed of the Director and his Assistant, theHeads of the several Departments in the College of Engineering, andthe Professor of Industrial Chemistry. This Staff is responsible forthe establishment of general policies governing the work of the Station,including the approval of material for publication. All members ofthe teaching staff of the College are encouraged to engage in scientificresearch, either directly or in cooperation with the Research Corpscomposed of full-time research assistants, research graduate assistants,and special investigators.

To render the results of its scientific investigations available tothe public, the Engineering Experiment Station publishes and dis-tributes a series of bulletins. Occasionally it publishes circulars oftimely interest, presenting information of importance, compiled fromVarious sources which may not readily be accessible to the clienteleof the Station.

The volume and number at the top of the front cover page aremerely arbitrary numbers and refer to the general publications of theUniversity. Either above the title or below the seal is given the num-ber of the Engineering Experiment Station bulletin or circular whichshould be used in referring to these publications.

For copies of bulletins or circulars or for other information addressTHE ENGINEERING EXPIERIMENT STATION,

UNIVERSITY OF ILLTNOIS,

URBANA, ILUNoIs

;*' if* *** .*f" . *"' ' '*". ;

^ ' ^ * . - " ' * '. - * ' . ^ " r

,' * ~ , . * - * : ' * - ' *: ' *,

- Fi:^

UNIVERSITY OF ILLINOIS

ENGINEERING EXPERIMENT STATION

THE EFFECT OF THERMAL SHOCKON CLAY BODIES

A REPORT OF AN INVESTIGATION

CONDUCTED BY

THE ENGINEERING EXPERIMENT STATIONUNIVERSITY OF ILLINOIS

IN COOPERATION WITH

THE CLAY PRODUCTS ASSOCIATION

BY

WILLIAM R. MORGANSPECIAL RESEARCH ASSISTANT IN CERAMIC ENGINEERING

ENGINEERING EXPERIMENT STATIONPUBLISHED BY THE UNIVERSITY OF ILLINOIS, URBANA

BULLETIN No. 229 JUNE, 1931

o# 4 81 188 UNIVERSTYOF ILLINOISi. PRESS ..

CONTENTSPAGE

I. INTRODUCTION . . . . . . . . . . . . . 5

1. Previous Data . . . . . . . . . . . 5

2. Purpose of Present Investigation . . . . . . 53. Acknowledgment . . . . . . . . . . 6

II. THERMAL SHOCK TESTS . . . . . . . . . . 64. Testing Procedure . . . . . . . . . . 65. Experimental Data . . . . . . . . . 76. Results of Tests. . . . . . . . . . . 97. Relation between Initial Strength and Per Cent Re-

duction in Strength . . . . . . . . . 118. Relation between Final Strength and Per Cent Re-

duction in Strength . . . . . . . . . 119. Discussion of Mathematical Relationships . . . 12

10. Application to Experimental Data . . . . . 18

III. MODULUS OF ELASTICITY AND POROSITY . . . . . 18

11. Determination of Modulus of Elasticity . . . 1812. Results . . . . . . . . . . . . . 2013. Determination of Porosity. . . . . . . . 2014. Results . . . . . . . . . . . .. 22

15. General Relationships . . . . . . . . . 23

IV. EFFECT OF ADDITION OF GROG . . . . . . . . 24

16. Effect of Grog on Resistance to Thermal Shock . 24

V. GENERAL DISCUSSION OF EFFECT OF THERMAL SHOCK 24

17. Tentative Definitions . . . . . . . . . 24

18. Comparisons with Data of Other Investigators. 25* 19. Summary of Results of Other Investigators . 26

20. Improvement of Test . . . . . . . . . 26

VI. SUMMARY . . . . . . . . . . . . . . 2721. Summary . ........... 27

LIST OF FIGURESNO. PAGE

1. Relation between Initial Strength and Per Cent Reduction in Strength. . 8

2. Relation between Final Strength and Per Cent Reduction in Strength. . 8

3. Relation between Initial and Final Strength . ... . . . . . .. 9

4. Method of Plotting Data for Individual Bodies . . . . . . . . 9

5. Relation between Initial and Final Strengths for All Bodies Tested . . 10

6. Relation between Initial and Final Strength for an Individual Body . . 10

7. Graphic Relationship between Initial and Final Strengths and Per CentReduction Based on Mathematical Derivation. . . . . . .. . 13

8. Graphic Relationship between Initial and Final Strengths and Per CentReduction Based on Mathematical Derivation. . . . . . .. . 14

9. Theoretical Relationship between Initial and Final Strengths Based onMathematical Derivation ..... . . . . . . . . 15

10. Relation between Maximum Values of Sf and Corresponding Values ofSi, P, andn . . . . . . . . . . . . . . . . . 16

11. Relation between Initial and Final Strength and Per Cent Reduction whenDecreasing Initial Strength Approaches a Constant C, as Final Strength

Approaches Zero . . . . . . .. . . . . . . . . 17

12. Relation between Initial Strength and Modulus of Elasticity Based onAverage Values for Each Body ... . . . . . . . . 19

13. Relation between Modulus of Elasticity and Per Cent Reduction Based onIndividual Values for Each Body Burned at Three Different Tempera-tures . . . . . . . .. . . . . . . . . . . . 20

14. Relation between Porosity and Per Cent Reduction Based on AverageValues . . . .I . . . . . . . . . . .. . .. . 21

15. Relation between Porosity and Per Cent Reduction Based on IndividualValues for Each Body Burned at Three Different Temperatures . . 21

16. General Relation between Initial and Final Strength, Per Cent Reduction,Modulus of Elasticity, and Porosity . . . . . . ... . 22

17. Effect of Grog on Physical Properties and Resistance to Thermal Shock . 23

THE EFFECT OF THERMAL SHOCK ONCLAY BODIES

I. INTRODUCTION

1. Previous Data.-Prior to 1927, published data on the effect ofthermal shock on clay bodies were confined to the results of water andair quenching tests in which the basis of comparison was the numberof quenchings required to produce failure.

