Aou

Oa

b

c

a

ARRAA

KBCBIMM

1

s(pDiastC2ipv

lT

h1

International Journal of Greenhouse Gas Control 28 (2014) 189–202

Contents lists available at ScienceDirect

International Journal of Greenhouse Gas Control

j ourna l h o mepage: www.elsev ier .com/ locate / i jggc

mixed integer nonlinear programming (MINLP) supply chainptimisation framework for carbon negative electricity generationsing biomass to energy with CCS (BECCS) in the UK

. Akgula, N. Mac Dowellb,c,∗, L.G. Papageorgioua, N. Shahb

Centre for Process Systems Engineering, Department of Chemical Engineering, University College London, London WC1E 7JE, UKCentre for Process Systems Engineering, Department of Chemical Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ, UKCentre for Environmental Policy, Imperial College London, South Kensington Campus, London SW7 2AZ, UK

r t i c l e i n f o

rticle history:eceived 25 June 2013eceived in revised form 9 January 2014ccepted 11 June 2014vailable online 17 July 2014

eywords:ECCSO2 capture

a b s t r a c t

The co-firing of biomass and fossil fuels in conjunction with CO2 capture and storage (CCS) has thepotential to lead to the generation of relatively inexpensive carbon negative electricity. In this work, weuse a mixed integer nonlinear programming (MINLP) model of carbon negative energy generation in theUK to examine the potential for existing power generation assets to act as a carbon sink as opposed to acarbon source. Via a Pareto front analysis, we examine the technical and economic compromises implicitin transitioning from a dedicated fossil fuel only to a carbon negative electricity generation network. Aprice of approximately £30–50/t CO2 appears sufficient to incentivise a reduction of carbon intensity ofelectricity from a base case of 800 kg/MWh to less than 100 kg/MWh. However, the price required to

iomass co-firingECM

ixed integer optimisationulti-scale modelling

incentivise the generation of carbon negative electricity is in the region of £120–175/t of CO2. In orderfor biomass to energy with CCS (BECCS) to be commercially attractive, the power plants in question mustoperate at a high load factor and high rates of CO2 capture. The relative fuel cost is a key determinant ofrequired carbon price. Increasing biomass availability also reduces the cost of generating carbon negativeelectricity; however one must be cognisant of land use change implications.

. Introduction

It is well accepted that any meaningful reduction in CO2 emis-ions will depend upon CO2 capture and storage (CCS) technologiesBoot-Handford et al., 2014; Mac Dowell et al., 2010a) with theossible addition of actually removing CO2 from the atmosphere.irect air capture technologies have been proposed several times

n the past (Goeppert et al., 2011; Lackner, 2009; Mahmoudkhanind Keith, 2009) with some quite astounding claims being made inome cases (Still, 2012). In this work we are proposing the combina-ion of the co-firing of biomass with fossil fuels in conjunction withCS; so-called Bio-Energy with CCS (BECCS) (Gough and Upham,011; Fuss, 2012) as a means of achieving indirect air capture. This

s a promising approach which has the potential to transform theower generation industry from a carbon source into a carbon sinkia the generation of carbon negative electricity.

∗ Corresponding author at: Centre for Process Systems Engineering, Imperial Col-ege London, South Kensington Campus, London SW7 2AZ, UK.el.: +44 020 7594 9298.

E-mail address: nmac-dow@ic.ac.uk (N. Mac Dowell).

ttp://dx.doi.org/10.1016/j.ijggc.2014.06.017750-5836/© 2014 Elsevier Ltd. All rights reserved.

© 2014 Elsevier Ltd. All rights reserved.

An attractive aspect of BECCS is that existing coal-fired powerstations can be readily converted to co-firing fossil fuels, and ifretro-fitted with CCS, provide a relatively low-cost path to carbonnegative energy generation (UK Dept of Energy and Climate Change,2012). This is particularly relevant in the context of the UK given theexpectation that a significant fraction of existing coal-fired powerplants are to be retired within the next decade (DECC, 2011). Fur-thermore, there is some evidence that BECCS technologies can helpreduce or eliminate the “not-in-my-backyard” (NIMBY) effect thathas traditionally been an important public acceptance barrier toCCS (Wallquist et al., 2012).

However, carbon negative energy generated via BECCS is likelyto be even more expensive than low-carbon energy generatedvia conventional CCS. Thus, in the absence of an internationalagreement to limit CO2 emissions (UN Earth Summit, 1992; UNConference on Sustainable Development, 2012), we would suggestthat a fixed price on carbon emissions is another way to incen-tivise low carbon power generation. This suggestion is in line with

the suggestion of a carbon price floor within the UK EMR exercise(DECC, 2011).

It is the main purpose of this contribution to provide someinsight as to what the costs associated with CO2 emission would

1 Green

htm

st2ShgTrit2

i2ltasrpaHtsaettetmtrhfecL

aAtwPt(cMppw

tctwqpaoie

by: ∑∑

90 O. Akgul et al. / International Journal of

ave to be in comparison to fuel prices, in order to incentivisehe generation of carbon negative energy, and establish a backstop

arginal abatement cost.The majority of the literature that considers biomass-to-energy

upply chains primarily focuses on the production and distribu-ion of biofuels for ground transport (Avami, 2012; Cucek et al.,012; Giarola et al., 2012; You et al., 2012; Akgul et al., 2012).ome previous work on biomass-based supply chains focuses onybrid systems where different production technologies are inte-rated (Elia et al., 2012; Bowling et al., 2011; Aguilar et al., 2011;ittmann et al., 2010). To the best of our knowledge, literature workelated to biomass-to-heat or biomass-to-electricity supply chainss limited compared to those focussing on biomass-to-biofuel sys-ems (Fortes et al., 2012; Gan and Smith, 2011; Rentizelas et al.,009; Dunnett et al., 2007).

In the context of BECCS, one very important point to considers the source of the biomass (McKechnie et al., 2011; Schulze et al.,012). If the co-firing of biomass becomes an important part of

arge-scale power generation, it would be vital to ensure the sus-ainability of the biomass supply chain. Co-firing has been proposeds a relatively cost-effective approach to mitigate CO2 emissions, ithould be highlighted that the degree to which biomass co-firingeduces the net GHG emissions depends on the methods used toroduce the biomass (Mosier et al., 2005). Failure to adequatelyccount for these emissions can lead to carbon leakage problems.owever, it has been suggested that from a whole-system perspec-

ive GHG emissions associated with co-firing are reduced at a ratelightly higher than the ratio of the biomass co-firing ratio (Mannnd Spath, 2001). Further, it is important to consider the additionalnergy penalty associated with co-firing arising from the poten-ially higher moisture content of the fuel. It has been reported thathe estimated heat rate degradation due to co-firing is 0.5% forvery 10% input of pellets (Zhang et al., 2010). Thus, this can leado the necessity to ensure that the biomass pellets have very low

oisture content–potentially raising their cost. Similarly, owing tohe reduced energy density, significant quantities of pellets will beequired, raising supply chain constraint questions. This obviouslyighlights the value in considering the exploitation of marginal land

or the cultivation of bioenergy crops, although as the value of thosenergy crops increases, the economic incentive to produce energyrops from more productive land would be strong (Bryngelsson andindgren, 2013).

It is worth pointing out that the combustion chemistry associ-ted with co-firing of fossil and biomass fuels is far from simple.n important area of inquiry has been the production of poten-

ially toxic emissions from the co-combustion of biomass andaste derived fuels with fossil fuels (European Union Joule III

roject, 2001). Here an important aspect to consider was the fate ofhe various trace element emissions arising from these processesMiller et al., 2002a, 2002b, 2003, 2004). An important result ofo-combustion processes is the potential loss of trace elements.ercury and cadmium in particular have been shown to cause

articular concern, although the addition of either pulp sludge orlastic waste is beneficial for improving mercury retention some-hat (Miller et al., 2006).

Distinct from conventional electricity generation, carbon nega-ive electricity generation via BECCS is subject to important supplyhain constraints associated with the availability of sufficient sus-ainably sourced biomass and adequate processing facilities withhich to convert the raw material into a fuel grade product. The

uestion we wish to address involves choosing between a set ofower stations and deciding what load factor, degree of CO2 capture

nd co-firing is most appropriate, subject to a constrained supplyf indigenous biomass. Thus, this question can be posed as a mixednteger nonlinear programming (MINLP) problem. Importantly, wexplicitly account for the costs associated with biomass co-firing

house Gas Control 28 (2014) 189–202

and CO2 capture as well as the supply chain constraints associatedwith the actual production and distribution of the biomass fuel.An important advantage of this approach is its capacity to provideinsight into the spatial and temporal effects of decision making, i.e.,what happens where and when based on a set of inputs.

