Geological Survey of Denmark and Greenland Bulletin 31, 2014, 27-30 27 A novel technique for obtaining representative water samples during CO2 core-flooding experiments on chalk at reservoir conditions Claus Kjøller and John Zuta There is a huge potential for using CO2 gas to recover ad- ditional oil after water flooding in reservoir chalk. However, the injection of CO2 into chalk reservoirs will disturb the chemical equilibrium between formation water, injection water and chalk. A proper understanding of these CO2- induced interactions and the resulting changes in the physical properties at representative reservoir conditions is required. Unfortunately, reliable chemical data are rare because data cannot be acquired directly at reservoir conditions with present-day techniques. In published experiments, water samples are in many cases obtained at atmospheric condi- tions with the aid of a back-pressure regulator. Thus, water samples are not representative of in situ reservoir conditions and if proper care is not taken, the collected data cannot be used to judge the magnitude of the chemical reactions taking place at reservoir conditions. However, in some cases water obtained at laboratory conditions can give information on in situ reservoir conditions by using geochemical speciation models to account for dissolved gases that are lost from the effluents during sampling (Bachu & Adams 2003). The objective of this study was to develop and test a new technique for obtaining water samples during CO2-brine– rock interactions in reservoir chalk under representative reservoir conditions and gain a better understanding of the chemical interactions which occur during the injection of CO2. The experiments were per- formed by injecting CO2-saturated seawater at supercritical CO2 conditions, at a pressure of 17.24 MPa (2500 psig; pounds per square inch of gauge pressure) and a temperature of 75°C. These values are typical of a region in a chalk field where the rock is exposed to long term reactions with flow of CO2-bearing water. In addition, the numerical code PHREEQC-3 (Parkhurst & Appelo 2013) was used to reproduce the experiments by assuming equilibrium between calcite and injected fluids. In this way, we can validate the sampling pro- cedure and investigate how the measured parameters (pH, CO2 pressure, calcium (Ca) and bicarbonate (HCO3–) con- centrations) compared with the calculated parameters. Reservoir chalk samples Chalk plugs for the experiment were sampled from the Maas- trichtian Tor Formation in the central North Sea and were first cleaned of salt and oil with methanol and toluene. Fol- lowing initial determination of porosity and permeability, the samples were saturated with synthetic formation water (FW). All plugs had a diameter of 3.81 cm and a length of approximately 7.5 cm; plugs with similar petrophysical prop- erties (Table 1) were used to check the reproducibility of the studied sampling procedure. The compositions of synthetic formation water, synthetic seawater and CO2 synthetic satu- rated seawater are listed in Table 2. Experimental setup The setup/rig for the experiments is shown in Fig. 1. It in- cludes three cylinders with pistons placed in an oven to main- tain constant temperature. Two of the cylinders were used © 2014 GEUS. Geological Survey of Denmark and Greenland Bulletin 31, 27–30. Open access: www.geus.dk/publications/bull Waste CO2-saturated waterLimit of oven Seawater Flooding plug Eq ui l. p lu g Primary sample loop (13 ml) Separation point Secondary sample cylinder (125 ml) Pressure transducer Pressure transducer pH-meter Pressure transducer ST1 ST2 PT1PT2 Primary sample loop (13 ml) Fig. 1. Experimental setup/rig within the oven. The rig has two primary sample loops. 