Paschals Paper

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/331535478
SIMULATION AND CONTROL OF A REACTOR FOR THE NON-CATALYTIC

HYDROLYSIS OF ETHYLENE OXIDE TO ETHYLENE GLYCOL
Research · March 2019

DOI: 10.13140/RG.2.2.14841.72801
CITATION READS
1 9,412
2 authors, including:
Paschal Onuorah
Plurimus Technologies, Nigeria
2 PUBLICATIONS 3 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Mathematical Modelling and SImulation View project
Chemical Engineering Process Design View project
All content following this page was uploaded by Paschal Onuorah on 05 March 2019.
The user has requested enhancement of the downloaded file.

Mathematical Theory and Modeling www.iiste.org
ISSN 2224-5804 (Paper) ISSN 2225-0522 (Online)
Vol.8, No.2, 2018
SIMULATION AND CONTROL OF A REACTOR FOR THE NON-CATALYTIC

HYDROLYSIS OF ETHYLENE OXIDE TO ETHYLENE GLYCOL
1
Akpa, Jackson Gunorubon, Onuorah, Paschal2
Department of Chemical/Petrochemical Engineering
Rivers State University,
Port-Harcourt, Rivers State, Nigeria
1
(corresponding author: akpa.jackson@ust.edu.ng), 2(onuorahpaschal@gmail.com)
Dynamic Models for predicting the concentration profiles of a Continuous Stirred Tank
Reactor for the Non-Catalytic hydrolysis of Ethylene Oxide to Mono Ethylene Glycol (MEG)
have been developed. The models validated by solving model equations using ode45 solver of
Mathlab 7.5 with industrial plant data from Batiha, (2004a) had a deviation of 2.47% and
0.93% for ethylene oxide and water respectively when the model outputs were compared with
the industrial plant outputs. Simulations of reactor to (±) step changes in the inlet flowrates
showed great effect on Mono-Ethylene Glycol production. The dynamic models were used for
control studies by incorporating a Proportional-Integral-Derivative (PID) feedback Controller.
The closed loop response of the reactor output showed continuous oscillatory response with
offset. Hence the controller parameters (proportional gain 𝐾𝑐 , Integral time 𝜏𝐼 and derivative
time 𝜏𝐷 ) were tuned using the ‘On-Line Trial and Error Method’ implemented using MathLab
Simulink to obtain optimum values that ensured quick stability of the closed-loop system,
reduced or no oscillatory response and no offset. The optimum controller parameter:
proportional gain 𝐾𝑐 = 0.1, Integral time 𝜏𝐼 = 14 and derivative time 𝜏𝐷 = 3 were obtained
and the reactor attained stability in about 180 seconds
key words: simulation, control, Ethylene Glycol production, reactor model
1. Introduction
Ethylene Glycol (Ethane -1, 2- diol) is an organic compound primarily used in the
manufacture of Polyester, Fibres, Fabric, solvents for paints and plastics, as hydraulic brake
fluids and in the preparation of Polyethylene Terephthalate (PET) resin used in making plastic
bottles for soft drinks (Landau & Ozero, 1982). Ethylene Glycol can be manufactured
industrially via the catalytic process where the reaction is catalysed by either an acid or base
or the non-catalytic process where the reaction occur at neutral pH and at elevated
temperatures.
The direct un-catalyzed neutral hydrolysis of ethylene oxide to ethylene glycol at high
temperature is the most widely used method (Samoilov & Mnushkin, 2012); in this process,
ethylene oxide is converted to ethylene glycol through uncatalyzed neutral (pH 6 –10)
23
Vol.8, No.2, 2018
hydrolysis (with excess water) at high temperatures and pressures (Rebsdat and Mayer 2005).
The primary product is mono-ethylene glycol (MEG) with higher glycols such as di and tri
ethylene glycols (DEG and TEG) formed in smaller amounts as the reaction proceeds through
a series of consecutive – parallel reactions. The use of excess water to control selectivity
towards Mono-ethylene glycol has been reported by Yang, et al., (2010) to increase
production cost due to high energy consumption during the distillation process to remove
water. The use of efficient catalysts that uses low water/ethylene oxide ratio, thus reduce cost;
has also been reported - catalytic processes where Ethylene Oxide is hydrolyzed over various
catalyst to produce ethylene glycol. These processes incudes ethylene oxide hydrolysis over:
Phosphorus-based Halide Catalysts (Kawabe, 2000); anion exchange resins such as
halogenate or bicarbonate-type: Strickler et al, (1999); polymer of a vinylaromatic compound:
Iwakura and Miyagi, (2000); HCO3-/CO32- catalyst form: Shvets, et al., (2005); supported
metal oxides and zeolites: Li, et al., (2004), Li, et al., (2005), Li, et al., (2006); polymeric
organosilane ammonium salts: Kruchten, (1999a) and macrocyclic chelating compounds:
Kruchten, (1999b). Other production methods include the direct oxidation of ethylene and
synthesis from carbon monoxide (Rebsdat and Mayer 2005) and the direct catalytic hydrolysis
of methanol and its derivatives such as formaldehyde to ethylene glycol (Celik et al., 2008),
this method has been reported to be cheaper as the carbon needed for this process is derived
