The sample point would not run clear, which is how it was found. A six inch rich amine isolation valve sat between the absorber sump and the lean rich exchanger of a 200 tonne per day capture unit on a gas fired boiler, and after nine months in service it was passing enough solvent to keep a two inch drain line warm. The valve had been hydrotested at one and a half times its rating, had held API 598 rate A on the bench, and carried a 316L body with 316L trim. When it was cut out, the bore was bright. The ball was bright. Both seats were intact and dimensionally correct. Under the downstream seat pocket, in the ring of body wall that the seat ring sits on, there was a chain of pits, the deepest at 1.9 millimetres, in a wall that had started at 9.5.
That is the shape of almost every ball valve failure in amine service, and it is worth stating plainly because it inverts the usual intuition. In capture plants the enemy is not carbon dioxide. Carbon dioxide is the product, it is present at a partial pressure of a fraction of a bar, and it is chemically gentle to stainless steel. The enemy is what the solvent becomes after a few thousand hours of contact with flue gas, with oxygen, with sulphur and with chloride, and the damage happens in the small still volumes where that altered solvent can sit and concentrate. A ball valve contains three such volumes by construction, and it contains them in exactly the places an inspection plan does not look.
This article works through the four streams a capture plant actually has, the two chemistry limits that decide the material, the three places the body thins, and the two behaviours unique to a ball valve that make an amine loop worse than the piping around it.
1. Four streams, four different valves
An amine unit is usually drawn as one loop, which is misleading for specification purposes, because the solvent at four points along that loop has four different damage mechanisms. The lean amine leaving the stripper is hot but CO2 lean, so it is barely corrosive and its problem is velocity. The rich amine leaving the absorber is cool but loaded, so it is corrosive in a different way and its problem is wherever flow stops. Between the two, in the cross exchanger, the solvent is both hot and loaded, and that is the single worst location in the plant for carbon steel.
Stream | Temperature | Loading, mol CO2 per mol amine | Dominant damage | Default body material |
|---|---|---|---|---|
Absorber sump, rich amine | 48 to 58 degrees C | 0.45 to 0.52 | Erosion corrosion where velocity is high; pitting in still pockets | 316L, or carbon steel with 3 mm allowance |
Cross exchanger, hot rich side | 95 to 112 degrees C | 0.45 to 0.52 | General and crevice corrosion, worst with heat stable salts | 316L minimum, 2205 above 500 ppm chloride |
Stripper bottom, lean amine | 116 to 122 degrees C | 0.18 to 0.25 | Thermal degradation of the solvent, then corrosion by its products | 316L, 2205 if chloride is high |
Lean amine after the cooler | 40 to 50 degrees C | 0.18 to 0.25 | Velocity driven erosion corrosion, negligible general loss | Carbon steel with allowance, or 316L |
Reclaimer, batch | 130 to 150 degrees C | Not applicable, caustic dosed | Caustic cracking and severe general loss | 316L or 2205, never carbon steel |
Two things follow from that table. The first is that a single material callout for "amine service" is not a specification, because the same alloy that is generous in the lean cooler is marginal in the cross exchanger. The second is that the valve that fails first is almost always the one on the hot rich side, and it fails for a reason that has nothing to do with the valve's pressure class.
2. Why 120 degrees C is a chemistry limit, not a corrosion limit
Monoethanolamine begins to degrade measurably above about 120 degrees C, and the degradation is not a slow loss of strength in the way a lubricant degrades. It is a set of reactions in which the amine reacts with carbon dioxide to form carbamates that cyclise into oxazolidinones, which then react further into oligomers and, in the presence of oxygen and iron, into organic acids. Every one of those products is either a stronger acid than the amine or a solid, and both are problems the plant cannot strip out.
Each ten kelvin above that threshold multiplies the degradation rate by about two, so a stripper running at 128 is not a little worse than one at 118. It is consuming solvent at a multiple, and it is doing so continuously, because the chemistry does not stop when the plant is at steady state. This is why reboiler temperatures are held near 120 degrees C even though the solvent is sold with a considerably higher boiling point, and why the first batch of heat stable salts in a new unit usually appears during commissioning, when steam is admitted to a reboiler that has no flow through it, or when a trace heating circuit is left on with the loop isolated.
