R. H. Weiland, Optimized Gas Treating, Buda, Texas; C. E. JONES, Ineos, Houston, Texas; and E. van HOORN, Hocon BV, Breda, Netherlands
Hydrogen cyanide (HCN) can greatly degrade amine system performance. Its presence is one of the main reasons refinery amine systems suffer from accelerated corrosion, operability problems and reliability issues. When HCN enters the amine system, its hydrolysis produces ammonia (NH3) and formate, a heat stable salt (HSS). The reaction of HCN with oxygen and hydrogen sulfide (H2S) generates another HSS, thiocyanate, which is also produced by the reaction of HCN with iron. Accelerated corrosion leads to faster formation of particulate iron sulfide (FeS), which in turn leads to filter element plugging, fouled equipment, reduced capacity and more stable foams.
Mass transfer rate-based simulation has been used to study HCN ingress and accumulation in amine systems in unprecedented detail. In this article, the chemistry, prevention and possible mitigation using water washing arrangements and reflux purging are discussed quantitatively. With this knowledge, the incursion and accumulation tendencies for HCN in refinery amine systems are quantified. A separate investigation (results to be reported in a follow-up article) showed the benefits of water washing very high HCN-content gas in a coal tar plant in which a monoethanolamine (MEA) regenerator became fatally corroded in only a few months.
Background and context. Over any substantial period of operation, refinery amine systems can be expected to show an increasing buildup of HSSs. This is especially pronounced for units that scrub gases primarily from coking and catalytic cracking units, where the HSSs are primarily formate (HCOO–) and thiocyanate (SCN–). Amine systems treating gases from various sources experience different rates of HSS incursion. An extreme example is the raw gas generated by the gasification of coking coal with its normally high sulfur and nitrogen contents, which produces the coke used in blast furnaces. The HCN concentration in coke oven gas can be significantly higher [several mole percent (mol%)] than what is ever seen in refinery gases. The actual rate of HSS anion buildup depends not only upon the incursion rate, but also upon the amine losses from the system itself. The primary sources of HSS incursion are summarized in TABLE 1. Note: Amine is a participant in most of these reactions, and there is a direct chemistry link between HCN, formate and thiocyanate.
If left unchecked, the buildup of HSSs will eventually permanently bind part of the amine by protonation, resulting in the loss of treating solution capacity. This is the primary acute symptom. HSSs are also known to complex with iron and accelerate corrosion in the hot, lean section of the amine unit—e.g., amine regenerator reboilers. When the complexed iron contacts higher concentrations of H2S in the absorber, FeS particles are generated. These particles can foul equipment and stabilize foam, leading to a loss of hydraulic capacity, so the operator usually resorts to trading these costs with the cost of replacing filter elements. Conversely, catastrophic corrosion of an amine regenerator and reboiler failure may occur well before solvent capacity loss becomes evident.
HCN is a byproduct of cracking the heavier fractions of crude oil in a refinery [either thermally as in a coker, or catalytically as in fluid catalytic cracking units (FCCUs)], and it is a product of gasifying coking coal to produce metallurgical coke. Gasoil [with a boiling point of 750°F (399°C)] and heavier fractions tend to have greater concentrations of nitrogen than diesel and lighter fractions. Because of the way it was formed, coal is always high in nitrogen. Cracking breaks up the larger nitrogen-containing molecules at high temperatures and low H2 partial pressure conditions that may not be as conducive to complete conversion of byproduct molecules, such as HCN to NH3 as in a high-pressure hydrotreater or hydrocracker.
Therefore, HCN has many sources and occurs quite naturally in refineries and coking plants. Some processes are high producers; others do not seem to produce HCN at all. Once produced, HCN finds its way into the amine system with the H2S-containing gases. HCN forms in various processes within a refinery, whereas HSSs form in the amine system. Once in the amine system, various conditions and the presence of other contaminants allow the HCN to be converted into HSS anions. An ounce of prevention is worth a pound of cure—the most obvious way to prevent HSS anion formation in amine systems is to prevent the ingress of HCN in the first place. One approach that has been tried repeatedly by many refiners is water washing the raw gas and sending the wash water to a sour water stripper (SWS). However, this approach seems invariably to fail because the amine system continues to experience the buildup of HSS anions in operation. The reason for this will become evident.
Water washing refinery gas to remove HCN. Because the thinking tends to be that only a relatively small flowrate of wash water ought to be sufficient to remove HCN from a large gas flow, the tendency has been to use a large wash water flow recirculating through the wash column to ensure proper hydraulics. As shown in FIG. 1, adding a small makeup water flow requires a similarly small flow of blowdown and prevents the SWS from being overloaded.
The rest of this article is restricted to examining refinery conditions. The HCN content in coke oven gas is orders of magnitude higher and will be examined in a subsequent article. This work begins with a typical refinery gas intended to be treated for H2S removal in an amine system. TABLE 2 shows the raw gas analysis and its conditions of temperature, pressure and flowrate. The water wash column contains 20 valve trays sized for 70% of jet and downcomer flood.
In addition to water and H2S, the components HCN and NH3 are of the most interest. They are all modeled by the co-author’s company’s simulatora based on their mass transfer rates. This approach to column simulation accounts not just for vapor-liquid equilibrium, but also for the effects of chemical reaction rates and the mass transfer characteristics of the tower internals on the rate of separation of all the components that are vital to the gas purification process. In the present context, HCN absorption into the wash water is of particular concern.
There are two distinct cases to consider. In the first, the total water circulation rate (Stream 31 in FIG. 2) is 100 gallons per minute (gpm) to which 5 gpm of fresh water is added as makeup and later purged. The other case involves no recirculation of wash water—only freshwater is used to feed the wash column. TABLE 3 shows the effect of adding various freshwater flows to the 100-gpm recirculating flow on the HCN and NH3 removal. TABLE 4 shows the effect of eliminating the recirculation altogether and pretreating the gas with various flows of fresh wash water. With more than 50 gpm of freshwater addition to the basic 100 gpm of recirculation, the use of recirculation has questionable benefits to the column’s hydraulic performance because it is not needed to keep trays functioning properly. At more than 20 gpm of water makeup, NH3 removal is hardly affected, H2S removal is rather minimal in any case and HCN removal is uniformly rather poor; however, NH3 removal is almost complete, while H2S is only minimally removed. The next question then is whether using scrubbing with freshwater alone (no recirculation) can achieve a better HCN-removal result.
TABLE 4 shows the performance of the wash column when scrubbing is done using freshwater alone—no recirculation. One advantage of using only freshwater is immediately obvious: virtually 100% NH3 removal can be achieved, completely preventing the ingress of NH3 into the amine treating unit. This will also result in correspondingly lower NH3 contamination of the amine acid gas feeding a sulfur recovery unit downstream. In contrast to the recirculation case, the same 250 gpm of water used in a freshwater-only case will remove virtually all the HCN and leave H2S as the only impurity to be treated by the amine system.The decision about what freshwater flowrate to use will depend on:
The ability of the sour water system to handle the additional sour water load
Whether there is a minimum acceptable acid gas content of the sour water (see TABLE 4 for sour water compositions under various scenarios)
Whether if below certain acid gas levels in the water it can be used elsewhere before being added to the refinery’s sour water system
The acceptable buildup rate of HSS anions in the amine system.
Regardless of the extent of HCN removal, prewashing with water is an excellent way to remove NH3 from the gas before it enters the amine system. More importantly, this approach can be used to remove as much of the HCN as the SWS is capable of handling through incremental sour water. Water wash will also reduce other HSS such as chlorides, although this was beyond the scope of the present study.
Because NH3 is a weak base and HCN a weak acid, there is a question as to the role NH3 might play in HCN removal. TABLE 5 compares the extent of both HCN and H2S removal with and without NH3 in the sour feed gas going to the wash column. This case has a 100-gpm recirculating flow. TABLE 6 refers to the same situation when there is scrubbing using once-through water. It is readily apparent that regardless of whether there is recirculation or not, NH3 only very marginally improves HCN removal, but it significantly increases H2S pickup in the wash water. H2S is a stronger acid than HCN and preferentially reacts with NH3—HCN seems hardly to react at all. Gas prewashing using only freshwater appears viable for removing undesirable contaminants in the amine system. However, using water wash with recirculation and only relatively modest blowdown does not appear to be a viable strategy. This is in agreement with the experience of refiners who report that high levels of HSS anions continue to build up in the amine system even after implementing the recirculating water wash.
The disposition of HCN in amine systems. If steps are not taken to remove HCN from the sour gas prior to entering the amine system or if the degree of removal achieved is inadequate, the amine contactor will absorb part of the HCN and allow the rest to escape with the treated gas. FIG. 2 shows a schematic of the amine system with a flash tank and a flash gas reabsorber, and the recirculating water wash scheme discussed in the previous section. The raw gas analysis is identical to that in TABLE 2, having 100 ppmv HCN and 1,000 ppmv NH3. The wash water purge rate is 5 gpm with the wash water recirculation (Stream 31) at 100 gpm, and this leaves the composition of the washed gas feed to the amine unit, as shown in TABLE 7.
The treated gas from the amine contactor is simulated to contain only 6.2 ppmv H2S, 1.9 ppmv HCN and 0.47 ppmv NH3. Essentially, 98.3% of the HCN is removed by the solvent flow of 995 gpm. Apart from 0.002% of the absorbed HCN lost in the flash gas, the remainder travels through to the amine regenerator. It is in the regenerator that HCN and NH3 exhibit what might be called interesting behavior.
The regenerator is a 4.5-ft diameter valve-tray column with rich amine feed to the 4th tray from the top (Tray 4) and with the first three trays of a single-pass design, and the bottom 20 trays of a two-pass design. The solvent flows to the regenerator (13 psig head pressure) at 1,012 gpm and enters at 215°F. The reboiler consumes 50,000 lb/hr of 50 psig saturated steam (0.82 lb/gal). This corresponds to a molar stripping ratio of 0.842 (moles water vapor per mole of total acid gas in the overhead vapor line). This results in lean amine loadings of 0.0006 and 0.00055 moles of H2S and HCN, respectively, per mole of methyldiethanolamine (MDEA) with just 0.00015 wt% NH3.
