For crude-unit overhead systems, pH is the main process parameter that impacts corrosion rates. To control corrosion conditions, many operators use various neutralizers at optimum ranges determined by site-specific conditions. A four-year study (20052008) was conducted at a Petrobras refinery using amine-blend solutions to control pH. Over this period, corrosion rates were measured through ultrasonic inspections and weight-loss coupons. Important process parameters monitored included:
pH, chloride and iron concentrations at the bottom of the overhead drum
Neutralizer and inhibitor flowrates.
A qualitative comparison was done with two refineries, using two other neutralizers: sour water from fluid catalytic cracking (FCC) unit and an ammonium aqueous solution. This investigation proved that maintaining a low chloride level and stable pH levels were the most effective ways to control equipment damage from corrosion. Also, the study found that several inspection techniques were particularly useful in estimating service life for pipes and other crude-unit equipment. Applying better pH control and improved monitoring and inspection programs can reduce equipment damage from corrosion.
Hydrocarbon-processing companies follow different methods in controlling crude distillation unit (CDU) overhead corrosion. Common approaches include inhibitors to neutralize acid solutions. Even with a good control on crude oil in storage tanks and the desalting process, hydrochloric acid (HCl) will still be present at the atmospheric tower overhead and it demands proper chemical treatment. Merrick and Auerbach performed a study on 129 different distillation units. From these studies, it was observed that the average chloride concentration was 10 ppm to 30 ppm in accumulator overhead drums.1 Chlorides are generated from some salts contained in the crude oil that is processed in the CDU; thus HCl is formed.
There are three main ways to neutralize acidic aqueous solutions at the CDU overhead; they include injecting:
Gaseous ammonia (NH3)
Ammoniac water (NH4OH solution)
Neutralizing amine solutions.
Regardless of the neutralization technique applied, the pH is lower than the dew point of water. This adds more challenges in measuring pH when condensation occurs; this is the preferred region for the corrosion process to begin. Neutralization equations are:
HCl (aq) + NH3 (aq) => NH4Cl (aq)
HCl (aq) + RNH2 (aq) => RNH3Cl (aq)
One concern for neutralization is the difficulty of controlling the ammonia or amine flowrates, which depend on the varying HCl levels in the CDU. The neutralizer injection levels can be too low and the pH in the overhead can drop. Excess neutralizer levels, especially in the presence of hydrogen sulfide (H2S), contribute to precipitation of salts, such as ammonia or amine disulfides or chlorides. Once formed, these salts (molten or solid) deposit on pipe surfaces, likewise, they can cause localized corrosion with a high rate of thickness loss. If salt formation occurs after condensation, then its dissolution into water represents minimal corrosion risk.2
In this article, some field results are presented, including chemical analysis, pH and corrosion rate for a CDU tower overhead. A qualitative comparison was conducted investigating the different ways to control corrosion, as listed in Table 1.
At Refinery A, the observed corrosion rates in pipes, heat exchangers and accumulator drum, were obtained from thickness measuring via ultrasonic testing. Corrosion rates reached values of 0.15 mm/yr. The heat exchanger tubes presented an average service life of only seven years, and corrosion deposits were found in baffle-plate regions. Some thickness loss and stress-corrosion cracking were also reported on the shells of the equipment, which can be attributed to high H2S levels (2,553 ppm) in the neutralizing solution used.
At Refinery B, high chloride levels, caused by inefficiencies in the crude preparation and desalting processes, generated high corrosion rates. The average corrosion rate observed by the coupon weight loss over two years was 0.29 mm/yr. The pipes connecting the top of the atmospheric tower and condensers have flaws from previous campaigns. Localized under-deposit corrosion in the lower blank of the condenser shell was observed.
At Refinery C, monitoring results for coupons installed in the overhead condenser (air cooler), had average corrosion losses of 0.16 mm/yr. This reflects a uniform thickness loss expected in equipment and pipes. But there were failures in pipes caused by localized under-deposit corrosion. In the air cooler, the average tube service life was five years, and failures were reported immediately after the flow entrance, where condensation begins.
For this study, two regions of the overhead pipe were selected at Refinery B to conduct thickness measurement via ultrasonic testing. Initially, the testing was separated into two areas:
Overhead atmospheric tower and condenser
Between the condenser and accumulator drum.
Different behaviors are expected from the pipes carrying the fluid before the condenser in the vapor phase (by design), then after, where the water is already in liquid phase. Thus, the observed corrosion rates were different, as shown in Fig. 1.
| Fig. 1. CDU showing locations identified for |
thickness monitoring by ultrasonic testing.
The locations for thickness measurement are always chosen based on the experience of the inspection team supervising the unit and the measurements usually apply these aspects:
In the curves, the corrosion rates may be higher due to an increased propensity for the occurrence of corrosion associated with erosion.
