DEN5200 LABORATORY REPORT

SILVIA MEHILLI
190619147

QUEEN MARY INIVERSITY OF LONDON

SCOOL OF ENGINEERING AND MATERIAL SCIENCES
Abstract

The report purpose, controlling the water levels in a tank, will describe the ways these levels
are reached using controller parameters based on the two systems which will demonstrate
how the predicted outputs are obtained using simulations, block diagrams. These are designed
on the CE117 Process trainer for the open loop and Simulink with the CE2000 software for
the closed loop system. The report will focus in comparing the open loop step response
system and the PID controlled system and testing the effectiveness of both systems.
The Potential Integrator Derivate(PID) controller is used for majority of works within control
systems that undergo consistent alternations in a specific period of time to produce the
needed outputs from a given system. During the procedure, the recorded data will be shown
demonstrated graphically with a more detailed analysis of settling time, undershoot and
overshoots and steady state error.
Content List

1. Abstract……………………………..……………………………..………………….2
2. Introduction and background theory…………………………..……………………...2
3. Problem Setup………………………………………………………………………...2
4. Controller Design and Numerical Simulation………………………………………...4
5. Experiment Validation……………………………..…………………………………6
5.1. Results……………………………..…………………………………………….7
5.2.Calculations……………………………..……………………………..………….9
5.3. Discussion……………………………..………………………………………..11
6. Conclusion……………………………..……………………………..……………...13
7. References……………………………………………………………………………14
Introduction

The two types of systems, used to put an input into a system and a response from the very
system, are known as the closed loop and open loop. The open loop system is defined as a
control system where the control action is absolutely independent of output from the system.
Additionally, a manual control system is also an open loop control system. The other type of
control system is a closed loop system, where output decides how the input variables are
varied in order to get the most accurate value to the ideal targets set, with the PID involved.
The PID controller consists of three components and work simultaneously to set values
dependant on the characteristics they have. The following formula defines the equation of the
output of a PID controller.
#
𝐾% 𝑑𝑒(𝑡)
𝑢(𝑡) = 𝐾! 𝑒(𝑡) + 𝐾" * 𝑒(𝜏)𝑑𝜏 +
$ 𝑑𝑡
, where K defines the type of PID, 𝜏 is the time constant, e represents the error and de/dt is
the change of error with respect of time(Guan, 20121).

The report aims is to investigate the control of fluid flow and level throughout several
processes within the reservoir to the process vessel with the help of the two systems. This
report will discuss about the problem setup with its components, graphical demonstration of
the recorded data, the block diagram used for the subject and last but not least the validation
of the series of calculations with discussion. The main objectives of the experiment are listed
below:

Firstly, designing block diagrams, the two loop systems and modifying the PID
controllers with the help of MATLAB/Simulink software.
- Secondly, reporting the validate readings with graphical figures and calculations
- Thirdly, analysing and comparing readings with a professional engineering style and
resulting in an accurate conclusion on the experiment at the end

Problem Setup

The problem setup of this experiment consists of a reservoir [1], useful as a storage facility
filled all with water as needed for the experiment. As seen in the figure below, number 2 is
the pump, which is located right next to the reservoir and is the main element which initiates
the circulation of the fluid in the flow path. Additionally, the cooler [3], is placed within the
flow path and it was not used, however was present for the whole purpose of transfer from
one end to the other of each pipe. At this point the flow has reached the proportional valve
[4] which is used to control the flow of the fluid. Although the valve is left open within the
experiment and had no restrictions to the fluid flow. Close to the proportional valve there is a
flow meter (5) used to measure values of flow rate across the device. After leaving the flow
meter, the fluid is transferred in the process vessel [6], known as a transparent cylindrical
shaped device, which is used to keep the fluid at the end of the circulation and further act as a
flow supplying device when the drain valve is open. Parallel to the vessel there is a level
transmitter [7] used for measuring the voltage signal by moving the fluid level inside the
vessel. Additionally, number 8 and 9 represent the air vent valve and drain valve
respectively, and where kept opened during the experiment procedure to stop the pressure
build up in the vessel.
Figure 1 and 2: Representation of open loop system and closed loop system(Alleman, 2019).

