Simultaneous oil recovery and residual gas storage: A pore-level analysis using in situ X-ray micro-tomography

Research paper
of 10
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Simultaneous oil recovery and residual gas storage: A pore-level analysis usingin situ X-ray micro-tomography S. Iglauer a, ⇑ , A. Paluszny b , M.J. Blunt b a Curtin University, Department of Petroleum Engineering, 26 Dick Perry Avenue, 6151 Perth, Australia b Imperial College London, Department of Earth Science and Engineering, Prince Consort Road, SW7 2AZ London, United Kingdom h i g h l i g h t s Waterflooding a reservoir prior togas injection leads to significantlylower oil recovery. More oil can be produced by directlyinjecting gas into a virgin oilreservoir. Waterflooding a reservoir prior togas injection leads to significantlylower gas storage capacities. More residual gas can be stored bydirectly injecting gas into an oilreservoir. Residual oil and gas clusters have alarge cluster size distribution andsurface area. g r a p h i c a l a b s t r a c ta r t i c l e i n f o  Article history: Received 28 April 2012Received in revised form 7 June 2012Accepted 21 June 2012Available online 28 August 2012 Keywords: Enhanced oil recoveryCarbon geo-sequestrationResidual oilResidual gas a b s t r a c t We imaged sandstone cores at residual gas saturation ( S  gr ) with synchrotron radiation at a nominal res-olutionof (9 l m) 3 . We studied two three-phase flooding sequences: (1) gas injection into a core contain-ing oil and initial water followed by a waterflood (gwprocess); (2) gas injection into a waterflooded corefollowed by another waterflood (wgw process). In the gw flood we measured a significantly higher  S  gr (=20.6%;  S  gr  in the wgw flood was 5.3%) and a significantly lower residual oil saturation ( S  or ;  S  or  in thegwflood was 21.6% and S  or  in the wgwflood was 29.3%). We also studied the size distribution of individ-ual trapped clusters in the pore space. We found an approximately power-law distribution  N  / s  s withan exponent  s  = 2.02–2.03 for the residual oil clusters and  s =2.04 for the gas clusters in the gw flood.  s (=2.32)estimatedforthegasclustersinthewgwprocesswassignificantlydifferent.Furthermore,wecal-culatedthesurfacearea  A –volume V   relationshipsfortheclusters.Againanapproximatepower-lawrela-tionship was observed,  A / V   p with  p  0.75. Moreover, in the gw flood sequence we identified oil layerssandwiched between the gas and water phases; we did not identify such oil layers in the wgw flood.Theseresultshaveseveralimportantimplicationsforoilrecovery, carbongeo-sequestrationandcontam-inant transport: (a) significantly more oil can be produced and much more gas can be stored using a gwflood; (b) cluster size distributions for residual oil or gas clusters in three-phase flow are similar to thoseobserved in analogue two-phase flow; (c) there is a large cluster surface area available for dissolution of the residual phase into an aqueous phase; however, this surface area is significantly smaller than pre-dicted by percolation theory (  p  1), which implies that CO 2  dissolution trapping and contamination of aquifers by hazardous organic solvents is slower than expected because of reduced interfacial contactareas.   2012 Elsevier Ltd. All rights reserved. 0016-2361/$ - see front matter    2012 Elsevier Ltd. All rights reserved. ⇑ Corresponding author. E-mail address: (S. Iglauer).Fuel 103 (2013) 905–914 Contents lists available at SciVerse ScienceDirect Fuel journal homepage:  1. Introduction With a growing global population and fast economic develop-ment coupled with dwindling fossil fuel resources – and the factthat world energy consumption is currently mainly based on fossilfuels (they account for more than 80% of the total world’s energyconsumption [1]) – it is important to develop advanced technolo-gies that can recover additional fossil fuel. Another challenge con-cerns the carbon dioxide (CO 2 ) emissions associated with burningfossil fuels and the changes to global climate that may result. Onetechnology to deal with this problem is CCS – Carbon Capture andStorage – where CO 2  is collected from fossil-fuel burning powerstations and other industrial sites, transported and injected deepunderground into saline aquifers or depleted oil or gas fields [2].Crude oil is the most important fuel; in 2008 it contributed41.6% (equivalent to an energy of 3505Mtoe) to the world’s totalfinal consumption[1]. Crudeoil, whichis not producedbyprimaryproductionornatural drivemechanismssuchassolutiongasdrive,water influx or gravity drainage, can be produced by enhanced oilrecovery (EOR) methods [3]. EOR processes include miscible orpartially miscible gas flooding, thermal stimulation [3], surfactantflooding[4]orpolymerflooding[5]. GasinjectionEOR(GEOR, with natural gas, carbon dioxide CO 2 , or nitrogen) is usually employedto displace and recover residual oil that