Extension cycling of sealants

(1)

READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE. https://nrc-publications.canada.ca/eng/copyright

Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.

Questions? Contact the NRC Publications Archive team at

PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.

NRC Publications Archive

Archives des publications du CNRC

This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at

Extension cycling of sealants

Karpati, K. K.

https://publications-cnrc.canada.ca/fra/droits

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

NRC Publications Record / Notice d'Archives des publications de CNRC:

https://nrc-publications.canada.ca/eng/view/object/?id=8048d241-90bb-487d-9153-b753fce23002 https://publications-cnrc.canada.ca/fra/voir/objet/?id=8048d241-90bb-487d-9153-b753fce23002

(2)

(3)

(4)

XII. FATIPEC KONGRESS

xllth

FATIPEC CONGRESS

XII"

CONGRES FATIPEC

Sonderdruck

Reprint

(5)

O Vcrlag Chemie GmbH, WeinheimiBergstr., 1974

Die Wicdcrgahe von Gehrauchsnarnen, Handelsnamen, Warenhezeichnungen und tlgl. berechtigt nicht zu der Annahrne, daB solche Narnen ohne weiteres von jedermann henutzt werden durfen. Vielmehr handelt es sich haufig urn gesetzlich geschiitzle eingetragene Warenzeichen, auch wenn sie nicht als solche gekennzcichnet sind.

The use of registered trademarks and materials designation< does by no means justify the assumption that such names can be freely used by everybody. Such names are often registered tradernilrls protected by law, even i f they are not expressly designated as such.

L'eniploi, dans cet ouvrage, de marques et noms diposi.s, designations de matcriaux etc. ne justifie pas utilisation par tous. 11 s'agit souvent de marques deposees et protCgCes par la loi, merne lorsqi~e nulle mention n'en est faite.

(6)

Karpati: Extension C y c l ~ n g 01 Sealants 455

Extension Cycling of Sealants

By K. K. Karpati*)

The amplitnrle anrl [he rate of daily joint movement in buildings was estimated from meteorological data. When model specimens of one-part chemically-clrring silicotte .senlat~t.s rvere cycler1 at these conr1itiotl.s otlrl at vnriolrs tetnperatlrres, the sealat~r ntanifesterl ruhhery ela.sticiry. If ntl itlitin1 estettsiort was itnposerl previolrs to the cycling ex~etl.siot~s, frrilrrre corrlrl be prorhrcetl. The re.s~r1l.s o f cycling experin1ettt.s have beett orrelaterl rvitlt the charocterisric fail~rre poitlls of tet~sile tests o t ~ .sitnilor specimetts. Thi.s rrllorvs a getleroliznriotr of the re.~rrlts otrrl also

rggesls how cycling rests for specificoliotl.~ corrlrl be tleriverl.

Cyclage en allongement des mastics d'etancheite

On a ivalrri l'at?~plitrrrle et la viresse rlrr ntorrvetnent qrrotirlien de joittts rle cottstrrrctiot~ srrr lrr base rles cot~rlitions clit?~ntir}rres. 0 1 1 a .solftni.s il rles cot~tmintes nlrerntes senthlal~les, ?I tlifftretltes tetnpiratrrres, rles iprolrverres-motleles cot~fectiot~t~ees ovec des rnnstics ?I lrn seirl cotnposrrnt

il base rle silicone el ri rl~rrcissetner~~ chimir/rre. Le tnastic tnor~ife,slnit role ilosticiti cnorrtchorrrerr.se. Toutefois, 1or.sclrr'on avail fflit srrbir rrt1e estettsiotl it~itinle art mastic ovnt~t le cycle tle cotrtrairttes, rrne rrrplrrre pouvnit se prorlrrire. Les ri.s~rllols rles e.ssni.s cyclic]rres ottt i t i tnis en relation avec les caructiristiqrtes rle ruptrcre des essais de traction effectuis sur ties iprorrvetres semblables. Ceci permet urte gtniralisation

1

rles ris1r1tat.s et forrrtlit rles it~tlicntions srrr In n~arlitre rlont les essais cyclic1rre.s pocrncriettt &re cot1r111it.s en ~ ~ r r e rle l'ita/~li.sseme~~t de spici- ficoriotu.