Since that time data concerning thermal shock have been pub-lished by Goodrich* with relation to the compressive strength offirebrick; by Parmelee and Westmant with reference to the transversestrength of firebrick; by Heindl and Mongt correlating transversestrength and other physical properties with quenching tests on saggerbodies; and by Heindl and Pendergast¶ correlating transversestrength and other physical properties with quenching tests onfirebrick.

2. Purpose of Present Investigation.-The present study of ther-mal shock has been made as part of a general investigation of thephysical properties of twenty commercial shale and fireclay bodies,ranging in softening point from cone 7 to cone 30, correspondingapproximately to temperatures of 2280 and 3000 deg. F., respectively.It was undertaken primarily for the purposes of developing a methodfor making comparisons of the resistance of bodies to thermal shockon a quantitative basis, and of establishing relations between physicalproperties and resistance to thermal shock by means of which resist-ant bodies could be designed and plant control of their manufactureeffected.

The purposes of this bulletin are:(1) To describe the method briefly, presenting only sufficient data

to indicate the general relation between physical properties, and toindicate, in part mathematically, the relation between these prop-erties and resistance to thermal shock.

*H. R. Goodrich, "Spalling and Loss in Compressive Strength of Fire Brick," Jour. Amer. Cer.Soc., 10, 784-94, 1927.

tC. W. Parmelee and A. E. R. Westman, "The Effect of Thermal Shock on the TransverseStrength of Fireclay Brick," Jour. Amer. Cer. Soc., 11, 884-895, 1928.

$R. A. Heindl and L. E. Mong, "V. Progress Report on Investigation of Sagger Clays and Prep-aration of Experimental Sagger Bodies According to Fundamental Properties," Jour. Amer. Cer. Soc.,12, 457, 1929.

J¶R. A. Heindl and W. L. Pendergast, "VI. Progress Report on Investigation of Fireclay Brickand the Clays Used in Their Preparation," Jour. Amer. Cer. Soc., 12, 640, 1929.

ILLINOIS ENGINEERING EXPERIMENT STATION

(2) To discuss the published data and conclusions of other inves-tigators on the basis of the results obtained in the present investi-gation.

3. Acknowledgment.-The data presented in this bulletin were ob-tained in an investigation which was conducted by the EngineeringExperiment Station of the University of Illinois, of which M. S.KETCHUJM, Dean of the College of Engineering, is director, in co6per-ation with the Clay Products Association. The research was carriedout in the Department of Ceramic Engineering of which C. W.PARMELEE, Professor of Ceramic Engineering, is the head, and ofwhich R. K. HURSH, Professor of Ceramic Engineering, has been theacting head, during the completion of the work.

Acknowledgment is made to Professor PARMELEE for his co6per-ation and assistance during the early part of the investigation; toProfessor HURSH for his interest and advice and especially for hiscritical survey of the material during its preparation for publication;and to Mr. G. H. DUNCOMBE, Jr., Ceramic Engineer for the ClayProducts Association, for his many helpful suggestions throughoutthe progress of the work.

II. THERMAL SHOCK TESTS

4. Testing Procedure.-Twenty 1 in. x 1 in. x 6 in. test pieces ofeach commercial body were burned in laboratory kilns to cone 04,twenty to cone 2, and twenty to cone 6, approximately equivalent totemperatures of 1925, 2075, and 2175 deg. F., respectively.

The average transverse strength or modulus of rupture for eachbody was determined by breaking ten of the specimens in each sampleby application of a concentrated load at the center of a five-inch span.The remaining ten specimens in each sample were subjected to ther-mal shock before transverse strength determinations were made. Theper cent reduction in strength, based on initial strength, was thencalculated.

The thermal shock test was planned to be equivalent to the moresevere conditions to which the material might be subjected in service,and, in addition, the test temperature selected was above the inver-sion temperature of quartz.

Thermal shock was produced by heating the samples for fifteenminutes entirely within a gas fired muffle furnace maintained at 1100deg. F., and then removing them and cooling them for fifteen minutes

THE EFFECT OF THERMAL SHOCK ON CLAY BODIES

by an air blast applied to one face. The cycle was repeated eighttimes.

The method was essentially that of Parmelee and Westman, butdiffered in the size of the test pieces, the temperature of the furnace,and the length and number of heating and cooling periods.

5. Experimental Data.-All modulus of rupture values, or initialstrengths, for the bodies were plotted against the corresponding percent reduction values, and a smooth curve was drawn showing thegeneral trend of the field (Fig. 1). Each point represents the averageresults from ten specimens before thermal shock and ten specimensafter thermal shock. There are three such points for each materialtested. The values of modulus of rupture after thermal shock, orfinal strengths, were also plotted against per cent reduction in strengthin Fig. 2. The curve in the figure was made to correspond to that ofFig. 1 by calculating the final strength from values on the curve inFig. 1. The general relationship between final and initial strengthcorresponding to the curve values of Figs. 1 and 2 is shown in thecurve of Fig. 3. The points representing the experimental data arealso plotted for comparison.

These are "average curves," and are intended to indicate only thegeneral trend of the data. The use of broken line curves indicates theuncertainty of the trend in the region of the higher values of per centreduction, in Figs. 1 and 2, and of initial strength, in Fig. 3.

The method of plotting is shown in Fig. 4 for an individual body,the data consisting of the three experimental points.

Individual curves, each based on three points and representingone body, are shown in composite form in Fig. 5. The general formof each curve is fixed to some extent by the three plotted points inthat they indicate the presence of a maximum value for final strength.The curves in general have the same form as the average curve.

In order to secure a more positive indication of the trend of therelation between initial and final strength, additional tests were madeon an individual body, covering a burning range from cone 09 tocone 9 corresponding approximately to temperatures of from 1700 to2300 deg. F. Initial strength ranged from approximately 800 to3200 lb. per sq. in.

The data are plotted in Fig. 6, in which each point represents theaverage results from ten test pieces before thermal shock and tentest pieces after thermal shock.