2828 for the delivery of CO2-saturated seawater and seawater. The third cylinder was used as a waste tank and for maintaining fluid pressure. All the fluids were injected at a constant rate of 6.5 ml/h with the plug in a horizontal position. The injection of CO2-saturated seawater was preceded by injecting close to 3.5 pore volume (PV) of seawater to mimic the present-day water flooding. At the end of the injection of CO2-saturated seawater, seawater was again injected through the plug. At this point, equilibrium with calcium carbonate was obtained in the injection fluid by first passing the seawater through the equilibrium plug (Fig. 1). This was done in order to avoid any further dissolution of chalk in the plug under study. The fi- nal seawater flooding was performed to displace any residual CO2 gas and bring the concentrations back to initial levels. Water sampling technique The system for obtaining water samples was located at the downstream end of the rig (Fig. 1). It consists of two primary sample loops each with a volume of c. 13 ml. This enabled continuous collection of water samples for every 13 ml (c. 0.5 pore volume) of plug flooded. The sampling was done by al- ternating the flow process between the two primary loops with the aid of computer-controlled valves (CV-210 valves) without disrupting the injection process. The primary sample loops were connected to a secondary sample cylinder with a volume of 125 ml by a separation point. Water samples were transferred from the primary sample loop to the secondary sample cylinder by a de-pressurisation step. This made it pos- sible to collect water samples for a set of chemical parameters (Ca and HCO3– concentrations) at predefined sampling intervals. The pH and CO2 pressures of the de-pressurised effluents were measured on-line in the secondary sample cyl- inder after the de-pressurisation step. The sampling from the secondary sample cylinder was subsequently done as fast as possible, in less than 5 min., thereby providing a consistent chemical dataset for the conditions prevailing in the second- ary sample cylinder. In this way it was possible to validate the subsequent numerical calculations used to estimate the in situ chemical conditions. A detailed description of the sampling procedure as well as documentation for the consistency of the collected dataset was published by Kjøller & Zuta (2012). Batch modelling with PHREEQC-3 The modelling was performed at the same conditions – pres- sure and temperature of 17.24 MPa (2500 psig) and 75°C as the experiments, in order to estimate the true in situ reser- voir condition chemistry. Three equilibrium steps in a closed system were investigated: (1) equilibrium between CO2 and seawater corresponding to the injected solution, (2) equilib- rium between CO2-saturated seawater and calcite, represent- ing reservoir conditions, and (3) composition of the solution after de-pressurisation in the secondary sample cylinder. A good fit between experimental and numerical data after step Table 1. Petrophyscial properties of plugs prior to the injection of CO2-saturated water at 17.24 MPa (2500 psig) and 75°C 17A 32.60 26.94 1.36 19A 30.60 25.59 1.49 Porosity Pore volume Permeability Plug ID (%) (ml) (mD) Table 2. Composition (in mg/l) of synthetic formation water (FW), seawater (SW), and CO2-saturated seawater (CSW) Na 22866 11090 11090 K 175 408 408 Mg 226 1370 1370 Ca 1244 434 434 Sr 142 6.8 6.8 Cl– 38383 20173 20173 HCO3 – 22 30 70 SO4 2– – 2780 2780 Ions (FW) (SW) (CSW) 0 0.033 0.067 0.100 0.133 0.167 0.200 0.233 4 4.5 5 5.5 6 6.5 7 7.5 0 5 10 15 20 25 C O 2 p re ss ur e (M Pa ) pH Pore volume pH Plug 17A pH Plug 19A CO2 pressure Plug 17A CO2 pressure Plug 19A Fig. 2. Measured pH and CO2 pressure for plug 17A and parallel plug 19A versus cumulative, injected pore volume. pH 3.2 4.8 6.3 Ca (mg/l) 438 1970 1970 Alkalinity, HCO3 – (mg/l) 82 3274 3308 Saturation index (calcite) –3.78 0 1.43 Chemical parameter Step 1 Step 2 Step 3 Table 3. Composition of resulting solutions at the two different equilibrium steps followed by the de-pressurisation step based on calculations with phreeqc.dat database in PHREEQC-3 29 3 is considered as a validation of the numerical estimate of reservoir conditions in step 2 (Kjøller & Zuta 2012). Results Figures 2–4 show pH, CO2 pressure, Ca, and HCO3– con- centrations measured as a function of injected cumulative pore volumes. The compositions of the resulting solutions calculated at the three steps with PHREEQC-3 are shown in Table 3. Equilibrium of CO2 and seawater results in an ini- tial CO2-saturated seawater solution with a pH of 3.2 and a saturation index of calcite of –3.78 (Table 3, step 1), an indi- cation