Aville online t www.sciencedirect.com ScienceDirect Energy Procedi 76 (2015 ) 573 581 Europen Geosciences Union Generl Assemly 2015, EGU Division Energy, Resources & Environment, ERE Stle lrge-scle CO 2 storge in defince of n energy system sed on renewle energy - Modelling the impct of vrying CO 2 injection rtes on reservoir ehvior Andres Bnnch *, René Huer, Mrtin Streiel, Michel Kühn, Gerrd Stienstr c ESK GmH, Hlsrücker Strße 34, Freierg, 09599, Germny GFZ Germn Reserch Centre for Geosciences, Telegrfenerg, Potsdm, 14473, Germny c DNV GL, Utrechtseweg 310, Arnhem, 6812 AR, The Netherlnds Astrct The IPCC Report 2014 strengthens the need for CO 2 storge s prt of climte chnge mitigtion options. The further expnsion of electricity genertion y solr nd wind nd its preferentil usge in Germny is leding to strong fluctutions in the CO 2 output from former se lod col fired power plnts. This study tkes look t the fesiility of lrge scle industril CO 2 injection into sline quifer structure with the min focus on vrying injection rtes. By mens of simultion the influence of the most importnt prmeters is nlyzed. 2015 The Authors. Pulished y Elsevier Ltd. This is n open ccess rticle under the CC BY-NC-ND license 2015 The Authors. Pulished y Elsevier Ltd. (http://cretivecommons.org/licenses/y-nc-nd/4.0/). Peer-review Peer-review under under responsiility responsiility of the of GFZ the GFZ Germn Germn Reserch Reserch Centre Centre for Geosciences for Geosciences. Keywords: CO 2 storge; numericl simultion; vrying injection rtes; industril size project 1. Introduction In 2014, electricl power genertion in Germny from renewle energy sources ccounted for 26.2% [1]. Wind power hd shre of 9.1% nd solr power of 5.7%. A totl of 11.4% cme from other renewle sources like * Corresponding uthor. Tel.: +49 3731 365418; fx: +49 3731 365-432. E-mil ddress: ndres.nnch@rwe.com 1876-6102 2015 The Authors. Pulished y Elsevier Ltd. This is n open ccess rticle under the CC BY-NC-ND license (http://cretivecommons.org/licenses/y-nc-nd/4.0/). Peer-review under responsiility of the GFZ Germn Reserch Centre for Geosciences doi:10.1016/j.egypro.2015.07.874
574 Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 hydropower or iomss. The growing shre of wind nd solr power in the energy mix mkes the wether ecoming the controlling fctor for electricity production for fossil fuel power plnts. Despite the current legl sitution in Germny, tht gives ech federl stte the possiility to n CO 2 sequestrtion on its territory nd the generl non-cceptnce of Cron Cpture nd Storge (CCS) y lrge prts of the Germn popultion, there seems to e no other solution thn storing CO 2 in the deep underground, if the mitious climte gols y the Germn government until 2020 (20% less CO 2 output compred to 1990), 2030 (40% less CO 2 output compred to 1990), nd 2050 (self-sufficient y renewle energy) wnt to e chieved, with col still plying sustntil role in the energy mix nd lso to void emissions from industril sources. The Federl Institute for Geosciences nd Nturl Resources (BGR) hs developed specil ctlog nd mp-sed informtion system contining ll relevnt storge horizons suitle for CO 2 storge s well s cp-rock horizons, necessry to hold the injected CO 2 within the structures [2]. According to BGR there is totl storge cpcity etween 6.3 nd 12.8 illion t for CO 2 in sline quifers in Germny [3]. Totl CO 2 emission ccording to the Federl Environmentl Agency (UBA) mounted to 822 million t in 2012 [4]. Annully, out 375 million t of CO 2 come from lrge point sources like power plnts nd industril fcilities. The figures suggest tht the storge cpcity in Germny would e sufficient for severl decdes pplying CCS preferly t some selected lrge power plnts. The study presented here simultes the geologicl storge for cluster of col fired power plnts, emitting in totl 11.7 million t of CO 2 per yer. The figure is derived from study on the CO 2 trnsport structure in Germny y 2050 [5]. The CO 2 emission of this cluster is hevily influenced y renewle energy sources tht hve voltile production pttern (wind nd sun) nd ccounts s well for the demnd driven chnges in conventionl power genertion. In this study it ws nticipted tht the fluctuting CO 2 output is injected into one on-shore geologicl structure lrge enough to hold the entire CO 2 relesed during time intervl of 10 yers, i.e. 117 million t of CO 2. A simultion model for the lrge structure ws set up to test the generl fesiility of the storge