UNIVERSITY OF CALIFORNIA COLLEGE OF AGRICULTURE AGRICULTURAL EXPERIMENT STATION BERKELEY, CALIFORNIA SPARK ARRESTERS FOR MOTORIZED EQUIPMENT J. P. FAIRBANK AND ROY BAINER BULLETIN 577 JULY, 1934 UNIVERSITY OF CALIFORNIA BERKELEY, CALIFORNIA Digitized by the Internet Archive in 2012 with funding from University of California, Davis Libraries http://www.archive.org/details/sparkarrestersfo577fair SPARK ARRESTERS FOR MOTORIZED EQUIPMENT 1 J. P. FAIRBANKS AND ROY BAINER3 INTRODUCTION The inflammable nature of vegetative ground cover during the dry summer season, throughout the major area of California, contributes to the fire hazards in fields and forests. In the rural districts, many fires have been attributed to ignition by the exhaust systems of motorized equipment. Fires started by internal-combustion engines may result from the emission of hot carbon particles in the exhaust stream; from contact between dry vegetation and hot exhaust pipes; or, under certain conditions, from actual contact between flames from the exhaust system and dry vegetation. Carbon residue collects within the cylinder heads, in the mufflers, and on the piston heads of practically all internal-combustion engines. The amount and nature of this carbon depends somewhat upon the condition of the engine and upon the kind of fuel and lubricating oil used. Under certain conditions, pieces of carbon residue break loose and are emitted with the exhaust gases. At the time of combustion, the gases within the engine cylinder range in temperature from 3000° to 4000° Fahrenheit; and they are exhausted at temperatures of 1200° to 1600° F. Since the temperature within the exhaust system is above the kindling point of the carbon, the carbon particles may well have temperatures approximating or even above exhaust temperatures, because they may be burning as they leave the engine. That they are hot enough to be incandescent is obvious when one watches a tractor, with no protective device on its exhaust system, working after dark. Sometimes a shower of sparks is thrown out in the exhaust stream when there is a sudden change in load or speed. Incandescent carbon particles may even be seen coming from the muffler tail-pipe of an automobile traveling the high- way at night, and they often glow for several seconds after passing from the engine. i Received for publication May 4, 1934. 2 Specialist in Agricultural Extension. 3 Assistant Professor of Agricultural Engineering and Assistant Agricultural Engineer in the Experiment Station. [3] 4 University of California — Experiment Station The law, as passed by the 1931 California Legislature, regarding pro- tective devices on internal-combustion engines, used under hazardous conditions, designates the following as a misdemeanor : Operating or causing to be operated any gas tractor, oil-burning engine, gas- propelled harvesting machine or autotruck in harvesting or moving grain or hay, or moving said tractor, engine, machine or autotruck in or near any grain or grass lands, unless he shall maintain attached to the exhaust on said gas tractor, oil-burning engine or gas-propelled harvesting machine an effective spark-arresting and burning carbon-arresting device. 4 One question was raised immediately : What constitutes an effective spark-arresting device? The Divisions of Forestry and Agricultural Engineering, of the University of California, had made a few prelimi- nary tests of spark arresters for tractors and harvesters in 1918. 5 On account of the development in recent years of new forms of arresters and of new types of tractors, the data on hand were not sufficient to give complete information on types of spark arresters fully complying with the law. In May, 1931, accordingly, the Stop Forest Fires Committee requested the Division of Agricultural Engineering to analyze the prob- lem further. In addition, the Equipment Committee of the Rural Fire Institute of California, which had already selected this phase of fire prevention as an important problem, desired that additional study be made. There was furthermore very little information available regarding the composition, size, and characteristics of the carbon particles emitted with the exhaust gases, or regarding the conditions under which fires might be started by carbon particles from internal-combustion engines or by contact with hot surfaces, such as exhaust pipes. The first step, accordingly, was to analyze some of this carbon material and to start fires in dry vegetation under different weather conditions, using carbon particles of various sizes and a, section of exhaust pipe, both heated through the range of critical temperatures. The first part of this bulletin deals with this aspect of the problem, and the results of experiments with spark arresters are given in a later section. CARBON COLLECTED FROM MOTORIZED EQUIPMENT The amount of carbon ejected from an engine for a given length of time was not definitely determined. On one 60-hp. tractor of the U. S. Forest Service, however, 100 grams of carbon were caught during 100 hours of operation, or 1 gram per hour. The tractor had run a total of 3,900 hours. 4 Calif ornia Penal Code, No. 9. Sec. 384. Stats. 1931. Chap. 311, p. 749. 6 Metcalf, Woodbridge. Fire protection for grain fields. California Agr. Exp. Sta. Bui. 295: 351-368. 