The results show definitely the presence of a maximum value forfinal strength, and indicate that the general form of the individual


(~)

(I

{J)

- 1 /9

Per Cent

FIG. 1. RELATION

~0 dO •4 ~,9 ?4 80 .~0 /00

Rea c//lo'O /12 Transverse Sfren2gfh, /PAfter Ther,7a/ 2Shoca-

BETWEEN INITIAL STRENGTH AND PER CENTREDUCTION IN STRENGTH

3rOO\ - - - - - - - - - - - * -

800o- o-Burned ato Cone 04S-Burnea or7 Cone 2 06 i* -Burned a' Cone 6 - a

2400 ---- o

1800 ~ ~ -------- -a^'*'-I- -------------/0 * a [

/6?00------- --- -- -- -- -

/200 -- /-/ ---- _____b lve-5q/7

800 -- y - - - - - - - - - -f Aver'g'e Curie P/o/!ea•'fo

4 A where / r ? 00 /bper sr./ :.

- ---- were' /7 /s 2. --- - - -- -

Per Cent Reac/ic/on , Trhnsverse Stregfth, P,Af/er T7erea'/ Shoc*

FIG. 2. RELATION BETWEEN FINAL STRENGTH AND PER CENTREDUCTION IN STRENGTH

v


I I I I I I- -Burne' at Cone

a- Burneo' ato' Co•~ *-Bt/r/neo a'^ Cone' 000

e~z2y oo.,00

goo

I I I I I I.4'eroge Curp-e P/o/fed fri~',':

$ s,.f-4~J7

3»/?/^/1

400 500 /26)0 /66)6' z6)•'o ~ao ~ ~

/ni//a/ TraTNsverse STtreE g/?, 5,, A /• /7 pe-R s5. /n.

FIG. 3. RELATION BETWEEN INITIAL AND FINAL STRENGTH

/1000

(i 800

400K

00 400 800 /200

X'n /.. pe qs. in.Per Cet /?R ed'uction /7

T7ra-7sierse 53renqg1/7After Th7ermroV ShocA-

FIG. 4. METHOD OF PLOTTING DATA FOR INDIVIDUAL BODIES

and average curves is correct. In addition it seems probable fromFig. 6 that, with increased vitrification and high initial strengthdevelopment, final strength will decrease and finally approach zero.

6. Results of Tests.-The data indicate, for the burning rangestudied, that:

(1) Per cent reduction in strength increases as initial strengthincreases, i.e., the stronger bodies lose a greater percentage of theirstrength than do the weaker ones.

I,

0 0S!__ 0

_ - _o/ _ _ -6

K

-^ /200

K,

g800

400•..•

I, z

where M,'_ /5 2900 lb. pe1' 5'. q/and n is 2.

0170' 17

- \

S\

S

6) 5. __---______

ri- * 3^-- ;* - -- - -0 *^ - - - -__ __

*.s "

^

7


.4.'k

4~(4)

KK

~z

SZ

Z000 - --- -7 - - - - -

800 ;100II

4° L/'e ?epreseni7ngi/600' --------- --- 7' - - -7TreoretA-/'cg/y--

/' ~Per fect Boady.--- T---o-- ---- / --

400--

^ \ A veroye' Curve-"\

0 400 300 /IOO /600 2000 2400 2800 3zoo/n/i/al Transverse Strengqh // Ik /7 Zpr sq /1

FIG. 5. RELATION BETWEEN INITIAL AND FINAL STRENGTHS FOR ALL BODIES TESTED

/na'/ Tiransv-erse Strengt/7 i17 /z per s. //.

FIG. 6. RELATION BETWEEN INITIAL AND FINAL STRENGTH FOR AN INDIVIDUAL BODY

(2) Strength after thermal shock, or final strength, increases to amaximum and then declines, as the strength before thermal shock, orinitial strength, increases.

If it is accepted that the average exponential curves of Figs. 1, 2,and 3 correctly represent the test results, the properties of thesemathematical curves define the following relations between initialstrength, final strength, and per cent reduction in strength.


7. Relation between Initial Strength and Per Cent Reduction inStrength.-All values of per cent reduction will lie between 0 and 100and can be expressed as some fractional part of 100, while the relationbetween initial strength and the strength corresponding to 100 percent reduction can also be expressed as a ratio. It is evident that thisratio and the per cent reduction are dependent variables which willchange as the initial strength changes.

While any particular equation selected may not completely satisfyall conditions, it will offer a means of systematic variation of the rela-tion between strength before and after thermal shock and per centreduction, from which a helpful analysis of the problem can be made.Such a relationship, which fits the test data fairly well, can be ex-pressed by the equation:

P = 100 (1)

Where P = per cent reduction in strength.Si = initial strength, modulus of rupture in lb. per sq. in.,

corresponding to per cent reduction P.Mioo = initial strength, modulus of rupture in lb. per sq. in.,

corresponding to 100 per cent reduction.n = a variable exponent.

In more convenient form Equation (1) becomes

S, = M,0o (0 (2)

8. Relation between Final Strength and Per Cent Reduction inStrength.-The relation between final strength and per cent reductionin strength can be obtained readily from Equation (2), since the finalstrength is always equal to the initial strength multiplied by the factor

1 - P-) where P is the per cent reduction.100

Let Sf = strength after thermal shock, or final strength, modulusof rupture in lb. per sq. in., corresponding to per cent reduction P.

Then Sf = Si 1 - - (3)

but Si = M 100o - (2)/ \1 / D\

so that Sf = M1 oo (I- 1 1 - -1 -\IUU/ \ IUIJ/


From Equation (4), Sf will be zero when P is 0 and when P is 100.Placing the first derivative of Sf with respect to P equal to zero,

Sf becomes a maximum when

(n + 1)P = 100 (5)

For any positive value of n, S1 will have a maximum value, i.e.,strength after thermal shock will increase to a maximum and thendecline.

9. Discussion of Mathematical Relationships.-

When M 1 0 0 is Variable and n is Constant

The solid line curves in Fig. 7 are plotted from Equation (2),where n is 2 and Mo10 is 1200, 1600, 2000, and 2400 lb. per sq. in.,respectively. They represent the relation between initial strengthand per cent reduction.

The broken line curves represent the corresponding relation be-tween strength after thermal shock and per cent reduction, and areplotted from Equation (4).

When n is kept constant and M1io varied, the per cent reductioncorresponding to maximum strength after thermal shock remainsfixed, for S. is a maximum when (n + 1)P = 100. A group of curvesfor a different value of n will have a fixed but different value for theper cent reduction corresponding to maximum strength after thermalshock.

