Geological Survey of Denmark and Greenland Bulletin 35, 2016, 43-46 43 Injection of chemically tuned, ‘smart’ water in oil reser- voirs may increase both oil recovery rates and the total recovery (e.g. Morrow & Buckley 2011; Austad 2013; Zeinijahromi et al. 2015). This kind of water management has gained increased importance in the Danish North Sea reservoirs due to decreasing sweep efficiency in maturing oilfields. Knowledge about the compatibility of the injected water with local formation waters is, however, a prerequi- site for successful implementation. Here, we present a re- gional overview of formation waters from oil reservoirs in the Danish North Sea, which comprise three main types of formation brine, and one type of modified seawater re- lated to extensive water flooding. The water types show a distinct geographical distribution, which reflects original connate waters that are modified by saline brine being ei- ther depleted or enriched in SO 4 2–. Formation water and produced water database In order to characterise the water types we have selected a total of 33 water analyses, 25 of produced water and six of formation water from North Sea wells (three core samples from the Francisca-1 well and production tests from the Boje-1, Elna-1 and M-9X, wells) and finally two analyses of seawater (North Sea mean water composition and a treated low-sulphate seawater), see Fig. 1. For characterisa- tion, samples analysed for Na, K, Ca2, Mg2, Sr2, Ba2, Cl–, and SO 4 2– were used. Water density had been measured for most of the samples, however, it was estimated for four samples. The data were collected from Samuelsen et al. (2009), Mackay et al. (2012), and Undall-Behrend (2012) and from final well reports for the Boje-1, Elna-1, Fran- cisca-1 and M-9X wells. Water type classification To classify the water types in our database, Principal Com- ponent Analysis (PCA) was applied, whereby a matrix X of measured data (N samples, P variables) is transformed into sets of projection subspaces delineated by Principal Components (each a linear combination of all P variables), which display variance-maximised interrelationships be- tween variables (Esbensen 2010, Esbensen et al. 2015). PCA score plots display groupings, or clusters, of samples based on compositional similarities, as described by the variable correlations (shown in accompanying loading plots). They also quantify the proportion of total dataset variance that can be modelled by each component, see Fig. 2. All data analyses in this work are based on auto-scaled data. The data analysis was performed in two steps. Step one is a PCA analysis of all 33 samples to investigate relation- ships between seawater and reservoir water (Fig. 2A, B). Based hereon, pure seawater and the samples produced from Skjold, Dan B, Dan F and Halfdan, which repre- sent extensively seawater-flooded reservoirs, were removed Types of formation water and produced water in Danish oil- and gasfields: i mplications for enhanced oil recovery by injection of ‘smart’ water Niels H. Schovsbo, Hanne D. Holmslykke, Claus Kjøller, Kathrine Hedegaard, Lars Kristensen Erik Thomsen and Kim H. Esbensen Valdemar A Kraka Regnar Dan FDan B Cecilie Lulita Gorm Roar Rolf M-9 X Svend Skjold Boje-1 Siri Nini Dagmar Halfdan Tyra SE S. Arne Tyra W Tyra E Valdemar B Fransisca-1 Harald E Harald W Valdemar A Kraka Regnar Dan FDan B Cecilie Lulita Gorm Roar Rolf M-9 X Svend Skjold Boje-1 Siri Nini Elna-1 Dagmar Halfdan Tyra SE S. Arne Tyra W Tyra E Valdemar B Fransisca-1 Harald E Harald W UK Germany Norway Denmark The Nether- lands 500 km 0 50 km Well-head Well Field deliniation Fig. 1. Location of the wells used for water-type characterisation in the Danish part of the North Sea. © 2016 GEUS. Geological Survey of Denmark and Greenland Bulletin 35, 43–46. Open access: www.geus.dk/publications/bull 4444 to examine the relationships between the primary water types, which is done in step 