Experiment Description - Multi-Tank System - Modified

Experimental Control Systems (XCOS)
Shouvik Chaudhuri
Experiment 3: Multi-Tank System (Quanser) Lab
Room
1. Brief Description of the Experiment:
Group
Fig. 1. Multi-tank system setup in CPL Fig. 2. Geometrical Parameters of the tanks
The Multi-tank System (Fig. 1) comprises three separate tanks (third tank not shown on
purpose) fitted with drain valves. The separate tank mounted at the base of the set-up acts as a
water reservoir for the system. Some of the tanks have a constant cross section, while others are
spherical or conical, so having variable cross section. This creates many nonlinearities in the
system. The following points should be noted regarding the system.
a. Liquid is pumped into the upper tank from the reservoir by a variable speed pump driven
by DC motor. The liquid outflows the tanks only due to gravity. The output orifices act as
flow resistors but can also be controlled from the computer.
b. The levels in the tanks are measured with a level sensor based on hydraulic pressure
measurement. The frequency signals of the level sensors are connected to the digital inputs
of the RT-DAC/PCI multipurpose I/O board.
c. There are four control signals send out from the board to the multi-tank system: three valve
controls and one pump control signal. The appropriate PWM control signals are transmitted
Shouvik Chaudhuri
from digital outputs of the I/O board to the power interface, and next to the valves and to
the DC motor. The speed of the pump motor is controlled by a sequence of PWM pulses
configured and generated by the logic of XILINX chip of the RT-DAC board.
d. The liquid levels in the tanks are the system states. The general objective of the tank system
control is to reach and stabilize the water level at a desired value in the tanks by an
adjustment of the pump operation and/or valves settings. For the real system, the levels in
the tanks as well as the flow rates of the pump are limited.
2. Safety Instructions: The risk of overflow in this experiment is more because the valves can be
closed. Please be careful regarding this since mechanical systems can cause injuries or
equipment damage if over-actuated. In case of emergency, immediately press the
EMERGENCY STOP BUTTON beside the experiment, which breaks off the power
connection to the system.
3. Objective of the experiment:
The objective is to control the height of water in the first (Upper) and second (Middle) tanks (𝐻1
and 𝐻2 ). This can be done by using 3 control inputs, i.e., by controlling the pump, the first and
the second control valves (𝑞𝑖𝑛 , 𝐶1 and 𝐶2 ). In this case, the third (Bottom) tank control valve
should be kept fully open. Also, the first, second and third manual valves (𝑀1 , 𝑀2 and 𝑀2 ) [with
the red knobs] should be kept open by 25-50% approximately. Since, it is difficult to measure
the percentage of opening of the manual valves, a close approximation of that will be enough.
This system is considered as a Multiple -Input Multiple-Output (MIMO) system, with 3 control
inputs (𝑞𝑖𝑛 , 𝐶1 and 𝐶2 ) and 2 system outputs (𝐻1 and 𝐻2 ).
4. Model of the Experiment: The nonlinear model of the tank system described above, can be
written as,
𝑑𝐻1 1 1 𝛼 𝛼
𝐻̇1 = = 𝑞𝑖𝑛 − (𝐶𝑚1 𝐻1 𝑚1 + 𝐶𝑐1 𝐻1 𝑐1 )
𝑑𝑡 𝛽1 𝛽1
𝑑𝐻2 1 𝛼 𝛼 1 𝛼 𝛼
𝐻̇2 = = (𝐶𝑚1 𝐻1 𝑚1 + 𝐶𝑐1 𝐻1 𝑐1 ) − (𝐶𝑚2 𝐻2 𝑚2 + 𝐶𝑐2 𝐻2 𝑐2 )
𝑑𝑡 𝛽2 𝛽2
Here, the outflow through the control and manual valves is modeled as: 𝑞𝑜𝑢𝑡 = 𝐶𝐻 𝛼 . The
symbols in the above equations denote the following,
𝐻1 → Height of water in the first (Upper) tank

𝐻2 → Height of water in the second (Middle) tank
𝑞𝑖𝑛 → Inflow rate to the first (Upper) tank from the pump
𝐶𝑚1 & 𝐶𝑚2 → Orifice resistance of the manual valves
𝐶𝑐1 & 𝐶𝑐2 → Orifice resistance of the control valves
𝛼𝑚1 & 𝛼𝑚2 → Flow coefficient of the manual valves
𝛼𝑐1 & 𝛼𝑐2 → Flow coefficient of the control valves
𝛽1 = 𝑎𝑤 → Constant cross-sectional area of the first (Upper) tank
(𝑎 and 𝑤 can be found from the geometry of the first tank given in Fig. 2 in cm)
𝐻2
𝛽2 = 𝑐𝑤 + 𝐻 𝑏𝑤 → Variable cross-sectional area of the second (Middle) tank
2𝑚𝑎𝑥
(𝑏, 𝑐, 𝑤 and 𝐻2𝑚𝑎𝑥 can be found from the geometry of the second tank given in Fig. 2 in cm)
Shouvik Chaudhuri
5. Starting values for necessary parameters: The initial values of certain parameters necessary
for the purpose of simulation are given below. Please note that during the experiments, it
possible that these values will differ from the ones given here.
𝐶𝑚1 = 2.7e − 04; 𝐶𝑚2 = 2.4𝑒 − 04;

𝛼𝑚1 = 𝛼𝑚2 = 𝛼𝑐1 = 𝛼𝑐2 = 𝛼 = 0.5 (May be assumed)
Pump Characteristics:
PWM 0.1 0.25 0.5 0.75 1.0
𝒒𝒊𝒏 (m3/sec) 0 0 0 0.9e-04 1.5e-04
Control valve orifice resistance characteristics:

PWM 0.1 0.25 0.5 0.75 1.0
𝑪𝒄𝟏 𝐚𝐧𝐝 𝑪𝒄𝟐 0 0.2e-04 1.0e-04 1.8e-04 3.6e-04
6. Required Tasks for Simulation:

a. Simulate the system described above in MATLAB/Simulink, based on the given information in
the objective and by using the given parametric values. To have a realistic simulation, add the
characteristics of pump and the control valve orifice resistances as look-up tables in the model.
Also, add noise to the system outputs with low variance (order of 10-6). Use a suitable filter after
the output to obtain the filtered output of the system which can be further used for controller
development. Notice the following points here,
• The command send to the pump is a PWM signal between 0 and 1 which corresponds to
0 and 100 % that determines the pump opening. Hence, pump characteristic table is
required for a realistic simulation.
• The input signal given to the manual and control valves is also between 0 and 1, where
“0” means that the valve is fully closed and “1” means it is fully open.
b. Assume that 𝑟1 (𝑡) is the reference for the first tank level and 𝑟2 (𝑡) for the second tank as shown
in the figure below. Linearize the nonlinear model around a proper point and find the state space
representation of the system.
c. Design a state feedback controller for the system (the LQR approach is recommended). Apply
the controller to the simulation model and make sure the results are acceptable by tuning the
LQR gains properly.
d. After getting acceptable results for the LQR, design and use a LQI controller. Explain the
difference in the plant output for these two controllers.
Shouvik Chaudhuri
7. Required Tasks for Experiment:

a. Pump Characteristics (𝑞𝑖𝑛 ): Identify the characteristics of the pump, i.e., the relationship between
PWM signal and Pump flow rate to the first tank based on the first equation of the model. To do
this, keep both the manual and control valves for the first tank (𝐶𝑐1 and 𝐶𝑚1) fully closed. Now,
for different PWM inputs (between 0 to 1), measure the difference in volume of the water level
in the first tank (𝛽1 ∆𝐻1) with respect to time interval (∆𝑡), which gives the estimation for the
flow rate. Repeat this step for several PWM inputs to come up with a look-up table which
resembles the table of the pump characteristics in the objective.
b. Estimate the Orifice resistance of the manual valves (𝐶𝑚1 and 𝐶𝑚2): In order to find the values of
the orifice resistances for the first and second manual valves, apply a constant PWM signal to the
pump (for example 0.75), keep the control valves fully closed (𝐶𝑐1 = 𝐶𝑐2 = 0) and the manual
valves fully open. Once the water level in both tanks stabilize (𝐻̇1 = 𝐻̇2 = 0), record the water
level (𝐻1 and 𝐻2 ) in the tanks. Use the nonlinear model equations along with the above
information, to calculate the values of 𝐶𝑚1 and 𝐶𝑚2.
c. Estimate the Orifice resistance of the control valves (𝐶𝑐1 and 𝐶𝑐2): In order to find the values of
the orifice resistances for the first and second control valves, apply a PWM signal (between 0
and 1) to first control valve (𝐶1 ), a constant PWM signal to the pump (say 0.75) and keep the
manual valves fully open. Let the water level settle in both tanks (𝐻̇1 = 𝐻̇2 = 0) and record the
water level (𝐻1 and 𝐻2 ) in the tanks. Use these measurements along with other known
parameters (𝛼, 𝐶𝑚1 , 𝐶𝑚2) in the model equations, to find the orifice resistance (𝐶𝑐1). Repeat the
experiment for other values of the control valve (PWM signal) to find more points. Finally, you
can do curve fitting or simply have a look up table. Since both control valves are similar in make,
you can consider 𝐶2 to have a similar lookup table.
d. Open-loop Model Validation: Log the output response data (𝐻1 and 𝐻2 ) from the experiment
using 𝑞𝑖𝑛 = 0.65 and 𝐶𝑐1 = 𝐶𝑐2 = 0.7 as inputs. Plug in the estimated values of the manual valve
orifice resistances (𝐶𝑚1 and 𝐶𝑚2), the look-up table for control valve orifice resistances (𝐶𝑐1 and
𝐶𝑐2), and the pump characteristics (𝑞𝑖𝑛 ) look up table into the simulation and use the same input
values as in the experiment (𝑞𝑖𝑛 = 0.65 and 𝐶𝑐1 = 𝐶𝑐2 = 0.7 ) for the simulation. Then, compare
the data between the simulation and the experiment. Don’t expect to have a perfect fit because
of modelling inaccuracies and measurement errors.
e. Controller Deployment to Experiment: Use the LQR and LQI controllers developed in the
simulations on the experimental setup, along with the pump and control valve characteristics
determined above. Here you should note that,
• The tank system operates at a different sample time than the one used for the calculations
in the simulation, rate transitions may be required.
• Limitations of the actuators should be implemented by adding saturations for 𝑞, 𝐶𝑐1 and 𝐶𝑐2.
f. Record and Compare data: Record the closed loop input-output data from the experiments.
Compare them with the data set obtained from simulations.
g. Also, try to record a video of the final results along with the setup for your final exam (this is
optional).

Experiment Description - Multi-Tank System - Modified

Uploaded by

Copyright:

Available Formats

Experiment Description - Multi-Tank System - Modified

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Experiment Description - Multi-Tank System - Modified

Uploaded by

Copyright:

Available Formats

Experimental Control Systems (XCOS)

Experiment 3: Multi-Tank System (Quanser) Lab

3. Objective of the experiment:

𝐻1 → Height of water in the first (Upper) tank

𝐶𝑚1 = 2.7e − 04; 𝐶𝑚2 = 2.4𝑒 − 04;

Control valve orifice resistance characteristics:

6. Required Tasks for Simulation:

7. Required Tasks for Experiment:

You might also like