You are on page 1of 30

Assignment #04 (Workshop 11)

CHEN-5352-04 Advanced Process Control

Feed Forward Control of the Jacketed Reactor

Subitted by: Samudra Gupta Vamsi Krishna Banda

Lamar University, Beaumont, Texas 19th March 2014

Page 1

Assignment #04 (Workshop 11)


I. Objective To explore feed forward controller design and implementation. Also, to compare disturbance rejection performance of a feed forward controller with that of a lone PI controller for the jacketed reactor process. II. Introduction A feed forward control element will be used in combination with a traditional feedback controller in this study (called feed forward with feedback trim). The feed forward element is comprised of a process model that describes controller output to process variable dynamic behavior and a disturbance model that describes the disturbance to process variable dynamic behavior. Feed forward, like cascade control, is used for improved disturbance rejection. Feed forward control is appropriate when one particular disturbance causes a significant disruption to process operation

Figure 1 showing jacketed reactor

Lamar University, Beaumont, Texas 19th March 2014

Page 2

Assignment #04 (Workshop 11)


Where, D is disturbance, C PV is measured process variable, C SP is set point, C CO is controller output, % III. First Order Plus Dead Time Order (FOPDT) The basic equation for first order is as follows:

G( s)
Where,

K s 1

K is steady state gain is the time constant d is the dead time The parameters of the process are obtained by going through following steps: The control loop is opened by switching it to manual This is done by making a doublet The time after each step change in the doublet should be enough so that manipulated variable reaches a steady state IV. Set Point Tracking By enabling set point tracking in a controller you make set point of the track it process value. This is to enable a bumpless transfer when the controler is switched to automatic from manual.

Lamar University, Beaumont, Texas 19th March 2014

Page 3

Assignment #04 (Workshop 11)


V. Second Order Plus Dead Time Order (SOPDT) Second-Order plus Dead-Time (SOPDT) models are commonly used to approximate systems in order to tune PID controllers. In simple process models three major control parameters dominates the control design in many applications. Those are, with at most three parameters Gain K, Time Constant T and Time-Delay. FOPDT models are limited when the process has under damped behavior or high order dynamics. This way, SOPDT models can better describe for over damped behavior kind of process. Transfer function and time domain of second order plus dead-time is shown below:

Figure 2 showing the equations for second order plus dead time order (overdamped)

The aim of this workshop is to present different tuning rules for designing Jacketed Reactor for SOPDT processes. These tuning rules are obtained by imposing various specifications on the closed-loop system, such as changing the above three parameters (Gain K, Time Constant T and TimeDelay). This is done by doubling and halving each parameter by keeping other two at constant value. This proposed tuning method has successfully been applied to the control of an experiment system that is modeled as an SOPDT process. The obtained experimental results verify the efficiency of
Lamar University, Beaumont, Texas 19th March 2014

Page 4

Assignment #04 (Workshop 11)


the proposed tuning methods that provide a very satisfactory performance of the closed-loop system. VI. Procedure Reactor exit temperature indicates percent conversion of reactants to products so maintaing the exit temperature is our control objective. Hence Tsetpoint = 92 oC. Process model describes the dynamic behavior of the process variable in response to the controller output changes. An open loop test is performed by stepping the control output from 42%5% (It is similar for both step 2 and 4) . A disturbance model describes the dynamic behavior of the process variable in response to disturbance changes. For this, the cooling jacket inlet temperature is increased from 50 oC to 60 oC and back to 50 oC and the response is recorded (It is similar for both step 3 and 5) . In Design Controller menu, select the PID with feed forward controller. Click on the process tab and FOPDT process model

parameters from step 4 and in Disturb tab enter FOPDT disturbance model parameters from step 5. Disturbance rejection performance of the feed forward with feedback trim implementation is tested by making step changes in the disturbance from 50 to 60 and back to 50 oC.
Lamar University, Beaumont, Texas 19th March 2014 Page 5

Assignment #04 (Workshop 11)

In Design Controller menu, select the PID with feed forward controller. Click on the process tab and SOPDT process model parameters from step 4 and in Disturb tab enter SOPDT disturbance model parameters from step 5. Disturbance rejection performance of the feed forward with feedback trim implementation is tested by making step changes in the disturbance from 50 to 60 and back to 50 oC. (It is similar for both step 6 and 7) . Disturbance rejection performance of static feed forward to dynamic feed forward is compared. Sensitivity study to determine how each fed forward model parameter impacts the disturbance rejection performance is conducted. Benefit of the feed forward control on set point tracking performance is explored. Step the set point from 92 oC up to 95 oC and back again.