2 of the analysis (Fig. 2C, D). In both PCA models the first two principal component axes resolve 80% of the total data variance, with the main trend expressed on the PCA-1 axis being salinity variation (seen as high positive PCA-1 loadings for Cl, Na and K). The PCA-2 axis displays high positive loadings of SO 4 2– and high negative loadings of elements such as Ba, Sr, Mg and Ca.This could indicate that SO 4 2– concentrations con- trol the concentration of Ba2+, Sr2+, Mg2+ and Ca2+ due to the low solubility of e.g. barite (BaSO 4 ) and anhydrite (CaSO 4 ). However, chemical speciation calculations using the numerical code PHREEQC suggest sub-saturation of SO 4 2–-bearing minerals except for barite in all water types. Composition and occurrence of water types From the PCA analysis, four water types can be identified based on natural groupings in the PCA-2 versus PCA-1 plot (Fig. 2). The most likely cause of the salinity varia- tion is a variable mixing of primary connate waters with brine originating from Permian Zechstein salt, which may or may not be SO 4 2–-rich (Warren et al. 1994). The char- acteristics and occurrence of each of the water types are presented below. Water types 1–3 have compositional char- acteristics similar to types presented in the comprehensive overview paper by Warren et al. (1994). Water type 1 plots within a very narrow group in the third quadrant of the PCA-2– PCA-1 plot (Fig. 2C), character- ised by low salinity water with an overall low abundance of all elements (Fig. 3). This water type is found in the Boje- 1, Francisca-1, Roar, Tyra E, Tyra SE and Valdemar fields and thus occurs in a broad range of reservoirs from Lower to Upper Cretaceous – Paleocene chalk to Oligocene sand- stone. The location borders the greater Tyra–Valdemar area, in geographical areas separated from saline Permian brines (Fig. 4). The water is interpreted to reflect unmodi- fied SO 4 2–-depleted formation water. Fig. 2. Principal component analysis, PCA. A: Score. B: loading plot of PCA on all data. C: score. D: loading plot of PCA from which seawater from the Dan B, Dan F, Skjold and Halfdan installations and produced waters were excluded. The plot models 79% (A, B) and 83% (C, D) of the total data vari- ance, respectively; variance proportions are shown along each component axis. Water types are classified according to their groupings in A (water type 4) and C (water type 1, 2, 3). F-1: Francisca-1. Vald: Valdemar. SW: seawater. LSSW: low SO 4 2–-treated seawater. P C A 2 ( 1 6 % ) -0.4 0.0 0.4 0.8 -4 0 4 8 PCA 1 (63 %) 0.0 0.2 0.4 P C A 2 ( 1 6 % ) -2 0 2 4 P C A 2 ( 1 6 % ) -2 0 2 4 PCA 1 (63 %) PCA 1 (67%) -4 0 4 8 0.0 0.2 0.4 PCA 1 (67%) -0.4 0.0 0.4 0.8 P C A 2 ( 1 6 % ) Boje-1 F-1Roar Tyra E Tyra SE Tyra W ValdA&B Dagmar Harald W Kraka Lulita M-9X Regnar Rolf Elna-1 Harald East S. Arne Siri A Siri C Siri D Gorm Na K Ca Mg Sr Ba Cl SO4 density Siri B Svend Na Ca Mg Sr Ba Cl SO4 density LSSW Dan B Dan F Halfdan Skjold KDagmar Harald WKraka Lulita M-9X Regnar Rolf Elna-1 Harald E S. ArneSiri ASiri C Siri B Siri D Svend Tyra SE Vald A Vald B Gorm F-1 B D Chalk field Sandstone field Produced water Formation water Sea water C A 45 Water type 2 is characterised by positive PCA-1 and nega- tive PCA-2 scores (Fig. 2C) and can compositionally be characterised by medium to high salinities, no SO 4 2–, and high to very high Ca and Ba concentrations (Fig. 3). This water type occurs in the Harald E and W, Lulita, S. Arne, Siri, Nini, Stine and Cecilie fields, all of which are locat- ed in the northern part of the Danish North Sea and in the Siri Canyon – i.e. in reservoirs that range in age from Jurassic to Paleocene and both in chalk and sand litholo- gies. Water type 2 is interpreted to reflect formation water modified by SO 4 2– depletion. Water type 3 plots with generally positive PCA-1 and PCA-2 scores in Fig. 2C, reflecting medium to high salini- ties and variable, low to high