11th

11th IFAC
IFAC International
International Symposium
Symposium on
on
11th IFAC International
Advanced
Advanced Control
Control of Symposium
of Chemical
Chemical on
Processes
Processes
11th IFAC International Symposium Available online at www.sciencedirect.com
on
Advanced
Venice,
11th IFAC
Venice, Control
Italy, June of Chemical
June 13-16,
International
Italy, 2021 Processes
2021
Symposium
13-16, on
Advanced
Venice, Control
Italy, June of Chemical
13-16, 2021 Processes
2021 IFAC
Advanced ADCHEM
Control of Chemical
2021 IFAC
Venice, ADCHEM
Italy, June 13-16, 2021 Processes
2021 IFAC
Venice,
2021
2021
IFAC
Venice,
IFAC
ADCHEM
Italy, June 13-16, 2021
ADCHEM
Italy, June 13-16, 2021
ADCHEM
ScienceDirect
Venice, Italy, June 13-16, 2021
Venice, Italy, June 13-16, 2021
IFAC PapersOnLine 54-3 (2021) 182–187
Control of a natural gas liquid recovery plant in a GSP unit
Control
Control of a natural gas liquid recovery plant in aa GSP unit
Control of a
a natural
ofunder
natural
under feed
gas
feed and
gas
and
liquid
liquid recovery
composition
recovery
composition
plant
plant in
disturbances
in
disturbances a GSP
GSP unit
unit
under feed
under feed and
and composition
composition disturbances
disturbances
Marta
Marta Mandis*,
Mandis*, Jorge
Jorge A.
A. Chebeir**
Chebeir**
Stefania Marta Mandis*, Jorge A. Chebeir**
Stefania Tronci*,Tronci*, Roberto
Roberto Baratti*,
Marta Mandis*, Jorge A. José
Baratti*, José A.
A. Romagnoli**
Chebeir** Romagnoli**
Stefania Tronci*, Marta Mandis*, Jorge A. José
Roberto Baratti*, Chebeir**
A. Romagnoli**
Stefania Tronci*, Roberto Baratti*, José A. Romagnoli**
*Dip.
*Dip. di di Ingegneria
Ingegneria Meccanica, Stefania
Meccanica, Tronci*,
Chimica
Chimica ee dei
deiRoberto
Materiali,
Materiali,
Baratti*,
Università
UniversitàJosédegliA. Romagnoli**
degli Studi
Studi di di Cagliari,
Cagliari, Cagliari
Cagliari 0912309123 Italy Italy
*Dip. di Ingegneria
(e-mail: Meccanica, Chimica stefania.tronci@dimcm.unica.it;
m.mandis@dimcm.unica.it; e dei Materiali,  Università degli Studi di Cagliari, Cagliari 09123).Italy
roberto.baratti@dimcm.unica.it
*Dip. (e-mail: m.mandis@dimcm.unica.it;
di Ingegneria Meccanica, Chimica stefania.tronci@dimcm.unica.it;
e dei Materiali, Università degli Studi roberto.baratti@dimcm.unica.it
di Cagliari, Cagliari 09123).Italy
*Dip. di Ingegneria
(e-mail:
** Meccanica,
m.mandis@dimcm.unica.it; Chimica stefania.tronci@dimcm.unica.it;
e dei Materiali, Università degli Studi di Cagliari,LA
roberto.baratti@dimcm.unica.it Cagliari 09123 ).Italy
** Department
(e-mail:Department of
of Chemical
Chemical Engineering,
m.mandis@dimcm.unica.it; Engineering, Louisiana
Louisiana State State University,
stefania.tronci@dimcm.unica.it; Baton
Baton Rouge, Rouge, LA 70809
University,roberto.baratti@dimcm.unica.it 70809 USA USA ).
(e-mail: m.mandis@dimcm.unica.it;
** Department of Chemical Engineering, stefania.tronci@dimcm.unica.it;
Louisiana
(e-mail: jchebe1@lsu.edu; State University, roberto.baratti@dimcm.unica.it
Baton Rouge, LA 70809 USA ).
** Department of Chemical Engineering, (e-mail: Louisiana Statejose@lsu.edu
jchebe1@lsu.edu; jose@lsu.edu
University, Baton )) Rouge, LA 70809 USA
** Department of Chemical Engineering, Louisiana Statejose@lsu.edu
(e-mail: jchebe1@lsu.edu; University, Baton ) Rouge, LA 70809 USA
(e-mail: jchebe1@lsu.edu; jose@lsu.edu )
Abstract:
Abstract: Recent technological improvements
Recent technological (e-mail: jchebe1@lsu.edu;
improvements have driven
have driven the jose@lsu.edu
the rapid increase
rapid increase ) in in natural
natural gasgas production
production from from
Abstract:
unconventional Recent technological
reservoirs. The improvements
heaviest have driven
hydrocarbon the rapid
fraction of increase
this fossil in
fuel,natural
the gas production
so-called natural from
gas
unconventional reservoirs. The heaviest hydrocarbon fraction of
Abstract: Recent technological improvements have driven the rapid increase in natural gas production from this fossil fuel, the so-called natural gas
Abstract:
unconventional
liquids (NGL), Recenthave
(NGL), technological
reservoirs.
greaterThe improvements
heaviestinterest
economic have
hydrocarbon driven
justifying the rapid
fraction
the of this
attention increase
fossil
on its itsin
fuel,natural gasprocess
the so-called
separation production fromfrom
natural gas
the
liquids
unconventional have greater
reservoirs. Theeconomic
heaviestinterest
hydrocarbonjustifying the
fraction attention
of this fossil on separation
fuel, process
the so-called from
natural the
gas
unconventional
liquids
raw gas. (NGL),
gas.(NGL),
Various reservoirs.
have greater
process The heaviest
economic
schemes have hydrocarbon
interest
been justifying fraction
the of this
attention fossil
on itsfuel, the
separation so-called
process natural
from gas
the
raw
liquids Varioushave process
greaterschemes
economic been developed
haveinterest justifyingand
developed the studied
and attentionfor
studied the
foronthe NGL
itsNGL recovery,
recovery,
separation including
including
process from the the
liquids
raw gas.(NGL),
conventional, Various have
cold greater
process
residue economic
schemes
recycle haveinterest
been justifying
developed the studied
and attention foronthe itsNGL separation
recovery, process from
including the
conventional,
raw gas. Various coldprocess recycle (CRR),
residueschemes (CRR),
have been and the
the gas
anddeveloped gas subcooled
subcooled
and studied process
processfor the(GSP).
