FOUR STEPS SOLVE CRUDE-TOWER OVERHEAD CORROSION PROBLEMS

Norman P. Lieberman
Process Improvement Engineering
Metairie, La.

Troubleshooting experiences have shown that, in the crude oil distillation process, a two-stage overhead system is the type of facility that most often suffers from severe corrosion.

Changing the point of injection of the neutralizing amine is one of four factors that, used together, can reduce corrosion in these systems.

CORROSION

The most common problem Process Improvement Engineering has encountered in performing crude-unit troubleshooting is corrosion-related failure caused by hydrochloric acid (HCl) attack (Fig. 1). Affected components are:

Atmospheric-tower overhead condenser tubes and reflux drums
Atmospheric-tower fractionation trays
Vacuum-tower overhead condenser tubes and jet-ejector bodies.

No less important than mechanical failure caused by metal loss is the plugging of trays, draw-off nozzles, and condensers as a result of the accumulation of corrosion products consisting of chloride salts, iron sulfide, and ammonium bisulfides.

PROBLEM ORIGIN

Hydrochloric acid attack is one of the oldest and best-understood refinery process problems. This would lead one to think that preventing localized hydrochloric acid attack would also be well-understood. Experience, however, indicates that this is far from true.

The problem originates with the presence of salt water in crude oils. Most of the salt content of the brine dispersed in crudes is in the form of sodium chloride (NaCl). About 10% of the salt, however, is present as magnesium chloride (MgCl2) and calcium chloride (CaCl2). In the presence of water and heat, these chloride salts hydrolyze to form HCl.

Of these salts, NaCl is the most heat-stable. CaCl2 is moderately heat-stable, and MgCl, is the least heat-stable. Most of the MgCl2 Will hydrolyze at typical crude-tower flash zone temperatures (about 650 F.). Most of the NaCl will remain unaffected at these conditions.

At typical vacuum-tower flash zone temperatures (770 F.), a substantial amount of the CaCl2 also will hydrolyze, as will a small fraction of the NaCl.

It would be misleading to further attempt to quantify the extent of hydrolysis because of the effects of:

High oil-film temperatures in preheat furnaces
Catalytic effects of naphthenic acids on the hydrolysis reaction
Hydrocarbon residence time in the bottom of the atmospheric and vacuum columns.

DESALTER OPERATION

A single-stage crude desalter should remove 90% of the salt content of the crude oil. A double-desalter operation should extract 99% of the salt.

These statements create the comforting but misleading impression that double desalting removes troublesome salts before they can cause downstream problems. It is unfortunately far more difficult to remove MgCl2 than NaCl.

For example, the following chloride-salt removal efficiencies were observed in one single-stage desalter:

90% NaCl removal
50% CaCl2 removal
40% MgCl2 removal.

Because the hydrolysis of MgCl2 proceeds far more readily than that of NaCl, its potentially lower removal efficiency in the desalter will create a disproportionate downstream corrosion problem.

Experience also indicates that the operations of desalters are, in many refineries, rather erratic. One client operating a two-stage desalter.

Experienced salt-removal efficiencies varying between 65 and 95% and averaging 90%. Another refiner experienced intermittent desalter water carry-over caused by interface level-control problems.

Of course, during such upset conditions, the operators did not sample the desalted crude because they realized the desalter efficiency would be low.

Based on observations, the principal problems with desalter efficiency are:

Level-control problems caused by plugging of level taps
Failure caused by shorting of electrical grids
Improper control of desalting chemicals and lack of pH control
Poor-quality desalter makeup water
Low mix-valve delta pressure
Low desalter-water flow rate
Vaporization in the desalter
Erratic desalter back-pressure control
Sludge buildup in the bottom of the desalter.

ACID EVOLUTION

Most crude-unit corrosion problems are more likely caused by erratic desalter operation than by consistently poor desalter efficiency. If, on the other hand, salt-removal efficiency is consistently 85%, a corrosion-control program can be implemented to deal with the HCl evolved.

Ammonia, morpholine, or higher-boiling-point neutralizing amines can be injected into the atmospheric-column overhead and judicious amounts of caustic can be added to the desalter effluent.

If, however, desalter efficiency varies between 0 and 95% and averages 85%, effective corrosion control is impossible.

From a short-term operational point of view, a loss of desalter efficiency for a few hours will have no observable effects at the control panel. On the other hand, the short-term corrosion rate during this period can increase exponentially.

And, if during this upset no samples are taken of the desalter effluent, the reflux-drum water, or the vacuum-seal drum effluent water the external effects of temporarily stopping the desalter water flow will go unnoticed.

CAUSTIC INJECTION

Injection of sodium hydroxide downstream of the desalter is an effective method of moderating hydrochloric acid attack in the crude-tower overhead system (Fig. 1). A rough rule of thumb based on observation in one plant is to inject 1 lb NaOH/lb salt in the desalter effluent.

The exact amount of caustic injection is based on the chloride content of the water in the crude-column overhead reflux drum. A reasonable target is 10-20 ppm chlorides, as measured in the second-stage reflux drum boot.

The caustic is injected through a quill in a horizontal run of piping in a direction opposed to the direction of flow. A good source of NAOH is spent caustic from a gasoline-sweetening operation.

For most situations, the maximum caustic injection rate is 5 lb/1,000 bbl crude.

HCL ATTACK

About one third of all crude-unit problems observed by Process Improvement Engineering have been caused by the hydroscopic attack of hydrochloric acid (i.e., below the water surface). The indications of this attack are typically:

Flooding of the atmospheric-tower top trays
Condenser-tube leaks
High pressure drop and low heat-transfer coefficients in the atmospheric-tower overhead condensers
Plugged kerosine draw-off nozzle
High iron content and erratically low pH in the reflux-drum water
Reduction in the temperature difference between the kerosine draw and the atmospheric-tower overhead
Plugging of orifice taps, level controls, and pressure points with iron sulfide deposits
Failure of the overhead vapor line.

Although the source of the corrosive attack is HCl, the product of corrosion is iron sulfide, not ferric chloride.

There is typically an excess of hydrogen sulfide in the atmospheric-tower overhead system. These reactions predominate:

HCl + Fe yields FeCl3
FeCl3 H2S yields FeS2 + HCl

Note that the HCl is regenerated by the H2S. Hydrochloric acid thus acts as a catalyst for the formation of iron sulfide.

HYDROSCOPIC ACID ATTACK

The HCl liberated from the hydrolysis of MgCl2 in the flash zone of the crude tower has a great affinity for water. As long as no water is present, the HCl remains a noncorrosive vapor. The first droplets of water that condense from the tower-overhead vapor, however, will dissolve all the HCl they contact.

Three important concepts arise from the discussion to this point:

The presence of HCl raises the aqueous-phase dew point temperature, as compared to the condensation temperature of steam. For instance, if the calculated condensation temperature of water vapor in the overhead vapor is 190 F., the dew point of the dilute HCl might be 220 F.
The aqueous phase that condenses in the absence of a neutralizing chemical will have a pH between 1 and 4.
Although the effluent from a condenser might be 230 F. and hence above the calculated dew point of the aqueous phase, low tube-wall temperatures will cause localized condensation of HCl and therefore aggressive, although restricted, condenser-tube corrosion.

To prevent the above scenario, a neutralizing agent-usually a high-boiling-point amine--has to condense either with or before the HCl.

AMINE CHLORIDES

If a neutralizing agent (amine or morpholine) is refluxed back to the atmospheric tower, some unpleasant phenomena will result.

First, although the tower top-temperature will be too high to permit the condensation of an aqueous phase, the amine chloride salts will be heavy enough to sublime on the upper trays of the column. These salts will act as a desiccant and absorb moisture from the crude-column vapors.

A strong solution of the chloride salts will form under the solid deposits and promote underdeposit corrosion (OGJ, May 24, p. 45). The downpour areas of the tray decks appear to be exceptionally subject to this type of attack (Fig. 2).

The top eight trays of one crude unit in the West Indies were so severely corroded by underdeposit corrosion that the tray efficiency had been reduced to 10% and the pressure drop per tray had fallen to zero.

Most amine chloride salts are too light (i.e., they vaporize at too low a temperature) to escape out of the bottom of the atmospheric tower. Hence they accumulate as a solid at an intermediate point in the crude column.

The jet fuel draw-off nozzles on several units have been partially plugged with these solids.

A sample analysis drawn from a low-point bleeder on the jet draw-off line showed:

85% dry amine chloride salt
15% ferric chloride.

Injecting steam back through this bleeder dissolved the offending deposits and restored the original jet fuel production rate.

VACUUM-SYSTEM CORROSION

Investigations into the appearance of HCl in vacuum-tower overheads indicate that none of the hydrochloric acid originates from amine chloride salts. The source of the HCl in the vacuum-system overhead is the decomposition of CaCl2, with lesser amounts contributed by the initial decomposition of NaCl and the decomposition of residual MgCl2.