SO 4 2– concentrations (Fig. 3). This water type is found in the Dagmar, Elna-1, Gorm, Kraka, M-9X, Regnar, Rolf and Svend fields, most clearly expressed in the intensely fractured Dagmar field sample. This field is situated on top of a salt dome that has reservoir oil in chalk and Zechstein carbonates. Type 3 waters are restricted to chalk reservoirs overlying salt domes in the southern salt dome province, and are interpreted as forma- tion water enriched in SO 4 2–. Water type 4 plots close to, or together with seawater with negative PCA-1 and positive PCA-2 scores in Fig. 2A, cor- responding to low to medium salinity with high SO 4 2– con- centrations (Fig. 3). This water type occurs in the Dan, Halfdan and Skjold fields and is interpreted to be the result of decades of extensive water flooding performed by the operator (Energistyrelsen 2013). Analyses of water from the Dan field (the M-9X well; Fig. 2C) prior to flooding sug- gest that it was originally filled with water type 3. Implications for enhanced oil recovery by injection of ‘smart’ water Water injection is currently applied in several of the Dan- ish oil fields, mainly in order to provide pressure support. However, such injection may also have secondary effects such as increased imbibition, alteration of the reservoir rock wettability or mobilisation of fines with a resulting in- crease in reservoir sweep. In some cases, the specific chemi- cal composition of the injection water may be important. Thus, it has been suggested that carbonate rocks become more water wet if the injection water contains SO 4 2– in combination with excess Ca2+ or Mg2+ (e.g. Austad 2013). The result is enhanced oil recovery, which is even more pronounced both for chalk and sandstone if the salinity of the injection water is significantly lower than that of the formation water (Morrow & Buckley 2011; Austad 2013). Na K Cl Ca Mg Sr Ba S e a w at e r - n o rm al is e d , p p m /p p m 0.01 1 100 10 000 Na K Cl Ca Mg Sr Ba SO4SO4 Water type 1 Water type 2 Water type 3 Water type 4 A B Fig. 3. A: Water concentration normalised to seawater composition. B: Calculated average compositions of the four water types observed in this study. For display purposes, Ba2+ and Sr2+ concentrations of 0 ppm in seawater have been increased to 1 ppm. Water types: 1: Low salinity 2: High salinity - SO4 3: High salinity + SO4 4: Sea water-modified Type 2-dominated Type 3-dominated Type 1-dominated Type 2-dominated Type 3-dominated Type 1-dominated Type 2-dominated Type 3-dominated Type 1-dominated 0 50 km Fig. 4. Occurrence of resolved water types in the Danish oil- and gas- fields. For location names see Fig. 1. The water types are geographically restricted and ref lect both structural basin development and reservoir conditions. The seawater modified water type 4 is assumed to have originated as water type 3 based on pre-waterf looding formation water analysis and its structural position within the salt dome province. 4646 Although several different mechanistic explanations have been suggested, a supposed change in carbonate rock wet- tability would involve surface chemical reactions such as ion exchange between SO 4 2– and oil molecules (e.g. Austad 2013). Following this argumentation, it is likely that in- jection of ‘smart’ SO 4 2–-bearing water in chalk reservoirs would have the largest effect in reservoirs with saline for- mation water depleted in SO 4 2– (water type 2). However, the application of SO 4 2–-rich water in reservoirs with this type of connate water is not straightforward, as there is a risk of scaling and subsequent clogging of the reservoir if the injected water is mixed with the connate water, due to its high concentrations of Ca2+, Ba2+, and Sr2+ (Samuelson et al. 2009; Mackay et al. 2012). Another risk related to in- jection of SO 4 2–-bearing water in SO 4 2–-depleted reservoirs is the possibility of hydrogen sulphide formation due to SO 4 