ARTICLE IN PRESS

Progress in Energy and Combustion Science 33 (2007) 580609

Department of Science, Technology and Society, Copernicus Institute for Sustainable Development and Innovation,

r 2007 Elsevier Ltd. All rights reserved.

3.4. Infrastructural requirements and costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589

www.elsevier.com/locate/pecs0360-1285/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.pecs.2007.02.002

Corresponding author. Tel.: +31 30 2537643; fax: +31 30 2537601.E-mail address: a.faaij@chem.uu.nl (A. Faaij).Keywords: Chain analysis; CO2 capture; CO2 transport; CO2 storage; Hydrogen; Infrastructure

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581

2. Chain analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583

3. Chain description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584

3.1. CO2 sources: electricity and hydrogen production technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 585

3.2. CO2 sinks: geological reservoirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 585

3.3. Hydrogen end-use markets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 587Utrecht University, 3584 CS Utrecht, Netherlands

Available online 20 April 2007

Abstract

Promising electricity and hydrogen production chains with CO2 capture, transport and storage (CCS) and energy carrier

transmission, distribution and end-use are analysed to assess (avoided) CO2 emissions, energy production costs and CO2mitigation costs. For electricity chains, the performance is dominated by the impact of CO2 capture, increasing electricity

production costs with 1040% up to 4.56.5 hct/kWh. CO2 transport and storage in depleted gas elds or aquifers typically

add another 0.11 hct/kWh for transport distances between 0 and 200 km. The impact of CCS on hydrogen costs is small.

Production and supply costs range from circa 8 h/GJ for the minimal infrastructure variant in which hydrogen is delivered

to CHP units, up to 20 h/GJ for supply to households. Hydrogen costs for the transport sector are between 14 and 16 h/GJ

for advanced large-scale coal gasication units and reformers, and over 20 h/GJ for decentralised membrane reformers.

Although the CO2 price required to induce CCS in hydrogen production is low in comparison to most electricity

production options, electricity production with CCS generally deserves preference as CO2 mitigation option. Replacing

natural gas or gasoline for hydrogen produced with CCS results in mitigation costs over 100 h/t CO2, whereas CO2 in the

power sector could be reduced for costs below 60 h/t CO2 avoided.A comparison of electricity and hydrogen production systemswith CO2 capture and storagePart B: Chain analysis of

promising CCS options

Kay Damen, Martijn van Troost, Andre Faaij, Wim Turkenburg

3.4.1. Electricity transmission and distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589

3.4.2. CO2 transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589

3.4.3. H2 transmission and distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592

3.4.4. Costs of CO2 and H2 infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594

3.4.5. Other H2 infrastructure requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596

3.5. CCS chains overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597

4. Results chain analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597

4.1. Electricity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597

4.2. Hydrogen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598

4.3. Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602

5. Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 607

1. Introduction

In recent years, carbon dioxide capture andstorage (CCS) has received much attention for its

potential to achieve major CO2 reductions in acarbon constrained world. In part A, we made aninventory and techno-economic comparison ofelectricity and hydrogen production technologies

ARTICLE IN PRESS

lant

CO2 transmission

CO2 transmission

power plant

electricity

transmission

CO2 injection

CO2 injection

electricity

distribution

coal

gas distribution

Fig. 1. Differ

power produc

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 581H2 plant

H2 transmission

gas production

gas transmission

CO2 transmissionCO2 injection

ent electricity and hydrogen production chains with CCS (from top to bottom: central power production, decentralisedH2 distributionpowerpgas production

gas transmissiontion, central hydrogen production).electricity

distribution

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609582with CO2 capture to identify promising options [1].The system boundary of this analysis was set at theproduction plant, i.e. only energy conversion, CO2capture and compression were considered. Thecomplete CCS chain also encompasses fuel extrac-tion and transport, CO2 transport and storage andenergy carrier transmission, distribution and end-

Nomenclature

ATR autothermal reformingAZEP advanced zero emission power plantBF blast furnaceBOF basic oxygen furnaceCCS CO2 capture and storageCG coal gasicationCHP combined heat and power productionCLC chemical looping combustionCOE cost of electricityCOH cost of hydrogenGT gas turbineDR direct reductionEAF electric arc furnaceEOR enhanced oil recoveryFCV fuel cell vehicleGHG greenhouse gasHMCMhydrogen mixed conducting membraneuse, causing additional CO2 emissions and costs andthereby affecting the overall chain performance. Asthe technologies studied in [1] differ in fuel type, theamount of CO2 captured and in scale, the impact ofthe chain elements outside the plant boundary mayaffect the performance of the technologies differ-ently (see Fig. 1). For this reason, the completechain has to be considered to compare (bituminous)coal with natural gas-red options, oxyfuel withpost-combustion capture options, and central withdecentralised technologies. A comparison of elec-tricity production with CCS versus hydrogenproduction with CCS as competing CO2 reductionoptions cannot be performed on plant level only. Itrequires the assessment of CO2 mitigation costsversus a reference system, such as gasoline ornatural gas in case of hydrogen. Since the distribu-tion and end-use of hydrogen is different withrespect to the fuels it substitutes, these elementshave to be incorporated.Relatively few studies have been performed in

which the entire chain is analysed. Most of thesestudies focus on large-scale combustion or gasica-tion systems and include only a number of chainelements. CCS at decentralised units is generallyconsidered incompatible due to economies of scale.Hendriks et al. performed a generic chain analysisfor several central electricity and hydrogen chainswith CCS to assess additional costs and avoidedemissions versus the average park and natural gas

HSD hydrogen separation deviceICEV internal combustion engine vehicleIGCC integrated gasication combined cycleLHV lower heating valueLS liquid steelMEA monoethanolamineMOB transport sectorMR membrane reformerNGCC natural gas combined cycleO&M operating and maintenancePC pulverised coal-red power plantPEMFC proton exchange membrane fuel cellRES residential sectorROW right of waySMR steam methane reformingSOFC solid oxide fuel cellST steam turbineSTL steel productionTCR total capital requirement[2]. Costs and energy use of CO2 and energytransmission (excluding distribution to end-usemarkets) for central power and hydrogen plantshave been studied in [3]. Fuel supply chains andassociated costs for different hydrogen productionand supply technologies have been assessed in [4].Ogden studied large-scale hydrogen productionand supply chains for the transport sector in theUSA, including CCS [5]. Also various Dutch CCSchains have been analysed, in which existingelectricity and industrial plants were linked topossible storage sites [6].However, no study could be identied in which

the aspects of time (short versus long term),scale (central versus decentralised units), energytransmission and distribution for different end-usersand choice of the reference system are explicitlyand consistently dealt with for both electricity andhydrogen systems with CCS. Therefore, in thisstudy, promising technologies identied in part Aare further assessed in a chain analysis that accountsfor these factors. Overall CO2 emissions, electricityand hydrogen production costs and CO2 mitigation

costs of various technologies are assessed fordifferent technological and infrastructural settings,and various reference systems. This allows us toconsistently compare a wide range of CCS technol-ogies in different contexts. By including differenthydrogen end-users, the performance of hydrogenin different sectors can be compared. The chainanalysis also provides insight into the economictrade-off between CO2 and energy transmission.This knowledge is of importance in site selection ofnew plants, for the construction of new infrastruc-

crucial parameters, followed by discussion andconclusions in Section 5.

2. Chain analysis

Fig. 2 shows the different elements of the CCSchains. In the analysis, greenhouse gas (GHG)emissions of fossil fuel extraction, transport anddistribution are accounted for. In order to captureCO2, additional fuel is required, which results in

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 583ture involves large investments. Finally, theCO2 emissions over the entire chain give anindication of the climate neutrality of decarbonisedelectricity and hydrogen versus its reference asapplied in [2].The geographical focus is on the Netherlands,

representing a densely populated industrialisedcountry, where CCS would typically be deployed.CCS may play a signicant role in the Netherlandsin the coming decades, as this country is char-acterised by numerous large CO2 sources andpotential sinks. The Dutch CO2 emission in 2003was circa 177Mt CO2, of which approximately100Mt emitted by the energy and manufacturingindustry [7]. Power and heat production accountedfor 55MtCO2. Large point sources (40.1MtCO2/yr) at which CO2 capture is feasible representedcirca 96MtCO2 [8]. The estimated technical storagecapacity of Dutch on- and offshore aquifers and gaselds exceeds 11GtCO2 [8], implying CCS couldpotentially be deployed for many decades to come.After the discussion of the main methodological

issues in Section 2, the electricity and hydrogenchains are described in Section 3. These chains areinspired by the future energetic and geological mapof the Netherlands, which gives an overview of CO2sources, hydrogen end-use markets and CO2 storagereservoirs. Section 4 presents the results of the chainanalysis, including a sensitivity analysis for severalFig. 2. Elements in electricity and hydroadditional GHG emissions due to coal mining andnatural gas extraction, as well as transport anddistribution of these fuels. Electricity and hydrogentransmission and distribution is accounted for assignicant energy losses and/or additional costs canoccur depending on the system layout [3,5]. Hydro-gen end-use is accounted for because specichydrogen conversion technologies (fuel cells) canbe fundamentally different from reference end-usetechnologies for conversion of hydrocarbons (tur-bines, boilers, internal combustion engines).For each chain, we set up an energy and CO2

balance and calculate levelised production costs ofelectricity and hydrogen (COE and COH forelectricity and hydrogen, respectively) and CO2mitigation costs. COE and COH are calculated bydividing the sum of annual capital and O&M costsof the conversion system and infrastructure, fuelcosts and CO2 storage costs by the annual energyproduction. Costs are converted to h2003 using GDPdeators [9] and annual currency exchange rates[10]. CO2 mitigation costs for electricity productionwith CCS are calculated using the followingformula:

CO2 mitigation costs COECCS COErefmCO2 ;ref mCO2 ;CCS

, (1)

in which COE is the cost of electricity (h/kWh) andm the CO2 emission factor (kg/kWh) of the CCSchain and reference chain.gen production chains with CCS.

3. Chain description

CCS chains, which combine specic electricityand hydrogen production technologies via specicinfrastructure to specic CO2 storage reservoirs andspecic end-users, have both a spatial and temporaldiminfcowitracava

(in h/kWh) over a certain distance are higher than

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609584The choice of the reference system has asignicant impact on calculated CO2 mitigationcosts. In this analysis, the impact of the variousreference systems will be assessed. In the mostcommon approach as applied in part A, identicalplants with and without CO2 capture are compared(i.e. the baseline varies per technology). This methodgives a good indication which technology inherentlyenables low-cost CO2 capture. However, a plantwith CO2 capture does not necessarily replace anidentical plant without capture, as the constructionof a new power plant is determined by market forcesand policies. In addition, the plant with low capturecosts can be expensive as such. In the Netherlands,the fuel mix for power generation consists primarilyof coal and natural gas, so it can be argued toconsider a PC and NGCC as reference system.Hendriks et al. [2] make a distinction between aproject independent approach using generic emissionand cost gures, and a project specic approach,using project specic data as reference. In the formerapproach, the reference system could be the currentDutch power generation mix, resulting in moregeneralised CO2 mitigation costs.Hydrogen produced with CO2 capture should be

compared to the fuel it substitutes: gasoline or dieselin the transport sector, natural gas in householdsand industry, or hydrogen (produced without CO2capture), e.g. for chemical purposes. In the longterm, hydrogen might replace cokes as reducingagent in steel production [11]. When hydrogenproduced with CO2 capture replaces conventionallyproduced hydrogen, the common formula tocalculate CO2 mitigation costs can be applied(Eq. (1)). For the cases where it substitutes gasoline,natural gas or cokes, the end-use technology shouldbe accounted for as well (Eq. (2)):

CO2 mitigation costs COH=Z CO&MH2 COF=Z CO&Mref

mCO2Zref mCO2

ZH2

,

2

in which COH is the hydrogen cost (h/GJ on LHVbasis); COF the reference fuel costs (h/GJ); Z theend-use efciency (functional unit/GJ); C the capitalcost of end-use (h/functional unit); O&M theoperating and maintenance costs (h/functionalunit); and m the CO2 emission factor (kg/GJ).A functional unit can be a kilometre, a GJelectricity/heat or a tonne of steel.for natural gas and CO2 (onshore conditions) [3]. Itmay be more advantageous, however, to produceelectricity nearby the CO2 storage reservoir pro-vided that this reservoir is closely located to thenatural gas source. Such trade-offs also exist forhydrogen, natural gas and CO2 transmission. It hasbeen estimated that hydrogen will cost between 30%and 50% more to transport than an equivalentenergetic quantity of natural gas [17]. Althoughhydrogen has a lower molecular weight andviscosity in comparison to gas, which makeshydrogen ow faster, the volumetric energy contentof hydrogen is about one third of that of naturalgas. The trade-off between CO2 and H2 transmis-sion is amongst others depending on the fuel used toproduce hydrogen and will be further studied here.The temporal dimension is related to the time-

frame considered for implementation. We distin-guish chains that may be implemented on arelatively short term (20102015) from long-termchains (42030), which differ in a number of aspects:

Extent of CCS: In the short term, the construc-tion of at most a few plants with CO2 capture canbe expected. In the longer term, a moresignicant contribution of CCS in the portfolioof CO2 emission reduction options is presumed.As a guiding line, we assume 20Mt CO2 will becaptured and stored annually by 2030.1

1The technical potential for CCS in 2030 has been estimated at

4060MtCO2 [18]. A recent update studying the potential of

various GHG reduction options revealed that the expected

contribution of CCS in 2020 is between 0 and 15Mt CO2avoided, depending on the emission reduction target set [19]. In

order to capture and store 20MtCO2 annually in the electricity

sector, the equivalent of at least six 500MWe state-of-the-art PC500plaension. The spatial dimension encompasses therastructural design to connect energy extraction,nversion, and end-use markets and CO2 sourcesth storage reservoirs. Therefore, insight into thensport costs of primary and secondary energyrriers and CO2 is required, for which quite ariation exists in literature [3,5,1216]. For aMWe NGCC, transmission costs of electricitynts needs to be installed with a chemical absorption unit.

dedicated CO2 pipelines from power plants to

pro

nosuelenoCCcelfor[28is olarof

3.2. CO2 sinks: geological reservoirs

Reservoirs suited for geological storage of CO2

ARTICLE IN PRESS

Interest rate 10% 515%

Capacity factor 85% 6090%

Economic lifetime for production

plants (yr)

20 1525

Energy+material costs

Steam coal (h/GJ)b 1.7 13

Natural gas for large industrial users

(h/GJ)b4.7 36

Natural gas for small industrial users

(h/GJ)b5.4 47

Natural gas for households (h/GJ)b 8.2 710

Electricity for large industrial users (h/MWh)b

( average Dutch electricity costs) 50 4070Electricity for small industrial users (h/

MWh)b80 70100

Electricity for households (h/MWh)b 90 80110

Gasoline (h/GJ)c 9 515

Coking coal (h/GJ)d 2.1

Steam (h/GJ)d 5

Iron ore (ne) (h/t)d 18

Iron ore (lump) (h/t)d 24

Iron ore pellets (h/t)d 36

Scrap (h/t)d 100

Electrodes (h/kg)d 2.4

GHG emission factorse

Natural gas (kg CO2-eq/GJ) 56 (57)

Coal (kg CO2-eq/GJ) 95

(103)

Gasoline (kg CO2-eq/GJ) 72 (87)

Average Dutch electricity 2020 (kg

CO2/MWh)

450 350500

aWe consider xed costs and emission factors to increase

transparency in outcome. The uncertainty in developments in fuel

costs is further accounted for in the sensitivity analysis.bForecasts for 2020 [23,24], including commodity, transmission

and distribution costs, excluding taxes and VAT. Electricity

prices are used for selling and buying.cRotterdam gasoline price (plus distribution costs, excluding

taxes and VAT) in 2003/2004, during which oil prices were

around 3035 $/bbl [25].d[26].eValues in parentheses include GHG emissions of fossil fuel

extraction, transport and distribution (and rening in the case of

gasoline) [2,27]. Note that indirect GHG emissions of coal and

natural gas use may change in the longer term as a consequence

of improved mining and transport practices, longer transport

distances and switch to LNG. The average emission factor of

Dutch electricity production is based on a scenario forecast given

in [23].

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 585Sm

rologies have been discussed in [1] and aremmarised in Tables 2 and 3. Decentralisedctricity and hydrogen production is generallyt considered due to the relatively high costs ofS at such scales. Advanced concepts using fuells and membranes, however, offer the potentiallow-cost CO2 capture at relatively small scales]. Decentralised electricity production with CCSf particular interest for the Netherlands given thege share of decentralised CHP units (circa 5GWetotal installed capacity near 20GWe in 2003) [29].all-scale H2 production with CCS might play aKduction technologies

ey techno-economic characteristics of the tech-3.1storage reservoir(s). In the long term, the CO2infrastructure is likely to be expanded to anetwork connecting various point sources andreservoirs. Similarly, we consider a hydrogennetwork to connect a large plant with variousend-use markets.

First, the electricity and hydrogen productiontechnologies are discussed, followed by the CO2storage reservoirs and the different hydrogen end-use markets. Combining the information on CO2sources, CO2 sinks and H2 end-use markets thenallows us to design the infrastructure required toconnect these elements. The economic assumptionsand emission factors applied in the analysis aregiven in Table 1.

. CO2 sources: electricity and hydrogen Technologies: We consider state-of-the-art tech-nologies for the short-term chains and moreadvanced technologies for the long-term chains[1]. Fuel cells for hydrogen end-use are expectedto become available in the longer term. In theshorter term, hydrogen could technically bedeployed for heat and power generation usingboilers and turbines and for traction in internalcombustion engines. However, forecasts of hy-drogen demand for energy purposes in20102015 appear too small to justify centralhydrogen production with CCS [4,2022].

Storage capacity: The capacity of natural gaselds becomes gradually available in the comingdecades with the depletion of these reservoirs.

