INDUSTRIAL AND ENGINEERING CHEMISTRY Brooks, B. T., J . Am. Chem. Soc., 56, 1998 (1934). Brooks, B. T., U. S. Patent 1,879,599 (1932). Ibid., 1,885,585 (1932) and 1,919,618 (1933). Ibid., 1,894,661 (1933). Ibid., 1,904,200 (1933). Brooks, B. T ,and Cardarelli, E., Ibid., 1,919,617 (1933). Brooks, B. T.. and Humphrey, I. W., J. Am. Chem. Soc., 40, 822 (1918).
Buc, H. E., and Clough, W. W., U. S. Patent 1,726,946 (1923). Butlerow, A., Ann., 144, 1 (1867): 1 8 0 , 2 4 5 (1875); Konowalow, D., Ber., 13, 2395 (1880). Curme. H. R., U. S.Patent 1,339,947 (1920). Davis, H. S., and Schuler, R., J . Am. Chem. Soc., 52, 721 (1930). Drushel, W. A., and Linhart, G. A,, Am. J . Sci., 32, 51 (1911). Ellis, C., Chem. & Met. Eng., 23, 1230 (1920). Ellis, C., “Chemistry of Petroleum Derivatives,” p. 301 et seq., New York, Chemical Catalog Co., 1934. Ellis, C., and Cohen, M. J., U. S. Patent 1,486,646 (1924). Engs, W., and Moravec, R. Z., Ibid., 1,864,581 (1932) and 1,912,695 (1933).
Fischer, E., and Speier, A,, Ber., 28, 3252 (1895). Fritzsche, P., Chem. Ind., 20, 266 (1897); 21, 33 (1898). Johannsen, O., and Gross, O., U. S. Patent 1,607,469 (1926). Keyes, D. B., Science, 77, 202 (1933). King, A. T., J . Chem. Soc., 115, 1404 (1919). Kreman, R., Monatsh., (3) 28, 13 (1907); 31, 165 (1910); 38, 5 3 (1917).
Lebo, R. M., U. S. Patent 1,865,024 (1932); Mann, M. D., and Williams, R. R.,Ibid., 1,365,043 (1921). McElroy, K. P., U. S. Patent 1,438,123; cj. Distillers Co., Ltd., British Patents 368,051 (1932) and 370,136 (1932); Rochlingsche Eisen v. Stahlw., British Patent 238,900 (1925). Maimeri, C., British Patent 215,000 (1924).
Vol. 27. No. 5
Menschutkin, N,,Ann., 197, 195 (1879). Merley, S. R., and Spring, O., U. 5. Patent 1,859,241 (1933); Basore, C. A., Ibid., 1,385,515 (1922). Nametkin, S., and Abakumovskaja, L., Ber., 66B, 358 (1933). Ormandy, W. R., and Craven, E. C., J . SOC.Chem. Ind., 47, 317T (1928).
Plant, S.G. P.. and Sidgwick, N., Ibid., 40, 14T (1921). Plimmer, R. H. A., and Burch, W. J. N., J . Chem. Soc., 1929, 288.
Rice, F. O., “Mechanism of Homogeneous Organic Reactions,” p. 117, New York, Chemical Catalog Co., 1928. Sanders, F. J., and Dodge, B. F., ISD. ENQ. CHEM..26, 2 0 8 (1933).
Senderens, J. B., and Aboulenc, J., Compt. rend., 152, 1672 (1911).
Simington, R. M., and Adkins, H., J . Am. Chem. Soc., 50, 1449 (1928).
Strahler J., and Hachtel, B., Brennstof-Chem., 15, 166 (1934). Taveau, R. M., U. S. Patent 1,845,007 (1932). Thomas, C. A., and Carmody, W. H., IND. ENG.CHEM., 24, 1125 (1932).
Ullmann, “EnsyclopLdie der technischen Chemie.” 2nd ed., Vol. 1, p. 717, Berlin, Urban & Schwarzenberg, 1928. Valette, F., Chimie & Industrie, 13, 718 (1925): Cie Bethune. British Patent 221,512 (1925); Compton, J., U. S. Patent 1,598,560 (1926).
