Cyanide Sources in Petroleum Refineries
B. VAIL PRATHER, Staff Specialist ROBERT BERKEMEYER, Laboratory Manager Williams Brothers Waste Control, Inc. Resource Sciences Park Tulsa, Oklahoma 74136
INTRODUCTION
It has been known for a number of years that some petroleum refinery wastewater effluents contain traces of cyanides. The nature or configuration of cyanide contained in the water has not been formally reported, but is usually assumed to be organic or possibly metallic complexes, subject to decomposition in acidic environment into hydrocyanic acid. Recent intensified interest in cyanide content of wastewater effluents from all sources and more particularly petroleum refinery effluent waters prompted Williams Brothers Waste Control, Inc. to embark on a cyanide research program under contract with an oil refining corporation. The customary point of initiation in any research project is a literature search. We followed this routine, and found there is considerable information on the treatment of cyanide wastes from metal finishing and plating which contain concentrations of 50 to 300 mg/1 of cyanides. Information on cyanides in petroleum refinery wastewaters, however, is scant and incomplete, with very little data, leaving much to the imagination. In order to define the cyanide problem in the refinery, it was logical to locate the sources of cyanides within the refining process. Accordingly, a plant survey was initiated to determine those sources as the initial plant action of the research project. Although there is no published data on cyanide production at refinery units, we did have private data from one refinery which illustrate the concentrations that exist at various points in the refinery. Review of this data is shown in Table I. Although Table I lists concentrations, it does not list water volumes; therefore, it is impossible to calculate pounds per day of cyanide produced. However, the data suggest that cyanides are present in the crude, since analyses of crude tank bottoms waters report them in small concentrations. No information was given regarding the source of the water which was used in the desalting process. If the source water was free of cyanide, then the high content shown in the desalter effluent might be due to washing of the cyanides from the crude, or the electric desalting process might be forming cyanide. Atmospheric crude unit fractionator overhead separator waters varied considerably in cyanide concentration. Again it should be noted that water volumes were not listed. 306 TABLE I CYANIDE CONCENTRATIONS IN PETROLEUM REFINERY WASTES
Source of Wastewaters Cyanide Concentration Range (mg/1) Crude tank water draw off 1.2 to 2.8 Desalter water 26.0 to 38.0 Crude unit overhead separator <0.5 to 26.0 Vacuum unit barometric condenser 20.0 to 26.0 Asphalt plant condenser 10.0 to 32.0 Gas recovery plant 16.0 to 26.0 Cat cracker overhead drum Sweet crude 23.0 to 80.0 Sour crude 82.0 to 144.0
The main fractionator overhead drum waters from the cat cracker operating on feed stocks derived from sweet crude oil had a much lower concentration range than the cat cracker operating on feed derived from sour crude oil, which poses an interesting question. Using this background information as a partial guide, we structured a survey program designed to furnish data from which we could not only determine the concentration of cyanide in the waters from the various units of the refinery, but also the pounds per day of cyanide produced. Crude tank bottoms water is drawn off at intervals when the water levels in the tanks are high. This occurs about once per week, and the volume is low. Since the concentrations of cyanide were low in this water (0.02 to 0.05 mg/1), it was agreed that no consideration should be given this source at this refinery. The analysis did serve to point out the fact that cyanide evidently existed in the crude oil as it was received. Crude unit overhead separator water draw-off was measured by bucket and stop watch, and the resulting data was checked against the steam consumption in the crude fractionation column. Correlations were made in order to translate steam consumption into gallons per minute of condensate waters in the overhead separators. We found this method of calculation at all distillation and fractionating units satisfactory for our survey. At the fluid catalytic cracking unit, two condensate streams were of importance: the fractionator overhead receiver drum and the second stage condensate waters. The latter contained some wash waters at times which were metered. Consequently, there was some variation of volumes from the second stage source. All of the waters from the cat cracker flowed to the sour water stripper for processing. Further discussion of the sour water operation follows later in this paper. The delayed coker was inoperative during a portion of the time span of the survey, and during those times was not a contributor to the total cyanide production. During the shutdown of the coker, the reduction of cyanide production was noticeable. The survey was conducted one day each week and continued from January 22 through April 30. Samples of all water streams suspected of containing cyanides were taken and preserved during the morning of one day each week. All samples were subjected to laboratory analyses in the afternoon of the same day as collected. Preservation and analyses of the samples were performed in accordance with Standard Methods for the Examination of Water and Waste Water, thirteenth edition, 1971. Due to the possibility of cyanide loss on standing over a period of time, even when the sample is preserved, no attempt was made to composite samples for more than one hour. Therefore, the samples will be considered as "grab" samples. However, care was taken to be sure that units in the refinery was operating normally at the time samples were drawn, and operating data were recorded during this 307 time. Figure 1 depicts graphically the total cyanide production in the refinery each week of the survey.
