Knowledge Center: Asset Optimization
Online Corrosion Monitoring
The subject of amine unit problems in petroleum refineries and petrochemical applications has been extensively addressed in the literature over the past 40 years. Despite this emphasis, amine units still are subject to design issues (erosion), process problems (build-up of contaminates, over-stripping of H2S and foaming), reliability issues and requirements for standby capacity. Unexpected corrosion leading to operational boundary uncertainty, unit outage, and lost capacity are still widely encountered. Regardless of the prevalence of amine corrosion problems there are still only limited quantitative corrosion data that can be directly utilized in developing better guidelines for corrosion control and more efficient unit operation.
One major limitation in amine unit operations is that historically the effects of velocity and local turbulence are not taken into account in a quantified manner. Even though, in many cases flow appears to be a critical variable, it has not been investigated over a wide range of unit operating conditions. Part of the problem is that most amine unit specialists have focused on empirical findings heavily relying on evaluations of operational experience. The shortcoming of this approach has been the lack of transportability of experience gain from one plant to seemingly similar units. There is also currently a need for more precise, quantifiable and timelier corrosion data that is directly linked to process conditions that can be used to define better operational boundaries for amine units and rapid mitigation of corrosion issues. With this type of information, it would be possible to achieve improved amine unit performance and system reliability. Most importantly, it allows for the assessment of the need for remedial action sooner and more accurately than with existing approaches.
Most of the variables that influence corrosion rates in other process systems are important in Amine systems as well. These variables include temperature, velocity and concentration of corrosive species. There are different commercial amines used in refining and gas treating systems: MEA, DGA, DEA and MDEA. Each amine solution can promote different corrosion reactions and consequently the extent of deterioration of the material. The areas of highest corrosion potential are the reboiler, hot lean amine piping, lean/rich exchanger, hot rich amine piping, stripper and stripper overhead. The main corrosion mechanism in amine units are:
• General and localized corrosion : depending on CO2/H2S concentration, Amine used, Temperature and pressures
• Hydrogen Induced Cracking, Stress Corrosion Cracking in H2S environments
• Hydrogen embrittlement due to H2S environments.
The combination of processing and operating conditions, materials of manufacture and previous failure become a potential location for corrosion problems.
This paper reviews the state-of-the-art in the understanding of amine unit corrosion and identification of outstanding issues. It also discusses two new tools in the evaluation of amine unit corrosion. These include:
•The use of online, real-time corrosion monitoring to assess operating conditions to rapidly identify and mitigate corrosion causing process conditions.
• The use of new corrosion modeling and prediction tools for quick assessment of corrosion risk versus changing operating conditions, and for use in assessment of inspection intervals and materials selection.
Experience with Corrosion in Amine Units Going back to the late 1950’s, an experience survey conducted by the American Petroleum Institute showed that corrosion by amines in gas treatment and sulfur recovery was a major concern in the refining and petrochemical industry. Major causes of amine corrosion identified included:
• Poor plant design and operating practices, and
• Solution contamination by impurities and heat stable salts.
In general terms, it was identified that amine systems used for removal of only CO2 were generally more corrosive than those involving only H2S, and MEA systems were generally more corrosive than DEA. More fundamental treatments of amine corrosion problems have identified that corrosion is usually most severe where the gases are absorbed or deabsorbed from rich amine solutions where temperature, flow and turbulence are important considerations.
The most common material of construction for amine units is carbon steel. Amine stress corrosion cracking (the combined effect of corrosion and stress) is also a consideration that has been extensively discussed particularly in lean amine systems. However, experience has proven that more failures in amine units have actually resulted from corrosion leading to severe consequences.
Process boundaries (for corrosion control) in amine systems including the following:
• Amine concentration (around 20% for MEA)
• Acid gas loading (0.3 to 0.6 mole/mole depending on the amine solvent utilized)
• Flow rate (rich amine – 6 fps; lean amine – 6 to 20 fps)
• Rich circuit temperature limits in the range 210-220 F)
• Lean amine reboiler temperature limits in the range (260 – 300 F)
• Limits for amine unit impurities and heat stable salts are typically based entirely on “rules of thumb” and may range from 1 to 2 percent.
These boundaries have been developed by evaluating service experiences in various plants with sometimes vastly different designs, throughput and operating conditions. Consequently, these boundaries vary unit-to-unit with little technical basis to help designers and operators optimize unit reliability and performance. There has been only limited use of predictions based on test data or plant monitoring data.
In the past five years or so, there have been a few studies on corrosion in the lean amine circuit . The findings indicate that lean H2S loading, impurities, polymer formation, temperature and flow rate can all influence corrosion in lean amine system. However, despite this recent emphasis, this work has not been able to fully define the corrosiveness of lean amine systems, but have shown directions that, if employed more fully, could improve amine unit throughput, reliability, and reduce unplanned outages and thereby reduce the need to standby capacity.
One effort showed the dynamic nature of corrosion in a lean amine circuit using online, real-time monitoring to identify specific operating conditions to periods of high corrosion rate . Despite the believe by many engineers that lean amine H2S loading is the major culprit, the use of online plant monitoring of corrosion showed that the root cause can sometimes be otherwise. Other process issues can predominate in lean amine systems as well, including polymer formation, and conditions that result in locally high wall shear stress (the mechanical forces on the metal surface that can remove protective corrosion scales). These include conditions of high throughput, turbulent flow around elbows, tees and weld protrusions, and flashing on heat transfer surfaces. This later situation leads to both high wall shear stress and increase in chemical aggressivity leading to severe corrosion in this portion of the lean amine circuit. Other studies have also related plant corrosivity with the formation of organic acids. But, it has been found that predicting corrosion rate based on acid concentration alone can yield misleading results when going from unit-to-unit. Hence, it appears that for accurate prediction of corrosion rate in lean amine systems, multiple variables must be considered, but only limited data are readily available.