Ausdehnungs-Wechselbeanspruchung der Dichtungsmassen

Amplitrtde lcnrl Mall der Tagesbervegung von Fugen wurden auf Grunri tler klimatischen Verhiiltnisse geschatzt. Moriellprobenkorper von chenlisch nbl~it~rler~rletr I-Kotnporretttert rotrl Silikot~rlic/~trrrtg.st~~ns~set~ rvrrrrlett er~t.s~~rec/tertrlert Wec/~sel/~entrs~~rltc/~~rt~gett bei ve~sc/~ierlet~en Ten~peratrrretl nrrsgesetzt. Die Dicklrrr~g.stnnsse verhielt sich grrt~~t~tieln.stisc/t. A1.s (lie Dic/~trrt~g.smn.s.sc uher vor rler Wec/~sel/~entr.s/~rrrc/~rct~g

einer arlfiit~glichett Arisdehnro~g rrt~terrvotfet~ ~vrrrde, kont~tc eitt Brrrclz verlrr.scrc/rt ~verrler~. Die Re.s~rltnte rler z)~kli.scher Verslrclre wrrrrletl ,nit rletz chnmkteri.s~isc/tett Brrrcltketntrvertet~ rler Zrrgver.;rrche, (lie at1 ii/~trlicltet~ Prohen riltsgefiihrr ~vrrrrlert, in Beziehrrrtg gebrrrclrt. Dies enniiglicht cirri, Verallgen~eirrer~o~g tler Resrrltati~ ~rt~rl nrrclr eitten Hit~ivei.~, tvie iykli.sc/le Versrrclle fiir rlie Alrf.s~el/rrt~g vott Spe;ifikatiotret~ ettt~vickelt rverrlen kotltlen.

Curriculum vitae

Klara K. Karpati has been associated with the Division of Building Research of the National Research Council of Canada since 1965 where her ficlcl of interest has hecn the mechanical properties of sealants. Prior to joining D B R she was employed by I. C. I. (England) and Trimetal Paint Co. (Belgium). She obtained the Licence Speciale en Chimie Industriclle from the University of Brussels and graduated in chemistry from the University of Sciences, Buclapest.

*) Division of Building Research, National Research Council of Canada, Ottawa, Ontario, K 1 A O R 6 (Canada).

(7)

Introduction Summary of previous results

T h e development of indu\tnal~zed building systems has The specimens used for tensile

eclmens were the size of wall element\ increased from conventional small used because a sealant bead has a very complex stre\s held bricks, so did the problem of sealing joints between building

elements. This outside skin of the building undergoes large dimensional variations due to changing weather conditions. For small units, such as bricks, these variations may be very small ancl hy s sing conventional mortar between bricks the building can be made weathertight. With large units, however, the amount of movenient at each joint will increase ancl thus the jointing material must undergo consiclerable deformation without failure to keep the building weather- proof.

A s conventional materials could not meet the more se- vere requirements. new materials were clevelo~ecl. It is

that changes as a function of the aforement~oned primary var~able\. A systematic investigation of the chang~ng stres\ field starting with specimens that undergo only uniaxial de- tormation would be the approach used In convent~onal, b a w \ c ~ e n t i f ~ c work. The u\eful result5 ach~evecl, however, would ~~ncluly extend the time of inve\t~gation and still woulcl only demonstrate whether the material would fall cohesively in

practice. T h e moclel \pecimen, however, can fail cohes~vely o r aclhes~vely and can, therefore, fully answer the question whether failure will occur in practice. The model specimen i\ illustrated In F i g ~ ~ r e 1.

claimed that some of them remain weaterproof at movements up to + 2 5 % andlor - 2 5 % of the joint wiclth occurring across the joint. The sealants, as the new materials canie to be callecl, have to withstanel the maximum extension at the lowest outdoor temperatures ancl the maximum compression at the highest temperatures. Because the cle- mancling performance requirements of sealants have occa- sionally been combined with poor joint design, little know- ledge of the mechanical properties of sealants, ancl poor workmanship, failures of sealed joints have occurred and a thorough investigation of the problem became necessary. A stucly was undertaken at the Division of Building Res- earch of sealant properties and the movement imposed o n the sealant. This paper reports the latest results in the investigation of the mechanical properties of sealants.

Fig. 1 Model specimen

General considerations

The test that comes closest to the functioning of sealants in practice is a fatigue-type test. The specimen is subjected to compression at high temperature and to extension at low temperature. The following factors are variecl concurrently: amount and direction of stress, strain, ancl temperature.The rate of cleformation shoulcl also be varied if the aim is to establish the relation between laboratory testing and performance in practice.