Infrastructure: In the short term, we considerle in the transition towards a hydrogen economy.Table 1

Economic parameters and emission factors used in this analysis

Parametera Value Rangecan be classied into (nearly) depleted oil and gas

ARTICLE IN PRESS

Table 2

Key parameters of electricity production technologies with CO2 capture and compression to 110 bar

Net electric

efciency (%)

CO2 capture

efciency (%)

TCR (h/

kW)

O&M

(%)

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609586Feedstock Conversion

technology

CO2 capture

technology

Reference technologies (600MWe)

Bituminous coal PC

Natural gas NGCC

State-of-the-art technologies (600MWe)elds, deep saline aquifers and unminable coalseams.2 In the Netherlands, gas elds offer a largestorage potential of circa 10GtCO2, of which 7.5Gtis represented by the Groningen gas eld. Most ofthe reservoirs provide less than 30MtCO2 storage

Bituminous coal PC Post-comb (MEA)

IGCC Pre-comb (Selexol)

Natural gas NGCC Post-comb (MEA)

Advanced technologies (600MWe except SOFC-GT)

Bituminous coal Advanced PC Improved post-comb

(MEA)

Advanced IGCC Pre-comb (Selexol)

IG-Water Oxyfuel (ASU)

IG-SOFC-GT Various (membrane/

cat. combustor)

Natural gas Advanced NGCC Improved post-comb

(MEA)

MR-CC Pre-comb (HMCM)

CLC Oxyfuel (separate

combustion)

AZEP Oxyfuel (oxygen

membrane)

SOFC-GT (20 MWe) Oxyfuel (afterburner

Table 3

Key parameters of hydrogen production technologies with CO2 captur

Conversion

technology

CO2 capture

technology

Fuel+feed

input

(GJ=GJH2 )

Electricity

input

(GJe=GJH2

Advanced ATR

(1000MWH2)a

MDEA 1.28 0.03

Advanced CG

(1000MWH2)a

Selexol 1.35 0.05

MR (2MWH2)b Pd membrane 1.26 0.13

aIncluding H2 compression to 60 bar.bEstimated capacity to supply a future hydrogen refuelling station, i

2Storage in the deep ocean is not considered. Injection depths

are at least 1000m [30], which would require transport by ship

over large distances as the North Sea is not deep enough.44 1500 5.2

56 540 3.7capacity,3 although nearly 30 reservoirs provide astorage potential between 30 and 300MtCO2[31]. In the coming decade various small and

35 88 2080 5.8

3235 85 17702170 4.85.2

47 85 920 4.3

40 85 1520 6.5

43 85 1500 5

41 100 1530 3.7

50 90 1760 3.3

55 85 650 4.8

53 100 940 4

51 100 900 4

50 100 900 4

) 59 80 1530 3

e and compression to 110 bar

)

Conversion

efciency (%)

CO2 capture

efciency (%)

TCR

(h=kWH2)O&M

(%)

74 90 280 4

69 90 600 4

65 70 610 9

ncluding H2 compression to 480 bar [28].

3Reservoirs should preferably offer sufcient potential to store

the captured CO2 of one plant over its lifetime. A 500MWeNGCC equipped with amines captures circa 1.5MtCO2/yr,

corresponding to 30Mt to be stored in a 20-year lifetime. For a

PC plant of equal capacity, the capture rate is more than twice as

large.

demand proles and spatial distribution of futurehydrogen end-use markets. In this analysis, thetransport, residential and industrial sector areconsidered (see Textbox 1 for details).In the transport sector, hydrogen can be used as

an alternative for gasoline and diesel in conven-tional vehicles with internal combustion engines(ICEV) and fuel cell vehicles (FCV). Various well-to-wheel analyses indicate that use in ICEV has noadvantages from an energetic point of view [27,34],so use in ICEV is not further considered. AlthoughFCV prototypes have been introduced into the

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 587medium-sized gas elds become available, whereasthe Groningen gas eld is not expected to becomeavailable before 2050 [32]. The Dutch oil elds areless interesting for CO2 storage as they represent arelatively low storage potential. The interest inenhanced oil recovery (EOR) by injecting CO2 intooil elds in the North Sea [33] might offer short-term opportunities for CCS. Aquifers and coalseams are not that well studied and characterised ashydrocarbon structures, which causes a relativelylarge uncertainty in the storage potential. Storage incoal seams is still in an experimental phase andneeds considerable testing before it might be appliedcommercially. In this analysis, we will thereforefocus on gas elds and aquifer traps as potentialCO2 storage reservoirs.There is signicant variation in storage cost

estimates due to differences in CO2 injection rate,storage capacity, reservoir type, features (pressure,thickness, permeability and depth) and location(onshore-offshore). Relations between storage costs,

Table 4

CO2 storage costs as derived from [6]

Reservoir type Depth (m) Storage rate

(Mt/year)

Storage cost

(h/t CO2)

Aquifer onshore 10002500 12 3

24 2

Aquifer offshore 15002500 12 8

24 5

Gas eld onshore 25003500 12 3

24 2

Gas eld offshore 30004000 12 10

24 6injection rate, reservoir type and capacity areapplied to generalise storage costs (see Table 4).We distinguish storage rates between 1 and 2Mt/yr(typical capture rate for 600MWe NGCC or1000MWH2SMR) and between 2 and 4Mt/yr(typical capture rate 600MWe PC/IGCC or1000MWH2CG). In some specic source-sink com-binations, more than one trap may be required tostore all CO2 in a 20-year lifetime. Storage costsmay increase with a factor 2 when three gas eldsare required instead of one to store 1Mt CO2/yr [6].

3.3. Hydrogen end-use markets

In order to set up a basic design of hydrogeninfrastructure, we need an indication of potentials,market, signicant market penetration is notexpected before 2020 [4,2022].In the residential sector, hydrogen could gradu-

ally take over the function of natural gas and couldsimultaneously reduce electricity demand from thegrid. It could technically be deployed in modiedboilers. In the longer term, proton exchangemembrane fuel cells (PEMFC) enable electricityand heat production on the level of a house/building(block) at a somewhat longer term. Hydrogenapplication is primarily foreseen for new buildprojects where an entire new infrastructure needsto be constructed.4 In the short term, hydrogendemand in the residential sector is likely to be verylimited [20,22].The manufacturing industry is an another inter-

esting sector to realise large CO2 emission reduc-tions by means of CCS due to the large amount ofenergy consumed and its clustered nature. Hydro-gen could play an important role in the decarboni-sation of industrial energy use. Nowadays,hydrogen is produced and consumed in industryon a large scale, mainly for fuel upgrading anddesulphurisation at reneries, and ammonia pro-duction. In the future, the hydrogen market can befurther extended by replacing fossil fuels applied inindustrial CHP units, boilers and heaters.5 At

4The question is whether pure hydrogen can be distributed

through the local natural gas grid to existing households,

considering the lower energy density and material issues. Much

of the distribution grid consists of polyethylene, which might be

susceptible to hydrogen leakage [35].5Hydrogen can generally not be combusted in existing boilers

and turbo-machinery. However, industrial burners using hydro-

gen are available [36]. Gas turbines need to be retrotted to

enable combustion of hydrogen-rich fuels [37]. In our analysis, we

assume the thermal efciency of hydrogen combustion is

unchanged, although there could be improvements in efciency[38].

ARTICLE IN PRESS

Textbox 1Estimating hydrogen demand for the NetherlandsIn estimating hydrogen demand, we focus on the Randstad, a large urban conglomeration in

the Midwestern part of the Netherlands, where major cities (accommodating over 7 millionpeople) and industries are located.Transport sector: In 2003, the fuel consumption of road transportation in the Netherlands

approximated 460PJ, causing a CO2 emission of 34Mt [7]. Projections indicate that theseemissions may reach circa 45Mt by 2020 [23]. These emissions, along with local air pollutants,can be reduced by means of hydrogen. The hydrogen demand for the Randstad area in 2030 isestimated as a function of the number of FCVs (determined by penetration level, new car salesand average car lifetime), FCV fuel economy and travel distance. Considering a fleet of 45 millionlight duty vehicles and assuming 525% of the entire light duty vehicle fleet consists of FCVs in2030 based on penetration scenarios presented in [4,2022], between 0.2 and 1.25 million FCVswill be on the road by that time. Given the range in fuel economy as specified in Table 5, the totalhydrogen demand for the Randstad area in 2030 varies between 3 and 40PJ/yr. We consider anaverage scenario assuming 4.5 million cars on the road, a FCV fraction of 20% and a fuel economyof 0.92MJ/km, which would require an annual hydrogen supply of 13.3PJ. The size of theindividual refuelling stations, ranging from 0.1 to 8MWH2 [40,41], is a function of the number ofFCVs served by a refuelling station, fuel economy, travel distance and the refuelling pattern. In ouranalysis, we consider a uniform capacity of 2MWH2 (see Table 6). This would imply that circa 250refuelling stations need to be installed to cover the demand of 13.3PJ/yr, which corresponds to0.05 hydrogen refuelling station per km2 on average. The average gasoline refuelling stationdensity is estimated at 0.31 per km2 using national statistics on car density and number of carsserved per gasoline station [42]. This implies a coverage factor of 16%.Residential sector: In 2003, the Dutch residential sector consumed circa 360PJ natural gas,

producing 19Mt CO2 [7], and consumed 80PJ electricity [47]. Considering 1,475,000 newdwellings to be constructed between 2011 and 2030 [48], we assume 200,000 dwellingsconstructed in this period in the Randstad area will be equipped with hydrogen infrastructure.This corresponds to circa 2.5% of the entire stock, which is within the range of estimates inother scenario studies (0.515% [20,22]). Based on heat and electricity demand of modernhouseholds and performance figures for micro-fuel cells for co-generation [49], we estimatedhydrogen demand for hydrogen districts comprising 5,000 households. We consider aconfiguration comprising a (heat) load following 1 kWe PEMFC per household supplementedwith a hydrogen burner and a heat buffer (see Table 7). In order to cover the specified energydemand, 10.53MWh hydrogen is required and 1MWh electricity is exported to the grid [49].This would imply an annual hydrogen demand of 7.6 PJ to supply 200,000 households.The service sector, consuming circa 200PJ fuels causing a CO2 emission of 11Mt/yr [7],

would also be appropriate to deploy hydrogen, but is not further considered here due to lack ofdata on heat and electricity demand patterns required to design a micro-CHP system.Industrial sector: In 2003, total fuel consumption in the Dutch industrial sector was circa 430PJ,

corresponding to a CO2 emission of circa 27Mt (excluding joint-venture CHP units, refineries andfeedstock use) [7]. The chemical industry is the main energy consumer, emitting 12MtCO2/yr.Refineries emit another 11Mt/yr from combustion of natural gas and oil products [7]. The fueldemand in the industry is projected to increase slightly the coming two decades [23].Of the total industrial energy use, a significant share (circa 260PJ) is represented by natural gas

consumption in CHP units, boilers and heaters [47]. In 2003, the installed CHP capacity atrefineries was about 400MWe [52], of which the majority steam and gas turbines in the range of1050MWe [53]. In the chemical industry, circa 1800MWe is installed, of which a few very largecombined cycles [52]. The capacity of CHP is expected to grow the coming decades, especially atthe chemical industry. We assume a part of this capacity will be fired with hydrogen: the excess ofhydrogen (production 1000MW minus demand transport and residential sector) equals nearly6PJ/yr, which is sufficient to fuel four 20MWe gas turbines with a net electric efficiency of 40%.