The remainder of this paper is organised as follows: The problemstatement is presented in Section 2, and the mathematical formula-tion is described in detail in Section 3. Some case studies performedwith this model are presented and discussed in Section 4 with someconclusions drawn in Section 5.

2. Problem statement

The problem addressed in this work can be stated as follows:Given the:

• geographical locations and capacities of current and potentialfuture electricity generation plants,

• total electricity demand,• different raw material types for pellet production and their geo-

graphical availabilities,• unit raw material supply, pellet production, and fossil fuel costs,• transport logistics characteristics (costs, modes, and availabili-

ties),• capital investment cost for the pellet production facilities, deter-

mine the optimal:• raw material supply, pellet production and electricity generation

rates,• locations and scales of the pellet production facilities,• flows of raw material and pellets between cells,• modes of transport of delivery for raw material and pellets;• fuel burn rates, capacity factors, generation efficiencies, degree

of CO2 capture and co-firing of pellets at each generation plant,so as to minimise the total annual cost (TAC) or the total annualemissions (TAE) of the supply chain.

The supply chain model introduced in this paper adopts a“neighbourhood” flow approach with 8 N configuration. Moredetailed explanation of the neighbourhood flow approach can befound in the work of Akgul et al. (2011). The main advantage of thisapproach comes from an increased computational efficiency whencompared to other approaches (Akgul et al., 2011).

3. Mathematical formulation

The problem for the optimal design of a bioelectricity supplychain is formulated as a spatially-explicit, static, multi-objective,mixed integer nonlinear programming (MINLP) model. In the fol-lowing sections, we describe the objective function, the variousconstraints to which our model is subject, and finally the modelsdeveloped to describe the behaviour of the carbon negative powerplant. In the interest of brevity, all notation and parameter valuesand ranges are presented in the appendices.

3.1. Objective function

The economic objective of the proposed model is the minimi-sation of the total annual supply chain cost (TAC) which is given

TAC =p∈P g∈G

CRF ICpEpg

Total pellet production plant capital cost (1a)

Green

∑

+

+

+

+

+

i

+

t

T

3

ppf

P

r

D

3

ttr

P

i

O. Akgul et al. / International Journal of

i∈R

∑g∈G

USCigPig Total raw material supply cost (1b)

∑i∈R

∑g∈G

∑p∈P

UPCip�i Dfipg Total pellet production cost (1c)

∑g∈G

UGCgPelec,g Total power generation cost (1d)

∑g∈G

UFCg mgϕfossil,g Total fossil fuel cost (1e)

∑g∈G

UCARC Pelec,gCIg Total carbon cost (1f)

∑i,g∗,g,l∈nig′gl

IMPCig∗ Qig∗gl Total product outsourcing cost (1g)

∑(/∈FP),g,g′,l∈nigg′ l

UTCilADDgg′lQigg′l

Total material transportation cost between cells (1h)

∑i∈R

∑g∈G

UTC∗ ALDgPig

Total local raw material transportation cost (1i)

The environmental objective of the proposed model considershe minimisation of the total annual emissions, TAE given by:

AE =∑g∈G

Pelec,g CIg (2)

.2. Demand constraints

The amount of raw material i consumed at a pellet productionlant of scale p located in region g, Dfipg, is related to the total pelletroduction rate at that plant, Pfi′pg, by the conversion factor, � i asollows:

fi′pg =∑i∈R

�iDfipg ∀i′ ∈ FI, p ∈ P, g ∈ G (3)

As a result, the total consumption rate of raw material type i inegion g, Dig is given by:

ig =∑p∈P

Dfipg ∀i ∈ R, g ∈ G (4)

.3. Production constraints

The material balance for each material i and region g states thathe production of a material in region g plus the incoming flows ofhat material to that region must be equal to the demand in thategion plus the outgoing flows from that region.

ig +∑l∈L

∑g′∈nig′gl

Qig′gl = Dig +∑l∈L

∑g′∈nigg′ l

Qigg′l ∀i ∈ I, g ∈ G (5)

The total pellet production rate of pellet type i in a region g, Pig

s given by:

Pig =∑p∈P

Pfipg ∀i ∈ FI, g ∈ G (6)

house Gas Control 28 (2014) 189–202 191

The total pellet production rate at a plant in region g is limitedby the minimum and maximum production capacities:

PCapminp Epg ≤

∑i∈FI

Pfipg ≤ PCapmaxp Epg ∀p ∈ P, g ∈ G (7)

where PCappmin and PCapp

max are the minimum and maximumplant capacities for a pellet production plant of size p, respectively.Epg is the binary variable that represents the establishment of apellet production plant of size p in region g.

Similarly to the pellet production rate, the local raw materialsupply rate is also limited by the minimum and maximum localavailabilities as follows:

BAminig ≤ Pig ≤ BAmax

ig ∀i ∈ R, g ∈ G (8)

where BAigmin and BAig

max are the minimum and maximum avail-abilities of raw material type i in region g.

The electricity generation rate at a power plant located in regiong, Pelec,g is related to the fuel mix burn rate, mg , the energy densityof the fuel mix, �g and the generation efficiency, �g as follows:

Pelec,g = mg�g�g

3600∀g ∈ G (9)

The total electricity production must meet the total electricitydemand, DEM:∑g∈G

Pelec,g = DEM (10)

The energy density of the energy density of the fuel mix, �g isrelated to the co-firing rates of the different fuel types, ϕig and theirenergy densities, �i as follows:

�g =∑i∈F

ϕig�i ∀g ∈ G (11)

The sum of the co-firing rates of all fuel types must be equal to1:∑i∈F

ϕig = 1 ∀g ∈ G (12)

The consumption rate of each fuel type i, Dig in the fuel mix isrelated to the fuel mix burn rate and the co-firing of that fuel asfollows:

Dig = ϕigmg ∀i ∈ F, g ∈ G (13)

The energy balance for a power plant located in region g is writ-ten as:

Pelec,g = ıgPPCg ∀g ∈ G (14)

where ıg is the variable that represents the capacity factor for apower plant located in region g and PPCg is the parameter thatrepresents the generation capacity of that plant.

The carbon intensity, CIg from a power plant located in regiong, is a linear function of the plant capacity, PPCg, the plant capacityfactor, ıg, the extent of carbon capture and storage, CCSg and theco-firing extent of the non-fossil fuels, ϕig as follows (see Section3.5):

CIg = CI + aCI

(PPCg − PPC

)+ bCI

(ıg − ı

)+ cCI

(CCSg − CCS

)+ dCI

(∑i∈FI

ϕig − ϕ

)∀g ∈ G (15)

where CI, PPC, ı, CCS and ϕ are the parameters that represent thereference values for carbon intensity, plant capacity, capacity fac-tor, extent of carbon capture and storage and the co-firing extentin the base case, respectively.

192 O. Akgul et al. / International Journal of Greenhouse Gas Control 28 (2014) 189–202

F o develop a set of inputs and outputs. Subsequently the meta-model is proposed andp

pccf

�

w

ro

U

w

3

c

Q

wb

3

wamcps

dt22m

Table 1Composition and energy density of the model coal and biomass used in this study.

Parameter Britishbituminouscoal

Biomass

GCV (MJ kg−1, as received) 24.6 18.7Moisture 12.0 7.0C 59.6 43.5H 3.8 4.5N 1.5 0.2O 5.5 42.6S 1.8 0.01Cl 0.2 0.01

TB

ig. 1. Meta modelling development process. We use a detailed modelling tool tarameters are adjusted to relate the inputs and outputs.

Similar to the carbon intensity, the generation efficiency of aower plant located in region g, �g is a linear function of the plantapacity, PPCg, the plant capacity factor, ıg, the extent of carbonapture, CCSg and the co-firing extent of the non-fossil fuels, ϕig asollows (see Section 3.5):

g = � + a�

(PPCg − PPC

)+ b�

(ıg − ı

)+ c�

(CCSg − CCS

)+ d�

(∑i∈FI

ϕig − ϕ

)∀g ∈ G (16)

here � is the reference generation efficiency in the base case.Finally, the unit generation cost in a power plant located in

egion g, UGCg is related to the plant capacity, PPCg, and extentf carbon capture and storage, CCSg as follows:

GCg = UGC + aUGC

(PPCg − PPC

)+ bUGC

(ıg − ı

)+ cUGC

(CCSg − CCS

)+ dUGC

(∑i∈FI

ϕig − ϕ

)∀g ∈ G (17)

here UGC is the reference unit generation cost in the base case.

.4. Transportation constraints

An upper limit on the flow of material i between regions can beonsidered such that:

igg′l ≤ Q maxil ∀i, g, g′, l ∈ nigg′l (18)

here Qilmax is the maximum flowrate of material i via mode l

etween regions g and g′.