1918. (Out of print.) Bul. 577] Spark Arresters for Motorized Equipment 5 On a 40-hp. tractor of the U. S. Forest Service, 14.45 grams were caught in 144 hours of operation, or 0.1 gram per hour. This tractor had been run only 580 hours. These figures merely bear out the common observa- tion that a worn engine produces more carbon than one in first-class condition. Size of Carbon Particles. — Carbon was trapped by spark arresters on eight tractors, five used in general farm work, and three on road con- struction. The three latter were U. S. Forest Service tractors to which TABLE 1 Screen Analysis of the Carbon Collected in Spark Arresters on Eight Tractors Tyler screen Size of opening, inches Per cent retained 0.185 0.093 046 0232 1.9 7 5 18.7 28-mesh 27.5 Pan 44 4 had been attached large spark traps of the inertia type with the final openings covered by 16-mesh-to-the-inch wire cloth. The carbon collected from each tractor was screened through a set of standard Tyler square-mesh screens (table 1). As the data (table 1) show, over half the carbon ejected by the tractors was coarser than the 28-mesh screen. In most samples the carbon parti- cles were hard and granular, but in one case they were soft and flaky. Fifty-three such flakes were not screened because of their large sizes, ranging from % to %-inch maximum diameter. One large flake, not included among these, measured % 6 by 1% 6 inches. For the three tractors on which the time of operation was known, carbon was caught at rates ranging from 0.1 to 1.0 gram per hour. The rate of carbon accumulation should be considered in the design and use of spark arresters. A former tractor-service man reported a case of fires being started from one of his firm's tractors equipped with an arrester. Upon investigation, the chamber proved to be filled with car- bon, which could be removed only by taking the arrester apart. - Kindling Temperature of Carbon. — A one-gram sample of each size of carbon caught from one tractor was analyzed to determine the quan- tity of ash and the kindling temperature (table 2). The ash was not analyzed qualitatively except to determine that it 6 University of California — Experiment Station was mainly a form of iron, presumably worn from cylinders and rings. 6 Since the kindling temperatures ranged from 887° to 1022° F, 7 it is evident that carbon can be ignited in the exhaust system of an internal- combustion engine. Weight of Carbon Particles. — In weight, carbon particles from differ- ent samples varied widely. For example, 50 pieces from one sample pass- ing through a 0.093-inch square mesh and caught on a 0.092-inch round- hole screen weighed 0.40 gram, or 0.008 gram each; while 100 pieces of another sample similarly screened weighed 0.11 gram, or 0.0011 gram each, a ratio of more than 7 to 1. TABLE 2 Kindling Temperature and Quantity of Ash tor One-Gram Samples of Each Size of Carbon Tyler screen Ash in grams Kindling tem- perature of carbon retained, degrees Fahr. 0.810 0.813 0.799 0.761 605 1022 1022 28-mesh 1004 Pan 887 SELECTION OF CARBON FOR FIELD TRIALS The carbon selected for the field tests must correspond in general type to that ejected by engines, must not break into smaller particles when handled, and must not coke in the furnace, later to be discharged as a wad rather than as separate particles. Furthermore, it must be available in quantity. Trials were made with the following forms of carbon : (1) carbon electrodes from dry cell batteries, (2) coal coke, (3) oil coke, (4) carbon scraped from pistons and cylinder heads of internal-combustion engines, and (5) cinders from coal-burning locomotives. Experiments comparing these carbons under controlled laboratory conditions are given in a later section. No one of the carbons used represented all types ejected from internal- combustion engines. The battery carbon, being very hard and heavy, did not burn readily nor coke. It was considered a useful indicator of the ease with which different forms of vegetation are ignited under various weather conditions; and it represented an inert or nonflaming type of material. 6 The analysis of carbon samples was made by H. W. Allinger, of the Division of Chemistry, University of California. 7 All temperatures in this bulletin are in degrees Fahrenheit. Bul. 577] Spark Arresters for Motorized Equipment 7 No field tests were made with coal coke. One form of oil coke was un- suitable because it coked readily. "Cokettes," an oil-coke product used in orchard heating, were extensively used. Carbon scraped from engines was probably the most representative of material ejected by exhausts. It had a tendency to burn after leaving the furnace and to break down into smaller particles. In some cases there was some coking effect in the furnace. -Comparative sizes of the five groups of carbon particles. Left to right, Nos. 1 to 5. Scale, in inches. Cinders from coal-burning locomotives, collected on car roofs in Mon- tana by members of the U. S. Forest Service, were used in a few field tests; but only small sizes were available. Being fairly inflammable, the coal cinders were not representative of the carbon ejected from internal- combustion engines. TABLE 3 Screens Used in Selecting Carbon Samples Carbon size No. Passing through Tyler screen Size of opening, inches Retained on Size of opening, inches 1 2 2-mesh 4-mesh 8-mesh 14-mesh 28-mesh 371 0.185 0.093 0.046 0.023 4-mesh hardware cloth 221 0.170 3 0.092 1 4 Tyler 28-mesh 023 : ] 5 Pan 000 Carbon particles caught in spark arresters on tractors were used in a few tests, but the procurable supply was limited. Five classifications of carbon sizes (fig. 1) were used. The basis for segregation was a set of Tyler square-mesh screens in which the diameter of each succeeding opening was one-half that of the one preceding. The range of sizes within each group was kept to a minimum by inserting other screens with openings slightly smaller than those of the corre- sponding square-mesh screen (table 3). All screening was done by means of a Ro-Tap machine. University of California — Experiment Station EXPERIMENTAL PROCEDURE AND TEST AREAS Equipment for Spark Tests. — Equipment was devised for heating car- bon to known temperatures and quickly ejecting it into vegetation, to determine the sizes and