(GSP). NGLThis
This study
study aims
recovery, aims to
to assess
including assess
the
raw gas. Various
conventional,
different cold process
residueschemes
recycle have
(CRR), been developed
anddetermine and studied
the gas subcooled processfor the(GSP). NGL recovery,
This including
studyprocess
aims to assess the
different control
conventional, control coldstrategies
residue for
strategies for aa GSP
recycle GSP unit
unit and
(CRR), and
anddetermine the
the most
the gas subcooled appropriate
most process
appropriate (GSP). and effective
andThis
effective aims tocontrol
studyprocess control
assess
conventional,
different For
scheme. control cold
this, theresidue
strategies
dynamic recycle
for GSP (CRR),
aresponses unitfor and
and thecontrol
gas subcooled
determine
each the most process
scheme appropriate
are (GSP).
evaluated andbyThis studyprocess
effective
changing aims
feed tocontrol
flow assess
rate
scheme. For
different this,strategies
control the dynamic for aresponses
GSP unitfor andeach controlthe
determine scheme are evaluated
most appropriate andbyeffective
changingprocess feed flow rate
control
different
scheme.
and control
For this,strategies
the dynamic for aresponses
GSP unitfor and determine
each controlthe most appropriate
scheme are evaluated andby effective
changing process
feed control
flow rate
and composition.
scheme.composition.
For this, the The main
main targets
Thedynamic targets are
are the
responses theforachievement
each controlof
achievement 84%
84% ethane
ofscheme ethane recovery
recoveryby
are evaluated and low
low levels
andchanging levelsfeedof methane
of flow
methane rate
scheme.
and For
composition.
impurity at this, the
The dynamic
main targetsresponses
are thefor each
achievementcontrol ofscheme
84% are
ethane evaluated
recovery by
andchanging
low levelsfeedof flow
methane rate
impurity at the
and composition. the bottom
bottom
The main of
of the demethanizer
thetargets
demethanizer column.
column. Due
are the achievement Dueofto the
the high
to84% high
ethane cost
cost of
of composition
recovery composition analyzers
analyzers
and low levels and
and the
of methane the
and
high composition.
impurity delaysat the The
bottom
introduced main
ofbythetargets
composition are
demethanizer the achievement
column.
controllers Due
underofto84%
the
the ethane
high
presence costrecovery
of
of and
composition
flow low levels
analyzers
disturbances, of
the methane
and
controlthe
high delays introduced by composition controllers under the presence
impurity at the bottom of the demethanizer column. Due to the high cost of composition analyzers and the of flow disturbances, the control
impurity
high
goals delays
are at the bottom
introduced ofbythe demethanizer
composition column.
controllers Due
under to the
the high
presence cost of
of composition
flow analyzers
disturbances, the and
controlthe
are reached
goalsdelays
high reached
introduced by
by the
thebyknowledge
knowledge
composition of
of on-line
on-line
controllers temperature
under themeasurements.
temperature measurements.
presence of flow This is
is done
done by
Thisdisturbances, by considering
considering
the control
high
goals
differentdelays
are introduced
reached by the by composition
knowledge of controllers
on-line under
temperature the presence
measurements. of flow This disturbances,
is done by the control
considering
different
goals are temperature
reached by control
temperature control
the knowledgestructures
structures of such
such
on-line as
as the
the direct
direct temperature
temperature temperature
measurements. control
controlThis and
and is cascade
cascade
done by control,
control, plus
plus aa
considering
goals
different
pressure are temperature
reached by control the knowledge
structures of such
on-line temperature
as the direct measurements.
temperature This
and is
controldisturbances, done by
cascade considering
control, plus a
pressure compensator.
different compensator.
temperature control The
The results
results are
are compared,
structures compared,
such as the in presence
in direct
presence of
of composition
composition
temperature and cascade with
controldisturbances, with the
the action
control, action
plus a
different
pressure
of a hybrid temperature
compensator.
cascade control
The results
control structures
that uses suchdelayed
arein-line
compared, as thein direct
presence temperature
concentrationof composition controldisturbances,
measurements and to cascade
update control,
with
the the plus a
action
controller
of a hybrid
pressure cascade control
compensator. that uses
The results arein-line
compared, delayed concentration
in presence measurements
of composition to updatewith
disturbances, the thecontroller
action
pressure
of a hybrid
reference compensator.
at cascade
each sampling The period.
control results
that uses are
Here, compared,
in-linethe delayed
hybrid in and
presence
concentration
the of composition
simple measurements
cascade disturbances,
controls toshow
update with
thethe the
best action
controller
control
reference
of a hybridatcascade
each sampling control period.
that uses Here,
in-linethe delayed
hybrid and the simple cascade
concentration measurements controls toshow
update thethebest control
controller
of a hybridatcascade
reference
performance. each sampling control period.
that uses in-line
Here, the delayed
hybrid and concentration
the simple cascademeasurements controls toshow
update thethebestcontroller
control
performance.
reference at each sampling period. Here, the hybrid and the simple cascade controls show the best control
reference
performance. at each sampling period. Here, the hybrid and the simple cascade controls show the best control
Copyright © 2021 The Authors. This is an open access article under the CC BY-NC-ND license
performance.
performance.
(http://creativecommons.org/
Keywords:
Keywords: Natural gas
Natural liquidslicenses/by-nc-nd/4.0)
gas liquids recovery, Dynamic
recovery, Dynamic process process simulation,
simulation, PressurePressure compensated
compensated temperature,
temperature,
Keywords:
Hybrid Natural
cascade gas
control, liquids
Discrete recovery,
composition Dynamic process simulation, Pressure compensated temperature,
control.