The valve consequence is specific. A valve body is not the hottest point in the loop, but a bonnet on an insulated body, or a cavity that is not swept, can sit 10 to 25 kelvin above the stream it is attached to and hold a few litres of solvent there for weeks. That volume is a batch reactor at the worst temperature in the plant, and it is producing acids that will be returned to the loop the next time the valve moves.
3. Heat stable salts: the number that decides the material
Heat stable salts are the salts that a stripper cannot break. The amine is a base, and any stronger acid that enters the loop binds it permanently. Chloride comes in with the flue gas and with cooling water leaks. Sulphate and thiosulphate come from sulphur dioxide that slipped past the flue gas desulphurisation, oxidised in the absorber. Formate, acetate, oxalate and glycolate come from oxidative degradation of the amine itself, catalysed by iron and by oxygen ingress. Together they are measured as a total, in weight percent, and reported alongside a chloride figure, because chloride is the one that decides which alloy is acceptable and the total is the one that decides how fast the whole loop is corroding.
The mechanism is pH. A clean 30 percent MEA solution sits at about 10.5 to 11, and at that pH carbon steel passivates behind an iron carbonate film that is thin, adherent and self repairing. As heat stable salts accumulate, they take amine out of circulation and the pH falls toward 8.5 or 9. Below about 9 the carbonate film stops forming and the steel corrodes as if it were in dilute acid, at rates that climb from a tenth of a millimetre per year to one or two. A 3 millimetre corrosion allowance, which is the normal design for lean amine piping, is consumed in under three years at the top of that range.
Condition | Indicator | What still survives | What does not |
|---|---|---|---|
Clean loop | Total salts under 0.5 percent, chloride under 100 ppm, pH above 10 | Carbon steel with allowance, 316L, any elastomer qualified for amine | Nothing in particular; specify on velocity |
Marginal | Salts 0.5 to 1.0 percent, chloride 100 to 500 ppm | 316L in the cold sections, 2205 on the hot rich side | Carbon steel in the cross exchanger, 304 anywhere |
Aggressive | Salts above 1.0 percent, chloride 500 to 1500 ppm, pH below 9 | 2205 to about 110 degrees C, 2507 above that | 316L in crevices, all carbon steel, most 400 series trim |
Out of control | Chloride above 1500 ppm, or pH below 8 | Superduplex, 6Mo, or alloy 625 as weld overlay | Every stainless grade below PREN 40, in a crevice |
The table is deliberately written as a ladder rather than as a rule, because the practical question a buyer asks is not "what is the corrosion rate" but "which grade do I stop at". The honest answer is that 316L is a sound default for a new unit with clean solvent and becomes marginal at the first chloride excursion, and that the cheapest way to stay on 316L is not metallurgy at all: it is a reclaiming system, a slipstream filter, and the discipline to keep chloride under a few hundred parts per million.
4. Three places the body thins, and none of them is the bore
The inspection finding at the top of this article is not unusual, and the reason is geometric. A crevice needs three things: a gap, an aggressive electrolyte, and a stagnant region where the chemistry can drift away from the bulk. A ball valve provides all three in the same place, twice, and provides none of them in the bore.
The seat pocket is the first. The seat ring sits in a recess with a clearance of a few tenths of a millimetre, sealed by an O ring at the back and open to the cavity at the front. Solvent enters that clearance, oxygen in it is consumed, chloride concentrates, pH drops by two or three units, and the stainless steel at the bottom of the recess becomes an anode with a large cathode around it. The result is pitting at a rate that is unrelated to the general corrosion rate of the same alloy in the same stream, which is exactly why a corrosion coupon in the line reads a healthy 0.02 millimetres per year while the pocket floor is at 1.9.