Wade, J., J . Chem. Soc., 87, 1657 (1905). Weismann, C., and Legg, P. A,, U. S. Patent 1,408,320 (1922). Whitmore, F. C., IND.Esa. CHEM.,24, 1125 (1933). Williams, R. R., and White, D. H., U. S. Patent 1,460,878 (1923).
RECEIVED September
6, 1933.
Contribution of the Standard Alcohol
Company.
Initial Inflammability of Construction A method of testing the flaming tendency of V Materials combustible materials has been developed which G. E. LANDTAND E. 0. HAUSMANN Continental-Diamond Fibre Company, Newark, Del.
T
HE selection of slow-burning materials for fire-resistant construction is of paramount importance to the engineer and architect. When possible, he eliminates the u8e of combustible materials, yet structural and use requirements are often such that he must rely on natural or manufactured materials which are combustible. From time to time empirical tests have been devised which are designed to permit the selection of materials on the basis of their flame-producing or flame-carrying properties. Such a characteristic test is given by the Navy Department and by the New York City Building Code (1); the Navy Department specification is as follows: A No. 2 Meeker gas burner shall be used. The flame shall be ‘/,inchin diameter and 4 inches high. The s ecimen shall be held horizontally, 3 inches above the to of t i e gas burner. The flame shall be applied to one corner orthe specimen (in the case of square rods) or to one side of the specimen (in the case of round rods and tubes) about 2 inches from one end. After the specimen has become well i ted it shall be withdrawn from the flame. The length of time E i n g which the material supports combustion after being withdrawn from the flame shall be determined.
The variables affecting the end result are very difficult to The fist requirements for the flaming test are that the rate of application of heat shall be constant and the rate control.
permits a classijlcation of these materials on the basis of this tendency. Data are presented to show that the spec$c heat of unit volume, the specific heat conductivity, and the existence of microscopic channels play a n important part in the rapidity with which flames derelop at the surface of combustible materials when subjected to intense heat. of heat dissipation by convection, conductance, and radiation shall be constant so as to permit a uniform rate of temperature increase on the sample under test. Such a method makes no allowances for the variables which may affect the end result, the accuracy with which gas flame can be regulated, the difference of B. t. u. contents of different gases or the influence of the surrounding air on the behavior of the sample, particularly the influence of air currents which may, depending on the circumstances of the test, exert an inhibiting influence on the burning, or may, on the other hand, stimulate it excessively. Truax and Harrison (2) have attempted t o standardize these variables. A simpler and more direct method of overcoming these objections was desired: The requirement of a specific and closely controllable heat source is achieved by passing a 110-volt current through a 7foot length of No. 30 B & S gage, nichrome wire (6.5 ohms per foot). The possible variations in heat developed by this wire are 2 per cent. Either alternating or direct current may be used, and, when necessary, the voltage is held a t the required value b y means of a rheostat in the circuit [Weiss and Price (3) acquired
March, 1935
INDUSTRIAL AND ENGINEERING CHEMISTRY
the same objective bJr means of a specially constructed furnace which was held to definite temperatures. Price likewise determined the time required for this sample to ignite.] The wire is wound about the sample under test. This prises a test piece I / , ~ x 1 x 10 inches. S uare grooves are milled into the edges over a 4-inch c e n t r j portion of the ~ ~ in ~ width, piece. They are '/32 inch in depth and ' s / ~inch and are spaced so as to permit ten turns of wire to the inch. The nichrome &.ire is lvound tightly through these grooves and is then fastened to binding posts at the ends of the sample. The width and depth of the sample are held to an accuracy of 6 / ~ ~ o o inch.
rials may be considered to take place in three separate steps. The first of these consists of a destructive' distillation which results in the production of gaseous material and of a residue consisting principally of carbon. The second phase consists of the union of oxygen with the carbon to form carbon manoxide, The third phase consists of the combination of oxygen with inflammable gases-namely, those which have been formed by destructive distillation as well as with the carbon monoxide formed in the second stage of burning. The phe-
TABLEI. IGNITION PERIODS OF VARIOGS MATERIALS 7 -
MATERIAL
COMPOSITION
GRADE
289
-1QNITION PERIOD, SECONDSNo. 2 No. 3 No. 4 No. 5
SP. GR.