of a strong dissolution potential for chalk. Thus, CO2 is dissolved and reacts with water to form carbonic acid (H2CO3). It is part of this acidity that is removed as the dis- solution process advances into the plugs. The dissolution of calcium carbonate in the plugs takes place according to the overall chemical reaction: During the injection of seawater within the first 3.5 pore volume, the pH increases to between 7.5 and 7.7 and subse- quently decreases to an average of 6.1 during the injection of CO2-saturated seawater (Fig. 2). At the final stage of the ex- periment, the pH increases again to the initial seawater pH level, an indication that all the residual CO2 was removed from the plugs during the injection of seawater in equilib- rium with chalk. The calculated decrease in pH at reservoir conditions (Table 3, step 2) corresponds with the measured decrease in pH after 3.5 pore volume, where breakthrough of the CO2-saturated seawater is observed. However, the calcu- lated reservoir condition pH of 4.8 is much lower than the average measured pH during the injection of CO2-saturat- ed seawater. This was expected since the pH was measured after the de-pressurisation step where CO2 gas had already degassed from the solution. Thus, taking the pH (6.3) of the degassed solution into account in the numerical simulation provides an excellent fit to the average measured pH of 6.1 (Table 3, step 3). The CO2 pressure in the secondary sample cylinder varied between 0.17 MPa (25 psig) and 0.24 MPa (35 psig) at the de-pressurisation stage during the injection of CO2-saturated seawater (Fig. 2). This matches with the calculated CO2 pressure of 0.20 MPa (28.9 psig) in the gas phase of the secondary sample cylinder (Table 4), and further validates the numerical calculations. According to the equation, degassing of CO2 from the water samples, caused by the transfer of water samples from the primary sample loop to the secondary sample cylinder, was expected to result in precipitation of CaCO3. Com- parison of Ca and HCO3– concentrations measured in both filtered and unfiltered water samples showed no evidence of mobilisation or precipitation of fine particles during the de-pressurisation stage, neither in plug 17A nor in plug 19A (Fig. 3A, B). The Ca concentration decreases during the first 3.5 pore volume to the Ca-concentration level in seawater H CO CaCO Ca 2HCO2 3 3 2 3+ +" + - Table 4. Properties of the gas phase generated after the de-pressurisation stage in the 125 ml secondary sample cylinder with the phreeqc.dat database in PHREEQC-3 Total pressure 2.00 MPa (28.9 psig) Gas volume 1.25e-001 litres Molar volume 1.38e+001 litres/mole P×Vm/RT 0.99228 (compressibility factor Z) 0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 0 5 10 15 20 25 H C O 3– c on ce nt ra tio n (m g/ l) C a co nc en tr at io n (m g/ l) Pore volume Ca Filt Ca Ca - SW Ca - FW HCO3 – Filt HCO3 – A Fig. 3. Measured Ca and HCO3– concentrations for filtered and unfiltered water samples in plugs 17A (A) and 19A (B) versus cumulative, injected pore volume. Filt: filtered. SW: synthetic seawater. FW: synthetic formation water. 0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 0 5 10 15 20 25 C a co nc en tr at io n (m g/ l) Ca Filt Ca Ca - SW Ca - FW HCO3 – Filt HCO3 – H C O 3– c on ce nt ra tio n (m g/ l) Pore volume B 3030 of c. 450 mg/l. Subsequently, the measured Ca concentra- tion increases to an average of 1200–1300 mg/l during the injection of CO2-saturated seawater, which is due to disso- lution of calcium carbonate in the plugs. At the final steps of the experiment, the Ca concentration again decreases to the seawater level during injection of seawater to remove re- sidual CO2 from the plugs. In the geochemical model, the calculated Ca concentration of 1970 mg/l at reservoir condi- tions (Table 3, step 2) is simulated by dissolution of calcite at reservoir conditions, and does not take into account the possible precipitation of calcium carbonate during sampling. Therefore, the calculated reservoir-condition Ca concentra- tion is higher than the concentrations measured in the ex- periments during the period with CO2-saturated seawater flooding. Since no mobilisation or deposition of suspended