This was exemplified based on the position of the valves, so if the valves’ direction were
perpendicular to one another, this meant there was a pressure build up occurring, and if the
valves were facing the same direction of the flow, then this would signify no pressure build
up. Throughout the cylinder, there was a heat exchanger [10] which did not have any
significant importance, even though it was important that when the fluid was filling the
vessel, the height of water should reach slightly above the heat exchanger in order to
minimise the possibility of interferences with the level of water in respect of the height to
volume ratio. Inside the process vessel there is the ruler [11], which was used to measure the
height level with the unit of centimetre. Additionally, the apparatus and the computer were
connected with each other through the CE117 process trainer which is demonstrated in the
figure below as n umber 12. As it can be seen in the setup figure, the red wires are installed to
convert the voltage signals from analogue current to digital current, whereas the black wires
were used to do the opposite. In this experiment, a data acquisition software named CE2000
was installed and paired with the apparatus with the help of a computer [13], in order to
transfer the readings onto the software whilst the system was functioning. This instalment
enabled further analysis used for analytical tools such as tables and graphs, which will be
shown during this report.

Figure 3: The setup diagram of this experiment.

This experiment operated in such a that the level of water flow was measured in the form of
pump voltage with the help of both the open loop and closed loop system. The closed loop
system did use a PID controller, whereas for the open loop system a controller was not
needed. This was tested by using Simulink within the MATLAB software so that block
diagrams of the system would be produced and the values for the three components would be
tested in order to identify the most accurate values for the system to have the desired
response. When talking about the control target, the PID values should adapt for the
parameters so that the control targets are satisfied to obtain the most accurate response. For
the sake of this, possibilities such as fast rising time, overshoot, least steady or no steady
reference signal changes must not overcome 10, with the given step reference signal between
6-6.5 V.

Figure 4: The Closed-loop Circuit, which is used in the 2nd experiment (Boult, 2020).

Controller design and numerical simulation

As previously explained, the feedback loop will deliver the error to the PID controller, which
will automatically reduce the error to keep the system output consistent. The PID controller
consists of three components that work together to minimise the error, shown in the diagram
below. Moreover, the gain in each component of the PID controller was adjusted in this
experiment so that it examines all the changes that occurred to the system.

Figure 5 – The following components of PID controller: Proportional Kp, Integral KI and
Derivative KD (Robots, 2019).

The first component which is known as the Proportional part of the controller will be the first
one to work on reduce the large errors in the system. At the same component, the gain Kp can
be commanded manually to maximise or minimise the amount of error that the proportion
branch must deal with. Even though the error within the proportional branch shows zero there
might still be error present. Consequently, the Integral component will have to focus and
minimise the very small errors that are still present after the use of proportional branch.
Additionally, the Integral component also acts as a memory device as it takes all pervious
errors and integrates them over time. This controller starts accumulating the error values in
the procedure of erasing the error values. As the integral component pushes the error to zero,
there will be some troubleshoot in the system. Consequently, the Derivative branch of the
controller is used reduce this troubleshoot and thus allowing for the error in the system to
somewhat disappear (Robots, 2019). At this point the derivative component will have to deal
with any sudden changes to system that has the potential to cause any errors.

The PID controller on Simulink

Figure 6 shows a PID controller created in Simulink as part of a closed-loop negative

feedback system. The PID controller could be easily tuned since each part of gain was
manually commanded on MATLAB. The setpoint values were used to configure the step
function of the circuit, consisting of a 6V initial value and a 6.5V final value. The step time
did set at 174.5 seconds, the time it took for the pressure vessel's height to change from Level
A to Level B. As it can be seen on figure 7, the graphs were produced on Simulink when the
simulation was completed, using the values of the PID as shown in the figure below.
KP = 15 KI =0.5 KD=0

Figure 6: The 2nd procedure block diagram with the PID controller.

Figure 7: The graph produced on Simulink for Procedure 2.