Hybrid cascade control, Discrete composition control.
Keywords: Natural gas liquids recovery, Dynamic process simulation, Pressure compensated temperature,
Keywords:
Hybrid cascade Natural gas liquids
control, Discrete recovery,
composition Dynamic process simulation, Pressure compensated temperature,
control.
Hybrid cascade control, Discrete composition control.
Hybrid cascade control, Discrete composition control. 
1. 
1. INTRODUCTION
INTRODUCTION  schemes was
schemes was the the conventional
conventional turboexpander
turboexpander process. process. In In this
this
1. INTRODUCTION  schemes
unit, was
refrigeration the conventional
is supplied turboexpander
through the process.
implementation In this
of
Among
1.
Among the
the different
INTRODUCTION fossil energy
different fossil energy sources,
sources, the the natural
natural gas gas unit, schemes refrigeration is supplied through
was the conventional the implementation
turboexpander process. In this of
1.
AmongINTRODUCTION
the different fossil energy sources, the natural gas schemes
unit,
pressure was
refrigeration
jump the conventional
is
between suppliedthe turboexpander
through
inlet the
stream and process.
implementation
the In
nominalthis
of
represents
Among theone
represents one of
of the
the cheapest
different cheapest
fossil energyand
and cleanest.
cleanest.
sources, This
Thistheenergy
energy
natural source
gas pressure
source jump between
unit, refrigeration is supplied the inlet
through stream and the nominal
the implementation of
Among
represents theone different
of the fossil energy
cheapest and sources,
cleanest. Thistheenergy
natural unit,
pressure
gas conditions refrigeration
jump is
between suppliedthe through
inlet the
stream implementation
and the nominal of
is
is mainly
mainly used
represents used
one of as an
an alternative
as the alternative
cheapest and to more
more traditional
to cleanest. traditional
This energy fossilsource
fossil fuels
fuels conditions
source pressure jump insidebetween
inside the column
the column by means
by
the inlet means
stream of and
of aa turbo-expander
turbo-expander
the nominal
represents one pressure
conditions jump
inside between
the columnthe inlet
by stream
means of and
a the nominal
turbo-expander
is
such
such
mainly
as oil
as
is mainly andof
oilused
and
used
as the
coal.
coal.
as
cheapest
an alternative
The use
The use of
an alternative
and
of to cleanest.
more gas
natural
natural
to more gas
This
traditional energy
to produce
to produce
traditional
fossil
fossil
source
fuels (Campbell
fuels (Campbell
energy
energy conditions inside and Wilkinson,
and Wilkinson,
the column 1981).
1981). Starting
Starting
by means from
of from this process
this
a turbo-expander process
is
such mainly
asa oilused
and as an alternative
coal. use of to
The(Kidnay more
natural traditional
gas to produce fossil fuels
energy conditions
(Campbell
scheme, plantinside the column
andmodifications
Wilkinson, andbyenergy
1981). means
Starting of from
a turbo-expander
integrations this
have process
been
is
is
such not recent
notasa oil
recent practice
practice
and coal. The(Kidnay
use of naturalet al.,
et al., gas2011),
2011), but in
but in the
to produce the last scheme,
(Campbell
last (Campbell
energy plantand modifications
Wilkinson, and
1981).energy
Startingintegrations
from have
this been
process
such
is notasa oil and
recent coal.
practice The use
(Kidnay of natural
et al., gas
2011), to produce
but in energy
the last scheme,
realized plant
to and
improveWilkinson,
modifications the 1981).
and
separation Starting
energy performance from
integrations this
have
and process
been
reduce
decades
decades its utilization
its utilization
is not a recent practicehas has increased
increased
(Kidnay due
due2011),
et al., to improvements
to improvements
but in the last in realized
scheme,
in scheme, to
plant improve
modifications the separation
and energy performance
integrations and
have reduce
been
is not
decades a recent
its practice
utilization has(Kidnay
increased et al.,
due 2011),
to but
improvementsin the last
in realized plant
operating tocosts. modifications
improve Among the and
the separationnewenergy
process integrations
performance schemes, have
and the been
reduce
Gas
extraction
extraction techniques
techniques including
including drilling
drilling
decades its utilization has increased due to improvements in realized and
and hydraulic
hydraulic operating
realized to costs.
improve Among the the new
separation process
performance schemes, and the Gas
reduce
decades
extraction its utilization
techniques has increased
including due
drilling to improvements
and hydraulic in operating
Subcooled tocosts.
improve
Process Among
(GSP)the separation
the
is new
one of performance
process
the most schemes,
widely and reduce
the
employed Gas
fracturing.
fracturing.
extraction This This has led
has
techniques led to
toincluding
an upsurge
an upsurge drilling
in natural
in natural and gas availability
gas availability
hydraulic Subcooled operating costs. Process (GSP) the
Among is onenew of process
the mostschemes,widely employed the Gas
extraction
fracturing. techniques
This has led to including
an upsurge drilling
in natural and
gas hydraulic
availability operating
Subcooled
units in costs.
Process
industry. Among
(GSP)
GSP the
is
utilizesnew
one of process
thethe most schemes,
split-vapourwidely the Gas
employed
concept to
and
and aa decrease
decrease
fracturing. Thisin inhas
itsled
its selling
selling price,
to anprice,
upsurge dropped
dropped
in natural dramatically
dramatically to less
to
gas availability less units Subcooledin industry.
Process GSP (GSP)utilizesis one of thethesplit-vapour
most widelyconcept employed to
fracturing.
and a This
decrease inhas
its led to
selling an upsurge
price, in
dropped natural gas
dramatically availability
to less Subcooled
units
obtain ina Processreflux
industry.
column (GSP)that
GSP is one
utilizes of
the
contacts thesplit-vapour
mostrectifies
and widelyconcept
employed
the to
vapor
than
than
and a30%30% of its
of
decrease itsinprevious
previous
its selling highs
highs
price,(Luyben,
(Luyben,
dropped2013).2013). Naturalto
Natural
dramatically gas
gas is obtain
is
less units ina industry.
column reflux GSP that contacts
utilizes and rectifiesconcept
the split-vapour the vapor to
and
than a30%
decrease itsinprevious
of occurringits selling price,
highs dropped
(Luyben, dramatically to less units
obtain ina industry.
column GSP
reflux utilizes
that the
contacts split-vapour
and rectifies concept
the to
vapor
aathan naturally
naturally
30% of occurring hydrocarbon
its previoushydrocarbon
highs (Luyben, gas2013).
gas mixture
mixture
2013).
