Tailoring MPCPlus Control Performance

DeltaV PredictPro application.

Refer to Tailoring MPC Control Performance.

Tuning dialog in the MPC Operate Pro application.

Refer to Tailoring MPC Control Performance for guidelines on setting the value.

Tuning dialog in the MPC Operate Pro application.

The value of the dynamic POE that the block calculates for CVs with a Target optimization type is based mostly on the configured/tuned value of POE_MID. Therefore, when the optimization type is Target, a good starting point for all three POE values (POE_ABOVE, POE_MID, and POE_BELOW) is 1.0. Use the response trajectory filter for the CV or the POM values of manipulated variables to tune the control response for target CVs. In general, if it is necessary to tune the POE values, adjust all three POE values equally. It may be appropriate for the POE_MID value to be lower than POE_ABOVE and POE_BELOW, but keep in mind that the value of POE_MID dominates.

For a constraint variable with an optimization type of None, a good starting point for POE_MID is 0.1 or lower, and 1.0 for POE_ABOVE and POE_BELOW.

Refer to Tailoring MPC Control Performance for basic guidelines on setting the error penalty.

Detail display for controlled variables in the MPC Operate Pro application.

Refer to Tailoring MPC Control Performance.

Detail display for manipulated variables in the MPC Operate Pro application.

Refer to Tailoring MPC Control Performance.

Refer to Tailoring MPC Control Performance.

The horizontal slider on the Tuning dialog in the MPC Operator Pro application.

The single slider simultaneously adjusts the target setpoint filter for all process outputs in the range of 0.1 (Robustness) to 0.8 (Performance). The larger the value, the faster the optimizer's optimal setpoint value is applied to the working setpoint of controlled variables.

The OPT_MV_WT parameter.

The parameter holds a value common to all manipulated variables. The default value of 1.0 results in the MV move plan being strongly influenced by the steady-state MV values determined by the optimizer. A value of 0.0 provides a light influence; a value of 0.1 provides a moderate influence. Control performance is subtly impacted by the value. A value of 0.0 provides slightly less integrated error when working setpoints of controlled variables remain at their setpoint (no constraints are being hit). A value of 1.0 provides slightly less integrated error when working setpoints tend to deviate from setpoint during a dynamic response (constraints are being hit).

Tuning dialog in the MPC Operate Pro application.

In the MPCPlus block, a CV with an economic objective (Max or Min) is given a lower preference by the optimizer than a CV with a target objective. If you want a Max or Min CV to have a higher preference than a Target CV, change the Target CV to a PSV CV by setting the optimization type to PSV. A PSV CV behaves the same as a Target CV with the exception that a PSV CV will relax its working setpoint if needed for an economic objective to be met when there are insufficient degrees of freedom.

Tuning dialog in the MPC Operate Pro application.

The optimizer uses the Priority value to differentiate preference when the solution is infeasible. If it becomes necessary for steady-state working setpoints to deviate from CV setpoints, the optimizer will relax variables in priority order, that is, start with the variables with the lowest priority (highest Priority value). Each process output has a priority setting because all have the potential to become constrained. The priority setting will not override the block's standard preference based on the optimization type, but will differentiate among outputs with the same optimization type. Gain ratios can have an influence on preference, so you may need to increase the spread between Priority values to achieve the desired result.

Tuning dialog in the MPC Operate Pro application.

The Value per % setting applies to economic variables and is used by the optimizer to essentially differentiate preference among economic variables. A higher value gives more weight to achieving the economic objective. The priority setting for an economic variable can have a supplemental influence when multiple economic objectives have already been achieved and there are insufficient degrees of freedom.

An expression in a CALC or Action block or an Action in a phase Sequential Function Chart (SFC).

The instruction in the following expression sets the gain multiplier of the transfer function for control/constraint variable 2, manipulated/disturbance variable 4 in the MPC-PLUS1 block of module MPCP-8CON to the value of the parameter named GAIN_C2_M4:

If the transfer function gain is set to 0.6 in PredictPro and the MODEL_GAIN[2][4] is 0.25, the actual gain becomes 0.15.