Flow rate and filling process control

Experiment 4: Process Control of Filling Level and Flow Rate4.1 Objective: 1. To understand the closed loop control system in three different processes. 2. To identify the manipulated and controlled variables 3. To identify the components (sensors, controllers, actuators) in a closed loop system for different types of process control. 4. To control and tune the process using PID controller.4.2 IntroductionProcess control is an engineering discipline that deals with architectures, mechanisms and algorithms for maintaining the output of a specific process within a desired range. Process control is extensively used in industry and enables mass production of consistent products from continuously operated processes such as oil refining, paper manufacturing, chemicals, power plants and many others. Process control enables automation, by which a small staff of operating personnel can operate a complex process from a central control room. There is some terminology which is commonly used in process control. Controlled variables are the variables which quantify the performance or quality of the final product, which are also called output variables. Manipulated variables are input variables that are adjusted dynamically to keep the controlled variables at their set-points. Disturbance variables are also called "load" variables and represent input variables that can cause the controlled variables to deviate from their respective set points. Set-point change is implementing a change in the operating conditions. The set-point signal is changed and the manipulated variable is adjusted appropriately to achieve the new operating conditions. It is also known as servomechanism (or "servo") control. Disturbance change is the process transient behaviour when a disturbance enters, also called regulatory control or load change. A control system should be able to return each controlled variable back to its set-point.A familiar example of feedback control is cruise control on an automobile. Here speed is the measured variable. The operator (driver) adjusts the desired speed set point (e.g. 100 km/hr) and the controller monitors the speed sensor and compares the measured speed to the set point. Any deviations, such as changes in grade, drag, wind speed or even using a different grade of fuel (for example an ethanol blend) are corrected by the controller making a compensating adjustment to the fuel valve open position, which is the manipulated variable. The controller makes adjustments having information only about the error (magnitude, rate of change or cumulative error) although settings known as tuning are used to achieve stable control. The operation of such controllers is the subject of control theory.In this experiment, we are going to determine the properties of P, PI and PID controller and learn how to tune this controller.

4.3 Materials and Equipment

4.3.1 FILLING LEVEL CONTROL 1. 1 Pump, 2. 1 Motor Controller, 3. 1 Analogue Ultrasonic Sensor, 4. 1 Digital Industrial Controller, 5. 2 containers, 6. 1 Control Console,7. Profile Plate, 8. Tubing, 9. Electrical and mechanical components, 10. Capacitive Proximity Sensors

4.3.2 FLOW RATE CONTROL

1. 1 Pump, 2. 1 motor controller, 3. I proportional valve, 4. 1 digital industrial controller, 5. Tubing, electrical and mechanical components,6. 1 container, 7. 1 Control Console,8. 1 profile plate,9. 1 flow sensor, 10. Capacitive proximity sensor

4.4 Procedure FLOW RATE CONTROL4.4.1 P Controller1. First, the set point is set to 3.5. Wait for the graph to stabilize at set point.2. Tr and Td is set to a fix value. 3. Value of Kp is manipulated.4. The set point then is change to 4.0.5. The graph of the response is captured.

4.4.2 PI Controller1. First, the set point is set to 3.5. Wait for the graph to stabilize at set point.2. Kp and Td is set to a fix value. 3. Value of Tr is manipulated.4. The set point then is change to 4.0.5. The graph of the response is captured.6. Step 1 to 5 is repeated, but this time, Tr and Td value is fixed and Kp value is manipulated.4.4.3 PID Controller1. First, the set point is set to 3.5. Wait for the graph to stabilize at set point.2. The Kp is set to a fix value. 3. Value of Tr and Td is manipulated.4. The set point then is change to 4.0.5. The graph of the response is captured.6. Step 1 to 5 is repeated but this time, Tr is kept constant and Kp and Td is manipulated.7. Step 1 to 5 is repeated but this time, Td is kept constant and Kp and Tr is manipulated. (Td must not exceed 1.5)

FILLING PROCESS4.4.4 P Controller6. First, the set point is set to 35. Wait for the graph to stabilize at set point.7. Tr and Td is set to a fix value. 8. Value of Kp is manipulated.9. The set point then is change to 40.10. The graph of the response is captured.