Checking Stability

The Bode plot below was produced using Simulink to check the stability of the closed-loop
system. For this system to be stable the phase at the cross over frequency where at the
magnitude of 0 dB should not have reached -180°. Based in figure 8 it can be said that at the
crossover frequency, at a magnitude of 0dB, the phase reached about -70°. Thus, showing
that the closed-loop system is stable.

Figure 8 – Diagram of Bode Plot of the closed loop system which checks stability.

Procedure 1 Simulink Setup

The first procedure (Figure 9) for the open loop circuit was run using simulation through
Simulink. The step time for the step function in this circuit was set at 652 seconds, which is
the time at which the pressure vessel’s water level started rising from Level A.
The step function’s initial and final values was set as 3.465V and 3.635V, respectively,
corresponding to the flow rate values at Level A and B of the pressure vessel and the stop
time of the simulation was settled at 1400 seconds. After running the simulation via
Simulink, a graph was developed, from which the simulation's time constant was determined,
using the cursor measurement function on Simulink which is attached next to the graph
below.

Figure 9: The block diagram for the Procedure 1 represented on Simulink.

Figure 10: The graph produced of the first procedure on Simulink with of the time constant of
143 seconds.

Experiment Validation

The experimental lab results were plotted in an excel graph, as shown in Figure 11. This is
the same graph that would have been created on the CE2000 Software, that is used for
calculating both the experimental and theoretical time constant values. Furthermore, this
Simulink Time constant is compared to the Simulink Time constant value, which was found
to have 143 seconds (Figure 10).

CE2000 software
9
8
7
6
VOLTAGE (V)

5
4
3
2
1
0
0 200 400 600 800 1000 1200 1400
TIME (S)

Pump 2 (V) LT (V) Flow Rate (V)

Figure 11: The graph that could be produced on the CE2000 Software.

1st Procedure: Sample calculations

As it can be seen, the graph above shows the different values at Level A and Level B and the
difference in height and flow rate between the 2 water levels. According to this data the
theoretical time constant was obtained using the following calculations below.

Table 1 – The calculated parameters used during this procedure

Test1 Flow Rate (V) Flow Rate (L/Minute) Flow Rate q ((𝑚! )/s) h (m)
Level A 3.465 3.465 5.78 × 10"# 0.105
Level B 3.635 3.635 6.06 × 10"# 0.127
Difference (D) 2.83 × 10"$ 0.022
S (𝑚% ) 0.017671459
R 7764.705882
Simulink τ (s) 143.000
Theoretical 𝜏 (s) 137.2136791
Experimental 𝜏 (s) 158.7
Percentage Difference of 𝜏 15.66%

The equation below is used to calculate the area using the given diameter of 0.150m as shown
in the lab handout.
$.()$ &
𝑆(𝐴𝑟𝑒𝑎) = 𝜋𝑟 & → 𝜋 4 &
5 = 0.01767𝑚& (I)

By using the following equation, the value of R was calculated by substituting the values
obtained from figure 11. Afterwards the theoretical value of time constant was calculated as
shown in equation 3.
∆ℎ 0.022
𝑅 = = = 7764.80. (II)
∆𝑞 2.8333 × 10*+

𝜏 = 𝑆𝑅 = (0.01767 × 7764.797) = 137.21 𝑠𝑒𝑐𝑜𝑛𝑑𝑠 (III)

Few calculations are needed to obtain the experimental time constant value. Firstly, the
change in the LT values is calculated as the water level changed from A to B:

∆𝐿𝑇 = 𝐿𝑇(𝐵) − 𝐿𝑇(𝐴) → 8.57 − 7.62 = 0.95V (IV)

Following, LT value at the time constant is calculated using equation 5. Based on the
previous section it was evaluated that the time constant occurred when there is a 63.2%
height change between A and B, which is exemplified by the following equation:

𝐿𝑇 𝑎𝑡 𝑡𝑖𝑚𝑒 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝐿𝑇(𝐴) + (0.632 × ∆𝐿𝑇) = 7.62 + (0.632 × 0.95) = 8.22𝑉 (V)

Furthermore, it can be said that the time when LT reached 8.22V, is 810.7 seconds. Thus, the
experimental time constant is obtained by subtracting this value with the time of 652 seconds
at level A.
𝜏,-!,./0,1#23 = 810.7 − 652 = 158.7 𝑠𝑒𝑐𝑜𝑛𝑑𝑠. (VI)
 It is important to calculate the percentage difference between the experimental and theoretical
𝜏, which is calculated using the formula below:
4!"#!$%&!'()* *4(+!,$!(%-)* ()5.6*(76.&(
% 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 = 4(+!,$!(%-)*
=4 (76.&(
5 × 100 = 15.66%. (VII)

2nd Procedure: Performance Criteria

As the procedure was at the end, the 12 sets of data were exported to Excel so that the
performance criterion values would be determined. Following there are listed the values
included:

1. The rise time, which us defined as the time taken for the LT value to jump from
10% to 90% of the final setpoint value of 6.5V.

2. The Percentage Overshoot, known as the percentage difference between the Peak LT
value to the final set point value of 6.5V.

3. The setting time that the LT value sets within a ± 2% region to the final set point
value.

4. The steady state error, which determines the difference between the final value of the
system and the final set point value(6.5V).

Table 2: Presenting the displaying performance criteria values and PID values for the 12
closed loop data sets.

Rise Overshoot(%) Settling time (s) Steady state error P I D

Time(s)
CL 1 7.92 2.34 26.90 0.00 10.00 0.50 0.00
CL 2 6.80 2.42 25.40 0.00 15.00 0.20 0.10
CL 3 7.30 1.15 7.20 0.01 15.00 0.50 0.00
CL 4 6.90 2.62 25.80 0.00 15.00 1.00 0.00
CL 5 19.80 0.00 15.10 0.01 5.00 0.00 0.00
CL 6 6.24 3.29 37.10 0.01 5.00 1.00 0.01
CL 7 8.70 0.54 7.60 0.01 10.00 0.20 0.02
CL 8 6.80 1.00 5.80 0.01 10.00 0.40 0.00
CL 9 6.90 1.11 5.70 0.01 12.00 0.50 0.02
CL 10 19.80 0.00 15.10 0.01 15.00 0.60 0.06
CL 11 6.90 1.72 6.60 0.00 20.00 1.00 0.08
CL 12 9.24 2.16 23.80 0.01 25.00 0.00 0.01
Excel Plot of the closed loop plot

The graph below was produced by one of the data sets to allow comparison with the
simulation on Simulink. It can be said that the closed-loop 3 PID values were identical with
the ones chosen for the simulation, which allowed a comparison between the experimental
and simulation results.

5
Voltage (V)

0
0 50 100 150 200 250 300
Time (s)
LT Set Point

Figure 13: The graph plot for the data of the closed-loop 3.

Discussion

According to the data of Table 1 it can be stated that the time constant taken from Simulink at
143s was closer to the theoretical value of 137.21s than the experimental value of 158.7s.
This can be explained by the fact that the simulation did not have the capacity to take into
consideration any of the real-life issues in the lab, thus resulting in the simulation being
highly closer to the theoretical value.

During the lab experiment there was a difficulty on reading off process vessel which caused
the higher experimental time constant value. The readings obtained for Level A and B were
different to the actual height, this came because of using a ruler for the measurement of the
water level, which caused parallax errors.
Additionally, due to any possible fault in the actual equipment, the calculated values carried
further errors, which affected the accuracy of the results. As we already know Simulink does
not take into consideration any lab issues, which can be explained by the large percentage
difference of 15.66% between experimental and theoretical value. Additionally, the lack in
repetition of this procedure validates the results as considerably unreliable and anomalies
present.

Thus, a main suggestion for a future experiment would be few repetitions, to reduce any
inaccuracies in values. Additionally, testing the open loop system at a greater vary of voltage
than the 0.5V, to prove if it would positively affect the system. These suggestions were made
so that the percentage difference produced would hit less that 5% for the theoretical and
experimental data to be considered in an agreement.
It can be seen from table 2 that the Integral component contributed to reduction of the steady
state error within the system. This can be exemplified by the closed loop 3 and 4 as the
Proportional gain and Derivative gain constants remained identical, whereas in the loop 4 the
Integral gain was doubled. As a consequence, this increase in the gain, the steady state error
droped from 0.01 to 0. Although, at same time the increased integral gain led to greater
overshoot, increasing by a percentage overshoot of 1.15% to 2.62%. The effect of the
increasing Proportional Component can be analysed by keeping I and D constant and
monitoring Closed Loop 1 and 3. For example, there was a decrease at the rise time vale
from 7.92s to 7.30 seconds, meanwhile a considerable decrease in the settling time and a
change from 26.9s to 7.2 seconds with the setting time. This came as e result of increasing the
Proportional Gain, helping the system run much faster with the system settling in a smaller
period. Apart from that, increasing the gain caused there to be a steady state error in closed
loop 3, as shown in Table 2.

When observing Closed loop 3 and 10, it is possible for the effects of Derivative Component
of the controller to be analysed. The increase of the derivative gain from 0 to 0.06 made the
system ran at a much slower step. For example, the results at Table 2 indicate the increased
rise time from 7.3 to 19.8 seconds and the increase from 7.2 to 15.1 seconds of the settling
time. The increase of the derivative gain caused the decrease to 0 of the overshoots within the
system. As it can be seen in Table 2, the main advantage of using the PID controller was that
it massively reduced the settling time of the system. This is proved by the fact that the
settling times in 2nd procedure were way lower than the time constant of the first procedure.
Thus, it can be stated that the PID controller is crucial in settling the stability of the system
through the reduction of error within the system. In more details this was enabled because the
feedback-loop did automatically create a dependency between the output value to the
controller, so the system would be faster(Alleman, 2019). Additionally, a second advantage
of the usage of the PID controller was that it was able to reduce the steady state error between
the LT and set point values, which is clearly shown in figure 7 and 12, where there is a
minimal error in the system. Being in contradiction with the open loop system where a large
steady state error was present in the system. This phenomenon can be seen in the figure 11
shown above. The PID controller is highly suggested to be used in this experiment as even
implementing it in the system is very easy and the cost of this controller is cheap.

Conclusion

In conclusion, the experiment is considered as a success as the aim of achieving both

precision and accuracy of the values collected from both the open and closed loop systems,
which were further used to decide which of the systems was best fit. As it was seen before,
the graphs and simulations that were created using Microsoft Excel and MATLAB
SIMULINK successfully delivered and illustrated the behaviour of the flow and the water
level in the system starting from the process vessel to the reservoir.

After a thorough evaluation, the open loop showed a more stable system compared to the
closed loop system. Meaning that this case contradicts the predictions of this coursework
reason being of the PID values not being as effective as possible. When observing the PID
sets it was shown that the 9th PID set had the best readings out of the all the PID sets and
moreover was found to be better than an open loop system. Meaning that it appeared to react
faster and reach the set-point sober than the open loop system. This report concluded that the
usage of PID controllers is a very good option when the chosen values are significantly
accurate as inaccurate values would possibly result in an unstable system thus causing errors
for any scenario being used.
Reference List

Alleman, S(2019). Closed Loop system. Available at:

https://www.researchgate.net/figure/Open-and-Closed-Loop-Systems-Both-can-be-found-in-
the-project-management-domain-But_fig2_2537163. (Accessed on: 29.04.2021).

Astrom, K. (2002). PID Controller. Available at:

https://www.cds.caltech.edu/~murray/courses/cds101/fa02/caltech/astrom-ch6.pdf.
(Accessed on: 0305.2021)

Boult, J. (2020). What is Closed-Loop Control System? [online] Electronics Coach. Available
at: https://electronicscoach.com/closed-loop-control-system.html (Accessed on 25.04.2021).

Guang, L (2021). Water tank level control lab handout. Controls Systems Analysis and
Design Qmplus: United Kingdom. (Accessed on: 02.05.2021).

Robots, P., (2019). PID Control for Robotics. Programming Robots. Available at:
<https://mjwhite8119.github.io/Robots/pid-control. (Accessed on: 28.04.2021).