The subject of alkaline sour water (ammonium bisulfide) corrosion in petroleum refineries has been addressed in the literature over the past 25 years.¹ However, despite the prevalence of sour water corrosion problems in refining operations, very little actual ammonium bisulfide corrosion data are published in the open literature. Surprisingly, most of these studies do not take into account the effect of flow conditions, which appears to be a critical variable. Those that have addressed this issue have not investigated a wide range of exposure and flow conditions. The approach used in many studies has been to focus on empirical findings heavily relying on evaluations of operational experience. These studies have used experience surveys with partially complete data to define an industry consensus on apparent plant operational limits. However, the variables of flow and the sour water environment have not been systematically approached to provide definable limits of serviceability for typical materials of construction.

Environmental regulations aimed at reducing pollution associated with emissions resulting from combustion of hydrocarbon fuels are requiring lower sulfur levels in refined products. This makes it necessary to either expand capacity with new hydroprocessing units (hydrocrackers, hydrotreaters, etc.) or increase throughput or treating severity of existing refinery processing units that remove and handle common impurities such as sulfur. Consequently, there is a critical need for more precise and quantitative data on ammonium bisulfide corrosion for a variety of materials under a range of refinery service conditions. These data can effectively serve as a technical basis for improved prediction of ammonium bisulfide corrosion for use in materials selection, control of process unit operation, and assessment of chemical treatments.

This paper reviews available information and the state-of-the-art in corrosivity prediction in alkaline sour water systems for assessment of plant operations, process control and materials selection. It also reveals some of the limitations that currently exist, and experimental and analytical methodologies that are currently being used to develop a quantitative understanding of aqueous ammonium bisulfide corrosion. The emphasis in this paper is particularly on reactor effluent systems, which are a particular concern due to their high levels of H₂S and ammonia, high pressure and high temperature operation, and high consequences of a failure including fire and explosion. Furthermore, there is a potential major economic liability of sour water corrosion and failures in these units in excess of $50 million per incident that more than justifies this effort.

As indicated previously, the subject of ammonium bisulfide corrosion has been addressed in the literature. In some cases the findings are experiential in nature, resulting from surveys of hydroprocessing unit operators. In other cases the findings are based on laboratory testing, but these have not adequately addressed the effect of flow. The three most influential papers are those by Piehl², Damin and McCoy³, and Scherrer, et.al.⁴ These are summarized below.

Perhaps the most notable article in the literature is Piehl’s paper² describing a survey conducted by the NACE T-8 committee covering corrosion in the reactor effluent air coolers and associated piping in 42 hydrocrackers and hydrotreaters. Analysis of the survey results established that sour water corrosion was mild to negligible when the concentration of ammonium bisulfide was 2 percent or less and the velocity was 20 ft/sec or less. This experience has been generally utilized by the industry for over 20 years for control of sour water corrosion in hydroprocessing unit reactor effluent air coolers, and has served the industry well when followed. Although an actual corrosion threshold was not identified by the survey results, it was noted that corrosion rates for steel may be severe above 3 to 4 percent ammonium bisulfide concentration in the sour water. Also emphasized was the use of the K_p factor (mol% NH₃ times the mol% H₂S in the reactor effluent) in assessment of potential system corrosivity with K_p values above 0.7 being particularly aggressive.

Damin and McCoy³ reported results of laboratory corrosion tests conducted in a stirred autoclave over the ammonium bisulfide concentration range of 10 to 45 percent. They measured low corrosion rates for carbon steel and type 316 stainless steel up to about 35 percent ammonium bisulfide concentration (See Figure 1). Above that, the corrosion rate of both materials increased rapidly to extremely high rates of attack. Their test results appear to demonstrate the presence of a threshold ammonium bisulfide concentration at which the corrosion resistance of the materials changes dramatically. They postulate that this is the result of the formation of a metal ammonium complex that could act to strip the normally protective iron sulfide film from the metal surface. It should be noted that the 35 percent concentration threshold was observed at near-stagnant conditions (actual velocity in the stirred autoclave was estimated to be 1 to 2 ft/sec).