remains in the reservoirafter natural depletion and waterflooding.In a GEOR process three fluid phases flow: oil, gas and brine;three-phase flow also occurs in carbon geo-sequestration (CCS) indepleted oil or gas reservoirs [6,7]. CCS can be combined withGEOR. The objective is to simultaneously maximise CO 2  storageandhydrocarbonrecovery[8]. Gas is injectedeitheras asecondaryprocess,intooilandinitialwater,orasatertiaryprocessintoresid-ual oil and water after waterflooding. For carbon dioxide storage itis valuable to trap the CO 2  as a residual phase, and so both gasinjection sequences can be followed by further waterflooding.We will compare these two processes in this paper.Several researchers have investigated three-phase flow at themeso (centimetre) scale, mainly with the focus on oil recovery[9,10], fluid distributions [11], relative permeability [12–15], or capillary pressure measurements [16]. Pore-scale displacementstudies have also been conducted [10,17–23]. These pore-scalestudies employed 2D models, which are, however, not necessarilyrepresentative of reservoir flow conditions as the connectivity of the pore network cannot be captured correctly (for example thepercolation threshold for 3D lattices is significantly lower thanfor 2D lattices [24]). In addition such 2D models typically usestrongly simplified artificial materials – not reservoir rock – whichmay not be representative of reservoir conditions. Furthermore,three-phase trapping has been measured in rock samples [25–27], which is important for CCS risk and capacity assessmentsand related residual trapping capacity predictions [28].To optimise GEOR, reservoir flow models are required that canpredict the efficiency of oil recovery and associated time scales.However, because of the complexities of rock pore morphology,fluid–fluid and fluid–solid interactions, theoretical understandingis currently limited to simple models which only have limitedpredictive capabilities with scant physical foundation, based onpore-scale displacement processes. To overcome this, we analysethree-phase flow (oil, brine, gas) in a sandstone at the pore-scale(micrometre scale) in 3D with micro-computed tomography( l -CT), and we compare two GEOR flooding sequences. 2. Experimental methodology  We compared two GEOR flooding sequences:(1) gas flood of a virgin oil reservoir; gas was directly injectedinto a core at connate water saturation ( S  wc ) followed by achase brine injection (gw sequence), and(2) gas flood of a waterflooded oil reservoir; where gas wasinjected into a waterflooded core at residual oil saturation( S  or ) followed by chase brine injection (wgw sequence).For these experiments we selected a clean, well-sorted rela-tively homogenous sandstone outcrop (Clashach, a quarried sand-stone from Elgin in Scotland). The brine permeability wasmeasured to be 8  10  14 m 2 (80mD) [29] and porosity was11.1%±0.5%. Clashach consists mainly of quartz ( P 96wt.%) withsmall amounts of K-feldspar, calcite and ankerite [30]. Oil (1-Bromododecane, purity P 99.5 mass%, supplied by Aldrich), gas(N 2 , purity>99.998mass%) and brine (10wt.% potassium iodide(KI) in deionized water) were selected as fluid phases. The brinewas doped with KI and the special brominated oil was used toguaranteesufficientCTcontrast.Thefluid–fluidinterfacialtensionsare listed in Table 1. We assume that the rock is water-wet.The spreading coefficient is defined by: S  ¼ c gw  c ow  c go  ð 1 Þ where  c gw  is the gas-brine interfacial tension,  c ow  is the oil-brineinterfacial tension and  c go  is the gas-oil interfacial tension. Thespreading coefficient  S   has a value of    11.05mN/m. All experi-ments were conducted at ambient laboratory conditions, that is293K and 0.1MPa. A small cylindrical core plug of 5mm diameterand 10mm length was drilled and placed into a fluoroplastic heat-shrink sleeve. Both plug ends were sealed with standard stainlesssteel Swagelok fittings, and this core system was heated so thatthe heatshrink sleeve strongly adhered to the plug and fluid by-passing was prevented. The fluids were injected with standard syr-inge pumps (Teledyne ISCO, model 500D, Lincoln, NE, USA) into theplug positioned horizontally. For both flood sequences, the cores were first completely satu-rated with brine under vacuum and then approximately 20 porevolumes (PV) of oil were injected at a low capillary number( l q/ c  5  10  6 , where  q  is the Darcy flow rate,  l  is the viscosityof the injected phase and c  is the interfacial tension), which is rep-resentativeofflowconditionsinareservoir.AfterinjectionofafewPV of oil, water production ceased. This represents a virgin oil res-ervoirwhereoildisplacedmostoftheformationwaterovergeolog-ical times except the immobile (connate) water  S  wc .  