4.4.5 PI Controller7. First, the set point is set to 35. Wait for the graph to stabilize at set point.8. Kp and Td is set to a fix value. 9. Value of Tr is manipulated.10. The set point then is change to 40.11. The graph of the response is captured.12. Step 1 to 5 is repeated, but this time, Tr and Td value is fixed and Kp value is manipulated.4.4.6 PID Controller8. First, the set point is set to 3.5. Wait for the graph to stabilize at set point.9. The Kp is set to a fix value. 10. Value of Tr and Td is manipulated.11. The set point then is change to 4.0.12. The graph of the response is captured.13. Step 1 to 5 is repeated but this time, Tr is kept constant and Kp and Td is manipulated.14. Step 1 to 5 is repeated but this time, Td is kept constant and Kp and Tr is manipulated. (Td must not exceed 1.5)

4.5 Result4.5.1 Flow Rate ControlSet point: 4.00 (desired value) Manipulated variable: pump (performance of pump) Measured value: the value measured by flow sensor Disturbance: No Kp= proportional control Tr= integral control Td= derivative control tr= rise time ts= settling time

4.5.1.1 P Controller Table 7: The value of overshoot, rise time, settling time and steady state error when P controller is tuned by changing the Kp and fixing the Tr123

Kp283520

Tr1.01.01.0

Td000

Set point444

Manipulated variable81.8382.1083.59

Overshoot, OS01.3250

Rise time, tr5.535

Settling time, ts55undefine11

Steady state error0.010.010.01

Figure 7(a): Flow rate control by controller P when Kp=28 and Tr=1.0

Figure 7(b): Flow rate control by controller P when Kp=35 and Tr=1.0

Figure 7(c): Flow rate control by controller P when Kp=20 and Tr=1.0

4.5.1.2 PI Controller Table 8: The overshoot, rise time, settling time and steady state error when PI controller is tuned by fixing the Kp and changing the Tr by 1.5, 5.0, 0.8123

Kp282828

Tr1.55.00.8

Td000

Set point444

Manipulated variable 81.9179.2178.49

Overshoot, OS000

Rise time, tr245

Settling time, ts3715

Steady state error0.00100




Table 9: The overshoot, rise time, settling time and steady state error when PI controller is tuned by fixing the Tr and changing the Kp by 28, 20.0, 15.0.123

Kp28.020.015.0

Tr1.51.51.5

Td000

Set point444


Overshoot, OS000

Rise time, tr433

Settling time, ts579





4.5.1.3 PID ControllerKp is fixed and Tr and Td variables changes.Set 1

Set 2

Set 3

123

Kp282828

Tr567

Td0.10.20.3

Set point444


Overshoot, OS000

Rise Time, tr (s)22.636317.6423.8264

Settling Time, ts (s)38.456450.621444.8536

Steady state error000

Tr is fixed and Kp and Td variables changesSet 4

Set 5

Set 6

123

Kp203035

Tr555

Td0.30.20.1

Set point444


Overshoot, OS00.40910.5909

Rise Time, tr (s)13.852113.021512.6345

Settling Time, ts (s)28.340928.0430-


Td is fixed where Kp and Tr variables changesSet 7

Set 8

Set 9

123

Kp203040

Tr1.51.31.1

Td0.30.30.3

Set point444


Overshoot, OS00.150.7

Rise Time, tr (s)11.15026.253213.6524

Settling Time, ts (s)9.811.2-


4.5.2 Filling Control4.5.2.1 P ControllerFix Tr and change Kp by 25% or more.

Table 1: The value of overshoot, rise time, settling time and steady state error when P controller is tuned.123

Kp283542.0

Tr0.50.50.5

Td000

Set point404040


Overshoot, OS0.10.10.1

Rise Time, tr (s)21.020.8323.21



Figure 1(a): Filling level control for P controller when Kp=28 and Tr=0.5 (run 1)

Figure 1(b): Filling level control by P Controller when Kp=35 and Tr=0.5 (run 2)

Figure 1(c): Filling level control by P Controller when Kp=42.0 and Tr=0.5 (run 3)

PI Controlleri. Fix Kp and change Tr by 25% or more.