No. 1
1.48 1.39 1.36 1.35 1.34 1.34 1.35 1.88 1.85
40 40 117 114 92 75 92 153 109
40 38 106 117 107 85 90 148 128
37 36 111 117 102 97
1.44 1.46 1.19 1.26 1.09
37 105 32 28 25
57 104 16 27 24
60 110 27 29 25
Mean
DEVIATION
% Hard Fiber Hard Fiber Dilecto Dilecto Dilecto Dilecto Dilecto Dilecto Dilecto Vulcoid (cotton paper) U. F. Dilecto Hard rubber Pressboard Pressboard a
2'/2 3 X
xx XP C L A
.$A 21/:
. .. . .. ... ...
Hard Fibre and synthetic resin Cellulose and urea resin Rubber, compn. unknown Cellulose plus size Cellulose DIUS size
90
148 130
42 37 114 123 102 91 91 157 130
41 37 127 122 99 83 89 150 122
58 109 I9
57 109 28 28
27 25
24
40
38 115 119 101
86
90 151 124 58 107 24 28 25
Av. 4.5 max. 7.6 A V . 3.4: max. 5.2
4v. Av. Av. Av. Av. Av. Av.
4.9, 2.7, 3.8, 7.2, 0.9, 2.0, 5.3,
max.lO.4 max. 3.3 max. 8.9 max.12.8 max. 2.2 max. 4.0 max.12.0
Av. 1.7, rnax. 3.4 Av. 1.1, max. 2.7 Av.23.2 max.32.5 Av. 2.1: max. 3.6 Av. 0.83,max. 4.2
Synthetic resin is the cresylic formaldehyde type.
The sample supported a t both ends is placed in a protecting container to shield it from air currents. This is a box 12 X 24 X 16 inches high, with the front and top removed. The container may not be completely closed, as the gases produced during the test may prevent the access of air to the sample. To permit this, and yet to guard against disturbing air currents, the box with the sample mounted in it is placed in a hood with the open side of the box toward the front. The front sash of the hood is then drawn down until a 2-inch opening remains a t the bottom. This permits a gentle stream of air t o pass over the surface of the sample, thereby supplying it with the necessary oxygen and concurrently enabling the products of decomposition and combustion to escape from the surface of the sample itself. A forced draft on the hood is not necessary. Samples are conditioned for 24 hours at 105' C. before test in a suitable oven. The test is started by turning on the current. The time from the snapping of the starting switch to the point where the sample bursts into flame is noted on a stop watch. This point has been very definite with all of the materials tested. The time required in seconds to bring about the production of a continuous flame is taken as a measure of the flame-carrying tendency of the material. Before the material actually bursts into flame, it shows more or less evidence of decomposition with the production of fumes and carbonization. But, inasmuch as flames cause the rapid communication of fire through a burning structure, the time
nomenon, flame, arises when the temperature at which the gases unite with oxygen reaches the temperature where luminescence takes place. The engineer's first concern is the development of this luminescence or flame, since it is the means of rapidly communicating a fire from one point to another in a burning structure. It may be necessary for him to consider other factors than this in selecting his materials, as various materials will show different characteristics during the disintegrating process that goes on during burning. Some will burn more rapidly than others, some with a more unquenchable flame. Some will develop a higher heat of combustion. This investigation is not concerned with these factors but with the determination of the relative rapidity with which different materials will develop flames; for the reasons previously mentioned. The kindling temperature of a given material is of great importance in determining its relative inflammability (3). Without further demonstration of proof we may accept the conclusion that a relatively high kindling temperature will be accompanied by low flame susceptibility; yet it will be shown that a measure of the kindling temperature of a material does not adequately express its tendencies to develop flames. The more direct method of measurement described here appears to be of greater value to the engineer.
