Liquid Level Control of Gravity Drained Tank
Liquid Level Control of Gravity Drained Tank
Lab Procedure
Goal: To investigate the dynamic behavior of gravity drained tanks and design control strategies to maintain the level through feedback systems.
References:
Cooper, Doug J. 2006-2008. Practical Process Control. controlguru. [Online] 2006-2008. [Cited: 10 10, 2013.]
James B. Riggs, M. Nazmul Karim. 2007.Chemical and Bio-process Control. s.l.: Ferret Publications, 2007.
In this experiment, you will study and design the control system of a liquid level in a 10 liter tank in the presence of feed flow rate and outlet flow rate disturbances. The self-regulating behavior of the process means that the tank will seek a steady operating level if the inlet flow and disturbances are held constant.
The cart is designed to use water from the lab faucet and is controlled by CV001, which then flows through a MicroMotion Flow Tube into the 10 liter vessel. The rate of water draining from the tank through the 0.25” ID hose is controlled by CV002. The water level in the tank is measured by a Milone eTape level transmitter that acts as a variable resistor providing instantaneous measurements to the microcontroller. The lab technician is the only person authorized to adjust the level transmitter. Students are not permitted to remove electrical tape.
Day 1
Before starting the experiment, trace the flow path from the faucet and identify all components from the P&ID. Draw the basic schematic and block diagram of the apparatus for PI control of the liquid level with the first and second control valve.
Now that you are familiar with the system you are ready for start-up. Hook up the ¼” air hose to the push-connect air line and the regulator on the cart. Check that all valves on the air-header are in the closed position. Turn the air on fully and adjust the regulator to supply 25 psi to the header. Next, screw the 1/2” OD polyvinyl hose to the water faucet and ensure a tight fit. Close the inlet ball valve before turning on the faucet. Turn the faucet handle for cold water fully open and check that no leaks are visible in any tube.
The cart is controlled by the Program LabVIEW 2013. It is a visual based programming that allows for real time data acquisition, and a user friendly user interface called the VI. Open the LabVIEW 2013 program titled LevelControllerFinal.vi from the desktop. This is the graphical user interface and visual programming that performs the control operations. The Front Panel
displays the sensor output for the flow transmitter on the top graph, the level transmitter on the bottom, as well as the Controller Outputs. The tuning parameters for both valves are displayed as are selections for the control of the valves.
You can access the Block Diagram that displays the programming of the VI by going to Window-Show Block Diagram. Students should not have to change any parameters at this point, but it is useful to become familiar with operation.
Plug all three USB cables into the computer and set the Output to COM11, Level transmitter to COM6 and Flow Transmitter to COM7. Click the Run button, the arrow in the top left of the toolbar to start the program. The default setting for both control valves is manual mode.
Open the green drain valve on the bottom of the cart and slowly open the inlet ball valve. Air that was trapped in the line will be forced out, so take care to open slowly until a smooth stream is seen entering the tank.
The first step is to become familiar with the process by manipulating the valve position manually and becoming familiar with the LabVIEW interface. The graphs display the controller output and sensor instantaneous sensor readings for the level transmitter and flow meter. To export this graph data to excel you simply right click on the graph, Export, export data to excel file. After each iteration of the experiment, or if other data is wished to be collected, the graphs must be cleared by right clicking-data operations-clear data, then right click again-data operations-make current value default. This will ensure the graph starts at time t=0 again.
Manually adjust the position of valve 1 and 2 until a steady state is seen on the level transmitter at 5-6”. This will be your Design Level of Operation (DLO) on which all your control strategies are based. Record the position of both valves and the inlet flow rate. Stop the program and clear the data, for you are about to perform the first “bump test”. Run the program for 2 minutes then manually set the first control valve to a position 1-3% away from the DLO. Use a stopwatch and visually record the level in the tank in addition to the automated data collection. Wait until the level in the tank reaches a new DLO. Export this data to Excel. Do this for both valves. Save the data from these experiments for next week
Day 2
This raw data will need to be filtered to make any useful calculations from it. Apply a filter method of your choice to acquire a reasonable looking step response. Mark when the valve position was changed and the time it took to first see a change in the level from its initial DLO. 1) This is the deadtime of the process, θd.