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609588

industries where no such energy carriers are avail-

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 589Table 5

Car characteristics [4,4345]

Parameter FCV ICEVreneries, hydrogen production using coal ornatural gas for combustion purposes is not themost obvious option, as renery gas and heavy oilresidues are often available. However, there are

Fuel economy (MJ/km) 0.51.3 12.8

Retail costs of drive train (h/kWpeak)a 60 25

O&M costs (h/yr) 400 400

Lifetime (yr) 15 15

Annual travel distance (km) 16,000 16,000

Hydrogen stored onboard (kg) 4.76

aIncluding transmission, fuel storage tank and motor. ICEV

and FCV have capacity of 100 and 74 kW, respectively.

Table 6

Refuelling station features [4,40,43,46]

Parameter Value

Station capacity (kg/d) 1440

Load factor 0.85

Peak production (kg/h) 153

Dispenser capacity (kg/h) 73

Tank rell (%) 80

# cars served (cars/day) 320

# cars served at peak hour (cars/h) 40

Maximum # cars per dispenser (cars/h) 10

# dispensers 4

Table 7

Cost and performance data for conventional and hydrogen

households after [49]

Parameter Value

Heat demanda 5.3MWh/yr

Electricity demand 3.4MWh/yr

Zth 10 kWth heater (NG/H2) 95%Ze PEMFC 66%Zth PEMFC 34%Costs 10 kWth heater (NG) 720 h+2% O&M

Costs 1 kWe,nominal PEMFCb 1500 h+2.5% O&M

Costs 8.2 kWth heater (H2) 630 h+2% O&M

Costs 7.7 MJth heat buffer 340 h+2% O&M

aState-of-the-art heat demand (space and water) for houses in

new residential area. In the longer term, heat demand is expected

to decrease further [23], but this is not accounted for in our

calculations.bComplete system costs. The value used here is an average of

the USDOE target of 1500 $/kW [4], the EU target of 2000h/kW

for micro-CHP applications in 2020 [50] and a target of

10001500 $/kW for large volumes quoted in [51].able and natural gas is purchased for electricity andheat production. This might create some opportu-nities for decentralised CHP units red withhydrogen. A recent study on decarbonisationstrategies for a number of remotely located dis-tributed gas turbines concluded that it would bemore cost-effective to produce hydrogen centrallyand distribute it to each turbine than scrubbing CO2from the collected ue gasses [39].In the longer term, hydrogen might be applied as

reducing agent in the steel industry. As the steelindustry is a large CO2 source in the Netherlands,emitting nearly 6Mt in 2003 [7], we explore theopportunities for steel production using coal ornatural gas with CCS. Direct reduction (DR)technology enables reduction of iron oxide withhydrogen or syngas, omitting the use of cokes in theblast furnace route to produce pig iron. Table 8gives the basic characteristics of the reference route(blast furnace+basic oxygen furnace (BF+BOF)),and various DR routes (+ electric arc furnace(EAF)) with inherent CO2 removal. A more detaileddescription of these technologies and their energyuse can be found in [5456].

3.4. Infrastructural requirements and costs

3.4.1. Electricity transmission and distribution

Central production units are connected to thehigh-voltage grid transporting electricity to theregional distribution networks. Decentralised powerplants can be connected directly to the localdistribution network, thereby avoiding transmissionlosses and costs. Typical electricity losses occurringduring transmission and distribution to nal con-sumers are estimated at 8%. Only 1% of these lossesare caused by transmission, the other 7% bydistribution on the medium and low-voltage grid[58]. Transmission and distribution tariffs forhouseholds are 4060 h/MWh. Transmission tariffsfor very large (50MWe) industrial users areapproximately 5 h/MWh [59].6

3.4.2. CO2 transmission

For large-scale CO2 transport, pipelines aregenerally considered to be most suitable [14,16].Textbox 2 discusses the CO2 infrastructure for theNetherlands. Ship transport is an alternative when

6Transmission costs for a 500MWe NGCC have beenestimated at 15 h/MWh for a distance of 50500 km [3].

ARTICLE IN PRESS

7]

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609590Table 8

Main inputs and outputs of different steel production routes [26,5the storage reservoir is located offshore, as it avoidslarge investments of pipeline construction and itoffers exibility in CO2 purchasing and delivering.

Process BF+BOFa Circored+EAFb

Input

Fine ore (t) 1.2 1.4

Lump ore (t) 0.2

Pellets (t)

Scrap (t) 0.1 0.1

Coking coal (t) 0.4

Other coal (t) 0.1

Natural gas (GJ) 11.4

Electricity (GJe) 0 (0.77) 2.2

Electrodes 2.9

Output

Liquid steel (t) 1 1

Electricity (GJe)c 0.09 (0)

By-product (GJ) 0.27 CO2 emission (t) 1.7 (0.6) 0.4

CO2 capture (t) 0 (1.1) 0.6

Costs (h/t liquid steel)

Capital costs 32.0 (37.6) 28.0

O&M 48.2 (50.2) 49.5

Labour 27.1 17.9

aValues in parentheses refer to a conguration in which BF gas is cobIncluding CO2 capture and compression to 110 bar.cWe assume the excess BF, BOF and coke oven gas is converted into e

export electricity to the grid.

Textbox 2CO2 infrastructureFig. 3 clearly illustrates the mismatch in CO2 so

majority of the gas fields are located in the Northethe majority of CO2 sources are located in the Soidentified aquifer traps are distributed more homelectricity and hydrogen plants and reservoirs, tr200 km. For the short term, we assume CO2 is tranonshore gas fields located at 100 km from the plato various onshore and offshore gas fields and aqNetherlands shows that there are not enough l20Mt/yr in total. As the majority of reservoirs in tprovinces and offshore, a backbone seems advantFor the costs of transporting CO2 through the truntariff is based on transport costs of 20MtCO2/yrstations [3] and generalised costs of branches towe consider a fixed storage tariff of 2 h/t CO2 bas(representing the largest storage potential).The disadvantage is the high cost for liquefaction andbuffer storage, which makes ships less interesting forshort transport distances. The turning point of

Circofer+EAFb Coal-based Midrex+EAFb

1.4

0.9

0.6

0.1 0.1

0.4 0.4

0.1 0.1

2.4 3.1

2.9 2.9

1 1

0.5 0.5

1.0 1.0

27.1 34.3

52.4 46.6

19.4 17.9

mpressed, shifted and CO2 is captured and compressed.

lectricity to cover the internal electricity demand and eventually to

urces and gas fields in the Netherlands; thern part of the country and offshore, whereasuthern and Western part of the country. Theogeneously. Given the potential locations ofansport distances vary between nearly 0 andsported in dedicated pipelines and stored in

nt. In the long term, a network supplying CO2uifers is considered. A rough screening of thearge reservoirs nearby the sources to storehe Netherlands are clustered in the Northernageous, to which CO2 sources are connected.k line, a fixed tariff of 1.3 h/t CO2 is used. Thisover 200 km, including costs of the boosterthe individual storage reservoirs. In addition,ed on CO2 injection into onshore gas fields

ARTICLE IN PRESS

he N

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 591Fig. 3. Location of geological reservoirs and large CO2 sources in t

a trunk line as part of a CO2 network.transporting 6.2MtCO2/yr is circa 700km7; beyond

that point ship transport becomes economically moreattractive than transport by pipeline [60]. Transportcosts for 20,000 t/day (equivalent of capture rate of1000MWe PC using MEA) over a transport distanceof 1000km are circa 13 $/t CO2 [60].Pipeline diameters are calculated using steady-

state equations applied for incompressible owunder isothermal conditions, for which the pressuredrop is derived from the Fanning, or Darcy,equation [61]:

Dp 2f rv2 LD 32f rLQ

2

p2D5. (3)

The friction factor (f) is a function of the Reynoldsnumber (Re) and the roughness of the pipeline insidesurface, according to the Moody chart. TheReynolds number is a dimensionless quantity thatindicates the type of ow:

Re rvDm

. (4)

7The transport distance from Rotterdam to the Gullfaks oil

eld, one of the reservoirs studied for EOR, is circa 1000 km.etherlands (courtesy of TNO-NITG) and possible conguration ofFor laminar ows (Reo2000), the friction factor iscorrelated to the Reynolds number according to

f 16Re

. (5)

A turbulent ow (Re43000) can be either fully orpartially turbulent. Various equations exist tocalculate the friction factor for turbulent ows,each of them being specic for a certain pipeline andow regime. The following generic approximationof the friction factor is used, which results inrelatively conservative pipeline design [61]:

f 0:04R0:16e

, (6)

where Dp is the pressure drop (Pa); f the frictionfactor; v the average uid velocity (m/s); Q thevolumetric ow rate (Nm3/s); L the pipe length (m);D the (internal) pipeline diameter (m); r the uiddensity; and m the uid viscosity.Fluid properties were derived as a function of

temperature and pressure [62]. Onshore pipelinetemperature is set at 10 1C, equal to the soiltemperature. Offshore pipeline temperature is circa6 1C [16]. The pipeline diameter is calculated

studies large quantities of hydrogen to be trans-ported over relatively small distances, transport bydedicated pipelines seems most appropriate and isprojected to give the lowest cost [12,40,68]. In thelonger term, part of the natural gas grid maybecome obsolete when the use of hydrogen takesover as gaseous energy carrier. This may openpossibilities for the conversion of parts of theexisting infrastructure to transmit and distributehydrogen.