temperatures necessary to start fires. A 2.5-kw. high-temperature, combustion tube furnace (fig. 2), rated at working temperatures up to 2500° F, was used for heating the carbon particles. Current was supplied by an engine-generator set mounted on a truck, and the temperature was controlled by voltage regulation. Fig. 2. — Portable electric tube furnace for heating carbon particles. J. Heating element; B, monel tube; C, thermocouple; D, carbon sample; E, title barrel; F, wads from cartridges; G, piping of the ejection system. Within the tube the temperature was measured by a "platinum — platinum 10% — rhodium" thermocouple in a Pyrstan case, inserted in the combustion tube. This thermocouple was calibrated in millivolts, and the cold-junction was packed in a vacuum bottle containing ice. The sensitivity of the instrument was indicated by the fact that after the charge of carbon particles was placed in the furnace the millivolt meter needle would drop for a few seconds and then gradually return to the initial temperature. Wind velocity was measured either with a standard Weather Bureau three-cup Tycos anemometer, mounted on a base so that the centers of the cups were 18 inches aboveground, or by a Biram-type 4-inch ane- mometer suspended 24 to 36 inches aboveground by a string from an arm on a staff and held into the wind by a light metal vane. Bul. 577] Spark Arresters for Motorized Equipment !) Air temperatures were read either from chemical thermometers sus- pended in the shade within 24 inches of the ground, or from the dry bulb of the psychrometer. In measuring soil temperatures, a soil thermometer, standard grade, with pointed metal end, was inserted at such a depth that the bulb was at the surface of the soil. Relative humidity was determined by sling or "egg-beater" types of psychrometers. Fig. 3. — Dry oat grass (Davis). Methods of Using the Equipment in the Field. — The truck carrying the 5-kw. engine-generator set was placed within 200 feet of the test plots, and the current was carried through 10-gauge weatherproof stranded conductors. The furnace was placed on the ground in the plot, and the vegetation adjacent to it and to the instrument table, was cleared away. Fire guards around the plots, together with shovels, wet sacks, and knapsack pumps, were used to control the fires. For each trial, two grams of carbon were used. The carbon particles burned but slightly when first put into the combustion tube, the flames ceasing within a few seconds, probably through deficiency of oxygen. Upon injection of the carbon, the millivolt meter needle dropped a few points. When it returned to the reading for the desired temperature, the carbon particles were ejected, either by shooting 0.22 or 0.32-caliber blank cartridges from guns, or by blowing into the pipe by mouth. The 10 University of California — Experiment Station carbon particles scattered widely, falling often within a few feet of the furnace but occasionally 25 to 30 feet away. In general, the shots were made with the wind. The results were considered negative unless the vegetation flamed. Smoke or a glow alone was not reported as a "fire." The number of spot fires resulting from a shot was not reported, although from one to eight spots may have ignited. Field trials were conducted during the summers of 1931 and 1932 in short, dry, immature oat and barley stubble at Davis, in grasslands in Fig. 4. — Barley stubble (Davis). Tehama County, in brush fields near Mt. Shasta, and in pine needles near McCloud. Dry Oats. — This field had been sown with oats that did not mature, because of a deficiency of moisture. The vegetative cover was mainly a mat of grasslike nature with small blades and stems up to 10 inches in length. There was a small percentage of barley, wild oats, and foxtail; but all were small plants (fig. 3) . Barley Stubble. — The barley field had been harvested with a combine, leaving the stubble 8 to 12 inches high. Blades, chaff, and some heads lay on the ground, clustered around the base of the stems (fig. 4). Moisture samples of 200 to 300 grams each were taken. The average moisture content of twelve samples was 7.1 per cent, the range being from 2.2 to 14.5 per cent. Range Grasses. — Field tests were made on two different plots of range grasses in Tehama County : one on the Owens ranch about 12 miles west Eul. 577] Spark Arresters for Motorized Equipment 11 of Red Bluff, and the other on the Mohr ranch about 8 miles north of the same city. Range land on the Owens ranch consisted of rolling hills covered with a heavy stand of fine, dry grass somewhat matted and trampled (fig. 5) . The larger plants still standing were from 10 to 18 inches high. The species of grass identified were as follows : soft cheat (Bromus hor- deaceus), smooth flowered soft cheat (Bromus racemosus) , Mediter- Fig. 5. — Dry grass on range land (Owens ranch, Tehama County). ranean barley (Hor oleum gussoneanum), and wild oat (Avena fatua). Other plants than grasses identified were : lupine (Lupinus), filaree (Erodium), fiddleneck (Amsinckia), and blow wives ( Achyrachaena mollis). The average moisture content for ten samples was 7.18 per cent, the range being from 3.1 to 20.2 per cent. The large variation in the moisture content was undoubtedly due to the relative proportion of the material in the sample from plants which were still in contact with their root system and those which had been broken from their root system and were lying on the surface of the ground. Although the two types of vegetation were apparently of the same degree of dryness, it became apparent in drying the samples that those which were still in their natural position with the roots and stems intact contained a much greater moisture content than those which had been broken off and were lying loosely