Article Content
A production tank does not treat hydrogen sulfide. It stores it, partitions it, and then releases it at the worst possible moment: when a hatch is opened, a truck is loaded, or a hot afternoon lifts the vapor pressure. Pipeline scavenger programs and gas-plant amine units get the engineering attention. The battery often gets a detector tube and a hope that yesterday’s assay still applies. That is how sour-service exposure events and failed shipping specifications keep showing up on tanks that were supposed to be downstream of treating.
Fixed storage is a different problem from a rail car or a marine cargo tank. Residence time is longer, the water leg is persistent, blanketing gas may itself be sour, and the same tank is filled, drawn down, and gauged for months. The chemistry is ordinary. The operating pattern is what makes the H2S number move.
Where the Sulfide Enters the Tank
Three streams put H2S into a crude tank, and they are easy to mix up in a sample log.
The first is the oil itself. Produced crude carries dissolved H2S from the reservoir, from a sour treater overhead that was not fully condensed and returned, or from an upstream chemical program that under-dosed. Once the oil is in the tank, temperature and agitation decide how much of that dissolved load reports to the vapor space. A cold morning assay and a hot afternoon hatch reading are not the same number.
The second is associated gas and blanket gas. If the tank is gas-blanketed with fuel gas, solution gas, or VRU recycle, the blanketing stream sets a floor on the headspace concentration. Sweet oil under a sour blanket will sour. Sour oil under nitrogen will still off-gas until the liquid and vapor re-equilibrate.
The third is generation in the tank. The basic sediment and water layer is anaerobic, sulfate is usually present, and retention of days to weeks is enough for sulfate-reducing bacteria. A tank that receives sweet oil can still develop a sour water leg and a sour interface. Killing the bacteria does not remove H2S already formed. It only slows the next batch.
Why the Hatch Reading Lies
Operators often treat the thief-hatch tube reading as the tank specification. It is a vapor measurement at one point in time. Liquid-phase H2S, the number that matters for a pipeline or a truck-loading limit, can be higher or lower depending on how recently the tank was agitated and how warm the oil is.
Henry’s law is the useful picture. Dissolved H2S in the oil is in equilibrium with H2S partial pressure in the headspace. Heat the oil and more H2S leaves the liquid. Pump in a fresh volume and the interface is renewed, so the headspace spikes even if the incoming oil assay was unchanged. Open a hatch on a tank that has been sitting static and the first reading can be high because vapor has accumulated, then fall as air dilutes it. None of those readings is the dissolved load.
A useful pair of numbers is liquid-phase H2S on a closed sample, plus headspace H2S at the vent or hatch under normal operating level. Shipping and transfer decisions should use the liquid number. Exposure decisions should use the headspace number, taken before the hatch has been open long enough to dilute it.
| Driver | What it does | Typical result |
| Temperature rise | Shifts dissolved H2S into the vapor space | Higher hatch and vent readings on hot days |
| Level change and pumping | Agitates the oil and refreshes the interface | Spikes during rundown and truck loading |
| Blanketing gas quality | Adds or dilutes H2S in the headspace | Fuel gas blanketing can sour a sweet tank |
| Retention time | Gives sulfate-reducing bacteria time to work in the water leg | BS&W layer becomes a sulfide source |
| Pressure and hatch leaks | Releases vapor that was held in solution | Odor and exposure at the thief hatch |
Exposure at the Battery
Tank batteries concentrate the hazard in a few routine jobs. Gauging, sampling, truck loading, and water draws all break the vapor seal. H2S is heavier than air and collects at the base of the tank, in the containment, and in the loading shed if there is one. A personal monitor at chest height can read clear while the hatch and the ground-level vent do not.
The practical controls are ordinary and still skipped. Crack the hatch and test before the head goes over it. Keep the loading connection closed until the vapor hose is made up. Treat the water draw as a sour stream, not as a utility drain. If the battery has a vapor recovery unit, confirm it is actually pulling during loading. A VRU that is down turns every loading event into an open vent.
Treating the Oil Versus Treating the Vapor
These are separate jobs. A scavenger mixed into the crude lowers dissolved H2S and, after re-equilibration, lowers the headspace. It does not instantly clear vapor already in the tank. A scrubber on the vent or on the VRU inlet lowers what leaves the hatch and the stack. It does not fix a truck that is about to haul sour oil down the road.
Oil-soluble scavengers are the usual choice in the crude phase. Triazine can work when a water phase is present to carry it, but a dry or low-water-cut tank often needs an oil-soluble product so the reaction happens in the oil rather than in a rag layer. Batch treating is appropriate when a tank is being prepared for shipment: dose, circulate or roll the tank, then re-sample the liquid before the truck is called. Continuous injection upstream of the tank is better when rundown is steady and the target is to keep the tank from going sour in the first place.