TABLE11. INFLUENCE OF SPECIFICGRAVITY ON IQNITION MATERIAL Hard Fiber (parchmentized) Hard Fiber (parchmentized) Hard Fiber (parchmentized)
COMPOSITION Cotton cellulose Cotton cellulose Cotton cellulose
SP. GR. 1.19 1.32 1.47
required for the material to burst into flame is considered the more important end point of the test. The results obtainable by this method are indicated by the data in Table I. Various materials used in construction and for electrical insulating purposes were tested in this manner. These are classified as inflammable by the existing methods of test. Five samples of each material were tested in order to determine the reproducibility of results.
THEORETICAL CONSIDERATIONS We may broadly consider combustible materials to be organic in their composition. The combustion of such mate-
ASH 0.74 0.54 0.53
No. 1 33 35 46
IQNITION PERIOD,SECONDS No. 2 No. 3 No. 4 No. 5 31 37 43
31 37 47
30 36 41
26 36 42
Av. 30 36 44
A significant example of the influence of kindling temperature is given in Table 11. Each of the samples represents material that comprises cotton fiber with only such amounts of foreign substance as are incidental to the process of manufacture. The attempt to determine the kindling temperature of these samples led to the conclusion that such a determination is untrustworthy. Assuming, however, from their similar composition that the kindling temperatures of the material are the same, we observe that one set of samples breaks into flame much more quickly than the others; that, in fact, the time required seems to be closely related to the specific gravity of the material.
INDUSTRIAL AND ENGINEERING CHEMISTRY
290
Let us consider briefly additional factors which may affect the time required. for a given material to reach its kindling temperature. If we assume that the material is heated a t a constant rate and that environmental influences such as fluctuation in air pressure, concentrations of oxygen, and temperature of the surrounding air are constant, we must expect that the time required for the material t o reach its kindling temperature will be determined by the rate of temperature increase; flaming, or burning, takes place when the gases leaving the surface reach the temperature of incandescence, or
FIGURE1. EFFECTOF SIZE OF SAMPLEON IGNITION the kindling temperature. Such a temperature will be attained more rapidly if the specific heat of the material is low. It will not be attained unless the rate a t which heat leaves the surface under test is less than the rate a t which heat is supplied to that surface, since this is the requisite condition for building up the temperature a t that surface to the kindling temperature. The thermal conductivity of the material will, therefore, contribute to the end result, since a material with a naturally low conductivity will provide a suitable temperature gradient more quickly than a material of high thermal conductivity. The size of the sample tested will likewise affect the time required to establish a suitable temperature gradient within the sample, because the heat input into a large sample must necessarily be greater for the surface temperature to reach the required value. Table I11 and the curves reproduced from it in Figure 1 illustrate these points very well.
Vol. 27, No. 3
nation of such a concentration is surrounded with exceptional experimental difficulties. It may be considered that such concentrations must acquire certain minimum values before flaming can take place.
DISCUSSION OF RESULTS The probable maximum error in measuring the time required for the sample to burst into flame is 0.5 second. Inspection of the results indicates that variations generally exceed this amount. However, the average deviation of the mean value for flaming is a matter of a few per cent, and the maximum deviation of any one observation from the average, though occasionally reaching a magnitude of 8 to 10 per cent, is usually a matter of a few per cent. The one notable exception to this occurs in the tests made on hard rubber. The series of results covering the observations made on Dilecto, grades X, XX, XP, C, and L, is of special interest. With the exception of grade XP the chemical compositions of these materials may be considered to be identical; the material is built up of paper or fabric (cellulose) impregnated with a phenolic synthetic resin. Grade XP has a similar composition, but a slight change has been made in the character of the synthetic resin. Yet, in spite of the similarity of their composition, the results of the observations show a marked difference in their respective inflammabilities. If certain facts are taken into consideration, these differences may throw light on the process which we know as combustion. The principal difference between these grades is the physical character of the cellulose entering their composition. Grade XX as compared with X and XP is of such a nature that the resin penetrates the cellulose fiber much more completely. I n grade L penetration is still further inhibited by the fact that the fibers are spun into threads. I n grade C
T
-' E
@
20
I
I I
0.05 at0 a15 azo Speci/ic Heat per Unit Volurna
Q25
FIGURE2. EFFECTOF SPECIFICHEATON IGNITION
OF SIZEOF SAMPLE ON IGNITION TABLE111. EFFECT
SECTION AREA
CROSS
So. cm.
IQNITION PERIOD Dilecto Hard Fiber (grade XX) (grade 21/z) Sec.
..
0.219 53 0.205 111 0.414 183 0.816 251 1.24 1.52 275 1.63 Thermal conductivity, tal./' C./ oc./sec. 6.5 X 10-1
SEC.
21.0 39:s
..
47:s
3.0 X 10-4
The tests conducted on samples of increasing size required progressively longer time intervals to reach the end point. That material with the lower thermal conductivity established a maximum constant flaming time more quickly because the low conductivity did not permit the dissipation of surface heat through the body of the sample so readily. An equally important factor, but less tangible in possibilities of evaluation, is the concentration a t which the combustible gases occur a t the surface of the material. The determi-
the threads are coarser than in grade L, and penetration is still less complete. We may consider that incomplete penetration of the fibers leaves microscopic channels in the structure of the material, and that an abundance of such channels enhances the flaming characteristics of the materials because they permit the gases formed by decomposition to reach the surface of the sample more readily and there build up the concentration requisite for combustion more quickly. The inflammability of these samples, therefore, decreases as the channeling or incompleteness with which the fibers are impregnated decreases. An indication of the relative channeling in these grades is given by the following data on 24-hour water absorption according t o A. S. T. M. method 229-32T: GRADE
HIO ABSORPTION
%
XX x XP L C
1.2 2.3 3.0
3.6 4.0
IXDUSTRIAL AND EKGINEERING
March, 1935
The close parallelism between inflammability and the imbibition of water where channeling again influences the end result further justifies the conclusion that such a channeling imparts the ready susceptibility to develop flames. The results obtained in Table I1 further bear out this conclusion, as the samples of lower specific gravity must for that reason have a higher porosity, thereby promoting the flaming tendency of the sample. However, this difference in specific gravity may have an independent influence on the flaming quality, as later discussions will show. The possibility of differentiating between the flaming characteristics of materials led to a further extension of these studies to some of the woods used in construction. The results are given in Table IV.
MATERIAL Masonite Balsa White pine Yellow pine Maple Oak (white)
-
SP. GR. d
CONTENT
SP,HEAT
PER a R A w
PER
%
1.04 0,094 0.31
2:i5 3.42 3.68 3.15 3.50
0.55 0.59 0.73
CC.
0.333 0.03008 0.1026
0.32 0.32 0.331 0.337 0.327 0.331
0.1854 0.1929 0.2416
This should follow from considerations related to the specific heats of woods. I n each test shown in Table IV the same volume of wood is being used. The weight of wood undergoing test increases directly as the specific gravity. The rate of temperature increase of the samples, since the specific heats are approximately constant, therefore, increa3es in direct proportion t o the specific gravity. It follows that the time required to reach a certain end temperature will increase in a similar ratio. The corollary t o this statement is that the inflammability of materials of the same approximate chemical composition will vary directly with the specific heat of unit volume of the material as shown in Figure 2. TABLEV. EFFECTOF MOISTURECONTENTOF WOODS ON IGNITION MOISTURE --IQNITION MATERIAL CONTENTNo. 1 No. 2
% 8.80 11.31 11.22 10.57 11.46
12
14.2
29 33 36
291
The pronounced effect of small amounts of moisture present in the wood in reducing inflammability is adequately shown in these tests, although the concordance of results is poore,. than that thus far experienced. The precise role that moisture plays in retarding the burning of the samples is an open question. Some heat is undoubtedly required to evaporate the moisture. An additional plausible explanation seems to be that the presence of small amounts of moisture causes the cellulose constituents to swell, thereby closing the microscopic channels which, when open, permit gaseous decomposition products to make their way to the surface more quickly and there build up the gaseous concentrations to the value required for combustion.
