Preprint, 11th IFAC Symposium on Dynamics and Control of Process Systems,
including Biosystems
June 6-8, 2016. NTNU, Trondheim, Norway
Agile control of CO2 capture technology for maximum net operating revenue
Norhuda Abdul Manaf, Abdul Qadir, Ali Abbas
*School of Chemical and Biomolecular Engineering, The University of Sydney, Sydney 2006, Australia
([email protected] ; [email protected] ; [email protected]).
Abstract: This paper uses a multi-layered control scheme consisting of model predictive control (MPC)
and mixed integer non-linear programming (MINLP) for the analysis of power plant net operating
revenue when retrofitted with a post-combustion carbon capture (PCC) plant. The capability of the
proposed control scheme is examined for 24-hour operations of an integrated plant (power plant with
PCC) in the years 2011 and 2020. The control scheme is tested in response to variability in upstream
power plant dynamic loads. The agility of the control scheme subject to forecast 2020 electricity and
carbon prices is shown to result in yearly net operating revenue of approximately 12% of the gross
revenue. Whereas, the same integrated plant generates a net operating revenue loss of roughly 13% of the
gross revenue under 2011 electricity and carbon prices.
Keywords: Algorithm, Automatic control (close-loop engineering), Optimization devices, Plant wide,
Multi-input multi-output system, Nonlinear system.
Where is the price of electricity and is the carbon price
(CO2 allowance). The net operating revenue composite
as the
consists of three individual costs which include
as the PCC operational
power plant operational cost,,
cost and finally, the cost of CO2 emission (indicated in the
second integration term). The first integration term in the
above equation represents the revenue generated through
selling of electricity. Here, we optimize the operational cost
of integrated plant where it is then used to determine the net
operating revenue of the system. Fig. 1 illustrates the
operation of MPC scheme (bottom layer).
1. INTRODUCTION
Low emissions fossil fuel technologies such as post
combustion CO2 capture (PCC) are expected to play a
significant role in combating climate change. Recently, a
‘carbon capture ready’ power plant emerged as a noteworthy
solution towards curbing global warming (Markusson and
Haszeldine 2010). However, profitability of operation needs
to be understood and costs reduced for this type of plant to
come to commercially feasible scale and operation.
This study uses a multi-layered control scheme consisting of
model predictive control (MPC) and mixed integer non-linear
programming (MINLP) for the analysis of 660 MW coalfired power plant net operating revenue when retrofitted with
amine-based PCC plant. Where, the power plant provides
part of its power to PCC plant operation via steam turbine
bleed. The higher layer (MINLP) computes optimal CO2
capture rate, CC and power plant net load at given electricity
and carbon prices with the ultimate objective to obtain
maximum net operating revenue (ideal revenue). While, the
bottom layer (MPC embedded with PCC plant) regulates lean
solvent flowrate and reboiler heat duty to ensure the PCC
plant operates in a way to achieve identical net operating
revenue as the ideal net operating revenue. This paper
demonstrates the agility and robustness of the control scheme
in maximising revenue of operation while matching the load
and considering carbon price input.
Fig. 1. MPC architecture (u1: flue gas flow rate, u2: CO2
concentration in flue gas, u3: lean solvent flow rate, u7:
reboiler heat duty, CC: CO2 capture rate, EP: energy
performance).
2. DEVELOPMENT OF A MULTI-LAYERED CONTROL
SCHEME
The innovative features of this work are that it offers a
temporal multiscalar multi-layered control scheme critical for
top-down management decision making and operation of
coal-fired power plant integrated with PCC. Where, the top
and bottom layers consider different interval times; 30
minutes and 10 seconds respectively. This big transmission
of interval time imposes a challenge to MPC scheme to
respond proactively to internal and external upsets stemming
from the integrated plant. A comprehensive explanation
The objective function at the high layer that maximizes net
operating revenue is described in (1).
∗
Copyright © 2016 IFAC
∗
∗
∗
(1)
332
IFAC DYCOPS-CAB, 2016
June 6-8, 2016. NTNU, Trondheim, Norway
faster/ahead response than the reboiler heat duty. The benefit
of employing MPC scheme and the agility performance in
terms of plant net operating revenue is explained in the next
sub-section.
pertaining to this multi-layered control scheme can be found
in (Abdul Manaf, Qadir et al. 2016).