.5. Power plant model

It is clear that an important aspect of the above described modelill be the model used to quantify the capital and operating costs

ssociated with the power plants. As introduced in Section 3.3, theodel can evaluate capital and operating costs, generation effi-

iency and carbon intensity of a power plant based on its namelate capacity, capacity factor, extent of co-firing of biomass andolid recovered fuels (SRF) and extent of carbon capture.

It would be preferable to use a detailed modelling approach inescribing the performance of a decarbonised power plant, such as

hose presented in (Mac Dowell and Shah, 2013; Mac Dowell et al.,010b, 2013; Arce et al., 2012; Biliyok et al., 2012; Lawal et al.,009, 2010). However, while we have previously employed suchodels in an input-output fashion (Mac Dowell et al., 2011), we

able 2ase input and output vectors obtained from IECM (IECM, 2012).

Base input vector, xn

Reference nameplate capacity,PPC (MW) 500

Reference capacity factor, ı (%) 100

Reference co-firing extent, ϕ (%) 22

Reference CO2 capture extent, CCS (%) 90

We note that the Orchid SRF product was given to be 80% biomass, 16% moistureand the remainder considered to be “ash”. The energy density of the SRF productwas given to be 12.5 MJ/kg.

cannot use them in a network optimisation context. Thus we adopta meta-modelling approach as illustrated in Fig. 1.

The meta-model is of the form

ym = ym(xn) + Amn (xn − xn) (19)

where the output vector ym is related to an input vector xn througha coefficient matrix Amn in a piecewise linear fashion by differencefrom a base input vector xn and a base output vector ym = f (xn).

In this work, our detailed modelling tool of choice was the Inte-grated Environmental Control Model (IECM) (IECM, 2012) as theeconomic and technical models which it contains are considered tobe well validated, and is thus a reliable tool. An important advan-tage of the IECM tool is the facility for the user to define a fuelcomposition, in addition to the default fuel compositions. In spec-ifying the composition and energy density of a fuel blend in thiswork, it has been assumed that the coal used was British bitumi-nous coal and that the available biomass source was wood pellets.The composition and energy density of the SRF was taken fromthe Renewable Power Fuel product from Orchid Environmental, aUK-based SRF producer (Orchid, 2014). The fuel compositions andenergy densities are provided in Table 1. Then the composition andenergy density of the co-fired fuel blend was obtained from a massaverage.

In this work, a 500 MW plant, operating at 100% capacity, co-firing 22 wt% biomass and 90% CO2 capture has been chosen as thebase case for the meta-model. The base input and output vectorsare presented in Table 2, the coefficient matrix Aij is given in Table 3and finally the operating range over which the proposed equations

are valid is given in Table 4.

Therefore, if one wishes to calculate the generation efficiencyof a 500 MW plant operating at 90% capacity, with 10% co-firingand 85% CO2 capture the corresponding equation is (Eq. (16) in the

Base output vector, ym

Reference capital Cost (k£/MW) 2079Reference unit generation cost, UGC (£/MWh) 16.57Reference generation efficiency, � (%) 34.5Reference carbon intensity, CI (kg CO2/MWh) −97

O. Akgul et al. / International Journal of Greenhouse Gas Control 28 (2014) 189–202 193

Table 3Parameter values for coefficient matrix, Amn .

Nameplate capacity, a (MW) Capacity factor, b (%) CO2 capture extent, c (%) Co-firing extent, d (%)

Capital cost (k£/MW) −1.66E+00 0.00E+00 4.71E+02 1.60E+020.001.682.88

m

�

cs

wpocp

cmat

4

tet(tte9priapICbcobtpcf

TO

Unit generation cost, UGCg (£/MWh) −2.70E−03Generation efficiency, �g (%) 6.00E−04

Carbon intensity, CIg (kg CO2/MWh) −1.00E−03

athematical formulation section):

g = 34.5 +(

6 × 10−4 (500 − 500))

+ (1.6758 × (90% − 100%))

+(−12.833 × 10−1 (85% − 90%)

)+ (−7.7 × (10% − 22%)) ≈ 35.9%

It can be observed that this corresponds to a slightly higher effi-iency in comparison to the base case, as might be expected with acenario corresponding to lower rates of co-firing and CO2 capture.

It must be noted that in implementing the meta-model in GAMS,e assumed that all power plants under consideration were com-osed of a number of 500 MW units, i.e., if the nameplate capacityf a given power plant is 2000 MW, this installation is supposed toomprise 4 × 500 MW units as is the case for the Ratcliffe-on-Soarower station operated by E.ON UK (2010).

We have tested the accuracy of the proposed meta models byomparing its outputs with those from the fully detailed IECModel. The average absolute relative deviation between the meta-

nd IECM models was 8.84%. The meta-model was therefore judgedo be sufficient for inclusion in our system-scale model.

. Results and discussion

We have applied our model to the problem of carbon nega-ive electricity in the UK. In this work, we use a sample of 10xisting generation assets which together provide a total genera-ion capacity of 19 GW. The total consumption from all consumersdomestic, commercial and industrial) in the UK has been reportedo be 308,034 GWh in 2011. Based on this total consumption andhe renewables target which requires UK to supply 30% of itslectricity from renewables by 2020, this corresponds to about2,410 GWh of renewable electricity generation. Our aim was torovide some insight into the costs associated with producing thisenewable generation of carbon-negative energy from the exist-ng capacity (DECC, 2012a, 2012b). It is worth noting that we havessumed operation at the same capacity factor for all the powerlants throughout a year (8000 hours of operating time per year).

n this case study, we consider power generation systems whereO2 capture and storage systems are combined with co-firing ofiomass with fossil fuels. We assume that the existing generationapacity is pre-installed with these systems, so we only considerperating costs associated with them but no capital costs. As cane seen in the mathematical formulation, the costs considered by

he economic objective are total raw material supply cost, totalellet production and pellet plant capital costs, total transportationost (of raw material), total power generation cost, total carbon andossil fuel costs associated with power generation and finally, total

able 4perating range of the meta model.

Input parameter Lower bound Upper bound

Nameplate capacity (MW) 300 1000Capacity factor (%) 60 100Co-firing extent (%) 0 50CO2 capture extent (%) 50 98

E+00 8.56E+00 0.00E+00E+00 −1.28E+01 −7.70E+00E+01 −7.73E+02 −1.01E+03

product outsourcing cost. Capital costs associated with retrofitting(for CO2 capture or biomass co-firing) or capital and operating costsof CO2 transportation are not considered. Here, we consider thatthe biomass and coal are co-milled in order to reduce the CAPEXassociated with converting the power plants to co-fire biomass andcoal.

In order to minimise carbon leakage, we use indigenous biomassonly. In order to maximise the potential for biomass co-firing,we use solid recovered fuels from municipal solid waste to aug-ment the biomass available from more traditional sources. Inthis case study, the UK is discretised into 34 square cells ofeach 108 km in length. The nameplate capacity, total annual CO2emissions and geographical location of each of the 10 coal-firedpower plants under consideration can be found in Table D1 inAppendix D.

The remaining inputs to the model are the costs associated withfuel (coal, biomass and SRF) and CO2 emissions, respectively. Wehave chosen three years; 2012, 2020 and 2050. For each of theseperiods, we have three decarbonisation scenarios correspondingto DECC’s low, central and high scenarios for carbon (DECC, 2013)and coal (DECC, 2012b) prices, respectively. We consider the lowprice scenario to be a pessimistic decarbonisation scenario owingto the low economic driving forces to implement either fuel switch-ing or CCS. Similarly, a scenario with both high CO2 and coalprices is considered an optimistic decarbonisation owing to theclear incentive to implement both CCS and fuel switching. Thecosts of both CO2 and coal for each scenario are presented inTable 5.

All the other related data related to raw materials, pellet pro-duction, transportation and power generation are given in theappendices.

We observe from Table 5 that in the low and central scenarios,coal prices appear to be significantly reducing in the period to 2030and beyond. We completely acknowledge that this is a matter ofsome debate, and there are good reasons to believe that this pricetrend may not occur. However, an exhaustive consideration of allthe possible pricing scenarios is beyond the scope of this work.

4.1. Pareto curve analysis

In this section, we present the results for each of the scenar-ios as a series of Pareto curves (Figs. 2–4). Two main conclusionscan be drawn from the comparison of these three figures. Firstly,as we move from the pessimistic decarbonisation scenario (Fig. 2)towards the optimistic one (Fig. 4), the carbon intensity of thesystem decreases at the minimum cost point. Secondly, it can beconcluded that increased carbon prices and increased availabilityof biomass (pellets and MSW) are the main drivers for reducing thecarbon intensity associated with power generation; this is espe-cially evident from the Pareto curves for the 2050 scenarios in eachof Figs. 2–4.