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609592iteratively until the chosen diameter meets thespecied maximum pressure drop, which is set at30 bar, assuming an inlet pressure of 110 bar and norecompression occurs along the main pipelinetrajectory. CO2 pipelines are operated at highpressure to guarantee high densities for optimalpipeline utilisation. In order to avoid phase transi-tion, pressures should be kept well above thesupercritical pressure (74 bar). According to Farris[63], the minimum pressure is between 83 and89 bar. The inlet pressure is generally set to over-come the pressure drop that occurs as a result offriction. Booster stations are required when a largepressure drop occurs, e.g. at long pipelines or hillyterrain. In order to optimise pipeline pressure anddiameter, the sum of pipeline and booster stationcapital and O&M costs plus electricity costs forpumping should be minimised. This theoreticaloptimum may differ substantially as there is a largevariation in capital costs of pipelines and boosterstations. The minimum distance between boosterstations specied in literature varies substantially,from circa 100250 km [3,16,64]. Recently, a 330 kmpipeline (diameter 0.30.36m) has been constructedtransporting circa 5000 tonnes CO2/day overlandwithout any booster station along the pipeline. Thepressure falls from 170 to 148 bar [65]. Offshore,where booster stations are very expensive, pipelinepressure and diameters are generally higher toovercome the pressure drop.

3.4.3. H2 transmission and distribution

Large quantities of hydrogen are generallytransported in gaseous form through pipelines.Liquid and gaseous hydrogen transport by meansof trucks may be an alternative for remote locationsor for small-scale applications [22,40]. As transitionoption, hydrogen could be added to the natural gasgrid up to 3% without any modications. Thehydrogen content might be increased to 25 vol%(peak concentration), which requires network up-grading and replacement of some end-use equip-ment [66]. The magnitude of CO2 abatement (versus100% natural gas) that can be achieved by blending25 vol% hydrogen produced by means of SMR withCCS is small (circa 4%) [66]. A current EU projectstudying the possibilities of hydrogen addition tothe natural gas grid shows that the margin of H2addition might be much lower [67], especially in thehigh pressure transmission network, which isdesigned for peak capacity. Reconsidering these

infrastructure options and the fact that this analysisIn order to design a H2 infrastructure, a hydrogendemand map should be set up, for which we useestimations of potential demand and possiblelocations of H2 production plants and end-usemarkets. The end-use markets vary in size andrequired hydrogen purity and pressure. We candistinguish high-purity (499.99%) hydrogen forapplication in PEMFCs from less pure, fuel grade(o95%) hydrogen suitable for combustion pur-poses. In this study, we simplify matters byconsidering a hydrogen grid transporting high-purity hydrogen, connecting a central hydrogenplant with refuelling stations, residential areas andindustrial CHP units (see Textbox 3). Hydrogen istransported from the production unit by means of ahigh-pressure transmission line, after which it entersa regional medium-pressure transmission ringline,similar to the natural gas transmission system. Atthe city gate stations (to which CHP units areconnected), hydrogen is transferred to the maindistribution grid, from where it continues its routeto refuelling stations and dwellings. The transportinlet pressure of the main transmission line is set at60 bar, allowing a pressure drop of 20 bar.8 In theregional transmission lines, pressure is between 40and 20 bar, after which it is further decreased to 10bar in the main distribution pipelines. At therefuelling stations, hydrogen is compressed to 480bar for buffer storage to enable cascade-lling,assuming hydrogen is stored onboard at 350 barin conventional pressurised tanks [43]. In the

8Typical transport pressures of hydrogen transmission are

between 10 and 30 bar [12], although pipelines have been

operated up to 100 bar [17,68]. We consider a value of 60 bar,

which is similar to that considered in recent studies on hydrogen

transmission [40,41,69]. The high-pressure natural gas transmis-

sion system in the Netherlands is operated at 5070 bar, after

which regional medium-pressure transmission pipelines supply

the gas to local gas distribution station at 2040 bar. The local

distribution grid is operated at maximally 8 bar, with the gas

nally reaching the consumer at a small overpressure of 250mbar.

ARTICLE IN PRESS

e totchatioingth oof Ulen

ics oarly

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 593Textbox 3H2 infrastructureFig. 4 illustrates a conceptual infrastructur

production unit to various markets. For the Dubetween 20 and 200 km, depending on the locthe regional transmission line is estimated us(R1 40 km, R2 20 km R3 5 km). The lengusing an algorithm developed by researchersare distributed over 10 conglomerations. Theindividual dwellings is estimated using statistdistribution network has a total length of nehousehold, on average.residential sector, hydrogen is delivered near atmo-spheric pressure.For pipeline design, we apply steady-state equa-

tions for an isothermal compressible ow [17]:

Qb 6:65Tb

pb

1

f

0:5p21 p22ZTGL

0:5D2:5, (7)

in which Qb is the ow rate at base (normal)conditions (Nm3/s); Tb the temperature at base(normal) conditions (293.15K); pb the pressure atbase (normal) conditions (101,325 Pa); f the frictionfactor; p1 the inlet pressure; p2 the outlet pressure; Zthe compressibility factor (1); T the average gastemperature; G the gas gravity ( 0.0696 for H2); L

Fig. 4. Conceptual hydrogen transmissideliver hydrogen from a central hydrogencontext, the main transmission line is variedn of the H2 production facility. The length ofan idealised representation of the Randstadf the main distribution system is estimatedC Davis [40], assuming 250 refuelling stationsgth of the low-pressure distribution grid ton the natural gas grid. The low-pressure gas85,000 km [66], corresponding to 12m perthe pipe length (m); and D the (internal) pipelinediameter (m).In high-pressure gas transmission lines with

moderate to high ows, the ow regime is eitherpartially or fully turbulent. Various specic owequations are applied in the gas industry to calculatethe friction factor [17]. The Panhandle A equation isapplied for partially turbulent ow conditions,medium to relatively large diameter pipelines withmoderate gas ow, operating under medium to highpressure. Panhandle B and AGA fully turbulentequations are appropriate for high-ow-rate, med-ium to large-diameter pipelines and high-pressuresystems. The pipeline variation is generally below20% using the different equations. We applied thegeneric Eq. (6) to calculate the friction factor,

on and distribution infrastructure.

ARTICLE IN PRESS

0.8

ion o

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609594resulting in average diameters in comparison tothose generated using AGA fully turbulent equation(large diameter) and Panhandle A/B (smallerdiameter). This generates sufciently reliable resultsgiven the generic nature of this study, the un-certainty in pipeline costs and the differencebetween internal and nominal pipe size. Averagevelocities calculated for the specied pressure drop,assuming no booster stations are installed along the

1.0

1.2

1.4

1.6

1.8

0.0

0.2

0.2

0.4

0.6

0.8

2.0

0.4 0.60

diameter (m)

investm

ent costs

(M

/k

m)

Fig. 5. Pipeline investment cost as a functpipeline, are generally between 5 and 20m/s, whichis reasonable for gas pipelines [17]. Given thespecied pressure drop, transporting large quanti-ties over small distances results in high averagevelocities that might exceed the erosional velocity[17]. Therefore, larger pipelines might be required,resulting in higher costs. Note that if the pipelinediameter is doubled, the pipeline capacity isincreased by a factor (2)2.5 5.66. This illustratesthe importance of considering possible futureexpansions (when setting up a large scale infra-structure) in the pipeline diameter selection [17].

3.4.4. Costs of CO2 and H2 infrastructure

Pipeline costs consist of material, labour, right ofway (ROW) and miscellaneous costs. In denselypopulated urban areas, pipeline construction costsare signicantly higher than for rural landscapes,due to safety requirements, higher ROW and thelarge number of infrastructural crossings. Construc-tion costs in urban terrain can be 20% higher [70]up to 10 times the costs for a pipeline installed inrural area [71]. The difference in landscape,population density, geographical location and steelprices explain the large variety in construction costsquoted in literature (see Fig. 5). Many studies usedata on natural gas pipelines. CO2 pipelines aregenerally more expensive than natural gas pipelinesbecause the former are operated at higher pressure,requiring a greater wall thickness. For H2 transmis-sion, embrittlement resistant steel pipelines are

1 1.2

f diameter [3,6,15,16,22,40,41,71,73,74,77].required, which are more expensive than naturalgas pipelines, generally by a factor 1.31.5[68,69,72]. We use the values of the IEA GHGstudy assuming a terrain factor of 1.2 [3], as itprovides a consistent set of equations suitable forboth CO2 and H2 pipelines for on- and offshoreconditions. Note that these costs are fairy low incomparison to other studies, which may result in anunderestimation of transport costs. The costs of H2distribution pipelines (and minor CO2 lines con-necting refuelling stations with larger trunk lines)are practically independent on pipeline diameter, asthey depend primarily on the labour costs asso-ciated with pipeline installation. We use values of500 h/m for the main distribution grid within thecity and 200 h/m for the low-pressure distributiongrid in new residential areas based on constructioncosts for small diameter pipelines [22,40,73]. Weassume pipeline O&M costs to be 2% of theinvestment costs, representing an average valueusing data from [3,7476], and an economic lifetimeof 25 years.