Table 2: The value of overshoot, rise time, settling time and steady state error when PI controller is tuned.123

Kp252525

Tr0.51.01.5

Td000

Set point404040



Rise Time, tr (s)21.5622.6123.83



Figure 2(a): Filling level control by PI Controller when Kp=25 and Tr=0.5 (run 1)

Figure 2(b): Filling level control by PI Controller when Kp=25 and Tr=1.0 (run 2)

Figure 2(c): Filling level control by PI Controller when Kp=25 and Tr=1.5 (run 3)

4.5.2.2 PI Controllerii. Fix Tr and change Kp by 25% or more.

Table 3: The value of overshoot, rise time, settling time and steady state error when PI controller is tuned.123

Kp203040

Tr0.50.50.5

Td000

Set point404040


Overshoot, OS 0.130.100.10

Rise Time, tr (s)25.020.2423.81



Figure 3(a): Filling level control by PI Controller when Kp=20 and Tr=0.5 (run 1)

Figure 3(b): Filling level control by PI Controller when Kp=30 and Tr=0.5 (run 2)

Figure 3(c): Filling level control by PI Controller when Kp=40 and Tr=0.5 (run 3)

4.5.2.3 PID Controller4.5.2.3 (I) Fix Kp and change other variables.Table 4: The value of overshoot, rise time, settling time and steady state error when PID controller is tuned.

123

Kp252525

Tr0.51.01.5

Td0.20.61.0

Set point404040



Rise Time, tr (s)18.4519.0521.4



Figure 4(a): Filling level control by PID controller when Kp=25, Tr=0.5 and Td=0.2 (run 1)

Figure 4(b): Filling level control by PID Controller when Kp=25, Tr=1.0 and Td=0.6(run 2)

Figure 4(c): Filling level control by PID Controller when Kp=25, Tr=1.5 and Td=1.0 (run 3)

4.5.2.3 PID Controller4.5.2.3 (II) Fix Tr and change other variables.

Table 5: The value of overshoot, rise time, settling time and steady state error when PID controller is tuned.123

Kp283542

Tr0.50.50.5

Td0.20.61.0

Set point404040



Rise Time, tr (s)22.6221.422.02



Figure 5(a): Filling level control by PID Controller when Kp=28, Tr=0.5 and Td=0.2(run 1)

Figure 5(b): Filling level control by PID Controller when Kp=35, Tr=0.5 and Td=0.6 (run 2)

Figure 5(c): Filling level control by PID Controller when Kp=42, Tr=0.5 and Td=1.0(Run 3)

4.5.2.3 (III) Fix Td and change other variables.

Table 6: The value of overshoot, rise time, settling time and steady state error when PID controller is tuned.123

Kp283542

Tr0.51.01.5

Td0.20.20.2

Set point404040


Overshoot, OS 0.05890.09610.99

Rise Time, tr (s)22.0222.6223.21

Settling Time, ts (s) 69.6467.8649.40


Figure 6(a): Filling level control by PID Controller when Kp=28, Tr=0.5 and Td=0.2(Run 1)

Figure 6(b): Filling level control by PID Controller when Kp=35, Tr=1.0 and Td=0.2(Run 2)

Figure 6(c): Filling level control by PID Controller when Kp=42, Tr=1.5 and Td=0.2(Run 3)