Natural
Natural
gas
consisting
consisting
is leaving
gas is leaving obtain a the the
columnturboexpander
turboexpander
reflux thatand and entering
entering
contacts and thethe column
column
rectifies the in in the
the
vapor
than 30% obtain
leaving a column
the reflux
turboexpander that contacts
and entering and rectifies
the column the vapor
in the
amainly
mainly of of
naturally
of
a naturally
its previous
occurring
methane,
methane,
occurring but
highs (Luyben,
hydrocarbon
but commonly
commonly
hydrocarbon
gas2013).
including
including
mixture
gas mixture
Natural
varying
varying
gas is lower
consisting
amounts
amounts
consisting lower stages
leaving stages (Pitman et
(Pitman
the turboexpander et andal., entering
al., 1998). The
1998). The Cold Residue
Cold
the column Residue
in the
amainly
naturally occurring
of methane, hydrocarbon
but commonly gas
including mixture
varying consisting
amounts Recycling leaving
lower
Recycling the(CRR)
stages turboexpander
(Pitman
process and
al., entering
etrepresents1998). an the column
The
extension Cold of the inGSP
Residuethe
of of heavier
heavier
mainly hydrocarbons.
hydrocarbons.
of methane, The presence
The presence
but commonly of these
of these
including hydrocarbons,
hydrocarbons,
varying amounts lower lower stages(CRR) process
(Pitman etrepresents
al., 1998). an extension
The Cold of the GSP
Residue
mainly
of of
heavier methane,
hydrocarbons. but commonly
The presenceincluding
of thesevarying amounts
hydrocarbons, Recycling
design. stages
CRR (CRR) (Pitman
retains process
all the et al.,
represents
therepresents 1998).
main advantages an
advantages The
extension Cold
of the
the ofof
GSP Residue
the GSP
while
known
known as NGL
as
of heavier NGL (Natural Gas
(Natural
hydrocarbons. Gas Liquids),
TheLiquids), represents
presencerepresents an additional
an
of these hydrocarbons,additional design. Recycling CRR (CRR)retains all
process main an extension of GSP thewhile
GSP
of heavier
known as hydrocarbons.
NGL (Natural TheLiquids),
Gas presence of these hydrocarbons,
represents an additional Recycling
design. CRR
adding a (CRR)retains
column process
all therepresents
reflux main
stream ofannearly
advantages extension the of
of pure GSP thewhile
GSP
methane
source
source of
of profit:
profit: these
these compounds
compounds are
are separated
separated
known as NGL (Natural Gas Liquids), represents an additional design. and
and used
used as
as adding
design. a
CRR column
retains reflux
all the stream
main of
advantages nearly of pure
the GSP methane
while
known
source as
of NGL
profit: (Natural
these Gas
compoundsLiquids), are represents
separated anandadditional
used as adding
(Wilkinson CRR
a retains
column
and Hudson, all the1992).
reflux main
stream advantages
Thisof nearly
enhances of pure
thethe GSPlevelswhile
methane of
feedstock
feedstock
source of forfor different
different
profit: industrial
these industrial
compounds processes.
processes.
are separated Consequently,
Consequently,
and usedthe the
as (Wilkinson
adding a column and Hudson, reflux1992). streamThis of enhances
nearly pure the levels
methane of
source
feedstockof profit:
for these
different compounds
industrial are
processes. separated and
Consequently, used as
the adding
(Wilkinson
ethane a column
recovery and in Hudson,reflux
the streamThis
1992).
separation of enhances
process. nearly
The pure
NGL methane
the levels
recovery of
NGL NGL havefor
have
feedstock significantly
significantly higher value
higher
different industrial value as separate
as
processes. separate marketable
marketable
Consequently, the ethane (Wilkinson recovery and in the separation
Hudson, 1992). This process. The NGL
enhances recovery
the levels of
feedstock for different (Wilkinson
ethane recovery and Hudson,
in the 1992).
separation This
process.enhances
The NGLthe levels
recovery of
NGL
products
have
products
NGL have than
than
significantly
as part
as part of
significantly ofindustrial
higher
the
the
higherraw
processes.
rawvalue
gas (Mokhatab
gas
value (Mokhatab Consequently,
as separate
as separateet etmarketable
al., 2015).
al.,
the process
2015).
marketable process is normally
is
ethane recovery normally influenced
in theinfluenced
separationby byprocess.
disturbances
disturbances
The NGL in operating
in operating
recovery
NGL have
products thansignificantly
as part of thehigherraw value as
gas (Mokhatab separate etmarketable
al.,on2015). ethane
process
variables recovery
is normally
such in the
as pressure,
pressure, separation
influenced by process.
temperature The
disturbances
(Chebbi in NGL in recovery
operating
et al.,
al., 2010)
For For this
this
products reason,
reason,
than as part numerous
numerous
of the raw studies
studies have
have
gas (Mokhatab focused
focused
et al.,on the variables
the
2015). process issuch as
normally influenced temperature (Chebbi
by disturbances et 2010)
operating
products
For this than
reason,asnovel
part of
numerous the raw gas (Mokhatab
studies have focused et al.,on 2015).
the and process
variables
and feed isconditions.
normally
such as pressure,influenced
The most by disturbances
temperature
common (Chebbi in
disturbances operating
et al.,
are 2010)
those
development
development
For this reason, of
of novel process configurations
process configurations
numerous studies have focused to separate
to separate these
onthese
the variables feed
variables conditions.