S  wc  was thestartingpointforbothfloodingsequences.Forthegwfloodthenextstep was to inject approximately 200 PV of gas at a capillary num-ber of approximately 10  5 until liquid production by viscous dis-placement ceased (visual observation). Then approximately 20 PVof chase brine were injected at a lowcapillary number (10  6 ) untilno oil or gas was produced and the residual gas saturation  S  gr  wasreached. In case of the wgw sequence, the core at  S  wc  was firstwaterflooded with approximately 20 PV of brine at a low capillarynumber (10  6 ) to  S  or , and then gas and chase brine were injectedusing the same procedure as in the gw process. During flooding,the cores were held horizontally. Bond numbers were estimated  Table 1 Interfacial tensions of the fluids used. Fluid–fluid system Interfacial tension (mN/m)Water/1-bromododecane a c ow  =52.09Water/nitrogen b c gw  =721-Bromododecane/nitrogen c c go  =30.96 a Measured at 295K [31]. b Measured at 293.15K and 0.101MPa [32]. c Surface tension of 1-bromododecane.906  S. Iglauer et al./Fuel 103 (2013) 905–914  to be   10  5 for the liquid–gas system and 10  7 for the oil-brinesystem; we therefore do not expect the residual clusters/residualsaturations to be influenced by buoyancy forces [33].Both specimens were then scanned with synchrotron radiationat the SYRMEP beamline of the Elettra light source facility in Trie-ste, Italy (photon energy=30keV). We analysed a subvolume of the resulting images consisting of 300 3 voxels (19.683mm 3 , nom-inal voxel resolution of 9 l m).All raw l -CT images were cleaned of ring artefacts by applyinga stripe removal algorithm based on combined wavelet—Fourierfiltering [34]. Salt-and-pepper noise was removed using aconservative anisotropic regularization filter [35]. The phaseswere then segmented according to their CT contrast using mul-ti-thresholding, i.e. by identifying peaks in the grey-level histo-gram of each image based on Otsu’s algorithm [36]. Fig. 1 shows slices through the rock and fluid phase distributions forboth flood sequences, and Fig. 2 displays observed residual oiland gas clusters in 3D. 3. Results and discussion  3.1. Fluid saturations – residual oil and gas saturations Fluid phase saturations measured from the  l -CT images arelisted in Table 2 for both flooding sequences. In addition severalmeso-scale literature values are added for comparison. It is clearfrom our datasets that much more gas can be stored and moreincrementaloilcanbeproducedifgasisdirectlyinjectedintoavir-gin oil reservoir (gw flood sequence).Furthermore, we have previously studied similar two-phase(brine and oil) flow processes at the micro- and meso-scale[30,37–40]. From the comparison of these datasets we reach sev-eral conclusions:1. More oil can be recovered by three-phase flow, i.e. a lower  S  or can be achieved, when gas injection is employed compared towater injection alone. (a) gw flood.(c) wgw flood. (b) gw flood. (d) wgw flood. Fig. 1.  2Dimageslicesthroughtherockandfluidsatresidualgassaturationafterchasebrineinjection.(a)gwFloodsequence, rawimage:oiliswhite,gasisblack,brinedarkgreyandsandstoneis lightgrey. (b)gwFloodsequence: segmentedimages, brineislight blue, oil isred, gasis yellowandrockisbrown. (c) wgwFloodsequence, rawimage:oiliswhite, gasisblack, brinedarkgreyandsandstoneislightgrey.(d)wgwFloodsequence: segmentedimages, brineislightblue, oilisred,gasisyellowandrockisbrown.All images show an area of 2.7mm  2.7mm=7.29mm 2 . S. Iglauer et al./Fuel 103 (2013) 905–914  907  (a) wgw – gas(b) gw – gas(c) wgw – oil(d) gw – oil(e) wgw – oil(f) gw – oil(g) wgw – gas(h) gw – gas Fig. 2.  3Dimagesofresidualoilandgasclusters(rockandbrinephaseswerecroppedout).(a)Residualgasclustersinthewgwflood.(b)Residualgasclustersinthegwflood.(c)residualoilclustersinthewgwflood.(d)Residualoilclustersinthegwflood.(a–d):Allvolumesdisplayedare2.7mm  2.7mm  2.7mm=19.683mm 3 .Theclustersarecolouredaccordingtosize: blue<235nl (nanolitre) (1000voxels), green235–2350nl (1000–10,000voxels), yellow2350–23,500nl (10000–100,000voxels), orange 23,500–235,000nl (10 5 –10 6 ), red>235,000nl (>10 6 voxels). (e) wgw flood: selected individual residual oil clusters: the largest clusters (6250–13000nl; 26,564–54,797voxels),several medium-sized clusters (1200–2350nl, 5000–10,000voxels) and some small clusters (2.35–235nl, i.e. 10–1000voxels) are shown. (f) gw flood: selected individualresidual oil clusters: the largest clusters (3800–6900nl, 16,180–29,251voxels), several medium-sized clusters (1200–2350nl, 5000–10,000voxels) and some small clusters(2.35–235nl, 10–1000voxels) are shown. (g) wgw flood: selected individual residual gas clusters: the largest clusters (75–100nl, 319–434voxels), several medium-sizedclusters (23.5–26nl, 100–110voxels) and some small clusters (11.7–16.5nl, 50–70voxels) are shown. (h) gw flood: selected individual residual gas clusters: the largestclusters (7400–11,000nl, 31,393–46,222voxels) and several medium-sized clusters (1200–2350nl, 5000–10,000voxels).908  S. Iglauer et al./Fuel 103 (2013) 905–914
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks