E. W. van Hoorn, Hocon b.v., Breda, The Netherlands; and R. H. WEILAND, Optimized Gas Treating Inc., Buda, Texas
Coke oven gas is a high-tonnage source of sulfur and high-value, byproduct coal chemicals, such as aliphatic and aromatic hydrocarbons, polycyclic aromatics, phenols, organic acids, etc.; coal tar pitch; and (after scrubbing) a clean environmentally friendly gaseous fuel. Historically, the byproduct chemicals were of high value in agriculture and in the chemical industry, and the profits made from their sale were often of greater importance than the coke produced. Note: Today, most of these same products can be more economically manufactured using other technologies, such as those of the oil industry.
Raw coke oven gas is predominantly hydrogen (H2), methane (CH4) and carbon monoxide (CO), making it an excellent high-calorific-value fuel. Some of the additional components—tar vapors, light oil vapors [aromatics, mainly benzene, toluene and xylenes (BTX) or BTX components], naphthalene, anthracene, pyrene and carbazole—provide coal tar chemicals used in dye synthesis, pharmaceuticals, etc., while others [ammonia, hydrogen sulfide (H2S) and hydrogen cyanide (HCN)] are lethal contaminants and must be removed.
In the first treatment step, raw coke oven gas is cooled either by direct contact with flushing liquor or via a heat exchanger to lower the temperature and greatly reduce the vapor volume. As the raw gas is cooled, tar vapor condenses, forming aerosols that are carried along with the gas flow. To prevent these tar particles from fouling downstream burner nozzles, for example, tar precipitators typically use high-voltage electrodes to charge the tar particles and collect them from the gas via electrostatic precipitation.
Before being combusted as fuel, coal tar gas is generally treated to remove H2S using either monoethanolamine (MEA) or diethanolamine (DEA) solutions. This is quite a challenging service because of the high particulates content and due to corrosion caused by the presence of HCN, a precursor to the formation of heat stable salts (HSSs). At one time, potassium salts of amino acids—so-called alkazids—were used as solvents for H2S absorption in several European steel mills because the alkazids show greater resistance to degradation. They have been supplanted by methyldiethanolamine (MDEA). The coal tar itself also contains valuable chemicals in addition to H2S, so it is treated separately by vacuum and atmospheric distillation. Removing residual components leaves tar pitch used as anodic material in aluminum production. Treating the volatiles stripped from the coal tar and mitigating corrosion by removing its HCN content are the subject of this article.
HCN can degrade amine system performance. When HCN enters the amine system, its hydrolysis produces ammonia and formate, an HSS. The reaction of HCN with oxygen and H2S generates another HSS, thiocyanate, which is also produced by the reaction of HCN with iron. HSSs are highly corrosive to steel and accelerated corrosion leads to faster formation of particulate iron sulfide, which in turn leads to filter element plugging, fouled equipment, lower capacity and more stable foams.
Mass transfer rate-based simulationa is used here to illuminate the effectiveness of water washing the raw coal tar gas to remove HCN and ammonia before they can enter the amine system and form HSSs. The results show that water washing can reduce HCN from mole percent (mol%) levels to parts per million (ppm), and can almost completely remove ammonia while absorbing very little of the H2S.
Coal tar gas treatment. The two main sources of H2S that must be removed in the gas treatment plant are gas generated by removing volatiles from the coal tar itself, and the coke oven light oil (COLO) gas remaining after aromatics have been extracted from the coke oven gas. As shown in the process flow diagram in FIG. 1, these two streams are processed individually in separate amine treaters. The loaded DEA solvent from the treaters are combined and regenerated in a single stripping column. The coal tar gas, however, is first water washed to prevent most of the HCN from entering the amine system.
TABLE 1 shows the raw coal tar gas and COLO gas compositions. Both gas streams are high in H2S, and both are dry but the coal tar gas has a very high HCN content—4.1 mol%—a value rarely if ever seen in oil refining but perhaps not unusual for coke oven and coal tar gases. As discussed later, the HCN content caused some of the original amine treating equipment to fail completely within only three months or so from startup when operating with MEA. At that time, the decision was taken to water wash the gas to remove the HCN.