from synthesis gas (obtained from natural gas, coal, or biomass), a cheaper carbon source than
petroleum-derived ethylene. A typical process scheme shown in Figure 1 is complex
consisting of a reactor, an evaporator and columns/vaporizers amongst other equipment.
Samoilov, & Mnushkin, (2012) proposed the use of a single apparatus – the reactive
distillation column for this process.
24
Vol.8, No.2, 2018
1.1. Process Description: Hydrolysis of Ethylene Oxide
A flow diagram of the manufacture of ethylene glycol through the un-catalyzed hydrolysis of
ethylene Oxide (Arrieta, 2001) is shown in Figure 1.
Figure 1: Flow diagram for a plant for hydrolysis of Ethylene Oxide (Arrieta, 2001).
a. Reactor b. Evaporator c. MEG Column d. DEG Column e. TEG Column f. Reboilers
Ethylene Oxide stream is first contacted with Water in a gas absorber at an Ethylene Oxide
(EO) to Water ratio of about 1:20 (to reduce the formation of higher homologues). The
Ethylene Oxide solution is preheated to a temperature of about 200oC and fed to a Continuous
Stirred Tank reactor (CSTR) where it is thermally hydrolyzed to Ethylene Glycol at an
operating temperature of about 190 – 200oC and pressure of about 14 – 22 atm. Ethylene
Oxide reacts further with Mono-ethylene Glycol (MEG) and higher homologues in a series –
parallel reaction to form di-ethylene and tri-ethylene glycol. The Water – Ethylene Glycol
mixture from the reactor is fed to the first stage of a multi-stage evaporator, where it is re-
boiled using high-pressure steam then sent to a sequence of columns where the ethylene
glycol is separated to mono, di, and tri ethylene glycol respectively.
Ethylene Oxide reacts faster with Ethylene Glycol than it does with Water (Melhem, et al.,
2001), hence without control, reaction proceeds normally with the successive formation of
25
Vol.8, No.2, 2018
higher homologues of glycol (di, tri, tetra ethylene glycol) through the continuous reaction of
ethylene oxide with the ethylene glycol homologue formed – by addition polymerization to
form higher ethylene glycols. The most important process variable affecting the glycols
(mono, di or tri ethylene glycol) distribution is the water to ethylene oxide ratio in the starting
material and only marginally on the reaction pH and temperature (Rebsdat & Mayer, 2005).
Hence an excess of water is usually used in commercial plants (almost 20 times more) for
high selectivity and to achieve at least a 90% conversion of Ethylene Oxide to mono ethylene
glycol (MEG) (Arrieta, 2001).
Therefore, to stop the continuous reaction, ensure selectivity and maximum mono ethylene
production (maximum volume for MEG), it is expedient that some form of control
mechanism be implemented to maintain the required Ethylene Oxide to Water ratio of the
feed entering the reactor. That is – maintain maximum volume of the mono ethylene glycol
(MEG) measured by the MEG/liquid level in the reactor (output variable) as changes in the
disturbance - inlet flowrate of the reactants occurs by manipulating the ethylene oxide to
water ratio – the manipulated variable. This involves the installation of appropriate controller
which returns the output variable back to its set point value whenever there is a disturbance
(changes in the inlet flowrate) in the process. Hence this work studied the behavior of the
continuous stirred tank reactor (CSTR) used for the hydrolysis of ethylene oxide to ethylene
glycol particularly to fluctuations in the feed flowrate and develop appropriate closed loop
models with Proportional-Integral-Derivative Controller capable of bringing the system
back/close to its original set point value and ensure stability of the process.
2. Material and Methods
2.1. Methods
The mathematical models of the continuous stirred tank reactor (CSTR) were developed by
applying the principle of conservation of mass and energy on the reactor. These models (open
26
Vol.8, No.2, 2018
loop process models) were ordinary differential equations and were solved with industrial
process data from Batiha, (2004a) and model parameters from Sinnott & Towler, (2009) using
the ode45 solver from MATLAB 7.5.0. The open loop models were simulated to determine
the effects of step-changes of ±25% and ±50% in the feed flowrate on the reactor
performance – the quantity or volume of mono ethylene glycol produced. The control
objective was to ensure maximum production (volume) of MEG in the reactor which was
measured by the liquid/mono ethylene glycol (MEG) height/level for a reactor of constant
cross-sectional area as fluctuations in the inlet flowrate occur. This was achieved by
manipulating the feed inlet flowrate through the ethylene oxide – water ratio of the feed. The
feedback control configuration with Proportional-Integral-Derivative (PID) controller was
used to effect the control action. A closed loop block diagram of the process was developed
from which the closed loop model of the process output (MEG height) was developed. The
closed loop model equation of the process was developed in terms of the transfer functions of
