Better Understanding Needed for Asphalt Tank-Explosion Hazards
Total Page:16
File Type:pdf, Size:1020Kb
Better understanding needed for asphalt tank-explosion hazards David G. Trumbore published has emphasized that the toring program in which we have Owens-Corning Fiberglas Corp. flash point of the material stored is the developed simple methods of evaluat- Summit, III. most important factor in determining ing asphalt tank vapor-space composi- if enough fuel is present in the vapor tions. Charles R. Wilkinson space for combustion.¹ ³ The methods are currently being 4 ¯ Owens-Corning Fiberglas Corp. Dimpfl concluded that factors un- used in our manufacturing facilities to Granville, Ohio related to simple evaporation of as- define specific tank hazards, thus pro- phalt are common causes of the build- viding us with an ever-expanding data Basic differences in the generation of up of combustibles in asphalt tanks. base to aid in predicting and control- combustible gas vapors in tanks stor- He cited smoldering coke deposits on ling high-risk situations. This in-plant ing four classes of asphalt materials the inner roof and shell of the tank, testing is being complemented by lab have been identified by laboratory, incidental thermal cracking, light hy- evaluations of materials, development pilot-scale, and plant-scale measure- drocarbons from solvent deasphalting of new testing methods, and more- ments. (SDA) processes, contamination from sophisticated characterization of actu- It has been clearly shown that clas- crude feed/vacuum residuum heat ex- al vapor spaces. sical methods of thinking about and changer leaks, and unstripped light A verification of the accuracy of the troubleshooting these hazards only hydrocarbons generated during air simpler tank-monitoring methods has work for flux and paving asphalts (and blowing as being potential sources of been completed, and that work is can break down even for these materi- vapor space fuel not detected by flash being prepared for publication. als), and are not applicable to solvent point tests. Vapor-space compositions. To deasphalted residuum and air-blown He emphasized smoldering coke’s date, our measurements of nearly asphalts. role in depleting oxygen and elevating 2,000 individual vapor spaces in over The latter have been shown, for the carbon monoxide and carbon dioxide 200 tanks has led to the view of first time, to pose a special problem in the tank vapor space, and pointed asphalt tank vapor-space composi- due to their continued reactivity after out its potential as an ignition source, tions summarized in Figs. 1, 2, and 3. undergoing the air-blowing process. with or without the catalyzing effect of All three figures present data on Time and temperature are critical iron sulfide. However, based on his four classes of hot asphalt tanks- parameters in determining and con- sampling of nine tank vapor spaces, those storing roofer’s flux (vacuum trolling the degree of this problem. Dimpfl was not able to predict what tower bottoms generally of low vis- Simple measurement techniques material and storage conditions tend- cosity and high flash point that can be were developed to monitor all these ed to give explosive vapor spaces. His processed into high softening point hazards, and these techniques are final conclusion was that the storage grades of roofing asphalt by air blow- considered to be necessary to more of asphalt is an unpredictable art. ing), paving asphalt, SDA, and air- accurately determine the vapor space The work described in this article blown asphalts. Included are exam- hazard in all asphalt tanks. was undertaken to build an under- ples of vapor-space compositions. Background. In spite of the exten- standing of the factors affecting the In the case of the flux and paving sive use of hot asphalt for over a accumulation of combustible vapors asphalt tanks, the values presented century, little has been published on in tanks storing various types of as- represent overwhelmingly typical val- the nature of explosion hazards in its phalt. The key to this effort has been ues. The SDA and air-blown asphalt storage tanks. Most of what has been an extensive tank vapor space-moni- tank values reflected hazardous ex- Fig.1 Fig. 2 Asphalt tank vapor Combustibles in asphalt tanks 9 8 C + 2 HC Flux Paving SDA Air blown II Flux Paving SDA Air blown Reprinted from the September 18, 1989 edition of Oil & Gas Journal Copyright 1989 by PennWell Publishing Company Fig. 3 Fig. 4 Hydrocarbons in tank vapor space Air-blown asphalt* L 4 *Causes more combustibles evolution l 380 400 420 440 460 480 C1 C3 C5 C7 C9 C11 Cl3 C15 B SDA x Air blown o Flux C Paving Storage temperature, °F. OGJ I OGJ . tremes that, while far from typical, was the only tank example to have a create a tank vapor space problem were encountered often enough to be significant amount of hydrogen sulfide because of solvent