Real-Time Monitoring in Amine Systems
In today’s market where losing 50 MMSCF of natural gas production can cost upwards of $400,000 per day of downtime, it is critical for these units to be online as much as possible. A rigorous corrosion monitoring and inspection program that includes predictive analysis and real-time monitoring can significantly contribute to maximizing uptime by minimizing failures and reducing inspection frequencies. Significant savings can also be realized through optimizing utility consumption within the unit. On-line corrosion monitoring can quickly help define the optimal lean acid gas loading windows which can prevent wasting steam by regenerating an amine solution further than required. An average sized (100 MMSCF) amine plant can spend up to $48,000 per day to generate the steam required to run the process. A two percent reduction in the steam consumption enabled by on-line corrosion monitoring would result in a $350,000 per year savings.
Honeywell recently introduced a new online, real-time corrosion monitoring transmitter that utilizes the existing plant distributed control system (DCS). The DCS serves as the primary data path for critical process data and implementation of advanced process control functions. Therefore, operators can effectively maximize productivity plant-wide and quickly responding to system changes and upsets on specific units. With application of Honeywell’s new SmartCET® corrosion transmitter (CET 5000), corrosion data can be easily acquired automatically, viewed in combination with process data and readily integrated into the global process optimization process. This integration is an important aspect since most corrosion data is still being obtained offline manually using corrosion coupons. The problem is that this manual process requires substantial time and effort to view leading to unexpected corrosion failures and unit outages.
In addition to directly impacting costs and profitability in the plant, predictive assessment and on-line corrosion monitoring can play a vital role in management of change processes within the facility. Changes in process conditions, gas conditions or chemical products can significantly impact corrosion rates within a facility. Utilizing SmartCET corrosion monitoring tools, plant management has the flexibility to quickly see and quantify the effects of any changes on corrosion and make clever decisions based upon real time data.
Honeywell field-proven corrosion solution offers measurable improvements and benefits:
1) Increase plant utilization due to improved reliability of assets
2) Reduce maintenance costs by moving from preventive to reliability-based maintenance
3) Maximize production throughput while protecting plant assets
4) Improve safety environmental compliance by minimizing the effects of process upsets and excursions
5) Optimize inhibitor costs
An example of the use and benefits of online corrosion monitoring in an amine unit is presented herein. This amine unit operated for approximately eight years without corrosion issues and then unexpectedly started to experience severe corrosion and integrity failures. A preliminary investigation recommended installing online corrosion monitoring equipment to relate corrosion rates to process changes. The corrosion monitoring system selected was the Honeywell online, real-time SmartCET system (CET5000). As is now being observed in more plant situations, the outcome of this investigation showed that corrosion was not a continuous process, but was related to periods of process instability that could only be efficiently address with real-time monitoring.
The amine unit in question was a mixed solvent system in Western Canada receiving a combined gas containing approximately 21 mol% H2S and 5 mol% CO2 with the target acid gas loading of 0.8 mol acid gas/mol amine. In this case, the pressure of the rich amine was decreased prior to entry into the flash drum. Any co-absorbed hydrocarbon flashes off the rich solution and is sweetened with lean amine in the flash gas contactor. The remaining rich amine was heated by the hot lean solution from the bottom of the regenerator prior to entry into the regenerator. The H2S, CO2, COS and mercaptan were stripped from the solution with steam which was generated by reboiling the liquid at the bottom of the column in two kettle type reboilers using 500 kPag steam.
The amine system had a good service history and solution monitoring was an integral part of the operation of the amine systems. Routine weekly and monthly testing of the solutions was on-going since plant start up. Ancillary tests included analysis for amine components, residue components, organic acids, metals and lean loading. The amine degradation products are generally referred to as residue and are mostly comprised of amine polymers. The polymers are believed to increase corrosion due to their potential chelating strength and ability to breakdown protective corrosion films.
Problem The service history and regular solution monitoring, indicated that there was basically minimal corrosion in the amine system for 8 years and then, the rate of corrosion attack increased substantially. At the time, inspection of the reboiler found extensive damage on the bottom of the shell at the weir plate with a general corrosion rate of 7 mm/yr. The hot liquid return lines had multiple indications of general corrosion concentrated primarily at the elbows as shown in Figure 1. Although there were increases in known corrosive species such as organic acids and residues, the results of lab tests indicated that none resulted in a major increase in the corrosion rate that was experienced in the actual unit. A possible contributing factor to the corrosion was postulated to be changes in process variables and hence, it was felt that online corrosion monitoring could facilitate the correlation of corrosion and variations in unit operating conditions.
Conventional corrosion coupon monitoring was not a viable option. Corrosion coupons exposed for periods of months would not identify short periods of process upset resulting in corrosion. Even use of conventional electrochemical techniques to produce corrosion versus time plots would be problematic since corrosion is viewed in isolation from process parameters. Without a direct link to corresponding process information, periods of variable corrosion rate could be observed, but without correlation to process conditions, the data do not provide a way to achieve root cause analysis.
Online, Real-Time Corrosion Monitoring In this case, SmartCET was used to obtain corrosion measurements in the lean amine circuit every seven minutes. This frequency of data sampling was compatible with normal process control measurements. Monitoring was achieved with a three electrode sensor and its unique transmitter that runs three electrochemical techniques remotely and automatically. Pre-processed corrosion signals were taken into the amine unit process control system and displayed in the plant data historian along with selected key process variables. Figure 2 is a typical screen view of corrosion and process information available to plant personnel including operators, maintenance and engineer functions. It shows the way that corrosion and process data trends can be viewed simultaneously. The buttons on the lower left hand corner link to trends of additional SmartCET data, as well as links to view process and corrosion data on an individual graph to fine tune the correlations between process changes and corrosion. Further links are incorporated to amine composition and corrosion data gathered by laboratory testing.
Data Analysis & Problem Identification Following commissioning of the online corrosion monitor, a drastic decrease in corrosion rate was observed. Review of the corresponding process data showed that there had been some adjustments made to the unit operating conditions. However, a review of the solution chemistry showed that most of the aforementioned corrosive species remained at near constant levels except the residue level had decreased. However, the process instability was identified as large variations in the differential pressure during the period just prior to the internal inspection of the system and when the corrosion rate decreased. The residue level increased to beyond 10 wt% at the onset of the corrosion episode and then decreased past 9 wt% when the corrosion rate decreased.
Figure 3 shows the real-time corrosion rate obtained during an upset in processconditions. This upset was characterized by sharp changes in the differential pressure across the regenerator as well as changes in steam rates to the reboilers, and amine to the regenerator as operations reacts to the upset. The change in corrosion rate was dramatic going from around only 0.15 mm/yr to over 1.25 mm/yr at the onset of the upset and remains high even after the regenerator appears to regain some stability. The corrosion rate did not decrease fully until the steam rate to the reboilers and the temperature at the top of the regenerator stopped fluctuating. The short duration of the high corrosion rate suggests that the instability causes a short period in which corrosive species exist and also shows that there must be certain other process conditions that allow the return of a protective iron sulfide scale.
The data from SmartCET was found to correlate with the extent of corrosion damage found during periodic non-destructive testing (NDT) in the hot lean amine system. But, the online corrosion data allowed for the identification of the specific process conditions that produced the corrosion rather than only the accumulated corrosion damage commonly found my inspection or use of corrosion coupons.
Figure 3 – Correlation between real-time corrosion rate (black line) and amine reboiler/regenerator process variables (red lines as indicated above: differential pressure across the regenerator, steam rate to reboilers, top temperature in the regenerator, and rich amine to the regenerator)
New Corrosion Prediction Tool for Rich Amine Circuits
Honeywell is also involved in the development of corrosion prediction software tools for assessment of refinery systems. It has just completed a multi-year joint industry program (JIP) sponsored by a group of leading refiners and service companies. The scope of this effort was to extend corrosion forecasting capabilities for use in risk. Corrosion rate assessment, definition of inspection intervals, alloy selection, process optimization, and asset management. The program was directed specifically at providing a better technical basis for predicting corrosion in rich amine system over a range of operating conditions for commonly used alloys. The culmination of this effort was the development of the Predict ®-Amine corrosion prediction software to facilitate rapid and consistent corrosion prediction by participating companies. A second phase of work is now about to start that will extend this methodology to lean amine circuits.
The approach used in the Amine JIP was to develop corrosion rate data under simulated service conditions using a sophisticated laboratory flow loop that simulated temperature, pressure, acid gas loading, CO2/H2S ratio, impurity concentration and flow conditions. This engineering data developed concentrated on three amine solvents that included: (a) 18% MEA, (b) 30% DEA, and (c) 45% DGA, with some limited testing of 45% MDEA. Alloys included in the study were carbon steel (base metal and weld heat affected zone structure), three stainless steels (AISI 304L, AISI 316L, Alloy 2205), and Alloy 825 (used to simulate the composition of a single pass Alloy 625 overlay).
The engineering database was used to examine the parametric influences of the primary process variables to address specific questions related to amine corrosion. This project specifically involved the development of corrosion rate data for carbon steel and commonly used alloys for a range of environments and velocities related to treatment of H2S-containing refinery streams where H2S loading and velocity are the principal variables for regularly used amine solvents and concentrations. The proposed work included parametric studies of the effects of temperature, impurities, and CO2 loading. This approach has already been shown to have substantial success in terms of being able to simulate corrosion in a previous development effort on refinery sour water (ammonium bisulfide) corrosion.
New Corrosion Assessment Methodology
As mentioned previously, most generally available industry guidelines for assessing amine corrosion have been empirical in nature and do not handle process variables in a quantitative manner. This JIP effort involved careful simulation and controlled laboratory experiments. Ionic modeling was used to determine the appropriate amounts of constituents required to achieve the desired loadings in equilibrium with the desired amine solvent at the test temperatures. This capability was critical to the design of suitable experiments to model actual process unit conditions. This type of process tool also provides a very important function in the use of the program results. Similar programs, such as Honeywell’s UniSim, are commonly used by process engineers to define specific process conditions in refinery processes.
Another important aspect of this JIP was that the characterization of flow was in terms of a scalable flow parameter – wall shear stress – and did NOT simply use a linear flow velocity. The wall shear stress is a measure of the mechanical force exerted on the internal surface of the material induced by the flowing medium that can assist in removal of normally protective corrosion scale. The determination of wall shear stress for specific flow conditions involves several factors that include the density and viscosity of the flow medium, pipe ID, surface roughness, flow regime, and the associated pressure drop. Figure 4 shows wall shear stress values for selected pipe sizes with varying linear flow rates. For proper flow assessment, this parameter also takes into account geometrical factors as related to changes in the flow path, such as weld protrusions, elbows, tees and bends. These geometrical aspects are dealt with as wall shear stress magnification factors that increase the magnitude of the wall shear stress at these locations.