When n is Variable and M 1 00 is Constant

In Fig. 8 the light solid lines are plotted from the same generalEquation (2), but in this case M1oo is kept constant at 2000 lb. persq. in., and n is allowed to vary. When n = 1, the equation repre-sents a straight line.

For values of n greater than one, a series of curves can be plottedwhich will rise more or less sharply from the origin, will be concavedownward, and will approach M1io with gradually decreasing slope.

For values of n less than one, the curves will rise very slowly atfirst, will be concave upward, and will approach M 0io with increasingslope.7 The broken line curves again represent the corresponding relationsbetween strength after thermal shock and per cent reduction. Theyare plotted from Equation (4).

As the relationship between transverse strength and per centreduction changes from that represented by a straight line, when9 = 1, to relationships represented by curves which are concave


Per Ce -X eC1d/0'oer? /7 -Transverse 1 5-re7g17'After T77erm'a// 51oc/ e

FIG. 7. GRAPHIC RELATIONSHIP BETWEEN INITIAL AND FINAL STRENGTHS ANDPER CENT REDUCTION BASED ON MATHEMATICAL DERIVATION

downward when n is greater than 1, the point of maximum strengthafter thermal shock changes from 50 per cent reduction to lesservalues as n increases, i.e., the more abruptly the curve representingthe relation between transverse strength before thermal shock andper cent reduction rises from the origin, the lower will be the per centreduction at which the body will attain its maximum strength afterthermal shock.

The heavy solid line curve represents the maximum strength afterthermal shock corresponding to different values of n.

Relation between Initial and Final Strength

The direct relation between initial and final strength is obtainedby combining Equations (1) and (4) and eliminating P.

Thus Sf = Si 1_ -(W)1 (6)\ Mlm


N 86

246

206?(N

/66L

S86.

(I

^4

Per Cent AReacI/o /n T47 i'V'erse Stre6-1gt2After T17errn7/ S5o'acl

FIG. 8. GRAPHIC RELATIONSHIP BETWEEN INITIAL AND FINAL STRENGTHS ANDPER CENT REDUCTION BASED ON MATHEMATICAL DERIVATION

from which it follows that Sf will be zero when Si is equal to zeroand M 0oo.

Placing the first derivative of Sf with respect to Si equal to zero,Sf becomes a maximum when

S--M100 (7)(n + 1)»

The direct relation between initial and final strength is showngraphically in Fig. 9, having been plotted from Equation (6). Thecurves indicate that, for a fixed value of Mioo, resistance to thermalshock increases with increase in value of the exponent n, and that thehigher the value of n, the more abruptly the resistance falls off afterthe maximum point is reached.

The relation between initial and final strength for a fixed value ofn with Mioo variable is not shown. It is evident, however, that theincrease in maximum final strength is directly proportional to theincrease in the value of M1oo.


/7///7/a Trotnsverse S/-rel/i, 5,; i, /7. per sq. i?.

FIG. 9. THEORETICAL RELATIONSHIP BETWEEN INITIAL AND FINALSTRENGTHS BASED ON MATHEMATICAL DERIVATION

Sf (max) in Terms of M 1 00 and n, with Corresponding Values of S i and P

The direct relation between maximum final strength, initialstrength corresponding to 100 per cent reduction (M100) and n isobtained by substitution in Equation (4) of the value of P corre-sponding to maximum final strength, given in Equation (5); or bysubstitution in Equation (6) of the value of Si corresponding tomaximum final strength, given in Equation (7).

In either case

Sf (max) = [n+ I M 100 (8)n(n+l) n

From Equations (7) and (8) it is seen that, in the limit, as n becomesinfinite, both S, (max) and the corresponding value of Si approachMioo; and as n approaches zero, Sf (max) approaches zero, and Si

approaches Mo0, where e is the base of Naperian logarithms.e


&/00

80

604

t 40N

S3^ .,

~0Va/&les of the Expone/nt 72?"

FIG. 10. RELATION BETWEEN MAXIMUM VALUES OF Sf AND CORRESPONDINGVALUES OF Si, P, AND n

Values of P corresponding to maximum values of Sf are given byEquation (5)

(n + 1) P = 100

As n approaches zero, P approaches 100, and as n becomes infinite,P approaches zero.

These general relationships are shown graphically in Fig. 10 forvalues of n from 0 to 10.

It has been shown that, for a given characteristic relationshipbetween initial and final strengths, there is a corresponding value of n.The curves indicate that, for any given value of n, the maximum finalstrength and corresponding initial strength are fixed fractional partsof M1 oo. For any value of n there is also a fixed value of per centreduction corresponding to maximum final strength.

Relations When Initial Strength Approaches a Minimum Value Greater ThanZero as Final Strength Approaches Zero

In the case in which the initial strength approaches a minimumvalue greater than zero, as final strength approaches zero, the curverepresenting the relationship may be considered to have been trans-


k/i/7a/ Transverse Streng/h, S,

0 80 40 60 8o /OOPer Cenl/ £vl cA'/On 7i Transverse Streng'/h

FIGa. 11. RELATION BETWEEN INITIAL AND FINAL STRENGTH AND PER CENT

REDUCTION WHEN DECREASING INITIAL STRENGTH APPROACHES A

CONSTANT C, AS FINAL STRENGTH APPROACHES ZERO

lated parallel to the X axis a distance equal to the intercept C on theX axis. Such a relationship is shown in Fig. 11. The range of valuesfor final strength remains unchanged, while each value of initialstrength is increased by amount C. Thus the relation between initialand final strength becomes:

Sf = (Si - C) 1 - (Si - C

As initial strength increases from C to Mo0o, and the final strengthincreases from zero to a maximum and then decreases to zero, the percent reduction decreases from 100 to a minimum value and increases-again to 100.

The relationships, when n is 2, are shown graphically in Fig. 11.


10. Application to Experimental Data.-The characteristics ofindividual bodies are shown by curves in Fig. 5. In form, these curvesclosely resemble the "average curve" of Fig. 3 which is repeated forcomparison. In each case the final strength increases to a maximumand then decreases with increasing initial strength. While the dataavailable are not sufficient for exact determinations of the constantsfor the several bodies, the values of M 00oo range from 1200 to 4000 lb.per sq. in., and n is, in all cases, greater than 1.