such as The
pressure, most common
temperature disturbances
(Chebbi et are
al., those
2010)
For this
development reason,
of novelnumerous
process studies
configurationshave focused
to separate on the
these and feed
related to such
conditions.
inlet as
flow pressure,
The
rate most
and temperature
common
composition (Chebbi
disturbances
variations. et al.,
are
The 2010)
those
flow
products.
products.
development of novel process configurations to separate these and related
and feed to inlet
conditions.flow rate
The and
most composition
common variations.
disturbances The
are flow
those
development
products. of novel process configurations to separate these rate rate feed
relatedmay to conditions.
inlet
be flow
subject toThe
rate most
and
changes common
composition
dictated disturbances
by variations.
the need toare
The those
flow
treat or
products. relatedmay to be subject
inlet flow to
rate changes
and dictated
composition by the need
variations. toThetreat or
flow
products. related
rate may
produce toabeinlet flow to
subject
smaller ratechanges
or and composition
larger dictatedof
amount byvariations.
the need while
product, toThe flow
treat or
the
Different process
Different process schemes
schemes have have beenbeen studied
studied and and developed
developed rate produce
rate may a
be smaller
subject or
to larger
changes amount
dictated of
by product,
the need while
to treat the
or
Different process schemes have beenfirst studied and developed producemay
composition be
a subject
smaller
can to
beor changes
larger
subject dictated
amount
to by
of the need
product,
fluctuations due to
while treat
to or
the
the
for for the
Different NGL
the NGL processrecovery.
recovery.
schemes One
Onehave of the
of the
beenfirst studiedproposed
proposed industrial
industrial produce
and developed composition
produce a can
smaller beor subject
larger to
amount fluctuations
of product, duewhile to the
Different
for the NGL process schemes
recovery. Onehave beenfirst
of the studied and developed
proposed composition
industrial characteristics a smaller
can
of the beor
thebenatural larger
subject
natural amount
to
gas to
extraction of product,
fluctuationsbasin. due duewhile to
To maintain
maintain the
for the NGL recovery. One of the first proposed industrial characteristics composition can of subject gas extraction
fluctuationsbasin. To to the
for the NGL recovery. One of the first proposed industrial composition characteristicscan
characteristics
of thebenatural
of the
subject
natural
gas to
gas
fluctuations
extraction
extraction
basin. due
basin.
to the
To maintain
To maintain
2405-8963 Copyright © 2021 The Authors. This is an open access article under the CC BY-NC-ND
characteristics of the license. gas extraction basin. To maintain
natural
Copyright © 2021 IFAC
Copyright 182
Peer review©under
2021 responsibility
IFAC 182Control.
of International Federation of Automatic
Copyright © 2021 IFAC 182
10.1016/j.ifacol.2021.08.239
Copyright © 2021 IFAC 182
Copyright © 2021 IFAC 182
 Marta Mandis et al. / IFAC PapersOnLine 54-3 (2021) 182–187
the product specifications of NGL recovery plants, it is
necessary to analyse the process dynamics and design a control
system capable of mitigating or eliminating the effect of these
disturbances.
Several NGL recovery processes have been studied in the
literature by Manning and Thompson (1991), Kidnay et al.
(2011), Chebbi et al. (2010), Park et al. (2015) and Kherbeck
and Chebbi (2015). However, these studies do not include the
inherent complexities involved in the dynamics of the process
and their control. The direct control of concentrations in the
presence of disturbances in feed flow rate and inlet
composition has been analysed by Luyben (2013) and Chebeir
et al. (2019). In terms of realization, this type of control
strategy is simple, but it possesses disadvantages related to the
long delay times inside the control loops resulting from the
measurement times of instruments and the high purchase and
maintenance costs of equipment. An alternative to direct
control is the indirect control of composition through
temperature measurements (Hori and Skogestad, 2007). With
this type of control, the choice of the tray with the highest
temperature sensitivity and the consideration of non-key
components impacting on the column temperature profile are
essential in cases of multi-component columns.
In this work, the dynamic behaviour of the system resulting
from the separation of NGL in a GSP unit is investigated under
the effect of disturbances in flow rate and input composition,
by comparing different control strategies. Particularly, an
improvement of the bottom methane impurity control in the
demethanizer column is achieved through the implementation
of a ratio flow controller in cascade with temperature
controller, with and without remote setpoint selection by
delayed composition measurements. Additionally, a pressure
compensator has been added in the separator to enhance the
ethane recovery control. The purpose of this study is to
evaluate the use of alternative control schemes concerning the
maintenance of product specifications in the demethanizer
column.
2. BACKGROUND
2.1 Flowsheet
The NGL separation process that takes place in a GSP unit,
depicted in Figure 1, has been analysed by means of a dynamic
simulation using Aspen HYSYS® and based on realistic
operating conditions (Chebeir et al., 2019). The raw gas is fed
to the plant with a flow rate of 4980 kmol/h, at the pressure
condition of 5818 kPa and a temperature of 35 °C. The
composition with a low content of liquids is reported in Table
1. The specifications required by the GSP separation unit are
to achieve an ethane recovery of 84% and a methane
composition of 1 mol% at the bottom of the demethanizer
column (T-100).
The main part of the separation unit consists of a 30 stages
demethanizer column with a reboiler (E-103) and no
condenser. A separator (TK-100) is placed upstream of the
column to remove the liquid fraction of the feed gas stream.
183
183
Then, this condensed vapor is introduced into stage 26. Part of
the vapor is sent to a turboexpander (TE-100) and fed on stage
8. The remaining vapor is cooled, expanded and fed to the top
of the column.
An important variable for ethane recovery is given by the
separator temperature TI-100, which depends on the
conditions achieved in the chiller (E-101) used to precool the
feed gas. The remaining refrigeration needed to meet the
column nominal temperature is provided by expanding the
column feeds through the turboexpander and Joule-Thompson
valves (JT-100 and JT-101).