Scherrer, et.al.⁴ presents the only major laboratory corrosion study⁴ documenting the effect of velocity on ammonium bisulfide corrosion (See Table 1). In their tests, conducted with 4.0 [Russ: Table 1 says 4.0 – one of these is not correct] to 10 percent ammonium bisulfide, the corrosion rate of carbon steel was shown to increase by 40 to 64 percent, respectively, when increasing the velocity from 3.5 to 6.5 m/s, the highest velocity tested. The data also show that the worst case conditions were those with combined high ammonium bisulfide concentration and high flow velocity. Based on these limited results, it appears that both bisulfide concentration and velocity are major parameters that work together to produce a range of corrosion responses in engineering materials. However, flow was only qualitatively handled in this study since no hydrodynamic evaluation was performed and the results were not presented in terms of flow-induced mechanical parameters such as wall shear stress.

The abovementioned studies were conducted in the time frame of 1976 to 1980. Since then, guidelines for sour water service, and the hydroprocessing unit REAC system in particular, have been heavily influenced by this work but have not really progressed much further in recent years. The amount of new information developed was limited, but service experience with sour water systems was growing and in some cases documented.

A paper by Shargay, Coombs, Bagdasarian and Jenkins was written in response to a need for a short course on refinery corrosion prepared by NACE International.⁵ Within the course material, these individuals attempted to coalesce the existing guidelines in sour water corrosion relative to hydroprocessing units and the REAC system in particular. This publication provides a good representation of the general understanding of the important factors in sour water corrosion in the REAC system:

• Corrosion results from the presence of ammonium bisulfide and corrosion rate generally increases with its concentration.
• The Piehl Kp factor (mole fraction ammonia times the mole fraction hydrogen sulfide in the reactor effluent) may be used to monitor the potential corrosive severity of process environments and potentially in materials selection.
• Sour water corrosion involves flow velocity as a primary parameter with excessive corrosion generally experienced at flow rates greater than 6 m/s. Stainless steels may provide acceptable corrosion rate up to 9 m/s as defined primarily on service experience.
• Maintaining water injection rate to result in at least 25 percent unvaporized (liquid) water in the effluent stream also appears important.
• The injected water must be free of dissolved oxygen (<50 ppb) to prevent rapid corrosion.
• Good flow distribution (in terms of vapor, liquid hydrocarbon and water phase) is important. Therefore, tube velocities of less than 3 m/s in the air cooler can result in problems due to separation (stratification) of phases and the formation of corrosion deposits on tubes.
• Deposition of solid ammonium bisulfide salt may occur due to flow maldistribution which starves some tubes of wash water, and exacerbates both fouling and corrosion problems.
• U-bend tube designs may cause problems in the REAC system due to the possibility of velocity accelerated corrosion in tubing bends.

A more recent survey was conducted by Harvey and Singh⁶ of UOP that included many licensed hydroprocessing units with over 700 operating years of total experience. It was described that many hydrocracking and high severity hydrotreating units have experienced severe local corrosion attack of the REAC and associated piping. In all cases, the common factors have been processing of high sulfur and high nitrogen feedstocks which increased sour water corrosivity and created further demands on the REAC system.

This study also evaluated system parameters such as the Piehl K_p factor, water wash rate, sour water ammonium bisulfide concentration, and estimated peak tube velocity in the REAC. It also examined the influence of balanced and unbalanced header conditions on the survey results. In general the results of this study supported the 2 percent ammonium bisulfide concentration limit presented in the Piehl paper² for the onset of significant corrosion problems and also identified 0.2 as a K_p threshold. But, more importantly, this study revealed that flow conditions in the header (e.g. balanced and non-balanced) were an important factor affecting the severity of corrosion. For example, whereas a limiting value of 2 percent ammonium bisulfide was shown for the onset of corrosive conditions in REAC systems, for systems with balanced inlet conditions, this concentration increased to 8 percent. Additionally, a velocity limit for balanced systems was 6.7 m/s whereas for unbalanced systems, the velocity limit could be as low as 3 m/s. Therefore, the limiting operating conditions of 6 m/s and 8 percent ammonium bisulfide was only applicable for REAC systems which had a completely balanced header.