Figure 1 — Refinery production of cya nide (pounds per day at weekly intervals).
During the winter months, the production was low, and increased perceptably from March 26 until the survey's end. This was attributable to a combination of increased crude throughput, operation of the delayed coker and increase in the cat cracker charge volume.
CYANIDE SOURCE CLASSIFICATION Biocides used for control of algae and bacteriological growth in the cooling towers contained a complex organic cyanide compound, and therefore must be considered as a cyanide source. Samples of cooling tower blowdown were analyzed for cyanide content, and the pounds per day contribution to the wastewater were calculated, based on the blowdown volume. This was considered as a controllable source. Since effluent waters from all molecular rearrangement units were fed to the sour water stripper, process sources of cyanides could be divided into two groups: namely the crude distillation units and the molecular rearranging or "sour water" units. Thus the classification of cyanide sources became noncontrollable and controllable. The noncontrollable source is further divided into the "existing cyanide" group which includes the crude distillation units and the "synthesis cyanide" group which includes molecular rearranging units such as the fluid catalytic cracking unit and the delayed coker unit. Table II shows the cyanide production from all sources in the refinery in pounds per day during the three month survey arranged according to classification and groupings. Obviously, only the sum of the unit production on an average basis will equal the total production in Table II. However, the total production shown in the high and low categories will not be off more than three percent. It is quite pertinent that the great bulk of the cyanide contribution is found in those units which have to do with molecular rearrangement of the hydrocarbon molecules such as the cat cracker and delayed coker. It is also pertinent that these same units produce the bulk of sulfide and phenolic waters.
CRUDE THROUGHPUT Refinery crude throughput during the survey period varied from 25,000 barrels per day to 33,000 barrels per day. Table III shows the cyanide production in pounds per 1,000 barrels of crude throughput at three time intervals. 308 TABLE II CYANIDE CONTRIBUTION FROM REFINERY SOURCES IN POUNDS PER DAY
Sources Noncontrollable Controllable Existing Synthesis Crude (Sourwater) Cooling Total Units Units Towers Production High 1.6583 65.7000 0.4390 66.3831 Low 0.0119 3.0700 0.0090 3.1769 Average 0.4387 24.0334 0.1372 24.6093 % of Total 1.7800 97.6600 0.5600 100.0000
TABLE III POUNDS CYANIDE PRODUCED PER 1,000 BBLS. CRUDE THROUGHOUT
High Low Average 1-22/2-19 0.6041 0.1442 0.3718 2-26/3-26 1.3672 0.1232 0.6156 4-2 /4-30 2.0732 1.2529 1.5424
The average charge to the fluid catalytic cracking unit during the first period (1 / 22 to 2/19) was 13,540 barrels per day. During the second period (2/26 to 3/26) it was 14,261 barrels per day, and during the third period (4/2 to 4/30) it was 18,480 barrels per day. Since the cat cracking unit is responsible for a large portion of the cyanide produced, an increase in the pounds per thousand barrels of crude throughput would be expected. No correlation could be made in regard to the crude source. The excessively high cyanide production during the April period of the survey might be attributed to the operation of the delayed coker during this period as well as the increase in cat cracker charge volume.