Values for initial and final strength for all bodies below maximumresistance to thermal shock lie in a comparatively narrow field slightlybelow the 45 deg. line. It is evident, however, that each body has itsindividual characteristics, for, as the maximum points are approachedand passed, the curves become widely divergent. In preliminaryexperiments, only small differences were obtained in final strengthcorresponding to quite large differences in initial strength. Thereason is now apparent, for, at the flat upper part of the averagecurve (Fig. 5) representing maximum resistance, final strength appar-ently remains unchanged at 1080 lb. per sq. in., while initial strengthchanges from 1400 to 1940 lb. per sq. in.

This also may offer some explanation of the discrepancies betweenservice tests and laboratory thermal shock tests. Laboratory testsusually are made on refractories which in their manufacture haveattained an initial strength considerably below that which would cor-respond to 100 per cent reduction in strength in a properly designedtest. Most refractories tested would lie in the best range, below orapproaching maximum resistance. In severe service, however, atleast the exposed surface approaches complete vitrification, and atthis point the refractory is far beyond its maximum resistance.

Refractories which compare favorably at or below their maximumresistance might be widely different under service conditions, as in-dicated by the divergence of the curves representing the relationbetween initial and final strength after they have passed their maxi-mum value.

III. MODULUS OF ELASTICITY AND POROSITY

11. Determination of Modulus of Elasticity.-Modulus of elasticitywas determined by loading the specimen as a cantilever beam with aconcentrated load of 2 pounds suspended 7 inches from the support.Deflections were measured by an Ames dial graduated to 0.0001 in.The breadth and depth of the specimens were measured to 0.01 in.The values were calculated from the standard equation for modulus


/inf,"a/ Trnsiers6e Stre "te, fAverages) lb //l A/er . 1

FIG. 12. RELATION BETWEEN INITIAL STRENGTH AND MODULUS OFELAsTIcIr BASED ON AvERAGE VALUES FOR EACH BODY

Pl3of elasticity, E - -, which for specimens of rectangular cross-

3AIsection becomes

E =4P13

Abd3

whereE = modulus of elasticityP = load in poundsI = length of span in inches

A = deflection in inches at point of application of the loadb = breadth of specimen in inchesd = depth of specimen in inches

The values determined for modulus of elasticity are not sufficientlyaccurate for a determination of the absolute value of that elasticconstant, but are of sufficient accuracy for the determination of therelation between modulus of elasticity, initial strength, and finalstrength.


Per Cernt ea'dcc/o of Tr'asverse Stfre,/g'f

FIG. 13. RELATION BETWEEN MODULUS OF ELASTICITY AND PER CENTREDUCTION BASED ON INDIVIDUAL VALUES FOR EACH BODY

BURNED AT THREE DIFFERENT TEMPERATURES

12. Results.-The modulus of elasticity increases with the initialstrength, as shown by average values of the several bodies plotted inFig. 12.

The relationship between the modulus of elasticity and the percent reduction in strength due to thermal shock is shown by the"average curve," and by curves for three individual bodies in Fig. 13.

The linear relationship shown in Fig. 12 and the similarity of thecurves of Fig. 13 with that in Fig. 1 shows that the mathematicalrelationship between the modulus of elasticity and the per cent re-duction in strength due to thermal shock would take the same form asthe equation given.

13. Determination of Porosity.-Apparent porosity is the per cent,by volume, of open pore space based on the bulk volume of thetest piece.

The test pieces were weighed dry and were then saturated byimmersing in kerosene under a 28-in. vacuum for twelve hours.The saturated specimens were then weighed both in air and sus-pended in kerosene.


FIG. 14. RELATION BETWEEN POROSITY AND PER CENT REDUCTIONBASED ON AVERAGE VALUES

35 -

030

00

" 0

5 0

/0-

^ _______^^ ^^ -^___ ___\

- ± - - .1 - I - L - .1 - L - .J - L - .1 - L - .I........A - .1 - .1 - J. - .1 - L - .1 - I/0 wO 30 40 s0 60 70 80 90Per Cen't Reduct'orn7 /?2 Transverse Streng/h

FIG. 15. RELATION BETWEEN POROSITY AND PER CENT REDUCTIONBASED ON INDIVIDUAL VALUES FOR EACH BODY BURNED AT

THREE DIFFERENT TEMPERATURES

K


30

;?ot /

K

1<

• S0 0 " .

/0 /000 -

W-W

where

, = weight suspended in kerosene

0 0 40 60 80 /00 ,0

in strength increase. Rur erse

FIG. 16. GENERAL RELATION BETWEEN INITIAL AND FINAL STRENGTH,PER CENT REDUCTION, MODULUS OP ELASTICITY, AND POROSITY

The porosity was then calculated from the equation

p 100 ( W - W d

whereP = per cent apparent porosity

WyV = saturated or wet weightWd = dry weightW, = weight suspended in kerosene

14. Results.-As the porosity of the body decreases (with greaterdegree of vitrification) the initial strength and the per cent reductionin strength increase.

The general relation between porosity* and strength is an estab-lished fact and will not be discussed.

The relation between porosity and per cent reduction is shown inFig. 14 in which each point is the average of all values for each body.

*A. V. Bleininger: Trans. Amer. Cer. Soc., XII, 564, 1910.


60

z50

40

20

la

0

4000-

SN3000

/--5/7 Grog'

0 20 40 60 80 /00

Per Cent PReuki3 w in Trawsverse StrenghFIG. 17. EFFECT OF GROG ON PHYSICAL PROPERTIES AND

RESISTANCE TO THERMAL SHOCK

All individual values are plotted in Fig. 15 and give a scatteredfield because the data represent a comparison of bodies, of widelydifferent characteristics, which have been burned at different tem-peratures.

The broken line curve shows a fairly definite lower limit in poros-ity for corresponding values of per cent reduction.

The average values in Fig. 14 represent the same group of bodies,each burned at the same temperature, and show the general trendclearly.

