Journal of Natural Gas Science and Engineering 71 (2019) 102974

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

journal homepage: www.elsevier.com/locate/jngse

Operability assessment on alternative natural gas liquids recovery schemes T

a a,b a,∗
J. Chebeir , S.D. Salas , J.A. Romagnoli
a
Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, LA, 70803, United States
b
Escuela Superior Politécnica del Litoral, ESPOL, Facultad de Ciencias Naturales y Matemáticas, Campus Gustavo Galindo Km. 30.5 Vía Perimetral, P.O. Box 09-01-5863,
Guayaquil, Ecuador

ARTICLE INFO ABSTRACT

Keywords: Natural gas liquids (NGLs) are valuable by-products of the raw natural gas. A variety of process schemes have
Natural gas liquids recovery been developed over the years for an economical, safe and efficient extraction of these components including the
Multivariable control conventional, cold residue recycle (CRR), and the gas subcooled process (GSP), each comprising a range of
Controllability competing advantages. In this work, the operability of these NGL recovery technologies is evaluated in terms of
Dynamic process simulation
the controllability and process dynamics. The evaluated configurations are analyzed qualitatively in order to
recognize their ability of achieving main control targets and reducing the process upsets due to typical dis-
turbances. The main goal foresees the assessment of control schemes capable of guaranteeing robustness and
performance during the operation while highlighting the importance of developing a structurally sound control
strategy.

1. Introduction exchange with column overhead, side reboilers and/or external re-
frigerants (Mehrpooya et al., 2010). Thereafter, a decompression pro-
Natural gas represents a global source of clean and cheap energy cess takes place to rapidly expand and cool even further these gases in
which production is rapidly increasing, especially in North America. order to favor the condensation of ethane, propane, and other heavier
The technological advancements developed in drilling and multi-stage hydrocarbons.
fracturing have allowed the access to enormous volumes of natural gas Various comprehensive reviews on the existing NGL extraction
trapped in the shale formations. In the United States, the production of technologies have been presented in the literature (Manning and
dry natural gas has experienced a robust increase from less than 20 Tcf Thompson, 1991; Arnold and Stewart, 1999; Kidnay et al., 2011; Park
to approximately 30 Tcf in the last two decades (EIA, 2019). This rapid et al., 2015). Among the numerous NGL extraction structures, it is
production surge has generated a tectonic impact on virtually all the possible to mention that the conventional process and its variants such
segments involved in the value chain. Big volumes of the feedstock as the gas-subcooled process (GSP), and the cold residue recycle (CRR)
material demand an economical and continuous operation of the gas process are the most renowned and utilized in gas processing plants
processing plants. Main activities involve the cleaning, purification and (Kherbeck and Chebbi, 2015). Each of these recovery processes have
conditioning of the raw natural gas to meet desired conditions for demonstrated different advantages and disadvantages, and their ap-
maintaining the pipeline specification for gas transportation, and to plicability depends on numerous factors that can be from the require-
guarantee an environmentally clean-burning gas (Faramawy et al., ments and process constraints to the technological culture of the op-
2016). erator.
Due to the higher marketable value of heavier hydrocarbons (C2+), The processing of natural gas incorporates a wide variety of op-
the NGLs are often extracted from the original feed gas. In this context, erations, where distillation is by far the most important stage of the
the use of turbo-expanders represents one of the most predominant process. With methane occupying a majority of the natural gas (< 80%
strategies applied in the gas industry for the separation of these valu- mol/mol), the first unit typically consists of a cryogenic high-pressure
able hydrocarbon liquids (Li et al., 2017). A large number of variations column to recover the valuable and energy dense hydrocarbon. This
of the original turbo-expander process scheme have been proposed in unit is commonly referred to as the demethanizer, where methane is
the last years. In most of the proposed schemes, the gases are first removed as an overhead gas, and the balance NGLs are processed
cooled to reach very low temperatures through the utilization of heat downstream as the bottoms product (Luyben, 2013). With the NGL

∗
Corresponding author. 3315N Patrick F. Taylor Hall, Louisiana State University, Baton Rouge, LA, 70803, United States.
E-mail address: [email protected] (J.A. Romagnoli).

https://doi.org/10.1016/j.jngse.2019.102974
Received 21 May 2019; Received in revised form 17 July 2019; Accepted 17 August 2019
Available online 22 August 2019
1875-5100/ © 2019 Elsevier B.V. All rights reserved.
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Table 1 2. Flowsheet of different process schemes