opertion. Strting from se-cse, different prmeters nd well ptterns were modified within relistic rnges tht limit the storge of CO 2. Every simultion run is providing n injection potentil tht is different nd chnging over the opertionl lifetime of the storge. A model of structure used for commercil nturl gs storge which lies in the vicinity of the lrge structure is used for plusiility checks nd comprison of simultion results in the first yer of CO 2 storge opertion. The model hs the dvntge, tht it is qulified for simultion eing history-mtched over storge period of more thn 35 yers. However, its volume cpcity, is y fr too smll for storing the mount of CO 2 coming from the considered cluster. 2. Model description The potentil of suitle structures for CO 2 sequestrtion is high in northern Germny. For this study the structures illustrted in Fig. 1 were used. The smll model is the existing gs storge, where the Volpriehusen sndstone (Middle Bunter) with n verge thickness of 16 m is used s storge horizon. This is lso used s storge horizon of the lrge structure. The geologicl model (structure nd property distriution) is sed on logging dt from 17 wells nd time-lps seismic interprettion. The structurl mp for the lrge model ws tken from the se Upper Bunter mp of the Geotectonicl Atls of Northwestern Germny [6] nd shifted y -360 m to dpt to the known top of the Volpriehusen sndstone. The tops of oth structures re t distnce of pprox. 40 km. As reservoir properties for the lrge structure re not exctly known it ws ssumed tht reservoir thickness nd property distriution re similr to the Volpriehusen sndstone in the nery gs storge. Therefore in generic modelling pproch property modelling workflow for the lrge model using the well logs from the smll model ws pplied. The structurl models of the smll nd the lrge structure re shown in Fig. 2.
Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 575 Fig. 1. Geologicl structures for lrge nd smll model. Fig. 2. () Lrge generic reservoir model; () Smll qulified reservoir model.
576 Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 3. Vrying CO 2 injection rtes The injection rtes come from study on the CO 2 trnsport structure in Germny y 2050 [5]. The study is sed on rel dt CO 2 output in Germny considering the yer 2013 nd presents week CO 2 output profile (Fig. 3) from power plnts with the strongest vrition (peks) ove se lod. 90,000 80,000 CO2 Rte 90,000 80,000 CO2 Rte 70,000 70,000 60,000 60,000 CO 2 Rte [t/d] 50,000 40,000 30,000 20,000 10,000 vrying injection se lod CO 2 Rte [t/d] 50,000 40,000 30,000 20,000 10,000 0 0 1 2 3 4 5 6 7 Time [d] 0 0 50 100 150 200 250 300 350 Time [d] Fig. 3. Vrying CO 2 injection rtes; () week schedule; () yer schedule. This week s profile represents the worst cse scenrio in terms of fluctutions [5] nd cn hppen ny time during the yer nd therefore will e pplied throughout the entire yer to generte mximum stress for the storge reservoir. As there is lso sesonl vrition oserved in electricity production y wind nd solr in 2013, sesonl vrition ws dded to the schedule, hving the highest CO 2 output in Ferury nd the lowest during summer nd utumn. 4. First-up pprisl clcultions With the ojective to test the volume cpcity nd injection rtes of the lrge reservoir, simultion runs were relized using 10 injection wells nd pplying constnt injection rte. The runs showed tht the volume cpcity would e sufficient to hold pprox. 120 million t of CO 2. However, due to the given flow dynmics, especilly the limited permeility cpcity, only one third of the required verge constnt rte ws possile to e injected, no mtter how mny more injection wells were pplied in the top of the structure. With this limittion in flow dynmics se-cse for simultions with vrying injection rtes nd prmeter vritions ws developed considering only 25% (pprox. 3 million t/) of the initilly required injection rte. As for the se-cse with constnt injection rte, four similr structures like the selected lrge structure would e needed to ccommodte the yerly CO 2 output of lmost 12 million t. 5. Configurtion of simultion runs In totl 8 simultion ctegories were set up, ech ctegory specifying the rnge of one prmeter. The prmeters, shown in Tle 1, re very diverse considering reservoir properties, structurl fetures, numer nd position of well loctions s well s quifer size. The se-cse ws given reservoir prmeters tht re expected to e fvorle for lrge scle gs injection. The prmeter rnge is within relistic limits.
Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 577 Tle 1. Vrition ctegories nd prmeters (lrge model). Vrition ctegory se-cse configurtion prmeter vrition from to 1) Permeility cpcity k*h= 4000 md 1000 md 4000 md 2) Distnce of wells d= 2000 m 1000 m 3000 m 3) Numer of top structure wells n= 10 4 10 4) Top wells vs. wells on structurl flnk depth= -2360 m -2360 m -3500 m 5) Well ore injection (perfortion intervl) entire reservoir section upper reservoir section lower reservoir section 6) Impct of hysteresis no hysteresis no hysteresis hysteresis 7) Impct of quifer size V Aqu= 500 illion m³ 5 illion m³ 500 illion m³ 8) Impct of ordering fults no oundry no oundry two oundries The mximum injection pressure for the top wells is set to 1.4 times the initil pressure llowing pressuriztion of the reservoir of 40%. This is done in ccordnce to the storge opertion of the gs storge (smll model). The mximum injection pressure for the wells positioned t the structurl flnk is set s such tht the reservoir pressure in the top of the structure is not reching higher vlues thn 40% ove the initil pressure in order to remin well elow the formtion rekdown pressure in this geologicl setting. 6. Evlution of simultion results The model results re evluted in terms of pressure rnge utiliztion. Fig. 4 shows the ottom hole pressure ehvior of n injection well for the se-cse (lue line) rnging from the initil pressure to the mximum injection pressure. Fig. 4. Pressure rnge utiliztion.
578 Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 After the mximum injection pressure is reched the injection rte is utomticlly reduced in order not to violte the mximum pressure limit. With respect to the pressure response division of the timeline in first yer, 2-10 yers nd 11-20 yers is pproprite. During first yer of injection the ville pressure rnge is used up to 100% resulting in sustntil injection rte reduction. In the yers 2-10 the pressure rnge is used up to 84% nd in the yers 11-20 only up to 75%. For comprison, the pressure response from n verged constnt injection profile is shown s green line in Fig. 4. As for the se-cse with the vrile injection profile, the injection pressure rnge ove the initil pressure is etween 25% (first yer) nd 10% (lter yers) higher compred to the scenrio with n verged constnt injection rte. 1.6 1.4 4000 mdm 2000 mdm 1000 mdm yer 1 yers 2-10 yers 11-20 1.6 1.4 n=10 n=9 n=8 n=7 n=6 n=5 n=4 yer 1 yers 2-10 yers 11-20 1.2 1.2 rte profile multiplying fctor rte profile multiplying fctor 1 1-1 1-1 1 1-3 1-3 1-3c 1-3d 1-3e 1-3f Fig. 4. () Impct of permeility cpcity on injectivity; () Impct of numer of wells on injectivity. There is dditionl potentil for higher injection rtes in yers 2-20 with respect to the se cse. For ech simultion run either the rte reduction within ech time intervl or the higher injection rte potentil reltive to the se cse is determined. In the first vrition ctegory the impct of permeility cpcity (k*h) on the reliztion of the vrile injection rte ws nlyzed (Fig. 4). As for the se-cse, using the full pressure rnge etween initil nd mximum reservoir pressure stedy increse in the injection rte cn e identified. On the y-xis in Fig. 4 the multiplying fctor 1 denotes the 25% initil rte schedule. In first yer of the se-cse scenrio for exmple, pplying the multiplying fctor of, only 15% of the initil rte schedule cn e relized. In the yers 2-10 nd 11-20 the injection rte cn e incresed, ecuse fter the first yer the pressure rnge in the 25% rte schedule scenrio is not entirely used. At lower permeility cpcity scenrios (runs 1-1 nd 1-1) we still see the generl tendency to higher injection rtes with progressive storge filling, however, compred to the se-cse the injection rtes decrese significntly. A cut of 50% of the permeility cpcity is roughly equl to injection rte reduction of 50%. At lower permeility cpcity no further increse in injection rte cn e relized in lter yers. Vrition ctegory 3 is deling with the numer of top structure wells (Fig. 4). Strting with the se-cse the numer of wells is continuously reduced from 10 to 4 wells. In first yer we see strong correltion in numer of wells nd injection rte. This is ecuse the effective pressure rdius of ech well is comprtively smll nd wells do not much interfere with ech other or the reservoir limit. With respect to the defined time intervls we cn see tht the sitution, i.e. the increse in injection rte from first yer to yers 2-10 nd yers 11-20, is not significntly chnging compred to the se-cse. We do however see tht the injection rtes in the yers 2-20 fll when less thn 8 wells re pplied. The use of more thn 8 wells on the other hnd will not result in higher injection rtes, s we encounter restrictions in the flow dynmics from the top of the structure to the surrounding reservoir.
Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 579 In vrition ctegory 4 the impct of the well loction (top vs. flnk) on the reliztion of the injection rte ws nlyzed (Fig. 5). Wheres in the se-cse ll wells re plced in tringulr pttern t equl distnce of 2000 m from ech other in the top of the structure, the wells in the cses 1-4 nd 1-4 re plced t isoths of -3000 m nd -3500 m, respectively, keeping lso distnce of 2000 m etween neighoring wells. When we look t the simultion results we see some minor increse in injection rte in first yer. Due to the ner wellore wter-gs replcement injection rtes get not much higher thn in the se-cse, even though the mximum injection pressure is higher ecuse of lrger depth. In yers 2-20, injection rtes for the injection scenrios t lrger depth increse significntly compred to the se-cse. One reson for tht is the higher mximum injection pressure. More importnt however is the fct tht the wells re plced in slightly ended line long the isoths, s opposed to the se-cse, where the interference etween wells is much stronger due to their tringulr pttern. The stronger influence of the well pttern cn e deduced from the injection rte increse etween se-cse nd cse 1-4 nd etween cses 1-4 nd 1-4. In summry, pprox. 40% higher injection rtes cn e relized when CO 2 is injected t the flnk of the structure. rte profile multiplying fctor 2.2 2.0 1.8 1.6 1.4 1.2 depth: -2360 to -2680 m depth: -3000 m depth: -3500 m yer 1 yers 2-10 yers 11-20 rte profile multiplying fctor 1.6 1.4 1.2 surrounded y quifers no-flowoundry to the west (~10 km) no-flowoundry to the north (~10 km) no-flowoundry to the west nd the north ssumed to e rel yer 1 yers 2-10 yers 11-20 1 1-4 1-4 1 1-8 1-8 1-8c Fig. 5. () Impct of well loction (top vs. flnk) on injectivity; () Impct of ordering fults on injectivity. In vrition ctegory 8 the impct of ordering fults on the reliztion of the injection rte ws nlyzed (Fig. 5). This ws done in view of the structurl sitution of the used lrge structure. The structure is in fct limited to the west end the north y fults t distnces of pprox. 10 km from the structurl top. As for the generic se-cse, quifer regions re defined t the model edges extending the reservoir lterlly to ll sides. In first yer, ccording to the simultion results, we do not see ny influence due to the ordering fults, ecuse the pressure rdius is not extending tht fr yet. In yers 2-20 however, we see strong influence of the ordering fults. Ech fult on its own (runs 1-8 nd 1-8) gives similr reduction in injection rtes compred to the se cse. In the yers 11-20 the injection rte reduction is out 25%. Both fults in comintion (run 1-8c) give n even stronger reduction of out 66% in the yers 11-20. 7. Comprison of simultion results from lrge nd smll model Hving similr verticl setups (reservoir thickness nd property distriution) the lrge generic model covers n re tht is pprox. 80 times lrger thn the smll qulified model. If the mximum injection rte is to e pplied, oth models cn only e compred for the first yer of injection, s the volume cpcity of the smll model is reltively smll. The simultion of the CO 2 injection in the smll model is relized with either 3 or 4 wells nd either