Table 1. Feed composition (Chebbi et al., 2010)
Components Mole fractions
Nitrogen 0.01
Methane 0.93
Ethane 0.03
Propane 0.015
Butanes 0.009
Pentanes 0.003
Hexanes 0.003
%C2+ 6
2.2 Temperature sensor placement
For implementing an indirect composition controller to
maintain the level of methane impurity in the column bottom
product below a specific value, the selection of the temperature
sensor position is of most importance. In a multicomponent
column, the best position to place a temperature sensor is in
the section where the sensitivity to temperature changes is
higher. Consequently, an analysis of the column temperature
and composition gradients has been carried out. To relate the
influence of components variation on the temperature gradient
inside the column, the analysis of the temperature gradient
with a per-component contribution diagram (Porru et al., 2013)
has been performed. In this manner, the best position for the
temperature sensor is determined to be in tray 28.
3. SYSTEM DYNAMICS AND CONTROL STRUCTURE
Under the effect of feed and composition disturbances, the
achievement of the control objectives has been evaluated
considering different control configurations. This work
proposes a modification of the control structure realized by
Chebeir et al. (2019) for the same plant, as it is depicted in
Figure 1. First, basic indirect controls of concentrations
through temperature control have been applied in order to
achieve both ethane recovery and methane concentration
targets in the column bottom product. Subsequently, several
improvements have been examined to eliminate or mitigate, as
far as possible, the steady states offsets.
3.1 Indirect temperature controllers
To reach the target values of 84% for ethane recovery and 1
mol% for methane concentration in the demethanizer bottom,
two indirect composition controllers (hereafter indicated as C1
configuration) have been implemented using on-line
184 Marta Mandis et al. / IFAC PapersOnLine 54-3 (2021) 182–187
temperature measurements in the separator and column. In the
first control loop, the controlled variable is the separator
temperature TIC-100, which is directly connected with the
ethane recovery. The manipulated variable is the chiller duty.
In the second control loop, the controlled variable is
represented by the tray temperature TIC-103 at stage 28,
related to the concentration of methane, by manipulating the
reboiler duty.
Figure 1: Schematic of the GSP unit and control structures: in
red the conventional configuration (C1); in blue the C3
configuration.
3.2 Pressure compensate temperature
The feed flow variations are induced by manipulating the lines
pressure through the valve upstream the separator (valve FCV-
100 in Figure 1). This implies that the separator pressure PI-
100 also undergoes variations. Since the separator temperature
TIC-100 is maintained constant by the control action, pressure
variations result in changes in the separator outflow
composition that influences the ethane recovery. With the
purpose of improving the separator direct temperature control
and, consequently, the ethane recovery, a pressure
compensator is placed in this unit. A new control variable has
been considered within the loop instead of the separator
temperature. This variable is represented by a fictitious
temperature, calculated on basis of Antoine’s law. This is a
Pressure Compensated Temperature (PCT) estimated by using
(1) (Brambilla, 2014):
𝑃𝑃
𝑃𝑃𝐶𝐶𝑇𝑇 = 𝑇𝑇𝑚𝑚 − 𝐶𝐶 ln
𝑃𝑃𝑟𝑟𝑟𝑟𝑟𝑟
where Tm is the measured separator temperature TI-100; P is
the separator pressure PI-100; Prif is the separator pressure
nominal value; and C is the compensation coefficient obtained
from temperature data recorded during the imposition of
pressure variations at constant composition with Aspen
HYSYS®. By controlling the PCT (control configuration
indicated with C2), it is possible to link the temperature
variations to the pressure changes in the separator and mitigate
the resulting concentration disturbances.
3.3 Indirect methane cascade control
Since the flow dynamics are faster than the ones of
temperature and composition, a ratio flow controller FIC-101
is introduced in the control strategy. The resulting control is a
cascade control system (this combined with PCT control is
hereafter indicated with C3) used to speed up the responses
obtained by the direct column tray temperature controller. The
external loop is represented by a temperature controller TIC-
103 that gives the setpoint to the internal loop. The internal
loop is composed by a flow controller, the purpose of which is
to maintain constant the ratio between the boilup and the
column bottom product, by manipulating the reboiler duty. It
is important to note that the ratio controller could guarantee the
required methane composition at the bottom of the
demethanizer column in case of feed flow variations, but it
fails when feed composition changes. In this case, a
temperature controller is required (Shinskey, 1996).
3.3.1 Boilup approximation
Considering that a boilup measurement is not generally
available in real plants, an estimation of this variable is needed
to implement the cascade control structure. A boilup
approximation is obtained by the process information
available and based on a balance around the reboiler expressed
by (2):
𝜆𝜆𝑉𝑉 + 𝑐𝑐𝑐𝑐(𝑇𝑇𝐵𝐵 − 𝑇𝑇𝐿𝐿 )(𝑉𝑉 + 𝐵𝐵) = 𝑄𝑄 (2)
Here, 𝐵𝐵 is the bottom product stream FI-102; 𝑉𝑉 is the actual
boilup; 𝑄𝑄 is the reboiler duty; λ is the latent heat of
vaporization of the mixture; cp is the specific heat capacity at
constant pressure; TB and TL are the temperature of the stream
𝐵𝐵 (TI-102) and the temperature of the liquid stream entering
the reboiler (TI-101), respectively. Considering different
values of inlet flow rate, the open-loop response of 𝑉𝑉, 𝐵𝐵, TL,
and TB are registered by using the model testing function of
Aspen HYSYS® on the reboiler duty with an amplitude of 2%.
These data are used to perform a multi-linear regression
considering the regression model (3):
𝑉𝑉̂ = 𝑝𝑝0 + 𝑝𝑝1 𝐵𝐵 + 𝑝𝑝2 𝑇𝑇𝐵𝐵 + 𝑝𝑝3 𝑇𝑇𝐿𝐿 (3)
where 𝑉𝑉̂ is the boilup estimation. The regressors considered
(1) are the registered data of 𝐵𝐵, TL, and TB while p0, p1, p2, p3 are
the model parameters.