Figs. 6 and 7 show the dependence of CO2 and H2transmission costs on scale and distance. As CO2transmission costs decrease considerably with in-creasing ow rate, especially at larger distance, itmay be more advantageous to construct a networkinstead of dedicated pipelines. A network consists of

a large trunk line, which is connected to variousCO2 sources and sinks by smaller branches, similarto the existing natural gas grid. The advantage of alarge backbone structure has been investigated forEuropean conditions in [75]. The results indicatethat a backbone offers reduced transport cost in

ARTICLE IN PRESS

0

1

2

3

4

5

6

7

8

0 50 100 150 200 250 300

distance (km)

CO

2 t

ran

sm

issio

n c

osts

(

/t C

O2)

0.1 Mt/yr

1 Mt/yr

2 Mt/yr

4 Mt/yr

10 Mt/yr

20 Mt/yr

40 Mt/yr

2 Mt/yr Hendriks

Fig. 6. Onshore CO2 transmission cost as a function of distance and mass ow rates based on IEA GHG values for cultivated land [3],

excluding compression. We also included the transmission costs using the regression t from Hendriks et al. [77], which are generally in

good agreement with various industrial estimates given in [6,16,78].

0.5

1.0

1.5

2.0

2.5

3.0

1

istan

H2 t

ran

sm

issio

n c

osts

(

/GJ)

ow r

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 5950.0

0 50 100

d

Fig. 7. H2 transmission cost as a function of distance and energycompression. We also included the transmission costs for a 1000MWH50 200 250 300

ce (km)

ates based on IEA GHG values for cultivated land [3], excluding2unit using pipeline costs derived by Parker [73].

ARTICLE IN PRESS

1

istan

coal

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609596comparison to dedicated pipelines when the avail-ability of reservoirs is restricted to offshore hydro-carbon reservoirs, probably as these elds are

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 50 100

d

tra

nsm

issio

n c

osts

(

/GJ)

CO2 (SMR) IEA GHG

CO2 (CG) IEA GHG

H2 IEA GHG

CO2 (SMR) Hendriks

CO2 (CG) Hendriks

H2 Parker

Fig. 8. CO2 and H2 transmission cost of a 1000MWH2 unit using

GHG values [3] and regression ts derived from [73,77].clustered and remote from CO2 sources. Dedicatedpipelines would be more attractive when there aresufcient large structures nearby each source. Fig. 6also makes clear that decentralised CCS plants (20MWe SOFC-GT and 2 MWH2 MR, capturing 40and 2 kt/yr, respectively) to be connected to a CO2network need to be constructed close to a trunk linein order to keep transport costs reasonable.Now insights can be created into the trade-off

between the costs of CO2 and H2 transmission,which is essential for the optimisation of a hydrogenproduction plant location. Fig. 8 shows the trans-mission costs of CO2 and H2 for a 1000 MWH2plant red with coal (CG) or natural gas (SMR). Itillustrates that the costs to transport the CO2captured at a SMR unit are lower than H2transmission over the same distance. For a coal-red unit, H2 transport costs are only slightly lowerin comparison to CO2 transport costs. This ob-servation is in fairly good agreement with the resultsfrom Ogden [5]. However, the additional costs toconnect to the natural gas grid and the presence ofcoal terminals should be considered as well inchoosing the plant location. Given the substantialtransmission costs of natural gas [3], it may be moreattractive to produce hydrogen close to the naturalgas source as proposed by Mintz et al. [79], providedthat CO2 is captured and can be stored in nearby

50 200 250 300

ce (km)

(CG) and natural gas (SMR), (excluding compression) using IEAdepleted natural gas reservoirs.

3.4.5. Other H2 infrastructure requirements

Apart from pipelines, the H2 infrastructurerequirements consist of buffer storage, compressorsand dispensers, for which the main features aresummarised in Table 9.The work (head) required for compression can be

approximated as an isentropic process according tothe following formula:

W ZRT1kMk 1

p2p1

k1k 1

" #. (8)

The compressor power, also referred to as brakehorsepower or shaft power, can be derived from theisentropic head:

P W _mZisZm

, (9)

in whichW is the work or isentropic head (kJ/kg); Zthe compressibility factor (set at 1 for H2); R theuniversal gas constant (8.3145 J/molK); T1 thesuction temperature (K); k the specic heat ratio(Cp/Cv 1.41 for H2);M the molar mass (kg/kmol);

ARTICLE IN PRESS

Base

1.63.1

14 kW

634 k

1740

].

s, a s

oduct

large

or fo

gen p

n cap

repres

n hyd

droge

ed to

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 597p1 the suction pressure (Pa); p2 the dischargepressure (Pa); P the compressor power (kW); _mthe mass ow rate (kg/s); Zis the isentropic efciency(0.70.75) and Zm the mechanical efciency (0.99).The choice of a compressor depends mainly on the

Table 9

Costs hydrogen supply components

Componenta Investment costs

Storage at production unitc 10100Mh

Compressor at refuelling stationd 20 kh

Storage at refuelling statione 375 kh

Gaseous dispenserf 18.3 kh

aWe consider 15 years lifetime for all components.bAnnual costs as share of investment costs derived from [40,80cIn order to safeguard hydrogen supply in case of plant outage

Estimations for required capacity range from 0.5 to 2 days of pr

Although geological storage seems the least expensive option for a

for which detailed industry gures are available [40].dRepresenting costs to install a 9-stage reciprocating compress

including drive motor and intercooling [43].eStorage capacity at refuelling stations is required to store hydro

storage capacity required is estimated at 70% of the daily productio

hydrogen recovery with cascade storage/dispensing [43]. Costs are

[43]. Note that storage requirements may be somewhat lower whe

pipeline).fDerived from costs of compressed natural gas and prototype hy

We assume the refueling station is unattended. Another 20% is add

for miscellaneous costs and contingencies [43].ow rate and the differential pressure required. Forcompression of large, continuous quantities of H2 upto 60 bar, we consider multi-stage centrifugalcompressors with an isentropic efciency of 75%[3,17,81]. Compressing relatively small quantities ofH2 up to 480 bar requires a multi-stage reciprocatingcompressor. The overall efciency of a reciprocatingcompressor is circa 70% [43]. For multi-stagecompressors, the power requirements may be over-estimated when using Eq. (9) (for a single-stagecompressor), as the gas is cooled between thedifferent stages, which reduces the work requiredfor compression. Power requirements for multi-stagecompression with N stages can be calculated fromEq. (10) (derived from [82]). For H2 compression,typically a compression ratio of 2 is applied [43].

W ZRT1M

Nk

k 1p2p1

k1Nk 1

" #. (10)

3.5. CCS chains overview

Tables 10 and 11 give an overview of the chainsconsidered in this study.4. Results chain analysis

4.1. Electricity

Fig. 9 shows that the contribution of CO2

scale Scale factor O&M (%)b

041.63.105 kgH2 1 1

0.75 4

gH2 0.95 1

kgH2/day 1

torage facility is needed for large plants considered in this study.

ion [40,41]. We assume a buffer capacity equivalent to 0.5 days.

unit [12], we consider aboveground storage in pressurised tanks,

r compression of maximally 4.8 kg H2/h from 1 up to 480 bar,

roduced at night time and as back-up for demand variations. The

acity based on typical refueling station demand patterns and 58%

entative for large steel tanks suitable to store hydrogen at 480 bar

rogen is delivered by pipeline (as hydrogen could be stored in the

n dispenser adjusted for hydrogen specs and mass production [43].

the capital costs of compressor, storage and dispensers to accounttransport and storage to the overall electricity costsis relatively small and does not affect the rankingamong technologies. This is especially true for thelarge natural gas-red technologies, due to therelatively small quantities of CO2 transported andstored. Storage costs are low as we consideredrelatively large onshore gas elds. Offshore reser-voirs might deserve preference, although this goes atthe expense of higher transport and storage costs.Offshore transport costs for a 600 MWe PC unitover 100 km are estimated at 0.15 hct/kWh, circatwice the costs for onshore transport. Storagecosts for this unit increase to 0.5 hct/kWh for atypical offshore gas eld in comparison to circa0.2 hct/kWh for an onshore gas eld. In the longerterm, transport costs per km are expected todecrease slightly due to economy of scale of theCO2 network.The calculations show that COE of decentralised

SOFC units with CCS are higher than COE ofcentral units, as the advantage of omitting electricitytransmission does seem to be outweighed by theadditional costs of CO2 transport. However, SOFCefciency forecasts are higher than for advanced

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609598Table 10

Electricity and hydrogen chain denitions

Chain code Conversion+capture technology

Short-term electricity

ST-E1 PC+chem. absorption (600MWe)

ST-E2 IGCC+phys. absorption

(600MWe)

ST-E3 NGCC+chem. absorption

(600MWe)

Long-term electricity

LT-E1 Advanced PC+chem. Absorption

(600MWe)

LT-E2 Advanced IGCC+phys. absorption

(600MWe)

LT-E3 IG-Water (600MWe)

LT-E4 IG-SOFC-GT+HSD (600MWe)

LT-E5 Advanced NGCC+chem.

absorption (600MWe)

LT-E6 MR-CC (600MWe)

LT-E7 CLC (600MWe)

LT-E8 AZEP (600MWe)

LT-E9 SOFC-GT+afterburner (20MWe)

Long-term hydrogena

LT-H1-MOB/RES/CHP Advanced ATR+chem. absorptionBrayton-cycles with CO2 capture, which makes it amore safe technology with respect to energy pricevolatility.Fig. 10 illustrates why CO2 mitigation costs

should be used and interpreted with care, as theoutcome is strongly depending on the context, i.e.the choice of the reference system. When the fuelchoice is xed for a specic plant (e.g. due topresent/absent infrastructure or strategic choices),one should compare a coal-red power plantwithout CCS with a coal-red unit equipped withCCS. In this setting, CO2 mitigation costs of coal-red power plants with CCS are lower than for gas-red power plants with CCS. When replacing amothballed PC unit for a NGCC with CCS insteadof building a new PC unit, CO2 mitigation costs arelower than when installing a PC unit with CCS, upto gas prices around 6.5 h/GJ. Mitigation costs willbe relatively high for coal-red units with CCS whenthe reference is a NGCC. This is mainly explainedby the signicant CO2 emissions of a PC unit withCCS in comparison to an NGCC without CCS. CO2

(1000MWH2)

LT-H2-MOB/RES/CHP Advanced CG+phys. absorption

(1000MWH2)

LT-H3-STL SMR/CG+absorption+direct

reduction

LT-H4-MOB MR (2MWH2)

aMOB transport sector, RES residential sector,CHP industrial CHP, STL steel production.emissions of power production with CCS arebetween practically 0 kg CO2-eq/kWh for advancedgas-red units up to 0.2 kg CO2-eq/kWh for a PC,of which circa 60% are direct CO2 emissions and theremaining 40% are caused by GHG emissions ofcoal mining and transport.

