Every cycle, the best feasible move.
Advanced multivariable controllers that use dynamic process models to predict future behaviour and optimise every control move — within your operating constraints.
How Model Predictive Control works for your plant.
Model Predictive Control uses a dynamic model of your process to predict how it will respond over the coming minutes to hours. Each cycle it computes the optimal set of control moves that respects every constraint — then does it again, continuously pushing the unit toward its best feasible operating point.
Watch it look ahead.
The dashed line is the controller's prediction for the next horizon; the bars are the moves it plans. Tighten the high limit, ask for a setpoint near it, or kick in a disturbance — each cycle it re-solves for the best move that keeps the prediction inside the limits.
- Optimises over a prediction horizon, re-solved every cycle
- Constraints on the output and on the move size
- Move suppression trades speed for smoothness
Dynamic behaviour captured from step tests and historian data.
Multivariable dynamic model built and reviewed.
Controller tuned with constraints and operator limits.
Closed-loop moves computed every cycle.
A PID loop reacts. MPC plans.
| PID loop | Model predictive control | |
|---|---|---|
| Sees | Error now | Predicted trajectory over the horizon |
| Variables | One in, one out | Many MVs and CVs, coupled |
| Constraints | Clamped after the fact | Respected in the optimisation |
| Dead time | Detunes the loop | Built into the model |
| Objective | Hold the setpoint | Push to the best feasible point |
“MPC does not replace your operators. It gives them a unit that holds steady while they run the plant.”
From the field · SMHcoders
Does MPC replace our DCS?
No. MPC sits above your existing control layer and writes setpoints to it. Operators keep full authority, including one-click handback to manual.
What data do we need to start?
Most engagements start with 6–24 months of historian, lab or document data. We audit coverage in the first week and tell you plainly if it is not enough — before you commit.
How long until we see results?
A proof of concept on your historical data typically lands in 4–8 weeks. Production deployment follows once the KPI target is met and your team signs off.
Where does it run, and who sees our data?
On your infrastructure — plant-side servers, your Azure or AWS tenancy, or an air-gapped network. Data never leaves your boundary, and nothing is used to train third-party models.
Who owns the result?
You do. Handover includes source code, documentation, training sessions and a retraining plan.