It is interesting to consider how technology selection changesas we move along the Pareto curve from a least cost objective to aleast carbon objective. As can be seen in Fig. 2, the trend in the opti-

mal selection of technologies as we proceed along the Pareto curvefor 2012 from the minimum cost to the minimum carbon intensitypoint is as follows: coal only, coal + CCS, co-firing of biomass pel-lets with coal + CCS and co-firing of biomass and SRF pellets with

194 O. Akgul et al. / International Journal of Greenhouse Gas Control 28 (2014) 189–202

Table 5Values for carbon and coal used in this work, we note that the DECC coal projections are limited to an end date of 2030. We therefore assume that the 2030 prices arerepresentative of the 2050 scenario. We note that the coal values are provides in $US–we assume a constant exchange rate of £1 = $1.6 in this work (DECC, 2013, 2014).

Low scenario (pessimistic) Central scenario (central) High scenario (optimistic)

CO2 (£/t) Coal (£/t) CO2 (£/t) Coal (£/t) CO2 (£/t) Coal (£/t)

ct

tajifoot

as

Fd

Fb

2012 13 80 222020 14 52 252050 100 52 200

oal + CCS. This is quite expected and these results are in line withhose of Morrow et al. (2007).

However, as the CO2 and coal prices start to increase in the cen-ral scenario, we see CCS becoming the cost-optimal choice as earlys 2020 (Table E2 in Appendix E). This implies that given our pro-ected fuel and CO2 prices, the decarbonisation of power generationn the 2020s would be a cost optimal solution. As can be concludedrom the comparison of Figs. 2 and 3, the trend in the optimal choicef power generation technologies in this scenario as we move fromne end to the other end of the Pareto curve for 2012 is similar tohat observed in the pessimistic decarbonisation scenario.

Finally, in our optimistic decarbonisation scenario (high coalnd CO2 prices) – illustrated in Fig. 4 – CCS technology starts beingelected in the cost optimal solution for each case.

ig. 2. Trade-off between total cost and total carbon intensity in the pessimisticecarbonisation scenarios for years 2012, 2020 and 2050.

ig. 3. Trade-off between total cost and total carbon intensity in the central decar-onisation scenarios for years 2012, 2020 and 2050.

84 28 8971 31 9871 300 100

4.2. Optimal network configurations

For the representation of the optimal network configurations,the central decarbonisation scenario (with carbon and coal prices)has been selected as an average estimate of the future market con-ditions. The optimal minimum cost and minimum carbon intensitynetwork configurations for 2020 with central carbon and carbonprices are shown in Fig. 5a and b, respectively, where the opti-mal solutions are presented at the regional level (including the 10regions in the UK) for ease of visualisation. The mapping betweenthese 10 regions and the 34 square cells is given in Table E1 inAppendix E. The corresponding detailed optimal configurations atthe cell level and the optimal power generation variables can alsobe seen in the Figs. E1 and E2 and Tables E2–E5 in Appendix E. Theoptimal configuration figures represent the total optimal rates ofbiomass and MSW supply as well as the total optimal number ofpellet plants established in each region (as indicated by the num-bers inside the yellow squares). They also provide information onthe optimal selection of co-firing plants (indicated by their loca-tions) and the optimal electricity generation rates in these plants.As can be seen from the comparison of the two optimal config-urations, in 2020 under the central decarbonisation scenario, theminimum cost configuration chooses electricity generation fromcoal only whereas the minimum carbon intensity configurationchooses to co-fire all the available domestic biomass and MSWresources with coal to make maximum use of the potential benefitsof the non-fossil fuel resources in reducing the emissions. This leadsto a reduction of the average carbon intensity from 43 kg CO2/MWhto −183 kg CO2/MWh with a corresponding increase of the averagecost from 47 £/MWh to 86 £/MWh. In the minimum carbon inten-sity configuration, the number of pellet plants established in eachregion is in general proportional to the availabilities of biomass and

MSW in that region.

The minimum cost and minimum carbon intensity configura-tions for 2050 with central carbon and coal prices are presented inFig. 6a and b, respectively. In contrast to the 2020 scenario, owing

Fig. 4. Trade-off between total cost and total carbon intensity in the optimisticdecarbonisation scenarios for years 2012, 2020 and 2050.

O. Akgul et al. / International Journal of Greenhouse Gas Control 28 (2014) 189–202 195

Fig. 5. The optimal bioelectricity supply chain configuration for the (a) minimum cost and (b) minimum carbon intensity options for the 2020 central decarbonisationscenario. The pink and green symbols indicate the total optimal rates of MSW and biomass supply in each region. The yellow squares correspond to the total optimalnumber of palletisation plants established in each region. Finally, the grey symbols represent the total optimal power generation within that region. (For interpretation ofthe references to colour in this figure legend, the reader is referred to the web version of the article.)

Fig. 6. The optimal bioelectricity supply chain configuration for the (a) minimum cost and (b) minimum carbon intensity options for the 2050 central decarbonisationscenario. The pink and green symbols indicate the total optimal rates of MSW and biomass supply in each region. The yellow squares correspond to the total optimalnumber of palletisation plants established in each region. Finally, the grey symbols represent the total optimal power generation within that region. (For interpretation ofthe references to colour in this figure legend, the reader is referred to the web version of the article.)

196 O. Akgul et al. / International Journal of Greenhouse Gas Control 28 (2014) 189–202

Fb

ttsMsaaw−1aat

4

ttti(Cb

mftsfebscdintigp

4

w

ig. 7. Carbon tipping point analysis for the 2012, 2020 and 2050 pessimistic decar-onisation scenarios.

o the significantly increased cost associated with CO2 emissions,he minimum cost configuration chooses to use of the completetock of domestic biomass in addition to 16% of the total availableSW. This results in complete decarbonisation of the electricity

ector in the cost optimal solution. Further decarbonisation can bechieved with the minimum carbon intensity configuration wherell the available domestic biomass and MSW are used for co-firingith coal. This has an effect of reducing the carbon intensity from109 kg CO2/MWh to −238 kg CO2/MWh with a small increase of0 £/MWh in the average cost. Due to the higher use of biomassnd MSW in the minimum carbon intensity configuration, therere more pellet plants established in that configuration comparedo the minimum cost configuration.

.3. Carbon tipping point

One of the objectives of this work is to provide some insight aso the cost associated with CO2 emission which would incentivisehe generation of carbon negative electricity – a tipping point. Inhe carbon tipping point analysis, we take a pessimistic decarbon-sation scenario where there is little incentive for decarbonisationi.e., low carbon and low coal prices) and try to determine theO2 price range which will prompt the co-deployment of CCS andiomass co-firing.

Fig. 7 shows change of the total carbon intensity of a mini-um cost system with increase in the carbon price levels. As seen

rom the figure, there is a similar trend in the system responseo the change in carbon price for the three snapshot years beingtudied. As can be concluded from the comparison of the figuresor 2012, 2020 and 2050, the earliest switch to carbon negativelectricity generation occurs in the case of 2050 at an average car-on price level of £120/t. The latest switch is in 2012 pessimisticcenario at an average price level of £175/t. This arises from theombined effect of higher coal price and lower availabilities ofomestic biomass and MSW in 2012 compared to 2020 and 2050. It

s, however, interesting to note that in both the 2012 and 2020 sce-arios, a relatively modest price in terms of £/t of CO2 is requiredo incentivise the generation of low carbon electricity–somethingn the region of £35/t CO2 would appear to be sufficient. This isood agreement with the bottom-up analysis we have presentedreviously (Mac Dowell and Shah, 2013).

.4. High biomass availability

We also investigated how an increase in biomass availabilityould affect the carbon intensity of a minimum cost configuration.

Fig. 8. Results of the high biomass availability analysis with the 2020 central decar-bonisation scenario.

For this purpose we considered three cases of availability: basecase with the nominal biomass availability, high availability case(double the base case) and very high availability case (five timesthe base case). Fig. 8 shows the lower end of the Pareto curves forthese three cases in the 2020 central decarbonisation scenario. Inthe interest of clarity, the upper portion of the Pareto curves i.e.,the high carbon intensity extreme, is not shown here as each ofthe cases has the same point of origin. As can be concluded fromthe comparison of the three curves, cost of decarbonisation can bereduced significantly with higher biomass availability; for exampleto achieve a carbon intensity of −190 kg CO2/MWh with the basecase corresponds to a cost of approximately £82/MWh, whereaswith increased biomass availability, this carbon intensity can beachieved for approximately £73/MWh and £66/MWh high biomassand very high biomass availability scenarios, respectively. Finally,the minimum carbon intensities for the high and very high biomassavailabilities reduces the carbon footprint of the power plants toabout −27 MT CO2/yr and −31 MT CO2/yr for the high and very highbiomass availability scenarios, respectively.