As one can see, a f a t i g ~ ~ e test is very complex. T h e choice of test conclitions, as useel in specifications, is at present arbitrary. Consequently, by using many differentconclitions, one can approximate the practical performance to various degrees but unless the effect of each factor is known sepa- rately, one cannot say with certainty which set of conditions used in a fatigue test is the best approximation. A tensile test clescribes material behaviour in terms of primary variables: stress, strain, time, and temperature. Although a relaxation o r a creep test does the same ancl in a simpler fash- ion, the tensile extension test was chosen for this investigation because: (1) it is the least time-consuming; (2) most laboratories are equipped for it; ancl (3) among the various ways sealants can be loaded in practice, this type of loading is most likely to lead to failure. The results obtained with tensile tests have been published on one-part chemically- curing silicone sealants [ l ] and on two-part polysulfides [2]. For a better understanding of the progress reported in this paper, a brief summary of the pertinent results achieved previously is given here.

Model specimens were subjected to tensile testsatvarious strain rates and temperatures. The results were treated by the same basic approach as reporteel for s o m e polymers testecl in uniaxial tension 131. The feasibility of this approach has been statistically verified and found to be acceptable. T h e success of this approach to the problem made it possible to characterize the sealant behaviour in a simpli- fiecl manner ancl relate laboratory testing t o behaviour in practice.

It was found in tests made at various strain rates that the stress-strain curves reduced t o unit strain rate gave a single cumulative curve at each temperature. The resulting curves are iclentical for the silicone sealant stuclied, i.e., the material is inclepenclent of temperature within the range the sealant is expected to function.

The remaining variables s t ~ ~ d i e c l - stress, strain, ancl time -can represent the mechanical properties of t h e sealant in a three-climensional co-ordinate system. From the point of view of sealant performance the most important aspect of the three-dimensional representation is the projection of break points in the strain vs time plane. This projection is the key to performance testing because it includes testing time to break and time to reach failure in practice. T h e failure points are scattered along a straight line sloping towards longer times. If this regression line is extrapolated to six months, one obtains the extensibility of the sealant during a yearly cycle. This is a characteristic figure on which it was proposed to base the acceptability of the material. This type of plot obtained with a sample of the specimens

(8)

L O G T I M E , M I N .

Fig. 2 Time dependence of strain at break a t 22 "C

Experimental reference [-I] were ~15ed. It was found that the average daily T h e material usecl was a one-part chemically-curing silicone

sealant. The type of specimen lrsecl is illustrated in Figure 1. The size of the sealant bead is 1. 3 X 1. 3 x 5 . 1 cm ( ' l a X

' I 2 X 2 in.) and is supporteel on aluminum bars 1. 3 X 2. 5 X 7.6 crn ( ' I z x I x 3 in.). Special jaws werc prepared to grip both sicles of the specimen along its f ~ i l l length and toensure

operation without backlash c l ~ ~ r i n g cycling. T h e specimens were cast at 22.2 " C (72 OF) anel 5 0 % relative humidity on the primed substrate. They were curecl at the same conditions for two weeks, then subjecteel to accelerated aging at 70 " C t ₁" C (158 k 2 OF) in a ventilated oven for four weeks

Discussion

The sealant ~inclergoes climensional changes becauseofjoint movements induced by climatic conditions. The main climatic factor. temperature change, follows a sinusoidal pattern (both through the year and through each day) and so, in most cases, the sealant deformation also follows a sinusoidal pattern [4]. When the temperature is at a m a x i m ~ ~ m , the joint width reaches a minimum and vice versa.

The tensile part of the yearly cycle can be predicted from laboratory tests by the method summarized above. Thenext step necessary in the investigation of sealant behavioi~r is to examine the influence of daily cyclic movements on the tensile part of the yearly cycle.

T o be able to investigate the effect of the daily variations, one must know the average daily and the annual total temperature changes. These values vary with geographical re- gions and can be derived from meteorological records. For this work, the values nleasured previously and reported in

temperature change in the Ottawa region is ahout 1 1 OC (20 OF); the total yearly temperature change is 6 7 "C (1 2 0 OF). As the silicone seaIant is claimed to be able to stand up to

+

2 5 % movement of the joint iviclth in a yearly cycle. joints arc designed in such a way that this total yearly movement should not be exceeded. If a temperature difference of 6 7 " C (1 20 OF) is allowecl by the design toproduce _t 25 % movement in the joint, then a difference of 1 1 " C (20 OF) can procluce k1.2%. This value is only an approximate estimate of the daily movement and it was rounclecl to t4% for experimental purposes.