One of the major accomplishments of the program was the development of a standardized methodology for representing corrosion data in rich amine systems. This methodology consisted of the development of isocorrosion diagrams for each solvent (MEA, DEA and DGA) and material (carbon steel, 304L, 316L and alloy 2205 stainless steels and alloy 825). The basic isocorrosion diagrams utilized the values of the measured corrosion rates plotted using the flow velocity (wall shear stress) on the vertical axis and the mole/mole H2S gas loading for the horizontal axis. These isocorrosion diagrams were developed for each solvent type at a nominal temperature of 54 C. The range of H2S loading used varied from a “low-rich” loading level of 0.2 mole H2S /mole amine to the “high-rich” loading level of 0.5 to 0.8 mole H2S/mole amine depending on the solvent type. This served as the baseline data for the program. Wall shear stress levels examined in this program were from 0 Pa (near static) to over 1600 Pa (typical of highly turbulent flow).
Once the baseline iso-corrosion diagrams were developed, parametric tests were conducted to evaluate the influence of selected process parameters on the corrosion rates. These parameters included: Temperature (54 C to 121 C), CO2/H2S ratio (0:1 to 1:1), and amine impurity concentration (up to 3 to 6 wt% depending on solvent using a cocktail of acetate, formate and oxalate). Additionally, sensitivity tests were conducted on the influence of higher amine solvent concentrations for MEA and DEA and the performance of 45% MDEA evaluating selected operating conditions for this amine solvent.
New Predictive Software Tool – Predict-AmineThe results from this program constitute a large body of test data that require proper modeling of flow conditions to be accurately utilized. To facilitate query of the program data, a software tool called Predict-Amine was developed as part of the JIP. It incorporates the program data, industry experience and streamlines the calculation of corrosion rates with incorporation of the appropriate wall shear stress characterization for the process stream. This software tool provides a data screen for input of the amine solvent type along with application and process stream variables. This information is then used to predict the corrosion rate of materials of construction using the following sequence:
• Calculation of the effective wall shear stress from process flow conditions
• Conversion of the field wall shear stress into an equivalent laboratory test flow velocity
• Use of the laboratory test flow velocity and the H2S loading to predict corrosion rates at 54 C for the five materials (carbon steel, 304L, 316L, alloy 2205 and alloy 825) from their respective isocorrosion diagrams (i.e., baseline conditions)
• Correction of the baseline corrosion rates for the effects of process temperature, CO2/H2S ratio and impurity concentration.
The software tool input screen is shown in Figure 5. The top left section of the screen is used for input of the environmental parameters and the top right section is used for application specific parameters, such as pipe size, corrosion allowance, design life and flow geometry descriptors. The bottom portion of the input screen is used to input the process stream properties and flow rates.
Upon calculation, the software tool provides the output screen listing the five materials and their predicted corrosion rates (See Figure 6). The acceptable corrosion rates (based on the input corrosion allowance and design life) are shown with a green box. Non-acceptable corrosion rates appear with a red box. In addition to the corrosion rate results, the results screen displays the flow regime, calculated wall shear stress, superficial liquid and gas velocities and equivalent laboratory flow velocity (for 100% liquid flow). A comments section is also provided to document the user assumptions, scenarios, etc. Input and results can be saved as a consultation file and printed.
The program data and beta software were reviewed by the program participants who obtained a single user version of the final Predict-Amine software. Now that this effort is complete, these participants also qualify for additional single or multi-user licenses at preferred rates. Non-sponsors that agree to the confidentiality restrictions of the JIP and pay the applicable non-sponsor fee can also gain access to the program results and software. A secure website has been maintained to archive all JIP meeting presentations, minutes, discussion topics, deliverables and software downloads.
New JIP Effort on Lean Amine CorrosionWith the completion of the JIP on rich amine corrosion, it was felt that there was a firm technical basis for the database and the corrosion prediction methodology. Therefore, it was decided to extend this methodology to analyze corrosion in lean amine circuits, as well. This new JIP effort will start in mid-2007 with the formation of a new industry group to fund and lead the program on lean amine corrosion prediction. As in the previous case, the program will develop a critical database of corrosion data developed under simulated lean amine process conditions and finish with the development of software code that will extend the currently available Predict-Amine to lean amine conditions.
This work will focus on the developing a data to examine the effects of several parameters on lean amine corrosivity to steel, stainless steels (304L, 316L and alloy 2205) and alloy 825. The concept will be to utilize a similar testing apparatus as used in the rich amine JIP effort, but with the modification that it will also reproduce variable conditions of heat transfer (?T) found in amine unit regenerators and reboilers. In the case of lean amine corrosion, experience also indicates that lean amine (H2S) loading will also be a critical variable controlling corrosion. The loading of acid gases in the lean amine, particularly H2S, is known to influence corrosivity of lean amines to carbon steels and possibly other material. Industry refers to “over-stripping” where reduced H2S levels cause the FeS film to be unstable which results in increased corrosion rates.
In this new program, the baseline conditions will be one H2S level (approx. 100 ppmw), one base solution temperature with three different ?Ts (15, 25 and 50 F tube skin temperature above mean amine temperature) per test simulating reboiler conditions, one level of contaminants (nominal maximum concentration), and three flow velocities. Additionally, parametric data will be developed. Based on discussions with major oil companies, the major factors to be included in this program are (1) H2S loading, (2) contaminant concentration (3) service temperature. In the cases where the residual level of H2S in the lean amine will be controlled, three different H2S levels will be utilized in the range of approximately 50 to 1000 ppmw. Ionic equilibrium calculations will be used to set the initial loadings and as a means to assure reasonably constant acid gas loadings will be used for all tasks. In each case, four amine solvents will be evaluated: MEA, DEA, DGA and MDEA. When combined with the results from the previous richamine JIP, the corrosion prediction methodology will be able to handle a complete range of amine unit conditions.