Acid gas to the SRU is 94.8% H2S (wet basis) with 645 ppmv HCN and 23 ppmv NH3. Most of the NH3 in the overhead vapor ends up in the reflux condensate, 80% of which is blown down. The resulting 17 gpm of blowdown contains 170 ppmw HCN, 1,950 ppmw NH3 and 8,750 ppmw H2S to be sent to the sour water system. However, possibly the most interesting occurrence in the regenerator is the disposition of NH3 and HCN.
FIG. 3 shows how the simulated molar flow of HCN in the vapor varies from tray to tray in the regenerator. FIGS. 4 and 5 show corresponding molar flows of NH3 and H2S, respectively. Both HCN and NH3 show an obvious accumulation in a region of several trays below the feed tray. Because of its much higher concentration and flowrates compared with the other components, H2S does not show a peak; however, its stripping is obviously retarded, as can be seen from its slow rate of decrease across several trays immediately below the feed tray. In searching for the reason for the HCN bulge and H2S retardation, the first place the authors looked was to the NH3 traffic. Both HCN and H2S react with NH3: as the solvent descends through the column, it was rationalized that their reactivity with NH3 might explain the simulated behavior. However, even when there is no NH3 in the system, the simulated HCN vapor-phase molar flow was nearly identically unchanged, as was the H2S molar flow profile. The HCN profile and H2S retardation have nothing to do with the presence of NH3.
As FIG. 6 shows, the HCN (and NH3) bulges are related to the temperature profile. Tray temperatures are close to bubble point values, and the bubble point is a strong function of the H2S partial pressure, which is loading dependent. The H2S loading is 0.377 mol/mol in the rich amine feeding the regenerator, and the H2S partial pressure is quite high. The bubble point is the temperature at which the vapor pressure of the liquid phase is equal to the local system pressure, and with the H2S partial pressure being high, this causes the bubble point of the amine solution to be much lower than even that of water. As H2S strips out over the first few trays, the bubble point of the amine solution increases rapidly.
The molar flow (and composition) bulges are present over the top 10 stripping trays, and in this part of the regenerator, the temperature decreases by nearly 40°F across the column. This is exactly where the bulge is observed. In the reboiler and across the lower 10 stripping trays, HCN, NH3 and H2S are stripped from the solvent, but upon reaching the 9th or 10th stripping tray, the temperature begins to fall. By the 4th or 5th stripping tray (Trays 7 or 8 in the figures), the temperature decreases very rapidly, and the already stripped gases begin to reabsorb so their component molar vapor flowrates start to drop. They continue to reabsorb all the way up to the feed tray. The result of this reabsorption is a maximum in the HCN and NH3 vapor phase molar flowrate profiles. The bulges are nothing more than a manifestation of the regenerator temperature profile. These maxima are seen only for trace components. H2S is flowing at some 400 lbmol/hr–600 lbmol/hr in the same region where the HCN flow is 0.5 lbmol/hr. This makes a small change in HCN obvious, whereas even a substantial change in H2S is seen only as a trailing off the stripping rate because it is a dominant component. The authors speculate that the hydrolysis of HCN is most likely to occur in the high-temperature bulge region where the HCN is concentrated. For example, in laboratory experiments, one of the co-authors reported that boiling sodium cyanide (NaCN) dissolved in MDEA under reflux produced HCOO–.
Takeaways. In this article, the production of HCN in refineries was first reviewed with an emphasis on HCN’s role in HSS formation and the corresponding impacts on the amine system. Mass transfer rate-based simulation both confirmed and quantified previously anecdotal observations that recirculating-type water wash systems are, at best, only marginally effective in removing HCN as a feed contaminant. To seriously prevent HCN ingress into refinery amine systems, fresh makeup water is required, and a lot of it. An economic trade-off between SWS capacity and energy usage vs. amine system operating costs (e.g., heat stable cleanup, filtration, capacity, reliability) will ultimately dictate whether water washing is an appropriate choice for a specific refinery system. Finally, the buildup and accumulation of HCN and NH3 in the amine regenerator were quantified in unprecedented detail. The observations of HCN accumulation in the amine regenerator may go some way towards explaining observed regenerator corrosion in existing plants and may inform material selection decisions for plants still in the design phase. However, this surmise may need to be moderated by the fact that the HCN bulges occur where loadings are high, whereas corrosion rates are worst where loadings are lowest and temperatures high (the reboiler area).
In summary, several facets of HCN ingress, prevention and accumulation in refinery amine systems have been quantified that were previously only poorly understood. This real-world behavior would not be reflected in the simulations without the ability to model the system as a virtual plant as available in a proprietary mass transfer rate-based simulatora. HP
NOTE
a Optimized Gas Treating’s ProTreat® simulation software