Safeguards for Handling, Storage and Employment of Flammable Solvents in Industry
In connection with the manufacture of munitions it became necessary for the E. I. du Pont de Nemours & Company to handle and store very large amounts of inflammable liquids, principally ether and alcohol and light oils.
For operating reasons it was necessary to interconnect series of tanks by means of filling and draw-off connections, and in some instances through vents. This introduced the question of the possibility or probability of fire or explosions in one tank spreading throughout the system because of the presence of inflammable or explosive vapors in the piping connections.
The Fire Protection Division of the Engineering Department of the E. I. du Pont de Nemours & Company, under the direction of Harold. L. Miner, Eire Protection Engineer, had this matter under Investigation.
It was, of course, of utmost importance that these essential materials be safeguarded against fire in order to protect plant operations as well as to protect property and human life.
Mr. Miner canvassed the entire country very carefully to ascertain if there was any reliable data available on this subject.
Finding that there was nothing of this sort available in this country, nor probably in the world, the question was taken up with the Underwriters’ Laboratories of Chicago, and being satisfied as to their ability to conduct such an investigation, they were authorized to proceed at the earliest possible moment and to render a full and definite report.
Bear in mind that the Underwriters’ Laboratories is maintained for service—not profit, but service must be recompensed and the whole cost of the investigation so far has been borne by E. I. du Pont de Nemours & Company of Wilmington, Del., constituting in its movement prevision in engineering and a handsome contribution to Fire Engineeing practice.
Recently the Underwriters’ Laboratories, A. H. Nuckolls, Chemical Engineer, issued a report of 64 pages with charts, illustrations and a great deal of data. “Special Investigation No. 528.”
“Report on Propogation of Flame in Pipes and Effectiveness of Arrestors Including Gasolene. Ether, Alcohol, Acetone, Amyl Acetate and Light Oil.”
which while most informing, interesting and quite luminous, is by no means complete, as witness what Mr. Nuckolls himself says in his summary:
“EXTENSION OF INVESTIGATION.”
“The results inidcate that an extension of this work, including tests in larger size and longer lengths of pipe to determine the nature of the propagation over a wider range and the effectiveness of arrestors in connecting pipe when explosion originates in a large tank would yield additional data of much importance in safeguarding industrial systems as well as of scientific interest. Further study of certain phases of this work and an investigation of propagation of flame through pipe of small dimensions and under high initial pressures are under way, the result of which it is expected will supplement certain phases of this investigation, enabling a closer correlation of experimental results.”
The value of extending this investigation into its final detail is of prime importance in the scheme of fire protection of various systems of industry and The FIRE ENGINEER urges upon the management of such plants that they collaborate earnestly and fully with the du Pont people and, all together, arrange to finance to completion an investigation whose results must prove beyond valuation to each one of them as well as to the scientific determination so clearly needed.
The FIRE ENGINEER will lend what aid it may with most earnest interest and eager good will.
If there were the space it would reproduco Mr. Nuckolls report in full, but for lack of space must be content to give the summary only and tbe sharp significance of this summary should surely incite the interest and indeed constrain the prompt ocoperation of every Industrial Unit affected. Many of them are affected now and a great many more of them, of necessity, are going to be.
The Introductory and Summary:
In the industries, tanks containing flammable solvents and light oils are commonly connected by a piping system for operating and other purposes. About a year and a half ago, E. I. du Pont de Nemours & Company brought up for discussion the possibility of fire and explosion spreading from one tank or part of a system to others, and finding that there was practically no information available regarding the propagation of flame in pipes or the effectiveness of so-called flash arrestors, initiated a movement to obtain experimental data bearing on safeguards for such systems and arrangements were made with Underwriters’ Laboratories to conduct an investigation of the propagation of flame in mixtures of air and vapors of flammable solvents in different sizes and lengths of pipes and the effectiveness of arrestors in preventing the propagation of such explosions. As a contribution to fire prevention work on safeguarding the handling and storage of flammable solvents the expense was borne by the du Pont Company.
Assistance was rendered by Mr. Win. K. Munn and Mr. C. Bockius, formerly Assistant Chemical F2ngineers, in the conduct of the work, also by Messrs. J. I. Banash and Garrett B. James. The scope of the work was determined and valuable suggestions were contributed by Mr. H. L. Miner of the F’ire Protection Division of the du Pont Com. pany.
The investigation was conducted to obtain data of practical application in the industries for the protection from fire and explosion of tank systems and pipe lines using flammable solvents.
The work included tests for explosion pressures in iron pipe ranging from 1/2 to 6 in. in size and 2 to 50 feet in length, and for the effectiveness of flat and cone screen flash arrestors in retarding propagation of flame through pipe of various dimensions up to 6 in. in size and 20 ft. in length.
The propagation of flame was produced by filling the pipe with vapor-air mixtures containing known percentages of vapor, including lean and rich mixtures (moisture content also varied), and igniting it by means of an electric spark. The pressures developed were measured by recording gas-engine indicators ; and the effectiveness of the arrestors by examination of the products remaining in the pipe and tests for their reignition. A few experiments on the speed of propagation were started but discontinued.
The results are in accord with the theory that the pressure developed is dependent on the nature of the vapor and the proportion of vapor in air, and is influenced by the size and length of the pipe.
Gasolene and ether produced maximum pressures ranging from about 75 lb. (per sq. in. above atmospheric) in 1/2-in. pipe* to about 140 to 150 lb. in 4 and 6-in. size (maximum for ether in combining proportions in 4-in. 50-ft. length was probably 500 lb. plus), while alcohol and acetone gave 50 to 100 lb.: amyl acetate and light oil about 70 lb. under conditions employed.