Feed gas composition for the different process
schemes. The different process models are developed in the process simulator
Component Composition HYSYS® in dynamic mode for the selected schemes including the con-
ventional, CRR and GSP structures. This process simulator is selected
Nitrogen 0.01 due to its reliable performance in petrochemical, petroleum refining, oil
Methane 0.93
assays and all related industry. The process models are constructed
Ethane 0.03
Propane 0.015 based on realistic operating conditions. The raw natural gas feed flow
Butanes 0.009 rate is 4980 kg-mol.h−1 with a composition constituted by nitrogen and
Pentanes 0.003 a mixture of different hydrocarbons including methane, ethane, pro-
Hexanes 0.003
pane, butanes and heavier hydrocarbons (C5 and C6). Table 1 sum-
% C2+ 6
marizes the inlet gas flow composition for the different process schemes
analyzed in this work. This composition is based on the case study
evaluated by Chebbi et al. (2010) to determine the optimal design and
recovery process being a high-throughput and energy-intensive com- operation of a conventional NGL recovery unit. In addition, the raw gas
ponent of natural gas processing, it has been a central target of opti- is supplied to the separation unit at a pressure of 6015 kPa, and at a
mization towards capturing loss profits. The GSP and CRR process temperature of 35 °C.
embrace a range of competing advantages, making them candidates of In the conventional process scheme for NGL recovery (Fig. 1), the
further improvement to achieve an economic, safe, and reliable op- raw gas feed is first cooled to −4.8 °C by the gas-gas heat-exchanger E-
eration (Kherbeck and Chebbi, 2015). 100 and then further cooled by the chiller E-101 in order to reach an
In this contribution, the dynamic response of different NGLs re- appropriate temperature for the separation of condensates in the flash
covery process schemes is evaluated. Operability and controllability of separator TK-100. The chiller is a propane-based refrigeration cycle. In
alternative process configurations are of paramount importance when the chiller, the feed temperature is reduced from −4.8 °C to −29.3 °C
seeking a deep understanding of the process dynamics as well as op- to simultaneously maximize the cooling and to avoid the air leakage
portunities for further improvement of the operation (Bahri et al., 1996, into the system (Manning and Thompson, 1991). After separating from
1997). These alternative configurations are analyzed dynamically, and the liquid phase, the gases leaving the separator TK-100 are further
different control strategies are tested and evaluated in terms of cooled by the gas-gas heat exchanger E-102 from −29.3 °C to −59.6 °C
achieving control targets, and by reducing the process upsets due to with the overhead flow coming from the tower T-100. A second se-
typical disturbances. This is accomplished by developing dynamic paration of condensates takes place in the flash separator TK-101 (cold
model simulations based on realistic operating conditions. Perturba- separator). The gas leaving the cold separator is depressurized in the
tions of operating conditions are performed to test the robustness of the turbo-expander TE-100 to achieve a pressure of 1010 kPa and a tem-
proposed control architectures. Proper control schemes around the perature of −116.3 °C and a vapor/liquid ratio of 0.88. This gas-liquid
demethanizer unit may additionally lead to profits being propagated flow is then fed at the top of the tower. The liquids coming from the
downstream by reducing process upsets due to input disturbances, thus bottom of both separators are expanded through the utilization of
highlighting the importance of developing a structurally sound control Joule-Thompson valves to achieve pressure values determined by the
strategy. operating pressure of the tower. The liquids leaving the separator TK-

Fig. 1. Conventional scheme for NGL recovery.

2
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 2. GSP scheme for NGL recovery.

Fig. 3. CRR scheme for NGL recovery.

100 are flashed from 4672 kPa to 1026 kPa while the liquids leaving the to 6081 kPa. The recompression process is also accompanied by a sig-
cold separator TK-101 are flashed from 5199 kPa to 1015 kPa. The nificant increase of temperature from −11.7 °C to 184.4 °C. This
bottoms product of the demethanizer is transported to the fractionator foresees the need of an extra air cooler to condition the sales gas flow.
for further separation of the valuable heavier hydrocarbons. These This equipment is not included in the flowsheet analyzed in the present
products leave the unit with a pressure of 1000 kPa and a temperature work.
of −9.6 °C. After its utilization as cooler in the heat exchangers E-100 The GSP process scheme was first proposed by Campbell and
and E-102, the gas leaving the top of the tower T-100 has to be re- Wilkinson (1981) as an improvement of the conventional process
compressed for its commercialization. Part of the pressure for the re- scheme. A portion of the gas from the cold separator is sent to a heat
compression of the sales gas is provided by the turbo-expander TE-100 exchanger where it is partially condensed with the column overhead
with an increase from 821.3 kPa to 1265 kPa. Finally, the recompressor stream. This stream is then flashed to the top of the column in order to
K-100 is utilized to increase the pressure of the sales gas from 1265 kPa act as a reflux and to improve the ethane recovery. The remaining gas

3
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 4. Basic control structure implemented in the conventional process scheme.

resides in the incorporation of a reflux stream in the column overhead.

In other words, this process scheme retains all the advantages of the
GSP process while incorporating a reflux stream of nearly pure me-
thane. This includes the utilization of a cryogenic compressor to boost
the reflux to a slightly higher pressure so that a fraction of the methane
can be condensed. The compressed overhead is cooled, and then ex-
panded through a valve before being supplied to the top of the tower.
The intention of the methane reflux is to improve the rectifying section
so that only a negligible amount of ethane and heavier components
Fig. 5. Ratio control configuration. escaped from the column overhead. A more detailed explanation of the
advantages related to the incorporation of this reflux was presented by
Pitman et al. (1998). The flow sheet corresponding to the CRR process
Table 2
Typical tuning values of process controllers. scheme is given by Fig. 3. Again, the different operating conditions are
depicted in this schematic representation of the process.
Controller KC,Conv τI,Convv KC,GSP τI,GSP KC,CRR τI,CRR