2) The change in the Process Variable (PV) divided by the change in the controller output (CO) is the process gain kp.
3) Next, find where 63% of the change in process variable occurred and the time it took from the initial change in height to get there. This is the process time constant τp.
Alternatively, you may perform a doublet test by manipulating the CO back to its original position after seeing a clear response in the PV.
You are now ready to write a First Order Plus Dead Time equation for the model from the form
Using this FOPDT model you can create your PI controller equation
CO = controller output signal
CObias= controller bias or null value; set bybumpless transfer
e(t) = current controller error, defined as SP – PV
SP = set point
PV = measured process variable
Kc = controller gain, a tuning parameter
Ti = reset time, a tuning parameter
The tuning parameters and type of controller must then be chosen.(Cooper, 2006-2008)
For this self-regulating process, PI control will be sufficient. The closed loop time constant τc can be chosen between .1 and 10 for aggressive to conservative response, respectively.
You will notice that the same CO for one control valve has the reverse effect on the liquid level as the other. What is this called?
Input your tuning parameter for Kp and Ki on the front panel of the VI. Return the cart to steady state as in Day 1 and turn on automated control. Implement disturbances by adding 1 liter of water to the vessel from a graduated cylinder and by closing the faucet 50%. Record the controller’s and process response and export to excel.
Day 3
Using the fundamentals you have learned in your Process Controls class, develop your own process model and calculate a theoretical gain. Is there a way to reject disturbance in the flow before the level in the tank changes?
Of course there is, by cascading the dynamic relationship of the flow control valve to the draining of the tank. This classic example of cascade control can be seen in the block diagram below.
The block diagram displays two controllers, two process variable sensors and only one control element, the Flow Control Valve. This setup allows for the rejection of disturbances to the flow before they affect the liquid level in the tank.
Tuning a Cascade Control Loop
Since the cascade control consists of two nested control loops it becomes necessary to perform two separate bump tests to find the correlation of the process dynamics.
1)Allow the process to reach a steady state at your chosen DLO as before. Place Valve 1 in manual mode and implement a 1-3% change in its position.
2)Collect the dynamic data of the Inner Controller Output, Process Variable 2 (Flow), and Process Variable 1 (Height).
3)Fit a FOPDT model to the data collected as in the method for PI control.
Inner Loop Outer Loop
- Process Gain : Kp2=( ) Kp1=( )
- Time Constant :
- Dead Time:
4)Which dynamic process is faster?
The control of the inner loop is dependent on the speed of each process. It is therefore imperative that the inner loop be controlled with the proper scheme. The inner control parameters must be set on the block diagram of the VI, once set they must not be altered.
Discussion Questions
1)Define the terms Process Gain, Process Variable, Control Variable, Dead Time, and Time Constant.
2)If I/P 1 receives a 13mA signal from the microcontroller, what is the resulting theoretical position of CV 1?
3)Why is it necessary to apply a filter to the sensor readings?
4)Using Bernoulli’s equation, calculate the volumetric flow rate out of the drain. How does this compare to actual results?
5)What is an integrating process?
6)What are the parts of the FOPDT model?
7)Develop a process model using your knowledge of an overall mass balance on the system, and the instantaneous sensor readings.
8)How does the results from part 4 compare to the FOPDT model you found from experimental data? Discuss any differences and potential reasons.
9)Discuss the advantages and disadvantages of P-only, PI- control and Cascade control in respect to this process.
10)If you were to do the experiment again, what in the procedure or cart design itself would you improve upon to facilitate a better lab experience?
Figure 2. P&ID