contamination of general concern. in the vapor space. from the deasphalting process if the At the other extreme, the vapor Finally, Fig. 3 breaks the “other steam stripping of the material is not spaces in tanks storing these two ma- hydrocarbon” fraction of Fig. 2 into a adequate. This problem appears to be terials look very much like the vapor profile of carbon number molecules. a strong function of the supplier’s spaces in the flux and paving tanks. These data indicate that the flux, pav- process, the solvent used, and the The vapor space of an SDA tank was ing, and air-blown asphalt tanks had a amount of ventilation in the tank generally at one of these two ex- broad spectrum of hydrocarbons pre- where the material is stored. tremes, while in air blown asphalt sent in their vapor spaces, whereas Because of the high volatility of the tanks both extremes and all points in the SDA tank had a single spike at the solvents in question (propane through between are common. carbon number that in every case of pentane), these hazards are not de- Fig. 1 divides the example tank high combustible gas readings in our tected by flash-point tests. vapor space compositions into four experience corresponded to the sol- Finally, from the data in Figs. 1 and components: combustibles, carbon vent used in producing this material. 2, it is obvious that air-blown asphalt dioxide, water vapor, and oxygen, The data summarized in the above is an entirely different material from a and presents them in a stacked bar examples give a view of hot asphalt tank-hazard standpoint, and it de- chart. The difference between the to- tanks as follows. serves some more extensive discus- tal bar height and 100% is nitrogen Combustible gas buildup in flux sion. The basic uniqueness of this from air. Note that none of the tanks and paving tanks can typically be material from a loss-prevention stand- discussed in this study were inerted. characterized by their flash points. point has not, to our knowledge, been From the data, it is clear that typical Since these materials are generally recognized before. flux and paving tanks have little more stored at temperatures substantially Again, the high levels of volatile than nitrogen, oxygen, a little water below their COC flash point, the va- combustibles, like methane and car- vapor,and a very small amount of por spaces are typically low in com- bon monoxide, make flash-point test- combustibles in their vapor spaces, bustible gas levels. ing inadequate. the SDA residuum tanks had, in the Flux and paving tanks thus tend to Air-blown asphalts. An experiment extreme case, a large combustible gas follow the traditional view of asphalt done in Trumbull’s pilot plant conver- fraction, and again in the extreme tank safety as being well-defined by tor illustrates the basic difference in case, the air-blown asphalt tanks had flash point measurements. Contami- the air-blown asphalt hazard. Roofer’s a large combustible gas fraction as nation problems with these materials flux was loaded into the converter well as large water vapor and carbon are not unknown but not commonly (1,000-gal capacity) and stored at a dioxide levels. encountered. specified temperature overnight. Fig. 2 divides the important com- When contamination does occur, The heating medium in all these bustible gas fraction into four addi- the control of hazards by flash point tests was hot oil circulating through a tional levels: methane, other hydro- can break down, depending on the heat exchanger countercurrent to the carbons, carbon monoxide, and hy- contaminant and the flash point meth- asphalt flow. The system was such drogen sulfide. The data on our exam- od generally used. We have moni- that the hot oil was never more than ples, once again, indicate nothing tored over 100 tanks storing these 25° F. hotter than the target asphalt hazardous in the typical flux and pav- materials and have only seen elevated storage temperature. ing tanks and only the “other hydro- combustible gas levels with flux The vapor space in the convertor carbon” fraction in the SDA tank. stored at high temperatures where the was sampled in the morning for very The extreme air-blown asphalt flux was inadvertently aerated, and light combustibles. This sampling, tanks, on the other hand, had large with paving that was made from which will be discussed later, used an methane and carbon monoxide com- blending solvent contaminated SDA activated charcoal tube to adsorb hy- bustible fractions in addition to a large material. drocarbons of carbon number three “other hydrocarbon” fraction, and Solvent deasphalted residuum can and higher so that only very light-end combustibles were measured. and shell of the tank. This phenome- The flux was then air blown to one The authors... non could lead to some of the obser- of a variety of endpoints ranging from vations discussed in this article, but 160°F.