It is apparent that there is currently insufficient corrosion data (particularly that which involve rigorous treatment of flow effects) to fully understand the corrosivity of dynamic ammonium bisulfide environments over a wide range of service conditions from which corrosion in REAC systems can be assessed and/or predicted. A recent API survey on corrosion in hydroprocessing unit REAC systems also indicated similar findings. The effect of other parameters such as pH, temperature, partial pressures of H₂S and NH₃, solution contaminants such as chlorides and the presence of oil/water mixtures and inhibitors have not been well quantified. Additionally, compared to carbon steel, there is even less corrosion data available for many alloys commonly used in this service. For example, no data were found for alloy 2205, which in recent years has been used in the higher ammonium bisulfide concentration REAC systems. In the past, experience surveys have been conducted. However, these studies have restricted applicability since they typically lack quantitative information required to apply the data on a broader scale across refinery units. The studies conducted to date have identified the potential extent of corrosion problems in process units handling alkaline sour water and the critical needs for improving predictive capabilities as well as system efficiency and reliability.

Some operating companies and process licensors have developed their own procedures for controlling ammonium bisulfide corrosion of carbon steel based on operating experience. In many instances, these permit concentrations of ammonium bisulfide exceeding the 2 percent recommended by Piehl, perhaps up to the 8 or 10 percent range, while maintaining the 6 m/s maximum velocity criteria. However, corrosion problems have occurred in some units operating in this concentration range.

There are some hydroprocessing units with carbon steel effluent systems that have actually operated for periods of time with ammonium bisulfide concentration in the 15 to 20 percent range but with limited documented experience. It is also increasingly common for hydroprocessing units, designed to handle a given ammonium bisulfide concentration, to be exposed to higher concentrations when the nitrogen level in the feed is increased without a corresponding increase in the injection rate of wash water. This usually results in increased corrosion and, in some cases, unit reliability problems and unscheduled shutdowns. Over the last 5 to 10 years, there have been several major incidents where ammonium bisulfide corrosion caused loss of containment in hydroprocessing units that resulted in damage/lost production on the order of $50 million. There have also been failures in the overhead systems of some sour water stripper columns that resulted in significant reliability impacts. Some of these involved rapid corrosion of alloys such as type 316 SS and alloy 800 that were previously thought to be resistant in this service. Currently no reliable methodologies or assessment techniques exist which allow prediction of the corrosive severity of these types of operational scenarios.

Understanding of Flow and Multiphase Conditions

The discussion of existing experience presented herein shows that sour water corrosion involves a velocity sensitive mechanism that can be influenced by a combination of factors, of which ammonium bisulfide concentration and velocity are key parameters. However, none of the studies published to date provide quantitative relationships based on both relevant chemical and mechanical factors as produced by the flowing service environments present in reactor effluent systems. Most commonly these environments are multiphase in nature involving the presence of sour water, liquid hydrocarbon and gas phases under complex flow regimes.

There are many factors that need to be considered when conducting corrosion assessments in multiphase environments that can be systematically evaluated using basic flow modeling techniques and parameters shown in Table 2. These include important factors related to the dynamic (or flowing) nature of the phases present that determine the dominant flow regime and the kinetic shear forces that are imparted by the flowing fluids on the pipe wall (wall shear stress). There have been major studies involving very sophisticated simulations of three phase flow outside the specific scope of sour water corrosion which are, nevertheless, still very applicable. These studies have been facilities intensive and costly because major investments were made to prepare, pump and dispose of corrosive materials required for such tests. However, these studies have helped to show the way in which flow rated corrosion problems involving multiphase (oil/water/gas) environments under service conditions can be systematically analyzed. Furthermore, these studies have also shown the impact of changes in flow regime and/or flow path anomalies on the level of mechanical action of the exposed surfaces of equipment. As shown in Table 3, one investigation involving hydrodynamic modeling revealed that such changes can result in a magnification of wall shear stress of over 100 times the nominal levels for smooth bore single phase flow.⁷ This illustrates why methodologies involving simply the linear flow velocity as a measure of mechanical action cannot provide rigorous solutions. It is necessary for flow modeling of systems to identify both flow regime and the likely maximum levels of wall shear stress that the materials of construction will have to withstand. It also provides a basis for understanding and representing the impact of process conditions (e.g., related to balanced and unbalanced header operation) on corrosive severity in REAC systems.