15. General Relationships.-Initial strength, per cent reduction instrength, and modulus of elasticity increase and porosity decreasesas strength after thermal shock increases to a maximum and thendeclines.

The general relationships are shown graphically in Fig. 16.


IV. EFFECT OF ADDITION OF GROG

16. Effect of Grog on Resistance to Thermal Shock.-Grog ordina-rily is considered to be any non-plastic and inert material used inceramic products for the purpose of altering the physical properties.Such material may be prepared from clay by burning to remove plas-ticity and shrinkage. The resulting product is inert, and after grind-ing and screening is used as grog.

To determine the general effect of additions of grog, the followingbodies were subjected to thermal shock:

(1) 100 per cent clay (through 100 mesh)(2) 85 per cent clay, 15 per cent grog (through 8 on 14 mesh)(3) 70 per cent clay, 30 per cent grog (through 8 on 14 mesh)

The results are plotted in Fig. 17 and indicate that, as the per-centage of grog increases, initial strength and modulus of elasticitydecrease, while strength after thermal shock increases to a maximumand then declines.

For the particular bodies tested, 15 per cent grog gave less thanmaximum resistance to thermal shock, while 30 per cent carried thebody beyond maximum resistance. Examination of the curvesshows that 20 per cent grog would have given best resistance tothermal shock for the bond clay and size of grog tested.

V. GENERAL DIscussION OF EFFECT OF THERMAL SHOCK

17. Tentative Definitions.-The American Society for TestingMaterials has tentatively defined "spalling,"* with reference to re-fractories, as "Breaking or cracking of refractories to such an extentthat fragments are separated, presenting newly exposed surfaces ofthe residual mass." Comparisons of resistance to spalling are usuallymade by heating the specimens to a specified temperature, followedby quenching in air or water. The relative number of cycles requiredto spall off 20 per cent of each of the materials being compared indi-cates their relative resistance to spalling.

Such a test does not consider the effect of repeated thermal shockunder test conditions which are not sufficiently severe to cause com-plete failure.

In the present investigation, comparison of the resistance of claybodies to thermal shock has been made on the basis of the reduction intransverse strength produced by thermal shock.

*"Tentative Standards," A.S.T.M., 1927.


The most resistant body is considered to be the one which has thehighest transverse strength after thermal shock.

18. Comparisons with Data of Other Investigators.-Goodrich com-pared the resistance of various brands of firebrick to thermal shock bydetermining their loss in compressive strength. He pointed out that avitrified or brittle structure was the worst handicap for good resist-ance to thermal shock.

As pointed out by Parmelee and Westman, his correlation be-tween initial and final strength is good, the probable error of the aver-age percentage of the original strength which was retained being±3.1. This was attributed to the fact that comparisons were madebetween the two halves of the same brick. The small probable erroris quite large, however, when compared with the values of the per-centage of strength retained, which vary from 68.0 per cent to 85.7per cent, a range of only 17.7 per cent.

Parmelee and Westman were able to predict the per cent loss instrength of firebrick after thermal shock, from the initial strength,with a probable error of + 6.8, and measure the per cent loss instrength with a probable error of about ± 3.1. Their values for percent loss in strength varied approximately from zero to 86 per cent.

They concluded that, "It is evident that the stronger brands losta greater proportion of their strength when subjected to thermal shockthan did the weaker brands, although their final strength was, ingeneral, higher than that of the weaker brands."

It seems probable that the accuracy of their predictions may beattributed, partially at least, to the fact that, with two exceptions,the bodies are below or near maximum resistance so that the valuesfor initial and final strength lie in a comparatively narrow fieldbelow the 45 deg. line.

These particular conditions will not exist for all groups of bodieswhich might be tested, and for this reason their assumption of astraight line relationship will not have general application.

It is evident also from the mathematical relationships that, asthe strength of the brick increases, the final strength of the strongerbrands will not continue to be greater than the final strength of theweaker brands.

Heindl and Mong in their work on sagger bodies concluded thatthe resistance of saggers to thermal shock "decreases with increasedtemperature of firing saggers" and that "there is a direct relationbetween modulus of rupture and modulus of elasticity."


Obviously, these conclusions contradict their empirical formulawhich states, in part, that resistance to thermal shock is directlyproportional to modulus of rupture and inversely proportional tomodulus of elasticity.

Heindl and Pendergast in their work on firebrick concludedthat "the modulus of elasticity and transverse strength increase asthe temperature of firing of the brick is increased" and that "resis-tance of firebrick to spalling on an average decreased with increaseof modulus of elasticity... "

Again, the conclusions obviously contradict the empirical formula,devised as a result of the study, which states, in part, that resistanceto thermal shock is directly proportional to modulus of rupture andinversely proportional to modulus of elasticity.

19. Summary of Results of Other Investigators.-Goodrich con-tended that high firing and the development of vitrification decreasedresistance to thermal shock, which agrees in general with the statedconclusions of Heindl, Mong, and Pendergast, but contradicts themathematical expression of their results, which, of course, theythemselves contradict.

Parmelee and Westman contended that resistance to thermalshock in general increased with increased initial strength whichcontradicts the statements of other investigators.

It has been shown in the present investigation that as initialstrength increases, resistance to thermal shock increases to a maxi-mum and then declines.

It is possible then that the disagreement in results may be ex-plained by the nature of the materials studied in each investigation,for it is evident that, for a limited field, resistance to thermal shockmight either increase or decrease as initial strength increases.

Parmelee and Westman studied carburetor brick of compara-tively low strength, while Heindl, Mong, and Pendergast studiedsagger bodies and firebrick which either developed higher strengthnormally, or were preheated at high temperatures before testing. Itseems probable that the carburetor brick in the first instance allwere below or near maximum resistance, while the sagger bodies andfirebrick in the other instances were beyond maximum resistance.

20. Improvement of Test.-The progressive effect of repeatedthermal shock must be determined before a comparison of bodies ofdifferent types can be made safely on the basis of a limited numberof thermal shocks. It is possible that a body of low initial and finalstrength with low per cent reduction in a standard test might have


a higher final strength than a body having high initial and finalstrength with a high per cent reduction in a standard test, if com-pared after a large number of thermal shocks.