TABLEIV. FLAMINQ CHARACTERISTICB OF WOODS MOI~TURBSP.REAT
In order to put these samples on an equal basis, they were conditioned in an oven a t 60" C. for 24 hours before test. Moisture content was determined by drying the samples for 24 hours a t 105" C. The deviations of the individual readings from the mean reading are, in this case, somewhat wider. The data, however, clearly permit a differentiation of these materials on the basis of their flaming qualities. These grow less as the specific gravity of the wood increases, and, in fact, the reduction in inflammability is found to be inversely proportional to the specific gravity and, within the limits of experimental error, to follow an equation of the form: a=-KtfC
Balsa White pine Yellow pine Maple Oak
CHEMISTRY
11.4 13.4 28 29 27
TINE, SECONDH No. 3 No. 4 No. 5 9 15.4 25 22 24
9.8
14.4 27 31 20
11.2
15.0
24
..
25
Mean 10.7
14.5 26.6 28.8 26.5
The inflammability of yellow pine does not follow this rule so closely. The discrepancy may result from the resinous constituents filling the pore space of the wood and thereby slowing up the flow of gaseous decomposition products to the surface. It is well known that the burning qualities of woods are influenced by their moisture content. The above woods were, therefore, conditioned for 4 days a t 90 per cent relative humidity and 25" C. The results are given in Table V.
No. 1
No. 2
20 4.8 11.2 15.5 11.6 17.4
24 5.8 9.4 15.6 12.4 18.6
IQNITION PERIOD, SECONDS No. 3
No. 4 22
23 4.6 9.4
9.2
10.4 13.6 12.6 13.8
14
17 14.2
No. 5
Mean=t
22 4.2
22 6.2 10.2
10.2
15.1
16.8 15.2 18
13.8
16.4
Large amounts of moisture undoubtedly absorb the heat conducted into the material and thereby extend the time required for the material to reach the temperature requisite for incandescence. The flaming and burning qualities of materials involve several questions of practical interest-namely, the effect which pretreatment of the material with foreign substances has on its flaming qualities. I n Table VI a number of tests are given on samples variously treated. TABLEVI. EFFECTOF PRETREATMENT ON IGSIT~ON OF WOODS -1QNITION
MATERIAL Masonite Whitepine Maple Maple Oak a b
No. 1 No. 2 34.1 16.4 27.1 12.8
16.0
33.0 20.2 23.2
10.8 16.1
PERIOD, SECONDSNo. 3 No. 4 No. 5 Mean 37.0 39.0 36.2 38.1 25.0 28.1 2 4 . 4 33.0 25.1 12.1 14.7
26.1
12.0
11.0
25.1
10.1 14.0
25.2 11.5 14
PRETREATNENT
Varnishedn Varnished" Varnisheda (NH4)SPOib Alz(S0d)ab
Ordinary air-drying varnish. Soaked in 5 % solution for 24 hours and then dried.
Comparison of these results with those in Table V shows that varnished materials are more flame-resistant than unvarnished materials, in spite of the inflammable character of varnish. The probable reason has already been suggested. The varnish effectively seals the channels that otherwise would lead combustible gases to the surface of the material where oxidation and combustion can take place. On the other hand, treatment of the samples with chemicals that are ordinarily supposed to retard burning had indeterminant effects. Masonite was undoubtedly rendered more flame-resistant while the woods thus treated were presumably less flame-resistant than formerly. I n the latter case we may again have a result that is dependent on the enlargement of interior channels through dissolving out cell substances with the salt solutions. LITERATURE CITED (1) N a v y Dept., Spec. No. 17-1-14, paragraph 6(P) ( l ) , Oct. 1, 1929; New York C i t y Building Code 639-27-SR. (2) Truax a n d Harrison, Proc. Am. SOC.Testing Materials, 29, 11, 973 (1929). (3) Weiss, H. F., Orig. Coin. 6th Intern. Congr. A p p l . Chem., 5 , 279 (1912); Price, R. E., PTOC. .VatZ. Fire Protection Assoc., 1915. RECEIVED August 31, 1934