FIC-100 0.77 0.07 0.17 0.07 0.17 0.07

FIC-101 – – 0.31 0.07 0.31 0.07 3. System dynamics and control
TIC-100 1.06 0.36 2.76 0.36 2.68 0.36
LIC-100 3.40 1.46 3.90 2.14 3.86 2.34 3.1. Design of different process schemes
LIC-101 3.60 0.75 3.82 0.73 3.63 0.73
LIC-102 2.94 1.70 – – – –
PIC-100 2.72 0.18 1.53 0.11 1.87 0.11
Dynamic simulations of the different process schemes are developed
XIC-100 3.62 5.13 3.70 7.33 3.73 6.60 for the implementation of the control structures. Previous literature and
RFIC-100 – – – – 1.16 0.07 default values provided by the process simulator are utilized for the
ERIC-100 (Primary Loop) 0.13 1.02 0.11 3.06 0.11 3.80 design of the different equipment in each scheme (Luyben, 2013;
Branan, 2002). The demethanizer column is constituted by 30 stages
with a reboiler. The diameter of the column is 1.72 m with a tray space
flow is also expanded to the tower operating pressure in a turbo-ex-
of 0.5 m. In the case of the conventional and GSP process schemes, the
pander and fed several stages below the top of the column. The cold
UA of the heat-exchangers E-100 and E-102 is 2.038 × 105 kJ/°C h and
liquids supplied to the top of the demethanizer act as reflux, contacting
1.961 × 105 kJ/°C h, respectively. The chiller has a volume of 0.10 m3.
and rectifying the vapors leaving the expander by absorbing the ethane
The extra heat-exchanger required in the CRR scheme (E-104) has a UA
and heavier hydrocarbons (Pitman et al., 1998). Typically, a GSP pro-
of 1.576 × 105 kJ/°C h. The reboiler of the column (regular HYSYS®
cess scheme can achieve an ethane recovery of 70–80% (Mokhatab
heater and separator reboiler) has a diameter of 1.193 m and a length of
et al., 2013). Another advantage of the cold reflux liquid stream is that
1.789 m. The flash separators, TK-100 (the three schemes) and TK-101
it can significantly reduce the risk of carbon dioxide solid formation
(conventional scheme), have both a diameter of 1 m and a height of
(Getu et al., 2013). The flowsheet corresponding to the GSP process
2.5 m. The expansion and compression sections of the turbo-expander
scheme is depicted by Fig. 2.
TE-100, recompressor K-101 and cryogenic compressor K-102 (CRR
The CRR process scheme was introduced in the original design of
process) have an adiabatic efficiency of 75%. Joule-Thompson valves
GSP by Campbell et al. (1989) to improve ethane recovery efficiency.
(JTV-100 and JTV-101) are also utilized in the different schemes to
The main difference of this process structure and the GSP scheme
generate a sudden expansion of fluid (pressure decrease) in the lines

4
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 6. Cascade structure schemes with the recovery controller as the primary control loop and the temperature controller as the secondary control loop.

Fig. 7. Cascade structure implemented for CRR and GSP schemes with the recovery controller as the primary control loop and the ratio controller as the secondary
control loop.

Fig. 8. Schematic representation of cascade control system for CRR and GSP configurations utilizing the ratio control as secondary loop.

entering to the demethanizer. The default values of tolerance error the main objectives during the operation of the demethanizer unit. A
defined by the process simulator for the energy and mass balances in major objective is to achieve a recovery of at least 82% of ethane. This
each equipment are maintained for the different models. can be achieved through temperature control in the first separator by
manipulating the amount of refrigerant (proppant) utilized in the
3.2. Control structures for alternative process configurations chiller. Although the influence of the column pressure in the ethane
recovery is not as significant as in the case of the separator temperature,
To define the different control structures, it is necessary to establish it has still certain influence in the operation of the unit. In previous

5
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 9. Profiles for the different process schemes with basic control structure: (a) Compositions; (b) Temperatures.

Fig. 10. Time evolution of temperature in different stages of the column for CRR configuration against 10% increase of feed flow rate: (a) basic control strategy; (b)
cascade control strategy.

Fig. 11. 3-D visualization of temperature profile in demethanizer for CRR Fig. 12. 3-D visualization of the temperature profile in the column for the CRR
process scheme and basic control strategy against a 10% increase of gas feed process scheme and cascade control strategy against a 10% increase of feed flow
flow rate. rate.