Since vitrification is known to decrease resistance to thermalshock it is important to know how a body would behave whenvitrified to a degree corresponding to that which would be producedby service conditions.

The greatest difference between bodies, as determined in a ther-mal shock test which does not affect their vitrification, appears afterthey have passed their maximum resistance and are approachingvitrification. For this reason it is probable that comparisons ofbodies could best be made by determining their resistance through-out the range of burning from a very soft to a vitrified structure.

VI. SUMMARY

21. Summary.-Previously published conclusions of other in-vestigators are contradictory because they apply only to limitedfields in each case.

The results of the present investigation indicate that:(1) In general, initial strength, modulus of elasticity, and per

cent reduction in strength increase and porosity decreases as strengthafter thermal shock increases to a maximum and then declines.

(2) As grog is added to a clay body the initial strength and mod-ulus of elasticity decrease, while strength after thermal shock in-creases to a maximum, and then declines.

RECENT PUBLICATIONS OFTHE ENGINEERING EXPERIMENT STATIONt

Bulletin No. 183. Tests of the Fatigue Strength of Steam Turbine Blade Shapes,by Herbert F. Moore, Stuart W. Lyon, and Norville J. Alleman. 1928. Twenty-fivecents.

Bulletin No. 184. The Measurement of Air Quantities and Energy Losses inMine Entries. Part III, by Alfred C. Callen and Cloyde M. Smith. 1928. Thirty-fivecents.

Bulletin No. 185. A Study of the Failure of Concrete under Combined Com-pressive Stresses, by Frank E. Richart, Anton Brandtzaeg, and Rex L. Brown. 1928.Fifty-five cents.

*Bulletin No. 186. Heat Transfer in Ammonia Condensers. Part II, by AlonzoP. Kratz, Horace J. Macintire, and Richard E. Gould. 1928. Twenty cents.

*Bulletin No. 187. The Surface Tension of Molten Metals. Part II, by Earl E.Libman. 1928. Fifteen cents.

Bulletin No. 188. Investigation of Warm-Air Furnaces and Heating Systems.Part III, by Arthur C. Willard, Alonzo P. Kratz, and Vincent S. Day. 1928. Forty-five cents.

Bulletin No. 189. Investigation of Warm-Air Furnaces and Heating Systems.Part IV, by Arthur C. Willard, Alonzo P. Kratz, and Vincent S. Day. 1929. Sixtycents.

Bulletin No. 190. The Failure of Plain and Spirally Reinforced Concrete inCompression, by Frank E. Richart, Anton Brandtzaeg, and Rex L. Brown. 1929.Forty cents.

Bulletin No. 191. Rolling Tests of Plates, by Wilbur M. Wilson. 1929. Thirtycents.

Bulletin No. 192. Investigation of Heating Rooms with Direct Steam RadiatorsEquipped with Enclosures and Shields, by Arthur C. Willard, Alonzo P. Kratz,Maurice K. Fahnestock, and Seichi Konzo. 1929. Forty cents.

Bulletin No. 193. An X-Ray Study of Firebrick, by Albert E. R. Westman.1929. Fifteen cents.

*Bulletin No. 194. Tuning of Oscillating Circuits by Plate Current Variations,by J. Tykocinski-Tykociner and Ralph W. Armstrong. 1929. Twenty-five cents.

Bulletin No. 195. The Plaster-Model Method of Determining Stresses Appliedto Curved Beams, by Fred B. Seely and Richard V. James. 1929. Twenty cents.

*Bulletin No. 196. An Investigation of the Friability of Different Coals, by CloydeM. Smith. 1929. Thirty cents.

*Circular No. 18. The Construction, Rehabilitation, and Maintenance of GravelRoads Suitable for Moderate Traffic, by Carroll C. Wiley. 1929. Thirty cents.

Bulletin No. 197. A Study of Fatigue Cracks in Car Axles. Part II, by HerbertF. Moore, Stuart W. Lyon, and Norville J. Alleman. 1929. Twenty cents.

*Bulletin No. 198. Results of Tests on Sewage Treatment, by Harold E. Babbittand Harry E. Schlenz. 1929. Fifty-five cents.

*Bulletin No. 199. The Measurement of Air Quantities and Energy Losses inMine Entries. Part IV, by Cloyde M. Smith. 1929. Thirty cents.

*Bulletin No. 200. Investigation of Endurance of Bond Strength of Various Claysin Molding Sand, by Carl H. Casberg and William H. Spencer. 1929. Fifteen cents.

*Circular No. 19. Equipment for Gas-Liquid Reactions, by Donald B. Keyes.1929. Ten cents.

Bulletin No. 201. Acid Resisting Cover Enamels for Sheet Iron, by Andrew I.Andrews. 1929. Twenty-five cents.

Bulletin No. 202. Laboratory Tests of Reinforced Concrete Arch Ribs, byWilbur M. Wilson. 1929. Fifty-five cents.

*Bulletin No. 203. Dependability of the Theory of Concrete Arches, by HardyCross. 1929. Twenty cents.

tCopies of the complete list of publications can be obtained without charge by addressing theEngineering Experiment Station, Urbana, Ill.

*A limited number of copies of the bulletins starred are available for free distribution.

30 ILLINOIS ENGINEERING EXPERIMENT STATION

*Bulletin No. 204. The Hydroxylation of Double Bonds, by Sherlock Swann, Jr.1930. Ten cents.

Bulletin No. 205. A Study of the Ikeda (Electrical Resistance) Short-Time Testfor Fatigue Strength of Metals, by Herbert F. Moore and Seichi Konzo. 1930.Twenty cents.

*Bulletin No. 206. Studies in the Electrodeposition of Metals, by Donald B.Keyes and Sherlock Swann, Jr. 1930. Ten cents.

*Bulletin No. 207. The Flow of Air Through Circular Orifices with RoundedApproach, by Joseph A. Poison, Joseph G. Lowther, and Benjamin J. Wilson. 1930.Thirty cents.

*Circular No. 20. An Electrical Method for the Determination of the Dew-Pointof Flue Gases, by Henry Fraser Johnstone. 1929. Fifteen cents.