580 Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 injection into the structurl top or the flnk. In comintion these re four simultion runs tht re compred to the simultion run of the lrge model with the corresponding vrition cse (run 1-1). In Fig. 6 the multiplying fctors for the injection rte schedules re shown. Regrding the mximum injection rte, the simultion results from oth models re quite similr. In the first yer of injection only 20%-30% of the se-cse vrying injection rte cn e relized independent from the size of the model. The simultion with 10 topstructure wells of the lrge model is out equivlent to the simultion with 4 wells on the structurl flnk of the smll model. A higher numer of wells in the smll model would, due to restrictions in the flow regime, not result in significntly higher injection rtes. The generl differences in the simultion results etween the lrge generic model nd the smll qulified model, used for predicting reservoir ehvior in nturl gs storge, come from differences in depth, numer of wells, nd model dpttions to the smll model s result of the history mtch. The generl good mtch etween the mximum injection rtes in oth models gives us n indiction tht the lrge model cn e pplied with good confidence. Long-term experience in gs storge opertion cn therefore e used to secure the prediction results of similr geologicl settings even if the extent of the reservoir is significntly smller. 1.1 0.9 rte profile multiplying fctor 0.7 0.5 0.3 10 top-structure wells, k*h/2 3 top-structure wells 3 wells on structurl flnk 4 top-structure wells 4 wells on structurl flnk 0.1 1-1 Storge_1 Storge_2 Storge_3 Storge_4 Fig. 6. () Comprison of lrge model nd smll model simultion results; () Smll model gs sturtion. 8. Conclusions The ojective of the project presented here ws to nlyze y mens of simultion the fesiility of lrge scle industril CO 2 injection into sline quifer structure with vrying injection rtes. For tht purpose structure lrge enough to hold pprox. 120 million m³ CO 2 ws chosen from lrge numer of potentil structures in northern Germny. For the chosen structure nd strtigrphic horizon (Volpriehusen sndstone - Middle Bunter) the sme reservoir properties were pplied s known from the nery nturl gs storge. The generlly existing uncertinties hve een ddressed within comprehensive sensitivity nlysis. The vrile injection rte schedule over 20 yers is sed on n out of one yer week scenrio with the highest fluctutions in CO 2 output, due to wind nd solr power electricity genertion, nd sesonl component. Under the given circumstnces stle lrge-scle CO 2 storge could e demonstrted y simultion. Permeility cpcity, quifer size nd ordering fults hve the lrgest influence on the relizle vrying injection rte. Key findings hve een proven y comprison to model results chieved from simultion runs with qulified gs storge model. Vrying injection rtes show most significnt influence on overll injectivity within the first yers of
Andres Bnnch et l. / Energy Procedi 76 ( 2015 ) 573 581 581 opertion when compring it to respective constnt injection scenrios. With incresing opertionl lifetime the dynmiclly induced pressure spred ecomes more mrginl. Acknowledgements The uthors cknowledge the funding for the AUGE project y the Federl Ministry of Eduction nd Reserch in Germny within the GEOTECHNOLOGIEN Progrm. References [1] Stromerzeugung 2014 nch Energieträgern; http://strom-report.de. [2] GIS-sierte Krtennwendung Informtionssystem Speicher-Ktster Deutschlnd. http://www.gr.und.de/de/themen/ CO2Speicherung/ Downlods/Speicherktster_Krtennwendung.html. [3] Knopf S, My F, Müller C, Gerling JP.: Neuerechnung möglicher Kpzitäten zur CO 2-Speicherung in tiefen Aquifer-Strukturen. Energiewirtschftliche Tgesfrgen 2010; 4: 76-80. [4] Emissionen von direkten und indirekten Treihusgsen und von Schwefeldioxid. http://www.umweltundesmt.de/dten/ klimwndel/treihusgs-emissionen-in-deutschlnd. [5] Buit L et l.: CO 2-Trnsportinfrstruktur in Deutschlnd - Notwendigkeit und Rhmenedingungen 2050. DNV GL, Groningen, 2014 [6] Bldschuhn R, Binot F, Fleig S, Kockel F.: Geotektonischer Atls von Nordwest-Deutschlnd und dem deutschen Nordsee-Sektor. Geologisches Jhruch, Reihe A, Heft 153, 3 CDs. Hnnover, 2001.