3.3.2 In-line methane measurements
To compare this contribution with the studies developed by
Luyben (2013) and Chebeir et al. (2019), the presence of a gas
chromatograph is assumed in the plant. In general, only
delayed and discontinuous composition measurements are
available due to the typical delay in the in-line analysers, not
considered in the mentioned contributions. To approach a real
situation, it is assumed that the gas chromatograph is used to
analyse the product composition of all the distillation units
present in the distillation train. This is a common practice in
184
 Marta Mandis et al. / IFAC PapersOnLine 54-3 (2021) 182–187 185

industrial plants, where only discrete and delayed composition By observing the top graph in Figure 2, it is possible to notice
measurements are available for the controller. Since further that the two temperature controllers are able to bring the
NGL separation requires other two distillation columns process variables to the setpoint value after approximately 80
downstream the demethanizer and considering an analysis min with the proposed control and 95 min with the
time of 10 minutes, reasonable for this type of mixtures, a conventional one. In the bottom graph, we can observe that
sampling time of 30 minutes has been assumed. This takes into none of the two control structures is able to meet the ethane
account the time needed by the instrument to complete a recovery specification.
measuring cycle in the entire plant. Thanks to the knowledge
of this measurement, another controller is added to the above-
mentioned cascade. In particular, the discrete methane
composition controller only acts in each delayed sampling
time to adjust the shot of the temperature controller setpoint,
leading to the implementation of a hybrid cascade controller
(hereafter indicated with C4). To compare the hybrid control
results with the conventional control ones, the methane
concentration controller is also used in cascade with the
conventional temperature controller (control scheme C5). The
tuning parameters, obtained using the improved internal model
control (IMC) approximate model controller tuning rules
(Ogunnaike and Ray, 1994), are reported in Table 2 for the
different control structures.

Table 2. PID tuning parameters

Controllers Configuration Kc τi
Temperature C1; C5 1.222 19.45
Concentration C5 0.18 9
Temperature C3; C4 0.802 11.5
Ratio C3; C4 10.992 8.6
Concentration C4 0.3 9
Figure 2: Responses obtained by a decrease of 10% in the feed
4. RESULTS nominal value of 4980 kmol/h for the ethane recovery.
With the purpose of meeting the control objectives, the
different control structures have been evaluated under the The new steady state values for the ethane recovery are 76%
presence, in the feed gas to the plant, of 10% variation in the with the conventional control, and 85% by controlling the PCT
flow rate and 40% variation in the ethane concentration. These with the C2 control. The ethane recovery profile obtained by
variations are chosen in accordance with the feed changes used means of PCT control shows a lower initial deviation and a
by Chebeir et al. (2019) and the composition changes used by lower offset at the new steady states, demonstrating that the
Luyben (2013). In this section, a comparison between the concentration upsets due to the feed variations are reduced. In
results obtained through the conventional control and the the middle graph, the behaviour of the manipulated variable is
proposed control structures is performed. These results are reported. Here, the control action obtained with the proposed
relative to the worst-case variations corresponding to a C2 control shows a lower variability.
decrease in the feed flow rate and the ethane concentration.
4.2 Methane bottom impurity level under feed disturbances
4.1 Ethane Recovery under feed disturbances
The action of the proposed C3 (cascade control while
The ability to reach the ethane recovery target has been controlling the PCT variable in the separator) and conventional
evaluated by analyzing the dynamic responses of the GSP unit C1 (direct temperature control) control schemes to limit the
under feed disturbances. A comparison between the methane impurity concentration have been evaluated by
conventional control (C1: direct temperature control in the comparing the profiles obtained in response to feed
separator) and the proposed control (C2: direct PCT control) disturbances. The results are depicted in Figure 3.
is performed. Particularly, the use of a PCT control instead of By looking at the tray temperature profiles depicted in the top
temperature control is addressed to overcome the pressure graph of Figure 3, it can be noticed that the C3 control structure
changes induced in the separator by the inlet flow rate has a faster response (it reaches the setpoint value of -90°C
variations to the plant. This is an important aspect when feed after around 98 min while the C1 control after around 190 min)
flow upsets are present. The ethane recovery responses, and a lower initial variation. Observing the methane
performed without concentration control, for a decreasing of concentration profiles, in the bottom graph, it is noted that the
10% in the feed flow rate are depicted in Figure 2. conventional configuration cannot bring the methane impurity
back to the target value of 1 mol%, even if the controller brings
the controlled variable to the setpoint value. On the contrary,

185
186 Marta Mandis et al. / IFAC PapersOnLine 54-3 (2021) 182–187
the value reached with the cascade control is very close to the
target. The methane concentration achieves a new value of 2%
with the C1 control and a value of 1.03% with the C3 control,
which also exhibits a lower initial deviation.
Figure 3 : Responses obtained by a decrease of 10% on the
feed nominal value of 4980 kmol/h for the methane impurity
level.
Under inlet flow disturbances, as displayed on Figure 2 and 3,
the use of the PCT as a controlled variable has a positive
influence not only on the ethane recovery but also on the
methane impurity concentration. Overall, the offset values of
ethane recovery and methane concentration are drastically
reduced. With the pressure compensator, we can maintain
almost constant the inlet column compositions, leading to an
improvement of the control performances. The responses
obtained by utilizing the cascade control are generally faster
than those using direct temperature control on the plate. This
is due to the implementation of the ratio controller. As the
dynamics of the flows in the column are faster than those of
temperature and concentration, a more aggressive control is
attained. Thus, a cascade control combined with pressure
compensator can reduce the deviation from the target.
4.3 Cascade control under composition disturbances
To consider the ability to meet the specifications of ethane
recovery and methane impurity level in the presence of inlet
composition disturbances, the conventional C1 and the
proposed C3 control schemes are compared under variations
of ethane inlet concentration. The profiles of ethane recovery
and methane concentration obtained with the conventional
control of the separator and column tray temperatures (C1) and
with the proposed PCT control and column tray temperature
controller in cascade with the ratio controller (C3) are shown
in Figure 4.
186
Observing the profiles displayed in Figure 4, we can notice that
none of the two control strategies can reach the selected ethane
recovery and methane impurity level values. In the top graph,
looking at the ethane recovery profiles, it can be noted that
improvement is not registered in the case of ethane
concentration upsets. Nevertheless, the distance between the
value obtained with the PCT control is not excessively large,
with a value of 1.8%, and a final offset of 6.6%. In the middle
graph, the methane concentration offset at the new steady-state
obtained with the C3 control is higher than the one obtained
by the C1 control. A value of 0.70% is obtained with the
cascade control while a value of 0.82% is achieved by the
column tray direct temperature control.
Figure 4: Responses obtained by a decrease of 40% in the
ethane inlet concentration nominal value of 0.03 for ethane
recovery and methane composition with C1 and C3 control.
The cascade control results under composition disturbances
can be explained by considering that the composition
dependent coefficient C in (1) has been evaluated at nominal
feed compositions. When changing inlet composition with
respect to the nominal one, the compensated temperature could
not be properly estimated.
4.4 Hybrid cascade control under composition disturbances
Exploring the possibility of improving the profiles obtained for
the methane concentration by the column tray temperature
control in cascade with the ratio controller, a hybrid cascade
(C4) control is proposed by introducing discrete methane
measurements. The comparisons between the conventional
control in cascade with a methane composition controller (C5
control) and the proposed C4 control for an increase of 40% in
the inlet ethane concentration are shown in Figure 5.
Observing the top graph in Figure 5, it can be noted that the
ethane recovery profile exhibits the same behaviour in both
control structures. By considering the methane impurity
 Marta Mandis et al. / IFAC PapersOnLine 54-3 (2021) 182–187
profiles reported in the middle graph, it is possible to observe
that the C5 control depicts a higher initial variation and a
higher speed of convergence, arriving at the steady-state
around 30 min earlier. Both configurations achieve the
methane control target at expense of speed of response. The
higher response speed achieved with C5 control is probably
due to the initial error of the composition controller. Since the
conventional initial error is higher than the proposed cascade
control error, this leads to a more aggressive initial control
action. Besides, by looking at the methane concentration
offsets depicted in Figure 4, it is possible to observe that the
conventional control initially had to overcome a lower offset,
leading to a faster response.
Figure 5: Responses obtained by a decrease of 40% on the
ethane inlet concentration nominal value of 0.03 for ethane
recovery and methane composition with the C5 control and
proposed C4 control.
5. CONCLUSIONS
In the presence of input flow rate and composition variations,
various control structures were compared with the main
objective of achieving a methane composition of 1 mol% while
maintaining an ethane recovery of 84% in a GSP recovery unit.
In the presence of variations in the inlet flow rate, it was shown
that the use of PCT in the separator as a controlled variable
reduced the collateral composition disturbance in the column
feed. Due to the implementation of the PCT control, the
compositions coming out of the separator and then entering the
column are kept almost constant. From the comparison of the
indirect methane impurity level control obtained with the
cascade control and the conventional direct temperature
control in presence of inlet flow rate disturbances, it was
possible to assert that the cascade configuration had the best
control performance. This controller had the fastest response
and maximum speed of convergence for controlling the
temperature and composition of the methane in the bottom
column stream. Under inlet composition disturbances, the
187
187
implementation of the in-line delayed methane concentration
controller in the cascade control structure, eliminated the
steady-state offsets at the cost of a slower response while
obtaining a better performance on the manipulated variable.
REFERENCES
Brambilla, A. (2014). Distillation Control and Optimization.
McGraw-Hill Education, New York, NY
Campbell, R.E., Wilkinson, J.D. (1981). Hydrocarbon gas
processing. Patent 4278457, USA.
Chebbi, R., Al-Amoodi, N.S., Abdel Jabbar, N.M., Husseini,
G.A., Al Mazroui, K.A. (2010). Optimum ethane
recovery in conventional turboexpander process. Chem.
Eng. Res. Des. 88, pp. 779–787.
Chebeir, J., Salas, S.D., Romagnoli, J.A. (2019). Operability
assessment on alternative natural gas liquids recovery
schemes. J. Nat. Gas Sci. Eng. 71, pp. 102974.
Hori, E.S., Skogestad, S. (2007). Selection of control structure
and temperature location for two-product distillation
columns. Chem. Eng. Res. Des. 85, pp. 293–306.
Kherbeck, L., Chebbi, R. (2015). Optimizing ethane recovery
in turboexpander processes. J. Ind. Eng. Chem. 21, pp.
292–297.
Kidnay, A.J., Parrish, W.R., McCartney, D.G. (2011).
Fundamentals of Natural Gas Processing. CRC Press,
Boca Raton, FL.
Luyben, W.L. (2013). NGL demethanizer control. Ind. Eng.
Chem. Res. 52, pp. 11626–11638.
Manning, F., Thompson, R. (1991). Oilfield Processing:
Natural Gas. PennWell Publishing Co. Tulsa, OK .
Mokhatab, S., Poe, W.A., Mak, J.Y. (2015). Handbook of
Natural Gas Transmission and Processing: Principles
and Practices. Gulf Professional Publishing, New York,
NY.
Ogunnaike, B.A., Ray, W.H. (1994). Process dynamics,
modeling, and control. Oxford University Press, New
York, NY.
Park, J.H., Khan, M.S., Andika, R., Getu, M., Bahadori, A.,
Lee, M. (2015). Techno-economic evaluation of a novel
NGL recovery scheme with nine patented schemes for
offshore applications. J. Nat. Gas Sci. Eng. 27, pp. 2–17.
Pitman, R., Hudson, H., Wilkinson, J., Cuellar, K. (1998).
Next generation processes for NGL/LPG recovery. Gas
Processors Association. Dallas, TX.
Porru, M., Alvarez, J., Baratti, R. (2013). Composition
estimator design for industrial multicomponent
distillation column. Chemical engineering transactions
32, pp. 1975–1980.
Shinskey, F.G. (1996). Process Control Systems: Application,
Design, and Tuning, fourth ed. ed. Mcgraw-hill
Professional, New York, NY.
Wilkinson, J.D., Hudson, H.M. (1992). Improved NGL
Recovery Designs: Maximise Operating Flexibility and
Product Recoveries. Tulsa, OK.