T h e rate of movement at which a specimen has to be cycled to acconlplish o n e cycle per day is determinecl by t h e t _{4 % extension anel the specimen width. This rate is} 0.000 14 cm/min, assuming a constant rate of movement. This rate is not available on tensile testing machines. F r o m the available rates close to this value, 0.0002 cm/min was chosen, which accomplishes the cycle in 16.0 hr.

Effect of daily cycles

Specimens were cycled at 22.2 t l . I " C (72 k 2 OF) for five cycles. T h e cycles gave identical curves well within t h e errors of the machine. T h e area of the loop was very small. T h e maximum width was about 2 % of the maximum load occurring in a cycle and within the accuracy of the machine. It could be reproduced, however, repeatedly. Repetition of the test with different specimens gave identical results ex- cept for a slight difference in the maximum load. T h e maxi- nlum loads ranged between 0.78 and 0.54 kg/crn2 both in tension anel in compression.

T h e tests were repeated at - 34 " C ( - 30 OF). T h e maximum loads reached in tension and in compression w e r e identical within experimental error at room temperature.

(9)

45 8 Karpat~: Extension Cycling of Sealants

At low temperature, the width of the loop was difficult to (0 OF) and at room temperatures. The results were identical estabish with accuracy. It was somewhat larger than that of within experimental error using the 5ame specimen and dif- the loop recorded at room temperature. On return to zero ferent specimens.

gauge length the specimen showed a slight compression load that relaxed to the initial load within a few hours.

These results illustrate that the material has rubbery elasticity at

*

4 % deformation and 0.0002 cm/min rate. This cycling does not affect the chemical bonds in the material, only the long chains are coiled to different degrees at various stages of the cycle. As the same behaviour and same maximum load value occurred at - 34 "C, the temperature- independent nature of the silicone sealant, as previously demonstrated [I], is confirmed here for the temperature range observed.

After establishing that the average daily movement cycle had no effect on the specimen,-the strain of the cycle was increased. The increase was based on meteorological data [5] for Ottawa showing that a maximum daily temperature change (drop or increase) of 19.4 "C (35 OF) can he experienced within six hours in return periods of 25 years. In Calgary the same temperature change would occur in about a five-year return period and 25 " C (45 OF) change in about a 15-year return period. If the yearly temperature change is 67 " C then, with the 25 " C daily temperature change a joint width change of

*

9.4% would occur. This figure was increased to ₂15% strain cycle based on the principle that if a larger movement does not produce permanent changes in the material properties then neither will smaller ones. With the increased strain of the cycle the rate of deformation must also be increased. At 0.0005 cmlmin crosshead speed available on the machine one cycle was accomplished in 25.4 hr: very close to one f u l l day. The test results illustrated that the area enclosecl by the hysteresis loop remained negligible in spite of the increased strain although the loads went up to 2.2 ancl 2.8 kg/cm2 for the maximum tensile and the maximum compressive load respectively. Even this large amount of strain, i~nlikely to occur in practice in a daily cycle, did not cause permanent changes in the material behaviour.

Up to this stage of the work only unloaded specimens were subjected to cycling at the estimated daily rate. As no permanent property changes occurred in the material, the daily cycling was imposed on the specimens at the extreme conditions expected in a yearly cycle.

It is claimecl that a silicone sealant will maintain

*

25 % joint movement in building joints without failure. As failure is more likely to occur in the tensile part of the yearly cycle, the specimens were subjected to 25% extension at - 31 OC ( - 3 0 ° F ) and, at this initial extension, a *4% strain cycle was imposed making the specimen cycle between 29 and 2 1 %. The specimen was extended to 25 O/c at 0.02 cmlmin rate and held at this extension until relaxation, i.e., uncoil- ing of the long chains, tapered off. Thirty minutes were allowed for relaxation because the load drop that occurs between the 20th and 30th minutes was within reading error. The load reached at 25% extension was in the order of 3 kg/cm2, decaying by about 0.16 kg/cm2 in half an hour.