The case being investigated here is a troubleshooting exercise that began nearly 10 yr ago. At that time, an MEA unit with a reclaimer was installed to remove H2S and other components from the coal tar gas. However, corrosion almost completely destroyed this MEA unit within a span of only 3 mos. The regeneration section was fatally corroded although the two parallel MEA absorbers survived. Note: The two parallel absorbers were installed so that in this severe service, they could be taken offline one at a time for cleaning. Carbon steel was a poor materials choice for these levels of HCN and the accompanying high HSS level at regenerator temperatures.
From 2017 to the present, the processor repurposed the two parallel MEA absorbers into water wash columns to remove as much of the corrosive contaminants as possible before the gas entered the amine system. Based on lab results, they found that water washing removed most of the HCN as well as most of the ammonia, but almost none of the H2S. In other words, in this instance water was a very selective and effective solvent.
One of the important questions for the present investigation is whether DEA can be used for both the coal tar gas and COLO gas streams with solvent stripping in a common DEA regenerator—without destroying the stripper. Another objective is to determine whether simulation can confirm the ability of water washing to achieve the observed results. Recent analyses have confirmed the original measurements made by the processor. A set of simulations was carried out to provide a substantiable answer to these questions.
Simulation results are shown in the callouts in FIG. 1. Simulation shows that the water wash column removes all the ammonia and reduces HCN from 4.1% to 0.6 ppmv. It allows 94% of the H2S to pass through to the DEA system where it is removed to 30 ppmv in the tail gas destined for incineration. The wash column is remarkably selective for the HSS-forming components HCN and ammonia. All the ammonia and virtually all the HCN end up in the sour water. However, the simulated HCN and ammonia levels in the washed gas do not correspond to measured data nearly as closely as one might hope. Two gas samples taken minutes apart showed HCN levels of 681 ppmv and 376 ppmv. A third sample taken 21 hr later showed 292 ppmv HCN. The simulated HCN concentrations are single ppm, far removed from measured values.
FIG. 2 is a photograph of a 5/8-in. Pall ring as it appeared when removed from the wash column. Severe fouling of the packing is evident, to the extent that the available interfacial area has likely been reduced to a small fraction of its original clean-packing value. The simulations are based on gas- and liquid-side mass transfer coefficients and interfacial areas for clean packing. Apart from its effect on interfacial area, it is next to impossible even to estimate the effect of the level of fouling shown here on mass transfer coefficients.
Nonetheless, reducing the wetted area by a factor of 0.25–0.30 gives HCN levels similar to measured data. However, for reasons unknown, predicted ammonia levels remain less than 1 ppmv vs. 64 ppmv–71 ppmv measured. Ammonia and HCN absorption are controlled by gas-liquid interfacial area and by resistance to mass transfer in the gas phase. Although comparatively minor, the liquid also offers resistance, but there are too many parameters to juggle—predicting the performance of heavily fouled packing in this case cannot be convincingly done using mass transfer correlations for clean packing.
It is worth noting that, as evidenced by FIG. 3, small packings are much more susceptible to severe performance degradation from fouling than are large packings. Fouling by deposits of gums, tars and coke more quickly fill the volume within the packing as well as the space between the packing pieces in the packed bed.
The important learning from this part of the study is that corrosion in amine systems caused by even very high levels of HCN and ammonia can probably be eliminated by water washing the gas before it enters the amine system. Water is very selective towards these components and removes only a little of the H2S (and CO2), leaving most of the H2S to be removed by the now HCN-free amine solvent and recovered as elemental sulfur in a solvent recovery unit (SRU), for example. However, it must be said that in line with the findings reported earlier,1 even though only a small water flow is needed to remove most HCN and ammonia, it is critical that HCN be removed to a few ppm at most, so a small recirculating water flow is probably an inadequate process configuration. As quantified below, HCN builds to too high a level in such a recirculating flow.
Sensitivity of water washing to fouling. Disregarding the effect of fouling on the gas and liquid film coefficients for mass transfer, fouling certainly reduces the interfacial area available for interphase mass transfer. FIG. 4 shows how the reduced fractional interfacial area affects the HCN removal performance of the water wash column. Therefore, if fouling reduces the interfacial area to 20% of its clean-packing value, HCN is removed to only 1,000 ppmv. This is not nearly low enough to mitigate severe corrosion. FIG. 4 also indicates that even with perfectly clean packing, a 5-m bed depth under normal operation is just barely enough to reduce HCN to a single-digit ppmv level.