on the ground. It is probable that ignition took place in most cases in the latter. This material in general, did not run higher than 3 per cent moisture. On the Mohr ranch the range land consisted of a small flat along a 12 University of California — Experiment Station dry creek where the vegetation was a sparse stand of grass 4 to 5 inches high (fig. 6). The species of grass were identified as follows: soft cheat (Bromus hordeaceus), few-flowered fescue (Festuca reflexa), hairy- leaved fescue (Festuca confusa), and red brome (Bromus rubens). Other plants than grasses identified were as follows : Turkey mullein (Eremocarpus setigerus), filaree (Erodium), knotweed (Polygonum calif ornicum), and clover (Trifolium microcephalum). Fig. 6. — Dry vegetation (Mohr ranch, Tehama County). The average moisture content of nine samples was 9.4 per cent, the range being from 2.9 to 20.7 per cent. Pine Needles. — The test plot was located in a stand of second-growtli ponderosa pine about fifty years old, near McCloud. The crown density, as expressed by foresters, was about 0.5, and the depth of needles on the ground averaged 2% inches. There was no undergrowth or other ground cover (fig. 7). The average moisture content of five samples was 7.1 per cent, the range being from 5.1 to 9.3 per cent. Brush Field Litter. — This test plot was in a brush field near the town of Mt. Shasta. The vegetation consisted of manzanita, whitethorn ceano- thus, and bracken (fig. 8) . The litter on the ground, about an inch deep, was composed of twigs, branches, and leaves, typical of brush fields. The average moisture content of three samples of the ground cover was 14.1 per cent, the range being from 11.5 to 17.7 per cent. Weather Data, — The field tests were all made in August, when many range and forest fires are expected, but few grain fires, because the grain has been harvested. Bul. 577] Spark Arresters for Motorized Equipment 13 During the 1931 field tests, the temperature readings ranged from 83° to 108° ; the relative humidity, from 13 to 40 per cent; and the wind velocity, from to 6 miles per hour (table 4). ■ Fig. 7. — Pine needles (Shasta County). Table 4 shows that, with the exception of one day, the 1932 tests were not made during a period of high fire hazard. The average temperatures were lower and the relative humidities higher than for the 1931 tests. Fig. 8. — Brush field (Shasta County). In general, at an air temperature of 80° the relative humidity was 30 per cent or less; at 100°, 20 per cent or less. At 110°, with a relative humidity of 10-12 per cent, the fire hazard was extremely high. 14 University of California — Experiment Station OS CM CO W N "O ^H CO CM CO * N M rt CM o o o ©i o co SO iO a ■ CO CO N N O CM rt i-H i-l CM o o o o IN O) CO ■* O -f 1 .-H i-i r-< ^ CO CM r- co O cr. o o o o K) O * to CO —1 03 © CO i-l o o 00 OS 00 CO o CO 00 OS oo it o o o o o c o O O o O CM CO Oi oo CO uo CO t» Q © 00 I- 00 00 a a a a a 2 a a a a s a ft ft o. d, ft & o. a o. ft <* CO o SO «5 o c o o o 04 o CO U0 8 >o co •cr 00 «0 -* CM tn -r >o 'SX C z o o c O o ~ a s a a a a a £ a a a a ft o. C3 ca ft 02 03 p. 99 o o CO KS © o q o q o CO O q o CO o 5 «3 O CR i-i ^h co co co OS OS OS CO 35 CM CO S9 ~". Ol CO oa CM CM CM CM CO CO CO CO O O) o> o -h" cm" iO CM CM CM co" CM o" 3 CI co" t-~" co' OS t* M M M 3 3 3 3 9 o o s (A Id -1 o 111 z u z Q. UI u. o z § -> U U. I- • ft Pu s o-u >-ui "< 0< "£ sS 5> =>> <* I< o ® d < i 01 1- 2 o _l tl I a. m • 3 " ( J 08 '.'"■'? D B { . 1 u 9 • : o H-^l i coo • 8 3 "SB. 8 < < z ' §§§§ i z i 3 >; u. J ^ z "> Ol z Z ui O Ml <" O - 5? Z o CO < o _l i UJ K U. O z u>< u u- *- u o "• •Si? ^UJ 5-LJ IIP s in" OJ o' o < < Q _i CD m D ■ u _l cc < m CC o z g ^« -A a. Jj >- " ?■ ° ^ >J '"I 5 « < $ I< kl tr •-a LJ * < z D 2 : - 2 i t - £ i ■ ♦ 2 < >r» * i 1 2 ! ', s n oj oj Z u o r^ u. CD *°3 - . m * z o S ooc - M "> "T J\ <») <3> >n rj ru CM O < < o < CC o QC o I If) AIR TEMP= 90°FTOM0*F LEGEND AVERAGE • 98°F # m , F!RE HUMIDITY = I3%T026* o . I TRIAL NO FIRE AVERAGE- 18% (M n 0) l-- v d i" o z < CC OC i o < a. o a o i UI or o D z "^ _l •" •^\t So! 00 oo S O 1- O < < i < a. => IX a S Q u "J i- g u z O Z> < I z — 1 DC 3 s 3 U) 11 o Fig. 14. — Cletrac cyclone spark arrester. The efficiencies for this arrester, in its original form, were 98.0, 96.8, and 69.3 per cent, respectively, for carbon particle sizes 2, 3, and 4. Tn the modified design these efficiencies were raised to 99.6, 99.3, and 97.6 per cent for the three respective sizes of carbon particles (table 11). The efficiency of the rebuilt arrester was from 1.3 per cent to 3.2 per cent lower when tested in a horizontal position than when tested in a vertical position (table 11). The factory now builds this improved model. The Cletrac Cyclone Arrester. — The spark arrester known as the cyclone, goose-neck, or question mark, consists of an outer shell, stamped in two parts from 18-gauge steel, then clinched together and held by a seam completely encircling it (fig. 14). Three-inch openings through the center of each side permit the exhaust gas to pass outside. A spark trap located in the path of the gas collects and retains the carbon particles. Upon entering the arrester, the gas and carbon particles are imme- diately directed through a circular course. The centrifugal force thus set up throws the particles to the outer edge of the exhaust stream, where Bul. 577] Spark Arresters for Motorized Equipment 29 they are separated from the gas by the baffle and fall into the trap. The trap may be removed for cleaning by taking out one bolt. After a complete series of tests with the arrester in its original form, the outlets were covered with iron-wire screen of 16 meshes per inch, and another complete series was run. The efficiencies for the Cletrac Cyclone were 98.7, 