5. Conclusions

This paper analyses co-firing of biomass in conjunction withCO2 capture and storage in 10 existing coal-fired power stationsin the UK. Both municipal solid waste and conventional biomassresources have been considered, which has the important advan-tage of increasing the biomass availability in the system. We haveinvestigated three different scenarios (low, central and high pricesfor CO2 and coal) for each of 2012, 2020 and 2050.

We have investigated the levels of CO2 price that would berequired to incentivise the generation of carbon negative electric-ity in a low carbon and low coal price scenario. A tipping pointanalysis has been carried out for this purpose where the changein the carbon intensity of a cost optimal system is compared withan increasing CO2 price. The results indicate that a CO2 price inthe region of £120–175/t will be necessary to incentivise the gen-eration of carbon negative electricity. Increasing availability ofbiomass is key to reducing this cost, however due considerationshould be given to the potential for resultant competition for avail-able crop land.

We have observed that the availability of biomass is an impor-tant constraint on the degree to which carbon negative energy canbe generated. Thus the exploitation of waste derived biomass as asolid recovered fuel source provides a useful route to at least soften

Green

tHuittftlm

ftictpolwarcnv“UtIiumasam

tcnopbtbgts

A

Iefgi

lp

SFFF

I

O. Akgul et al. / International Journal of

his constraint whilst additional energy crops can be produced.owever, it has been observed previously that the promotion of these of biomass for energy applications can cause a rapid increase

n the price of biomass (Wianwiwat and Asafu-Adjaye, 2013). Ashe sharp increase in electricity costs associated with the transi-ion from low carbon to carbon negative electricity arises in partrom the cost of biomass, we would suggest that before ambitiousargets of biomass utilisation for electricity production are estab-ished, one would first ensure that there is sufficient biomass to

eet the subsequent demands.We propose that the co-firing of biomass and solid recovered

uels in conjunction with CO2 capture provides a promising routeo the near to medium term generation of carbon negative electric-ty. Importantly, given existing coal-fired power plants, co-firingan be implemented with relatively low capital costs within aimeframe of fewer than 5 years. The retro-fitting of CO2 capturerocesses may take somewhat longer, but the relative immediacyf co-firing provides one possible route to extending the operatingife of existing assets within a carbon constrained energy system. It

ill likely be necessary that power plants generating carbon neg-tive electricity operate at a high load factor with similarly highates of CO2 capture. Achieving this could prove challenging unlessoal + biomass-based CCS becomes a base-load generating tech-ology of choice. A potential option where BECCS could becomeery attractive and allow it to become a base-load technology is acarbon-bubble” concept, analogous to the “Clean Air Act” of theS Environmental Protection Agency (EPA) which allows the EPA

o grant emissions permits for certain pollutants (Crandall, 2008).mportantly, these permits are tradable between polluters, allow-ng one to pay another to reduce their emissions. This concept restspon the specification of a geographic region – or bubble – whichust reduce its total level of CO2 emission by a given amount by

given date. The use of this bubble concept has been shown toignificantly reduce the whole system cost of emission reductionnd may well be attractive option in the context of CO2 emissionitigation as well.We would finally note that this approach is sufficiently general

o be extended to consider other CO2 capture routes, e.g., oxyfuelombustion or high temperature solids looping. All that would beecessary to include this in our model would be the generationf an appropriate meta-model for the performance of a co-firedower plant equipped with an oxyfuel system. This is importantecause preliminary analysis indicates that the sharp “elbow” inhe trade-off curve can be avoided and a much deeper decar-onisation take place once “advanced” technologies (e.g. biomassasification with CCS or biomass-based chemical looping combus-ion with carbon dioxide compression) are available at commercialcale.

ppendix A. Nomenclature

ndices:lec Bioelectricityossil Fossil fuel, g′ Square cells (regions),i′ Material (biomass, MSW, biomass pellet, SRF pellet, fossil fuel,

bioelectricity) Transport mode

Pellet production plant scale

ets: Set of fuels (biomass pellet, SRF pellet, fossil fuel)P Set of final products (bioelectricity)I Set of fuels that can be produced from raw material i (biomass

pellet from biomass, SRF pellet from MSW) Set of materials (biomass, MSW, biomass pellet, SRF pellet,

fossil fuel, bioelectricity) (I = R ∪ F ∪ FP)

house Gas Control 28 (2014) 189–202 197

R Set of raw material types (biomass, MSW)Totaligg′ l et of total transport links allowed for each material i via mode l

between regions g and g′

nigg′ l Subset of Totaligg′ l including all regions g′ in the neighbourhoodof region g for each material i and mode l

P Set of pellet production plant scales

Parameters:ADDgg′ l Actual delivery distance between regions g and g′ via model l

(km)ALDg Average local delivery distance in region g (km)aCI Nameplate capacity coefficient in the carbon intensity

equationa� Nameplate capacity coefficient in the generation efficiency

equationaUGC Nameplate capacity coefficient in the unit generation cost

equation˛ Annual operating hours (h/year)bCI Capacity factor coefficient in the carbon intensity equationb� Capacity factor coefficient in the generation efficiency equationbUGC Capacity factor coefficient in the unit generation cost equationBAig

min/max Minimum/maximum availability of raw material i (i ∈ R) inregion g (t/h)

cCI Extent of carbon capture and storage coefficient in the carbonintensity equation

c� Extent of carbon capture and storage coefficient in thegeneration efficiency equation

cUGC Extent of carbon capture and storage coefficient in the unitgeneration cost equation

CCS Reference extent of carbon capture and storage used in thecarbon intensity, generation efficiency and unit generationcost equations (%)

CI Reference carbon intensity in the carbon intensity equation(kg CO2/MWh)

CRF Capital recovery factor (1/year)dCI Extent of co-firing coefficient in the carbon intensity equationd� Extent of co-firing coefficient in the generation efficiency

equationdUGC Extent of co-firing coefficient in the unit generation cost

equationı Reference capacity factor used in the carbon intensity,

generation efficiency and unit generation cost equations (%)DEM Total electricity demand (MW)� i Conversion factor of raw material i to its pellet (t pellet/t raw

material)ICp Investment cost of a pellet production plant of scale p (£)IMPCig* Unit import cost for importing material i from foreign supplier

g* (£/t)ϕ Reference extent of co-firing used in the carbon intensity,

generation efficiency and unit generation cost equations (%)� Reference generation efficiency in the generation efficiency

equation (%)PCAPp

min/max Minimum/maximum pellet production capacity of a plant ofscale p (t/h)

PPCg Nameplate capacity of the power plant located in region g(MW)

PPC Reference nameplate capacity used in the carbon intensity,generation efficiency and unit generation cost equations (MW)

�i Energy density of fuel type i (i ∈ F) (MJ/t)Qil

max Maximum flowrate of material i via mode l (t/h)UFCg Unit fossil fuel cost at a power plant located in region g (£/t)UCARC Unit carbon cost (£/kg CO2)UGC Reference unit power generation cost in the unit generation

cost equation (£/MWh)UPCip Unit pellet production cost from raw material i at a plant scale

of p (£/t pellet)USCig Unit supply cost of raw material i in region g (£/t)UTCil Unit transport cost of product i via mode l (£/t km)UTC* Unit transport cost for local raw material transfer (£/t km)

Binary variables:Epg 1 if a pellet production plant of scale p is to be established in

region g

Continuous variables:

CCSg Extent of carbon capture and storage in a power plant located

in region g (%)CIg Carbon intensity of a power plant located in region g

(kg CO2/MWh)Dig Demand for raw material i (i ∈ R) in region g (t/h)

1 Greenhouse Gas Control 28 (2014) 189–202

D

ım

�P

PP�

QT

T

U

ϕ

TU2

TUB

Table A3Unit supply cost of biomass per cell in the UK (Wikipedia, 2014; Eunomia, 2013).

Cell 2012 Cell 2012

1 38.3 18 54.72 38.3 19 54.73 38.3 20 62.74 38.3 21 46.55 38.3 22 46.56 38.3 23 67.77 38.3 24 62.78 38.3 25 62.79 38.3 26 46.5

10 38.3 27 46.511 51.4 28 57.612 38.3 29 65.113 51.4 30 65.114 53 31 57.615 53 32 57.616 46.5 33 57.6

98 O. Akgul et al. / International Journal of

fipg Demand for raw material i (i ∈ R) at a pellet production plant ofscale p located in region g (t/h)

g Capacity factor of a power plant located in region g (%)˙ g The fuel mix consumption rate in a power plant located in

region g (t/h)g Generation efficiency of a power plant located in region g (%)fpg Pellet production rate at a plant of size p located in region g

(t/h)ig Production rate of material i in region g (t/h) (i /= elec)elec,g Electricity generation rate in region g (MW)g Energy density of the fuel mix for a power plant located in

region g (MJ/t)igg′ l Flow rate of material i via mode l from region g to g′ (t/h)AC Total annual cost of a bioelectricity supply chain network

(£/year)AE Total annual emissions resulting from a bioelectricity supply

chain network (kg CO2/year)GCg Unit power generation cost at a power plant located in region

g (£/MWh)ig Extent of co-firing of fuel i (i ∈ F) in a power plant located in

region g (%)

able A1K biomass availability data per cell in t/h for years 2012, 2020 and 2050 (CEBR,010; Wikipedia, 2014).