The

+

25 2 4 % strain cycles gave similarly narrow loops as the experiments without the 25% initial extension. The maximum loads were also of the same orcler (calculated as the difference between the load at the tip of the loop and at the end of the 'I2 hour relaxation). After return to zero extension a small compression load developed but disappea- red within a few hours. The

+

25

*

4 % extension cycles were repeated several times at - 40 C ( - 40 OF), - 18 O C

Cycling to reach failure

Because neither load nor temperature gave any ba5is for predict~ng failure, further ~nvestlgation of strain and time was required with temperature constant at 22 'C. It was also necessary to ascertain when the number of cycles is suf- ficient to stop the experiment. In this respect an e5timate can be made assuming sinusoidal movement.

During the yearly cycle the jo~nt w~dth will be between the maximum joint width and 87% of that width within a period n/3 t ~ m e s the 365 daily, i.e., 60.8 cycles. As the accelerated heat aging increases the property change equivalent to about 2'1, years (based on I0 " C doubling the reac- tion rate), 120 cycles (2 years) were attempted acd, if reached, cycling was stopped and the specimen considered pass the test.

In a first series of tests both the initial extension and the rate of cycling (time factor) were increased experimentally until failure was produced. Attempts to increase the initial extension to 50% were i~nsuccessful because the majority of the specimens had failed at lower extensions. The initial extension had to be lowered to 30% to produce failures during cycling. The extension of the cycles was progressive- ly increased from

*

4 % to

+

8 and to 2 12%. The rate of cycling was accelerated sim~~ltaneously, first to achieve the increased extension per day, then to produce failure within a time considered reasonable to spend on one experiment. Extension rates used were 1 .O, 0. l and 0.0 1 cm/min. At these rates 120 cycles are accomplished within 73 min.,

12 hr., and 5 days, respectively, using

+-

12% cycles. The most important conclusion drawn from this phase of the work (consisting of 25 tests ancl 1000 cycles in total) was that one shoulcl aim at cycling conditions wherechances of failure are 50%, because if all of the specimens fail, or if none fails there is no indication of the safety limit of the material. The large scatter in the reproducibility of failures also made it necessary to aim at a 50% failure using a large number of specimens.

With the experience gained, a series of tests was planned to approximate conditions that woulcl give failures in the orcler of 50%. Table 1, which summarizes the results, shows that the extension cycles at

+

30

+

12% gave too many failures at both rates of movement and lowering the initial extension to

+

25% with

+

12% cycles did not procluce a significant improvement. The extension cycles, therefore, were further lowerecl to k 8%. At this cycle and at the rate of 1.0 cmlmin over 50%. of the specimens did not fail. At 0.1 cmlmin the number of failures was significantly higher. The same increase in the number of failures with the de- crease in the rate of movement can be observed at

+

30

2 12% cycles.

Most failures were cohesive and most cohesive failures started adhesively. The few specimens that failedadhesively dicl so before cycling began. At the slower rates thecohesive failure was propagated closer to the interface. The specimen was considered failed when a split of about 1 cm long occured; shorter split cloes not necessarily lead to failure. The increased number of failures with decreasing rates of cycling can be understood if the characteristic log strain vs log time plot of the breakpoints is prepared for the material

(10)

Karpati: Extension Cycling of Sealants 459

[I]. Eighteen specimens were subjected t o tensile tests, the represented by the filled circles of Figure 2. The correspon- break points of which a r e illustrated with circles in Figure 2. ding best fitting line and the 95 % confidence limits a r e The centre line of the three continuous lines of the figure is drawn with dashed lines. These specimens do not possess the best fitting line of the break points. T h e outer continu- the flaws of the average rpecimen and, therefore, they all fit ous lines represent the 95 % confidence limits. in the upper half of the area delimited by the 95% con- O n e can approximately locate the cycling tests of Table 1 fidence level of the randomly chosen samples. The slope of in this figure. T h e breakpoints a r e arranged in five vertical the regression line is, however, about the same.