Contact is the failure mode. A tote pumped into the hatch of a static tank treats the surface. Without circulation, the bottom of a 400-barrel tank is still sour when the top sample looks clean. A circulating pump, a mixer, or injection into the rundown line ahead of the tank is the difference between a passed assay and a failed load.
Vapor treating belongs on the common vent header, the VRU suction, or a dedicated loading vent, not as an afterthought at a single thief hatch. A small caustic, oxidant, or triazine scrubber, or a polishing adsorbent vessel, can hold the stack and the loading hose in specification. Size it on the displacement volume during truck loading, not on the quiet breathing rate of a static tank. Loading is the peak.
Blanketing and Vapor Recovery
Nitrogen blanketing stops oxygen ingress and slows iron sulfide formation on the shell, but it does not remove H2S. It dilutes it only to the extent that blanket gas is added and vapor is displaced to a treated vent. Fuel-gas blanketing is cheaper and often sour. If the blanket is richer in H2S than the equilibrium headspace, the blanket becomes a source.
Vapor recovery is the right destination for displaced gas, provided the compressor and the downstream treating can take the H2S. A VRU that discharges to a flare without a scrubber has only moved the emission. A VRU that recycles sour gas back onto the tanks has created a loop. The scrubber belongs on the VRU discharge to flare or fuel, or on the common tank vent upstream of the compressor if the machine metallurgy is not sour-rated.
The Water Leg
BS&W is where tanks go sour after the oil has been treated. Sulfate-reducing bacteria in the water layer produce sulfide that partitions back into the oil and the headspace. A monthly oil assay can look stable while the interface gets worse, then a water draw or a mix before shipment releases it.
Draw the water on a schedule instead of letting it accumulate for a turnaround. If the water is sour, scavenge or oxidize it before it goes to a pond or a disposal tank that is not rated for the vapor. Biocide in the water leg slows regrowth. It is not a substitute for removing sulfide already there. Iron sulfide on the floor and the shell is both a corrosion product and a reservoir: acid or a biocide program that drops the pH can redissolve it and spike the headspace.
| Method | Best use | What it does not fix | Waste or side effect |
| Oil-soluble scavenger, batch or continuous | Dissolved H2S in the crude | Vapor already in the headspace | Spent chemical in the oil or water draw |
| Vapor scrubber on the vent or VRU inlet | Hatch, vent, and loading vapors | H2S still dissolved in the oil | Spent scrubber liquid or adsorbent |
| Nitrogen or sweet-gas blanket | Oxygen exclusion and dilution | A large dissolved load | Blanket gas consumption |
| Mix and circulate after a batch dose | Poor contact in a static tank | Channeling if the mixer is undersized | Temporary vapor release while mixing |
| Water-leg scavenger or biocide | SRB in BS&W | H2S already in the oil | Water-treating chemical and solids |
Sampling That Matches the Decision
Use a closed liquid sample for the custody or shipping number. A bottle filled at the hatch, left with a headspace, and driven to town will lose H2S before it is titrated. Fill it full, cap it, and run it the same day, or use a field method on site and record the temperature.
Use a detector tube or a meter on the hatch only after the reading has stabilized and before anyone leans in. Note the tank level, the oil temperature, and whether a truck was loading. A log that says “12 ppm” without those three notes cannot be compared with next week’s number.
Two failures show up in audits. Sampling only the top after a surface batch dose, and sampling the vapor and calling it the oil specification. Either one ships a sour load or shuts in a tank that was already in spec.
A Battery Layout That Holds Specification
The sequence that works on a Western Canadian or similar production battery is short.
- Scavenge in the rundown line if the incoming oil is chronically sour, so the tank is not the primary reactor.
- Blanket with nitrogen or confirmed-sweet gas, and route displaced vapor to a scrubber or a treated VRU.
- Draw the water leg before it becomes a sulfide generator, and treat that water if the pond or disposal tank is shared.
- Batch-treat and circulate any tank that must meet a truck or pipeline liquid limit, then re-sample the liquid, not the hatch.
- Test the hatch before gauging, and treat loading vapor as the design case for the scrubber.
That layout uses a metering pump, a circulating pump, and a small vent scrubber. It does not need a gas plant. It does need the liquid number and the vapor number kept separate, because they answer different questions and they do not move together.
What to Set Before the Next Load
Three decisions prevent most tank-battery H2S surprises. What liquid-phase limit the truck or the pipeline actually enforces. Whether the blanket gas is sweeter than the oil. And whether anyone circulates after a batch dose. A scavenger program that cannot answer those three is a chemical cost, not a specification.
Hydrogen sulfide in a crude tank is a partitioning and contact problem. Treat the oil for the load that will be shipped, treat the vent for the vapor that will be released, and do not use a hatch tube to stand in for either one.