Bulletin No. 208. A Study of Slip Lines, Strain Lines, and Cracks in Metalsunder Repeated Stress, by Herbert F. Moore and Tibor Ver. 1930. Thirty-five cents.

Bulletin No. 209. Heat Transfer in Ammonia Condensers. Part III, by AlonzoP. Kratz, Horace J. Macintire, and Richard E. Gould. 1930. Thirty-five cents.

Bulletin No. 210. Tension Tests of Rivets, by Wilbur M. Wilson and William A.Oliver. 1930. Twenty-five cents.

Bulletin No. 211. The Torsional Effect of Transverse Bending Loads on ChannelBeams, by Fred B. Seely, William J. Putnam, and William L. Schwalbe. 1930.Thirty-five cents.

Bulletin No. 212. Stresses Due to the Pressure of One Elastic Solid uponAnother by Howard R. Thomas and Victor A. Hoersch. 1930. Thirty cents.

Bulletin No. 213. Combustion Tests with Illinois Coals, by Alonzo P. Kratzand Wilbur J. Woodruff. 1930. Thirty cents.

*Bulletin No. 214. The Effect of Furnace Gases on the Quality of Enamels forSheet Steel, by Andrew I. Andrews and Emanuel A. Hertzell. 1930. Twenty cents.

Bulletin No. 215. The Column Analogy, by Hardy Cross. 1930. Forty cents.Bulletin No. 216. Embrittlement in Boilers, by Frederick G. Straub. 1930.

Sixty-five cents.Bulletin No. 217. Washability Tests of Illinois Coals, by Alfred C. Callen and

David R. Mitchell. 1930. Sixty cents.Bulletin No. 218. The Friability of Illinois Coals, by Cloyde M. Smith. 1930.

Fifteen cents.Bulletin No. 219. Treatment of Water for Ice Manufacture, by Dana Burks, Jr.

1930. Sixty cents.*Bulletin No. 220. Tests of a Mikado-Type Locomotive Equipped with Nichol-

son Thermic Syphons, by Edward C. Schmidt, Everett G. Young, and Herman J.Schrader. 1930. Fifty-five cents.

*Bulletin No. 221. An Investigation of Core Oils, by Carl H. Casberg and Carl E.Schubert. 1931. Fifteen cents.

*Bulletin No. 222. Flow of Liquids in Pipes of Circular and Annular Cross-Sections, by Alonzo P. Kratz, Horace J. Macintire, and Richard E. Gould. 1931.Fifteen cents.

*Bulletin No. 223. Investigation of Various Factors Affecting the Heating ofRooms with Direct Steam Radiators, by Arthur C. Willard, Alonzo P. Kratz, MauriceK. Fahnestock, and Seichi Konzo. 1931. Fifty-five cents.

*Bulletin No. 224. The Effect of Smelter Atmospheres on the Quality of Enamelsfor Sheet Steel, by Andrew I. Andrews and Emanuel A. Hertzell. 1931. Ten cents.

*Buletin No. 225. The Microstructure of Some Porcelain Glazes, by Clyde L.Thompson. 1931. Fifteen cents.

*Bulletin No. 226. Laboratory Tests of Reinforced Concrete Arches with Decks,by Wilbur M. Wilson. 1931. Fifty cents.

*Bulletin No. 227. The Effect of Smelter Atmospheres on the Quality of DryProcess Enamels for Cast Iron, by A. I. Andrews and H. W. Alexander. 1931.Ten cents.

*Circular No. 21. Tests of Welds, by Wilbur M. Wilson. 1931. Twenty cents.*Bulletin No. 228. The Corrosion of Power Plant Equipment by Flue Gases, by

Henry Fraser Johnstone. 1931. (In press.)*Bulletin No. 229. The Effect of Thermal Shock on Clay Bodies, by William R.

Morgan. Twenty cents.

*A limited number of copies of bulletins starred are available for free distribution.

UNIVERSITY OF ILLINOISTHE STATE UNIVERSITY

URBANAHARRY WOODBURN CHASE, Ph.D., LL.D., President

The University includes the following departments:The Graduate SchoolThe College of Liberal Arts and Sciences (Curricula: General with majors,

in the Humanities and the Sciences; Chemistry and Chemical Engi-neering; Pre-legal, Pre-medical, and Pre-dental; Pre-journalism, HomeEconomics, Economic Entomology, and Applied Optics)

The College of Commerce and Business Administration (Curricula: Gen-eral Business, Banking and Finance, Insurance, Accountancy, RailwayAdministration, Railway Transportation, Industrial Administration,Foreign Commerce, Commercial Teachers, Trade and Civic SecretarialService, Public Utilities, Commerce and Law)

The College of Engineering (Curricula: Architecture, Ceramics; Architec-tural, Ceramic, Civil, Electrical, Gas, General, Mechanical, Mining, andRailway Engineering; Engineering Physics)

The College of Agriculture (Curricula: General Agriculture; Floriculture;Home Economics; Landscape Architecture; Smith-Hughes-in con-junction with the College of Education)

The College of Education (Curricula: Two year, prescribing junior stand-ing for admission - General Education, Smith-Hughes Agriculture,Smith-Hughes Home Economics, Public School Music; Four year, ad-mitting from the high school-Industrial Education, Athletic Coaching,Physical Education. The University High School is the practice schoolof the College of Education)

The School of Music (four-year curriculum)

The College of Law (three-year curriculum based on a college degree, orthree years of college work at the University of Illinois)

The Library School (two-year curriculum for college graduates)

The School of Jburnalism (two-year curriculum based on two years ofcollege work)

The College of Medicine (in Chicago)

The College of Dentistry (in Chicago)

The School of Pharmacy (in Chicago)

The Summer Session (eight weeks)

Experiment Stations and Scientific Bureaus: U. S. Agricultural ExperimentStation; Engineering Experiment Station; State Natural History Sur-vey; State Water Survey; State Geological Survey; Bureau of Educa-tional Research.

The Library collections contain (May 1, 1930) 836,496 volumes and 221,800pamphlets.

For catalogs and information addressTHE REGISTRAR

Urbana, Illinois

I

Documents

Effect of thermal shock on clay bodies : a report