Model Predictive Control for small applications

Transcription

1 Model Predictive Control for small applications Slide 12-1 MPC for Small Applications

2 Advantages of Model Predictive Control Automatically compensate for process interaction, measurable load disturbances and constraints using MPC function block. Difficult process dynamics may be effectively addressed e.g. deadtime dominant Allow production to be increased by automatically adjusting throughput to maintain a process at its input operating constraint. Slide 12-2

3 Multi-variable Control Made Easy Example: Control loop with one disturbance and one constraint Traditional Approach Slide 12-3 MPC

4 Saving by Operating Closer to Product Specification or Operating Limit $$ Savings Limit or Spec Target Setpoint Before MPC Slide 12-4

5 Throughput May be Operating at Process Input Limit or Operating Constraint Throughput MPC Production Increase $$ Throughput at limit Operator Setpoint Time Slide 12-5

6 MPC - Addressing Difficult Dynamics MPC Temperature Process (1X1) Slide 12-6

7 MPC -Addressing Difficult Dynamics Slide 12-7

8 MPC -Addressing Difficult Dynamics Slide 12-8

9 Automated process testing to identify the process model Slide 12-9

10 Step Response Model Slide 12-10

11 Verification of identified model Slide 12-11

12 Testing of control using simulated environment Slide 12-12

13 Example Operator interface to MPC Future Values of control Slide 12-13

14 MPC Replacement for PID with Feedforward MPC Process (2X1) Measured Disturbance Slide 12-14

15 MPC Replacement for PID with Feedforward Slide 12-15

16 MPC Replacement for PID Overrride MPC Process (1X2) Constraint Unmeasured Disturbance Slide 12-16

17 MPC Replacement for PID Overrride Slide 12-17

18 Addressing Process Interaction MPC Process (2X2) Slide 12-18

19 Addressing Process Interaction Slide 12-19

20 Addressing Process Interaction Slide 12-20

21 Layering MPC on Existing Strategy MPC AI MPC PID RCAS_OUT AO RCAS_IN Unmeasured Disturbance Process (1X1) Slide 12-21

22 Exposing RCAS_IN & RCAS_OUT Right click on the control or AO block to expose the RCAS_IN as an Input parameter and RCAS_OUT as an Output parameter. Slide 12-22

23 Layering MPC on an Existing Strategy Slide 12-23

24 Example - Long-tube Rising Film Evaporator Feed enters the bottom of the heating tubes and as it heats, steam begins to form. Usually there must be a rather high temperature difference between the heating and boiling sides Extensively used for food, pharmaceutical, and many industrial applications. Slide 12-24

25 Evaporator Products Solids Control With Temperature and pressure Constraints Traditional Strategy - 3 PID s and Control Selector Blocks Slide Predict Strategy - 1 MPC Block

26 Installation Experience 5 Evaporator units commissioned to-date using DeltaV Predict Much faster to config! Much easier to Tune. No de-coupling of loops, 13 minutes and your done! MPC is easier to manipulate from startup SFC's. This is a no-brainer, manipulating the mode of one block would always be easier than four. Simple and easy to use!!!!!!!!! No grey matter needed! Slide 12-26

27 Installation Experience Operators love the MPC Operator Display. Evaporator Graphics are less cluttered with one MPC, than 3 PID dynamo's MPC operator interface is much more simple and easier to use. On the older plants with overide control the operators never know which of the 3 loops is controlling the final control element. They never know if they are constraining the plant via temp or heater pressure and therefore sometimes run the process at lower throughput rates than they need to $$$$$$$. Slide 12-27