InterCorr International, Inc., in association with Equilon Enterprises, LLC has developed the experimental and analytical capabilities to assess concerns resulting from ammonium bisulfide corrosion in flowing systems. These capabilities involve analysis of the multiphase flow conditions that can be obtained, which are then used as a basis for experimental simulation using three-phase emulation techniques. These techniques provide for the establishment of multiphase conditions (water/gas or water/oil/gas) in a reservoir autoclave of a laboratory flow loop using rigorous ionic modeling of test environments. To simulate the alkaline sour water conditions in the REAC system, the primary corrodent (H₂S) is dissolved in an ammonium hydroxide solution. Facilities for totally anaerobic handling and addition of the constituents to the autoclave reservoir are involved so that a correct chemical simulation can be obtained in the flow loop.

The mechanical simulation of the effects of flowing media is based on hydrodynamic analysis of the flow regime and wall shear stress produced in service by the flowing multiphase media. These have been discussed in greater detail in a previous paper.⁸ Once the wall shear stress for actual conditions is determined, the flow loop can produce equivalent levels of wall shear stress by flowing a sour water solution (or sour water/oil mixture) in the loop. The main assumption utilized in this technique is that the major contribution to the wall shear stress is made by the liquid phase. This approximation technique is valid because the contribution of liquid phase density and viscosity to the resultant shear stress under multiphase flow conditions predominates over that of the lower density gas phase.

Of significance in many flowing multiphase systems is the handling of flow regimes that can magnify the wall shear stress over nominal levels. For example, the main attribute of slug flow is the dynamic conditions of high turbulence in the region of the flow at the leading edge of the moving slug (See Figure 2). Likewise, an additional concern involving corrosion in multiphase flow conditions is droplet impingement. Under conditions of high velocity gas flow, entrained liquid droplets are accelerated to near the superficial gas velocity which substantially increases their momentum and the resultant wall shear stress that they produce when they impact metal surfaces. As in the case of slug flow, the problem can be approached more simply and cost effectively with flow modeling techniques combined with the abovementioned laboratory three phase flow emulation procedures. An important aspect to keep in mind is that the flow velocity required in the loop to produce the high levels of wall shear stress associated with these conditions will not usually be the same as the nominal velocity in the actual process unit.

InterCorr International, in collaboration with Equilon Enterprises, LLC, has initiated a joint industry sponsored program utilizing the analytical and experimental techniques mentioned above. This program has been specifically designed to generate useful engineering data and expand the understanding of the ammonium bisulfide corrosion process. This information will be used as a basis for the development of a more accurate and comprehensive predictive tool including assessment methodologies for control of ammonium bisulfide corrosion of a wide range of materials of construction to help attain safe and reliable operation of process units handling ammonium bisulfide.

The program involves the tasks that focus initially on corrosion in H₂S-dominated sour water systems most applicable to hydroprocessing unit reactor effluent air cooler (REAC) systems. It is also investigating parametric effects that may affect system corrosivity including H₂S partial pressure, chloride concentration and operating temperature. Additional areas of emphasis include the influence of oil/water mixtures (e.g., the effect of water carryover), and performance limits of inhibitors (e.g., ammonium polysulfide and filming amine) in sour water systems. The outcome of the program will include a series of engineering curves involving corrosion rate versus velocity, wall shear stress and ammonium bisulfide concentration along with a series of iso-corrosion curves for various materials versus these same variables. The program will also provide for the development of a software corrosion prediction tool (Predict-SW).