4.6. Discussion4.6.1 FLOW RATE CONTROLFirst, P controller was used. From the Graph, The values of Kp were selected by 28, 35 and 20. The initial value was 28 and other variable were set to zero. However, Tr value cannot be set to zero as the smallest value of Tr will be 1.0.The rise time was 5.5, settling time was 50 and steady state error of 0.01.It has overshoot of 0.The second reading was 35. The rise time was 3, settling time was cannot be defined and steady state error of 0.01.It has overshot which is 1.325. The third reading was 20. The rise time was 5, settling time was 11 and steady state error of 0.01.It has no overshoot. The second reading cannot be defined due tu error occur during experiment. The rise time decreases with the increase in Kp. This agrees to the theory. Overshoot was reduced, and the theory says that after certain value of steady state error, the overshoot will increase. The settling time decreases but theoretically it will have small change. The steady state error remains constant. According to the theory, increase in Kp will decrease rise time, increase overshoot, decrease steady state error and little effect on settling time. The proportional component depends only on the Error, which is the difference between the Set Point and the Process Variable. The Proportional gain(Kc)determines the ratio of Output response to the Error. For instance, if the Error signal has a magnitude of 10, a Proportional Gain of 5 would produce a proportional response of 50. In general, increasing the Proportional Gain will increase the speed of the control system response and also decrease the steady-state error which is the final difference between Process variable and Set Point. However, if the Proportional Gain is too large the Process Variable will begin to oscillate. If Kc is increased further, the oscillations will become larger and the system will become unstable.For Proportional-Integral (PI) controller, there were six sets tested. The first three sets were carried out by setting the Kp value fixed, Td was set to 0 and integral time, Tr was set to 1.5-Set 1, 5.0-Set 2 and 0.8-Set 3. For set 1, it resulted in 0.001 % of overshoot assume to be 0%, 2s of rise time, 3s of settling time and 0.001 of steady state error. For set 2, it resulted in 0% of overshoot, 4 s of rise time, 7s of settling time and no steady state error. For set 3, it resulted in no overshoot, 5s of rise time, 15 s of settling time and no steady state error. Theoretically, the increase in Tr will decrease rise time, increase overshoot, increase settling time and eliminate steady state error. Based on the results obtained, the rise time decreases from set 1 to set 2.The set 3 showed increases in rise time which did not agree to theory. For settling time, set 2 and set 3 agreed to theory, from set 1 to set 2 it showed decrease in settling time, which differed from theory. The overshoot and steady state error showed readings around zero. Theoretically, the overshoot will increase but it was opposite in the results. This could be due to very small change in the Tr value, which turned to exhibit only some of the properties or maybe there is an error occur during handling the experiment.The three other sets were determined by fixing Tr value constant= 1.5 and changing the Kp value, Kp= 28-Set 4, Kp = 20-Set 5, Kp = 15-Set 6.For set 4, it resulted in no overshoot, 3 s of rise time, 5s of settling time and no steady state error. For set 5, it resulted in no overshoot, 4 s of rise time, 7 s of settling time and 0.2 of steady state error. For set 6, it resulted in no overshoot, 5 s of rise time, 9 s of settling time and 0.001of steady state error. Theoretically, the increase in Kp will decrease rise time, increase overshoot, decrease steady state error and has less effect on settling time. The presence of integral controller must eliminate steady state error. Based on the results, the rise time decreases as the Kp increase and with the presence of integral action. The settling time showed fluctuations, which means there is need to introduce the PID controller. The steady state error was very small and can be considered the value approaching zero for the three sets.The major advantage in introducing the integral controller is the removal of steady state error. PI controller has overall better performance than p controller. However, the integral action may cause overshoot, oscillation, and/or instability problems if the selected integral gain(Tr)is too small. Integral controllers tend to respond slowly at first, but over a long period of time they tend to eliminate errors. The integral controllereliminates the steady-state error, but may make the transient response worse. The controller may be unstable.For PID Controller, from the graphs, we can see that PID is better than system with only P controller or P and I controller. In control system, the rise time of a system can be improved by adding a proportional controller or to