6
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 13. Closed-loop responses using basic control configuration for conventional scheme against a 10% feed flow rate increase: (a) Ethane Recovery; (b) Methane
Composition; (c) NGL Flow Rate; (d) Reboiler Level.

works, it has been demonstrated that these types of systems are more feed gas stream in the chiller.
effective at lower pressures (Kherbeck and Chebbi, 2015). A tight
pressure control is also essential to avoid flashing of the liquid in the A similar basic control structure is utilized for the GSP and CRR
trays and weeping and dumping of trays (Luyben, 1992). The other schemes with the following modifications:
main objective is to control the level of methane impurity in the bot-
toms NGL stream. A pre-established value of 1% mol composition of 1. Feed flow rate is controlled by manipulating the inlet valve GV-100
methane is considered for the liquids separated in the demethanizer. instead of the turboexpander TE-100 power. This is done due to the
Other group of controllers such as level and flow control loops are presence of a split ratio that bypasses part of the TE-100 inlet flow.
mainly related to maintain certain stability in the operation (i.e. the The higher the bypassed flow the higher is the delta pressure
prevention of liquid in entrance of the expander). available for the inlet flow control.
The basic control structure developed for the conventional NGL 2. The level of the only flash separator in the system (TK-100) is
recovery unit (Fig. 4) is summarized as follows: controlled by varying the size of the discharge flow with the control
valve LCV-101. The other level control loop corresponds to the re-
1. Feed flow rate is controlled by manipulating the power of the turbo- boiler E-103 (same implementation of the conventional scheme).
expander. 3. A ratio control system is established by manipulating the flow
2. Column pressure is controlled by manipulating the supplied power through the expander f and leaving the bypassed flow g as the wild
to the gas compressor located in the outlet of the unit, which dis- variable. As recommended by Shinskey (1996), it is preferred to
charges at 6081 kPa. locate the ratio calculation (ratio station) outside the closed-loop, in
3. The level of the first separator (TK-100) is controlled by varying the the path of the required set-point (Fig. 5). This ratio station can be
size of the discharge flow with the valve LCV-100. considered a manual control where the wild flow g is multiplied by
4. The level control of the cold separator (TK-101) is performed by an operator-introduced parameter R. The selection of this parameter
manipulating the discharge flow with the control valve (LCV-101). in conjunction with the separator temperature is crucial in the level
5. Liquid level in the reboiler (E-103) of the column is controlled by of recovery achieved in the column. Later, it will be shown that the
manipulating NGL bottoms flow stream. manual selection of the parameter can be modified by the im-
6. Methane composition in the NGL bottoms stream is controlled by plementation of a cascade control scheme to satisfy a required re-
manipulating the heat provided to the reboiler E-103. The typical covery (higher-level objective).
delay in the on-line analyzers is not considered in this case study. 4. Finally, the control structure of the CRR scheme also incorporates a
7. The temperature in the first separator TK-100 is controlled by recycle control in the column overhead by manipulating the power
varying the amount of refrigerant utilized in the chiller E-101. In our to the cryogenic compressor.
simulation model, this is translated as the amount of heat loss by the

7
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 14. Comparison of closed-loop responses using basic and cascade control configuration for conventional scheme against 10% feed flow rate increase: (a) Ethane
Recovery; (b) Methane Composition; against 10% feed flow rate decrease: (c) Ethane Recovery; (d) Methane Composition.

The procedure implemented for the determination of the tuning a direct impact on the ethane recovery (Luyben, 2013). A lower tem-
parameters of the PI controllers in the different process schemes is perature not only increases the ethane recovery but also increments the
given as follows: refrigeration duty in the chiller. In this sense, the value of this tem-
I. The relay auto tuning strategy (Astrom and Hagglund, 1984) with perature is manually selected by the operator to obtain the desired
Tyreus-Luyben method (Tyreus and Luyben, 1992) is utilized for setting separation in the demethanizer. A better approach can be implemented
the different controllers of the system. This strategy is available in by manipulating the temperature set-point to achieve the necessary
HYSYS®. ethane recovery. Of course, this implies the utilization of a cascade
II. The controllers with high gains are detuned by certain a factor F control structure with an outer loop defined by the ethane recovery
in order to generate a smoother response of the system. This detuning (primary controller) and an inner loop established by the TK-100 outlet
includes the level, pressure, and methane composition controllers of the temperature (secondary controller). The critical issue in the proposed
different schemes. control strategy resides in the indirect measurement of the ethane re-
III. In the case of implementing a cascade control in the system, the covery. The control structure requires the incorporation of a new ana-
tuning is based on the approach proposed by Romagnoli and Palazoglu lyzer to indirectly determine the ethane recovery. This system can be
(2012). First, the secondary loop is tuned using any technique. In our constituted by a flow indicator and an ethane analyzer in the NGL
case, it is implemented the Tyreus-Luyben tuning technique. Secondly, bottoms stream. The existing inlet flow indicator plus an extra ethane
the secondary loop is placed in automatic and the primary loop is tuned analyzer should complete all the necessary instrumentation for the in-
with the same tuning technique. For this control loop, the integral direct determination of the recovery. In the different process simulation
constant is detuned by the same factor F to avoid a sluggish response of models, the determination of the ethane recovery is performed using
the control system. the information from the indicators in a spreadsheet feature and then,
The procedure is implemented in the different schemes and the exported to the recovery controller. This control structure can be im-
detuning factors are the same in all systems for comparing the control plemented in all the process schemes presented in this work. Fig. 6
architectures under fair conditions. The values of the different tuning depicts a schematic representation of the cascade control structure for
parameters including the gain and integral time for the different process the different process schemes.
configurations are listed in Table 2. A possible variation of the previous cascade structure can be im-
plemented in the GSP and CRR process schemes. This control structure
3.3. Monitoring and control of ethane recovery is based on the construction of a system with the same primary con-
troller and utilizing the ratio control for the second control loop.
None of the previous proposed structures allow the direct mon- Similar to previous cascade control, a flow indicator and an ethane
itoring and control of the ethane recovery in the column bottoms. It is analyzer are required in the demethanizer bottoms stream.
well known that the temperature in the first flash separator TK-100 has Additionally, an extra ethane analyzer in the feed gas is necessary to