The clean wash column operates at just below 20% of flood, so a common strategy to achieve better contacting in such a lightly loaded column might be to recirculate a large portion of the wash water flow and add just enough fresh water to keep the HCN level in the washed gas below some nominally acceptable value. This would also unload a sour water system operating near its capacity limit. Unfortunately, the wash water must only contain less than 1 ppmw HCN to limit the treated gas HCN level to ~4 ppmv.
As FIG. 5 shows, the presence of even a trace of HCN in the wash water cripples the water wash column. This is also why trying to mitigate corrosion by water washing and using the feed and bleed approach to HCN removal from refinery gases almost invariably fails. In this particular application of coal tar gas treatment, any amount of HCN in the wash water would throw the wash column into a pinched state where the washed gas contaminant level is dictated by the cleanliness of the wash water and the amount, size and type of packing, rather than by mass transfer.
H2S removal from the coal tar and COLO gases. The treatment of washed coal tar gas and COLO gas uses conventional amine treating. Both gas streams are contacted with 10 wt% DEA in separate packed absorbers. As FIG. 6 shows, both contactors are only lightly loaded—indeed, the COLO contactor is so lightly loaded, the ability of this column even to perform properly can be questioned. However, both columns are so severely lean-end pinched that the coal tar and COLO contactors are effectively using only 2/3 and 1/2 of their respective packed heights. This makes them less susceptible to fouling and, perhaps to a lesser extent, maldistribution. Due to its lower pressure, the coal tar column treats the gas to ~30 ppmv H2S while the higher-pressure COLO column treats to ~6 ppmv.
Using only 10 wt% DEA allows a higher solvent flowrate, which keeps the packing in the COLO column wetted at least to some extent (albeit only 2.5% flood). The combination of solvent flow and amine strength is adequate to produce quite low H2S gas, but the rich solvent has a rather high H2S loading from the COLO column (0.73 mole/mole) and in the combined stripper feed (0.63 mole/mole). However, there is only a very small amount of CO2 in the stripper feed (0.021 mole/mole), so the high H2S level is likely enough to keep mild steel passivated.
Takeaways. The most important conclusion is that water is a highly selective solvent for HCN (and ammonia). However, to achieve the HCN levels needed to obviate corrosion by the HCN itself—and by the HSSs it produces—very clean water must be used for gas washing. Even extremely small amounts of HCN in the wash water will likely thwart this approach to corrosion mitigation.
Attempting to prevent overloading the sour water system by using a bleed-and-feed scheme has been the common approach taken to HCN removal by water washing in refineries; however, this has almost invariably met with failure. Almost any amount of recycled water contains too much of the very contaminant targeted for removal, severely limiting the extent of removal possible. However, if a clean water source is available and the sour water system can handle the additional load, water washing even in refinery applications can be a viable strategy.
Impactful levels of fouling make it extremely difficult to obtain quantitatively reliable answers via simulation. Under such circumstances, the best outcome is to obtain a qualitative appreciation for how seriously fouling can affect a column’s performance. Conversely, calculating the detailed temperature and composition profiles in columns helps to develop better understanding of the root causes of problems, as well as which proposal has the best chance of success. HP
NOTES
ProTreat® v. 8.1
LITERATURE CITED
Weiland, R. H., C. E. Jones and E. Van Hoorn, “Hydrogen cyanide in amine systems, Hydrocarbon Processing, December 2024.
Egbert W. van Hoorn is an independent consultant. He obtained his academic degree in chemical engineering from the University of Twente. Ir. van Hoorn worked 10 yr for Shell and 4 yr for Huntsman before working as a consultant from 1999 for Hocon.
Ralph Weiland is a graduate of the University of Toronto with a PhD in chemical engineering. Dr. Weiland formed Optimized Gas Treating in 1992 and serves as its Chairman. OGT developed the ProTreat® mass transfer rate-base gas treating simulator and the sulfur plant simulator SulphurPro®.