98.2, and 79.0 per cent, respectively, for carbon sizes 2, 3, and 4 without screens; 100, 99.9, and 80.8 per cent, respectively, when the outlets were screened (table ii). Fig 15. — Caterpillar spark arrester. The back pressure was 4.75 inches of water for the arrester without screens and 5.75 inches when the screens were applied (table 10) . The Caterpillar Spark Arrester. — This arrester consists of a one-piece iron casting (fig. 15). It works on the same principle as the cyclone, except that the exhaust gas is carried through an additional 270° and discharged in line with the entrance after the carbon particles have been separated from the gas stream. The arrester, therefore, can be located close to the manifold; and a tail pipe can carry the gas away. No bene- ficial results, as far as separating the carbon is concerned, are gained in turning the exhaust gas through the last 270°. The efficiencies were 94.0, 93.3, and 77.0 per cent, respectively, for carbon sizes 2, 3, and 4 (table 11) . The back pressure amounted to 12.15 inches of water (table 10). 30 University of California — Experiment Station A field study of this arrester showed that very little carbon was ever retained in the trap. The reason might possibly have been the nearness, in some cases, of the arrester to the exhaust manifold, resulting in tem- peratures high enough to ignite the carbon. The most likely explanation, however, is that the space provided for collecting the carbon is too small; after a small amount of carbon accumulates, the rest is gradually carried out in the exhaust stream. 4Ho£ea flinch Cff Q55 SE CTIO N MM Fig. 16. — Funke experimental spark arrester. (Cross section drawn on one-half larger scale.) The Funke Spark Arrester. — An experimental arrester (fig. 16) de- signed and built by F. W. Funke, Senior Forest Ranger, U. S. Forest Service, showed considerable promise. An outer cylinder is enclosed at the lower end and tapered to a 5-inch outlet at the top; and an inner cylinder, concentric with it, is enclosed at the top and opens at the bot- tom directly into the exhaust pipe. A series of % 6 -inch holes, four per inch horizontally and five vertically, near the base of the inner cylinder, permit the gas to flow into the outer cylinder and thence to the outside. A cone centered and attached, inverted, to the top of the inner cylinder splits the stream of carbon particles, directing them through an opening made between it and the walls of the frustum of another cone. The latter frustum is open at both ends, with the larger end attached to the inside wall of the inner cylinder 2 inches from its base. The two cones Bul. 577] Spark Arresters for Motorized Equipment 31 taper toward each other, providing at their closest point an opening of y 2 inch (fig. 16). Upon entering the arrester, the carbon particles, because of their inertia, continue in a straight line until deflected by the inverted cone, while the gas turns more or less abruptly, flowing through the holes in the side of the inner cylinder. Having passed the narrow opening be- tween the two cones, the carbon particles fall into a space formed by the 5//X W£W SECTiQft fl-fl Fig. 17. — John Deere spark arrester. outside wall of the frustum of the cone and the inside wall of the inner cylinder, in the top of which an opening with a removable covering per- mits one to clean the entrapped carbon from the arrester. The efficiencies were 99.9, 96.0, and 94.3 per cent, respectively, for carbon sizes 2, 3, and 4 (table 11). The back pressure amounted to 2.75 inches of water (table 10). The John Deere Spark Arrester. — Though not in the inertia group, the John Deere arrester does resemble inertia types, in that the gas is set into a swirling motion by the use of vane-type baffles between the different sections. This arrester (fig. 17) is made up of similar sections superimposed, each consisting of a central spacer with six holes equally spaced around the periphery to permit passage of the gas between the outer and inner chambers, and with an outer shell. A baffle, with twenty-five vanes which impart a swirling motion to the gas, forms the top of each section, be- 32 University of California — Experiment Station tween the central spacer and the outer shell. The arrester consists of five such sections, each slightly telescoped together, held between two cast- ings. The bottom casting is so shaped that it can be connected to the exhaust manifold, while the top one provides a connection for a tail pipe to carry the gas away from the engine. The lower end of the inner cylinder, formed by the five spacers, is tight, so that the gas passes through the holes in the spacer, between the outer and inner chambers. After reaching the first section, the gas has two possible paths: it may pass immediately through the holes in the first spacer into the inner chamber, whence it is free to go into the open; or it may travel vertically from one section to the next, passing through the baffles between each section until the last is reached, and then leaving the arrester through the inner chamber. The gas is set into a swirling motion by the baffles as it goes from one section to the next. Some carbon remained in the lower end of the inner cylinder at the end of a test, but much passed out of the arrester with the exhaust gas, as the tests indi- cated. Obviously, in the design of this arrester, no place is provided for the retention of carbon material. The efficiencies were 17.6, 17.9, and 11.8 per cent for carbon sizes 2, 3, and 4, respectively (table 11). The back pressure amounted to 4.25 inches of water (table 10) . TESTS OF SCREEN-TYPE ARRESTERS The McCormick-Deering Spark Arrester. — This device (fig. 18) is so constructed that the exhaust gases must pass through two thicknesses of 10-mesh, No. 20 