In LNG production, the amine unit decides whether gas moves on to liquefaction. Treated gas must stay below 50 ppmv CO₂ before it reaches the cryogenic section, because CO₂ above that level can freeze in the main cryogenic heat exchanger (MCHE) and damage it and other equipment. The IEA's World Energy Outlook 2024 projects LNG capacity expansion through 2030, so more pretreatment capacity will have to run inside the same tight limits.
When foaming, corrosion or solvent contamination erodes that margin, a treating problem quickly becomes a train availability problem. The amine unit sets the pace for the whole train. For the regeneration economics of the acid gas removal unit as a whole, see acid gas removal optimization.
How the treating cycle works in LNG service
Sour feed gas enters the bottom of the absorber in the acid gas removal unit (AGRU) and flows up against descending lean amine. Countercurrent contact transfers CO₂ and H₂S from the gas into the solvent. Gas sweetened to a pipeline specification has more room. In LNG service the treated gas target is absolute, with almost no margin for exceedance.
Rich amine leaves the absorber, passes through the lean/rich heat exchanger and enters the regenerator, where steam from the reboiler strips the absorbed gases. The lean/rich exchanger recovers heat that would otherwise add to reboiler demand, which makes it the most important energy recovery point in the loop. Regenerated lean amine is cooled and returns to the absorber.
Promoted MDEA
Solvent choice in LNG service follows from the CO₂ specification. Pure methyldiethanolamine (MDEA) reacts too slowly with CO₂ to reach the target within typical absorber contact time, so promoted MDEA, often activated with a faster-reacting component such as piperazine, is common.
The promoter concentration becomes an operating variable. With too little promoter, absorption slows. With too much, regeneration energy rises, because the faster-reacting component forms a more stable bond with CO₂.
Regenerator performance
Regenerator performance depends on reboiler temperature, reflux and stripping steam rate. If the regenerator does not strip enough acid gas, lean loading rises and the absorber loses capacity on the next pass. In tropical LNG plants, day-to-night ambient swings can change lean amine cooling, and with it absorber performance, within a single shift.
Contamination control
Inlet filter-coalescers upstream of the absorber catch contaminants before they reach the solvent. With this little operating margin, contamination control affects both solvent condition and treating stability. Plants working toward self-optimizing gas processing treat filtration and solvent management as optimization variables as well as maintenance tasks.
Where amine systems put train availability at risk
Foaming, corrosion and heat stable salt accumulation are the main amine system threats to LNG train availability, and they reinforce each other.
- Foaming. Liquid hydrocarbons, upstream production chemicals or degradation products lower solvent surface tension and stabilize gas bubbles across the contactor trays. Minor foaming may be manageable in pipeline service. In LNG service even a small event can breach the treated gas limit and block feed to the liquefaction train.
- Corrosion. Corrosion produces iron sulfide particles that circulate through the system, foul equipment and stabilize foam, as Hydrocarbon Engineering describes in its review of amine filtration. That accelerates further corrosion and adds to regeneration energy demand across the loop.
- Heat stable salts. Acids stronger than H₂S or CO₂ react with the amine to form salts that regeneration cannot reverse. As they accumulate, they reduce absorption capacity, accelerate corrosion and promote foaming.
Recovery takes longer in LNG service than in a conventional gas plant. When treated gas goes off-spec, feed to the MCHE stops. A long disruption can extend the restart and cooldown sequence by hours or days, depending on how far temperatures move during the event. A solvent contamination problem then becomes a downtime event that affects production schedules and cargo commitments.
Because the three failure modes feed each other, breaking the cycle means acting on several variables at once instead of responding to the most visible symptom.
Why conventional control struggles with amine systems
Conventional advanced process control (APC) delivers benefits on many process units. In amine treating, the constraints themselves move.
Nonlinear chemistry
Amine absorption is ionic chemistry with strongly nonlinear response surfaces. Temperature, acid gas loading, amine concentration and partial pressures interact in ways that linear dynamic models cannot represent across a realistic operating range. Feed composition adds to the problem, because a change in the H₂S/CO₂ ratio alters the equilibrium the design-case model assumed.
Late feedback
Confirming solvent quality and heat stable salt formation in the lab takes hours. In between, the unit is effectively open loop on the variables that set treated gas quality and corrosion risk. Trending solvent condition indicators in the historian between lab results narrows the gap, but the lag remains.
Eroding confidence
Operators are left choosing between conservative constraints that cost energy and throughput and a wider envelope that accepts more risk. Neither is satisfactory on a unit already running near its treated gas limit. Over time the gap between model predictions and actual behavior widens. Operators lose confidence, engineers stop updating models, and the optimization tool sits unused or runs on outdated assumptions. Amine systems change faster than static models can follow, especially where feed varies with upstream conditioning.
Where the recoverable energy is
Over-circulation
One of the most useful improvements needs no capital: correcting over-circulation. Reboiler heat duty rises with solvent circulation rate. When circulation exceeds what the actual gas composition and loading require, the unit pays for that extra margin in steam.
Operators often run higher rates as insurance against off-spec gas, especially after a foaming event or a feed change. Correcting it needs real-time visibility into the relationship between actual acid gas loading and circulation rate. It is often the quickest energy efficiency improvement available on an amine unit.
Solvent management
- Filtration that reduces total suspended solids removes the particles that stabilize foam.
- Heat stable salts need active management to hold their concentration within a range.
- Holding lean amine temperature above inlet gas temperature prevents the hydrocarbon condensation that triggers foaming.
These levers interact. Higher filtration rates reduce foam risk but do nothing for heat stable salts, while aggressive salt removal by reclaiming can create solvent balance problems of its own if it is not managed carefully.
Learning the unit's nonlinear behavior
Each lever depends on knowing the unit's current state in near real time. A model trained on the plant's own operating data can learn the nonlinear relationships that static models were not built to capture, including the interactions among circulation rate, solvent loading, feed composition and ambient temperature. The objective is lower solvent, steam and power consumption with treated gas held on spec as conditions shift.
Building operator trust
Deployments that last start in advisory mode. The model recommends adjustments, and experienced operators check each recommendation against their own judgment before any setpoint moves. They see the predicted effect on treated gas quality, reboiler duty and solvent health beside the model's reasoning, and they can accept, modify or reject it. That phase also builds process engineering capability on site. Plants then move to operator-approved operation, and on to closed loop when the team is comfortable.
When maintenance, operations and planning share one model of amine unit behavior, decisions on circulation rate and solvent health reflect how each adjustment affects treating performance, energy cost and equipment health together. That matters when one group is focused on fouling, another on train rate and another on utility consumption.
No model captures every instinct behind a veteran operator's judgment. The observable relationships between process states and the actions that produced good outcomes can be preserved in a model that is kept current as conditions change.
How the Imubit Platform applies
Imubit applies Map, Model and Control to amine treating as part of its work in gas processing and LNG. Map traces lost energy and train availability to the recurring decisions behind them, such as circulation rate, reboiler duty and lean amine temperature. Model learns the unit's nonlinear behavior from its own history, including absorber loading, regenerator performance, solvent quality and ambient conditions, with Deep Learning Process Control® (DLPC®). Control takes those decisions to closed loop, writing setpoints to the existing APC and DCS, inside limits set by plant engineers and leaving regulatory loops and base-layer controls in place.