The cavity floor is the second. Solids settle there: iron sulphide, degradation sludge, filter fines, and the polymer dust that the seat generates as it wipes. A layer of sludge on the bottom of a cavity holds solvent that is not renewed, and under the layer the surface is both starved of oxygen and held at temperature. Cavity drains are often fitted on large valves and just as often never opened.
The stem guide and packing region is the third, and it is the one that produces the surprise failures, because it is above the flow and therefore dry most of the time and wet some of the time. Alternate wetting and drying with chloride present is the classic recipe for chloride stress corrosion cracking in 300 series austenitic steel, and a 316L stem in a hot rich amine valve with 800 ppm chloride is a candidate.
The practical instruction that follows is simple and rarely followed: ultrasonic thickness mapping on an amine valve has to include the pocket floor and the cavity bottom, which means either a profile scan from the outside at the correct angle or an internal inspection at every shutdown. Mapping only the bore produces a report that says the valve is fine, which is what the report said in this case.
5. Velocity is the other half of the number
Corrosion rate in amine service is a function of chemistry and of wall shear stress, and the design guidance that captures the second is a velocity limit that depends on material and on whether the stream is rich or lean. Rich amine is the stricter of the two because it is loaded and because its corrosion product film is weaker. Carbon steel in rich amine is normally limited to about 0.9 to 1.5 metres per second; 316L roughly doubles that; lean amine in carbon steel is allowed 4.5 to 6 metres per second. These are not conservative round numbers, they are the values at which erosion corrosion begins to strip the film faster than it reforms.
A full bore ball valve, fully open, does not violate any of them. That is the trap. The same valve at 20 percent open has a flow area that is a lens between the ball and the seat, and the mean velocity through that lens is ten to twenty times the line velocity. On a 6 inch rich amine line at 3 metres per second, a throttled ball valve produces a jet at 30 to 60 metres per second, and because erosive loss scales with velocity raised to a power between 2.5 and 3, the wear rate at the seat and on the downstream body is two to three orders of magnitude above the pipe. The damage signature is a bright ring on the ball, an oval wash on the body opposite the port, and a seat that looks perfectly machined because it has been polished by the jet.
The conclusion is a division of labour that should be written into the line list. Ball valves in amine service are isolation devices: fully open or fully closed, never in between. Where the process needs a pressure letdown or a flow control function, it gets a different valve with a cage, an angle body or hardened trim, and it gets a material selected for the jet rather than for the pipe.
6. Drop the pressure and the CO2 leaves the liquid
The one place a ball valve is most often proposed, and least suitable, is the rich amine letdown. Rich amine leaves the absorber at 1.2 to 2 bar absolute and has to reach the stripper at a similar pressure after being heated, so on some configurations there is a level control valve that drops several bar across a single seat. What happens there is not a liquid flow.
A 30 percent MEA solution at a loading of 0.5 holds about 0.99 kilomoles of CO2 per cubic metre for every 0.2 of loading change, which is 43 kilograms. Released at 2 bar and 50 degrees C that is around 13 cubic metres of gas per cubic metre of liquid, so a letdown that liberates even 0.2 of loading converts a cubic metre of solvent into thirteen cubic metres of two phase mixture. The mixture accelerates through the seat at 20 to 40 metres per second, the CO2 comes out of solution partly as foam rather than as bubbles, and the trim sees a flashing, cavitating, foam-forming stream that no ball valve seat is designed for. The failure mode is not a leak, it is a seat ring eroded out of square in a few thousand hours, followed by a leak.
The same physics explains why the cavity is a pressure vessel on this service. Trapped rich amine at 100 degrees C continues to desorb carbon dioxide until the liquid's equilibrium partial pressure is reached, so the cavity pressure rises by the CO2 partial pressure of the solvent at that temperature, which for rich amine at 100 to 110 degrees C is several bar, and it does so slowly, over hours, without any thermal expansion being needed. A seat with a self relieving design, or a ball with a vent hole on the upstream side, gives that gas a route back to the line. A double piston seat arrangement with a cavity that has no relief does not, and the result is a body joint that weeps after a hot shutdown, or a seat that is extruded into its own clearance.