Lamar University, Beaumont, Texas 19th March 2014

Page 6

Assignment #04 (Workshop 11)


VII. Process The following are the steps to complete the workshop Step-1: The following is the figure showing the control station 3.7 user interface when jacketed reactor case study is chosen.

Figure 3: showing jacketed reactor interface

Lamar University, Beaumont, Texas 19th March 2014

Page 7

Assignment #04 (Workshop 11)


Step-2: After making sure the setpoint is set at 92C we do a doublet test by firstly start saving the live data in a text file, then stepping the controller output from 42% to 37% then to 47% and back to 42%. Below is the figure showing the controller output.

Figure 4 showing the variable which is to be changed

After performing the test we stop saving the live data to the file and open that file in design tools. Upon opening we choose first order plus dead time sequence and start fitting the curve, after fitting the data in first order plus dead time sequence we fit the data for second order plus dead time sequence (SOPDT). Below is table showing the values for process gain, over all time constant, dead time, sum of squared errors and goodness of fit for both the curve fittings.
Table 1: Showing values of model parameters for FOPDT and SOPDT sequence

Model Parameters

First Order Plus Dead Time Second Order Plus Dead (FOPDT) Time (SOPDT) Process Gain, Kc -0.3549 -0.3544 First Time Constant, 1 2.03 1.86 Second Time Constant, 2 0 1.2 Dead Time, 0.8155 0.4888 Sum of Squared Error 1.79 1.66 (SSE) Goodness of Fit, R2 0.9979 0.9981
Lamar University, Beaumont, Texas 19th March 2014 Page 8

Assignment #04 (Workshop 11)


Below is the figure showing the curve obtained after fitting for first order plus dead time sequence.

Figure 5 showing controller output response with step change (K c=-0.3549, i=2.03, =0.8155, 2 R =0.9979)

Below is the figure showing the model parameters obtained after fitting for first time order plus dead time sequence.

Figure 6 showing model parameters after fitting the data in FOPDT sequence

Lamar University, Beaumont, Texas 19th March 2014

Page 9

Assignment #04 (Workshop 11)


Below is the figure when the data is fitted in curve for second order plus dead time (SOPDT) sequence.

Figure 7 showing controller output response for SOPDT sequence (K c=-0.3544, i=1.86, 2=0.4636, 2 =0.4888, R =0.9981)

Below is the figure showing model parameters after the data is fitted in a curve in second order plus dead time (SOPDT) sequence.

Figure 8 showing model parameters for SOPDT sequence

Lamar University, Beaumont, Texas 19th March 2014

Page 10

Assignment #04 (Workshop 11)


Step-3: Now we manipulate the disturbance variable by stepping the cooling jacket inlet temperature from 50C to 60C and back to 50C and save the data in another text file. Below are the figures showing the data fitting in a curve for first order plus dead time (FOPDT) sequence and second order plus dead time (SOPDT) sequence.

Figure 9 showing cooling jacket inlet temperature response (Kc=-0.6728, i=2.52, =1.04, R =0.9993)

Figure 10 showing model parameters when data is fitted in FOPDT sequence

Lamar University, Beaumont, Texas 19th March 2014

Page 11

Assignment #04 (Workshop 11)

Figure 11 showing the cooling jacket inlet temperature response in SOPDT sequence (K c=-0.6709, 2 i=1.85, 2=1.2, =0.3982, R =0.9998)

Figure 12 showing model parameters when data is fitted in SOPDT sequence

Lamar University, Beaumont, Texas 19th March 2014

Page 12

Assignment #04 (Workshop 11)


Below is the table showing model parameters when the data for cooling jacket inlet temperature is fitted in first order plus dead time (FOPDT) and second order plus dead time (SOPDT) sequence
Table 2 showing model parameters for FOPDT and SOPDT sequence

Model Parameters

First Order Plus Dead Time Second Order Plus Dead (FOPDT) Time (SOPDT) Process Gain, Kc 0.6728 0.6709 First Time Constant, 1 2.52 1.85 Second Time Constant, 2 0 1.2 Dead Time, 1.04 0.3982 Sum of Squared Error 3.25 0.7572 (SSE) Goodness of Fit, R2 0.9993 0.9998