the components of the closed loop block diagram. The close loop model was used to study the
effects of the PID controller to step changes in the inlet flowrate on the liquid (MEG) level in
the reactor. To effectively ensure this control – minimize the effects of disturbances, obtain
rapid and smooth responses to set point changes with little or no oscillations and eliminate
steady state error (offset); the optimum values of the controller parameters (controller gain
𝐾𝐶 , the integral or reset time 𝜏𝐼 and the derivative time 𝜏𝐷 ) had to be determined (controller
tuning) - a systematic selection of controller parameter values that achieves control
objectives; an iterative procedure using the On-Line Trial and Error Method (Luyben, 1990)
implemented with MathLab Simulink was developed. In this procedure, several/different
combinations of the controller’s parameters values were used in solving the closed loop
model to obtain the response of the output/controlled variable (liquid/MEG level) to (±) step
changes in the disturbance (inlet feed flowrate). The combination values of these parameters
27
Vol.8, No.2, 2018
that stabilizes the output (MEG level) in the reactor, nullifies the effect of the disturbance -
maintains a constant/maximum value of the liquid/MEG level (the desired/set point value) is
the optimum and required controller setting for effective control of the process.
2.2. Model Development
2.2.1. Kinetic Model
The elementary steps of the series-parallel reactions for the non-catalytic process of Ethylene
Glycol synthesis are as follows - reaction of Ethylene Oxide with water to give Mono
Ethylene Glycol (MEG) and subsequent reaction with ethylene oxide to form di and tri
ethylene glycol:
𝐾1
𝐶2 𝐻4 𝑂 + 𝐻2 𝑂 → 𝐶2 𝐻6 𝑂2 (1)
𝐾2
𝐶2 𝐻4 𝑂 + 𝐶2 𝐻6 𝑂2 → 𝐶4 𝐻10 𝑂3 (2)
𝐾3
𝐶2 𝐻4 𝑂 + 𝐶4 𝐻10 𝑂3 → 𝐶6 𝐻14 𝑂4 (3)
The rate expressions for any component j from equations (1), (2) and (3) can be written
For j = A,B,C,D, E: A represent 𝐶2 𝐻4 𝑂; B: water; C: 𝐶2 𝐻6 𝑂2; D: 𝐶4 𝐻10 𝑂3; E: 𝐶6 𝐻14 𝑂4 as:
(−𝑟𝐴 ) = −𝐾1 𝐶𝐴 𝐶𝐵 − 𝐾2 𝐶𝐴 𝐶𝐶 − 𝐾3 𝐶𝐴 𝐶𝐷 (4)
(−𝑟𝐵 ) = −𝐾1 𝐶𝐴 𝐶𝐵 (5)
(𝑟𝑐 ) = 𝐾1 𝐶𝐴 𝐶𝐵 − 𝐾2 𝐶𝐴 𝐶𝐶 (6)
(𝑟𝐷 ) = 𝐾2 𝐶𝐴 𝐶𝐶 − 𝐾3 𝐶𝐴 𝐶𝐷 (7)
(𝑟𝐸 ) = 𝐾3 𝐶𝐴 𝐶𝐷 (8)
The reaction rate constant (𝐾𝑖 ) for any reaction path i is Arrhenius law dependent given as:
𝐸𝑖
𝐾𝑖 = 𝑘𝑖0 𝑒 −𝑅𝑇 (9)
The rate constant for reaction 1 as a function of temperature is given by Batiha, (2004b):
−21193
𝐾1 = 7.123𝐸 + 08 × 𝑒𝑥𝑝 ( 𝑅𝑇
) (10)
28
Vol.8, No.2, 2018
The ratios of the rate constants for the reaction paths as given by Batiha, (2004b) are:
𝐾2 𝐾3
= 2.946, = 3.229 (11)
𝐾1 𝐾1
2.2.2. Reactor Model
The following simplifying assumptions were made in the development of the process model:
The reactants are perfectly mixed, hence outlet and inlet condition of the reactor are the same,
density is constant and heat capacity varies negligibly with temperature, the reactor is
perfectly insulated and there is no heat loss. The continuous stirred tank reactor used for the
hydrolysis reaction is represented by a single input - single output system as shown in Figure
2.
𝐶𝑗 , 𝑞𝑖 , 𝑇𝑖
𝑜
𝐶𝑗 , 𝑞, 𝑇
Figure 2: Schematic of the CSTR
The Continuous Stirred Tank Reactor (CSTR) model was obtained by applying the
conservation principle on the fundamental quantities which characterize the system: Total
mass, Component mass and Total Energy as follows:
i. Total mass balance
A Total mass balance on the CSTR gives:
𝑑(𝜌𝑉)
= 𝜌𝑖 𝑞𝑖 − 𝜌𝑞 (12)
𝑑𝑡
But 𝜌𝑖 = 𝜌; 𝑉 = 𝐴ℎ,
Substituting and rearranging gives:
29
Vol.8, No.2, 2018
𝑑ℎ
𝐴 𝑑𝑡 = 𝑞𝑖 − 𝑞 (13)
ii. Component mass Balance
A component mass balance for any component (j) of the CSTR yields:
𝑑(𝑉𝐶𝑗 )
= 𝑞𝑖 𝐶𝑗𝑜 − 𝑞𝐶𝑗 + (𝑟𝑗 )𝑉 (14)
𝑑𝑡
substituting 𝑉 = 𝐴ℎ and expanding gives:
𝑑𝐶𝑗 𝑑ℎ
𝐴 [ℎ + 𝐶𝑖 𝑑𝑡 ] = 𝑞𝑖 𝐶𝑗𝑜 − 𝑞𝐶𝑗 + (𝑟𝑗 )𝐴ℎ (15)
𝑑𝑡
Substituting equation (13), rearranging and simplifying gives:
𝑑𝐶𝑗 𝑞𝑖
= (𝐶𝑗𝑜 − 𝐶𝑗 ) + (𝑟𝑗 ) (16)
𝑑𝑡 𝐴ℎ
iii. Energy Balance
The principle of Conservation of energy applied on the reactor gives:
𝑑(𝜌𝑉𝐶𝑝 𝑇)
= 𝜌𝑞𝑖 𝐶𝑃 𝑇𝑖 − 𝜌𝑞𝐶𝑃 𝑇 + (−∆𝐻𝑟 )(𝑟𝑗 )𝐴ℎ (17)
𝑑𝑡
Expanding and assuming 𝜌 and 𝐶𝑃 vary negligibly with temperature, equation becomes:
ℎ𝑑𝑇 𝑇𝑑ℎ
𝜌𝐴𝐶𝑃 [ 𝑑𝑡 + ] = 𝜌𝐶𝑃 (𝑞𝑖 𝑇𝑖 − 𝑞𝑇) + (−∆𝐻𝑟 )(𝑟𝑗 )𝐴ℎ (18)
𝑑𝑡
Substituting eqn. (13) and simplifying gives:
𝑑𝑇 𝑞𝑖 (−∆𝐻𝑟 )
= (𝑇𝑖 − 𝑇) + (𝑟𝑗 ) (19)
𝑑𝑡 𝐴ℎ 𝜌𝐶𝑃
2.3. Process Control
The Control Objective was to ensure the liquid/MEG level in the reactor is maintained at a
maximum pre specified level (is stable - does not fluctuate with time) in the face of ever-
changing disturbances (step changes in the inlet flow rate – a Servo problem). The Closed
loop diagram of the process is shown in Figure 3.
30
Vol.8, No.2, 2018
𝐶𝑗 𝑜 , 𝑞𝑖 , 𝑇𝑖 Gc (s)
C
hm
DPC hsp
- +
𝐶𝑗 , 𝑞, 𝑇
Figure 3: Closed-Loop diagram of the Process.
*DPC – Differential Pressure Cell
The general model equation for the output (liquid/MEG level, ℎ(𝑆) ) of the closed loop process
in the Laplace domain due to set point (ℎ𝑆𝑃(𝑆) ) and load/disturbance (𝑑(𝑠) ) changes in terms of
transfer functions of the various components in Figure 3 was developed to give:
𝐺𝑓(𝑠) 𝐺𝑐(𝑠) 𝐺𝑝(𝑠) 𝐺𝑑(𝑠)