8
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 15. Closed-loop responses using basic control configuration for CRR scheme against a 10% feed flow rate increase: (a) Ethane Recovery; (b) Methane
Composition; (c) NGL flow rate; (d) Reboiler Level.

complete the instrumentation for the indirect determination of the re- of the column). The variations observed on the top stages in the CRR
covery. In this structure, the selection of the parameter R in the Ratio and GSP schemes are mostly related to the cold liquids (fed on the top
Station is not performed manually by an operator. This parameter is of the column) contacting and rectifying the turbo-expander vapors
now defined by the recovery controller (primary control loop), which coming from the lower stages of the demethanizer. This would in
means an indirect definition of the flow controller set-point. After principle allow a better control on both recovery and methane com-
tuning the primary loop, the values of the gain, KC, and integral time, τI, position.
are 0.10 and 2.30, respectively. A schematic representation of the Fig. 10a illustrates the time evolution of the temperature profiles in
proposed cascade control structure is depicted by Fig. 7. A schematic different stages of the column for the CRR configuration under basic
representation of the cascade control system for the CRR and GSP control configuration when the gas feed flow rate is increased by 10%.
process schemes is depicted in Fig. 8. As shown, the profile moves, and the methane composition controller
brings them back close to their original position. This allows the system
4. Results and discussion to adjust for the gas feed flow rate changes and maintain the level of
impurity in the column bottoms stream at the target value. As expected
4.1. Temperature and composition profiles from our previous analysis of the composition and temperature profiles,
larger changes (due to the disturbance) are observed at the bottom of
Before comparing the closed-loop behavior of the alternative con- the column while slightly variations can be seen as we move to the top.
figurations, it is important to analyze the internals of the demethanizer This indicates that the most appropriate location for the implementa-
column to better understand the main differences between the com- tion of a simple methane composition control with the column tem-
peting schemes. Fig. 9 illustrates the composition and temperature perature is between the trays 27–28. These trays of the stripping section
profiles for the three alternative configurations. These profiles were are the best in terms of sensitivity and capture the main separation in
obtained for the particular operating regime and under the basic control the column. Fig. 10b depicts the time evolution of the temperature
configurations described above. Clearly, both CRR and GSP configura- profiles for the CRR scheme under cascade control with the ethane
tions have similar profiles (but not exactly the same), and they differ recovery. In this case, the temperature is recuperated throughout the
from the profiles of the conventional configuration. The conventional different stages of the column. This is explained in more detail in the 3-
scheme shows basically no main separation and temperature changes in D visualizations of the dynamic responses for both the basic and cas-
the top of the column while (as expected) sharp changes occur at the cade control structures.
bottom of the column. On the other hand, the CRR and GSP config- The 3-D temperature profiles for both the basic and cascade control
urations display both separation and temperature changes at both the structures are depicted in Fig. 11 and Fig. 12. These figures show the
ends of the column (although still sharper changes occur at the bottom effect of the inlet flow perturbation in the temperature profiles and the

9
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 16. Comparison of closed-loop responses using basic and cascade control configuration for CRR scheme against 10% feed flow rate increase: (a) Ethane
Recovery; (b) Methane Composition; against 10% feed flow rate decrease: (c) Ethane Recovery; (d) Methane Composition.