iron-wire screen, twice before they are discharged. Any carbon material smaller in diameter than the size of screen opening passes through the arrester with the gas. The coarser carbon material remains in the exhaust line, pounding about till its size lias been reduced sufficiently to pass the screen. The openings in the two layers of screen do not always coincide; and, since they are rather loosely held together, carbon particles are some- times caught and held between the screen walls, especially near the point. Any large amount retained by the screen increased the restrictive effect of the arrester. For example, when a 100-gram sample of size 2 carbon was gradually introduced into the line, the back pressure in- creased from 10.5 inches of water at the beginning of the test to 76.07 at the end, while the pulley speed dropped 20 revolutions per minute. On the other hand, when a test was run with No. 4 carbon, the back pressure increased only 2.1 inches of water (table 10). In the latter case, most of the carbon was small enough to pass through the screen and the speed did not fluctuate to any great extent. Bul. 577] Spark Arresters for Motorized Equipment 33 Sometimes, under operating conditions, the carbon caught and held between the screen layers and inside the cone tip may catch fire because of the high temperatures of the exhaust gas. Thus the lower tip may burn off, decreasing the efficiency of the arrester. This tendency was reported in one case. The efficiency was 100, 99.9, and 18.5 per cent, respectively, for car- bon sizes 2, 3, and 4 (table 11). s S ZCT/ON 3-3 Fig. 18. — McCormick-Deering spark arrester. Experimental Spark Arrester. — An experimental device of the screen type was built up primarily to test screens of different composition for resistance to heat and vibration. It consisted of a pipe, 4 inches in diam- eter and 30 inches long, made of 18-gauge black steel (fig. 19). The upper end was closed off; and three rows, each including nine 1-inch holes, were drilled around its circumference. The first row was 1 inch below the top of the pipe; and the next two rows were 3 inches and 5 inches, respectively, below the top. A shell of the same weight steel, 8 inches in diameter and 9 inches long, was placed concentric with the upper end of the pipe. This shell was closed around the pipe 2 inches below the bottom row of holes. Its top extended 2 inches above the end of the center pipe and had an opening 5 inches in diameter, covered with 16-mesh, No. 1 Nichrome wire screen. The size of the wire in the screen was 0.019 inch, which gave a net opening of 48 per cent of the total area. The screen had a total area of 19.6 square inches, with a net 34 University of California — Experiment Station opening of 9.4 square inches. It was held in place by an annular plate bolted to the top of the shell. The efficiency of this arrester was 100, 100, and 71 per cent for carbon Fig. 19.- -Experimental spark arrester. sizes 2, 3, and 4, respectively (table 11). The back pressure created amounted to 2 inches of water (table 10). The Case Harvester Arrester. — The Case arrester (fig. 20) consists of a pipe closed off at the upper end and having six vertical rows of %-inch holes punched near the upper end. The outer shell is 5*4 inches in diam- Bui,. 577] Spark Arresters for Motorized Equipment 35 eter and 11 inches long, with a closed top and with a 2-inch opening in the bottom for slipping over the end of the inner pipe. Lugs on the pipe 11 inches from the top, and a bolt through the top of the outer shell and Cftg. OOCO o o o O () o oo io o © o 6 l/erf/ea/ /Ebtv* 7 tfo/es fact, Of z \Q6*?e> GtefcS&s/ SecrtQN fi-rt Fig. 20. — Case spark arrester for combines. the end of the pipe, keep the outer shell concentric to the pipe. A series of eighteen vertical rows of five holes each, % 6 inch in diameter, extend through the outer shell. The top holes are 1 inch from the top of the shell and extend down 4 inches. 36 University of California — Experiment Station As some of the holes through the pipe are in line with the holes in the outer shell, obviously carbon can pass directly through the arrester. The efficiency of the Case arrester was 82.6, 64.6, and 41.4 per cent ■3i -- CZuTAVVdtY VtZStf Fig. 21. — Vacuum muffler and spark arrester. — <3/ner 1/s*\a/ Fig. 22. — Hercules muffler and spark arrester. for carbon sizes 2, 3, and 4, respectively (table 11). The back pressure amounted to 4.75 inches of water (table 10). The Vacuum Muffler and Spark Arrester. — The Vacuum device (fig. 21) consists of a somewhat intricate casting. The outer shell, cast in two Bul. 577] Spark Arresters for Motorized Equipment 37 parts, is held together by three bolts. The shell is 4 inches wide and 8 inches in diameter, having at the center an inlet 3% inches in diameter. The outlet is near the outer edge on the opposite side from the inlet. TABLE 10 Back Pressure Caused by Various Arresters • Name of arrester Back pressure at beginning of test Back pressure at end of test Yuba (original) Yuba (modified) inches water 4*25 1 75 4 75 5.75 2 10.5 5.35 2 75 4 75 4 25 5 35 12 15 inche 4 1 4 5 2 76 5 2 4 4 15 12 ? water 25 75 Cletrac cyclone 75 75 Experimental 07* 35 Funke 75 75 John Deere Hercules 25 55 1 15 * Test in which size 2 carbon was used. Back pressure at end of test for size 3 was 48.9 inches water, at end of test for size 4 carbon, 12.6 inches water. t For size 2 carbon only. Back pressure for sizes 3 and 4 carbons at end of test, same as for beginning. TABLE 11 Efficiency of Spark Arresters, in" Per Cent Arrester Carbon size No. 2 Carbon size No. 3 Carbon size No. 4 Yuba 98.0 99.6 98.3 99.9 98.7 100 94.0 99.9 99 9 17 6 100 100 82.6 81.7 83 96.8 99.3 97.7 99.6 98 2 99.9 93 3 96 1 99 6 17 9 99.9 100 64.6 28.2 18 5 69.3 97.6 94.4 98.7 79 • 80 2 Caterpillar Funke 77 1 94 3 98.2 118 18.5 58.7 41 4 15 3 Hercules „ 9 5 Twenty finger-like projections iy 2 inches long, cast integral with the side walls, extend toward the middle of the arrester; and two other flat castings with similar fingers are located in the middle. The fingers on the inner walls and on the flat casting dovetail together, leaving a small clearance for the passage of gas and, at the same time, preventing car- bon particles from passing until their size is approximately No. 3. 