3. CASE STUDY
3.1 Annual flexible operation mode
4.2 Net operating revenue – MPC performance
The agility of the proposed control scheme is tested in
response to variability in upstream power plant dynamic
loads based on annual flexible operation mode of the
integrated plant. The case study was demonstrated based on
2011(-$147/ MWh - $ 12,136/MWh) and 2020 ($3/MWh $17,959/MWh) electricity prices. For the year 2020, the
projected electricity prices were estimated by assuming a 5%
yearly increment from the base year 2008 (Australian Energy
Market Operator (AEMO)). Carbon price for year 2011 was
adapted and extracted based on European Union Allowance
(EUA) historical data available in Intercontinental Exchange
Futures Data. For the year 2020, a hypothetical carbon price
trend was estimated by assuming a 5% yearly increment and
3% inflation rate induced yearly from the base year 2008.
This estimation is based on the core policy scenario proposed
by the Treasury Department, Australia (The Treasury).
Agility performance of the MPC scheme was evaluated based
on the deviation in ideal and actual net operating revenue
(controller error). From (1), the total plant net operating
revenue was segregated into four individual terms as given in
(2).
(2)
Where A represents the plant revenue generated through
selling of electricity, B is cost of CO2 emission (carbon price
paid), C and D are the power plant and PCC operational costs
respectively.
Fixed, ideal and actual gross revenue of operation and the
resultant net operating revenue of the system for year 2011
and 2020 are presented in Table 1. Fixed operation mode
(base condition) was performed by implementing 90%
capture rate throughout one-year operation. This was
implemented by constraining the lower and upper bounds of
CO2 capture rate at 90% while maintaining the objective
function (maximize net operating revenue) at corresponding
power plant loads. It can be observed, for flexible mode, the
system subject to 2020 electricity and carbon prices
generated annual net operating revenue of approximately
12% of the gross revenue. While for 2011, the system
incurred a net operating revenue loss roughly 13% of the
gross revenue. This negative net operating revenue occurred
possibly because of the lower bounds set for the power plant
output (0 MW) and CO2 capture rate (25 %). For instance,
during times of very low electricity prices (possibly even
negative), the cost of operation of the power plant and PCC
plant would exceed the revenue generated from selling the
electricity. As expected, net operating revenue forecasted
from fixed operation mode (year 2011 and 2020) is much
lower compared to that in flexible operation mode. This
outcome illustrates that the application of flexible operation
mode enhances plant net operating revenue and provides
significant cost saving.
4. RESULT AND DISCUSSION
4.1 Annual flexible operation mode
Figs. 2-3 show the outputs from the multi-layered control
scheme for year 2011 and 2020 respectively. For year 2011,
it can be observed that the power plant operated in load
following mode for a high proportion throughout the year.
This reflects that during load following operation, power
plant combined with PCC generated low cost corresponding
to its prevailing electricity and carbon prices. Short-term
maintenance (shutdown) plans were observed during month
of May, June, July and December. For year 2020, the power
plant is operated in recurrent maximum load with intermittent
narrow unit turndown and load following. This result is
particularly relevant to power plant operation since energy
systems do not all operate in the same way (Mac Dowell and
Shah 2015). Moreover, these mixed operation modes of
power plant in line with scheduled shutdowns are actually
assisting to reduce the running cost of both power plant and
PCC. Both these dataset (power plant load profile and CO2
capture rate) obtained in this study are beneficial for future
insight if ‘carbon capture ready’ power plants are to become
a reality. Where, these analyses/data may serve as a baseline
or reference information to the contractor, engineer and plant
operator during design, commissioning and troubleshooting
stages.
Capture energy penalty and other main performance
parameters for the power plant associated with PCC are listed
in Table 1. The fixed operation mode has a capture penalty of
9%, being 0.1% higher than the flexible operation mode
(actual) for year 2011 and 2020 respectively. Where, those
energy penalties fell within the range of the state of the art
amine-based technology (3.0 - 4.5 MJ/kg CO2). Besides that,
via MPC scheme, PCC plant (in flexible operation mode) was
capable of attaining average of 84% from the ideal
cumulative CO2 capture in both years, concurrently obtaining
high net operating revenue compared to the fixed operation
mode. MPC scheme exhibited superior control performance
by minimizing the controller error to an average of 4% for
both years. This is illustrated based on the deviation between
ideal and actual net operating revenue from which the agility
At the bottom layer, for both years, the MPC scheme has
shown satisfactory control performance in tracking the ideal
CO2 capture rate (CCideal). It can be observed that the lean
solvent flow rate, u3 and reboiler heat duty, u7 were
compensating each other in responses to set point changes of
CO2 capture rate. The responses showed that the lean solvent
flow rate is relatively more sensitive compared to the reboiler
heat duty in its reaction to the fluctuation of CO2 capture set
point (CCideal). In other words, lean solvent flow rate gives a
333
IFAC DYCOPS-CAB, 2016
June 6-8, 2016. NTNU, Trondheim, Norway
(a)
and
d robustness oof proposed multi-layered
m
control schem
me is
eviident. Thesee combinedd results reflect
r
that the
imp
plementation of advanceed control for
f
flexible PCC
opeeration associaated with coall-fired power plant under caarbon
pricing/trading sschemes can reduce the net
n cost of caapture
sysstems. Lowerr capture eneergy penalty gained from
m the
flex
xible mode pprovides econnomic value to the power plant
com
mpany to conssider a penetraation of PCC technology.