responses of the system. The basic control strategy (Fig. 11) does not control structure (temperature controller TIC-100 as secondary loop) is
recover the original profile and in the rectifying section of the column, illustrated by Fig. 14. In the case of a 10% gas feed flow rate increase,
the temperature stabilizes in a lower value which is correlated with a the system with cascade control recovers the pre-established ethane
higher ethane recovery. Different is the case of the temperature in the recovery set-point at the expense of a lower temperature in the flash
stripping section, where the action of the methane composition con- separator and the higher deviation of the methane composition in the
troller brings the temperatures to their original values. On the other column bottoms. Of course, this implies a higher power consumption in
hand, the cascade control structure (Fig. 12) does recover the tem- the reboiler to stabilize the methane composition at the required set-
perature profile in the rectifying section of the column, which is related point. If a 10% decrease in the feed flow rate is performed, the closed-
with the action of the ethane recovery control. loop returns the ethane recovery to its set-point value at the expense of
a more oscillatory behavior of the system, and a higher deviation of the
4.2. Conventional process configuration methane composition. As can be observed in both increase and decrease
of the flow rate for cascade control, there is competition between the
The dynamic responses of the conventional scheme utilizing a basic ethane recovery control structure and the methane composition con-
control configuration is illustrated by Fig. 13. An increase of 10% (step- troller.
change in set-point) in the feed flow rate is performed by manipulating
the turbo-expander power. A negligible decrease of 2% in the ethane 4.3. CRR process configuration
recovery is observed as a consequence of this flow change. This is re-
lated to the slight increase of temperature observed due to the action of The dynamic responses of the basic control system for the CRR
the methane composition controller (increase of reboiler power heat), scheme against a 10% of the feed flow rate increase is depicted by
which has a direct detrimental effect on the recovery. The methane Fig. 15. The change of feed flow rate is followed by an increment of
composition controller recovers the required purity after approximately 8.5% in ethane recovery during a period of approximately 50 min
56 min of introducing the step change. During the transition to the new (Fig. 12a). The proposed control system is also able to bring the me-
steady-state condition, the purity of the NGL bottoms stream goes off- thane composition to its pre-established set-point in about 65 min. Si-
specification in certain periods. In addition, an increment in the feed milar to the case of the conventional scheme, an increase in the NGL
flow rate is followed by a rise in the amount of sales gas and NGL bottoms stream ( 14.3% increase in 23 min) is observed as a con-
generated. The flow increase in the NGL bottoms stream is achieved sequence of the change in the inlet flow. Similar as in the case of the
after 70 min. Finally, the reboiler level returns to its set-point after methane composition, the reboiler level is recovered after approxi-
98 min. mately 26 min. The remaining controlled variables also return to their
A comparative study between the basic control and the cascade corresponding set-point values after the perturbation of the inlet gas

10
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 17. Dynamic using basic control configuration for CRR scheme against a 10% feed ethane composition increase: (a) Ethane recovery; (b) Methane impurity; (c)
NGL flow rate; (d) Reboiler level.

flow rate. basic control configuration. As depicted by Fig. 17, periods of ap-
Fig. 16a and b depict a comparative study between the basic and proximately 30 and 20 min are required for stabilization of the ethane
cascade (TIC-100 as secondary loop) control structures for different recovery and the NGL flow rate, respectively. Although the flow rate of
monitored variables in the CRR unit against a 10% increase of gas feed the NGL bottoms stream experiences an increase of 3.8%, the recovery
flow rate. In both the basic and cascade control structures, the ethane is negatively affected with a slight decrease of 0.85%. A slightly longer
recovery experiences an important overshoot before reaching its final time period of 49 min is required for the stabilization of the methane
values. The cascade control recovers the pre-established set-point value composition. Finally, the reboiler level control presents the fastest re-
for the ethane recovery after 98 min. To maintain the ethane recovery sponse to the composition disturbance with a stabilization period of
in the desired set-point, an increase in the separator temperature is approximately 20 min.
performed by decreasing the amount of refrigerant supplied to the
chiller (energy dissipation). The methane composition is also stabilized 4.4. GSP process configuration
to the pre-established set-point after 90 min of unit operation. The
overshoot observed in the response of this controller in the cascade Fig. 18 illustrates the closed-loop responses for the GSP scheme
structure is lower than the one observed in the basic control system. If a using basic control configuration against a gas feed flow increase of
10% decrease of feed flow rate is implemented in the system, a more 10%. Similar to the case of the CRR process scheme, there is an increase
rapid and oscillatory response to the changes is observed in the ethane of 7.3% in ethane recovery in a period of approximately 49 min. The
recovery control (Fig. 16c). Ethane recovery achieves its pre-established proposed control system is able to bring the process to the required
set-point after 89 min. In the case of the methane composition control purity and recovery levels. The control system is able to bring the re-
(Fig. 16d), the required purity is obtained after 134 min. When com- maining variables (Fig. 18b and d) to their pre-established set-points. In
pared to the basic control, the response of this controller presents a the case of the methane composition, the impurity achieves even higher
slightly lower overshoot and similar speed to achieve the pre-estab- levels than in the CRR process scheme in a period of 50 min. The se-
lished set-point. Similar to the conventional process scheme, the cas- parator level recovers its required set-point in approximately 26 min.
cade control structure opposes to the action of the methane composition Fig. 19 compares the performance of the basic control against the
control loops. cascade configuration for the GSP process scheme. Like in the case of
Given the high variability observed not only in flow rate but also in the CRR scheme, the improvements on the performance are evident in
the composition of the raw gas, the dynamic response of the system to terms of the effect of the feed disturbance on the ethane recovery. The
variations in the content of the gas constituents can represent another pre-established ethane recovery set-point is achieved after perturbing
important study. In this case, a 10% increase in the ethane mol com- the feed flow rate in approximately 90 min. Similar to the previous
position of the raw gas is performed in the CRR process scheme with the process schemes, the cascade control structure and methane

11
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 18. Dynamic response of GSP process scheme using a basic control configuration against a 10% feed flow rate increase: (a) Ethane recovery; (b) Methane
Composition; (c) NGL flow rate; (d) Reboiler Level.