In flu epee of Proportion of Vapor.
Rich and lean mixtures (i. e., mixtures containing vapor or air in excess of the proportions required for complete combustion) gave lower pressures, ranging from a few pounds upwards.
Influence of Pipe Dimensions.
Other conditions being constant, the pressure increased as the size of the pipe (being lowest in the 1/2-in.) but not as the length, apparently being less in 20-ft. lengths than in lengths of 2 to 6 ft. The propagation in 6 to 10-ft. lengths of 1/2 to 1-in. pipe, particularly 1/2-in., continued from end to end of pipe only when the vapor was present in or very near combining proportions, the flame in lean and rich mixtures when in these sizes (1/2 and 1-in.) of pipe (lengths over 6 ft.) being extinguished a few feet from point of ignition. The range giving effectve pressures was found to be much narrower than the range of flammabilitiy.
Nature of Propagation.
Execpt for ether as noted above the propagation in the units included appeared to be progressive combustion of a complex nature of highest intensity in the region along the axis of the pipe and vibratory in character, having a velocity apparently not uniform in the longer lengths (closed ends).
For ether the mode of propagation appeared to be similar to that of the other vapors except in the 50-ft. 4-in. pipe when ether was present in or very near combining proportions, in which case a much more violent explosion was produced. This behavior of ether vapor suggested the possibility of the occurrence of such explosions in the case of the other vapors in tanks or pipes of larger dimensions under favorable conditions.
As indicated by the above results, the effectiveness of flash arrestors in pipe was influenced by the nature and proportion of vapor, the size and length of pipe and the location of the arrestor as well as by the arrestor itself.
Propagation of the flames of ether and gasolene was more difficult to arrest than in the esae of the other vapors, gasolene being second to ether in this respect.
Influence of Proportion of Vapor.
The larger the amount of vapor or air in excess of combining proportions (i. e., the richer or thinner the mixture) the more readily the propagation was arrested, a single screen being sufficient to extinguish the flame in mixtures bordering on the limits of flammability.
Influence of Pipe Dimensions.
In the and 1-in. sizes the flame was more readily retarded than in the 2 and 3-in. pipe; the conditions presented in the 4-in. pipe approached the margin of safety for screen arrestors while in the 6-in. size the gauze was melted. In the ½ and 1-in. sizes increasing the length of pipe from 6 to 10 ft. and upwards augmented the retarding effect to a marked extent: increasing the length of 2 and 3-in. pipe from 6 up to 20 ft. apparently favored the arrestor.
Location of Arrestor.
Increasing the distance between the ignition end of the pipe and the point of location of the screen within the range 1 to about 6 ft. decreased the retarding effect somewhat. When the end of the pipe where ignition was initiated was closed and the other end beyond the arrestor was open the propagation was more difficult to retard.
Effectiveness of Arrestors.
The screen type of arrestors appeared to he more satisfactory than glass, wool or metal chips tor the purpose indicated; 40 mesh No. 31 B. & S. brass wire gave the best results; the cone shape, axis twice diatneter of base, proved superior to the flat form. Double cones (no spacing) and also triple flat screens except for ether in 4-in. unit arrested flame propagation in pipe 4 in. in size and under, including lengths up to 10 ft. for 3 and 4-in. and 20 ft. for smaller sizes, but failed in 6-in. size. Multiple coarse mesh failed. Flash back in 2-in. vents opening into the air was prevented by a single 40-mesh flat screen.
Protection of Industrial Systems.
Applying the results to the protection of industrial systems, it will be noted that flame propagation from tank to tank through connecting pipe is a possibility that should be guarded against.
Connecting pipe should be not over 4 in. in size (not over 3 in. for ether) and preferably 1/2 or 1 in. Length of pipe may be up to 20 ft. for 1/2, 1 and 2-in. and 10 ft. for 3 and 4-in. sizes. Data on longer lengths are not included.
For gasolene and acetone triple flat or for ether, gasolene and acetone double cone screens (no spacing) may be employed. For alcohol and amyl acetate a single cone or double flat screen can be used. The length of the cones should be twice the diameter of the base and have hooked seams (no soldering). Screens should be properly installed to prevent displacement bv pressure. Vertex of cone should be preferably toward possible source of ignition.
For vents in 2-in. pipe a flat screen (40-mesh) or preferably a double flat screen is indicated.
Cold weather favors the formation of dangerous vapor-air mixtures in the case of the more volatile liquids such as ether, gasolene, and acetone, which tend to produce rich mixtures at higher temperatures. Alcohol, amyl acetate and oil are likely to produce dangerous mixtures in warm weather while at lower temperatures their vapors condense tending to produce thin mixtures.
Obviously the exclusion of air from the system or where practical the substitution of carbon dioxide for air would safeguard against propagation of flame*
Extension of Investigation.
The results indicate that an extension of this work, including tests in larger sizes and longer lengths of pipe to determine the nature of the propagation over a wider range and the effectivness of arrestors in connecting pipe when explosion originates in a large tank would yield additional data of much importance in safeguarding industrial systems as well as of scientific interest. Further study of certain phases of this work and an investigation of propagation of flame through pipe of small dimensions and under high initial pres* sures are under way, the results of which it is expected will supplement certain phases of this investigation, enabling a closer correlation of experimental results.
Note: A few tests were conducted in pipe mounted vertically, the results beinK apparently similar to those obtained with pipe in horizontal position.