4.2. Hydrogen

As illustrated by Fig. 11, the additional costs ofCO2 transport and storage are negligible for large-scale H2 production. As the additional costs ofcapture are modest as well [1], the carbon pricerequired to induce CCS at advanced large-scalehydrogen plants is relatively low: 1020 h/t (seeblack bars in Fig. 12).Hydrogen costs are to a large extent determined

by the infrastructure requirements. For centralplants supplying the transport and residentialsector, costs of hydrogen infrastructure are equalor higher than the costs to produce hydrogen. Itappears that the higher costs of H2 production andCO2 transport inherent in decentralised productionunits do not compensate for the omitted costs ofhydrogen infrastructure in centralised chains. Inaddition, H2 compression costs for (decentralised)MR units are signicantly higher in comparison tocentralised chains as H2 at MR is produced at 1 bar.From that perspective, MR units are more suitablefor markets where hydrogen is required at near-atmospheric pressure (e.g. in households). Anotherdisadvantageous factor is that natural gas andelectricity costs are signicantly higher for decen-tralised production. Assuming the refuelling sta-tions owner enforces identical price agreements fornatural gas and electricity as large industrial users,COH might be decreased to circa 18 h/GJ.Taking a closer look at H2 infrastructure makes

clear that the costs of distribution are a majorcontributor to overall production costs for thetransport and residential sector. A breakdown ofH2 infrastructure costs shows that especially thecosts of the low-pressure distribution grid are themain cause for the high costs to deliver hydrogen tothe front door of individual dwellings (80% oftransmission and distribution costs). For the trans-port sector, also the costs of the refuelling stationsare relatively high.Although CO2 emissions are greatly reduced with

CCS, the remaining emissions are substantial:1040 kgCO2-eq/GJ in comparison to 50120 kg

CO2 captured/GJ. This is part of the explanation

ARTIC

LEIN

PRESS

Table 11

Infrastructure assumptions

Chain code Electricity/H2 CO2

ST-E1-E3 Transmission line+distribution grid 100 km transmission line to onshore

gas eld

LT-E1-E8 Transmission line+distribution grid 10 km transmission line to CO2 trunk

line

LT-E9 Directly connected to distribution grid 50 km transmission line to CO2 trunk

line

LT-H1-

MOB

200 km transmission line+125km regional

transmission line+70km high-P distribution line

10 km transmission line to CO2 trunk

line

LT-H1-RES 200 km transmission line+125km regional

transmission line+70km high-P distribution line

+2500km low-P distribution line

10 km transmission line to CO2 trunk

line

LT-H1-CHP 200 km transmission line+125km regional

transmission line+10km high-P distribution line

10 km transmission line to CO2 trunk

line

LT-H2-

MOB/RES/

CHP

20 km transmission pipeline, further infrastructure is

identical to LT-H1

10km transmission line to CO2 trunk

line

LT-H3-STL 10km transmission line to CO2 trunk

line

LT-H4-

MOB

70km satellite line +125km

transmission line to CO2 trunk linea

aFor all refuelling stations.

K.Damen

etal./Progress

inEnerg

yandCombustio

nScien

ce33(2007)580609

599

ARTICLE IN PRESS

Fig. 10. CO2 mitigation costs of electricity production with CCS vis-a`-vis different reference systems, including indirect emissions of fuel

extraction and transport.

0

1

2

3

4

5

6

7

8

PC IGCC NGCC PC adv IGCC

adv

IG-water IG-

SOFC-

GT

NGCC

adv

MR-CC CLC AZEP SOFC-

GT

Co

st o

f e

lectr

icity (

ct/

/kW

h)

power plant capital power plant fuel power plant O&M CO2 transport CO2 storage electricity transmission

Fig. 9. Electricity production costs including CCS and electricity transmission. The dashed horizontal lines represent COE of PC and

NGCC without CO2 capture.

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609600

ARTICLE IN PRESS

0

5

10

15

20

25

MR-MOB ATRadv-MOB CGadv-MOB ATRadv-RES CGadv-RES ATRadv-CHP CGadv-CHP

Cost of hydro

gen (

/G

J)

CO2 storage

CO2 transport

H2 refueling station

H2 transmission + distribution

Central H2 storage

H2 plant O&M

H2 plant electricity use/export

H2 plant fuel+feed

H2 plant capital

Fig. 11. Hydrogen production costs including CCS and H2 infrastructure for delivery to various end-users (MOB transport sector,RES residential sector, CHP industrial CHP). Costs of H2 compression in centralised production have been accounted in the H2 plant(up to 60 bar) and refuelling station (up to 480 bar). In membrane reformers, H2 compression is accounted for in the production.

0

50

100

150

200

250

ATRadv-MOB CGadv-MOB MR-MOB ATRadv-RES CGadv-RES ATRadv-CHP CGadv-CHP DR-STL

CO

2 m

itig

ation c

osts

(/t C

O2)

versus identical plant versus gasoline/natural gas/cokes

Fig. 12. CO2 mitigation costs of hydrogen application vis-a`-vis different reference systems, including indirect emissions of fuel extraction

and transport. MOB transport sector, RES residential sector, CHP industrial CHP, DR-STL steel production by means ofdirect reduction.

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 601

ARTICLE IN PRESS

ed+E

n the

K. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609602why CO2 mitigation costs, depicted in Fig. 12, are

0

50

100

150

200

250

BF+BOF BF+BOF+CCS Circor

Co

st o

f ste

el (

/tL

S)

Fig. 13. Steel production costs for a tonne of liquid steel, based o

routes (all with CCS).relatively high for sectors where natural gas (emis-sion factor of 57 kgCO2-eq/GJ) is the reference fuel.For the residential area, the other part of the story isobviously formed by the high costs of H2 supplyand PEMFC, which is not outweighed by the higherefciency of the PEMFC. Mitigation costs in thetransport sector are mainly determined by therelatively high costs of the fuel cell. This isillustrated by the costs per driven kilometre:from the 0.06 h/km for an FCV (versus 0.04 h/kmfor ICEV), circa 20% can be attributed to H2 costs,the rest being costs of the drive train and storagetank.In the steel industry, replacing cokes in the

traditional BF route for hydrogen or syngas inDR routes results in CO2 mitigation costs of2040 h/t CO2. Costs of CO2 capture from BFgas in the traditional BF route are estimated at15 h/t CO2, comparable to the value calculated in[83]. Fig. 13 shows the production costs of varioussteel production technologies. Circored technologyresults in relatively high production costs due to thehigh fuel costs. For coal-based Midrex, the explana-tion lies mainly in higher expenses in iron ore. Thecosts of Circofer and the traditional BF route withCO2 capture from BF gas are relatively close.Although the additional costs due to CCS is very

AF Circofer+EAF coal-based Midrex+EAF

labour

O&M

electrodes

oxygen

electricity

natural gas

coal

scrap

iron ore

capital costs

blast furnace route (with and without CCS) and direct reductionsmall for Circofer (as CO2 is already separated), theadditional indirect emissions due to electricityrequirements of EAF make that mitigation costs(versus BF without capture) appear slightly higherin comparison to capture of CO2 from BF gas in theclassical route. So the primary reason to switch toDR would not be CO2 emission reduction, but theomission of the polluting coke oven plant.

4.3. Sensitivity analysis

Figs. 14 and 15 show the results of sensitivityanalyses performed on several promising or likelytechnologies that may be implemented, encompass-ing the complete spectrum of technology (combus-tion, fuel cells, gasication), fuel (natural gas versuscoal) and scale (central versus decentralised units).Apart from the basic economic variables, for whichthe variation is specied in Table 1, the impact ofuncertainty in plant capital costs (730% for state-of-the-art technologies and 50% for advancedtechnologies such as SOFC and gasication, whichare not mature yet) is assessed. Note that the fuelprice and capacity factor are correlated to eachother, which may result in even a wider range asshown in Fig. 14, especially for NGCC. A power

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 603capacity factor

TCR plantplant dispatch model is required to study the jointimpact of these parameters in more detail, which isbeyond the scope of this study.

4 65

fuel price

interest rate

fuel price

interest rate

capacity factor

TCR plant

fuel price

interest rate

capacity factor

TCR plant

fuel price

interest rate

capacity factor

TCR plant

CO

Fig. 14. Sensitivity analyses of various electricity production chains with

the variation in parameters is given in Table 1.