2– reducing microbial activity. For reservoirs already enriched in SO 4 2– (water type 3) or with water of relatively low salinity (water type 1), other types of injection water may have greater effects on oil re- covery. In shaly sand reservoirs, injection of low-salinity ‘smart’ water can also mobilise clay fines, in order to intention- ally clog current flow paths and redirect the flow in the reservoir (e.g. Morrow & Buckley 2011; Zeinijahromi et al. 2015). In this case, the mobilisation of non-swelling clays is provoked solely by the change in salinity. Therefore, ap- plication of this type of water technology seems to be most relevant in reservoirs with connate water of relatively high salinity, such as most of the reservoirs in the Siri Canyon (water type 2). Conclusions Four water types are present in the Danish North sea: SO 4 2–-bearing, medium- to highly saline water (type 3), SO 4 2–-depleted medium to high saline water (type 2), SO 4 2–- -depleted low saline (type 1), and a seawater-modified man- ifestation (type 4 water). These water types reflect variable mixing of connate water with deeper brines and are tied in with the known hydrocarbon provinces. Type 2 represents the Siri Canyon and the South Arne – Svend areas. Water type 3 is characteristic of salt dome reservoirs, while water type 1 represents the greater Tyra–Valdemar area. The vari- able chemistry of the formation water in the Danish North Sea imposes regional differences in production strategies and hence in the designing of ‘smart’ water for enhanced oil recovery. The classification of water types presented here shows that their composition is predictable and related to geographical domains in the North Sea. This may be useful when designing procedures for optimal water man- agement in the Danish North Sea, e.g. application of low salinity water flooding on mature fields, or in some cases even during the exploration stage. References Austad, T. 2013: Water-based EOR in carbonates and sandstones: New chemical understanding of the EOR potential using ‘smartwater’. In: Sheng, J.J. (ed.): Enchanced oil recovery field case studies, 301–335. Waltham, MA, USA: Elsevier. Energistyrelsen, 2013: Danmarks olie- og gasproduktion 2013, 105 pp. København: Energistyrelsen. Esbensen, K.H. 2010: Multivariate data analysis – in practice. An in- troduction to multivariate data analysis and experimental design, 5th edition, 598 pp. Oslo: CAMO Software AS. Esbensen, K., Schovsbo, N.H. & Kristiansen, L. 2015: Down-hole per- meability prediction – a chemometric wire-line log feasibility study from a North Sea chalk well. Geological Survey of Denmark and Greenland Bulletin 33, 13–16. Mackay, E., Ginty, W.R. & Jones, T.J. 2012: Oilfield scale manage- ment in the Siri asset – paradigm shift due to the use of mixed PWRI / seawater injection. 74th EAGE Conference and Exhibition incor- porating EUROPEC 2012. Copenhagen, Denmark, 4–7 June 2012. SPE 154534, 1–12. Morrow, N. & Buckley, J. 2011: Improved oil recovery by low-salinity waterf looding. Journal of Petroleum Technology 63, 106–112. Samuelsen, E.H., Frederiksen, R.A., Heath, S.M., Thornton, A., Sim, M., Arefjord, A. & McAra, E.K. 2009: Downhole scale control through continuous injection of scale inhibitor in the water injection – a field case. Conference Tekna Geilo paper 240309, 23 pp. Undall-Behrend, G. 2012: Produceret vand på Tyra Øst F. Bachelorpro- jekt Århus Maskinmesterskole, 75 pp. Warren, E.A., Smalley, C.P. & Howarth, R.J. 1994: Compositional variations of North Sea formation waters, Part 4. Geological Society, London. Memoirs 15, 119–208. Zeinijahromi, A., Ahmetgareev, V., Badalyan, A., Khisamov, R. & Bedrikovetsky, P. 2015: Case study of low salinity water injection in Zichebashskoe field. Journal of Petroleum Science Research 4, 16–31. Authors’ address Geological Survey of Denmark and Greenland, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark; E-mail: nsc@geus.dk