Cell 2012 2020 2050 Cell 2012 2020 2050

1 5.2 28.0 56.0 18 6.5 82.7 118.52 3.9 21.1 42.2 19 3.9 74.9 96.23 7.5 40.6 81.2 20 0.4 48.3 50.74 9.6 52.0 104.0 21 0.8 4.3 8.65 2.5 13.4 26.8 22 4.5 24.7 49.46 6.5 35.3 70.6 23 4.3 66.5 89.97 9.1 49.1 98.2 24 9.4 97.5 149.18 0.4 2.0 4.0 25 4.3 69.3 92.79 4.1 22.3 44.6 26 0.6 3.5 7.010 9.6 52.0 104.0 27 3.4 18.4 36.811 1.7 31.7 41.1 28 2.4 29.4 42.412 0.7 4.1 8.2 29 5.4 60.9 90.713 0.6 15.9 19.0 30 0.9 36.2 41.314 5.6 56.2 86.9 31 0.9 23.4 28.115 1.2 36.5 43.2 32 1.1 24.7 30.716 0.2 1.2 2.4 33 0.6 19.7 23.017 3.0 16.2 32.4 34 0.6 34.4 37.7

TOTAL 2012 121TOTAL 2020 1196TOTAL 2050 1858

able A2K MSW availability data per cell in t/h for years 2012, 2020 and 2050 (The Worldank DataBank, 2013; Almansoori and Shah, 2006; DECC, 2010).

Cell 2012 2020 2050 Cell 2012 2020 2050

1 30.2 31.3 35.5 18 179.7 186.3 211.22 22.8 23.6 26.7 19 106.7 110.6 125.43 43.8 45.4 51.4 20 11.5 12.0 13.64 56.0 58.1 65.8 21 12.4 12.8 14.55 14.4 15.0 17.0 22 72.0 74.7 84.66 38.1 39.5 44.7 23 360.6 373.7 423.77 53.0 54.9 62.2 24 246.3 255.3 289.48 2.2 2.3 2.6 25 111.6 115.7 131.19 24.1 25.0 28.3 26 10.1 10.5 11.910 56.0 58.1 65.8 27 53.5 55.4 62.811 170.8 177.0 200.6 28 163.0 168.9 191.512 4.4 4.5 5.1 29 817.3 847.0 960.213 460.6 477.4 541.2 30 140.5 145.6 165.014 278.5 288.6 327.2 31 58.6 60.7 68.815 60.3 62.5 70.9 32 75.1 77.9 88.316 3.4 3.5 4.0 33 40.8 42.2 47.917 47.3 49.0 55.5 34 89.4 92.6 105.0

TOTAL 2012 3915TOTAL 2020 4058TOTAL 2050 4600

17 46.5 34 65.1

Appendix B. Raw material parameters

The biomass availability data is presented in Table A1. It must benoted that it is assumed that only woodfuel is assumed to be avail-able as biomass resource in 2012 whereas miscanthus will also beavailable by 2020 and 2050 as an additional biomass resource apartfrom woodfuel. Therefore, the biomass availability data for 2012corresponds to the availability of woodfuel only whereas those for2020 and 2050 represent the total availabilities of both miscanthusand woodfuel (CEBR, 2010; Wikipedia, 2014).

The municipal solid waste availability for 2012, 2020 and 2050is given in Table A2. This data has been derived based on the factthat the UK produces 0.5 t of waste per person per year on aver-age (Wikipedia, 2014). The current and projected populations areused to calculate the total availability of municipal solid waste (TheWorld Bank DataBank, 2013) and this total value is distributedbetween different cells in the UK based on their population den-sities (Almansoori and Shah, 2006).

Table A3 shows the unit supply cost of biomass per region inthe UK. This data has been derived based on the fact that the aver-age biomass price is 50 £/t (DECC, 2010) and the cost variation isassumed to be the same as that of wheat prices per region in theUK (Akgul et al., 2012). The cost of MSW has been taken as 50 £/tfor all regions (Eunomia, 2013; DEFRA, 2009).

Table B1Pellet production parameters used in this study (Harvard Green Campus Initiative,2013).

Plant size(p)

PCappmin

(t/h)PCapp

max

(t/h)UPCip

(£/t)ICp (£)

1 1 10 105 94,5752 11 20 98 182,8313 21 30 90 248,405

Appendix C. Pellet production parameters

The cost parameters for different scales of pellet productionplants as well as the corresponding minimum and maximum plantcapacities are given in Table B1 (McCartney, 2007). The conversion

factors for pellet production from biomass and MSW are taken to be1/3 and 1/5, respectively (Harvard Green Campus Initiative, 2013).

O. Akgul et al. / International Journal of Green

Table C1Unit transportation cost for each mode and resource (Wikipedia, 2014).

Unit transportation cost (£/t km)

Mode Biomass MSW Biomasspellet

MSWpellet

A

t

TNt

TOe

TOr

TOc

TOr

Road 0.47 0.47 0.47 0.47Rail 0.17 0.17 0.17 0.17

ppendix D. Transportation parameters

The tortuosity factors for road and rail are 1.4 and 1.2, respec-ively. The actual distance between two cells is calculated by

able D1ame plate capacity, average annual emissions and geographical location of each of

he power plants considered in this study.

Name of powerplant

Fuel Nameplatecapacity(MW)

Averageemissions(Mt CO2/year)

Longitude Latitude Cell

Drax COAL 3870 21.6 −1 53.73 19Cottam COAL 2008 9.4 −0.78 53.31 13Ratcliffe COAL 2000 8.6 −1.25 52.86 18West Burton COAL 1972 8.7 −0.81 53.36 10Eggborough COAL 1960 7.3 −1.13 53.71 14Kingsnorth COAL 1940 7.1 0.6 51.42 30

Didcot A COAL 1925 5.3 −1.26 51.62 28Tilbury B COAL 750 4.2 0.39 51.45 29Ferrybridge C COAL 1955 6.7 −1.29 53.72 18Rugeley COAL 1006 4.2 −1.91 52.75 23

able E2ptimal power plant variables for the 2020 central decarbonisation scenario (minimum

lectricity is generated using coal in conjunction with CO2 capture. This results a CO2 foo

Cell Pelec,g (MW) ıg (%) �g (%) mg (t/h) ϕig (%) (i:biom

13 1766 88% 36% 726 0%

14 1960 100% 36% 802 0%

18 3955 100% 37% 1565 0%

19 3870 100% 37% 1533 0%

able E3ptimal power plant variables for the 2020 central decarbonisation scenario (minimum

ecovered fuels with coal in conjunction with CO2 capture results in a CO2 footprint of ap

Cell Pelec,g (MW) ıg (%) �g (%) mg (t/h) ϕig (%) (i:biom

10 1780 90% 34% 849 0%

14 1766 90% 34% 842 0%

18 3955 100% 35% 1795 22%

28 1727 90% 34% 824 0%

29 579 77% 33% 280 0%

30 1744 90% 34% 832 0%

able E4ptimal power plant variables for the 2050 central decarbonisation scenario (minimum conjunction with CO2 capture results in a CO2 footprint of approximately −10 MT/CO2 pe

Cell Pelec,g (MW) ıg (%) �g (%) mg (t/h) ϕig (%) (i:biom

13 2008 100% 34% 896 19%

14 1718 88% 34% 763 12%

18 3955 100% 36% 1689 9%

19 3870 100% 36% 1651 12%

able E5ptimal power plant variables for the 2050 central decarbonisation scenario (minimum

ecovered fuels in conjunction with CO2 capture results in a CO2 footprint of approximate

Cell Pelec,g (MW) ıg (%) �g (%) mg (t/h) ϕig (%) (i:biom

10 1842 93% 34% 901 0%

13 1885 94% 34% 922 0%

18 3955 100% 35% 1890 0%

19 3870 100% 34% 1794 34%

house Gas Control 28 (2014) 189–202 199

multiplying the tortuosity factors with the linear delivery distance.Road and rail modes are assumed to exist between all elements.The unit transportation cost data is given in Table C1 (Wikipedia,2014).

Appendix E. Power generation parameters

The characteristics of the 10 power plants considered in thisstudy are given in Table D1.

Table E1Discretisation of the UK into square cells.