Table 1 Cycling at 1.0 and 0.1 cmlmin

Conclusions

Rate, cmlmin Number of Specimen\

Total Failed Failed Passed T h e sealant tested manifested rubbery elasticity when before between 120cycles loaded in cycles equivalent to the daily cycles experienced 5th cycle 5and120th in building joints. This elastic behaviour was observed cycle _{whether the cycles were imposed on unloaded specimens o r}

Initial and cycling added to the maximum yearly extension and was tempera-

extensions: _{ture independent in the temperature range ob5erved. A s}

+

30 -t 12% _{the elastic behaviour provides no guidelines for prediction}

1.0 23 10 9 4

0.1 27 21 5 1 of failure, cycling conditions were sought that produced

+

25 -t 12% failure.

1.0 16 4 9 3 It was estimated that 60 daily cycles per year occur a t a

+ 2 5 - t 8 % joint width that is 87% o r more of the maximum yearly

1 .O 14 2 3 9 joint width. T h e artificial aging of t h e specimens tested 0.1 15 4 8 3 was in the order of 2 years, thus 120 daily cycles were im- posed on specimens that had previously undergone an initial extension equal to the yearly maximum extension. rows slightly tilted to the right. They correspond to test During the course of the experiments it became clear that rates of 1.0,0.1, 0.01, 0.001 and 0.0001 cmlmin going from one had to aim at approximately 50% failure to locate t h e left to right. The extension cycle that produced failure inthe safety limits of the material in cycling experiments. T h e range of 50% was the +25 f8% cycle. If the f 25% initial extension and rate of cycling necessary to achieve this was extension is located on the vertical rows of failure points of related to a plot previously determined as characteristic of the 1.0 and 0.1 cmlmin rates, a line connecting the two a material and a batch

[I].

This is the log strain vs log time

25% points crosses the lower confidence level. This ex- plot of the break points of tensile tests. The points can b e plains why at the lower rate the majority o f specimens fitted with a straight line with 95% confidence limits. T h e failed, while a t the higher rate more than 50% passed 120 lower of these can be considered the safety limit for t h e cycles without failure. It becomes equally clear that few o f material. T h e cycling tests that produce failures in the o r d e r the tests at 30% initial extension can pass, especially at the of 50% had an initial extension that falls on the lower con- lower rate of cycling. fidence limit within a few percent, with thecyclingextension O n e can coliclude that to produce an indication of the penetrating not more than half way to the best fitting line. safety limitinacycling experiment, the initial extension cho- T h e rate of cycling was found to be 1.0 and 0.1 cmlmin t o sen has to be close to the 95% lower confidence limit. The finish 120 cycles within a time acceptable in a laboratory extension used for cycling should not penetrate more than experiment. T h e characteristic plot and, therefore, the cy- approximately half of the dirtance between the regression cling that produces about 50% failures varies from batch t o line and the lower confidence limit. T o stay within the time batch. A cycling test that any batch should pass, and there- acceptable in a laboratory experiment with the 120 cycles fore be suitable for specifications, should be established by given, one has to confine the rates to 1.0 and 0.1 cmlmin. investigatingbatch-to-batch variations.

A s batches vary, the cycling extensions should vary ac- cordingly. In fact the batch used in these tests would pro-

bably be regarded as a poor one by the manufacturer. This _{Acknowledgement} does not affect the conclusions, however, because they are

related to the characteristic regression line o f the break T h e author is indebted to Mr. L. R. Dubois for his careful points and its confidence limits. For example, t o produce and indefatigable work of specimen preparation.

the same number of failures with the batch used in the This paper is a contribution from the Division of Building work of reference

[I],

the

+

25 f 8% cycling should be Research, National Research Council of Canada and is replaced by approximately the

+

45 f 15% extension cy- published with the approval of the Director of the Division. cles.

The batch-to-batch variations and the corresponding variations required in the extensions of the cycles should be

part of a separate study. Based o n this study, o n e could References

choose the constant cycling extensions necessary for speci- _{[ I ]}_{K ,}_{K , K} _~ _~_{J ,}_{paint ~}_~ _~ _~_~_, _~_{44,75 (1972)}_: _h _~ _~ _l _, fication purposes, at which n o failure would be allowed [2] K . K . Karpnti. J . Paint Techno]. 45,49 (1973) during cycling of any batch. [3] T. L. Smith. J . Polymer Science XX, 89 (1956)

As a matter of interest, the break points of tensile tests

-,

L41

K . K . K f l r ~ a ' i and E. '. 'lbbons: Mat'1s. Res. and Std. lo, l 6 (1970)

O n 'pecimens that had 120 _{[5] G.}_R._{Kendoll: Conference on Ice Pressures AgainstStructures,}