CRUDE UNIT STUDY A study of the cyanide production in the crude unit operation revealed some information of interest. Figure 2 shows the production of cyanide in pounds per 1,000 barrels of crude charged during the survey. Wide fluctuation in the early part of the survey prompted a look at the nitrogen content of the crude in the hope that some clue could be found for the fluctuation. Only a general trend could be established. Only a small quantity of cyanide is contributed by the crude units. This was to be expected since there is only slight molecular rearrangement in crude fractionating units, if any. But it does make the point that higher nitrogen crudes may contain higher cyanide concentrations. Figure 3 displays this trend. Our interest at the crude units was also concerned with the question of which crude distillation process produced the most cyanides, the atmospheric or the vacuum distillations. Table IV shows the distribution of cyanides in the two processes. Most of the cyanide appeared in the atmospheric column overhead condensate water. It should be mentioned here that this particular refinery utilized sour water stripper effluent as a water supply for desalting. Removal of cyanides from stripped foul water used 309 in the desalting process is discussed later in this paper. This may account for a small portion of the cyanide found in the atmospheric distillation unit condensate water.
Figure 2 — Crude unit study. Cyanide production (pounds cyanide per 1,000 barrels crude).
Figure 3 — Crude unit study. Pounds cyanide per 1,000 barrels
4 0 5 06 07 08 0 9 10 II 1.2 I 3 14 IS crude versus crude nitrogen CRUDE NITROGEN CONTENT % BY WEIGHT content.
SOUR WATER UNITS STUDY As previously mentioned, the bulk of the cyanide production was contributed by those units which rearrange and split the hydrocarbon molecule. These units produce the so- called "sour water" which contains, among other things, sulfides, ammonia and phenols. These are toxic materials and must be removed from the wastewaters before discharge to receiving bodies. Best practice is to treat these waters at the source before discharging them to the plant sewers. The treatment usually consists of stripping the sour waters with steam, flue gas or natural gas under controlled conditions. At this refinery, the sour waters from these units are collected in a surge tank from which they may be charged at a reasonably constant rate to the sour water stripper. Hence
310 we designate this portion of the survey as the "sour water units" study. Figure 4 shows the total cyanide production from these units in pounds per day during the survey.
TABLE IV DISTRIBUTION OF CYANIDE PRODUCTION IN CRUDE DISTILLATION UNITS (Cyanide in Pounds/Day)
Atmospheric Unit Overhead Vacuum Unit Barometric No. 1 No. 2 No. 3 No. 4 Separator Separator Condenser Condenser High 0.6443 0.9936 0.0180 0.0190 Low 0.0019 0.0009 0.0012 0.0024 Average 0.2278 0.1958 0.0075 0.0075 0.2118 0.0075
Figure 4 — Sour water study. Cyanide production (pounds per day).