For more information Please visit Honeywell website at www.honeywell.com/ps/sea
SmartCET is a registered trademark Honeywell International and a product of Honeywell Process Solutions. Predict is a registered trademark Honeywell International and a software product of Honeywell Process Solutions
The subject of amine unit problems in petroleum refineries and petrochemical applications has been extensively addressed in the literature over the past 40 years. Despite this emphasis, amine units still are subject to design issues (erosion), process problems (build-up of contaminates, over-stripping of H2S and foaming), reliability issues and requirements for standby capacity. Unexpected corrosion leading to operational boundary uncertainty, unit outage, and lost capacity are still widely encountered. Regardless of the prevalence of amine corrosion problems there are still only limited quantitative corrosion data that can be directly utilized in developing better guidelines for corrosion control and more efficient unit operation.
One major limitation in amine unit operations is that historically the effects of velocity and local turbulence are not taken into account in a quantified manner. Even though, in many cases flow appears to be a critical variable, it has not been investigated over a wide range of unit operating conditions. Part of the problem is that most amine unit specialists have focused on empirical findings heavily relying on evaluations of operational experience. The shortcoming of this approach has been the lack of transportability of experience gain from one plant to seemingly similar units. There is also currently a need for more precise, quantifiable and timelier corrosion data that is directly linked to process conditions that can be used to define better operational boundaries for amine units and rapid mitigation of corrosion issues. With this type of information, it would be possible to achieve improved amine unit performance and system reliability. Most importantly, it allows for the assessment of the need for remedial action sooner and more accurately than with existing approaches.
Most of the variables that influence corrosion rates in other process systems are important in Amine systems as well. These variables include temperature, velocity and concentration of corrosive species. There are different commercial amines used in refining and gas treating systems: MEA, DGA, DEA and MDEA. Each amine solution can promote different corrosion reactions and consequently the extent of deterioration of the material. The areas of highest corrosion potential are the reboiler, hot lean amine piping, lean/rich exchanger, hot rich amine piping, stripper and stripper overhead. The main corrosion mechanism in amine units are:
• General and localized corrosion : depending on CO2/H2S concentration, Amine used, Temperature and pressures
• Hydrogen Induced Cracking, Stress Corrosion Cracking in H2S environments
• Hydrogen embrittlement due to H2S environments. The combination of processing and operating conditions, materials of manufacture and previous failure become a potential location for corrosion problems.
This paper reviews the state-of-the-art in the understanding of amine unit corrosion and identification of outstanding issues. It also discusses two new tools in the evaluation of amine unit corrosion. These include:
•The use of online, real-time corrosion monitoring to assess operating conditions to rapidly identify and mitigate corrosion causing process conditions.
• The use of new corrosion modeling and prediction tools for quick assessment of corrosion risk versus changing operating conditions, and for use in assessment of inspection intervals and materials selection.
Experience with Corrosion in Amine Units Going back to the late 1950’s, an experience survey conducted by the American Petroleum Institute showed that corrosion by amines in gas treatment and sulfur recovery was a major concern in the refining and petrochemical industry. Major causes of amine corrosion identified included:
• Poor plant design and operating practices, and
• Solution contamination by impurities and heat stable salts.
In general terms, it was identified that amine systems used for removal of only CO2 were generally more corrosive than those involving only H2S, and MEA systems were generally more corrosive than DEA. More fundamental treatments of amine corrosion problems have identified that corrosion is usually most severe where the gases are absorbed or deabsorbed from rich amine solutions where temperature, flow and turbulence are important considerations.
The most common material of construction for amine units is carbon steel. Amine stress corrosion cracking (the combined effect of corrosion and stress) is also a consideration that has been extensively discussed particularly in lean amine systems. However, experience has proven that more failures in amine units have actually resulted from corrosion leading to severe consequences.
Process boundaries (for corrosion control) in amine systems including the following:
• Amine concentration (around 20% for MEA)
• Acid gas loading (0.3 to 0.6 mole/mole depending on the amine solvent utilized)
• Flow rate (rich amine – 6 fps; lean amine – 6 to 20 fps)
• Rich circuit temperature limits in the range 210-220 F)
• Lean amine reboiler temperature limits in the range (260 – 300 F)
• Limits for amine unit impurities and heat stable salts are typically based entirely on “rules of thumb” and may range from 1 to 2 percent.
These boundaries have been developed by evaluating service experiences in various plants with sometimes vastly different designs, throughput and operating conditions. Consequently, these boundaries vary unit-to-unit with little technical basis to help designers and operators optimize unit reliability and performance. There has been only limited use of predictions based on test data or plant monitoring data.
In the past five years or so, there have been a few studies on corrosion in the lean amine circuit . The findings indicate that lean H2S loading, impurities, polymer formation, temperature and flow rate can all influence corrosion in lean amine system. However, despite this recent emphasis, this work has not been able to fully define the corrosiveness of lean amine systems, but have shown directions that, if employed more fully, could improve amine unit throughput, reliability, and reduce unplanned outages and thereby reduce the need to standby capacity.
One effort showed the dynamic nature of corrosion in a lean amine circuit using online, real-time monitoring to identify specific operating conditions to periods of high corrosion rate . Despite the believe by many engineers that lean amine H2S loading is the major culprit, the use of online plant monitoring of corrosion showed that the root cause can sometimes be otherwise. Other process issues can predominate in lean amine systems as well, including polymer formation, and conditions that result in locally high wall shear stress (the mechanical forces on the metal surface that can remove protective corrosion scales). These include conditions of high throughput, turbulent flow around elbows, tees and weld protrusions, and flashing on heat transfer surfaces. This later situation leads to both high wall shear stress and increase in chemical aggressivity leading to severe corrosion in this portion of the lean amine circuit. Other studies have also related plant corrosivity with the formation of organic acids. But, it has been found that predicting corrosion rate based on acid concentration alone can yield misleading results when going from unit-to-unit. Hence, it appears that for accurate prediction of corrosion rate in lean amine systems, multiple variables must be considered, but only limited data are readily available.