UK region Cell

North East 11North West and Merseyside 13Yorkshire and The Humber 14, 15East Midlands 18, 19West Midlands 23Eastern 20, 24, 25South East and London 29, 30, 34South West 28, 31, 32, 33Scotland 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12Wales 16, 17, 21, 22, 26, 27

Appendix F. Optimal configurations

This mapping between the 10 UK regions and the 34 cells isgiven in Table E1. The detailed optimal configurations at cell levelare given in Figs. E1 and E2 whereas the corresponding optimalpower generation variables are given in Tables E2–E5.

cost). In this scenario, there is no co-firing of biomass or solid recovered fuels andtprint of approximately 4 MT/CO2 per year.

ass pellet) ϕig (%) (i:MSW pellet) CCSg (%) CIg (kg CO2/MWh)

0% 100% 420% 100% 450% 100% 430% 100% 43

carbon intensity). In this scenario, a combination of co-firing biomass or solidproximately −16.9 MT/CO2 per year.

ass pellet) ϕig (%) (i:MSW pellet) CCSg (%) CIg (kg CO2/MWh)

20% 100% −16120% 100% −161

5% 100% −22720% 100% −16018% 100% −14420% 100% −161

ost). In this scenario, a combination of co-firing biomass or solid recovered fuels inr year.

ass pellet) ϕig (%) (i:MSW pellet) CCSg (%) CIg (kg CO2/MWh)

0% 100% −1453% 100% −1095% 100% −973% 100% −104

carbon intensity). In this scenario, a combination of co-firing biomass or solidly −22 MT/CO2 per year.

ass pellet) ϕig (%) (i:MSW pellet) CCSg (%) CIg (kg CO2/MWh)

24% 100% −19524% 100% −19626% 100% −223

0% 100% −294

200 O. Akgul et al. / International Journal of Greenhouse Gas Control 28 (2014) 189–202

Fig. E1. The optimal bioelectricity supply chain configuration for the (a) minimum cost and (b) minimum carbon intensity options for the 2020 central decarbonisationscenario. The pink and green symbols indicate the total optimal rates of MSW and biomass supply in each region. The yellow squares correspond to the total optimal numberof palletisation plants established in each region. Finally, the grey symbols represent the total optimal power generation within that region, e.g., in cell 18 the number 3995represents the combined generation capacity of the Ferrybridge C and Ratcliffe power plants operating at 100% of their capacity. (For interpretation of the references to colourin this figure legend, the reader is referred to the web version of the article.)

Fig. E2. The optimal bioelectricity supply chain configuration for the (a) minimum cost and (b) minimum carbon intensity options for the 2050 central decarbonisationscenario. The pink and green symbols indicate the total optimal rates of MSW and biomass supply in each region. The yellow squares correspond to the total optimal numberof palletisation plants established in each region. Finally, the grey symbols represent the total optimal power generation within that region, e.g., in cell 18 the number 3995represents the combined generation capacity of the Ferrybridge C and Ratcliffe power plants operating at 100% of their capacity. (For interpretation of the references to colourin this figure legend, the reader is referred to the web version of the article.)

Green

R

A

A

A

A

C

2

I

A

A

B

B

B

B

C

C

D

D

D

D

D

D

D

D

E

E

E

F

F

Tittmann, P.W., Parker, N.C., Hart, Q.J., Jenkins, B.M., 2010. A spatially explicit techno-

O. Akgul et al. / International Journal of

eferences

guilar, J.E.S., Campos, J.B.G., Ortega, J.M.P., Gonzalez, M.S., El-Halwagi, M.M., 2011.Optimal planning of a biomass conversion system considering economic andenvironmental aspects. Ind. Eng. Chem. Res. 50, 8558–8570.

kgul, O., Zamboni, A., Bezzo, F., Shah, N., Papageorgiou, L.G., 2011. Optimization-based approaches for bioethanol supply chains. Ind. Eng. Chem. Res. 50,4927–4938.

kgul, O., Shah, N., Papageorgiou, L.G., 2012. An optimisation framework for ahybrid first/second generation bioethanol supply chain. Comput. Chem. Eng.42, 101–114.

lmansoori, A., Shah, N., 2006. Design and operation of a hydrogen supply chain.Chem. Eng. Res. Des. 84, 423–438.

EBR Ltd., 2010. The Economic Value of the Woodfuel Industry to the UKEconomy by 2020, Available from: http://www.biomassenergycentre.org.uk/pls/portal/docs/pAge/BeC teCHNICAL/reseArCH%20AND%20stUDIes/eCoNoMIC%20stUDIes/2010-06-30%20eCoNoMIC%20VALUe%20oF%20WooDFUeLreVIseD%20report.pDF (accessed May 2013).

010. E.ON UK’s Ratcliffe-on-Soar Power Station, http://www.eon-uk.com/generation/ratcliffe.aspx

ECM, 2012. Integrated Environmental Control Model (IECM), Available from:http://www.cmu.edu/epp/iecm/

rce, A., Mac Dowell, N., Shah, N., Vega, L.F., 2012. Flexible operation of solventregeneration systems for CO2 capture processes using advanced control tech-niques: Towards operational cost minimisation. Int. J. Greenh. Gases Control 11,236–250.

vami, A., 2012. A model for biodiesel supply chain: a case study in Iran. Renew.Sustain. Energy Rev. 16, 4196–4203.

iliyok, C., Lawal, A., Wang, M., Seibert, F., 2012. Dynamic modelling, validationand analysis of post-combustion chemical absorption CO2 capture plant. Int.J. Greenh. Gases Control 9, 428–444.

oot-Handford, M.E., Abanades, J.C., Anthony, E.J., Blunt, M.J., Brandani, S., MacDowell, N., Fernández, J.R., Ferrari, M.C., Gross, R., Hallett, J.P., Haszeldine, R.S.,Heptonstall, P., Lyngfelt, A., Makuch, Z., Mangano, E., Porter, R.T.J., Pourkasha-nian, M., Rochelle, G.T., Shah, N., Yao, J.G., Fennell, P.S., 2014. Energy Environ. Sci.7, 130–189.

owling, I.M., Ortega, J.M.P., El-Halwagi, M.M., Floudas, C.A., 2011. Facility loca-tion and supply chain optimization for a biorefinery. Ind. Eng. Chem. Res. 50,6276–6286.

ryngelsson, D.K., Lindgren, K., 2013. Why large-scale bioenergy production onmarginal land is unfeasible: a conceptual partial equilibrium analysis. EnergyPolicy 55, 454–466.

randall, R.W., 2008. Pollution Controls, Available from The library of Eco-nomics and Liberty, http://www.econlib.org/library/Enc/PollutionControls.html

ucek, L., Varbanov, P.S., Kelemes, J.J., Kravanja, Z., 2012. Total footprints-basedmulti-criteria optimisation of regional biomass energy supply chains. Energy44, 135–145.

ECC, 2010. Biomass Prices in the Heat and Electricity Sectors in theUK [online]. Available from: http://www.rhincentive.co.uk/library/regulation/100201Biomass prices.pdf (accessed May 2013).

ECC, 2011. Electricity Market Reform. http://www.decc.gov.uk/en/content/cms/legislation/white papers/emr wp 2011/emr wp 2011.aspx#

ECC, 2012a. Digest of United Kingdom Energy Statistics [online]. Available from:http://www.decc.gov.uk/assets/decc/11/stats/publications/dukes/5949-dukes-2012-exc-cover.pdf (accessed May 2013).

ECC, 2012b. UK Renewable Energy Roadmap Update [online]. Available from:https://www.gov.uk/government/uploads/system/uploads/attachment data/file/80246/11-02 13 UK Renewable Energy Roadmap Update FINAL DRAFT.pdf(accessed May 2013).

ECC, 2013. Carbon Prices. http://www.decc.gov.uk/en/content/cms/emissions/valuation/valuation.aspx

ECC, 2014. Coal Prices. http://www.decc.gov.uk/assets/decc/11/about-us/economics-social-research/2936-decc-coal-price-projections.pdf

EFRA, 2009. Solid Recovered Fuel Regional Assessments [online]. Available from:http://archive.defra.gov.uk/environment/waste/residual/widp/documents/SRF-Hull-East-Riding.pdf (accessed May 2013).

unnett, A., Adjiman, C.S., Shah, N., 2007. Biomass to heat supply chains: applicationsof process optimization. Trans IChemE B 85 (B5), 419–429.

lia, J.A., Baliban, R.C., ad Floudas, C.A., 2012. Nationwide energy supply chain anal-ysis for hybrid feedstock processes with significant CO2 emissions reduction.AIChE J. 58 (7), 2142–2154.

unomia, 2013. Costs for Municipal Waste Management in the EU [online]. Availablefrom: http://ec.europa.eu/environment/waste/studies/pdf/eucostwaste.pdf(accessed May 2013).

uropean Union Joule III Project, 2001. Reduction of Toxic Emissions from Co-Combustion of Coal, Biomass and Waste in Fluidised Beds. Final Report [JOR3-CT97-091].

ortes, M.P., Aguirre, J.M.L., Piera, P.A., Velo, E., Puijganer, L., 2012. Design of regionaland sustainable bio-based networks for electricity generation using a multi-objective MILP approach. Energy 44, 79–95.

uss, S., 2012. BECCS status and gaps: results from the IEA-IIASA Workshop onBECCS. In: Proceedings of IEA/IIASA BECCS Experts Workshop, Available from:http://www.iea.org/media/workshops/2012/bioenergyccsandbeccs/1IIASABECCS wksp Laxenburg.pdf

house Gas Control 28 (2014) 189–202 201

Gan, J., Smith, C.T., 2011. Optimal plant size and feedstock supply radius: a mod-elling approach to minimize bioenergy production costs. Biomass Bioenergy 35,3350–3359.