increase the value of proportional gain of the controller. As for steady-state error can be improved by adding an integral controller. In order to improve the overshoot of a system, a derivative controller will be added. The above PID graphs have all the three proportional, integral and derivative controllers. Hence the system will experience less oscillation, fast rise time, short settling time and also less steady-state error. For PID system, the type of controller used was Proportional-Integral-Derivative (PID) controller. The first three sets were determined by fixing the Kp value constant to 28 and increasing both Td and Tr values. For set 1, Kp= 28 , Tr= 5, Td= 0.1, it resulted in almost no overshoot, 22s of rise time, 38s of settling time and almost no steady state error. For set 2, Kp= 28, Tr= 6, Td=0.2, it resulted in no overshoot, 17s of rise time, 50s of settling time and almost no steady state error. For set 3, Kp= 28, Tr= 7, Td= 0.3, it resulted in no overshoot, 23s of rise time, 44s of settling time and no steady state error. Since there are three types of controlled being integrated, it becomes very difficult to predict the outcomes. These trials helped to determine the pattern that could be obtained when Kp fixed, and the increase in Tr and Td.The results obtained showed there is decrease in rise time and settling time and with little or no overshoot and settling time. For set 2 values are inaccurate as there was no enough time for the graph in system to settle down before keying in new values Tr and Td. Hence the result of set 2 shows a great deviation from set 1 and set 3. Next step(set 4-6) the Tr value was set to constant and increasing Kp and decreasing Td values. For set 4, Kp= 20, Tr=5, Td= 0.3, it resulted in no overshoot, 13.85s of rise time, 28s of settling time and no steady state error. For set 5, Kp= 30, Tr=5, Td= 0.2, it resulted in 40.91% of overshoot, 13.02s of rise time, 28s of settling time and no steady state error. For set 6, Kp= 35, Tr= 5, Td= 0.1, it resulted in 59.09% of overshoot, 12.63s of rise time and very small steady state error. Based on the results, the rise time and settling time reduced and the steady state error and overshoot can be considered almost zero. This pattern also makes the response good and better than P and PI controller. For set 6 the graph shows very large oscillation and the system didnt settle down near to the setpoint. Hence there is no settling time for set 3 trial. In order to have a shorter rise time and settling time, both Kp and Td shall be increase or decrease instead of increasing one and decreasing the other. Sets 7-9 were determined by fixing the Td value constant to and increasing Kp and decreasing Tr values. For set 7, Kp= 20, Tr=1.5, Td=0.3, it resulted in no overshoot, 11.15 s of rise time, 9.8 s of settling time and no steady state error. For set 8, Kp=30, Tr=1.3, Td=0.3, it resulted in little overshoot, 6.25s of rise time, 11.2s of settling time and no steady state error. For set 9, Kp=40, Tr=1.1, Td=0.3, it resulted in little overshoot, 13.6s of rise time and no steady state error. For set 9 the graph shows very great oscillation and the graph didnt settle down to the setpoint. Hence the result of rise time and settling time from set 9 is greatly deviate from the trials. For this pattern, the rise time and settling time were decreased; the steady state error and overshoot can be considered as zero. This method is also very successful as all the criteria have been fulfilled.Differences occur between the PID controller and normal controllers in the way they work. The PID controller uses an advanced formula to try and prevent any errors from occurring. This ensures the devices or system being controlled performs as flawlessly as possible. This formula type is known as an algorithm. An algorithm directs actions based on what is happening. The algorithm would have specific directions on how to react to certain changes. In this way, an algorithm is a series of different procedures that can be followed or altered based on what the device receiving the orders is doing Finally, the PID controller participates in a feedback loop. Information is sent out by the controller, received by the devices, and information from the devices is sent back to the controller. The controller then makes a decision on how to proceed based on the information it receives and sends it out, creating a continuous loop. One main advantage stands out above the rest when using a PID controller. It can control various systems or devices with little human interaction. Not only does this allow the workers to concentrate on other tasks, but it also allows many processes to run at once. The drawback to this method comes from the fact that the controller must be tuned, meaning the instructions that tell it what to do must be tweaked, to keep it functioning properly. To do this, advanced knowledge for setting up this type of controller is required to avoid error.

