An acid gas removal unit (AGRU) protects the LNG train from CO₂ and H₂S. Running it conservatively keeps treated gas on specification, but it also raises amine circulation and regeneration duty every hour the train runs. When CO₂ slip rises, the cryogenic equipment and the whole plant's throughput come under pressure. The IEA expects around 300 bcm per year of new LNG export capacity to be added by 2030, so more amine treating systems are entering service while existing facilities face tighter margins and more variable feed gas.
The gap between where most AGRUs operate and where they could operate has less to do with absorber capacity than with how regeneration, circulation and feed variability are managed together.
How the amine loop works
Sour feed gas enters the bottom of the absorber and flows up against lean amine coming down the column. Treated gas leaves the top and goes on to dehydration and liquefaction. Rich amine leaves the bottom, is flashed to release dissolved hydrocarbons, picks up heat in the lean/rich exchanger and enters the regenerator. There, the steam-heated reboiler boils up stripping vapor that drives the CO₂ and H₂S out of solution, and the acid gas leaves overhead. Hot lean amine returns through the lean/rich exchanger and a cooler to the top of the absorber.
Lean loading, the acid gas left on the regenerated amine, sets how much more acid gas the absorber can pick up. Circulation rate sets how much solvent the regenerator has to strip. A lower lean loading and a higher circulation rate both cost reboiler steam.
Why regeneration economics drive AGRU performance
Operations teams naturally focus on the absorber, where treated gas quality is measured and foaming is most visible. The economic burden sits mostly in solvent regeneration.
The reboiler is the largest energy consumer and the hottest point in the amine circuit. It can also be a throughput constraint. When reboiler duty is too low, lean loading climbs. CO₂ slip to treated gas follows, and solvent condition can deteriorate over time.
In LNG service the consequences chain together. CO₂ that slips through can freeze in the main cryogenic heat exchanger, so poor regeneration turns the AGRU into a plant-wide constraint. Operators may have to cut train throughput to protect the exchanger. At that point the cost of conservative regeneration includes lost production as well as energy.
Amine circulation is one of the most important economic variables in acid gas treating. Higher circulation means more solvent to regenerate, more reboiler steam and more cooling duty. A small, persistent overshoot in circulation adds up to a measurable energy loss over a full operating year.
How feed variability compounds the problem
Feed gas composition changes continuously. U.S. Gulf Coast terminals draw gas from several producing basins, each with its own CO₂ and H₂S profile, and those profiles shift over the life of supply contracts.
New trains that meet unexpected acid gas composition during startup or ramp-up can face capacity limits and foaming. A foaming event can force a rapid throughput cut and an extended recovery that affects production schedules across the facility.
Operators usually respond by holding high lean amine flow as a buffer against feed changes. The extra flow protects against off-spec gas, but it makes the regenerator run harder. Some facilities hold treated gas well below the specification limit to absorb sudden acid gas spikes, and the margin between actual CO₂ content and the limit can be several times what current conditions need.
That conservatism reflects the tools available. When confirmation of lean amine quality arrives hours after conditions change, extra margin is the rational choice. It still has a cost: circulation and regeneration duty stay above what the process needs until someone readjusts them by hand.
Why conventional control falls short in amine systems
Most large LNG facilities run advanced process control (APC), but the AGRU exposes limits that standard multivariable controllers struggle with.
Linear models on a nonlinear unit
Standard model predictive control relies on linear process models, and amine systems are strongly nonlinear across their operating envelope. MDEA selectivity for H₂S over CO₂ can fall away at high pressure, high temperature and highly sour conditions. Across that part of the range the controller model may be wrong.
Late measurement of lean loading
Many plants still rely on laboratory analysis to confirm lean loading, the main indicator of regeneration quality. By the time degraded lean amine is confirmed, the absorber may have produced marginal gas for much of a shift. Good historian practice narrows the gap, but inferring lean loading in real time from the available measurements is still hard with conventional tools.
Solvent condition that drifts
Real amine systems accumulate heat stable salts, degradation products and hydrocarbon contamination. These change absorption kinetics and reduce effective capacity. A controller built on steady-state assumptions cannot track those shifts reliably. Engineers re-tune or update the models periodically, and between updates the controller drifts further from how the unit behaves.
Because the AGRU feeds dehydration and liquefaction directly, a fixed treated gas specification may also stop reflecting the best operating point when downstream conditions change.
What a learned model changes in AGRU operation
Instead of setting reboiler steam for worst-case conditions, a model learned from the unit's own operating data can target the minimum duty that holds product specifications under current conditions. It learns the nonlinear relationships between circulation rate, reboiler duty, lean loading and treated gas quality across the range of feed compositions and solvent conditions the plant has seen.
Treating feed gas CO₂ as a disturbance variable lets the controller adjust circulation and regeneration before the disturbance works its way through the absorber. That anticipation matters with variable feed. By the time a conventional controller sees the change in treated gas quality, the absorber has already been running at the wrong rate.
The same model can infer lean loading from available measurements between lab samples. As solvent condition shifts through normal operation, the model is updated from recent plant data rather than re-tuned by hand. When one model covers the AGRU and the downstream units, setpoint decisions can account for how regeneration affects dehydration performance and liquefaction stability.
Keeping operator knowledge across shifts
A model does not capture every instinct behind a veteran operator's judgment. It can preserve the observable relationships between process states and the actions that produced good outcomes. That is most useful at shift change, when the context built up by the operator leaving does not fully transfer in a handover.
Building operator trust in steps
Most deployments begin in advisory mode. The model recommends setpoint moves and operators keep full control authority. Engineers and board operators compare the recommendations with their own judgment and with how the unit responds. Trust comes from repeated exposure to plant-specific decisions over many shifts.
Advisory mode also changes how experienced and newer operators work together. Senior operators can check whether the recommendations follow the logic they learned through years of upset handling. Newer operators get a clearer reference for a good response under changing conditions.
The next step is closed loop. The controller writes setpoints to the APC or DCS and operators can take over from the board. Teams set the operating envelope the controller may use and widen it as their confidence in the model grows.
One view of the AGRU for operations, engineering and planning
When operations, process engineering and planning share one model of AGRU behavior, they can settle the energy and throughput trade-off on evidence:
- Planning can see when a conservative treated gas target is consuming regeneration duty it does not need.
- Process engineering can test whether a suspected regeneration bottleneck is real, or whether operators are compensating for a different constraint.
- Operations can weigh the cost of extra circulation against the current risk to downstream stability.
That shared view speeds agreement on what the unit is constrained by at any moment. It matters more as gas processing plants face tighter margins and more variable feed.
How the Imubit Platform applies
On an AGRU, the Imubit Platform follows the same Map, Model, Control approach it applies to other units in refining and gas processing. Map traces the energy and throughput lost in the treating section to the recurring decisions behind it, such as circulation and reboiler targets. Model learns the unit's nonlinear behavior from its own history with Deep Learning Process Control® (DLPC®). Control takes those decisions to closed loop, inside limits set by plant engineers, writing setpoints to the APC and DCS already in place.