7. The elastomer will fail first, and it will fail quietly
Fluorocarbon elastomer, the material most purchasing departments mean when they write FKM or Viton, is a poor choice for amine service and is nevertheless found in amine valves constantly, because it is the default for hydrocarbons. Amines attack the cure system rather than the polymer backbone: the material hardens by 10 to 20 points on the Shore A scale, takes a permanent set, loses its compression recovery, and then cracks at the point where it is doing the most work. The failure is not a sudden blowout. It is a stem seal that starts to weep at a rate that a operator wipes up with a rag for six months.
Elastomer | Behaviour in 30 percent MEA at 100 degrees C | Where it is used | Watch out for |
|---|---|---|---|
FKM, fluorocarbon | Hardens, takes set, cracks in months | Only where nothing else fits and the temperature is low | Looks fine on a swell test, fails on compression set |
FEPM, TFE propylene | Good; the standard amine elastomer | Seat O rings, stem seals, body seals | Higher cost, longer lead time, fewer stock sizes |
EPDM | Good in amine, poor in hydrocarbons | Lean amine, water side, utility connections | Swells badly if any oil contaminates the loop |
PTFE and modified PTFE | Chemically inert, mechanically creeps above 150 degrees C | Seat inserts, chevron packing, gaskets | Cold flow under seat load; needs a backup ring |
PEEK | Inert and dimensionally stable to 250 degrees C | Seats on hot lean amine, bearing rings | Hard, so it scores the ball if any solids are present |
Two details matter more than the table. The first is that a manufacturer's chemical resistance chart, which reports volume swell after 70 hours of immersion at ambient, will pass almost every one of these materials including FKM. The test that predicts field behaviour is compression set after ageing at temperature, and it is worth asking for. The second is that the lubricant inside the valve is itself a contaminant. Silicone grease is not amine stable at these temperatures, and any grease that reaches the solvent is a hydrocarbon that will find the absorber and stabilise foam.
8. Foam, fines, and the loop that hates particles
Amine absorbers foam, and when they foam the solvent leaves the top of the column with the treated gas, the water wash cannot recover it, and the plant loses solvent at a rate that shows up as make up consumption long before it shows up as a trip. The three causes are hydrocarbons above a few hundred parts per million, fine solids, and heat stable salts that act as surfactants. A ball valve contributes to two of the three.
The contribution is mechanistic rather than accidental. Every stroke drags a loaded polymer seat across the ball, and the wear products are polymer particles plus iron oxide from the ball surface, in the size range that stabilises foam best. A valve that strokes twice a day for a year produces grams, which in a loop holding a few hundred cubic metres is enough to matter. Add the grease that was packed at the factory and the hydrocarbon source appears at the same time. This is why amine specifications increasingly call for solvent compatible lubricant, or for a dry assembly, and why the isolation valves should sit upstream of the filters rather than downstream.
Filtration itself deserves a number. A well run capture loop filters a slipstream of 10 to 20 percent of circulation to 10 or 25 micrometres absolute, and the filter catches both the iron sulphide and the polymer. A valve that sheds into the clean side of that filter is shedding into the absorber.
9. What goes on the requisition
A requisition for an amine valve is mostly a matter of replacing three comfortable words with numbers. Instead of "amine service", state the stream, the temperature, the loading, the design chloride and the expected total salt level, because those four decide the alloy. Instead of "stainless steel", state the grade for the body, the ball and the stem separately, since the stem is the part exposed to alternate wetting and the body is the part exposed to a crevice. Instead of "suitable elastomer", state FEPM for all dynamic seals, PTFE or PEEK for seats, and no silicone anywhere in the assembly.
Three further clauses are worth the paper. Specify the seat as self relieving, or the ball as vented, and say which direction the relief has to work, since a rich amine cavity generates gas rather than expanding liquid and needs a route out. Specify that the cavity drains, where fitted, are plug and not needle valves, because a needle drain is never opened and a plug drain sometimes is. And require that the seat pocket and cavity floor are accessible for ultrasonic mapping from outside, which in practice means a drawing with the pocket geometry dimensioned and a note that no weld or boss obscures the pocket floor.