Fluctuations in cyanide production may be accountable to refinery operations. During the earlier part of the survey, the coker was not operative, and charge to the fluid catalytic cracking unit was at a minimum. Later, as demand for manufactured products changed, refinery operations changed, and greater volumes of feed stock were sent to these units for processing with the resultant increase in cyanide production. Sampling of each sour water stream at the source was undertaken to determine the contribution of cyanide from each unit. Table V shows the results of these studies. Table V is divided into three different time periods to reflect the differences due to changes in refinery operations. The first time period does not reflect coker operations, and the total cyanide production in each of the first two periods is less than thirty percent of the production during the third period. The "other" sources reported here are from light oil processing units which contribute some sour waters to the total charge to the sour water stripper. During the first period the cyanide produced from "other" sources is about thirty-six percent of the total produced from sour water sources. During the second and third periods when the coker was operating, the cyanides produced from the other sources calculates about three percent of the total. 311 TABLE V CYANIDE PRODUCTION OF SOUR WATER UNITS
Average Pounds Per Day Cyanide Produced From Period Total FCCU Coker Other 1-22/2-19 12.62 8.05 4.57 2-26/3-26 13.40 6.03 7.37 4-2 /4-30 46.10 33.56 10.78 1.75
FCCU SOUR WATERS Sour water samples were obtained from two sources at the fluid catalytic cracking unit. One source was from the second stage unit in the gas condensing section of the cracker. The other was from the FCCU overhead receiver drum which contained condensate from the FCCU fractionator overhead and waters from the second stage gas condenser as well. Quantities of water and cyanide from fractionator overhead were determined by subtracting the second stage flow and cyanide content from the total flow and cyanide content from the receiver drum. Figure 5 shows the distribution of the cyanide production of the FCCU. It is presumed that cyanides which escape solution in the overhead condensate will be dissolved in the condensate or wash water in the second stage section. Operational modes may account for fluctuations in the distribution.
Figure 5 — FCCU sour water. Percent of total cyanide in overhead.
No correlations could be made in cyanide production per 1,000 barrels of charge stock versus charge rate. Neither was there any correlation of the cyanide production per 1,000 barrels of charge stock versus reactor temperatures. However, we did find that the crude nitrogen content did have a definite effect.
EFFECT OF CRUDE NITROGEN CONTENT Where catalytic cracking processes are utilized it is general refinery practice to take the FCCU charge stock direct from the crude throughput each day. With such an operation we could compare the cyanide production from crudes with various nitrogen contents. Figure 6 shows the effect of crude nitrogen content on the cyanide production at the FCCU. Note that as the percent nitrogen by weight in the crude increases, the cyanide production in pounds per 1,000 barrels of charge increases accordingly.
312 1.40 /O 1.35 1 .30 1.25- / 1 .20 / 1 . 15 O / 1 . 10 1 05- / 1 .00 / .95 / .90 / .85 • / .R0' / .75- cr o .70 .65 V .60 .55 <3 / O .50- .45 .40 .35" .30 .25
-«rt*niDS««O-Nl'lt«(0N9l0lC Figure 6 -- FCCU effect of crude nitro FCCU CN. POUNDS / 1,000 BBLS CH4RGED gen on cyanide production.
No data was recorded on the charge rate to the delayed coker; therefore, no attempt was made to relate the nitrogen content of the crude to the cyanide production at this unit.
SOUR WATER STRIPPER STUDY The sour waters contain the bulk of the cyanide produced in the refinery and consequently the disposition of this water becomes very important. There are other polluting components in these waters as well as cyanides. Sulfides, ammonia and phenols are the most prevalent. Most petroleum refineries subject these waters to treatment by stripping to remove the sulfide, and this refinery is no exception. In the stripping of sulfides, additional benefits are obtained by the stripping of some of the ammonia, and, in some instances, a portion of the phenols. Our attention therefore was directed to the fate of the cyanides in the stripping process.
Figure 7 — Sour water stripper cyanide 0»TE removal. 313 Figure 7 gives the results of this study. The data obtained demonstrates the ability of the stripper to remove most of the cyanide even when high charge rates were experienced. The percent removal does not necessarily follow the charge rate, but it can be seen that the last residual cyanide is difficult to remove. This particular stripper is highly efficient in sulfide removal and has been designed for this purpose. Ammonia and phenol removal is secondary to sulfide stripping. The effluent from the stripper therefore must undergo further treatment to further reduce those contaminants. Biological oxidation of the phenols is recognized as a viable process, but strict controls generally are necessary to level out the phenol loading to the biological units. Furthermore, low loadings are desirable. To further reduce the phenolic content of the stripper effluent, and to practice water conservation by reuse, this effluent is utilized in the crude desalting process. We therefore turned our attention to the fate of cyanides in this unit.