4.6.2 FILLING LEVEL CONTROL In this experiment, filling level control was being control by P controller, PI controller and PID controller. From the result by using P controller for filling level experiment, the Tr was fixed at 0.5 and change Kp by 25% or more by an initial value of 28. . Based on result obtained as shown in Table 1, it is clearly shown that graph in Figure 1(c) is more oscillatory than others two when P controller is tuned. This is because by increased value of Kp increase the rise time and overshoot causing it to be oscillated as high Kp will lead to instability. However, from the results obtained overshoot does not increase and rise time is not decrease with the increased Kp. This may due to technical error cause by filling level control device when taking the performance of filling level control.For the PI controller, the Integral is added to the controller to control the filling level system. PI controller is tuned by fixing Kp and change Tr at first. Theoretically, Ki is proportional to the 1/Tr, when the value of Tr increases, the value of Ki will decreases. Therefore, the rise time is supposed to be increased, overshoot and settling time should be decreased and steady-state error should be eliminated in this process. Based on the results obtained, it can be clearly seem that when the integral increases, the rise time had increased from 21.56s to 23.83s. This result had corresponded to the theory. However, the data of settling time and the overshoot obtained did not agreed with the theory. For the settling time, it increased from 51.28s to 64.78s. For the overshoot, it is constant to which at 0.1% . These might due to the parallax error in this experiment since the graph is too small and quite difficult to calculate the settling time and overshoot. However, the integral had success to eliminate the steady -state error from 0.875 to 0.7. When Tr is fixed at 0.5 and Kp is kept increasing, the rise time increases as shown in the Table 2 and overshoot had reduced from 0.13% to 0.10%. The steady state error also decreased. The added of Integral controller into the system had proved it can successfully eliminate the steady state error and producing a more efficient control system with faster rise time.Then, PID controller is used in this process. Firstly Kp was fixed and Td was increased. Initially, overshoot from 0.1 increased to 0.14. Rise time increase from 18.45s to 21.4s.while the settling time decrease from 75.0s to 63.09s.this means that increasing number of td and tr needs shorter rime to became stable. It is proven as the steady state error is reduce from 0.2% to 0.04%. Secondly, Tr is fixed to 0.5 while Kp and Td are changes throughout the process. When Kp is increases, the oscillation response and large overshoot will be produced. Besides that, the effect of Td also will cause the oscillation response because Kd is proportional to Td. When the Td decrease, the effect of Kd also reduce. The double effect of Kp and Kd produce an unstable response Thus, in this second trial, the overshoot is quite large which is 0.8%. However, the rise time is decrease from 22.62s to 21.4s and then increase again at 22.02. This inconsistent in the values may due to parallax error when calculate the readings. Next, the settling time is decreases from 79.76s to 63.10s. Moreover, the steady-state error is decrease from 0.1 to small value 0.03. Lastly, Td is fixed at 0.2 while Kp and Tr are changes throughout the process. Overall, the results of rise time is increased, settling time decreased. But the overshoot show increased. This may due to the parallax error while because of the small scale of the graph make it difficult to read the data. However, the steady-state is successfully eliminated.

4.7 Conclusion In conclusion, the experiment showed that PID controller is the best among the controllers. PID controller produced response with fast rise time, lowest overshoot, short settling time and almost no steady state error. However, a sudden change in set point will cause derivative terms momentarily to become very large and thus provide a derivative kick to the final control element.

4.8 Reference1. PID Controller Simplified. (2008, May 11). Retrieved from https://radhesh.wordpress.com/2008/05/11/pid-controller-simplified/2. PID Controller Design. (2012, January 1). Retrieved from http://ctms.engin.umich.edu/CTMS/index.php?example=Introductionion=ControlPID

0 | Page

Flow rate and filling process control

Documents

cruise control

introductionprocess

regulatory control

servo control

stable control

control console

controlled variables

flow rate control