Real-Time Monitoring in Amine Systems
In today’s market where losing 50 MMSCF of natural gas production can cost upwards of $400,000 per day of downtime, it is critical for these units to be online as much as possible. A rigorous corrosion monitoring and inspection program that includes predictive analysis and real-time monitoring can significantly contribute to maximizing uptime by minimizing failures and reducing inspection frequencies. Significant savings can also be realized through optimizing utility consumption within the unit. On-line corrosion monitoring can quickly help define the optimal lean acid gas loading windows which can prevent wasting steam by regenerating an amine solution further than required. An average sized (100 MMSCF) amine plant can spend up to $48,000 per day to generate the steam required to run the process. A two percent reduction in the steam consumption enabled by on-line corrosion monitoring would result in a $350,000 per year savings.
Honeywell recently introduced a new online, real-time corrosion monitoring transmitter that utilizes the existing plant distributed control system (DCS). The DCS serves as the primary data path for critical process data and implementation of advanced process control functions. Therefore, operators can effectively maximize productivity plant-wide and quickly responding to system changes and upsets on specific units. With application of Honeywell’s new SmartCET® corrosion transmitter (CET 5000), corrosion data can be easily acquired automatically, viewed in combination with process data and readily integrated into the global process optimization process. This integration is an important aspect since most corrosion data is still being obtained offline manually using corrosion coupons. The problem is that this manual process requires substantial time and effort to view leading to unexpected corrosion failures and unit outages.
In addition to directly impacting costs and profitability in the plant, predictive assessment and on-line corrosion monitoring can play a vital role in management of change processes within the facility. Changes in process conditions, gas conditions or chemical products can significantly impact corrosion rates within a facility. Utilizing SmartCET corrosion monitoring tools, plant management has the flexibility to quickly see and quantify the effects of any changes on corrosion and make clever decisions based upon real time data.
Honeywell field-proven corrosion solution offers measurable improvements and benefits:
1) Increase plant utilization due to improved reliability of assets
2) Reduce maintenance costs by moving from preventive to reliability-based maintenance
3) Maximize production throughput while protecting plant assets
4) Improve safety environmental compliance by minimizing the effects of process upsets and excursions
5) Optimize inhibitor costs
An example of the use and benefits of online corrosion monitoring in an amine unit is presented herein. This amine unit operated for approximately eight years without corrosion issues and then unexpectedly started to experience severe corrosion and integrity failures. A preliminary investigation recommended installing online corrosion monitoring equipment to relate corrosion rates to process changes. The corrosion monitoring system selected was the Honeywell online, real-time SmartCET system (CET5000). As is now being observed in more plant situations, the outcome of this investigation showed that corrosion was not a continuous process, but was related to periods of process instability that could only be efficiently address with real-time monitoring.
The amine unit in question was a mixed solvent system in Western Canada receiving a combined gas containing approximately 21 mol% H2S and 5 mol% CO2 with the target acid gas loading of 0.8 mol acid gas/mol amine. In this case, the pressure of the rich amine was decreased prior to entry into the flash drum. Any co-absorbed hydrocarbon flashes off the rich solution and is sweetened with lean amine in the flash gas contactor. The remaining rich amine was heated by the hot lean solution from the bottom of the regenerator prior to entry into the regenerator. The H2S, CO2, COS and mercaptan were stripped from the solution with steam which was generated by reboiling the liquid at the bottom of the column in two kettle type reboilers using 500 kPag steam. The amine system had a good service history and solution monitoring was an integral part of the operation of the amine systems. Routine weekly and monthly testing of the solutions was on-going since plant start up. Ancillary tests included analysis for amine components, residue components, organic acids, metals and lean loading. The amine degradation products are generally referred to as residue and are mostly comprised of amine polymers. The polymers are believed to increase corrosion due to their potential chelating strength and ability to breakdown protective corrosion films.
Problem The service history and regular solution monitoring, indicated that there was basically minimal corrosion in the amine system for 8 years and then, the rate of corrosion attack increased substantially. At the time, inspection of the reboiler found extensive damage on the bottom of the shell at the weir plate with a general corrosion rate of 7 mm/yr. The hot liquid return lines had multiple indications of general corrosion concentrated primarily at the elbows as shown in Figure 1. Although there were increases in known corrosive species such as organic acids and residues, the results of lab tests indicated that none resulted in a major increase in the corrosion rate that was experienced in the actual unit. A possible contributing factor to the corrosion was postulated to be changes in process variables and hence, it was felt that online corrosion monitoring could facilitate the correlation of corrosion and variations in unit operating conditions.
Conventional corrosion coupon monitoring was not a viable option. Corrosion coupons exposed for periods of months would not identify short periods of process upset resulting in corrosion. Even use of conventional electrochemical techniques to produce corrosion versus time plots would be problematic since corrosion is viewed in isolation from process parameters. Without a direct link to corresponding process information, periods of variable corrosion rate could be observed, but without correlation to process conditions, the data do not provide a way to achieve root cause analysis.
Online, Real-Time Corrosion Monitoring In this case, SmartCET was used to obtain corrosion measurements in the lean amine circuit every seven minutes. This frequency of data sampling was compatible with normal process control measurements. Monitoring was achieved with a three electrode sensor and its unique transmitter that runs three electrochemical techniques remotely and automatically. Pre-processed corrosion signals were taken into the amine unit process control system and displayed in the plant data historian along with selected key process variables. Figure 2 is a typical screen view of corrosion and process information available to plant personnel including operators, maintenance and engineer functions. It shows the way that corrosion and process data trends can be viewed simultaneously. The buttons on the lower left hand corner link to trends of additional SmartCET data, as well as links to view process and corrosion data on an individual graph to fine tune the correlations between process changes and corrosion. Further links are incorporated to amine composition and corrosion data gathered by laboratory testing.