Based on the SSE, does the SOPDT model show improved capability in describing the disturbance behavior over the FOPDT model? From the data above collected we can say that SOPDT model does show improved capability in describing the disturbance behavior over the FOPDT model. Step-4: Now we select the controller design menu and in that we first set the setpoint at 92C, then we select from the drop down menu PI controller and we input the value Kc and i which we obtained from step 2 for FOPDT sequence in the process model section. Now we choose the disturbance model tab and fill in the values which we got from step 3 for FOPDT sequence. After filling in the values we test the disturbance rejection performance of the feed forward with feedback trim implementations by making step changes in the disturbance (cooling jacket inlet temperature) from 50C to 60C and back to 50C

Lamar University, Beaumont, Texas 19th March 2014

Page 13

Assignment #04 (Workshop 11)


Below is the figures showing the response of the step changes in disturbance without feed forward (PI controller)

Figure 13 showing the disturbance response without feed forward implementation (Tuning Gain=-3.9, Reset time= 2.03)

Lamar University, Beaumont, Texas 19th March 2014

Page 14

Assignment #04 (Workshop 11)


Below is the figure showing the response of the disturbance (cooling jacket inlet temperature) with feed forward controller.

.
Figure 14 showing the disturbance response with feed forward (K p=-0.3549, 1=2.03, 2=0.0, =0.8155, KD=0.6728, 1=2.52, 2=0.0, =1.04)

How does the two controllers compare? Does the feed forward element show clear benefit? Is the PI tuning you used for the non feed forward study correct or to make the comparison fair, do you need to adjust the tuning values? From the above figures we can say that controller with feed forward is better than the controller without feed forward as the temperature rise in controller without feed forward is higher than in controller with feed forward, the rise in the value of process variable is in consideration with the controller output for the feed forward controller which is a bit off in controller without feed forward and also the time taken to reach a steady state is higher in case of controller without feed forward as compared to controller with feed forward.
Lamar University, Beaumont, Texas 19th March 2014

Page 15

Assignment #04 (Workshop 11)


The PI tuning used for comparison is fair and we do no need to adjust the tuning values. Step-5: Now we run the disturbance test with feed forward controller for first order plus dead time (FOPDT) sequence and for second order plus dead time (SOPDT) sequence. By making step changes in the disturbance from (cooling jacket inlet temperature) from 50C to 60C and back to 50C. Below is the figure showing disturbance response with feed forward controller in first order plus dead time sequence (FOPDT).

Figure 14 showing the disturbance response with feed forward (K p=-0.3549, 1=2.03, 2=0.0, =0.8155, KD=0.6728, 1=2.52, 2=0.0, =1.04)

Lamar University, Beaumont, Texas 19th March 2014

Page 16

Assignment #04 (Workshop 11)

Below is the figure showing disturbance response in second order plus dead time (SOPDT) sequence.

Figure 14 showing the disturbance response with feed forward (K p=-0.3544, 1=1.86, 2=0.4636, =0.3982, KD=0.6709, 1=1.85, 2=1.02, =0.3982)

The SOPDT model had lower SSE value than the FOPDT model, implying better model fits of the data. Does the SOPDT model show clearly improved feed forward disturbance rejection performance. Why or why not? The SOPDT model does show an improved feed forward disturbance rejection as it can be observed from figure 14 the rise in process variable setpoint is in correlation with rise in controller output as well when the disturbance value drops from 60% to 50% the drop in process variable is in correlation with controller output as well, in both case for rise and drop the correlation is steady whereas from figure 13 we can see that the correlation is not steady.
Lamar University, Beaumont, Texas 19th March 2014 Page 17

Assignment #04 (Workshop 11)

STATIC FOPDT

Lamar University, Beaumont, Texas 19th March 2014

Page 18

Assignment #04 (Workshop 11)

STATIC SOPDT

Lamar University, Beaumont, Texas 19th March 2014

Page 19

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 20

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 21

Assignment #04 (Workshop 11)

STEP 10

Lamar University, Beaumont, Texas 19th March 2014

Page 22

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 23

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 24

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 25

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 26

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 27

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 28

Assignment #04 (Workshop 11)

SETPOINT STEP CHANGE FEED FORWARD

Lamar University, Beaumont, Texas 19th March 2014

Page 29

Assignment #04 (Workshop 11)

Lamar University, Beaumont, Texas 19th March 2014

Page 30

You might also like