(b)
4. CONC
CULSION
Results presentted in this paper
p
show that it woulld be
ben
neficial to im
mplement the proposed
p
conttrol scheme w
which
cou
uld allow for a PCC plant retrofitted to a coal-fired ppower
plaant to operatte flexibly and
a
profitably
y under scennarios
sim
milar to the onnes presentedd in this paper (i.e. for thee year
202
20). Howeverr, under histooric 2011 elecctricity pricess, the
opeeration costs oof PCC plant retrofitted to a coal power plant
excceed the grosss revenue of thhe power systtem. This indiicates
thaat the implemeentation of PC
CC plant is un
neconomical aat the
preevailing electrricity and carbbon price market conditionss (i.e.
forr the year 20111). These results highligh
ht the value oof the
preesented schem
me in informinng cost assessm
ments and anaalyses
forr capture readdy power plannts, and in low
wering the riskk and
red
ducing the uncertainty associated with investtment
deccisions. Finaally, cautionn is warran
nted in sccheme
com
mputations annd interpretaations becausse the resultss are
hig
ghly dependennt on the assumptions th
hat come withh the
sceenarios (eg. prrice and load forecasts).
f
(c)
(d)
Fig. 3 Control-opttimization respponses from flexible
f
operattion
mod
de for year 202
20.
Tablle 1 Comparison of net operratining reven
nue, its individdual
cost and performaance of power plant with PC
CC systems foor
20.
year 2011 and 202
2011
Revenue composite
c
2020
Fixeed
Fixed operation
Flexible
Flexible
Flexible opeeration
operattion
mode (at CC =
operation
operation
mode (ideal)
mode (att CC =
90%)
mode (ideal)
mode (actual)
90%
%)
Flexxible
operationn mode
(actuual)
Net operatting revenue ($ Million/year)
-30.6
-18.4
-17.8
47.88
58.3
55..9
Gross revenue generated through selling of
electricity ($ Million/year)
162.0
144.7
144.7
436..1
460.5
4600.5
Cost of CO2 emission ($ Million/year)
3.1
14.0
17.9
18.44
40.3
49..2
Power plaan operational cost
($ Million//year)
69.1
66.1
66.1
155..5
164.2
1644.2
PCC operrational cost ($ Million/year)
120.00
83
78.5
214..3
197.7
1911.3
Power plaant cumulative auxiliary energy
consumpttion (MWe)
243,830
241,060
241,060
393,6670
331,370
331,3370
PCC cumu
mulative energy consumption
(MWth )
2,106,200
854,800
673,710
7,659,300
4,530,400
4,2688,300
Cumulativve CO2 captured (tonne)
1,804,100
867,980
634,670
8,795,200
5,222,200
4,9066,200
4.20
3.55
3.82
3.144
3.12
3.113
Capture sppecific energy penalty
(MJth/kgC
CO2)
REFERE
ENCES
"Ausstralian
En
nergy
Marrket
Operaator
(AEM
MO).
<http
p://www.aemo
o.com.au> [acccess 10.06.14
4]."
"Inteercontinental
Exchannge
Futtures
D
Data.
<http
ps://www.quaandl.com/data//ICE/CZ2014-ECX-EUA>..
[acceess 05.04.15]."
"Thee Treasury." Strong
S
Growthh, Low Polluttion: Modellinng a
Carb
bon Price. 30 Februarry 2015. Retrieved
R
fr
from
http://carbonpricem
modelling.treaasury.gov.au
Fig
g. 2 Control-optimization reesponses from
m flexible operration
mo
ode for year 20011.
334
IFAC DYCOPS-CAB, 2016
June 6-8, 2016. NTNU, Trondheim, Norway
Abdul Manaf, N., A. Qadir and A. Abbas (2016). "Temporal
multiscalar decision support framework for flexible operation
of carbon capture plants targeting low-carbon management of
power plant emissions (Unpublished)." Journal of Applied
Energy, 169, 912-926.
Mac Dowell, N. and N. Shah (2015). "The multi-period
optimisation of an amine-based CO2 capture process
integrated with a super-critical coal-fired power station for
flexible operation." Computers & Chemical Engineering
74(0): 169-183.
Markusson, N. and S. Haszeldine (2010). "‘Capture ready’
regulation of fossil fuel power plants – Betting the UK’s
carbon emissions on promises of future technology." Energy
Policy 38(11): 6695-6702.
335