Data Analysis & Problem Identification Following commissioning of the online corrosion monitor, a drastic decrease in corrosion rate was observed. Review of the corresponding process data showed that there had been some adjustments made to the unit operating conditions. However, a review of the solution chemistry showed that most of the aforementioned corrosive species remained at near constant levels except the residue level had decreased. However, the process instability was identified as large variations in the differential pressure during the period just prior to the internal inspection of the system and when the corrosion rate decreased. The residue level increased to beyond 10 wt% at the onset of the corrosion episode and then decreased past 9 wt% when the corrosion rate decreased.
Figure 3 shows the real-time corrosion rate obtained during an upset in processconditions. This upset was characterized by sharp changes in the differential pressure across the regenerator as well as changes in steam rates to the reboilers, and amine to the regenerator as operations reacts to the upset. The change in corrosion rate was dramatic going from around only 0.15 mm/yr to over 1.25 mm/yr at the onset of the upset and remains high even after the regenerator appears to regain some stability. The corrosion rate did not decrease fully until the steam rate to the reboilers and the temperature at the top of the regenerator stopped fluctuating. The short duration of the high corrosion rate suggests that the instability causes a short period in which corrosive species exist and also shows that there must be certain other process conditions that allow the return of a protective iron sulfide scale.
The data from SmartCET was found to correlate with the extent of corrosion damage found during periodic non-destructive testing (NDT) in the hot lean amine system. But, the online corrosion data allowed for the identification of the specific process conditions that produced the corrosion rather than only the accumulated corrosion damage commonly found my inspection or use of corrosion coupons.
Figure 3 – Correlation between real-time corrosion rate (black line) and amine reboiler/regenerator process variables (red lines as indicated above: differential pressure across the regenerator, steam rate to reboilers, top temperature in the regenerator, and rich amine to the regenerator)
New Corrosion Prediction Tool for Rich Amine Circuits
Honeywell is also involved in the development of corrosion prediction software tools for assessment of refinery systems. It has just completed a multi-year joint industry program (JIP) sponsored by a group of leading refiners and service companies. The scope of this effort was to extend corrosion forecasting capabilities for use in risk. Corrosion rate assessment, definition of inspection intervals, alloy selection, process optimization, and asset management. The program was directed specifically at providing a better technical basis for predicting corrosion in rich amine system over a range of operating conditions for commonly used alloys. The culmination of this effort was the development of the Predict ®-Amine corrosion prediction software to facilitate rapid and consistent corrosion prediction by participating companies. A second phase of work is now about to start that will extend this methodology to lean amine circuits.
The approach used in the Amine JIP was to develop corrosion rate data under simulated service conditions using a sophisticated laboratory flow loop that simulated temperature, pressure, acid gas loading, CO2/H2S ratio, impurity concentration and flow conditions. This engineering data developed concentrated on three amine solvents that included: (a) 18% MEA, (b) 30% DEA, and (c) 45% DGA, with some limited testing of 45% MDEA. Alloys included in the study were carbon steel (base metal and weld heat affected zone structure), three stainless steels (AISI 304L, AISI 316L, Alloy 2205), and Alloy 825 (used to simulate the composition of a single pass Alloy 625 overlay).
The engineering database was used to examine the parametric influences of the primary process variables to address specific questions related to amine corrosion. This project specifically involved the development of corrosion rate data for carbon steel and commonly used alloys for a range of environments and velocities related to treatment of H2S-containing refinery streams where H2S loading and velocity are the principal variables for regularly used amine solvents and concentrations. The proposed work included parametric studies of the effects of temperature, impurities, and CO2 loading. This approach has already been shown to have substantial success in terms of being able to simulate corrosion in a previous development effort on refinery sour water (ammonium bisulfide) corrosion.
New Corrosion Assessment Methodology
As mentioned previously, most generally available industry guidelines for assessing amine corrosion have been empirical in nature and do not handle process variables in a quantitative manner. This JIP effort involved careful simulation and controlled laboratory experiments. Ionic modeling was used to determine the appropriate amounts of constituents required to achieve the desired loadings in equilibrium with the desired amine solvent at the test temperatures. This capability was critical to the design of suitable experiments to model actual process unit conditions. This type of process tool also provides a very important function in the use of the program results. Similar programs, such as Honeywell’s UniSim, are commonly used by process engineers to define specific process conditions in refinery processes.
Another important aspect of this JIP was that the characterization of flow was in terms of a scalable flow parameter – wall shear stress – and did NOT simply use a linear flow velocity. The wall shear stress is a measure of the mechanical force exerted on the internal surface of the material induced by the flowing medium that can assist in removal of normally protective corrosion scale. The determination of wall shear stress for specific flow conditions involves several factors that include the density and viscosity of the flow medium, pipe ID, surface roughness, flow regime, and the associated pressure drop. Figure 4 shows wall shear stress values for selected pipe sizes with varying linear flow rates. For proper flow assessment, this parameter also takes into account geometrical factors as related to changes in the flow path, such as weld protrusions, elbows, tees and bends. These geometrical aspects are dealt with as wall shear stress magnification factors that increase the magnitude of the wall shear stress at these locations.