ℎ(𝑆) = ℎ𝑆𝑃(𝑆) + 𝑑(𝑠) (20)
1+ 𝐺𝑓(𝑠) 𝐺𝑐(𝑠) 𝐺𝑚(𝑠) 𝐺𝑝(𝑠) 1+ 𝐺𝑓(𝑠) 𝐺𝑐(𝑠) 𝐺𝑚(𝑠) 𝐺𝑝(𝑠)
For the servo control problem where the load does not change (𝑑(𝑠) = 0), while the set point
undergoes step changes; the feedback controller acts to keep the liquid level (ℎ(𝑆) ) close to the
set point (maximum MEG height) (ℎ𝑆𝑃(𝑆) ). The output ℎ(𝑆) from equation (20) becomes:
𝐺𝑓(𝑠) 𝐺𝑐(𝑠) 𝐺𝑝(𝑠)

ℎ(𝑆) = ℎ𝑆𝑃(𝑆) (21)
1+ 𝐺𝑓(𝑠) 𝐺𝑐(𝑠) 𝐺𝑚(𝑠) 𝐺𝑝(𝑠)
The transfer function of each component was obtained by expressing the component model
equation in deviation variables form, laplace trasform the resulting equation and rearrange the
final equation in the input-output form. The transfer function for the various components in
the closed loop diagram (Figure 3) were therfore obtained following this procedure as
follows:
31
Vol.8, No.2, 2018
i. Process
The model of the liquid level at any time for a P% magnitude step change in inlet flowrate is
given as:
𝑑ℎ
𝐴 𝑑𝑡 = 𝑃𝑞𝑖 − 𝐴ℎ (22)
where: 𝑞 = 𝐴ℎ
This equation can be written in terms of deviation variables and rearranged as:
𝑑ℎ′
𝐴 + 𝐴ℎ′ = 𝑃𝑞′𝑖 (23)
𝑑𝑡
Taking the Laplace transform and rearranging equation gives the Transfer function as:
̅′ (𝑠)
ℎ 𝑃
𝐺𝑝 (𝑠) = = 𝐴(𝑆+1) (24)
𝑞′𝑖 (𝑠)
ii Controller: Proportional-Integral-Derivative (PID) Controller
The model equation for the controller in deviation variables is:
1 𝑡 𝑑𝜀̅ ′ (𝑡)
𝐶̅ ′ (𝑡) = 𝐾𝑐 (𝜀̅′ (𝑡) + 𝜏 ∫0 𝜀̅ ′(𝑡)𝑑𝑡 + 𝜏𝐷 ) (25)
𝐼 𝑑𝑡
Equation (25) was laplaced and rearranged to obtain the Transfer function as:
𝐶̅ ′ (𝑠) 1
𝐺𝑐 (𝑠) = = 𝐾𝑐 (1 + 𝜏 𝑠 + 𝜏𝐷 𝑠) (26)
𝜀̅ ′ (𝑠) 𝐼
iii Measuring Device (Differential Pressure Cell)
The mathematical models for the Differential Pressure Cell; Stephanopoulos, (2012) is:
𝐴𝐿𝜌 𝑑2 𝑥 𝐶 𝑑𝑥 𝐴
(𝐾𝑔 ) 𝑑𝑡 2 + 𝐾 𝑑𝑡 + 𝑥 = 𝐾 ∆𝑃 (27)
𝑐
Where, 𝐴 = Cross-sectional area of the Capillary tube,
𝐿 = Length of the Capillary Tube,
𝜌 = Density of the Liquid,
32
Vol.8, No.2, 2018
𝑥 = Fluid displacement in the Capillary Tube,
𝐾 = Hooke’s constant for the Diaphragm,
𝑔𝑐 = Conversion Constant,
𝐶 = Damping Coefficient,
∆𝑃 = Differential pressure across the cell.
This equation shows that the system’s response follows a second-order dynamics and can be
written in standard form as:
𝑑2 𝑥 𝑑𝑥
𝜏 2 𝑑𝑡 2 + 2𝜁𝜏 𝑑𝑡 + 𝑥 = 𝐾𝑃 ∆𝑃 (28)
𝐴𝐿𝜌 𝐶 𝐴
Where: 𝜏 2 = (𝐾𝑔 ) , 2𝜁𝜏 = 𝐾 , 𝐾𝑃 =
𝑐 𝐾
If 𝑥 = ℎ𝑚 ; Δ𝑃 = 𝛽ℎ equation becomes:
𝑑2 ℎ𝑚 𝑑ℎ𝑚
𝜏2 + 2𝜁𝜏 + ℎ𝑚 = 𝐾𝑃 𝛽ℎ (29)
𝑑𝑡 2 𝑑𝑡
This equation was expressed in deviation variables, Laplace transformed and rearranged to
obtain the transfer function as:
̅𝑚 (𝑠)
ℎ 𝑃 𝐾 𝛽
𝐺𝑚 (𝑠) = ̅(𝑠)
= (𝜏2 𝑠2 +2𝜁𝜏𝑠+1) (30)
ℎ
iv Final Control Element (The Control Valve)
Process Control Valves are designed so that the flow rate through the valve is a nearly linear
function of the signal to the valve’s actuator. A first-order transfer function provides an
adequate model for an installed valve (Stephanopoulos, 2012) as:

𝐾𝑣
ℎ̅′ (𝑠) = 𝜏 𝑠+1 𝐶̅ ′ (𝑠) (31)
𝑣
Hence the Transfer function is:
̅′ (𝑠)
ℎ 𝐾𝑣
𝐺𝑓 (𝑠) = =𝜏 (32)
𝐶̅ ′ (𝑠) 𝑣 𝑠+1
33
Vol.8, No.2, 2018
The block diagram for the closed-loop system with the transfer functions is shown in Figure
4:
ℎ̅𝑠𝑝 ′(𝑠) 𝜀̅′(𝑠) 𝐶̅ ′(𝑠) 𝑞̅𝑖 ′(𝑠)

Gc(s) Gf(s) Gp(s)
ℎ̅𝑚 ′(𝑠)
Gm(s)
Figure 4: Block Diagram of Closed-Loop System
3. Discussion of Results
3.1. Open Loop Response
The concentrations of the reactants as predicted by a solution of the model equations are
shown in Figures 5 and 6.
Figure 5: Plot of Variation in concentration of Ethylene Oxide with time.
34
Vol.8, No.2, 2018
Figure 6: Plot of Variation in concentration of Water with time.
These figures show a decrease in the concentrations of both reactants as reaction proceeds.
The validity of the models were ascertained using industrial data from the works of Batiha,
(2004a) and comparing the results predicted from the model with industrial plant output. This
comparison for the reactant’s outputs are shown in Table 1.
Table 1: Comparison of Model Predictions and literature results