38 University op California — Experiment Station The path of the gas through this arrester is into the center of one side, then through the space between the fingers, and thence to the outlet. There is no place for carbon particles to be retained. If too large to pass the small clearance between the fingers, they remain in the exhaust line until reduced sufficiently to pass through between the fingers. The efficiency was 81.7, 28.2, and 15.3 per cent for carbon sizes 2, 3, and 4, respectively (table 11). The back pressure amounted to .5.35 inches of water (table 10) . The Hercules Muffler. — This muffler (fig. 22), used occasionally as a spark arrester, consists of two stampings from 16-gauge steel riveted together to form a 4 X 10 inch dome-shaped chamber. A small space between the two halves of the arrester provides passageway for the gas and the smaller carbon particles (including size 3). The efficiency of this arrester was 83.1, 18.5, and 9.5 per cent for sizes 2, 3, and 4, respectively (table 11). The back pressure amounted to 5.35 inches of water under normal operating conditions, but increased to 15.55 after 100 grams of size 2 carbon particles had been introduced into the exhaust line. On the coarser carbon material the arrester tended to load up, causing greater restriction (table 10). FIELD TESTS OF ALLOY SCREENS Common 16-mesh window screen is often placed over the end of exhaust pipes of tractors and trucks to stop the discharge of carbon material. Since iron screens rust and burn out when subjected to the heat from exhaust gases, frequent replacements are necessary. Observations made on one harvester equipped with a screen type of arrester showed three replacements of screen within a week. This screen is liable to burn out without the operator's knowledge, presenting, until replaced, the fire hazard of an unprotected exhaust system. In its use, little attention has been given to proper openings (screen area) to insure that the velocity of exhaust gases, through the screen, will be sufficient to keep the screen relatively free from a carbon coating. For example, the screen type of arrester that was standard equipment on Harris harvesters had a screen area of approximately 113 square inches. As the gas velocity was too low to keep the surface clean, carbon collected on the screen; and later, when the engine was working under a heavy load, the carbon caught fire and was carried away still burning. An experimental arrester (fig. 19), on which the screen area over the final opening was approximately one-sixth that of the regular Harris arrester, was used to replace the latter. The back pressure on the exhaust system increased only 0.2 of an inch of water in consequence of the Bul. 577] Spark Arresters for Motorized Equipment 39 change. At the same time the screen was found to remain free of carbon coating because of the higher velocity of the gases through it. An attempt was made to find an alloy screen that would withstand the heat and vibration of the exhaust gases. Two such screens, 16 meshes per inch, were selected — namely, "Chromel" and "Nichrome" (trade names for nickel and chromium alloys) . Chromel A and Nichrome 4 are identi- cal, containing 80 per cent nickel and 20 per cent chromium. The cheaper grades contain small amounts of iron, which lowers their resistance to heat and vibration, disqualifying them for this type of work. Chromel A or Nichrome 4 withstands these conditions better than other combina- tions in these series. An experimental arrester employing Chromel A, 16-mesh screen, hav- ing a 46 per cent net opening, showed no signs of failure after 560 hours of use. It was attached to the exhaust pipe of a 30-hp. tractor, only 10 inches from the exhaust manifold, so that the conditions were rather severe. Failure occurred in 130 to 180 hours of operation when the lower grades of alloy screens, containing iron, were used. TEMPERATURE EFFECTS FROM COVERING EXHAUST PIPES A piece of seamless steel tubing, 1%-inch outside diameter, 8 feet long, was attached to the exhaust manifold of a 22-hp. 4-cylinder engine, 4-inch bore, 4^2-inch stroke, operating at 1,170 r.p.m., pulling its rated load. The temperature of the gases in the manifold was 1000° ; at the end of the line, 598°. The temperature on the surface of the pipe was 590° at points 2 and 5 feet from the manifold. The tubing was then wrapped with engineer's tape (asbestos, 7 / 1Q inch thick, 2y 2 inches wide), and the temperatures were noted at the same points as above. The temperature in the manifold was 985° ; at the end of the line, 728°. On the surface it was 320° at 2 feet and 270° at 5 feet from the manifold. Though the coverings materially reduced the surface temperatures, the temperatures of the gases at the end of the line were considerably higher. The tape was then replaced by a 3-inch galvanized iron pipe, held equidistant from the tubing, giving a clearance of approximately % inch between the tubing and the pipe. The temperatures of the gases in the manifold and at the end of the line were 975° and 760°, respectively, while the surface temperatures were 200° at 2 feet and 185° at 5 feet from the manifold. A third pipe 4 inches in diameter was placed around the 3-inch pipe and held equidistant from it. This pipe remained so cool during the test that it did not burn bare hands. 