10 12 14 16

C

MR-MOB

(LT-H4)

fuel price

interest rate

capacity factor

TCR plant

fuel price

interest rate

capacity factor

TCR plant

fuel price

interest rate

capacity factor

TCR plant

Fig. 15. Sensitivity analyses of various hydrogen production chains with

Table 10) are given in parentheses and the variation in parameters is giv

the increased CO2 and H2 transmission costs when pipelines are underSOFC-GTThe coal-red power plants and the SOFC-GTare, not surprisingly, more sensitive to variation incapital costs, interest rate and load factor, whereas

7 98 10

E (ct/kWh)

PC

(ST-E1)

NGCC

(ST-E3)

IGCC adv

(LT-E2)

(LT-E9)

CCS. The chain codes (see Table 10) are given in parentheses and

18 20 22 24 26

OH (/GJ)

ATR-MOB

(LT-H1)

CG-MOB

(LT-H2)

CCS supplying the transport sector (MOB). The chain codes (see

en in Table 1. The impact of variation in capacity factor includes

utilised.

COE of NGCC is affected especially by the fuelprice. Despite the coal price increase in recent years,prices are likely to moderate the coming decades

has a major impact. The costs of this grid depend onlocal conditions and might be a factor 2.5 more

uncertainties exists. Although current oil prices

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609604[84]. Gas prices are subject to various trends such asincreased gas-to-gas competition and shift to moreglobal markets. The exact impact in the long term ishard to establish. According to the IEA WorldEnergy Outlook, gas prices in Europe may drop thesecond half of this decade and then gradually rise toaround 4 h/GJ in 2030 (considering oil prices of2535 $/bbl) [84]. The natural gas breakeven price atwhich COE of PC with CCS and NGCC with CCSare equal, assuming a coal price of 1.7 h/GJ andboth are operated at equal capacity factor of 85%,lies between 66.8 h/GJ for CO2 prices between 0and 100 h/t CO2. Even if the capacity factor forNGCC with CCS is decreased to 60% at the defaultgas price (4.7 h/GJ), COE would still be slightlylower in comparison to PC with CCS. IGCC withCCS9 appears more competitive than PC with CCS,resulting in somewhat lower break-even natural gasprices.The uncertainty in CO2 transport costs hardly has

any effect on the COE. Assuming structurallyhigher pipeline costs (up to a factor 2 in comparisonto the base case) and a maximum transport lengthof 200 km, COE of a PC unit are increased with amere 0.3 hct/kWh. The uncertainty with respect tostorage costs is mainly related to reservoir char-acteristics. In the most pessimistic case, assumingstorage costs of 10 h/t CO2 for deployment ofseveral smaller, deep offshore reservoirs with lowpermeability, COE for a PC are increased withnearly 1 hct/kWh.As illustrated by Fig. 15, centralised chains with

large-scale hydrogen production and CCS offercheaper hydrogen than decentralised production bymeans of MR with CCS, for the range in parametersstudied here. Large-scale ATR units with CCS arecompetitive with CG units with CCS at gas prices of33.5 h/GJ, which is considerably lower than thebreakeven price of their electricity counterparts(NGCC with CCS versus IGCC with CCS). COHare strongly dependent on the interest rate due tothe capital-intensive infrastructure required tosupply hydrogen to refuelling stations and dwell-ings. The uncertainty in costs of the distributionnetwork, especially the residential hydrogen grid,

9Note that the IGCC in Fig. 14 represents an advanced

concept, which cannot be compared to state-of-the-art PC andNGCC.(6570 $/bbl) are historically high, it is unclearwhether such high prices will retain the comingdecades. In 2004, the World Energy Outlookforecasted oil prices between 25 and 35 $/bbl upto 2030 [84]. If investments in oil infrastructure inthe Middle East are not signicantly increased, oilprices might be as high as 50 $/bbl (nearly 9 $/GJ) inthe long term [85]. Assuming gasoline costs at therefuelling station of 15 h/GJ instead of the base casevalue of 9 h/GJ, CO2 mitigation costs woulddecrease to circa 8090 h/t CO2 for centralisedhydrogen chains. Mitigation costs can vary withroughly 100 h/t CO2 for the range of PEMFCefciencies quoted in literature [4,4345]. Theuncertainty in fuel cell costs has an even biggerimpact on the performance of hydrogen. If weconsider vehicle retail prices projected for 2010 asquoted in [34], assuming fuel cell costs of 100 h/kW,mitigation costs would be more than 700 h/t CO2.Note that a signicant improvement still has to bemade to achieve a level of 100 h/kW given currentfuel cell cost (produced in low quantities)[4,50,86].10 Although PEMFC costs for stationarypurposes are higher than mobile PEMFC costs, asthe former requires a longer lifetime and thecapacity is lower, it is unclear what cost differencecan be expected. A lower range of 500 h/kW forstationary fuel cell systems [87] would decrease CO2mitigation costs with 2530%. Apparently, theimpact of PEMFC costs on CO2 mitigation costsfor stationary applications is less severe than formobile applications; energy costs are a moreimportant factor.

5. Discussion and conclusion

A chain analysis has been performed for promis-ing CCS options, incorporating a wide varietyof technologies, infrastructural settings, hydrogen

10The cost of current fuel cell systems for mobile applications is

estimated at 250300 $/kW at a production volume fof 500,000expensive than our default value, resulting in COHaround 33 h/GJ.CO2 mitigation costs of hydrogen application as

presented in Fig. 12 are to a large extent determinedby the assumptions in fuel prices and end-useconversion efciency and costs, for which largeunits per year [51].

ARTICLE IN PRESSK. Damen et al. / Progress in Energy and Combustion Science 33 (2007) 580609 605end-use markets and reference systems to studyvarious CCS congurations under specic conditionsand assumptions. The results indicate that the overallimpact of CCS on CO2 emissions and electricityproduction costs is signicant. CO2 emissions perkWh are reduced with 75% up to practically 100%with respect to an equivalent power plant withoutCCS. Note that the overall chain emission of a PCunit with CCS is still more than half the emission of aNGCC without capture. Electricity costs are increasedwith 2550% and 1535% for state-of-the-art andadvanced concepts, respectively. The range of studiespresented in the IPCC special report on CCS [88]shows a higher upper boundary. Some studies forecasta COE increase of 90% for PC or NGCC with post-combustion capture, transport and storage, which canmainly be explained by more conservative estimationsfor capital costs and/or energy use of CO2 capture.The cost increase is dominated by CO2 capture

and compression, representing 6590% of theadditional costs due to CCS. Only 211% of totalproduction costs for central units can be attributedto CO2 transport and storage. According to Rubinet al. [89], the share of CO2 transport over 160 kmand storage into an aquifer is 510% of the COE.For typical Dutch conditions (i.e. a maximumtransport distance of 200 km), transport costs for alarge power plant are estimated at 17 h/t CO2 (theupper range representing offshore conditions),corresponding to 0.050.3 hct/kWh. Although CO2transport costs are low on a kWh basis, the totalinvestment costs to construct a CO2 network arelarge. Considering the length of the high-pressuretransmission grid for natural gas, we estimate theinvestments costs of a CO2 network at a few billionEuros. The extent of the CO2 network could beminimised when power plants with CCS areconstructed in the Northern provinces, where thebulk of storage capacity is located. Only a limitedcapacity can be t into the current network;investments may be required to strengthen theexisting electricity grid to carry the electrons fromthe North down to the major electricity demandcentres. More in-depth research is required to studythis trade-off for optimal decision making on newplant locations.The contribution of CO2 transport is signicantly

higher for decentralised power generation withCCS. The general consensus is that decentralisedtechnologies are not suited for CCS for this reasonand due to concerns over economies of scale in

capture technology. However, specic decentralisedtechnologies that enable inherently low-cost CO2capture such as SOFC may produce electricity withCCS at reasonable costs, provided that the plantcan be connected to a CO2 trunk line locatednearby. Another option would be to install the planton top of a CO2 storage reservoir. Note that storagecosts will increase when CO2 from a single 20MWedistributed unit is injected into a reservoir that ischaracterised by a permeability that allows formuch higher injection rates.CO2 storage in onshore and offshore gas elds or

aquifers costs 110 h/t CO2 at injection rates of atleast 1Mt/yr, corresponding to 0.050.9 hct/kWh. Inthe Netherlands, natural gas elds are the primarytarget reservoirs for CO2 storage. Whether enoughsuitable gas reservoirs (excluding Groningen) will beavailable to store large quantities in the longer termis open to question, considering future UGS demandand risk proles of reservoirs. Possibly, aquifers andcoal seams also need to be exploited.The prospects and competition between different

CCS technologies are depending on several uncer-tain parameters such as fuel prices, capacity factorand capital costs, for which a sensitivity analysiswas performed. COE for NGCC with CCS arelower than for PC with CCS over the studied rangein parameters (gas prices up to 6 h/GJ). Onlycombinations of high gas prices and low capacityfactors favour PC with CCS. The breakeven naturalgas price for IGCC with CCS is somewhat lower incomparison to PC with CCS. It should be notedthat NGCC without CCS is the most competitivetechnology in a world where natural gas pricesprevail below the range from 5.57 h/GJ and CO2prices of 055 h/t CO2. Kreutz and Williams [90]came to comparable ndings, although in theiranalysis the perspectives for IGCC are somewhatmore optimistic and NGCC with CCS enters theplayeld at much higher CO2 prices.The impact of CCS on costs of centralised

hydrogen production and supply is not as sub-stantial as for electricity production; they addanother 12 h/GJ for transport distances up to200 km. Of the total costs of centralised H2production (between 8 and 21 h/GJ for base caseassumptions on fuel price etc.), maximally 0.5 and0.7 h/GJ are added by CO2 transport and storage,respectively. Hence the carbon price required toequip hydrogen units with CCS is rather low;typically in the order of 1020 h/t CO2. Mitigationcosts for replacing natural gas or gasoline with

hydrogen produced with CC