CRUDE DESALTING STUDY Desalter water was shown in Table I as a cyanide source, and some conjecture was made regarding the mechanics of solution of the cyanide in this water. The desalting process at the refinery reported was electrolytic. The desalting unit utilized in the refinery under study here is a chemical process, and perhaps the two processes may not be compared in respect to cyanides. Table VI shows the results of the desalter water study. Although the pounds of cyanide per day charged to the desalter is small, there is a significant removal of cyanide in the desalting process. TABLE VI DESALTER WATER STUDY
Cyanide Pounds/Day Sample Charge Effluent Percent No. Water Water Removed Removal 1 0.23 0.10 0.13 56.5 2 0.43 0.09 0.34 79.0 3 0.48 0.31 0.17 35.4 4 0.58 0.08 0.50 86.2 5 0.86 0.21 0.65 75.6 6 1.40 0.09 1.31 93.6 7 1.44 0.13 1.31 90.9 8 1.76 0.23 1.53 86.9 9 3.48 0.89 2.59 74.4
The cyanides, fickle as they are, would show no correlation of percent removal with pounds per day charged to the desalter unit. The possibilities for cyanide removal through the desalter are promising, and present a bonus when the process is used for phenol removal from the stripped sour water.
SOUR WATER CONTRIBUTION TO SEWERS To determine the sour water cyanide contribution to the sewers, both the sour water stripper and the crude desalter were considered. Only a portion of the stripped sour water was used in the desalter, and the balance was discharged to the sewers. A material balance across these systems will show the total contribution to the sewers. 314 Figure 8 displays this balance, taken from the period of highest cyanide production in the sour water units. Only two percent of the cyanides produced entered the sewer system.
50- 46.1 3.60 2.75 0.85 0.17 1.02
!z Oi 40- i= or . • DISCUSSION The data presented was developed from laboratory analyses of samples which must be considered "grab samples." Results of these analyses were extrapolated into pounds per day data by using flow measurements of the various streams analyzed. The pounds per day concept was employed in order to gain perspective rather than to provide an absolute number. Technical personnel employed in petroleum refineries are familiar with the variations of operations during a twenty-four hour period and will understand the problems involved in such a survey as was conducted. Information from the survey indicates that those units in a refinery which are recognized as rearrangers of the molecular structure of the hydrocarbon molecules produce ninety-six to ninety-eight percent of all the cyanides found in petroleum refinery wastewaters. It also indicates that the cyanide production at these units is influenced by the nitrogen content of the stocks being charged to those units. If we may indulge ourselves a little, some speculation as to the mechanics of cyanide formation is presented. CYANIDE FORMATION SPECULATION Research often starts with speculation, and the case under consideration here is no exception. The greatest source of cyanide as indicated previously is the Fluid Catalytic Cracking Unit and the Delayed Coker. Both of these refining units operate on a partial decomposition or molecular rearrangement of the long chain hydrocarbons present in the charge stocks. Both processes rely on pyrolysis of the long paraffin hydrocarbons with or without catalysts. In the catalytic cracking process, the charge stocks are generally saturated hydrocarbons, although some unsaturation may be present. In some instances, these stocks may undergo a hydrogenation process to assure saturation before charging to the cat cracker, so that better yields may be experienced. 315 In the process, the longer chains are decomposed, or cracked, into the shorter chain components, some of which have a degree of unsaturation (1). Large quantities of gas are formed which consist principally of low boiling paraffins, olefins, diolefins and some aromatic cyclic structures, along with hydrogen. Nitrogen compounds contained in the stocks charged to the catalytic cracker may be the amines, amidines, nitriles or isonitriles. Figure 9 illustrates the configuration of such compounds. When such compounds are cracked, a number of nitrogen compounds may result such as ammonia (NH3), hydrocyanic acid (HCN), and, when sulfur is present, thiocyanic acid (HSCN) and thiocyanates (RSCN). The short chain nitriles and isonitriles may be present without changing the nitrogen configuration (see Figure 10). SCCOMMJtT O -O Figure 9 — Possible nitrogen com