This program focuses on developing corrosion data in ammonium bisulfide environments using a laboratory flow loop run under simulated service conditions focusing on the effects of flow (velocity) on corrosion and the performance of commonly used alloys. This approach has already been shown to have substantial success in terms of being able to simulate ammonium bisulfide environments common to sour water effluent streams. In fact, the results from initial tests are already complete and have been utilized to evaluate and improve plant operations. These data were made available to the program sponsors immediately upon startup of the program. The balance of the data will be made available in reports and in the software provided to the sponsors over the two year period of the program.

InterCorr will incorporate the data from the experimental effort into a new software program. This will be similar in nature and functionality to its currently available Predict^® program used for upstream oil and gas production environments. It provides a user-friendly Windows-based program that uses inputs of typical operating parameters to assess corrosion rates of steels in multiphase oilfield production environments containing H₂S and CO₂. The new software tool originating from this program (Predict-SW) will incorporate the program data and any other appropriate data and experience on ammonium bisulfide corrosion available in the published literature and submitted by sponsor companies on a voluntary basis.

Based on the information presented herein, the following conclusions were made:

1. Ammonium bisulfide (alkaline sour water) corrosion is a significant problem in hydroprocessing unit reactor effluent air cooler (REAC) systems and other refinery systems that has a particularly large risk related to failure, including lost production time and safety/environmental issues associated with refinery incidents.
2. Most of the available information on sour water corrosion is qualitative and mainly experiential in nature and has not adequately defined limits of operation for typical materials of construction.
3. Sour water corrosion involves a velocity-sensitive mechanism that must be quantified in terms of both mechanical and chemical factors to provide for accurate assessment.
4. Operational limits for various alloys can be developed based on laboratory flow loop tests conducted under closely controlled environmental simulation and hydrodynamic conditions. This approach has already been shown to have substantial success in terms of being able to simulate ammonium bisulfide environments common to REAC systems.

1. ASM Handbook, Volume 13, Corrosion, ASM International.
2. R.L. Piehl, "Survey of Corrosion in Hydrocracker Effluent Air Coolers", Materials Performance, Vol 15 (1), January 1976, pp 15-20.
3. D.G. Damin and J. D. McCoy, "Prevention of Corrosion in Hydrodesulfurizer Air Coolers and Condensers", Materials Performance, Vol 17 (12), December 1978, pp 23-26 (see also NACE Corrosion/78, paper # 131).
4. C. Scherrer, M. Durrieu, and G. Jarno, "Distillate and Resid Hydroprocessing: Coping with High Concentrations of Ammonium Bisulfide in the Process Water", Materials Performance, Vol 19 (11), November 1980, pp 25-31 (see also NACE Corrosion/79, paper # 27).
5. C.A. Shargay, A.J. Bagdasarian, J.W. Coombs and W.K. Jenkins, "Corrosion in Hydroprocessing Units", in Corrosion in the Oil Refining Industry, Short Course presented by NACE International, Houston, Texas, Sept. 1996, pp 10/1-18.
6. C. Harvey and A. Singh, "Mitigate Failures for Reactor Effluent Air Coolers", Hydrocarbon Processing, October 1999, pp 59-72.
7. B.V. Johnson, H.J. Choi, and A.S. Green, "Effects of Liquid Wall Shear Stress on CO₂ Corrosion of X-52 C-Steel in Simulated Oilfield Production Environments", Corrosion/91, Paper #573, NACE International, Houston, Texas, March 1991.
8. S. Srinivasan and R.D. Kane, "Experimental Simulation of Multiphase CO₂/H₂S Systems", Corrosion/99, Paper # 90014, NACE International, Inc., Houston, Texas, April 1999.

Figure 1 – Effect of ammonium bisulfide concentration on corrosion at 93 C.³

Figure 2 – Common regimes of multiphase flow. Note region
of high turbulence in front of advancing slug.