Regions of encounter between two pipes, in the form of T, are also preferred regions for erosion.
In the straight sections, fewer points are selected, which are expected to be representative of the system.
In 2004, measurements were made on about 70 points before the condensers. Another 70 points were inspected with the same technique in pipes after the condensers. The same 140 points were inspected again in 2005. The results allowed defining several average corrosion rates:
Before the condensers: 0.14 mm/yr
After the condensers: 0.16 mm/yr.
Also, there were high standard deviations in both cases, 0.16 mm/yr before and 0.15 mm/yr after the condensers. The highest rate observed in the first case was 0.63 mm/yr, and the lowest 0.02 mm/yr. After the condensers, the highest rate was equal to 0.61 mm/yr and the lowest 0.02 mm/yr.
Before the condenser, 25 points were measured again in 2008; the average corrosion rate (20052008) was equal to 0.17 mm/yr. The data from the corrosion coupons indicated an average rate of 0.19 mm/yr in the same period.
In December 2008 and June 2009, five points were measured in random areas of the pipes before the condensers, resulting in an average corrosion rate equal to 0.21 mm/yr. In the same period, some corrosion coupons were analyzed monthly, positioned on the inlet connections of overhead condensers; these coupons had an average corrosion rate equal to 0.28 mm/yr.
For Refinery B, the study was expanded to include how operating conditions had contributed to equipment deterioration. For example, in 2007, some studies researched the impact from the chloride content variations in the feed at the pre-flash tower overhead, while the corrosion rates were measured by coupons. Due to the ineffectiveness of desalting, there was a direct influence on the corrosive process in the overhead system, as is illustrated in Table 2. During these periods, electrical problems caused transformer problems that affected the inner electrodes to the desalting drums. Poor desalting of the crude led to chloride levels above 1,000 ppm in the overhead accumulator drum. Also, pH was affected, reaching values approximately 4. Weight-loss coupons were installed in the inlet connections of the atmospheric overhead condensers.
To improve desalter efficiency at full operating conditions, more tests were made by adjusting the differential pressure of the mixer valve. Usually covered with ∆P = 1 kgf/cm², this value was increased by 0.2 kg/cm². Result: Without any other changes in operating parameters, a reduction of 44% in the chloride content in the accumulator drum was obtained.
In the overhead atmospheric system, pH, chloride, iron and corrosion rates were monitored by weight-loss coupons. Fig. 2 shows the historic data of pH values measured in the accumulator drum since 2008. The figure shows the mean values and standard deviations for measurements over each month, except May and September, when there were no analysis reports.
| Fig. 2. Historic data of pH values in the |
overhead accumulator drum of the
atmospheric tower for refinery B.
It is observed that the average pH over the years has always been very close to or within the recommended range. But the high standard deviations showed a lack of control during some periods. There were some incidents in January in which a pH reaching 1.5 was observed and adjusted to 3 on the same day and recovered to a pH = 6 on the next day. On two days, the pH reached 4. The higher standard deviation observed in this month contributed significantly to increased corrosion rates.
Fig. 3 shows the measured chlorine values in the same drum. The target is 40 ppm as the maximum, which can only be guaranteed with efficient control in crude preparation at the storage tanks and desalter.
| Fig. 3. Historic data of chloride values in the |
overhead accumulator drum of the
atmospheric tower for Refinery B.
It is observed that the values remained above the recommended targets throughout the year, showing deficiency in the early stages of crude processing. As an immediate consequence, we can expect greater usage of neutralizers and corrosion inhibitors. What is not always sufficient to maintain is the appropriate pH and low corrosion rates over slack periods, as observed in January (average of 108.96 ppm chloride).
| Fig. 4. Historic data of iron values in the |
overhead accumulator drum of the
atmospheric tower for Refinery B.
The iron level in the water was also monitored. Iron can be another indicator of corrosion in the overhead system. In Fig. 4, the measurements from 2008 are shown; conditions exceeded the maximum value of 1 ppm over the year. Also, we can observe that the iron content was below the recommended limit 4 of the 10 months evaluated. These results vary greatly over the month, with standard deviations above the mean values; the data is not included in Fig. 4. Intakes of neutralizing solutions and corrosion inhibitors also represent relevant data on analyzing control parameters in the overhead system. Figs. 5 and 6 show the injection rates for neutralizers and inhibitors for Refinery B in 2008.
| Fig. 5. Historic data of neutralizer amine |
flowrates at overhead pipe in the atmospheric
tower for Refinery B.
| Fig. 6. Historic data of inhibitor flowrates at |
overhead pipe in the atmospheric tower for
The mass balance at the tower overhead is shown in Fig. 7. It is known that the chloride content measured in the top accumulator is directly linked to the presence of HC1 formed from the hydrolysis of salts present in the feed. Thus, it is possible to set base values for neutralizing agent flowrates. From the condensate analysis in the overhead drum, several periods were selected in which the chloride content was close to 100 ppm, or 50% of this, 50 ppm. On the same dates, the average flowrates of the neutralizing solution and pH were recorded, as listed in Table 3.
| Fig. 7. Mass balance of the atmospheric tower |
overhead systemRefinery B.