pounds in petroleum charge stocks. Figure 10 —Hydrocarbon cracking. The fact that a nitrogen source other than that contained in the charge stocks is possible in the cracking process must not be overlooked. Regeneration of the catalyst is accomplished through oxidation of the hydrocarbons which are adsorbed on it. Copious quantities of air are used for this purpose, and nitrogen may be entrapped in the regenerated catalyst returning to the cracking process. Since the cyanide production appears to increase with the nitrogen content of the charge stocks, it may be assumed with a reasonable degree of credibility that the nitrogen from the catalyst probably unites with the hydrogen released during the cracking process to form ammonia rather than the carbon which is already bound. Later in this discussion we shall see that nitriles and isonitriles are measured as cyanide in our testing methods. In the delayed coking process, decomposition of the hydrocarbon continues to a deposition of free carbon, bound by a certain quantity of heavy tars into a solid mass. During the decomposition process, light hydrocarbon by-products are recovered for use in blending fuel stocks. Condensate waters from this recovery contain the cyanides. Identification of each carbon/ nitrogen compound is highly desirable if an attempt is to be made to trace the mechanism of cyanide production in such processes as catalytic cracking. Our research has not progressed to this stage. Analyses for total cyanides has been used throughout the study thus far. 316 LABORATORY TESTING FOR CYANIDE In general, manuals of methods for the chemical analyses of water and wastewater provide procedures for TOTAL Cyanide (2, 3, 4). Some manuals provide procedures for cyanide amenable to chlorination (2, 4). In either case, both simple and complex cyanides are measured. In the EPA Manual 1974 under STORET No. 00720, cyanides are defined as "cyanide ion and complex cyanides converted to hydrocyanic acid (HCN) by a reaction in a reflex system of a mineral acid in the presence of cuprous ion." Other manuals follow the same conversion step in the procedures. The various complex cyanides which will yield to this reaction include the nitriles, isonitriles and the metal complexes. Further analytical research is indicated to identify the various complexes and to differentiate between them as well as the hydrocyanic acid. Where cyanide concentrations are high, analysis problems are usually minimal and precision and accuracy are acceptable, but in the ranges below the milligram per liter level, the reproducibility of results by several laboratories leaves much to be desired (5). Huge errors may be encountered in calculating pounds of cyanides produced per day from such low concentrations. Fortunately in this survey, those wastewater streams showing such low concentrations of cyanide were low in volume, and the overall computation would be only slightly altered by such errors. CONCLUSION Certain conclusions may be drawn from the reported study of cyanide sources, but further research is necessary to verify the data taken from a single refinery. Other refineries will have additional processing units, other crude oil types and other operational procedures. The survey in this particular refinery indicated that treatment of the effluent from the foul water strippers for cyanide removal would probably produce a plant effluent with no detectable cyanide concentration. In another refinery, however, different situations may emerge which could alter the conditions enough to render such treatment impractical. Each case must be judged on its own merits. The study pointed out that those processes which produced a molecular rearrangement and splitting of large molecules of hydrocarbons are the offending units of the refinery. It further disclosed that the crude oils with higher nitrogen content will produce a larger quantity of cyanide during processing. Therefore, the crude nitrogen content is of importance. Continuing research and study is contemplated at Williams Brothers Waste Control on this important subject. REFERENCES 1. Nelson, W.L., Petroleum Refinery Engineering, McGraw-Hill, New York, Fourth Edition (1958). 2. U.S. Environmental Agency, Manual of Methods for Chemical Analysis of Water and Wastes. (1974). 3. Standard Methods for the Examination of Water and Waste Water, Thirteenth Edition, APHA AWWA WPCF (1971). 4. ASTM Standards, Part 23. Water. (1973). 5. Illinois Petroleum Council, Subcommittee for Cyanide Testing Report (1974). 317