composition controller have an opposite action in the demethanizer. cascade control structures, it can be noticed that the dynamic responses
More power heat is required in the stripping section to maintain the of the system are similar. The difference only resides in the higher level
level of impurity in the bottoms stream while lower temperatures are of initial overshoot observed in the cascade control with flash separator
required in the column to favor the separation of ethane and heavier temperature. In the case of methane composition (Fig. 22b), the cascade
hydrocarbons in the rectifying section. control with temperature possesses a higher overshoot and sluggish
response due to the competing action with the secondary control loop
4.5. Comparative analysis of alternative process configurations TIC-100. Finally, the difference in the separator temperatures (Fig. 22c)
is given by the fact that one of the cascade control structures utilizes
Next, it is compared the cascade control structure performances of this variable to correct and restablish the ethane recovery set-point. In
the alternative process configurations when subject to the same dis- other words, the set-point of the TIC-100 is established by the primary
turbances and working under the same operating conditions. Fig. 20 control loop ERIC-100. The variation of temperature in the cascade
illustrates the comparative results in terms of the ethane recovery control structure with ratio is minimal because the set-point of the
against a 10% increase and 10% decrease, respectively, of gas feed flow controller TIC-100 is not modified to adjust the ethane recovery.
rate. Analyzing the dynamic responses of the different controllers, it is
possible to appreciate that the conventional process scheme presents 4.7. Recovery set-point changes
the fastest responses. Both the GSP and CRR process schemes present
similar responses to an increase of the raw gas flow rate. A less sensitive An analysis of the closed-loop changes against the set-point changes
response is observed in these two other process schemes. In the case of is considered for the different studied configurations. Fig. 23a illus-
the methane composition control (Fig. 21), the conventional process trates the alternative systems behavior when sequential changes of set-
scheme not only possesses the fastest response but also the lowest point from 0.82 to 0.902 and 0.902 to 0.7667 in the ethane recovery are
overshoot. Again, the GSP and CRR process configurations present si- applied. Fig. 23b depicts a zoomed version of Fig. 23a for a decrease of
milar responses of the methane composition to a 10% increase and the recovery set-point from 0.902 to 0.7667. Both CRR and GSP process
decrease, respectively, of the raw gas flow rate. schemes can achieve the target recovery set-points smoothly and in a
reasonable settling time. There is a minimal difference in the closed-
4.6. Comparative analysis of alternative cascade configurations loop responses of these two process schemes. On the other hand, the
conventional configuration shows a very unstable closed-loop behavior
After implementing the cascade structure with the ratio controller when the set-point changes from 0.82 to 0.902. Clearly, the system is
as the secondary loop, a comparative analysis with the original cascade achieving the maximum ethane recovery point to be operated in a
system can be performed for an increase in 10% of the gas feed flow stable fashion. This is also confirmed by the smoother behavior of this
rate. Observing the ethane recovery (Fig. 22a) implementing both variable when the recovery set-point decreases from 0.902 to 0.7667.

12
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 19. Comparison of closed-loop responses using basic and cascade control configuration for GSP scheme against a 10% feed flow rate increase and 10% decrease.

Fig. 20. Comparison of closed-loop responses for ethane recovery using cascade control configuration for all schemes against: (a) 10% feed flow rate increase; (b)
10% feed flow rate decrease.

At lower values of ethane recovery, the conventional process scheme 5. Conclusions

maintains a better controllability of the system. In the case of the me-
thane composition (Fig. 24), both GSP and CRR schemes have again a Different control structures were assessed, and their dynamic re-
stable behavior with a rapid settling time. The differences in the closed- sponse was analyzed for alternative NGL recovery process schemes in-
loop responses of these two systems is again minimal. The conventional cluding the conventional, CRR and GSP. The operability of these NGL
configuration presents a much more stable behavior, but it also depicts recovery technologies was evaluated through set-point changes and the
a much higher overshoot and a slower response when compared to the introduction of process disturbances. A comparative study between a
other two process schemes. basic control configuration and a cascade structure permitted to

13
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 21. Comparison of closed-loop responses of methane mol composition using cascade control configuration for all schemes against: (a) 10% feed flow rate
increase; (b) 10% feed flow rate decrease.

Fig. 22. Comparison of closed-loop responses using cascade control configuration with temperature and ratio split for CRR scheme against a 10% feed flow rate
increase: a) Ethane recovery; b) Separator Temperature; c) Methane Mol Composition.

14
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Fig. 23. Comparative of closed-loop responses for different process schemes with cascade control (temperature control as secondary loop): (a) sequential changes of
recovery from 0.82 to 0.902 and 0.902–0.7667; (b) zoomed version of plot for recovery change from 0.902-0.7667.

Fig. 24. Variation of methane composition for different process schemes with cascade control (temperature control as secondary loop): (a) sequential changes of
recovery from 0.82 to 0.902 and 0.902–0.7667; (b) zoomed version of plot for recovery change from 0.902-0.7667.

qualitative analyze the response and dynamics of these control archi- References
tectures. The basic configuration did not include the control of ethane
recovery in any of the schemes. On the other hand, the cascade struc- Arnold, K., Stewart, M., 1999. Surface Production Operations, vol. 2 Gulf Publishing
ture included a tight control of the ethane recovery in the bottoms Company.
Astrom, K.J., Hagglund, T., 1984. Automatic tuning of simple regulators with specifica-
product stream. tions of phase and amplitude margins. Automatica 20 (5), 645–651.
In the case of the conventional unit, the lowest level of variation in Bahri, P.A., Bandoni, A., Romagnoli, J., 1996. Operability assessment in chemical plants.
ethane recovery was observed. For the two other process schemes, the Comput. Chem. Eng. 20, S787–S792.
Bahri, P.A., Bandoni, J.A., Romagnoli, J.A., 1997. Integrated flexibility and controllability
ethane recovery followed the variation of the inlet flow. In the cascade analysis in design of chemical processes. AIChE J. 43 (4), 997–1015.
control strategy, a competing effect was observed between the ethane Branan, C., 2002. Rules of Thumb for Chemical Engineers: A Manual of Quick, Accurate
recovery and the methane composition controllers for maintaining their Solutions to Everyday Process Engineering Problems, third ed. Gulf Professional
Publishing, Houston TX.
pre-established set-point values. Additionally, the conventional process Campbell R.E., Wilkinson J.D. (1981). Hydrocarbon gas processing. U.S. Patent 4,278,
scheme depicted a less stable behavior at higher levels of ethane re- 457.
covery. Both the CRR and the GSP process schemes showed excellent Campbell R.E., Wilkinson J.D., Hudson H.M. (1989). Hydrocarbon gas processing. U.S.
Patent 4,889,545.
(similar) responses when the ethane recovery set-point was increased in
Chebbi, R., Al-Amoodi, N.S., Jabbar, N.A., Husseini, G.A., Al Mazroui, K.A., 2010.
the cascade control system. Optimum ethane recovery in conventional turboexpander process. Chem. Eng. Res.
Results demonstrated the importance of understanding the dy- Des. 88 (5–6), 779–787.
namics of different control strategies. Varying composition of the gas Energy Information Administration (EIA), 2019. Annual Energy Outlook 2019 with
Projections to 2050. Washington DC, US.
and volatility on hydrocarbon prices might change production goals; Faramawy, S., Zaki, T., Sakr, A.E., 2016. Natural gas origin, composition, and processing:
therefore, a robust and structurally sound control strategy is desired to a review. J. Nat. Gas Sci. Eng. 34, 34–54.
improve the economy of the enterprise. In the future, it is expected to Getu, M., Mahadzir, S., Long, N.V.D., Lee, M., 2013. Techno-economic analysis of po-
tential natural gas liquid (NGL) recovery processes under variations of feed compo-
approach the economic aspects of these types of processes with a focus sitions. Chem. Eng. Res. Des. 91 (7) 1272-12.
on the optimal operability of the unit to maximize the profits generated Kherbeck, L., Chebbi, R., 2015. Optimizing ethane recovery in turboexpander processes.
during a long-term planning horizon. Uncertainty in the different exo- Ind. Eng. Chem. Res. 21, 292–297.
Kidnay, A.J., Parrish, W.R., McCartney, D.G., 2011. Fundamentals of Natural Gas
genous parameters such as the costs of feedstock, cost of utilities and Processing, second ed. CRC press, Boca Raton FL.
the product prices are part of the future work derived from this study. Li, Y., Xu, F., Gong, C., 2017. System optimization of turbo-expander process for natural
gas liquid recovery. Chem. Eng. Res. Des. 124, 159–169.

15
J. Chebeir, et al. Journal of Natural Gas Science and Engineering 71 (2019) 102974

Luyben, W.L., 1992. Practical Distillation Control. Springer Science & Business Media, economic evaluation of a novel NGL recovery scheme with nine patented schemes for
New York NY. offshore applications. J. Nat. Gas Sci. Eng. 27, 2–17.
Luyben, W.L., 2013. NGL demethanizer control. Ind. Eng. Chem. Res. 52 (33), Pitman, R.N., Hudson, H.M., Wilkinson, J.D., Cuellar, K.T., 1998. Next Generation
11626–11638. Processes for NGL/LPG Recovery (No. CONF-9803159-). Gas Processors Association,
Manning, F.S., Thompson, R.E., 1991. Oilfield Processing of Petroleum. PennWell Tulsa, OK.
Publishing Company, Tulsa, USA. Romagnoli, J.A., Palazoglu, A., 2012. Introduction to Process Control, second ed. CRC
Mehrpooya, M., Vatani, A., Mousavian, S.A., 2010. Introducing a novel integrated NGL press, Boca Raton FL.
recovery process configuration (with a self-refrigeration system (open–closed cycle)) Shinskey, F.G., 1996. Process Control Systems: Application, Design, and Tuning, fourth
with minimum energy requirement. Chem. Eng. Process 49 (4), 376–388. ed. McGraw-Hill, New York NY.
Mokhatab, S., Mak, J.Y., Valappil, J.V., Wood, D.A., 2013. Handbook of Liquefied Natural Tyreus, B.D., Luyben, W.L., 1992. Tuning PI controllers for integrator/dead time pro-
Gas. Gulf Professional Publishing. cesses. Ind. Eng. Chem. Res. 31 (11), 2625–2628.
Park, J.H., Khan, M.S., Andika, R., Getu, M., Bahadori, A., Lee, M., 2015. Techno-

16