With the pH near the equivalence point, if we consider only the presence of HCl, neutralizer and water, the result is salt formation, N2Cl, which dissociates. We can determine the resulting pH; the reactions are:3
N2Cl j N2+ + Cl- (1)
N2+ + H2O = N2OH + H+ (2)
From the salt concentration, it is possible to determine the expected pH:
As Ka is very low, the salt concentration Cs = [N2+]:
There are many other contaminants in the overhead system, such as H2S, ammonia (NH4), sulfur oxides (SOx) and others that can alter conditions and force changes on the predicted pH values. We cannot establish a direct relationship between the chloride (Cl-), flowrate and pH neutralizer from field results.
However, we can determine the salt concentration (N2Cl) from the N2 solutions, as described in Table 4, and compare it with the expected resulting pH. Table 4 lists the results; observing that, in a few cases the values coincide, as in D1, D3, D4 and D5, and the neutralizing added on top is extremely diluted into the total water solution (264,000 l).
Corrosion monitoringweight-loss coupon.
Fig. 8 shows the historic data of the weight-loss coupons, installed on the inlet connection of the overhead condenser. There were many lack periods, in which the corrosion rate is at greater than the established limit (0.125 mm), such as in November 2004 (0.60 mm/yr), January 2008 (0.53 mm/yr) and November 2008 (0.55 mm/yr).
| Fig. 8. Weight loss monitored by corrosion |
coupons, installed at the inlet connection of
the overhead condenserRefinery B from
November 2004 to May 2009.
At the three refineries presented in this study, various problems caused by corrosion are sourced to low operating efficiencies in the crude desalting unit, which is initiated at the storage tanks. Checking field data and literature to find benchmark values for evaluating the effectiveness of existing desalters can help maximize salt-removal efforts.4 Also, leakages observed in pipelines in Refineries B and C were mainly caused by deficiencies in pH control. This is the main control parameter in the tower overhead, and it must be kept within the range with the minimum possible deviation. We could not associate a neutralizer type to observed failures.
The results of Refinery B showed that even with at stable pH behavior over the study period, corrosion increased. The standard deviation observed during 2008 was 0.54, with daily routine measurements. This value is consistent with observed deviation cited in the literature, equal to 0.78, when gaseous ammonia was used as a neutralizer in the same unit.5
The literature shows that low pH values lead to high corrosion rates on mild steel, even though the presence of inhibitors may be insufficient to alleviate this problem.6 Conversely, a pH too high can also bring negative consequences:
Using excess neutralizing solutions, based on amine or ammonia, favors the occurrence of deposits, leading to localized corrosion with extremely fast kinetics.
In stream containing H2S, such as the CDU, stability of the protective iron-sulfide film is compromised while increasing its solubility, thus accelerating corrosion.7
We can analyze a phase diagram for H2O-HCl and correlate it to the overhead corrosion process.8 It is possible to observe a temperature range of approximately 100°C to 102°C, in which an average concentration observed in the field (0.7% HCl), and in which two phases are present in equilibrium conditions: vapor (rich in water) and liquid (rich in HC1). At the temperature where condensation begins, the HCl concentration in the liquid is 10 times higher than vapor phase. Only below 100°C, in equilibrium condition, the steam is fully condensed, and the final concentration of the liquid is reached. In Refinery C, this behavior was well marked, as leakages occurred in the starting point of condensation on the overhead air cooler, while the rest of the pipes were found in good conditions.
To increase the process data analysis, the measured consumption of neutralizing amine in the overhead during 2008 were compared with values originally estimated by the supplierdata presented in Table 5. The comparison was done in a period in which the main process variables, such as pH and chloride content in the overhead drum, did not suffer interference from typical discontinuities, such as high levels of base sediment and water in oil. The selected period was the months of June 2008 to August 2008, in which the corrosion rate was below the recommended target of 0.125 mm/yr, as shown in Table 6.
Application of neutralizing amine can be varied for many reasons, such as incorrect pH measurement, which interferes directly in injection flowrate. If the quality of the crude is kept almost constant, the product amount injected into the overhead stabilizes. This is the condition studied in the chosen periodoptimum injection to compare the predicted with the far field. From Table 5, the relationship between the measured and predicted consumption of neutralizer is doable.
For a maximum chloride content of 50 ppm at the overhead using data of steam injection background and specific consumption provided by the manufacturer, the amount of amine provided at the top would be 60 l/d, while in practice, keeping variables under control, the measured consumption was 120 l/d. We can conclude that the predicted flowrate for the neutralizing solution can be a guide for the process, but only constant pH monitoring (preferably online) can promote adequate control for amine injection.
