Predictive control library PCR

PCR is a set of model predictive control blocks
 
TARGETS
The systems which get benefit of being controlled by PCR are those which are usually badly handled by controllers like PIDs.

It is mainly the case when :
  • the time delay is significant
  • the system is integrator (such as levels)
  • disturbance variables are measured
Chemical reactors are a specific target of PCR since that product had been designed originally to control the reactor temperature. The blocks used for that purpose fit with the most common configurations of heat exchangers met on these units.


On the other hand, PCR is well adapted for any other kind of process :
  • whose behaviour must be improved : better stability, reduction of standard deviation (quality control, temperature, humidity control, concentration,...)
  • on which the movements of the actuators should be minimized
     
 
 
 
 
Compared to PID control, PCR shows the predictive control advantages for the instrumentist advantages for the instrumentist PCR IS profitable where the PID is known to be poor.
CLICK FOR MORE DETAILS  and advantages for the producer advantages for the producer Stability of the unit and reduction of the quality variations.
CLICK FOR MORE DETAILS .

 
EXAMPLES OF APPLICATIONS
 
Control of furnace outlet humidity

The process variable to be controlled is the humidity of the product out of the furnace. The manipulated variable is the set point of the air temperature close to the burners.

That temperature is controlled by the existing PID controllers.
Plant tests : square waves applied to the set point of the air temperature

Measurements : temperature set point, temperature and humidity.

The model represents the link between the temperature set point and the measured humidity.

The PCR controller reveives the measured temperature and the measured humidity and computes the temperature set point to be applied to the PID controller. 
 
 
Temperature control of a jacketed reactor

The process variable to be controlled is the reactor mass temperature.

The manipulated variables are :
  • the steam valve for heating the thermofluid in the jacket
  • the cold water valve for cooling
The intermediate temperature is the thermofluid temperature (jacket inlet).

The PCR library contains a block which helps define a set point profile which will be satisfied by the control blocks.
 
PCR assets can solve the difficulties :
  • the dynamic linking the jacket temperature to the reactor temperature varies with the level of product in the reactor.
  • the transfers for heating and cooling are quite different and require separated models.
 
 
Other applications : reactor with external heat exchanger reactor with external heat exchanger Control of the reactor temperature by a cascade of 3 PCR blocks.
CLICK FOR MORE DETAILS and bottom level control of column
reactor with external heat exchanger reactor with external heat exchanger Control of the reactor temperature by a cascade of 3 PCR blocks.
CLICK FOR MORE DETAILS  and Column bottom level control Column bottom level control The level is much disturbed by the bottom outlet flow rate.
CLICK FOR MORE DETAILS
 

Download  : BASF REACTOR (184 Ko) - REPSOL reactor (83 Ko) - Food dryer (164 Ko)  - Steam generator (113 Ko) - Tower level YPF-REPSOL (92 Ko)

 
 

INTEGRATION


The PCR library is designed for being embedded in PLCs and DCS boards. It is integrated with success by end-users in control systems such as Schneider Electric, OMRON, Honeywell, Siemens, etc.

The PCR3 version is embedded in the UnityPro et CONCEPT environnements (QUANTUM, MOMENTUM & PREMIUM) by Sherpa and Schneider Electric. The PCR3 library is available from Schneider Electric and Sherpa Engineering.

 

CAD tools for model identification and control design

The model is so simple and the controller having almost no tuning parameters are in favor of the application of most of the PCR modules without the need of complementary design tools.

Anyway, a set of CAD tools makes it easy to build more complex control architectures when necessary (split range, cascade, tracking of set point profile).

 

MODEL IDENTIFICATION

The control blocks are based on a model which represents the dynamic relationships between process inputs and outputs. This model is to be built before the controller design.

If the model parameters cannot be evaluated from plotted responses, they can be estimated by an identification algorithm from plant tests collected data.

The IDENTIFICATION part of the CAD is directly integrated in the EXCEL file which contains the collected data. This avoids the boring transfer of data between different applications.

The results of the identification module (model parameters) are stored in a file for the controller design and closed loop testing on simulator.

 

CONTROL DESIGN and TESTING

This part of the CAD tools is to prepare each block of the full control architecture. 

 
 The user has access to all the test conditions (set point chnages, AUTO/MANU mode, adding noise and disturbances) from the closed loop block diagramme.

The signals are plotted in the real time of the simulator for evaluating the controller performances.

 

 

In case of a control structure including several control blocks, the CAD proposes several typical architectures in order to build and test the full control structure.

The example opposite is a set of two cascaded controllers whose design and testing can be performed globally.

A detailed printable report shows how to connect the blocks to each other. This simplifies the phase of integration into the control system (PLC, DCS, supervision).


This structure includes 4 blocks : two control blocks and two feed-forward ones. The process is simulated, based on the same hierarchy as the control structure.

 

Example of a cascaded control structure applied to the temperature control of a chemical reactor (master controller) via the intermédiate control of the thermofluid circulating in the jacket (slave controller).



In the cascaded structure one of the PCR assets is the rigorous handling of the way the contraints are transferred from the slave to the master.