Vacuum-tower overhead corrosion is normally less severe than atmospheric-tower overhead corrosion because of the large volume of steam that condenses along with the HCl in vacuum towers. This steam originates either as bottom-stripping steam or as motive steam to the first-stage jet (Fig. 1).

One interesting case illustrates this point. The overhead system and trays of a crude tower were suffering severe hydroscopic HCl attack. The vacuum-tower overhead was free of HCl corrosion, even though no neutralizing chemicals were injected at this point.

The reason for this unusual distribution of HCl was that the preheater outlet temperatures for both the atmospheric and vacuum towers were the same. The higher temperature needed to continue hydrolysis of chloride salts downstream of the atmospheric tower was missing.

TWO-STAGE OVERHEADS

Process Improvement Engineering is often asked whether there is some common feature of crude units suffering from severe corrosion failures and tray-deck plugging with corrosion products. Fig. 1, which shows a two-stage overhead system, is the type of facility that most often suffers such severe corrosion.

This may strike the experienced design engineer as strange. After all, the purpose of the two-stage overhead system is to minimize hydrochloric acid corrosion.

Unfortunately there is a fundamental error in the design of the two-stage overhead system. Note the position of the neutralizing-amine injection point. Amine injected at this point will react with the HCl in the first-stage condenser.

The first-stage reflux drum is designed to run at temperatures greater than the water dew point temperature to prevent hydroscopic HCl attack on the condenser tubes. No water is intended to be collected in this reflux drum and all the condensed hydrocarbon is refluxed back to the atmospheric tower.

Tower top-temperature control is maintained by supplementing the reflux from the second-stage cold reflux drum.

Reflection should reveal this to be a recipe for disaster. The chlorides are going to be trapped between the flash zone of the crude tower and the hot first-stage reflux drum. While these chloride salts are totally soluble in water, no water is condensed routinely to dissolve them.

Field observations have shown that small amounts of water condensed only occasionally do not help. The water so condensed is loaded with particulates and quickly plugs level-control taps, sometimes even filling the entire water draw-off boot with a sludgy liquid. This liquid is refluxed back to the crude column, thus accelerating tray-deck corrosion and plugging.

REFLUX DRUMS

On one crude unit, a substantial reduction in corrosion and fouling of the first-stage condensers and atmospheric tower internals was affected by:

Relocating the neutralizing-amine injection point to the inlet of the second-stage condensers
Maintaining the hot reflux-drum temperature at 230 F. (i.e., 35 F. hotter than its previous operation).

Still, even the moderated corrosion of the first-stage condensers and tower trays were excessive. While the 230 F. hot reflux-drum temperature was high enough to prevent aqueous-phase condensation, two problems remained:

Control of the tower top-temperature was still based on taking a slipstream of cold reflux from the second-stage reflux drum. This stream was saturated with water and created a localized aqueous phase on the tower top-tray.
The overhead condenser tube-side coolant was 90 F. crude.

This meant the tube-wall temperature was often less than the water dew point of 190 F.

The upper five 410 stainless steel trays of the tower were retrofitted with Monel decks and downcomers. This effectively stopped the tray corrosion but high pressure drop and tube failure of the first-stage condensers persisted.

Blocking in the cold reflux from the second reflux drum helped reduce tower internal corrosion and plugging by preventing neutralizing amine from returning to the crude column. This, however, necessitated running the first reflux drum colder to meet naphtha-product end point specifications, which caused a water phase to form in the first reflux drum.

Although only 3% of the total overhead water was condensed at this point, the water contained 1,800 ppm chlorides, 2,000 ppm iron, and was highly acidic.

RECOMMENDATIONS

The two-stage overhead system using the hot and cold drum is a design that, despite its widespread use, does not fulfill its objective of stopping crude-tower overhead corrosion. The following program is, in the opinion of the author, the only way to halt overhead corrosion in a crude tower:

Effective desalting that removes more than 60% of the MgCl2
Caustic injection downstream of the desalter to control chlorides in the reflux-drum water draw-off boot
Forced water condensation in the crude-tower overhead vapors (Fig. 3)
Sufficient addition of neutralizing amine to control water boot at a pH of about 6.

Finally, there can never be a truly effective replacement for proper desalting. Process Improvement Engineering's experience has repeatedly proven that desalting is the most cost-effective method of controlling hydroscopic hydrochloric acid corrosion.

When downstream residhydrotreating processes preclude the use of neutralizing caustic, the importance of the desalter is magnified even further.