40 University of California — Experiment Station When the 3-inch galvanized pipe was used over the 8-foot length of tubing, in a vertical position, the temperatures along the surface 2 and 5 feet from the manifold were 160° and 140°, respectively, because of natural convection through the space between the two pipes. Since both ends of the 3-inch pipe were open, air circulated freely through it. EFFECTIVENESS OF MUFFLERS AS SPARK ARRESTERS A few preliminary tests were run on two mufflers to determine whether or not carbon particles were separated from the exhaust gases. In some cases the carbon was retarded by the muffler; but very little was held permanently. In other words, when carbon was introduced into the line ahead of the muffler, some passed through immediately, the rest at inter- vals ranging from a few seconds to several minutes. A test, previously reported, in which a car was treated for carbon removal, revealed that the carbon particles passed through the muffler and tail pipe at temperatures sufficiently high to set cotton batting on fire. The batting was held 18 inches back of the tail pipe. To produce the same results, engine scrapings size 4 had to be heated to a temperature of 1400°. In general, the average muffler cannot be considered a satisfactory spark arrester. No doubt it assists somewhat, especially in cooling the particles. SUMMARY The amount of carbon ejected with the exhaust gas, for the average tractor engine, varies from less than 0.1 gram to 1.0 gram per hour, according to the condition of the engine. On the average, 50 per cent of this carbon is of a size that will not pass a 28-mesh screen. Since the kindling temperature for this carbon varied from 887° for size 5 to 1022° for size 2, evidently carbon can be ignited by the temperatures existing in the exhaust system of an internal-combustion engine. Fires can be consistently started in dry vegetation, by carbon par- ticles, sizes 1, 2, and 3 (see table 2 for the sizes of carbon samples) wild initial temperatures of 1500° to 1600°, during the normal summer con- ditions in California. In extremely warm dry weather, fires may be started with these sizes at temperatures as low as 1300°. Fires can be started by size 4 with an initial temperature of 1500° under conditions favorable to fire. Carbon size 5, when heated to an initial temperature of 1500°, can start a fire in dry grass on extremely hot dry days. This size of particle, however, lost its heat so rapidly that it ceased to glow within a distance of 9 feet from the furnace, when heated to an initial temperature of Buii. 577] Spark Arresters for Motorized Equipment 41 1800°. In practice, therefore, the exhaust pipe could be turned up, dis- charging into the air, thus providing the necessary distance for cooling the smaller particles before they reach the ground. The tendency for brush-field and pine-needle litter to be ignited by carbon sparks is about the same. Both catch fire less easily than dry grass. Punk, however, was found not only to be readily ignited but to cause "hang fires"; that is, the material did not blaze immediately, but continued to smoulder and finally burst into flame. In one case the time was 40 minutes. Exhaust pipes having surface temperatures of 1200° may start fires upon contact with dry grass. Surfaces with temperatures as low as 838° may ignite dry vegetation after several minutes of contact, which could occur if dry grass or straw lodged on an exhaust manifold. If the surface shows even the slightest red color, when viewed at night, it is dangerous and the part should be so placed or guarded that it will not touch vege- tation. Several commercial spark arresters now on the market were above 95 per cent in efficiency in stopping carbon sizes 2 and 3 ; but all were 80 per cent or below in efficiency for size 4. One commercial arrester was below 20' per cent in efficiency for all sizes of carbon. The efficiency of one arrester was increased from 69.3 to 97.6 per cent, for size 4 carbon, by a slight modification in design. One experimental arrester had an efficiency of 93 per cent or higher for carbon sizes 2, 3, and 4. The arresters with the highest efficiency created the least back pres- sure on the exhaust system. The average muffler cannot be considered a satisfactory spark arrester. No doubt it assists somewhat, especially in the cooling of particles. Iron screen over the exhaust pipe presents considerable danger, as holes may burn through without the operator's knowledge, thus pre- senting a fire hazard from an unprotected exhaust system until replaced. Alloy screens consisting of 80 per cent nickel and 20 per cent chromium stood up much better than iron screen. The cheaper grades of these alloy screens, however, contain varying amounts of iron and are, therefore, unsuitable for this type of apparatus. ACKNOWLEDGMENTS The authors acknowledge indebtedness to John R. Curry, Associate Silviculturist, U. S. Forest Service, for assisting in the planning and carrying out of work relative to the starting of fires under various con- ditions and for identifying the vegetation encountered; to F. W. Funke, 42 University of California — Experiment Station Senior Forest Ranger, U. S. Forest Service, for assisting in the develop- ment of experimental arresters and for collecting and supplying carbon for field trials; and to 0. C. French, Junior Agricultural Engineer, Uni- versity of California, for assisting in the work relative to burning, and with the efficiency tests of spark arresters, and for preparing all draw- ings of the spark arresters. In addition, the authors are grateful to Roy Owens, of Red Bluff, and Albert Mohr, of Cottonwood, for permission to carry on investigational work in their fields. 12m-7,'34