Corrosion rates are directly proportional to pH. Accordingly, field monitoring uses weight-loss coupons to validate the quality of process parameters control. At Refinery B, measurements were made from 2004 until early 2009, when only 45% of cases were below the limit0.125 mm/yr. Throughout 2008, the weight loss was framed in only 30% of the months monitored. Comparing these results with inspections by thickness measurement, we realized that the difference between the rates obtained with both techniques was short only at the second decimal number, as shown in Table 7.
This study listed a number of results available in many CDUs. But the relationship between them can generate even more support for inspection teams that manage equipment integrity. From the temperature (T) and pressure (P) in the overhead pipe, it is possible to estimate if water vapor and its components reach the dew point before the condenser. The pH measured in the accumulator drum indicates how the developed corrosive process will progress throughout the system. The chloride content, which is directly related to the flowrate of the neutralizer, also increases corrosion at high values, even if the pH is controlled. Injecting inhibitors can reduce corrosion rates but not with the same intensity as pH adjustments. Thus, we must work to meet the primary objective of the refinery integrity program: to reduce unplanned shutdowns, identify root causes for corrosion degradation of equipment and ultimately develop a good corrosion monitoring program.9
Conclusions. Among the available neutralizing solutions, refiners should use the one that provides the best efficiency, coupled with the cost benefit for each unit, while considering environmental aspects from waste generation and final treatment. There are pros and cons associated with each neutralizer.10 The results showed that the type of neutralizer used on the CDU atmospheric tower overhead was not the determining factor in minimizing corrosion. Only a good control of process parameters, especially the desalting efficiency (low chloride level at the overhead accumulator drum), can increase equipment service life. We can also establish a direct relationship between the historic data of the process parameters (chloride level, pH, temperature and pressure) and the expected thickness loss of the equipment and pipes.
Monitoring weight-loss coupons is essential to validate the quality of the process parameters control. At Refinery B, the rates obtained with the coupons were compared to results from inspections by ultrasonic thickness measurement, where only a small difference in the second decimal number (0.02 mm to 0.07 mm) was observed. With these low rates and constant monitoring, the likelihood of failure is minimized, and it becomes possible to predict damage to equipment and avoid unplanned shutdowns due to equipment failures by corrosion.
Plant results and literature data indicate that there is an optimal pH control range for the CDU overhead system. The main process parameter, defined in terms of two main corrosion mechanisms are:
At low pH (pH below 5.5) the HCl causes severe corrosion in the mild steel
At high pH (pH above 6.5), due to the presence of H2S, there is an increase in the uniform corrosion rate due to the breakdown of the iron sulfide layer, and localized corrosion under deposit is also more likely to occur because of the salts formed.
For each system, an optimal range should be specified. It will depend on the chemical composition of the final solution obtained in the accumulator drum. It is important to note that pH stability is dependent on system automation. More reliable online information enables low deviations if there is an instrumented injection control fed by online pH measurement. HP
1 Merrick, R. D. and T. Auerbach, Crude unit overhead corrosion control, Materials Performance, September 1983, p. 15.
2 Couper, A. S. Bothered by corrosion of your crude-unit condensers?, Oil & Gas Journal, July 1964, p. 79.
3 Harris, D. C., Quantitative Chemical Analysis, 7th ed., California, 2006.
4 Gutzeit, J. Controlling crude unit overhead corrosion by improved desalting, Hydrocarbon Processing, February 2008, p. 119.
5 Jambo, H. C. M., D. S. Freitas and J. A. C. Ponciano, Ammonium hydroxide injection for overhead corrosion control in a crude distillation unit, International Corrosion Congress, Granada, Spain, September 2002.
6 Gutzeit, J. Effect of organic chloride contamination of crude oil on refinery corrosion, Nace, Orlando, Florida, March 2000.
7 Sardisco, J. B. and R. E. Pitts, Corrosion of Iron in an H2S-CO2-H2O System: Composition and Protectiveness of the Sulfide Film as a Function of pH, Corrosion, November 1965.
8 Potolokov, V. N., V. A. Efremosv, S. V. Nikolashin, T. K. Menshchikova, E. G. Zhukov and V. A. Fedorov, Liquid-Vapor Equilibrium in the AsCl3-HCl-H2O System, Inorganic Materials, September 2006, p. 1027.
9 Ropital, F. Current and future corrosion challenges for a reliable and sustainable development of the chemical, refinery, and petrochemical industries, Materials and Corrosion, July 2009, p. 495.
10 Jahromi, S. A. J. and A. Janghorban, Assessment of corrosion in low carbon steel tubes of Shiraz refinery air coolers, Engineering Failure Analysis, November 2004, p. 569.