For the material side, J-VALVES supplies amine and capture duty valves in 316L, 2205 and 2507 duplex, superduplex and the nickel alloys, from DN15 up to DN1400, which is 1/2 inch through 56 inch, with body ratings spanning ASME Class 150 to 2500 and a qualified window of minus 196 to plus 650 degrees Celsius. Shell proof stands at 150 percent of the cold rating and the seat is proved at 110 percent, the two held as separate records, and PMI plus 3.1 material certificates travel with the valve as standard rather than as an extra. Send a line list carrying those four stream numbers and the reply, drawn and alloyed with a price against each item, is issued within three working hours.
10. The sentence worth keeping
A ball valve does not fail in amine service because the amine is corrosive. It fails because the design gives the amine three places to stand still and one surface to scrape, and because the chemistry that makes the solvent expensive to replace is the same chemistry that makes the steel thin. Pick the alloy from the salt and chloride numbers rather than from the word amine, keep the valve out of any throttling duty, give the cavity a way to vent the gas it will generate, and put the valve upstream of the filter. The bore will still be bright when the body goes through, and that is the point.
11. Frequently Asked Questions
Why does a 316L ball valve pit in rich amine when the bore looks perfect?
Because the bore is the one wetted surface with flow across it, and pitting needs a stagnant crevice. The damage collects under the seat ring, where a clearance of a few tenths of a millimetre holds solvent that is not renewed, so oxygen is consumed, chloride concentrates and the pH falls by two or three units. That pocket becomes an anode against a large cathode and pits at a rate unrelated to the general corrosion rate, which is why a coupon in the line can read 0.02 millimetres per year while the pocket floor is at 1.9. Thickness mapping has to include the pocket floor, not just the bore.
What chloride level forces a move from 316L to duplex in a capture plant?
Roughly 500 ppm in the hot rich amine, and lower if the design has a crevice at that temperature. Chloride is the number that selects the alloy, while total heat stable salts decide how fast the whole loop corrodes. Below 100 ppm with total salts under 0.5 percent, 316L is comfortable. Between 100 and 500 ppm 316L survives in the cold sections but becomes marginal in the cross exchanger, where 2205 is the usual answer. Above 500 ppm at 110 degrees C, superduplex or a nickel alloy overlay is the honest choice, because 2205 is at its stress corrosion limit rather than inside it.
Can a ball valve be used for rich amine letdown?
It can be installed, and it should not be asked to control. Dropping 0.2 of loading releases about 13 cubic metres of carbon dioxide per cubic metre of solvent at 2 bar, so the trim sees a flashing two phase mixture at 20 to 40 metres per second. A ball valve seat in that duty erodes out of square in a few thousand hours and then leaks. Use a cage or angle valve with hardened trim for the letdown, and keep the ball valve as an isolation device that is either fully open or fully closed.
Which elastomer should be specified for amine service?
FEPM, the TFE propylene family, for every dynamic seal, with PTFE or PEEK for seats and no silicone grease anywhere in the assembly. Fluorocarbon elastomer is the usual default and the usual failure: amines attack its cure system, so it hardens by 10 to 20 Shore A points, takes a permanent set and then cracks, which shows up as a stem seal that weeps for months before anyone treats it as a fault. A 70 hour ambient swell test will pass FKM, so ask for compression set after ageing at temperature instead.
Does NACE MR0175 apply to a carbon capture amine loop?
Not to a sweet flue gas case, because the standard is triggered by hydrogen sulphide partial pressure and a gas fired boiler flue has none. It does apply where the capture unit takes gas from a treating plant or a refinery stream that carries H2S, and then the hardness limits, the 22 HRC ceiling and the qualification route all come into force for the wetted metallic parts. The distinction matters at the purchase order stage, because a valve ordered to MR0175 costs appreciably more than the same valve ordered without it, and the decision belongs to the feed gas analysis rather than to the valve vendor.