S/N Component Concentration (kmol/m3) % Deviation
Model Prediction Literature
1. Ethylene Oxide. 0.0582 0.0568 -2.47
2. Water. 1.9582 1.9402 -0.93
Table 1 showed that the models accurately predicted the reactants outputs with a maximum
deviation of 2.47%. Hence the models were used to simulate the behaviour of the product
(Mono Ethylene Glycol) to fluctuations (changes) in the reactants inlet flowrate to the reactor.
The behaviour of the CSTR (formation of the desired product, Mono-Ethylene Glycol (Cc)) to
step-changes ±25% and ±50% in the feed flowrate Fi is shown in Figure 7.
35
Vol.8, No.2, 2018
Figure 7: Effect of feed flow rate on the production rate of the desired product, MEG.
Figure 7 shows that the rate of formation of the desired product, Mono-Ethylene Glycol is
greatly affected by fluctuations of the inlet feed volumetric flowrate.
3.2. Closed-Loop Response: Effect of the Controller
The closed-loop response of the process (the controlled variable – the amount/volume of
MEG produced measure by the liquid/MEG height in the reactor) to (±) step changes in the
inlet flowrate; (-) step changes are shown in Figures 8, 9 and 10 while (+) changes are shown
in Figures 11, 12 and 13. The changes in inlet flowrate caused a decrease or an increase in the
liquid level and are shown by the yellow lines in the figures. The closed loop responses of the
liquid/MEG level in the reactor with a feedback PID controller at different values of its
parameters aimed at stabilizing the liquid level are also shown in these figures. Some of the
many iterations performed are hereby presented.
36
Vol.8, No.2, 2018
Figure 8: Effect of PID Controller at 𝜏𝐼 = 2, and 𝜏𝐷 = 2
Figure 8 shows fluctuations due to instabilities of the liquid level. The liquid level responded
fast, oscillates progressively and continuously with increasing amplitude and a low period of
oscillation. The process is quite unstable and the controller requires more tuning.
37
Vol.8, No.2, 2018
In Figure 9, the liquid level response is slower, but still oscillates progressively and
continuously with increasing amplitude and higher period of oscillation. The process is still
unstable, with unbounded output and the controller requires more tuning.
Figure 10 shows that the liquid level response is a little sluggish, oscillates progressively with
increasing amplitude, a smaller period of oscillation but attains stability after a time span of
about 340 seconds.
These plots showed the effect of Integral Control in a composite control system such as a PID
controller on the process. Results showed that for constant controller gain, 𝐾𝐶 , as the integral
time increases, the response becomes more sluggish, less oscillatory with lower overshoots
and decay ratios.
38
Vol.8, No.2, 2018
Figure 11: Effect of PID Controller at 𝜏𝐼 = 2, and 𝜏𝐷 = 2 for positive unit step-change
Figure 11 shows that the liquid level responded fast, oscillates progressively and continuously
with increasing amplitude and a lower period of oscillation. The process is quite unstable and
the controller requires more tuning.
39
Vol.8, No.2, 2018
Figure 12 shows that the liquid level responded a bit sluggish, oscillates progressively and
continuously with increasing amplitude and a decrease in the overshoot. The process is quite
unstable and the controller requires more tuning.
Figure 13 shows that the liquid level response is more sluggish with a larger decrease in the
overshoot. The process is still unstable.
Following the trends observed from the results of these plots, the integral and derivative time
were further adjusted in a systematic/specific way to obtain the ‘best’ values of controller
parameters for the process – values that gave faster and less oscillatory response that becomes
stable in the shortest possible time. These are shown in Figures 14 and 15.
40
Vol.8, No.2, 2018

These figures (14 and 15) show that by increasing the integral time to about 14 and the
derivative time to 3 at a constant controller gain, the closed-loop response of the liquid level
was almost stable. Therefore, the optimum values of the controller parameters that achieved
the specified control objective were: 𝐾𝐶 = 0.1, 𝜏𝐼 = 14, 𝜏𝐷 = 3
41
Vol.8, No.2, 2018
4. Conclusion
Dynamic models for the industrial process of the non-catalytic synthesis of Ethylene Glycol
via the hydrolysis of Ethylene Oxide in a Continuos Stired Tank Reactor have been developed
by applying the principles of conservation of mass and energy. The developed model
equations were validated by solving the model equations with industrial plant data given by
Batiha, (2004a). The model predictions of the reactants output obtained from the model
compare favorably with industrial plant outputs with maximum deviation of 2.47%.
Simulation of the continuous stired tank reactor to fluctuations/chamges in inlet feed flowrate
showed a huge effect on the performance (volume of MEG produced) of the reactor. This
justified the need for process control to maximize production and ensure selectivity towards
the desired product. Process control was effected with a Proportional-Integral-Derivative
feedback controller. A closed loop model (process with controller) was developed in terms of
the transfer functions of the components of the closed loop block diagram. Solution of the
closed loop model to step changes in the inlet flowrate showed oscillatory responce in the
output (MEG level) with offset over a long period. Controller tunning was therefore
performed using the on-line trial and error method to obtain optimum values of the controller
parameters (controller gain 𝐾𝐶 , the integral or reset time 𝜏𝐼 and the derivative time 𝜏𝐷 ) of
𝐾𝐶 = 0.1, 𝜏𝐼 = 14, 𝜏𝐷 = 3 which gave rapid and smooth responses to the set point changes
with little or no oscillations and eliminated the offset (steady state error).
References
Arrieta, I. (2001). Ethylene Glycol. Retrieved from