One of the major accomplishments of the program was the development of a standardized methodology for representing corrosion data in rich amine systems. This methodology consisted of the development of isocorrosion diagrams for each solvent (MEA, DEA and DGA) and material (carbon steel, 304L, 316L and alloy 2205 stainless steels and alloy 825). The basic isocorrosion diagrams utilized the values of the measured corrosion rates plotted using the flow velocity (wall shear stress) on the vertical axis and the mole/mole H2S gas loading for the horizontal axis. These isocorrosion diagrams were developed for each solvent type at a nominal temperature of 54 C. The range of H2S loading used varied from a “low-rich” loading level of 0.2 mole H2S /mole amine to the “high-rich” loading level of 0.5 to 0.8 mole H2S/mole amine depending on the solvent type. This served as the baseline data for the program. Wall shear stress levels examined in this program were from 0 Pa (near static) to over 1600 Pa (typical of highly turbulent flow).
Once the baseline iso-corrosion diagrams were developed, parametric tests were conducted to evaluate the influence of selected process parameters on the corrosion rates. These parameters included: Temperature (54 C to 121 C), CO2/H2S ratio (0:1 to 1:1), and amine impurity concentration (up to 3 to 6 wt% depending on solvent using a cocktail of acetate, formate and oxalate). Additionally, sensitivity tests were conducted on the influence of higher amine solvent concentrations for MEA and DEA and the performance of 45% MDEA evaluating selected operating conditions for this amine solvent.
New Predictive Software Tool – Predict-AmineThe results from this program constitute a large body of test data that require proper modeling of flow conditions to be accurately utilized. To facilitate query of the program data, a software tool called Predict-Amine was developed as part of the JIP. It incorporates the program data, industry experience and streamlines the calculation of corrosion rates with incorporation of the appropriate wall shear stress characterization for the process stream. This software tool provides a data screen for input of the amine solvent type along with application and process stream variables. This information is then used to predict the corrosion rate of materials of construction using the following sequence:
• Calculation of the effective wall shear stress from process flow conditions
• Conversion of the field wall shear stress into an equivalent laboratory test flow velocity
• Use of the laboratory test flow velocity and the H2S loading to predict corrosion rates at 54 C for the five materials (carbon steel, 304L, 316L, alloy 2205 and alloy 825) from their respective isocorrosion diagrams (i.e., baseline conditions)
• Correction of the baseline corrosion rates for the effects of process temperature, CO2/H2S ratio and impurity concentration.
The software tool input screen is shown in Figure 5. The top left section of the screen is used for input of the environmental parameters and the top right section is used for application specific parameters, such as pipe size, corrosion allowance, design life and flow geometry descriptors. The bottom portion of the input screen is used to input the process stream properties and flow rates.
Upon calculation, the software tool provides the output screen listing the five materials and their predicted corrosion rates (See Figure 6). The acceptable corrosion rates (based on the input corrosion allowance and design life) are shown with a green box. Non-acceptable corrosion rates appear with a red box. In addition to the corrosion rate results, the results screen displays the flow regime, calculated wall shear stress, superficial liquid and gas velocities and equivalent laboratory flow velocity (for 100% liquid flow). A comments section is also provided to document the user assumptions, scenarios, etc. Input and results can be saved as a consultation file and printed.
The program data and beta software were reviewed by the program participants who obtained a single user version of the final Predict-Amine software. Now that this effort is complete, these participants also qualify for additional single or multi-user licenses at preferred rates. Non-sponsors that agree to the confidentiality restrictions of the JIP and pay the applicable non-sponsor fee can also gain access to the program results and software. A secure website has been maintained to archive all JIP meeting presentations, minutes, discussion topics, deliverables and software downloads.
New JIP Effort on Lean Amine CorrosionWith the completion of the JIP on rich amine corrosion, it was felt that there was a firm technical basis for the database and the corrosion prediction methodology. Therefore, it was decided to extend this methodology to analyze corrosion in lean amine circuits, as well. This new JIP effort will start in mid-2007 with the formation of a new industry group to fund and lead the program on lean amine corrosion prediction. As in the previous case, the program will develop a critical database of corrosion data developed under simulated lean amine process conditions and finish with the development of software code that will extend the currently available Predict-Amine to lean amine conditions.
This work will focus on the developing a data to examine the effects of several parameters on lean amine corrosivity to steel, stainless steels (304L, 316L and alloy 2205) and alloy 825. The concept will be to utilize a similar testing apparatus as used in the rich amine JIP effort, but with the modification that it will also reproduce variable conditions of heat transfer (?T) found in amine unit regenerators and reboilers. In the case of lean amine corrosion, experience also indicates that lean amine (H2S) loading will also be a critical variable controlling corrosion. The loading of acid gases in the lean amine, particularly H2S, is known to influence corrosivity of lean amines to carbon steels and possibly other material. Industry refers to “over-stripping” where reduced H2S levels cause the FeS film to be unstable which results in increased corrosion rates.
In this new program, the baseline conditions will be one H2S level (approx. 100 ppmw), one base solution temperature with three different ?Ts (15, 25 and 50 F tube skin temperature above mean amine temperature) per test simulating reboiler conditions, one level of contaminants (nominal maximum concentration), and three flow velocities. Additionally, parametric data will be developed. Based on discussions with major oil companies, the major factors to be included in this program are (1) H2S loading, (2) contaminant concentration (3) service temperature. In the cases where the residual level of H2S in the lean amine will be controlled, three different H2S levels will be utilized in the range of approximately 50 to 1000 ppmw. Ionic equilibrium calculations will be used to set the initial loadings and as a means to assure reasonably constant acid gas loadings will be used for all tasks. In each case, four amine solvents will be evaluated: MEA, DEA, DGA and MDEA. When combined with the results from the previous richamine JIP, the corrosion prediction methodology will be able to handle a complete range of amine unit conditions.
For more information Please visit Honeywell website at www.honeywell.com/ps/sea
SmartCET is a registered trademark Honeywell International and a product of Honeywell Process Solutions. Predict is a registered trademark Honeywell International and a software product of Honeywell Process Solutions