Giarola, S., Zamboni, A., Bezzo, F., 2012. Total environmentally conscious capacityplanning and technology selection for bioethanol supply chains. Renew. Energy43, 61–72.

Goeppert, A., Czaun, M., May, R.B., Prakash, G.K.S., Olah, G.A., Narayanan, S.R., 2011.Carbon dioxide capture from the air using a polyamine based regenerable solidadsorbent. J. Am. Chem. Soc. 133 (50), 20164–20167.

Gough, C., Upham, P., 2011. Biomass energy with carbon capture and storage (BECCSor Bio-CCS). Greenh. Gases Sci. Technol. 1 (4), 324–334.

Harvard Green Campus Initiative, 2013. Biomass Fact Sheet [online]. Available from:http://green.harvard.edu/sites/default/files/attachments/renewables/biomass-fact-sheet.pdf (accessed May 2013).

Lackner, K.S., 2009. Capture of carbon dioxide from ambient air. Eur. Phys. J. Spec.Top. 176 (1), 93–106.

Lawal, A., Wang, M., Stephenson, P., Yeung, H., 2009. Dynamic modelling of CO2

absorption for post combustion capture in coal-fired power plants. Fuel 88 (12),2455–2462.

Lawal, A., Wang, M., Stephenson, P., Koumpouras, G., Yeung, H., 2010. Dynamicmodelling and analysis of post-combustion CO2 chemical absorption processfor coal-fired power plants. Fuel 89 (10), 2791–2801.

Mac Dowell, N., Shah, N., 2013. Identification of the cost-optimal degree of CO2

capture: an optimisation study using dynamic process models. Int. J. Greenh.Gases Control 13, 44–58.

Mac Dowell, N., Florin, N., Buchard, A., Hallett, J., Galindo, A., Jackson, G., Adjiman,C.S., Williams, C.K., Shah, N., Fennell, P.S., 2010a. An overview of CO2 capturetechnologies. Energy Environ. Sci. 3, 1645–1669.

Mac Dowell, N., Galindo, A., Jackson, G., Adjiman, C.S., 2010b. Integrated solvent andprocess design for the reactive separation of CO2 from flue gas. Comput. Aid.Chem. Eng. 28, 1231–1236.

Mac Dowell, N., Alhajaj, A., Konda, M., Shah, N., 2011. Multiscale whole-systemsdesign and analysis of CO2 capture and transport networks. Comput. Aid. Chem.Eng. 29, 1205–1209.

Mac Dowell, N., Samsatli, N.J., Shah, N., 2013. Dynamic modelling and analysis of anamine-based post-combustion CO2 capture absorption column. Int. J. Greenh.Gases Control 12, 247–258.

Mahmoudkhani, M., Keith, D.W., 2009. Low-energy sodium hydroxide recoveryfor CO2 capture from atmospheric air—thermodynamic analysis. Int. J. Greenh.Gases Control 3 (4), 376–384.

Mann, M.K., Spath, P.L., 2001. A life cycle assessment of biomass cofiring in a coal-fired power plant. Clean Prod. Process. 3, 81–91.

McCartney, C., 2007. A Feasibility Study for Small-scale Wood Pellet Pro-duction in the Scottish Borders [online]. Available from: http://www.energyfarming.org.uk/resources/Wood%20Pellet%20final%20report%20CM%20Nov07.pdf (accessed May 2013).

McKechnie, J., Colombo, S., Chen, J., Mabee, W., MacLean, H.L., 2011. Forest bioenergyor forest carbon? Assessing trade-offs in greenhouse gas mitigation with wood-based fuels. Environ. Sci. Technol. 45 (2), 789–795.

Miller, B.B., Dugwell, D.R., Kandiyoti, R., 2002a. Partitioning of trace elements dur-ing the combustion of coal and biomass in a suspension-firing reactor. Fuel 81,159–171.

Miller, B.B., Dugwell, D.R., Kandiyoti, R., 2002b. Trace element emissions fromco-combustion of secondary fuels with coal: a comparison of bench-scale exper-imental data with predictions of a thermodynamic equilibrium model. EnergyFuels 16, 956–963.

Miller, B.B., Dugwell, D.R., Kandiyoti, R., 2003. The influence of injected HCl and SO2

on the behavior of trace elements during wood-bark combustion. Energy Fuels17, 1382–1393.

Miller, B.B., Dugwell, D.R., Kandiyoti, R., 2004. Trace element behavior dur-ing co-combustion of sewage sludge with polish coal. Energy Fuels 18,1093–1103.

Miller, B.B., Dugwell, D.R., Kandiyoti, R., 2006. The fate of trace elements during theco-combustion of wood-bark with waste. Energy Fuels 20 (2), 520–531.

Morrow, W.R., Griffin, W.M., Matthews, H.S., 2007. State-level infrastructure andeconomic effects of switchgrass cofiring with coal in existing power plants forcarbon mitigation. Environ. Sci. Technol. 41 (19), 6657–6662.

Mosier, A.R., Halvorson, A.D., Peterson, G.A., Robertson, G.P., Sherrod, L., 2005. Mea-surement of net global warming potential in three agroecosystems. Nutr. CyclingAgroecosyst. 72 (1), 67–76.

Orchid Environmental Ltd, 2014. http://www.orchid-environmental.co.ukRentizelas, A.A., Tatsiopoulos, I.P., Tolis, A., 2009. An optimization model for multi-

biomass tri-generation energy supply. Biomass Bioenergy 33, 223–233.Schulze, E., Korner, C., Law, B.E., Haberl, H., Luyssaert, S., 2012. Large-scale bioenergy

from additional harvest of forest biomass is neither sustainable nor greenhousegas neutral. GCB Bioenergy 4 (6), 611–616.

Still, D., 2012. Air Fuel Synthesis. http://www.imeche.org/news/archives/12-10-15/UK engineers create petrol from air.aspx

The World Bank DataBank, 2013 [online]. Available from: http://databank.worldbank.org/data/views/reports/tableview.aspx (accessed May 2013).

economic model of bioenergy and biofuels production in California. J. Trans.Geog. 18, 715–728.

UK Dept of Energy and Climate Change, 2012, August. Technology InnovationNeeds Assessments: Carbon Capture and Storage in the Power Sector Summary

2 Green

UU

W

W

1157–1180.Zhang, Y., McKechnie, J., Cormier, D., Lyng, R., Mabee, W., Ogino, A., MacLean, H.L.,

2010. Life cycle emissions and cost of producing electricity from coal, nat-

02 O. Akgul et al. / International Journal of

Report. Author: Low Carbon Innovation Coordination Group, Available from:http://www.lowcarboninnovation.co.uk/working together/technology focusareas/carbon capture and storage/

N Conference on Sustainable Development, 2012. http://www.uncsd2012.org/N Earth Summit, 1992. UN Conference on Environment and Development.

http://www.un.org/geninfo/bp/enviro.htmlallquist, L., Seigo, S.L., Visschers, V.H.M., Siegrist, M., 2012. Public acceptance of

CCS system elements: a conjoint measurement. Int. J. Greenh. Gases Control 6,77–83.

ianwiwat, S., Asafu-Adjaye, J., 2013. Is there a role for biofuels in promoting energyself-sufficiency and security? A CGE analysis of biofuel policy in Thailand. EnergyPolicy 55, 543–555.

house Gas Control 28 (2014) 189–202

Wikipedia, 2014. List of Urban Areas in the United Kingdom. http://en.wikipedia.org/wiki/List of urban areas in the United Kingdom

You, F., Tao, L., Graziano, D.J., Snyder, S.W., 2012. Optimal design of sus-tainable life cycle assessment and input–output analysis. AIChE J. 58 (4),

ural gas, and wood pellets in Ontario, Canada. Environ. Sci. Technol. 44 (1),538–544.