http://natsha.eng.usf.edu/gilbert/courses/modelandanalysis/portfolio_2001/1_project_1/a
rrieta_ethylene_glycol.pdf
42
Vol.8, No.2, 2018
Batiha, M. (2004a). Dynamic Modelling of the Non-Catalytic Process of Ethylene Oxide

Hydrolysis, Journal of Science and Technology, Vol. 9 (1 & 2), pp. 17 – 27.
Batiha, M. (2004b). Kinetic Investigation of Consecutive-Parallel Reactions in the Non-

Catalytic Process of Ethylene Oxide Hydrolysis, Journal of King Abdulaziz University:
Engineering Science, Vol. 15, (1), pp. 19-31
Celik, F. E., Lawrence, H. & Bell, A. T. (2008). Synthesis of precursors to ethylene glycol
from formaldehyde and methyl formate catalyzed by heteropoly acids, Journal of
Molecular Catalysis A: Chemical 288, 87–96
Iwakura, T. & Miyagi, H. (2000.) US Patent 6147, 265.
Kawabe, K., (2000). US Patent 6080,897
Kruchten, V. (1999a). US Patent 5874,653
Kruchten, V. (1999b). WO Patent 9923,053.
Landau, R. & Ozero, B. J. (1982). Ethylene Glycol, Encyclopedia of chemical processing and
Design. 2-52.
Li, Y., Yan, S., Yue, B., Yang, W., Xie, Z., Chen, Q. & He, H., (2004), Selective catalytic
hydration of ethylene oxide over niobium oxide supported on α-alumina, Applied
Catalysis A, General, 272, 1-2, 305 – 310.
Li, Y. C., Yan, S. R., Yang, W. M., Xie, Z. K., Chen, Q. L., Yue, B. & He, H. Y. (2005).
Effects of support modification on Nb2O5/α-Al2O3 catalyst for ethylene oxide
hydration J. Mol. Catal. A: Chem., 226, 285–290.
Li, Y. C., Yan, S. R., Qian, L. P., Yang, W. M., Xie, Z. K., Chen, Q. L., Yue, B. & He, H. Y.
(2006). Effect of tin on Nb2O5/α-Al2O3 catalyst for ethylene oxide hydration, J. Catal.
241, 173-179.
Luyben, W. L. (1990). Process Modeling, Simulation and Control for Chemical Engineers,
(2nd ed.). Singapore, McGraw-Hill Publishing Company.
Melhem, S. A., Gianetto. A., Levin, M. E., Fisher, H. G., Chippett, S., Singh, S. K. &
Chipman, P. I. (2001). Kinetics of the reactions of ethylene oxide with water and
ethylene glycols, Process Safety Progress, 20, 4, 231 – 246.
Rebsdat, S. & Mayer, D. (2005). Ethylene oxide. In: Ullmann's Encyclopedia of Industrial
Chemistry, 7th Ed., Weinheim, Wiley–VCH Publishers (on line)
Samoilov, N. A. & Mnushkin, I. A. (2012). Mathematical Modeling Fractionation Stage of

Reactive Distillation Process for Production of Ethylene Glycol, Electronic Scientific
Journal “Oil and Gas Business”, 4, 158 – 164.
Sinnott, R. & Towler, G. (2009). Chemical Engineering Design Volume 6. India: Elsevier
Limited.
43
Vol.8, No.2, 2018
Stephanopoulos, G. (2012). Chemical Process Control: An Introduction to Theory and