fine particles in the water samples were observed, the differ- ence in Ca concentration must be attributed to deposition of fine calcium carbonate particles on the inner surface walls of the secondary sample cylinder during the pressure-reduction stages. In line with this, the simulated saturation index (SI) with respect to calcite of 1.43 after de-pressurisation (Table 3, step 3) is higher than the saturation indices between 0.6 and 1.2 that can be calculated based on the measured water chemistry. If the saturation index with respect to calcite is con- strained to be between 0.6 and 1.2 in the numerical calcula- tions in step 3, the calculated Ca concentration at the sam- pling conditions varies between c.1150–1650 mg/l, which is in much better agreement with the actually measured Ca concentrations (Fig. 3). Similar considerations and conclu- sions can be made for the HCO3– -concentration trend dur- ing the experiment. Thus, the measured HCO3– concentra- tion averages 2350–2550 mg/l during the injection of CO2 saturated seawater, while the calculated HCO3– concentra- tion after de-pressurisation is 3308 mg/l when no constraints on the saturation index with respect to calcite are applied (Table 3, step 3). However, accounting for some calcium car- bonate precipitation during de-pressurisation will diminish the difference between calculated and measured concentra- tions, and overall it is suggested that the calculated reservoir condition concentrations shown in Table 3, step 2 represent the true in situ reservoir conditions prevailing in the plugs during experiments. The general decrease in measured Ca and HCO3– concen- trations during the injection of CO2-saturated seawater from 3.5 to 15 pore volume (Fig. 3), suggests that, with time, there is an increasing amount of calcium carbonate precipitating in the secondary sample cylinder during sampling. For future studies, this gradual change in calcium carbonate precipita- tion should be avoided in order to minimise the uncertainty in the measured data, and thereby also the uncertainty of the numerical estimate of the reservoir condition chemistry (Ta- ble 3, step 2). Conclusions • A new sampling technique was developed and used to study CO2-brine–rock interactions in reservoir chalk at reservoir conditions. • The similar results obtained in the parallel plugs vali- date the sampling procedure and show a high degree of reproducibility. • The results from the experiment and the modeling show increased levels of calcium and alkalinity (HCO3–) with decreasing pH during the injection of CO2-saturated water at reservoir conditions. There were, however, dif- ferences between the measured and calculated calcium concentration and HCO3– concentration which is probably due to accumulation of fine calcium carbon- ate particles on the inner walls of the secondary sample cylinder during the de-pressurisation stage. • Further refinement of the method may include modi- fication of the secondary sample cylinder, in order to minimise the accumulation of fine calcium carbonate particles on its inner surface walls. Acknowledgements We are grateful to Dansk Undergrunds Consortium (DUC) – a co-opera- tion between A.P. Møller Maersk, Shell, Chevron and Nordsøfonden (the Danish North Sea Fund) – for sponsoring the work. References Bachu, S. & Adams, J.J. 2003: Sequestration of CO2 in geological media in response to climate change: capacity of deep saline aquifers to sequester CO2 in solution. Energ y Conversion and Management 44, 3151–3175. Kjøller, C. & Zuta, J. 2012: CO2-brine–rock interactions in reservoir chalk rock – a coupled experimental and numerical approach for ob- taining hydrochemical results at reservoir conditions. Paper SCA2012- 04 presented at International Symposium of the Society of Core Ana- lysts, Aberdeen, Scotland, UK, 27–30 August, 2012, 12 pp. Parkhurst, D.L. & Appelo, C.A.J. 2013: Description of input and exam- ples for PHR EEQC version 3 – a computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. U.S. Geological Survey Techniques and Methods, http:// pubs.usgs.gov/tm/06/a43/. Authors’ address: Geological Survey of Denmark and Greenland, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark. E-mail: clkj@geus.dk