Practice. (Indian ed.). New Delhi, India: PHI Learning Private Limited.
Strickler G. R., Landon V. G., & Lee, G. J. (1999). US Patent WO1999031033A1.
Shvets, V. F., Kozlovskiy, R. A., Kozlovskiy, I. A., Makarov, M. G., Suchkov, J. P. &
Koustov, A. V. (2005), The Model of Catalytic Reactor of Ethylene Glycol Production,
Org. Process Res. Dev., 9 (6), 768 – 773.
Yang, G., Ren, N., Zhang, Y. & Tang, Y. (2010). Studies on mechanism for homogeneous
catalytic hydration of ethylene oxide: Effects of pH window and esterification,
Catalysis Communications 11, 447 – 450.
Numenclature
A = Cross Sectional Area (m2)
𝐶𝑗𝑜 = Inlet Concentration of component j (kmol/m3)
𝐶𝑗 = Outlet Concentration of component j (kmol/m3)
𝐸𝑖 = Activation Energy for reaction path i (cal/g mol)
𝐺𝑐 = Transfer Function of the Controller
𝐺𝑑 = Transfer Function of the Load
𝐺𝑓 = Transfer Function of the Final Control Element
𝐺𝑚 = Transfer Function of the Measuring Device
𝐺𝑝 = Transfer Function of the Process (CSTR)
h = Height of Reactor (m)
i = Reaction path
j = Component
𝐾𝑖 = Reaction rate constant for reaction path i
𝐾𝑖𝑜 = Pre-exponential Factor for reaction path i (m3)3 kmol-3S-1
𝑞𝑖 = Inlet Volumetric Flowrate of feed (m3/min.)
q = Outlet Volumetric Flowrate of feed (m3/min.)
44
Vol.8, No.2, 2018
𝑟𝑗 = Rate equation for component j
𝑇𝑖 = Inlet Temperature of feed (K)
T = Outlet Temperature (K)
V = Volume of Reactor (m3)
𝐶𝑝 = Heat capacity of Feed (kJ kmol-1K-1)
ρ = Density of Feed (kg/m3)
Δ𝐻 = Heat of Reaction for reaction (kJ kmol-1)
45
View publication stats

Paschals Paper

Uploaded by

Copyright:

Available Formats

Paschals Paper

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

Paschals Paper

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

SIMULATION AND CONTROL OF A REACTOR FOR THE NON-CATALYTIC

Research · March 2019

Mathematical Modelling and SImulation View project

Chemical Engineering Process Design View project

The user has requested enhancement of the downloaded file.

SIMULATION AND CONTROL OF A REACTOR FOR THE NON-CATALYTIC

Phosphorus-based Halide Catalysts (Kawabe, 2000); anion exchange resins such as

halogenate or bicarbonate-type: Strickler et al, (1999); polymer of a vinylaromatic compound:

organosilane ammonium salts: Kruchten, (1999a) and macrocyclic chelating compounds:

petroleum-derived ethylene. A typical process scheme shown in Figure 1 is complex

consisting of a reactor, an evaporator and columns/vaporizers amongst other equipment.

distillation column for this process.

1.1. Process Description: Hydrolysis of Ethylene Oxide

ethylene Oxide (Arrieta, 2001) is shown in Figure 1.

Stirred Tank reactor (CSTR) where it is thermally hydrolyzed to Ethylene Glycol at an

glycol is separated to mono, di, and tri ethylene glycol respectively.

glycol (MEG) (Arrieta, 2001).

models with Proportional-Integral-Derivative Controller capable of bringing the system

2. Material and Methods

feedback control configuration with Proportional-Integral-Derivative (PID) controller was

tuning) - a systematic selection of controller parameter values that achieves control

implemented with MathLab Simulink was developed. In this procedure, several/different

2.2. Model Development

2.2.1. Kinetic Model

For j = A,B,C,D, E: A represent 𝐶2 𝐻4 𝑂; B: water; C: 𝐶2 𝐻6 𝑂2; D: 𝐶4 𝐻10 𝑂3; E: 𝐶6 𝐻14 𝑂4 as:

(−𝑟𝐴 ) = −𝐾1 𝐶𝐴 𝐶𝐵 − 𝐾2 𝐶𝐴 𝐶𝐶 − 𝐾3 𝐶𝐴 𝐶𝐷 (4)

(−𝑟𝐵 ) = −𝐾1 𝐶𝐴 𝐶𝐵 (5)

2.2.2. Reactor Model

Figure 2: Schematic of the CSTR

mass, Component mass and Total Energy as follows:

i. Total mass balance

A Total mass balance on the CSTR gives:

Substituting and rearranging gives:

ii. Component mass Balance

substituting 𝑉 = 𝐴ℎ and expanding gives:

Substituting equation (13), rearranging and simplifying gives:

iii. Energy Balance

The principle of Conservation of energy applied on the reactor gives:

Substituting eqn. (13) and simplifying gives:

2.3. Process Control

loop diagram of the process is shown in Figure 3.

Figure 3: Closed-Loop diagram of the Process.

*DPC – Differential Pressure Cell

transfer functions of the various components in Figure 3 was developed to give:

𝐺𝑓(𝑠) 𝐺𝑐(𝑠) 𝐺𝑝(𝑠) 𝐺𝑑(𝑠)

𝐺𝑓(𝑠) 𝐺𝑐(𝑠) 𝐺𝑝(𝑠)

ii Controller: Proportional-Integral-Derivative (PID) Controller

The model equation for the controller in deviation variables is:

iii Measuring Device (Differential Pressure Cell)

Where, 𝐴 = Cross-sectional area of the Capillary tube,

𝐿 = Length of the Capillary Tube,

𝜌 = Density of the Liquid,

𝑥 = Fluid displacement in the Capillary Tube,

𝐾 = Hooke’s constant for the Diaphragm,

∆𝑃 = Differential pressure across the cell.

written in standard form as:

obtain the transfer function as:

iv Final Control Element (The Control Valve)