Bouillante geothermal f ield (Guadeloupe)

Transcription

1 Bouillante geothermal f ield (Guadeloupe) Geochemical monitoring during the stimulation operation Critical evaluation of chemical and isotopic field data December 1999 R BRGM."mm=..".WCE 0. ",Ems

2 Bouillante geothermal fleld (Guadeloupe) Geochemical monitoring during the stimulation operation Critical evaluation of chemical and isotopic field data Autlzored by B. Sanjuan, E. Lasne, M. Brach, L. Vaute December 1999 R 40646

3 Keywords :Bouillante geothermal field, Production test, Wells BO-2 and BO-4, Thermal springs, Stimulation test, Organic tracers, Scale inhibitor, Chemical, Isotopic and gas data. In bibliography, this report should be cited as follows : 5. Sanjuan A., Lasne E., Brach M., Vaute L. (1999) - Bouillante geothermal field (Guadeloupe). Geochemical monitoring during the stim~~lation operation. Critical evaluation of chemical and isotopic field data. BRGM report R p., 16 Figs., 14 Tables, 4 App. BRGM, This document may not be reproduced without the epress authorization of the BRGM. 2 BRGM Report R 40646

4 Abstract An operation of thermal stimulation during which around m3 of sea water have been injected into well BO-4, from loth to 271h August 1998, has been performed in the Bouillante geothermal field (Guadeloupe) in order to increase the production of this well. The aim of this study is to present and to discuss the analytical results obtained during the geochemical monitoring associated to this operation. During the stimulation operation, three organic tracers (sodium naphtionate, sodium benzoate and isophtalic acid) were injected with sea water into well BO-4. A geochemical monitoring was carried out from well-bo2 and three neighbouring thermal springs during a period of 45 days. Geochemical monitoring was also performed during the Reference and Evaluation Production Tests (RPT and EPT) accomplished in well BO-4. The multiple tracer test performed during the stimulation operation has shown and confmed that no direct hydraulic connection eists between the wells BO-4, BO-2 and the three neighbouring thermal springs even when a large volume of sea water is injected from BO-4. Both production tests designed to evaluate the modifications in BO-4 discharged fluid due to the stimulation operation were limited to single flow rate tests. The results obtained on brine, steam condensate and non condensable gases are similar in terms of phase proportions and chemical, isotopic composition in both RPT and EPT discharged fluid. The improvement of production by thermal stimulation had no significant consequences on chemical parameters of the total fluid discharged. According - to the results obtained from natural tracers (chloride, magnesium, - sulphate, deuterium and oygen-18 particularly) and organic compounds used during the stimulation operation, it can be observed that chemical composition of both RPT and EPT samples^is always little influenced by sea water. The percentage of remaining sea water can be estimated to 0.2% in discharged fluids collected at the end of the RPT and EPT. Consequently, it can be concluded that sea water is rapidly and completely mied with the reservoir geothermal fluid. This can be eplained by either a relatively significant circulation rate or a high water-rock ratio, which is in agreement with the isotopic oygen-18 signature of the deep geothermal fluid. The geochemical monitoring has allowed to better characterise the composition of the deep geothermal fluid from well BO-4 which was badly known before the two production tests. It has contributed to check and compare the previous results obtained for well BO-2. It has shown that the deep BO-4 fluid is very close to BO-2 one on a chemical, isotopic and physical point of view. The total proportions in weight of each phase of the fluid are similar for both wells. Some inaccuracies subsist on the values of BRGM Report R

5 key parameters such as alkalinity, ph, dissolved aluminium and sulphide concentrations and will have to be suppressed by additional works (accurate down hole samplings). The use of a scale inhibitor (IDOS 130) dunng al1 the stimulation operation has allowed to prevent anhydrite scale deposits predicted by modelling and observed during the RPT. These deposits probably occurred during the first interference test in March 1998 whereas only 100 m3 of sea water had been injected. Apart from the objective of the stimulation operation, the geochemical monitoring has shown that direct escapes of deep geothemal brine, mied with very small variable proportions of sea water after vaporisation and cooling, outcome from the littoral spring BO-3 Beach. The fumaroles situated near this thermal spring are probably the outcome of the stem resulting from the vaporisation of the deep geothermal fluid. During the humid penod (August - October), an additional contribution of superficial fresh water was obsewed in this spring. The geochemical monitoring of the thermal spring Tuyau, situated near the geothemal power plant, essentially constituted of fresh water (> 90%) with low amounts of deep geothemal brine, suggests an additional and increasing contribution of steam condensate. Presently, this contribution could have reached a stationaq state. According to Cl concentration and SD and F1'O values, it would represent about 45% of stem condensate. BRGM Report R 40646

6 Contents Introduction Description of the geochemical monitoring Tracer test associated to the stimulation operation Selected tracers Injection conditions Geochemical monitonng Production tests and down hole sampling in well BO Surface sampling Down hole sampling Geochemical monitoring results Tracer test associated to the stimulation operation Well BO Neighbouring themal springs Production tests and down hole sampling in well BO On site data acquisition dunng the Reference Production Test (RPT) down hole sampling On site data acquisition during the Evaluation Production Test (EPT) Discharge of solid materials during production tests Critical evaluation of Bouillante geothermal field data Chemical gas composition and gas ongin in wells BO-4 and BO Estimation of CO2 and HîS partial pressures in the reservoir conditions Estimation of the isotopic and chemical composition of the deep fluid Isotopic composition Chemical composition Use of chemical geothennometers Geochemical modelling and scale deposit nsks Conclusions 71 References BRGM Report R

7 List of Figures Fig. la - Fig. lb - Fig. 2a - Fig. 2b - Fig. 3 - Fig. 4 - Fig. 5 - Fig. 6 - Fig. 7 - Fig. 8 - Fig. 9 - Fig Fig Fig Fig Fig Fig Fig Topographic map of Bouillante area showing the location of three hot springs monitored during the interference test between the two wells BO-4 and BO A N150 cross section intersecting wells BO-2 and BO-4 showing the simplified geologic logs and the temperature profile measured in well BO-4 in Note than vertical and horizontal.. scales are simlar... 8 Detailed view of the sampling equipment for the collection of bnne samples after high pressure separator from well BO Detailed view of the testing loop and equipment used for fluid sampling during the production tests performed in well BO Well BO-2. Evolution of some on site measurements versus time Spring S1 (BO-3 Beach). Evolution of some on site measurements versus time Tracer test using sodium naphtionate. Analytical data obtained from S 1 spring... Spring S2 (Ravine Blanche). Evolution of some on site measurements versus time Tracer test using sodium naphtionate. Analytical data obtained from S2 spring Spnng S3 (Tuyau). Evolution of some on site measurements versus time Deuterium/Oygen-18 diagram (corrected values for weils BO-2 and BO-4) Oygen-18 values vs choride concentrations (corrected values for wells BO-2 and BO-4) Deuterium values vs choride concentrations (corrected values for wells BO-2 and BO-4) Bromide vs chloride concentrations (corrected values for wells BO-2 and BO-4) Tracer test using sodium naphtionate. Analytical data obtained from S3 spring Compared evolution of water flow-rate, conductivity and chloride content of water at weir bo during the Reference Production Test Evolution of conductivity, chloride and magnesium concentrations versus time and discharged volume during the EPT Evolution of AGA/Na naphtionate and benzonate concentrations versus time and discharged volume during the EPT BRGM Report R 40646

8 List of Tables Table 1 - Summary of the schedule of the stimulation operation Table 2a - Chemical analyses of brine samples collected from wells BO-2 and BO-4 before, during and after the BO-4 stimulation test Major species Table 2b - Chemical analyses of brine samples collected from wells BO-2 and before, during and after the BO-4 stimulation test. Trace species Table 2c - Chemical analyses of brine samples collected from wells BO-2 and BO-4 before, during and after the BO-4 stimulation. Infratrace species Table 2d - Isotopic analyses of fluid samples collected from wells BO-2 and BO-4 before, during and after the BO-4 stimulation test. Stable isotopes Table 3a - Chemical analyses of water samples collected from SI, S2 and S3 springs before, during and after the BO-4 stimulation test. Major species Table 3b - Chemical analyses of water samples collected from SI, S2 and S3 springs before, during and after the BO-4 stimulation test. Trace species Table 3c - Chemical analyses of water samples collected from S 1, S2 and S3 springs before, during and after the BO-4 stimulation test. Infratrace species , Table 3d - Isotopic analyses of water samples collected from SI, S2 and S3 springs before, during and after the BO-4 stimulation test. Stable isotopes Table 4 - Chemical composition of gases analysed in wells BO-2 and BO Table 5 - Summary of natural and artificial tracer analyses in brines collected from BO-4 weir bo (samples A) and in two stem condensates Table 6 - samples after separator (samples C) Isotopic analyses of fluid samples from wells BO-2 and BO-4. Corrected values Table 7 - Measured mean chemical data from the wells BO-2 and BO-4. Estimated chemical composition of the deep fluid. (DHS: Down Hole Sampling, LFSS: Low Flow Surface Sampling) Table 8 - ph, partial gas pressures and SI calculations performed from the BOS and BODP chemical compositions, using EQ3NR geochemical code List of Appendi App. 1 - Characteristics of the scale inhibitor IDOS : App. 2 - Geochemical monitoring results obtained during the multitracer test App. 3 - Geochemical monitoring results obtained during the RPT App. 4 - Geochemical monitoring results obtained during the EPT BRGM Report R 40646

9 Fig. la - Topographie map of Bouillante ara showing the location of three hot springs monitored during the interference test between the iwo wells BO4 and BO-2. Fig. Ib - A N150Q cross section intersecting wells BO-2 and BO-4 showing the simplified geologic logs and the temperare profile measured in wer BO-4 in Nore that veh'cal and horizontal scales are simiulr.

10 Introduction The research presented in this report was conducted as part of the project "Enhancement of productivity of high enthalpy geothemzal wells by cold water stimulation" in the framework of the CEC Thermie Programme. The scientist partners in this project were CFG, BRGM and ORKUSTOFNON. Financial support was provided by the Commission of the European Communities (contract N0GE/00194/97/FR/iS), by the Research Division of BRGM and by Geothermie Bouillante. The main objective was to cany out a thermal stimulation in well BO-4 (Bouillante geothermal field, Guadeloupe) by injecting cold sea water in order to increase its production. The analytical results obtained during the geochemical monitoring associated to the stimulation operation which occurred from loth to 27" August 1998 are presented and discussed in this study. A first tracer test using 30 kg of isophtalic acid (+ 15 kg of NaOH), 30 kg of sodium naphtionate and 25 kg of KNO3 dissolved in 1.15, and m3 of sea water respectively, had been carried out between the wells BO-4 and BO-2 at the end of 1996 (Herbrich, 1996; Sanjuan and Brach, 1997). A volume of 73 m3 of sea water had been then injected into well BO-4 to displace al1 the tracers within the reservoir. The geochemical monitoring performed from BO-2, during a period of 1 month, suggested the absence of direct hydraulic connection between wells BO-4 and BO-2 distant of about 400 m (Fig. 1). These first results were confirmed by a longer interference test carried out from march 27" to June 2nd 1998 (Sanjuan et al., 1998). 300 kg of sodium benzoate and 150 kg of fluorescein dissolved in 4 m3 of sea water were injected into BO-4 at a flow rate of 100 l/mn and pushed immediately after by about 77 m3 of sea water. This volume is approimately 2 times the interior volume of BO-4. Geochemical monitoring was performed from well BO-2 and three neighbouring thermal springs (BO-3 Beach or «Bord de mer», Ravine Blanche, Tuyau; Fig. 1). Analytical results indicated no change in the chemical fluid compositions caused by the injection of sea water into well BO-4. No iniected tracer was detected in BO-2 and in the thermal springs - - during - the monitoring period. Fluorescein was analysed on site using a spectrofluorimeter whereas sodium benzoate was determined by HPLC chromatography..(uv ~ detector) accompanied by a pre-concentration technique ii the BRGM lahoratories, at Orleans. Considering the results of the two first tests and knowing that the amounts of water injected into well BO-4 would be much larger in the stimulation operation, it was interesting to carry out a third tracer test associated to this operation. This test started on Saturday 15" August, after the injection of a volume of sea water around m3 in BO-4. Three tracers were injected into this well and the geochemical monitoring was performed from BO-2 and the three neighbouring thermal springs during a 45 day period BRGM Report R

11 (August 15'~ - September 30'3. Two other geochemicai monitorings (July 15" - 17" and October 31d - 10~) were conducted during the BO-4 discharge in the production tests carried out before and after the stimulation operation. These monitorings were performed in order to evaluate the chemicai modifications in BO-4 fluid due to the thermal stimulation. A fluid sample from BO-4 was sampled at the bottom hole on Saturday lst August, using a KUSTER sampler. BRGM Report R 40646

12 1. Description of the geochemical monitoring A summary of al1 the geochemical works performed during the stimulation operation is reported in Table 1. During al1 the stimulation operation, the scaie inhibitor DOS 130, developed by the Company REP (App. 1) and selected by CFG, was added continuously in the sea water at a concentration around 30ppm in order to prevent sulphate and carbonate scale deposits such as anhydrite and calcite at high temperature in well BO-4 (Sanjuan, 1998). At each moment, the injection rate of this inhibitor was adapted to the pump flow rate used for sea water injection. 1.l. TRACER TEST ASSOCIATED TO THE STIMULATION OPERATION 1.1.l. Selected tracers The ideai tracers should be inepensive, environmentally safe, conservative (non adsorptive, non reactant, heat-resistant), easily soluble, detectable at low concentrations, absent from natural geothermal fluids and ground waters. Because of the high natural background of the halides, the difficulty in obtaining permits for radioactive tracers, and the lack of diversity of the dyes, more tracers are needed for geothermal injection tests. A class of compounds that appears to be acceptable is the aromatic acids. These compounds have low natural background levels and are available commercially in many forms, many of which are non toic. In addition, several aromatic acids have been tested as groundwater tracers with great success and someone can be used at temperatures as high as 300 C (Adams and Davies, 1991; Adams et al., 1992). Consequently, isophtalic acid and sodium benzoate were selected. This last tracer was used in form of benzoic acid in association with fluorescein in a tracer field test conducted at the Diie Valley, Nevada geothermal system (Adams et al., 1989; Adams and Davies, 1991) where the maimum temperature in the eploited reservoir was approimately 250 C. Satisfactory results were obtained. The thermal stability of the third injected tracer, sodium naphtionate, is poorly known at hiph temperatures. Nevertheless, this tracer was selected because of its cost, high solubility (around 240 g/l), low detection lirnit and ease of analysis. Contrary to the other tracers and as for the fluorescein use in the previous test, sodium naphtionate was essentially chosen for its imrnediate detection because the spectrofluorimetry technique allows to obtain fast on site analyses. In contrast, the interpretation of results can be difficult and is only qualitative. Isophtalic acid and sodium benzoate were accurately analysed by HPLC chromatography. BRGM Repod R 40646

13 Table 1 - Summary of the schedule of the stimulation operation. Date Operation Injected volume, m3 (+) Discharged volume, m3 (-1 Comments July 1 51h - 17Ih Reference Production Test Geochemical monitoring in well BO-4, 1 sampling (B A, B, C) August 1%' Down hole sampling KUSTER sampler (sampling B04-98-F1 at a depth about 500 m) August 10'- 15' Starting of stimulation operation - Sea water injection Addition of scaling inhibitor (IDOS 130) - 30 ppm August 15Ih Tracer injection + 10 (Q = 600 Vmn) kg of sodium benzoate, kg of sodium naphtionate, kg of isophtaiic acid, Beginning of the geochemical monitoring in well BO-2 and neighbouring springs August 15" - 26Lbtimulation operation continuation - Sea water injection Addition of scaling inhibitor (IDOS 130) - 30 ppm, Geochemicai monitoring in well BO-2 and neighbouring springs August 26Ih Tracer injection kg of amino-g acid, Geochemical monitoring in well BO-2 and neighbouring springs August 26Ih Sea water injection + 76 Addition of scaling inhibitor (IDOS 130) - 30 ppm, Geochemical monitoring in well BO-2 and neighbouring springs August 27Ih Fresh water injection + 14 End of stimulation operation, Geochemicai monitoring in well BO-2 and neighbouring springs August 27Ih - October 3rd October 3'* - Continuation of tracer test 10~~valuation Production Test - 5 O00 Geochemical monitoring in well BO-2 and neighbouring springs Geochemical monitoring in well BO-4,5 samplings BRGM Report R 40646

14 (UV detector) in the BRGM laboratory at Orleans in order to quantify the different processes commonly observed in the case of a positive tracer test. Analytical procedures about al1 these techniques are reported in a technical BRGM report (Sanjuan et al., 1999). A special attention must be drawn to tracer photodegradability, ph range used for their analyses and contamination risks. The detection limits for sodium naphtionate, isophtalic acid and sodium benzoate analyses were 1, 50 and 10 ppb respectively. The precision of these methods is better than + 20%. Comparative analyses of sodium benzoate were performed using another HPLC chromatographic system and gave similar results Injection conditions The minimum quantities of tracer to be injected in BO-4 were estimated in two ways. The first, based on numerical sim~ilations of an injection test, was performed using the hydrogeological code CATTI, built by BRGM. In the second method, semi-empirical relations were considered. From the swept volume approach (Adams et al., 1993), the minimum tracer quantity required can be estimated from the equation: where: C,,,i, = tracer detection limit (103 kgil), Vinj = tracer mass (103 kg), D = distance separating wells (m), h = reservoir thickness = sum [producing zones] (m), 6> = porosity. Taking D = 400 m, h = 100 m and a porosity of 5% (fractured medium), it appears that a minim~im quantity around 30 kg of tracer is necessary for its detection. This estimation is in a good agreement with these obtained from other semi-empirical equations (involving volume considerations) and numerical sim~ilations. Considering these estimations and in order to increase the probability to detect these tracers, 300 kg of sodium benzoate, 200 kg of sodium naphtionate and 100 kg of isophtalic acid (+ 50 kg of NaOH) were dissolved in 10 m3 of sea water and injected in BO-4 at a flow rate of 600 I/mn. Tracer injection was performed using the reservoir tank of the pump located on BO-3 platform (Fig. 1) and the pipe line going up to well BO-4 installed for the stimulatioil operation. After the tracer injection, approimately m3 of sea water and 14 m%f freshwater were injected before closing the well. During the stage of tracer preparation and injection, fluid samples were collected in order to be able to analyse the aqueous major species and dissolved tracers in the case of their later restitution and detection. BRGM Report R

15 Bouillante geotherrnal field (Guadeloupe) Fig. 2a - Detailed view of the sampling equipment for the collorfion oj -...ze samples after high pressure separatorfrom well BO-2. Fig. 2b - Detailed view of the testing loop a d equipment used for fiid sampling during the production tests performed in weu BO-4. BRGM Report R 40646

16 Geochemical monitoring The detailed organisation of the geochemical monitoring is presented in a technical BRGM report (Sanjuan et al., 1999). Before the tracer injection during the test carried out in March-June 1998, 50 1 of fluid were sampled from well BO-2 in order to perform chernical and isotopic analyses of the initial fluid and to keep samples without tracers (blank samples). Al1 the water samples were systematically recorded after sampling and stored in the laboratory of the geothemal power station. a) Monitoring from well BO-2 Brine samples were collected after high pressure separator from a connection made in the pipe line and using a cooling system (Fig. 2a). Sampling temperature was about 35 C. High pressure stem condensate and non condensable gases were not monitored but were sometimes sampled. The automatic sampler equipped with 24 glass 300 ml bottles was especially used at the beginning of the monitoring and for night sampling. During the day, most of fluids was directly sampled. As epected, during the first five days after the tracer injection, a fluid sample was collected each 2 hours. After that, the sampling became progressively less frequent (a sample collected each 4 hours during the 15 following days, then each 6 hours and finally, 1 per day). If tracers had been detected, the frequency of sampling would have been increased. At each sampling, it was collected: - 60 ml of fluid in a brown polyethylene bottle for sodium naphtionate on site analyse; ml of fluid in an opaque glass bottle for sodium benzoate and isophtalic acid analyses in the BRGM laboratory at Orleans; ml of 0.45 pm filtered fluid in a polyethylene bottle for storage. Ecept for the temperature measurements, ail the on site analyses were carried out in the laboratory of the geothermal power station. Classical analytical techniques (pwehmeter, conductivity unit, Titration, Merck Colorimetry kits) were used. The precision of al1 these methods is better than a 5%. The determination of non conservative parameters such as temperature, ph, specific conductivity, Redo potential (Eh) and the other on site analyses (dissolved Ca, Cl, SiOz N&, H2S concentrations and alkalinity) were performed on an only sample per day, generally during the first manual sampling in the moming. Fluids conditioned for chemical and isotopic analyses in the BRGM lahoratories were also collected only dunng this sampling. Dissolved silica was analysed on brine diluted 10 times with mq water irnmediately after sampling in order to prevent the fast precipitation of amorphous silica due to the BRGM Report R

17 fluid cooling and stem separation. For the other determinations and for chernical analyses, 1 1 of additional brine was sampled in a polyethylene bottle. 100 mi of fluid were directly sampled in a polyethylene bottle for deuterium and oygen-18 isotopic analyses. Non conservative parameters and dissolved HS, NH4 concentrations were determined on non filtered and non acidified fluid. Specific conductivity, ph and Eh were measured at low but similar temperatures (close to 25 C) in order to be able to compare directly the analytical data and to follow their evolution. Dissolved Ca, Cl concentrations and alkalinity were analysed on 0.45 pm filtered fluid. For chemical analyses performed in the BRGM laboratones, it was collected: mi of 0.45 pm filtered fluid in a polyethylene for analyses of major anions and some traces; mi of 0.45 pm filtered and acidified fluid (HN03 Suprapur) for analyses of major cations and some traces. The samples collected for chemical and isotopic analyses as well as for determinations of dissolved magnesium (natural tracer of sea water), sodium benzoate and isophtalic acid were selected before to be sent at Orleans. if tracers had been detected, more samples would have been sent in the BRGM laboratories for more detailed analyses. b) Monitoring from the three thermal springs The fluids of the three neighbouring thermal springs BO-3 Beach, Ravine Blanche and Tuyau (Fig. 1) were also monitored. Ecept at the beginning of the monitoring, fluid sampling and on site analyses (in the laboratory of the geothermal power station) were canied out approimately each 2 days. For al1 the springs, it was sampled: of fluid in a polyethylene bottle for on site analyses and chemical analyses in the BRGM laboratories; - 60 mi of fluid in a brown polyethylene bottle for artificial tracer analyses (some samples were collected in glass bottles for comparative analyses); mi of fluid in a polyethylene bottle for deuterium and oygen-18 isotopic analyses. As for BO-2 brines, non conservative parameters, dissolved sodium naphtionate, Ca, Cl, NH4, Si02, HS concentrations and alkalinity were determined on site. Specific conductivity, ph and Eh were measured at low but similar temperatures (close to 25 C) in order to be able to compare directly the analytical data and to follow their evolution. Samples for chemical analyses in the BRGM laboratones were conditioned as those from BO-2. The chemical and isotopic analyses as well as the determinations of sodium benzoate and isophtalic acid were performed at Orleans on selected samples. BRGM Report R 40646

18 1.2. PRODUCTION TESTS AND DOWN HOLE SAMPLING IN WELL BO-4 During the two short term production tests carried out before and after the stimulation operation (Reference and Evaluation Production Tests, RPT and EPT respectively), a geochemical monitoring programme was performed (Table 1) in order to obtain detailed chemical and isotopic compositions of the fluids, and to assess any chemical modification in the course of the short term discharge. A down hole sampling in BO-4 was performed after the first production test (RPT). For the second production test, a fourth tracer (50 kg of amino-g acid dissolved in of sea water) was injected into BO-4 at a flow rate of mn one day before the end of the stimulation operation in order to cany out a back flow test. Approimately 100 m3 of sea water and 14 m3 of freshwater were introduced in BO-4 after the injection of amino-g acid Surface sampling When a geothermai fluid rises rapidly to the surface through a well, it boils to produce water and steam. Non-volatile components as silica remain in the liquid phase whereas some others as HS, CO2 or chloride enter the stem phase. Therefore, concentrations of ion or gas species in the total discharge are obtained by sampling and analyses of each phase. Between the well-head and the atmospheric separator, a cyclic sampling separator was installed with the intention of separating and collecting each phase of the fluid at the same time for a given separation pressure (Fig. 2b). According to the programme, the total fluid was sampled at surface conditions using a specific sampling tool inserted in the flow line. The weir bo sampling ailows to study the chemistry and scaling potentiai of the separated hot water in atmospheric pressure. Using a suitable cyclonic pressure mini-separator (geochemicai line), samples of separated brines, condensate and non condensable gases were collected. After flow rate stabilisation dunng both Reference and Evaluation Production Tests, GaslSteam Ratios (GSR) were determined from the on-line separator. In the two production tests, non conservative parameters (temperature, conductivity, ph, Eh, dissolved oygen), dissolved Cl, Ca, Si02, NH4 and HS concentrations and aikaiinity were determined on site in fluids (brines + steam condensate) collected at weir bo and separator. The frequency of sampling and measurements was adapted to the Production Test programme. Dissolved silica was anaiysed on brine diluted 10 times with mq water immediately afrer sampling. Non consemative parameters and dissolved BRGM Report R

19 HS, NH4 concentrations were determined on non filtered and non acidified fluid. Dissolved Ca, Cl concentrations and alkalinity were anaiysed on 0.45 Pm filtered fluid. During the Reference Production Test, only one sampling of fluids from weir bo and separator was collected for complete chemical and isotopic analyses in the BRGM laboratories because of the short monitoring duration. These samples were conditioned as those from BO-2. For non condensable gas analysis, airtight glass flasks were used. Some analyses of fiuorescein were performed on site. In the Evaluation Production Test, several samples were collected at weir bo and separator for complete chemical and isotopic analyses in the BRGM laboratories. Some non condensable gas samples were also collected. Sodium naphtionate and arnino-g acid were monitored and analysed on site by spectrofluorimetry in al1 the types of fluids (brines and condensates) and in fluids of the annulus. Some analyses of sodium benzoate were performed at Orleans Down hole sampling A down hole sampling using a KUSTER sampler was carried out on Saturday 1' August after the RPT but before the stimulation operation. Because of unepected problems with the sampler, only one sampling was successful (below the casing shoe at a depth around 500 m) and the quantity of collected brine was low (c 1 1). Consequently, al1 the planned analyses could not be performed. Non conservative parameters (temperature, specific conductivity, ph, Eh, dissolved oygen), dissolved Cl, Ca concentrations and aikalinity were determined on site. Samples were conditioned according to the required specifications for chemical and isotopic analyses in the BRGM laboratories. BRGM Report R 40646

20 2. Geochemical monitoring results Complete chemical and isotopic analyses of fluid samples were performed in the BRGM laboratones using standard water analytical techniques such as Titration, Potentiometry, Colorimetry, Ion electrode, Atomic Absorption Spectrophotometry, Ion Chromatography, Inductively Coupled Plasma-Mass Spectrometry (ICP-MS), Mass Spectrometry. The precision of major and trace species is better than t 5% and 15% respectively. The precision of deuterium, oygen-18 and strontium isotopes is 10.8%0, 2 O.l%o and rr (20) respectively. The ion balance values lower than 5% suggest a good quality and coherence of the analyses of aqueous major species TRACER TEST ASSOCIATED TO THE STIMULATION OPERATION Al1 the chemical data obtained on site are reported in Appendi 2. The results of the chemical and isotopic analyses carried out in the BRGM laboratories are presented in Tables 2 and 3. Analytical results obtained during the monitoring of the first tracer test (March-June 1998) also have been reported in these tables. The geochemical monitoring shows there is no substantial modification of the chernical and isotopic composition of the brines during the stimulation operation (App. 2 and Tables 2a,b,c,d). These results are similar to those obtained during the monitoring of the first tracer test performed in March-June 1998 (Sanjuan et al., 1999). The decrease in values of some parameters (ph, alkalinity, calcium concentration) observed for the sample collected on September 13'~ was due to a pressure drop at well-head. Ecept for this sample, from the careful monitoring of dissolved chloride, calcium and silica concentrations, alkalinity or non conservative parameters such as conductivity and ph (Fig. 3), it appears that the slight observed variations can be attributed to: - the analytical accuracy (c 5%); -the small changes (c 2%) in proportions of each phase (brine and steam) after fluid separation. These results are confirmed by the monitoring of the three artificial tracers. Among these last, no signal was detected. Al1 the analytical data were below the detection limits as duriug the monitoring of the fist tracer test (March-June 1998). After the injection of a volume of sea water of about m3 in BO-4 and a period of 45 days of monitoring from BO-2, it can be concluded there is no direct hydraulic connection between the wells BO-4 and BO-2. BRGM Report R 40646

21 Bouillante geothermal M d (Guadeloupe) Chloride (circles) and calcium (squares) concentrations conductivity (mslm~) SiO2 (mg) IWO W O 15/08/98 22/08/ /09/98 1îl /09/98 Conductivity (circles) and silica concentration (squares) ph (circles) and alkalinity (squares) Fig. 3 - WeZl BO-2 - Evolution of some on site measurements versus time. 20 BRGM Repofi R 40646

22 w n O Table 2a - Chemical analyses of brine samples collected from wells BO-2 and BO-4 before, during and after the BO-4 s -.. stimulation test Major species. Sample number B093-F2 B096-F2 B096-F2 (corr.) B098-F2-1 B098-F2-142 B098-F2-188 B098-F2-74OC B098-F2-402 B098-F2-470 BO96-F4 B04-98-F1 B A B B B A B Abis B A B B B A A B B Date May-93 (DHS) May-96 (SP) May-96 (SP) 27/03/98 14:OO (SP) 15/04/98 08:OO (SP) 24/04/98 08:OO (SP) 24/04/98 10:30 (SP) 26/08/98 08:OO (SP) 10/09/98 08:00(SP) May-96 (DHS) 01/08/98 17/07/98 (WB) 17/07/98 (SP) 10/05/98 14:OO (WB) :lO (WB) 10/06/98 08:00(WB) 10/06/98 08:00(SP) 10/10/9807:30(WB) 10/10/9808:30(WB) 10/10/98 08:30 (SP) T "C ph Eh,..,, mv Na mgll K mgil Ca mgll Mg mgll CI mgll IO : not analysed, DHS: Down Hole Sarnpling, SP: Sampling afier Separaior, WB: Sarnpling at Weir Bo Alk. meqll SO, mgll NO, mgll n.a < 1 < 1 < 1 < 1 < 0.5 <0.5 <5 < 0,5 <0,5 <0,5 <O,5 <0,5 <0,5 <0,5 <0,5 < 0,5 SiO, mgll TDS mgll I.B. % CUBr

23 NI NI Table 2b - Chemical analyses of brine samples collected from wells 80-2 and BO-4 before, during and after the BO-4 stimulation test. Trace species. Sample number BO93-F2 B096-F2 BO98-F2 (corr.) BO98-F2-1 B098-F2-142 B098-F2-188 B098-F2-74'C B098-F2-402 B098-F2-470 BO96-F4 B04-98-F1 B A B B B B B B Date May-93 (DHS) May-96 (SP) May-96 (Se 27/03/98 14:OO (SP) 15/04/98 08:OO (SP) 24/04/98 08:OO (SP) 24/04/98 10:30 (SP) 26/08/98 08:OO (SP) 10/09/98 08:OO (SP) May-96 (DHS) 1/08/98 09:OO (DHS) 17/07/98 (WB) 17/07/98 (SP) 10/06/98 08:OO (SP) 10/10/98 08:30 (SP) Br mgil F mgil B mgil I Po4 mgil < NH4 mgil ma NO2 mgil n. a. <0.01 <0.01 <0.01 < 0.01 <0.01 < < 0,Ol Li mgil Z Sr mgll Ba pgll Mn pg/l Fe mgll Al mgll <0.10 < 0.10 < 0.10 i < na.: not analysed, DHS: Down Hole Sampling, SP: Sampling afler Separator, WB: Sampling at Weir Bo

24 s" Stimulation. Infratrace species. 2 : 3 a 8 $ Table 2c - Chemical analyses of brine samples collectedfrorn welk BO-2 and BO-4 before, during and after the BO-4 Sample number BO93-F F2 B096-F2 (corr.) B098-F2-1 B098-F2-142 B098-F2-188 B098-F2-74 C B098-F2-402 B098-F2-470 B096-F4 B04-98-El B A B B B B B B Date May-93 phs) May-96 (SP) May-96 (SP) 27/03/98 14:OO (SP) 15/04/98 08:OO (SP) 24/04/98 08:OO (SP) 24/04/98 10:30 (SP) 26/08/98 08:OO (SP) 10/09/98 08:OO (SP) May-96 (DHs) 1/08/98 09:OO (DHS) 17/07/98 (WB) 17/07/98 (SP) 10/06/98 08:OO (SP) 10/10/98 08:30 (SP) Ag pg11 < 15 < 15 < 15 < 15 < 15 < 1 < 1 na. 4 <5 <5 < 5 < 5 As pgll n.a Be pg11 < 15 < 15 < 15 < 15 < 15 < 0.1 < 0.1 na. <0.1 <5 <5 < 5 < 5 Cd pgll 65 < 6 <6 < 6 < <2 <2 < 2 < 2 Co pgll < n.a Cr pg11 < 30 n.a < 15 < 15 < 15 < n.a 15 <5 <5 5 < 5 Cu pg Ni pg n.a n.a n.a Pb ~g11 n.a <6 <6 < 6 < nua <2 <2 Zn pg n.a < 15 < 15 < 15 < n.a < Cs pg Rb mgll : nit analysed, DHS: Down Hole Sampling, SP: Sampling after Separator, WB: Sampling at Weir Bo

25 P Table 2d - Zsotopic analyses offzuid samples collected from wells BO-2 and BO-4 before, during and after the BO-4 stimulation test Stable isotopes. Sample number B096-F2 BO96-F2 (corr.) B098-F2-1 B098-F2-142 B098-F2-188 B098-F2-74"C B098-F2-402 B098-F2-470 B04-98-F1 B A B B B C B Abis B B B C B A B A B B B C Date May-96 (SPJ M@y-96 (SP) 27/03/98 14:OO (SP) 15/04/98 08:OO (SP) 24/04/98 08:OO (SP) 24/04/98 10:30 (SP) :00 (SP) 10/09/98 08:00 (SP) 01/08/98 17/07/98 (WB) 17/07/98 (SP) 17/07/98 (SC) :10 (WB) :OO (SP) :OO (SC) :30 (WB) 10/ :30 (WB) 10/10/98 08:30 (SP) 10/10/98 08:30 (SC) 8"0 (%O) f0.1% D (%O) f 0.8% DHS: Down Hole Sampling, SP: Sampling afler Separator, WB: Sampling at Weir Bo, SC: Steam Condensate Sampling

26 Neighbouring thermal springs The analytical results of the geochemicai monitoring started since March 1998, during the first tracer test, were integrated to this study (App. 2). In opposition to the brine from BO-2, significant variations in the chernical and isotopic fluid compositions are observed for the thermal springs. a) Thermal spring BO-3 Beach (SI) The outflow of this spring, situated near sea (Fig. l), is not visible at shallow. At each sampling, a hole had to be dug to collect the fluid from this spring. Temperature is close to 100 C. For the geochemical monitoring, it was tried to sample the fluids always at the same place. Temperature of the sampled fluids varied from 73 to 91 OC. Fluid of this thermal spring had been sampled in May 1996 and studied by Sanjuan and Brach (1997). Its salinity was high (- 25 a); sodium and chloride were largely the dominant species. Its concentrations of dissolved Na, Cl, K, Mg, Br and its isotopic signature (deuterium, oygen-18, strontium) indicated a miing between sea water and a geothermal brine similar to that from well BO-2. Calculated proportions were approimately 34% of sea water and 66% of geothermal brine. Consequently, it was decided to incorporate this spring to the geochemical monitoring according to the possible connection with the geothermal hydraulic system. Results obtained on site (App. 2, Fig. 4) and in the BRGM laboratories (Tables 3a, b, c, d) during the geochemical monitoring show relatively large variations in the chemical and isotopic fluid composition. According to the magnesium and sulphate concentrations, it can be noticed a significant decrease in amounts of sea water between samples collected in May 1996 and in The observed variations seem to be essentially due to: - a quasi-direct sampling of deep geothermal brine afier vaporisation and cooling with very small variable proportions of sea water for the samples collected on lst and 25" April For these samples, Cl, Mg, S04, Br, B and Li concentrations are very close to those analysed in BO-2 and BO-4 brine samples collected after separator. According to the analysed chloride, calcium and magnesium concentrations, this same type of sampling was performed between on 1' April and 2nd June The fumaroles situated near this thermal spring are probably the outcome of the stem resulting from the leakage of the deep (partly-boiling) geothermal fluid or from the vaporisation of shallow aquifers; - an additional contribution of superficial freshwater for al1 the samples which indicate chloride concentrations lower than those estimated for deep geothermal fluid (around mgll), not noticed in 1996 and which is probably related to the rainy season. This is observed for numerous samples collected between on 25Ih August and on 2gth October Those collected on 7th and 281hOctober indicate the highest freshwater contributions. BRGM Report R

27 I Chloride (cireles) and caicium (squares) concentrations l Conductivity (circles) and süica concentration (squares) ph (cireles) and alkaüniiy (squares) Fig. 4 - Spring SI (BO-3 Beach). Evolution of some on site measurements versus time. BRGM Report R 40646

28 Lu Table 3a - Chemical analyses of water samples collected from SI, S2 and S3 springs before, during and afer the BO-4 s - stimulation test. Major species. Sample number BO9631 B098-SI-1 B098-S1-21 B098-S1-28 B098-SI-37 B096-S2 B098-S2-1 B098-S2-21 B098-S2-43 B098-S2-51 BO9643 B098-S3-1 B098-S3-21 B098-S3-29 B098-S3-30 Date May-96 01/04/ :OO 25/04/ : :OO 11/09/1998 1O:OO May :OO :45 25/08/ :20 11/09/ :35 May-96 19/04/ :OO 25/08/ /09/ / :OO T "C PH 6.80(74OC) 7.59 (25OC) 7.55 (27 C) 7.70 (3OoC) (25OC) 7.78 (27OC) 7.81 (25OC) (28 C) 6.25 (60 C) 6.27 (3OoC) 6.29 (29 C) 6,64 (25'C) Eh,., mv 126(74'C) 47 (25OC) 198 (27OC) -14 (30 C) (25OC) 137 (27OC) 17 (25OC) 110 O (28OC) -2 (60 C) 157 (3OoC) 42 (29OC) 91 (25 C) Na mgll K mgll Ca mgll Mg mgll CI mgll Alk. meqll SO, mgll NO, mgll < 1 < 1 <0.5 < SiO, mgll I TDS mgll I.B. % CVBr : not analysed

29 N Co Table 3b - Chemical analyses of water samples collected from SI, S2 and S3 springs before, during and after the BO-4 stimulation test. Trace species. Sample Date Br F number mgll BO9631 May B098-SI-1 01/04/ :OO B098-S /04/ : B098-SI-28 25/08/ :OO 35.8 < 0.1 ~098-S /09/ :OO 45.0 < '..." B096-S2 B098-S2-1 B098-S2-21 B098-S2-43 : not analysed B mgil PO4 NH4 NO2 Li mg11 mql1 ma. n. a < <0.1 < 0.1 < 0.1 < B098-S /09/ : < BO9643 B098-S3-1 B098-S3-21 B098-S3-29 May-96 31/03/199811:00 25/04/ :45 25/08/ :20 May-96 19/04/ :OO 25/08/ /09/ : <1 < < 0.02 < <0.1 < <0.1 < 0.1 A a au n. a Sr mgil Ba Pgll < 15 < Mn pgii < 15 < Fe mgll < Al mg11 < 0.10 < 0.10 < <0.10 <0.10 < 0.03 < 0.03 <0.10 <0.03 <0.03

30 Co 8 Table 3c - Chemical analyses of water samples collected from SI, S2 and S3 springs before, during and after the BO-4 S m I 2 stimulation test. Infratrace species. Sample number B096-SI B098-SI-1 B098-SI-21 B098-S1-28 B098-S1-37 BO9632 B098-S2-1 B098-S2-21 B098-S B098-S /09/ :35 < 1 7 <0.1 < BO9643 May-96 naa B098-S3-1 19/04/ :OO < < 15 <6 < 6 < 15 < 6 < 15 < 6 < 15 B098-S /08/ :50 < 1 27 < 0.1 < B098-S /09/ :55 < 1 28 < 0.1 < : not analysed Date May-96 01/04/ :OO 25/04/ :20 25/08/ :OO 11/09/ :OO May-96 31/03/ :OO 25/04/ :45 25/08/ :20 Ag pgll < 15 < 15 < 1 As pg/l < 1 < < 15 < 15 < 1 n.a <30 <30 7 Be pgll < 15 < < 15 < 15 <0.1 Cd ygll <6 <6 2 2 <6 <6 <1 Co ~gll n.a <6 <6 1 Cr pgll ma < 1 < 15 9 < 15 < 15 2 Cu pgll <6 <6 3 Ni pgll < 15 < 15 9 Pb pgll n.a <6 < <6 <6 5 Zn pgll n.a < 15 < < 15 < Cs pgll Rb mgll

31 W O Table 3d - Isotopic anabses of water samples collected from SI, S2 and S3 springs before, during and after the BO-4 stimulation test. Stable isotopes. Sample nurnber B096-SI B098-SI-21 B098-SI-28 B098-S1-37 B096-S2 B098-S2-21 B098-S2-43 sl'o (%O) f O.l%o B098-S /09/ :35 BO96S3 B098-S3-1 B098-S3-21 B098-S3-29 B098-S3-30 BOYS-S3-32 B098-S3-33 Date May-96 25/04/ :20 25/08/ :OO 11/09/1998 1O:OO 8D (%O) f O.8% May :45 25/08/ :20 May-96 19/04/ :OO 25/08/ :50 11/09/ :55 03/10/ :OO 07/10/ :OO

32 Sodium benzoate was analysed on 29 samples and the concentrations aiways were below the detection limit (C 10 ppb). In contrast, the concentrations of dissolved sodium naphtionate, analysed on site, appear to be close to 5-7 ppb before the injection of this tracer (Fig. 5, App. 3). Ten days after the injection, they show a significant increase that reaches values up to 31 ppb and decreases after. From October 7", a new increase is observed. It is difficult to eplain the values of sodium naphtionate concentrations observed during the monitoring of the first test before the injection of this tracer. Sodium benzoate cannot be used to confirm these values because its detection limit is higher (10 ppb). Two assumptions can be proposed: -an other species than sodium naphtionate is detected by spectrofluorimetry (interference problems); - analysed sodium naphtionate comes from the injection carried out in In this case, a very low hydraulic connection would eist between BO-4 and this thermal spring. The naphtionate degradation would be relatively slow (?) and it can be questioned if the tracers were injected in a deep level (high temperature) or in a more superficial aquifer (much lower temperature). Assurning a value of 2 I/mn for the flow rate of this spring and a mean value of 6 ppb for the naphtionate concentration, the tracer quantity recuperated from September 1996 to August 1998 is estimated to be about 12 g. Despite the naphtionate degradation, this amount significantly lower than that injected in BO-4 (30 kg), does not allow to reject this assumption and could confirm a very low interconnection with BO-4. In order to test the validity of the first assumption, additional analyses would have to be performed to confirm the presence of sodium naphtionate (characteristic spectra of emission and absorption) or to detect it by other methods (HPLC Chromatography with a pre-concentration technique, for eample). It can be added that this substance was never detected in the analyses of several sea water samples (< 1 ppb). The increases of sodium naphtionate obsewed in the fluid during the second interference test (Fig. 5) are probably related to the injection of this tracer. However, as naphtionate was injected from the BO-3 platform which is near the thermal spring, these increases could be caused by slight leaks that occurred from the reservoir tank during the injection and not by an interconnection between BO-4 and the thermal spring. It can be noticed that these increases in sodium naphtionate coincide with the additionai contribution of freshwater (decrease in fluid salinity). Using the same value as previously for the flow rate of the spring and a mean concentration of 25 ppb, the quantity of tracer recovered from August to October 1998 is estimated to 7 g. This quantity is negligible in relation to that injected in BO-4 (300 kg) and could be in good agreement with slight leaks during the injection. BRGM Report R 40646

33 Concentration (pgfi) ~ Sodium naphtionate injection! (15/08/98) i v ! & i ~ " " " ' " " ~ ~ ' ~ " " " ' " ~ " " " " " " ~ " " " " " " ~ " " " " " " " ' " ' " " " " i " " " " ' ~ " ~ " " ' ~ ~ ' ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0.0- m. m c! c! c! 3. c! 3 2 c! m * V) m P m d e e. - m ea m m m m m m e e e e e S. - * O\ P - N 2 2 O N N 3 $1 O O Time (days) 1 a O 2 O> Fig. 5 - Tracer test using sodium naphtionate. Analytical data obtained from SI spring.

34 Anaiytical data obtained for sodium benzoate (concentrations < 10 ppb), which was injected at the same time and place as sodium naphtionate, are not consistent with these last results. Consequently, results observed for sodium naphtionate must be confirmed (identification of the ernission and absorption spectra characteristic of sodium naphtionate in samples, use of other techniques) in order to draw appropriate conclusions. Nevertheless, most of results (sodium benzoate and fluorescein) suggests no direct connection with BO-4. b) Thermal spring Ravine Blanche-BO-4 (S2) This spring, located between well BO-4 and Ravine Blanche (Fig. l), forms a pool at its outflow. Temperature is close to 38 C and flow rate is very low. This spring was aiso sampled in May 1996 and studied by Sanjuan and Brach (1997). Fluid saiinity is about 1 gll. It is a Na-Cl-HC03 fluid, essentially constituted of superficial freshwater with a meteoric isotopic signature. Given its temperature and its proimity to well BO-4, very low proportions of geothemal fluid could contribute to its salinity although no geochemical evidence was found. The chemicai and isotopic compositions anaiysed during the monitoring of the first interference test are similar to that analysed in 1996 (App. 2 and Tables 3a, b, c, d). Significant variations were observed only after that fluid from BO-4 was poured out in the Ravine Blanche during the first production test (RPT - July 1998) and that a small part of this fluid was mied with water from the pool formed by the spring. This miing caused an increase in salinity and in most of dissolved species concentrations of the samples (Fig. 6). The chemical and isotopic composition slowly comes back to the initiai composition but has not still reaches it in September Benzoate sodium was analysed on 22 samples and the concentrations were always below the detection limit (< loppb). The concentrations of sodium naphtionate, anaiysed on site, are close to 2-3 ppb (App. 3). During the monitoring of the second interference test, it is clear that the injection of sodium naphtionate in August 1998 has no effect on fluid of this spring (Fig. 7). These results are in agreement with the absence of fluorescein detection during the monitoring of the first interference test. As for the spring S1, the values of naphtionate concentrations observed before the injection of this tracer are difficult to eplain and the two assumptions proposed for S1 spring can be applied to this spring. It can be again concluded that no evident direct hydraulic connection eists between BO-4 and this spring. C) Thermal spring Tuyau (S3) This spring, situated near the geothermal power plant, has flow rates varying from 2.5 to 3.8 llmn. Measured temperatures range from 55 (measured in 1963 by Tonani and reported in Sanjuan and Brach, 1997) to 74OC. Study carried out by Sanjuan and Brach (1997) has shown that salinity of the fluid sampled from this spring in May 1996 was about 2 g/l 'and the aqueous dominant species were Cl and Na. The chemical and BRGM Report R 40646

35 Boujllante geothermal field (Guadeloupe) I Chloride (circles) and calcium (squares) concentrations I Condudivity (ms/cm) Conductivity (circles) and silica concentration (squares) ph (circles) and alkalinity (squares) Fig. 6 - Spring SZ (Ravine Blanche). Evolution of some on site measurements versus time. BRGM Report R 40646

36 Lo D s m % 1 a 4 O 2 O> Concentration (pgh) Sodium naphtionate injection ( ) ~ Time (days) W ui Fig. 7 - Tracer test using sodium naphtionate. Analytical data obtained from S2 spring.

37 I Chloride (circles) and calcium (squares) concentrations I I 1 Condudivity (mslcm) sioz (W) Condudivity (circles) and silica concentration (squares) l ph (circles) and alkalinity (squares) I Fig. 8 - Spring S3 (Tuyau). Evolution of some on site measurements versus time. 36 BRGM Report R 40646

38 isotopic composition suggested that this fluid was essentially constituted of a freshwater (> 90%). However, contributions of sea water (- 2%) and of a geothermal fluid similar to that from BO-2 (- 5%) probably also occurred. Fluid temperature and flow rate measured in May 1996 were 60 C and 3.1 llmn respectively. Temperature measured during the monitoring of the two interference tests progressively increases from 68 to 74 C. Measurements of flow rate indicate two value ranges: llmn (from April 19" to July 10") and 3.8 Umn (from July 24" to October 1"). Since 1996, fluid of this spring and neighbouring zones indicate a progressive increase in temperatures. in contrast, a decrease in fluid salinity and concentrations of dissolved major species is observed (App. 3, Tables 3a, b, c, d, Fig. 8). This decrease is much larger before July 24" than after this date where the values become quasi-stable. For data obtained in 1998, this behaviour can be related to the evolution of the flow rate which increases before July 2dh and becomes stable after this date. Data observed in 1996 do not follow this trend. The monitoring of this spring would have to be continued. Additional data (in both dry and rainy season) are necessary to better understand the physical and chemical evolution of its fluid and to know if the observed evolution is seasonal or general. The increase in flow rate and the decrease in fluid salinity noticed up to July 24th could be eplained by a higher rainwater supply (rainy season) but it is difficult to interpret the observed increase in temperatures with this only eplanation. Another assumption could be an additional and increasing contribution of stem condensate which would have reached a stationary state. in this case, the evolution of such a contribution would have to be rapidly monitored to valid this assumption. A third assumption could be a combination between the two phenomena. The variations observed in the isotopic fluid composition indicate more negative values in 6D and 6180 for the fluid analysed in August 25'h. After this date, the values remain constant (Table 3d). This trend is in agreement with the previous observations and could be due to a contribution of steam condensate in the fluid of this thermal spring (depletion in heavy deuterium and oygen-18 isotopes). The relatively high concentrations of ammonium, volatile species, compared to those measured in the fluids of the two other Springs (Table 3b, App. 2), also would be in accordance with a contribution of stem condensate. The higher value analysed in the fluid smpled on April 19" could be due to a larger content of ammonium in the steam condensate. As deuterium and oygen-18 isotopes, chloride and bromide can be considered as consemative species in many cases and consequently, are often used as origin tracers. Figures 9, 10, 11 and 12 suggest that the sample analysed in April 1998 is probably a miing between sea water, geothermal fluid (BO-2 or BO-4) and superficial freshwater, as the fluid sampled in May However, a slight increase in the proportion of freshwater is observed in the sample analysed in Apparently, this additional contribution is mainly constituted of superficial freshwater but a small supply of a steam BRGM Report R

39 local meteoric waters rld meteoric waters SW evaporation (In closed sea) A Steam condensate O "0 (%O) Fig. 9 - Deuterium/Oygen-18 diagrain (corrected walues for wells BO-2 and BO-4).

40 Fig Oygen-18 values vs choride concentrations (corrected values for wells BO-2 and BO-4).

41 s? Fig. I l - Deuterium values vs choride concentrations (corrected values for wells BO-2 and BO-4).

42 condensate (< 3%) close to that analysed in well BO4 (BO C, BO C, B C) also can have occurred (Fig. 10). According to Figures 10 and 11, the fluids sampled from Aupst 25th are probably constituted of about: - 55% of fluid resulting from a miing between sea water, geothermal brine (BO-2 or BO-4) and superficial freshwater; - 45% of stem condensate close to that analysed in well BO-4. From an isotopic point of view, a contribution of local rain water (Sanjuan and Brach, 1997) cannot be completely discarded (Figs. 10 and 11). In contrast, the stem condensate represented by a sample collected in May 1996 from a fumarolle situated near the geothermal power plant cannot be used as one of the end members of miing (Figs. 10 and 11). Sodium benzoate was analysed on 4 samples. Concentrations were below the detection limit (< 10 ppb). The concentrations of dissolved sodium naphtionate, analysed on site, are close to 7-9 ppb (App. 3). During the monitoring of the second interference test, it is evident that the injection of sodium naphtionate has no effect on fluid of this spring (Fig. 13). These results are in agreement with the absence of fluorescein detection during the monitoring of the first interference test. As for the springs S1 and S2, the two assumptions can be proposed to eplain the values of naphtionate concentrations observed before the injection of this tracer. It can be concluded that no evident direct hydraulic connection eists between BO-4 and this spnng PRODUCTION TESTS AND DOWN HOLE SAMPLING IN WELL BO-4 In a geochemical point of view, the aim of the Reference Production Test (RPT) performed in July 1998 (15'~- 17'~) was to collect reliable and accurate physical and chemical data on separated brine, stem condensate and non condensable gas (ncg) fraction of discharged fluids from well BO-4 before the stimulation operations. The main objective of the Evaluation Production Test (EPT), camied out from 3'* to loth October, was to check any change in the fluid characteristics due to the injection operations On site data acquisition during the Reference Production Test (RPT) down hole sampling The last operation carried out on the well before the Reference Production Test was the injection of tracers (sodium benzoate and fluorescein in 4m3 of sea water) in March 1998 to check the eventuality of hydraulic connections between BO-4 and BO-2 wells. The tracers were pushed with 77 m3 of sea water. The well was opened on July 151h at 01:45 p.m. with a very tight opening of the master valve for the warming up of the well-head. BRGM Report R

43 a O 2 O> Fig Bromide vs chloride concentrations (corrected walues for wells BO-2 and BO-4).

44 % Concentration (pgfl) O s m 5 D O 2 0, Sodium naphtionate injection / ( ) œ - a : O. O ~ ' ~ " ' " " " ' ~ " " " " " ' ~ ~ " " " " ' " ~ ' ~ ~ ~ ~ m m m m 0: m s s e $ e 0: a oa s 4 e s 8 8 e S 3 * N N 3 2 O O O Time (days) - P W Fig Tracer test using sodium naplztionate. Analytical data obtained from S3 spring.

45 ConduetMty (squares) and chloride concentration (circles) vater flow rate (üh) Fluorescein (mgii) Fluorescein concentration (circles) and flow-rate (squares) Fig Compared evolution of waterflow-rate, conductivity and chloride content of water at weir bo during the Reference Production Test. BRGM Report R 40646

46 At weir bo, conductivity of the separated water was abnormally low at the beginning of the discharge. The three fust samples taken after the opening of the well were very low in conductivity and chloride concentration (App. 3). These samples were collected after a production of 80 m3 of fluid. Taking into account that only 81 m3 of sea water (chloride concentration around mg~l) were injected in the well before the test, this fluid dilution confirmed by fluorescein results is unepected (Fig. 14). Assurning the eventuality of a hole in the casing, a dilution by freshwater coming from a shallow aquifer could be responsible of this anomaly. This assumption seems to be consistent with a slight decrease in temperature observed at around 350m in the profiles temperature versus depth performed for well BO-4. In the following samples, conductivity and chloride content of separated water regularly increased and reached 4OmSIcm and m a respectively, after 21 hours of discharge. From sample006 (July 16" at ll:45 a.m.), conductivity and chloride concentration began to be stable. These values are close to those of deep geothermal fluid. A complete sampling of the fluid was carried out on July 17" at 9:00 a.m. (sample 009). Separated brine was sampled both in the weir bo and through the sampling separator. Steam condensate and non-condensable gas were also sampled through the separator. During the sampling, the thermodynamic conditions were the following ones: Total flow rate (WT) 39 th Well-head pressure 20 bar Flow-line temperature 150 C Flow-line pressure bar Separator pressure 1.3 bar The analyses on site on separated water and stem condensate gave the following results: July 17" at 9:00 a.m A (Weir bo) B (Separator) C (Steam condensate) Temp. (OC) ph Eh (mv) Conductivity (&/cm) Alkalinity (mesfi) Sulphide (mgnt The laboratory results (chernical and isotopic analyses) are given in Tables 2a, b, c, d. Knowing that 150 kg of fluorescein were injected with 81 m3 of sea water into well BO-4 and considenng the last available value of tracer content (17.7 mga; App. 3 and Fig. 14), it can be estimated that the discharged fluid is constituted of a maimum proportion of 1% of sea water. When the final sampling was perfonned, at least BRGM Report R 40646

47 P al Table 4 - Chemical composition of gases analysed in wells BO-2 and BO-4. Well BO4 BO4 BO4 BO2 BO2 (322111) BO2 BO2 BO2 BO2 Date /07/ CO2 (%) , (%) <0.005 n.d N, Ar He (%) (%) (%) < <0.005 < <0.005< H, (%) H,S (%) CH, (Yo) C,H, (%) < C3Hs O < n.d. Nz/Ar (vol.) n.d HelAr (vol.) n.d. n.d. n.d. n.d n.d. Tl (OC) T2 (OC) T3 (OC) n.d Tl : temperature calculated using the geothermometer and Panichi, 1980), pc02 = 1 atm T2 : temperature calculated using the geothennometer CO2iH21CH4 (Marini, 1987), pcoz = 1 atm T3 : temperature calculated using tlie geothermometer H2/Ar (Giggenbach and Goguel, 1989) : not analysed n.d. : not determined The atmospheric volume ratios N21Ar et HelAr are 84 et 5.7.1m4, respectively A volume ratio HelAr around 0.1 generally represents a magmatic arigin

48 Bouillante geotherrnal field (Guadeloupe) m3 of water were discharged from the well. The last magnesium and sulphate concentrations ( mga and 31 mga respectively; Table 2a) in comparison with those analysed in brines collected after BO-2 separator ( mga for Mg and mg/l for S04) seem to be high and are in agreement with the presence of a low proportion of sea water in the discharged fluid (around0.2%). However, when compared with the Mg and S04 concentrations determined in the down hole samples collected from wells BO-2 and BO-4 (Table 2a), it is not possible to conclude if there is a contribution of sea water. The similarity of the separation conditions and of the concentrations of the other species between brine samples collected after separator from wells BO-2 and BO-4 (Table2a) suggests that the first comparison is the most convenient. The GSR (GaslSteam Ratio) has been measured after the cyclonic separator. The weight ratio of non-condensable gases (NCG) in stem was estimated to %. Therefore, one ton of fluid discharged from BO-4 is composed of 707 kg of water, kg of stem and 1.5 kg of NCG. Chemical gas composition of the corresponding sample is reported in Table On site data acquisition during the Evaluation Production Test (EPT) During the production, 43 brine samples were canied out at weir-bo for on site analyses and 5 fluid samples were collected for laboratory chemical and isotopic analyses. Among these latter, two complete samplings from the separator for separated water, stem condensate and non condensable gas recovery were performed (samples B098-F4-231 on October 6" at 4:lOp.m. and B098-F4-244 on October 10" at 8:30 a.m.). The temporal evolution of the fluid at weir-bo is presented in Figure 15. The first 5 cubic meters of water discharged from BO-4 presented a low conductivity ( mslcm) and chloride content ( mga) corresponding to a miing between freshwater, sea water and probably deep geothemal fluid flowing from the aquifer through liner perforations at meters depth. Ecept for the three fust values, the general evolution of conductivity and chloride concentration curves (Fig. 15) is in good agreement with a miing between injected freshwater, sea water and reservoù fluid during the discharge of the 30 first cubic meters. The observed increase is related to the production of a more significant contribution of reservoir water from m, m, m production zones. After 6 hours of pumping (about 40 m3 of discharge), freshwater has been completely etracted from the well. Then, conductivity and chloride contents go up and stabilise to 38 mslcm and mg/l respectively. These values are close to those of deep separated geothemal bnne. Despite the injection of large volumes of sea water, it can be pointed out that the maimum measured chloride content was never supenor to m g. Consequently, it can be concluded that sea water is rapidly and completely mied with reservoir geothemal fluid. This can be eplained by either a relatively significant circulation rate or a high water-rock ratio. BRGM Report R

49 Conductivity (ms/cm)/ Magnesium (mg) cl (mgfi) 45,, Chloride (squares), conductivity (circles) and magnesium (triangles) versus time O O0 Conductivity (circles) and chloride (squares) evolution versus discharged volume (m3) Fig Evolution of conductivity, chloride and magnesium concentrations versus tirne and discharged volume during the EPT. BRGM Report R 40646

50 Bouillante geotherrnal field (Guadeloupe) The comple behaviour of the injected artificial tracers is difficult to eplain for several reasons. In theory, amino-g acid (AGA) would have to be detected at least after the first 100 m3 of discharged water. Sodium benzoate, sodium naphtionate and isophtalic acid would have to be observed after approimately m3 of discharged water. Analytical results (Table 5, Fig. 16) show this restitution sequence is not respected. The volumes of discharged water in which tracers are detected do not correspond to the volumes of water injected after their introduction into the well. This can be eplained by a comple fluid behaviour within the weil due to the presence of three main production zones, the use of different flow rates during the stimulation operation and resemoir heterogeneity (fissured reservoir). These factors lead to the absorption of injected water at different levels, following the discharge flow rate and the permeability of each level, which complicates the tracer restitution. Amino-G acid and sodium naphtionate have a similar behaviour. On site analyses indicate high values at the beginning of the discharge and a significant drop after the ' first m3 of discharged water. A slight increase is obsemed when 11 m3 of water are discharged. From this discharged volume, a regular downward trend can be noticed. The final measured values (for a discharged water volume of m3) are around 10 ppb. Because of the similarity between their wave length of ecitation and emission, interference were obsemed in the analyses of these two tracers. Tests in laboratory have shown that in a miture of amino-g acid and sodium naphtionate, the signal of this last substance is masked by that of amino-g acid. At least in the first 100 m3 of discharged water, amino-g acid is probably the only present tracer. The absence of sodium benzoate (injected with naphtionate) after 0.5 m3 of production is consistent with this assumption. Up to a production volume of 410 m3, the detection of naphtionate cannot be confirmed. From this volume where benzoate is detected (Table 5), it is probable that naphtionate concentration is largely higher than that of amino-g acid. After that, benzoate is detected up to the end of the monitoring. Given the similar amounts of injected benzoate and naphtionate, the large discrepancies observed between the concentrations of these two tracers (a factor about 150) remain uneplained (analytical problems?, adsorption phenomena in the well or reservoir?, comple chemical processes?). Additional detailed studies would be necessary to understand the behaviour of these tracers. Tests in laboratory have shown IDOS-130 inhibitor has no influence on the detection of the different injected tracers. Sodium naphtionate was detected in the two steam condensates collected on October 6th and loth (samples 231C and 244'2) at low concentrations (< 10 ppb). However, these concentrations are not negligible in comparison to those determined in the corresponding brines sampled at weir bo (Table 5). Apparently, this tracer is relatively fractionated between steam and brine after separation at temperatures superior to 100 C. Given the amounts of discharged fluid from well BO-4 and the analytical problems due BRGM Report R 40646

51 AGA - Na-Napht. (p@) Na-Benz. h g) 1 AGAiNa-Napht. (circlesltriangles) and sodium benzoate (squares) versus time 1 Amino-G-Acid concentration (pg) I O O0 Amino-GAcid concentration evolution versus discharged volume (m3) Fig Evolution of AGMNa-naphtionate and benzoate concentrations versus time and discharged volume during the EPT. BRGM Report R 40646

52 to the interference with sodium naphtionate, amino-g acid concentrations analysed on the two condensate samples are not probably representative and are rather close to O. The values of concentrations determined for sodium naphtionate are inferior to the detection limit of sodium benzoate analysis. The only determination performed for this last tracer on the sample 244C (Table 5) is in agreement with these values. As the sodium benzoate concentration in the corresponding brine sampled at weir bo (sample 244A) is relatively high, it can be concluded that this tracer is little fractionated after stem separation and is essentially present in the residual brine. On October 6", the flow rate was increased from an average total flow of 32 to 53 t/h. In the moming of October 7", the flow rate was adjusted to 44 th until to the end of the test. On October 6" and IO", two complete samplings from the cyclic separator were performed (B098-F4-231 and B098-F4-244 respectively). Sampling conditions for separation were the following ones: October 6" October loth Total flow rate (WT) 33 tnl 45 th Well head pressure 20.5 bar 21.0 bar Flow-line temperature 143 C 135 C Flow-line pressure bar bar Separator pressure 1.1 bar 1.1 bar For both samplings, on site analyses on separated water and stem condensate gave the following results: October 10'" at 8:30 a.m B098-F4-244A (Weir bo) B098-F4-244B (separator) B098-F4-244C(condensate) The laboratory results (chemical and isotopic analyses) are given in Tables 2a, b, c, d. For both samples, when compared with those determined in brines collected after BO2 separator, higher Mg and S04 concentrations (Table 2a) indicate a slight influence of injected sea water. This contribution is more significant in the first sampling (about 0.5 %, discharged volume around 1500 m3) than in the second one (about 0.2 %, discharged volume around m3). BRGM Report R

53 Table 5 - Summary of natural and arti.j?cial tracer analyses in brines collectedfrom BO-4 weir bo (samples A) and in two steam condensates samples after separator (samples C). Identification Number Date B09&F4402A B098-F4203A B098-F4204A B09?-F4205A B098-F4-206A B098-F-i-207A B098-F4208A B098-F4-209A B09&FMlOA B09&F4-211A.B098-FM12A B098-F4-213A B098-F4-214A B09&F4-215A B09&F4-216A B098-F4-217A B098-F4-218A B098.F4-219A BO98-F4-22OA B09&F4-221A B09&F4-222A B098-F4-223A B098-F4-224A B098-F4-225A BO98-F4-226A BO98-F4-227A B09&F4-228A B09&F4-229A B098-F4-230A B09&F423OAbis?8-F4-231A -8-F4231C B09&F4-232A B09&F4-233A BWSF4-234A B098-F4235A B098-F4436A B098-F4-Z37A B098-F4-238A B098-F4-239A B09&F4240A B098-Fb241A B098-F4-24ZA B09&F4-243A B098-F4-244A B098-F4-244C O : :OO : : :45 03/10/98 ll: : :OO O :30 O O :OO :OO : :OO :OO 04110/ O : :30 OS :OO O O O :OO :OO :OO O /10/98 07: : :OO : :45 09/ :OO 09/10/ : :30 AGA rd Na-Napht. rsn Tracers Na-Benz. rsn C C 10 Mg mg CI mm Physicochemistry ph Cond. mstcm Discharge Volume m3 O BRGM Report R 40646

54 Moreover, sodium benzoate concentrations measured before and after the first sampling decline from pg/l in samples B098-F4-230A, B098-F4-230Abis to pgll in sample B098-F4-233A (Table 5). For the second sampling, the decrease in sodium benzoate concentration from to pg/l is consistent with that observed for dissolved magnesium concentrations (Table 5). When final sample is collected, approimately m3 of water have been discharged. The total flow, taking into account stem production, is m3. This volume is less significant than total injected water volume during the stimulation operation. Conductivity measurements performed in fluids collected from annulus show values corresponding to low salinity waters (App. 4). A progressive decrease in conductivity is observed up to the sample 222D. After this sample, al1 the other values are close to 20 ps/cm. These low values associated to acid ph (< 5) are representative of stem condensates. The first fluid samples, more mineralised, were probably contaminated by a small contribution of another fluid with an unknown composition but with a higher salinity. Artificial tracers (sodium naphtionate, amino-g acid, sodium benzoate) were practically detected in al1 the fluids collected from annulus (App. 4). The behaviour of these tracers and their restitution sequence are very difficult to interpret and are complicated by the problems of analytical interference between sodium naphtionate and amino-g acid. As for the brines collected at weir bo, the apparition of these tracers do not correspond to the volumes of discharged water from which they would have to be detected. So, contrary to the analytical results, sodium naphtionate and benzoate would not have to be observed in the analysed samples. Amino-G acid which was injected towards the end of the stimulation operation would have to appear approimately after the sample 219D. The significant discrepancies found between sodium naphtionate and benzoate concentrations (tracers injected at the same time) are also badly understood. The presence of artificial tracers in fluids collected from annulus is probably related to a comple behaviour of these tracers dunng their injection. The eventuality of a hole in the casing could have a main influence on this behaviour but other unknown processes can also be responsible Discharge of solid materials during production tests During the Reference Production Test (RPT), solids were discharged with geothermal fluid from well BO-4. Grey-colored, sand-size materials filled the separator and weir bo. -Ray diffractometry indicates that these solid material are predominantly formed with anhydrite (CaS04). Other subordinate phases are calcite (CaC03), halite (NaCl), oypsum (CaS04.2H20), and clays (smectite, serpentine). No trace of rock-forming D BRGM Report R

55 silicates (mainly plagioclase and pyroenes) coming from the reservoir formations (lavas and tuffs) has been found. On the other hand, there was no abundant solid material discharged with fluids during the second production test carried out after the injection of cold sea water in BO-4. However, scaling was observed on the walls of the weir bo. These were mainly amorphous silica with low quantity of halite, anhydrite, magnetite, calcite, sylvite. Amorphous silica and magnetite formation results from the cooling of the geothermal fluid and stem separation. Halite, sylvite and anhydrite precipitation is due to localised evaporation processes. Calcite is probably formed by CO2 degassing at the surface. The origin of solid materials discharged during the RPT and the predominance of anhydrite must be discussed by taking into account the behaviour of sea water mied with hot geothermal fluids. Indeed, in 1996 and 1998 before the RPT, two tracer tests were carried out in well BO-4 using sea water for dissolution of chemical tracers and injection into the reservoir. The volume of sea water injected in BO-4 was about 70 m3 each time. In 1997, during the change of the BO-4 master valve, a volume of 100 m3 of sea water was also injected in order to keep the well under control. The occurrence of solid materiais containing anhydrite as a predominant phase discharged during the RPT is probably related to the heating of cold sea water and miing with the deep geothennal fluid (Sanjuan, 1998). We shall see this fluid is close to equilibnum with anhydrite in reservoir conditions. Fluid movement within the reservoir and discharge through the well bore displaced the crystallised anhydnte deposits and allowed the progressive cleaning of the well and near well reservoir. An isotopic determination of 634~ carried out in an anhydrite deposit collected during the RPT gives a value of 18.8 t 0.3%0. This analysis was performed by dissolving 100 mg of sample, constituted essentially of anhydrite with low arnounts of calcite, gypsum, halite, and smectite, in a 0.5M HCl solution. The solution was filtered and sulphate was precipitated in form of Bas04 using a Bac12 solution. The corresponding solid was treated following the usual protocol. As we shall see it, the found value is close to that measured for sea water in this area and for the brines collected from BO-2 after separator. This is in a good agreement with an anhydrite precipitation resulting from a miing between cold sea water and deep geothermal fluid. The lack of solid materials discharged during the second production test is ascribed to the addition of scale inhibitor (IDOS 130) in the injected sea water. BRGM Report R 40646

56 3. Critical evaluation of Bouillante geothermal field data Another aim of this study, through a critical evaluation of the obtained chemical data, is a best understanding of the local deep chemical and thermodynamic conditions. Availability of a large amount of field and laboratory data on the chemical and isotopic compositions of fluid produced by wells BO-2 and BO-4 allows the application of geochemical techniques to reservoir engineering. The reconstruction of the total composition of the fluid resemoir can be compared to the local measured or known physical and petrological conditions. 3.1.CHEMICAL GAS COMPOSITION AND GAS ORlGlN IN WELLS 80-4 AND 80-2 In the Fust sampling, GasISteam Ratio was measured twice and estimated to 0.51 and 0.49% in weight. In the second sampling, estimation of GasISteam ratio gave 0.55% in weight. Corresponding gas compositions are reported in Table 4. According to Demians d'archimbaud and Munier-Jolain (1975) and a GSR measurement performed in July 1998, the GasISteam Ratio can be estimated to 0.4% in weight for well BO-2. In well BO-4, from al1 the measurements carried out in 1998, this ratio is similar but seems to be slightly higher (- 0.5% in weight). Al1 available gas data conceming wells BO-4 and BO-2 have been reported in table 4 in volume percent. The samples collected on October 10" 1998 from BO-4 and in 1988 from BO-2 are considered as the most representative of these wells. In these samples, the Hz contents are the highest and the N2 and O2 contents are the lowest, which indicates a minimal contamination by atmospheric gases during the sampling. Chemical gas composition of these two samples is similar and is largely dominated by COz gas (94-95%). A commou origin is assumed. Considering a production of 120 tonsih of brine, 30 tonsih of stem and a degassing of 120 kgih of non condensable gases (0.4% in weight of steam) for well BO-2, it cau be estimated that about 115 kg of COz, 2 kg of H2S and 3 kg of N2 per hour are rejected in atmosphere. For well BO-4, during the Evaluation Production Test, an average production of 36 tonsk of brine, 14 tonsih of steam and a degassing of 70 kgih of non condensable gases (0.5% in weight of steam) were considered. The amounts of non condensable gases rejected in atmosphere per hour are then approimately 67 kg of CO2, 1.5 kg of H2S and 1.6 kg of N> The N21Ar volume ratios are close to the atmospheric value (84). Given the low oygeu contents (< 0.1 %), this ratio is probably the result of the presence of atmospheric N2 and BRGM Report R

57 Ar in resewoir. The only isotopic carbone-13 analysis that was performed in the BO-2 fluid indicates a value of -2.6%0. This value is far from typical data of basaltic magmas (close to -7%0). It is rather in agreement with a sedimentary origin of C02, produced by metamorphic reactions at high temperature and pressure. However, it could also be result from a mied origin of CO2 (sedimentary, magmatic and atmospheric origin). This assumption seems to be the most reliable. Indeed, it would be in agreement with the values obtained with the -/Ar volume ratios and the water miing that occurs in the resewoir. Gas geothermometers, applied on the selected BO-4 and BO-2 samples, give temperature values ranging from 157 to 251 C (Table 4). Most of these values are lower than those measured in reservoir ( C). According to Allard (personal communication, 1996), a significant He content would have a mantellic origin that would be related to deep fluid rises through tectonic fissures (3~eP~e ratio measured in BO-4 from 3 to 4 times higher than the atmospheric ratio). These thermal and gaseous leaks would be controlled by regional tectonics rather than by a relatively small magmatic etmsion. However, the value obtained in BO-4 is about 2 times lower than the value measured at the top of Soufrière. This suggests a miing with crustal helium. 3.2.ESTlMATlON OF CO2 and H2S PARTIAL PRESSURES IN THE RESERVOIR CONDITIONS ph measurements at surface in BO-4 and BO-2 brines cannot be considered as representative of the deep fluid in reservoir because of the degassing (essentially CO2 gas) and cooling. Consequently, the ph of deep fluid must be reconstructed in order to be able to cany out geochemical Saturation Inde calculations in resewoir and to predict possible scaling deposits in the wells after fluid separation and cooling at the surface. One of the parameters need for ph reconstruction is the CO2 partial pressure. According to D'Arnore and Truesdell(1985), the following equation: log P(i) = log (n(i)/n(hzo)) + log B(i) + log P(H20) can be used to calculate the partial pressure P(i) for each gas i at a given temperature in equilibrium with a single liquid phase. B(i) is the distribution coefficient between steam and liquid (Giggenbach, 1980; D'Arnore and Tmesdeii, 1988) and n(i) is the number of moles of gas i. At any temperature, P(i) is a function of the measured molal concentration of a gaseous species with respect to total H20. This means to consider the gasl(total H20) ratio measured at the discharge point to be representative of the value in the resewoir fluid. Given the speed of C02, H2S reactions with water at high temperature, these gases can be considered to have reached equilibrium with fluid in the reservoir. BRGM Report R 40646

58 Values of B(i) and P(H20) refer to reservoir conditions of temperature and salinity. Considering that the TDS values for BO-4 and BO-2 fluids were not very high, these two parameters were not corrected in function of salinity. At 240 C, P(H20) = 33.5 bar [log P(H20) = and log B(CO2) = 2.14 (Giggenbach, 1980). At 250 C, P(Hî0) = 39.7 bar [log P(H20) = and log B(CO2) = 2.03 (Giggenbach, 1980). Considering a fluid separation of 80% of brine and 20% of steam, a quantity of non condensable gases of 0.4% in weight of stem and the chemical gas composition of the sample collected in 1988 from BO-2 or that of the sample collected in October 1998 from BO-4, it is found: P(C02) = 1.4 bar at 240 C and P(C02) = 1.3 bar at 250 C respectively. In this case, temperature effect is insignificant on P(C02) values. If P(H20) is set to 42 bar at 242"C, value selected by Abou Akar et al. (1992) in geochemical simulations for well BO-2, we obtain a P(CO2) close to 1.8 bar. This value can be considered as representative of the reservoir conditions. The estimation of ph value for the deep brine will be performed and discussed in chapter As for CO2 gas, HîS partial pressure also can be calculated. At 240 C and 250 C log B(H2S) is equal to 1.70 and 1.60 respectively (Giggenbach, 1980). P(H2S) is estimated to be close to bar at 240 or 250 C. If P(H20) is set to 42 bar at 242 C for well BO-2 (see above), results become: P(H2S) = bar. Consequently, a P(H2S) value of bar can be considered as representative of the reservoir conditions ESTIMATION OF THE ISOTOPIC AND CHEMICAL COMPOSITION OF THE DEEP FLUlD lsotopic composition It must be observed that fluid samples were collected from wells BO-2 and BO-4 after fluid separator or at weir bo and then 6"0 and 6D were strongly fractionated. The following equations must be used to recalculate the original values of the deep fluid (DF) from the separation temperature mass balance: 6"0DF = g 6"0B +YS 618~s SDDF = B 6Ds + YS 6Ds BRGM Report R 40646

59 VI 03 Table 6 - Isotopic analyses offluid samples from wells BO-2 and BO-4. Corrected values. Sample number B096-F2 (6, calculated) B096-F2 (6, measured) B098-F2-1 B098-F2-142 B098-F2-188 B098-E2-74"C 8098-F2-402 B098-F2-470 Mean Standard deviation B B A B B B C B Abiis B B B C B A B A B B B C Mean Standard deviation Date May-96 (SP) May-96 (SP) 27/03/98 14:OO (SP) 15/04/98 08:OO (SP) 24/04/98 08:OO (SP) 24/04/98 10:30 (SP) 26/08/98 08:OO (SP) 10/09/98 08:OO (SP) 01/08/98 09:00 (DHS) 17/07/98 (WB) 17/07/98 (SP) 17/07/98 (SC) 10/05/98 16:lO (WB) 10/06/98 08:OO (SP) 10/06/98 08:OO (SC) 10/10/98 07:30 (WB) 10110/98 08:30 (WB) 10/10/98 08:30 (SP) 10/10/98 08:30 (SC) ~eparation temper. (OC) Brine (6) proport. (%) tea am (s) proport. (%) "0, ( O ) S"O~ (%) ô18~p SD, (%) ((%o O. 1 6Ds (9k) ÔDDF (%O), O. 4 DHS: Down Hole Sampling, SP: Sampling afler Separator, WB: Sampling at Weir Bo, SC: Steam Condensate Sampling Only measured values are represented in bold type

60 B is the mass proportion of brine obtained at separation temperature T, ys (= 1 - B) is the mass proportion of stem obtained at separation temperature T, &"OB and SDB are the isotopic values determined for the brine samples, 6180s and 6Ds are the isotopic values measured on the corresponding steam condensate samples. When 6"os and 6Ds values are not determined, it is possible to calculate them from measured 6"0~ and ~ DB, using the following fractionation equations: where co and ad are the fractionation factors between liquid and stem phase at separation temperature. The values of these factors at 100, 120 and 160 C, which were the separation temperatures, were taken from Amason (1977) (in Ferronsky et al., 1982). Corrected values of 6180~~ and ~DDF for wells BO-4 and BO-2 are reported in Table 6. The proportions of brine and steam after separation and the separation temperature used for correction also are presented in this table. The separation temperature for BO-4 separator was estimated by comparing the values of 6180s and 6Ds obtained from the fractionation equations for samples B and the values directly measured on the stem condensate in samples C. A value of 120 C was selected. This value is slightly lower than those measured at separator ( C) but similar to sampling temperature. For BO-4 bnnes sampled at weir bo (samples A), a separation temperature of 100 C was taken. For well BO-2, the separation temperature in the high pressure separator is close to 160 C. Al1 the values corrected from fluids sampled in wells BO-2 and BO4 are similar (Table 6). For each well, standard deviations are close to analytical precision. These values are consistent with those estimated for deep BO-2 fluid in May 1996 (Sanjuan and Brach, 1997) and those directly measured on the down hole sample performed in August 1998 in well BO-4. They also are in agreement with the assumption of the eistence of a deep fluid constituted of a miing of sea water and freshwater in similar amounts for each well (Fig.9). Proportions of each end-member will be accurately determined using chemical data. Isotopic data obtained during the geochemical monitoring relative to the production tests camied out before and after stimulation works in well BO4 show no major differences. Isotopic 87~r/86~r ratios measured on BO4 fluid collected after separator and at weir bo during the geochemical monitoring relative to the RPT indicate values of and e respectively. 87~r/86~r ratio of the rocks surrounding reservoir (andesite or basalt) was estimated to by Sanjuan and Brach (1997). It can be BRGM Report R

61 concluded that water-rock interaction is close to an equilibrium state. The values observed for BO-4 fluid samples slightly higher than that measured on the brine sample collected in May 1996 from well BO-2 after high pressure separator ( ; Sanjuan and Brach, 1997) could reflect a very small additional contribution of sea water. This assumption is consistent with dissolved magnesium and sulphate concentrations and oygen and deuterium isotopic data which show very slightly higher values for BO-4 fluid samples and then closer to sea water (Table 6). The only isotopic determination of S ~ performed ~ S in dissolved sulphate of a brine collected from BO-2, after high pressure separation, indicates a value of %0, close to that measured for sea water in this zone ( %0). As for oygen-18 isotopic values determined in BO-4 and BO-2 brines, which do not indicate an enrichment in heavy isotopes due to water-rock interaction processes, this suggests a fast circulation rate of the deep fluid or large water-rock ratios in reservoir Chernical composition As fluid samples were collected from wells BO-2 and BO-4 after fluid separator or at weir bo, corrections using mass balance equations considering separated phases (liquid and steam) must be performed to reconstmct the chemical composition of deep fluid (DF). It can be written for the concentration of each species : where B and ys (= 1 - B) are the respective mass proportion of brine and steam obtained at separation temperature and pressure. These values calculated from the mean analytical data obtained for each well are reponed in Table 7. Ecept for sulphide and ammonium, the specie concentrations in steam condensate can be considered as negligible with respect to the concentrations determined in brine. For both wells, sulphide and ammonium concentrations are higher in stem condensate than in brines (about mg11 instead of mg/l for total sulphide and 4-5 mgil instead of 1-2 mg/l for ammonium). The steam condensate is acidic (ph about ) whereas neutral conditions are found for brines (ph ). It can be pointed out that if H+ activity of the deep fluid is calculated using the above mass balance equation, a value of ph ranging from 5.0 to 5.1 is found. We shail see this value is close to these estimated from P(CO2) and alkalinity. On the whole, the chemical compositions reconstructed for deep fluid by a statistical approach using chemical data obtained in brine and condensate samples collected after separator or at weir bo from wells BO-2 and BO-4 (Table 7) are similar. This fluid is essentially constituted of dissolved chloride and sodium with a TDS of about20 gll. Ecept for arsenic concentration which is about 10 times higher than the admissible BRGM Report R 40646

62 ~ - ~ ~ Table 7 - Measured mean chemical datafrom the wells BO-2 and BO-4. Esîimated chemical composition of the deep fluid. (DHS : Down Hole Sampling, LLFSS : Low Flow Surface Sampling). A k mg/l HCO, so, mgd ES mg/l NO, mgfl SiO, mgil PO4 mg/l Na< mg4 NO2 m l Br mgil CüBr F mg/l B Li mg/l mgil Sr mg/l Fe mgil Al mg/l Ba p ' Mn pg/l Ag p l As g/l Be p l Cd pg/l Co p l Cr p Cu 1 Ni p l Pb.pg/l Zn pg/l Cs p Rb p 22.8 / j j 0.10 < , / ! 0.12 <0.01 i i j ,0.007,0.040~ ! 126 < 1 ; <0.1! f j j 13,6 I <6!. < 15 i 325 j f i! 34.5j [ j <0.01f 48.3! ( 21.8 j 6.18 f 22.8 j 0.410! i ' 5560 j < <5 i <2 i 33 j < ! i na. n.a j 27.2 ( 2.7 : i 30.97~ ! <' f 1.20 i < 0.01 i 52.0' ! f f j : f 381 ' <5 571 f 33 <5 i < i 5 j 28.0 i i 8.6 <z ] f : not analysed Onty the two chernical compositions reported in bold type (BOS and BODP) were used in the geochemical rnodelling n.e ne 457 na na na na n.e na na na na ne na ria na na na. na j j 0.3 : <0.4 i 474 i ~0.01 i j 7 ' 1.0 j ' f ! i '1 i i <0.1 j ' / i 10.9 c5 i <15 / < n.a na 658: na n.â na na na n.a 309 na. na. na n.a n.a na na. n.a ma na sz5 '. na 1.50 na ne < 15 n.e < <3û < na 770 na ne ! i ( 4.3. < 0.4 i..445 f <0.01 f i 0.9 f 15.7 f i ' ' c <4 j < j < j [.49.0 j 4.3 i na. ; i i 23.2 j j < 0.4 j < ! j f 'jo.0~ !0: f 285' 4032 j 200 <4! <4 j ' ! <4 i 21.0 i f < na 'na pa n.a na. 276 na. na. na. na. n.a iia ' 1499 ' na. < <O.l 2.40 n.a <

63 concentration in naturai waters (50 ppb), results show that the concentrations of most of trace species in deep fluid are relatively low. Only the chemical compositions obtained from down hole fluid samples collected from BO-2, at a depth of 322 m, in 1993 and from BO-4 in May 1996 (at a depth of 650 m; Sanjuan and Brach, 1997) and in August 1998 can be considered as representative of deep fluid without changes due to processes of phase separation and degassing. Ecept for the sample collected in May 1996 (Table2a), the concentrations of dissolved chloride, sodium, potassium, bromide, boron and fluoride of the two other samples are close to those estimated from the fluids collected at surface by statistical approach (Table 7). Discrepancies confirmed by previous analyses of down hole samples (Table 7) are observed for aikaiinity, concentrations of dissolved calcium, magnesium, sulphate, silica and numerous trace species (Sr, Mn, Ba, Li, Fe, Al, Ag, Cd, Co, Cr, Cu, Ni, Pb, Zn). The low value of dissolved silica analysed in the down hole sample collected in 1998 is due to the sampling. As the sample was not diluted with mq water immediately after sampling, precipitation of amorphous silica occurred after fluid cooling. In order to eplain the observed discrepancies, two assumptions can be considered: - a low precipitation of minerais such as carbonates (calcite with Mg, S04, Sr, Ba, Ni?, Pb?, Zn?), polymetallic sulphide (pyrite + sulphide of Cu, Zn, Ni and other metals), clays and amorphous silica during the rise of the deep geothermal fluid could be responsible of the decrease in the concentrations of these species in the bnne samples collected after separator or at weir bo from wells BO-2 and BO-4; - a contamination during the sampling, due to the presence of minerai deposits with the water sample (carbonate and clays) and to the KUSTER sampler (mainly metal ions). These perturbations will be discussed in more details in the chapter From the concentrations of conservative parameters such as chloride and bromide (Fig. 12), it appears that the deep geothermal fluid is constituted of 58-60% of sea water and 42-40% of freshwater for each well. These results are in agreement with previous estimations (Sanjuan and Brach, 1997), concentrations of other species and isotopic data (Figs. 10 and 11). They are in disagreement with those obtained in the down hole fluid sample collected from well BO-4 in 1996 for which it had been estimated an higher proportion of freshwater (46 against 42%; Sanjuan and Brach, 1997). The Iower proportion in sea water observed for the down hole sample collected in 1996 can be eplained by several assumptions: -the presence of a very smail contribution of sea water injected during the tracer and stimulation operations. Slightly higher values observed in magnesium, sulphate concentrations and in strontium, deuterium, oygen-18 isotopic values with respect to BO-2 brine seem to confirm this assumption. However, using magnesium and sulphate BRGM Report R

64 concentrations, it has been calculated that this contribution is inferior to 0.5% even for the highest magnesium concentrations; -the e~istence of an additional contribution of freshwater in the sample collected in No production test and consequently no significant fluid circulation had been performed before sampling. This additional contribution is no doubt predominant in comparison to the influence of sea water injected during the stimulation operation. It must be remember that an uneplained contribution of additional water coming probably from a superficial aquifer is suspected at the beginning of the monitoring of the first production test performed in well BO-4. In the chemical monitoring of the second production test, this dilution process could not be confirmed because 14m3 of freshwater had been injected at the end of the stimulation operation. In order to confinn dilution by freshwater and then to identify the mechanism responsible, it is recommended to perform a down hole sampling before the net production test. During this test, especiaily at the beginning, it will be essentiai to carry out a chemical monitoring of the produced fluid. Ecept for the down hole sample collected from BO-4 in 1996, al1 the isotopic and chemical results suggest a probable common origin of the deep fluid in wells BO-2 and BO-4. The geochemical monitoring performed in well BO-4 dows to notice that the influence of sea water injected during the tracer tests and the stimulation operation always has been practically negligible on the isotopic and chemicai composition of the deep fluid. This observation is in good agreement with sulphur and oygen-18 isotopic values which suggest a fast circulation rate of the deep fluid or large water-rock ratios in reservoir Use of chemical geothermometers Chemical geothermometers enable the temperature of the reservoir fluid to be estimated. They are therefore valuable tools in the evaluation of new fields and in monitoring the hydrology of systems on production. Solute geothemometers are based on temperaturedependent mineral-fluid equilibria. Reactions must be fast enough to reach equilibrium in the reservoir but, to ensure the reservoir composition is retained in the discharge water, no re-equilibration must occur as the fluid migrates to the surface. The main solute chemical geothermometers (Fournier and Rowe, 1966; Foumier and Tmesdell, 1973; Foumier and Potter, 1979; Michard, 1979, 1989; Giggenbach, 1988; Nicholson, 1993) applied on the reconstructed chemical compositions of deep bnne give temperature values vqing from C (Quartz, Na-K-Ca) to C (Na-K, K-Mg). The Na-Li geothermometer yields a lower value around 200 C. For the down hole sample collected from BO-2 in 1993 (Table 7), a temperature close to 260 C is found with Quartz and Na-Li geothermometers. With Na-K and Na-K-Ca, temperatures are estimated to 250 and 235'C, respectively. The value calculated with K- Mg is 208'C. The lower estimations of temperature given by the Quartz and Na-Li 64 BRGM Report R 40646

65 geothennometers for the chemical compositions of deep brine reconstructed from samples collected at surface are due to a decrease in dissolved silica and lithium probably related to the precipitation of small amounts of silica and clay phases during the rise of the deep geothermal fluid. Given the accuracy of these estimations (2 20 C), it can be concluded that they are very close to measured values and that deep brine is close to equilibrium with a specific mineralogical assemblage at a temperature ranging from 240 to 260 C Geochemical modeliing and scale deposit risks a) Preliminary considerations Starting from sampled waters and gas composition of the fluid at a fied reservoir temperature (242 C) and some other constraints, it is possible to calculate the deep water speciation and the Saturation Inde for a large number of rninerals, using the geochemical EQ3NR code and the DATAO.COM.R2 data base (Wolery, 1995a). The B- dot equation (etended Debye-Hückel model) was applied to calculate the activity coefficient of the ionic species. The Saturation Inde (SI) of any water with respect to a given mineral is defined as the logarithm between the ionic activity product IAP of the aqueous species and the equilibrium constant K relative to the considered reaction at a given temperature: SI = log (IAPJK) A value of SI = O is estimated as representative of an equilibrium state between water and a given mineral. Given the likeness of the chemical compositions obtained for reconstmcted deep îluid from fluid samples collected at surface in wells BO-2 and BO-4, an only chemical composition named BOS was selected (that estimated by Sanjuan and Brach (1997) with some additional and modified data, see Table 7). In order to consider the discrepancies observed between fluids collected at surface and directly in reservoir, the chemical composition of the down hole brine sampled in 1993 in well BO-2 was also chosen and named BODP (Table 7). This composition is probably the most representative of the deep geothermal fluid. For aluminium determination in both cases, it was assumed that this species was controlled by K-feldspar, a mineral probably present in the mineralogical assemblage in equilibnum with deep brine (Na-K geothermometer is based on equilibrium between water, albite and this mineral). ph was set using alkalinity (essentially constituted of dissolved bicarbonate) and the CO2 partial pressure estimated from gas analyses and Gas-Steam-Liquid ratios (P(CO2) = 1.8 bar). Sulphide concentration was caiculated by setting a P(H2S) value equal to 10-"~~ bar. BRGM Report R

66 b) ph reconstitution Applying these assumptions, geochemical simulations performed at 242 C have allowed to calculate a ph of 5.05 and 5.74 using the chemical BOS and BODP compositions, respectively. This discrepancy is caused by the significant difference between alkalinity values. The first ph value (5.05) is lower than that estimated to by Abou Akar et al. (1992) or to by Sanjuan and Brach (1997), who found corresponding P(C02) values around bar. However, Sanjuan and Brach (1997) estimated these values assuming the deep fluid in equilibnum with respect to calcite. For a ph of 5.05 and a P(C02) of 1.8 bar, we shall see that the deep fluid is under-saturated with respect to this mineral. The second ph value (5.74) is close to the previous ones but alkalinity used in calculation is higher. This akalinity measured on a down hole sample is probably more representative of the deep geothermal fluid. Taking into account al1 these results, the ph value for the deep geothermal fluid can be estimated to be close to 5.3 I 0.3. c) Redo potential (Eh) reconstitution Measurements of Eh in water are not representative of a total Redo equilibrium (Michard, 1989). Measurements commonly indicate mied Redo potentials that result from the presence of several Redo couples. For some fast couples such as Fe(II)/Fe(ilI), when the concentrations of the two species are relatively high, the measured Eh can be close to that calculated from aqueous Fe(II) and Fe(m) activities. However, this result does not involve that al1 the other Redo couples are in equilibrium with water. From measurements performed on site (Table 2a and App. 2), the determination of a representative value of Eh for deep fluid is not possible. Although the values obtained in stem condensate from BO-4 seem to be relatively stable (around -200 mv), changes in values in bines are too many frequent (from -330 to -220 mv for BO-2 and from -297 to -245 mv for BO-4). Moreover, the effect of steam separation on Eh is difficult to estimate. It can be only considered that the most negative measured values (less oygen) must be the closest to that of deep fluid. These values in brines are around -330 mv. After correction by subtracting the reference electrode potential at the measurement temperature (- 25"C), an approimate value of -100 mv is found. This value is given at 25 C and not at reservoir conditions. When this value was used in geochemical simulations at 242 C for both chemical compositions (BOS and BODP), oygen partial pressure values were much higher than that obtained using the empincal equation computed by D'Amore and Gianelli (1984), BRGM Report R 40646

67 from field data on several geothermal fields considering suitable alteration mineral equilibria with hydrothermal solutions. This equation is: log P(02) = / T("K) T'PK) At T = 242 OC, log P(02) = bar, At T = 250 OC, log P(O2) = * 0.5 bar, At T = 260 OC, log P(O2) = I 0.5 bar. Computed Hz partial pressure values were also very far from that estimated using chemical gas composition and gas-stem-liquid ratios in the D'Amore and Tmesdell (1985) equation, which is around 10-l.~ bar at 242 C. The value of -100 mv for Eh is probably too much higher. For the BOS composition, P(02) and P(H2) values computed from D'Arnore and Gianelli (1984) and D'Amore and Tmesdell (1985) equations are in agreement with an Eh of -280 mv (Table 8). P(CH4) value (10-l.~~ bar) is also similar to that detennined using D'Amore and Tmesdell (1985) equation at 242 C (10-l.~- 10-'.' bar). For this Eh value, water is under-saturated with respect to pyrite (SI = -0.73). However, at this same Eh or at close values, deep fluid can be saturated with pyrite by modifymg ph or akalinity values slightly. Saturation Inde (SI) for this mineral is very sensitive to variations of Eh and ph whose values are relatively inaccurate. Dissolved sulphide concentration (about 15 mgil) computed in the geochemical simulation is higher than that estimated from direct measurements on brine and stem samples (- 7 ma). A P(H2S) value of bar would lead to a value similar to that estimated by direct measurements (Table 8). In this case, water would be still more under-saturated with respect to pyrite (SI = -1.53). Another eplanation of the difference found between computed and measured values of sulphide can be a loss of this species before sampling and analysis: precipitation?, degassing? (sulphide is essentially in form of dissolved H2S and the main part of this species has been detected in stem phase). Using the chemical composition BOS, water is saturated at 242OC with pyrite (SI= -0.14) for a value of Eh around -250 mv (Table 8). Computed P(02) and P(H2) values ( and l ~ - bar ~ respectively). ~ ~ are then far enough from values estimated by the other methods. Dissolved sulphide concentration computed in simulation is always close to 15 ma. In this case, Eh would be essentially set by water equilibrium with pyrite and not by dissolved 02. For the BODP composition, geochemical modelling shows that P(O2) and P(H2) values computed from D'Amore and Gianelli (1984) and D'bore and Tmesdeii (1985) equations are in agreement with an Eh value of -350 mv. In this case and for Eh values > -350 mv, water is over-saturated with respect to pyrite (SI > 1.62). However, if ph2s is set to bar instead of 10-l.~~ bar, water is saturated with respect to this mineral at an Eh value of -350 m and dissolved sulphide concentration is closer to analytical data (Table 8). BRGM Report R

68 Bouillante geotherrnal field (Guadeloupe) Table 8 -ph, partial gas pressures and SI calculations pedormed from the BOS and BODP chemical compositions, using EQ3NR geochemical code. Anhydrite Barythe Boehmite Celestite Fluorite Pyrite Quark Carbonates Calcite Dolomite Dis. Dolomite Siderite Strontianite Clay minerais: Beideiiite-H Beideliite-K Illite Kaolinite K-Montmodlonite Na-Montmoriüonite Muscovite Pyrophillite Zeolite: Epidote Heulandite Laumontite Margarite Prehnite Wairakite Zoizite O. 02 O O O O II O O. 02 O O O O Il O. O O The numbers in italic type indicate that saturation state is achieved O O O O O. O O O. O O O O O. O BRGM Report R 40646

69 Given the present analytical data and the compleity of the Eh parameter, it is difficult to know if the deep water is saturated with respect to pyrite. Accurate analyses of sulphide on a bottom hole sample (brine and gas) could probably yield more information. d) Scale deposit risks Using the BOS composition, the dissolved aluminium concentration of the deep fluid was computed to 137 pg/l (Table 8). This value is approimately two or three times higher than that reconstructed from measured values on samples collected at the surface (Table 7). This suggests a probable precipitation of alumino-silicate minerals (formation of kaolinite or smectite?) or adsorption processes (on amorphous silica, for eample) during the rise of deep fluid. For the BODP composition, the calculated aluminium concentration (69 pga) is close to the analytical value (Tables 7 and 8). Table 8 shows the Saturation indices calculated for each chemical composition (BOS and BODP) using a Eh value of -280 and -350 mv respectively. Water equilibrium with similar minerals (albite, K-feldspar, quartz, smectite, heulandite, laumontite, wairakite, zoizite) was found for both chemical compositions. The deep fluid is saturated with respect to anhydrite using the BOS composition and slightly over-saturated for the BODP composition. Given the bad accuracy on the ph determination and on the thermodynamic data about muscovite (or illite), this fluid also could be saturated with this type of mineral. Most of these minerals were detected in the reservoir by mineralogical and petrological studies and are consistent with an high temperature alteration. Some of these minerals (smectite, zeolite) have also been found in hydrothermal deposits collected at the outflow of thermal spnngs for which water indicates a miing with a similar geothemal brine (Traineau et al., 1997). This is in absolute agreement with a cooling process of a fluid in equilibnum with these minerals at 242 C because alumino-silicate minera1 solubility becomes lower with decreasing temperature. For the BOS composition, simulations also were performed with a P(CO2) value of 1.4 bar. ph is slightly increased (5.15) but the other results are similar. On the contraq, as this is illustrated by the simulations performed using BOS and BODP, changes in alkalinity may lead to significant modifications, especially for ph value but also for the Saturation Indices with respect to carbonate minerals. Indeed, according to alkalinity values, water can be under-saturated, saturated or over-saturated with most of carbonate minerals. Even if calcite solubility increases at low temperatures, calcite can be precipitated during the CO2 degassing because of the increase in ph values in brine. In this case, alkalinity values decrease and measurements of this parameter at the surface are lower than in the deep fluid. As a result, it is absolutely necessary to perform direct accurate alkalinity measurements in down hole fluid samples to compare them with the present values determined at the surface and to confirm the real value of alkalinity in the deep fluid. BRGM Report R

70 One of the major risks of scale deposit is the precipitation of amorphous silica due to the fluid cooling during its rise. Indeed, this fluid in equilibrium with quartz at C has got high dissolved silica concentrations ( mgil) which can precipitate amorphous silica at lower temperatures. A small precipitation of amorphous silica during the rise of the deep fluid could eplain the slightly lower values in dissolved silica obsemed for the samples collected at the surface (Table 8). As this was eplained in chapter 3.3.2, the eceptionaily low value of dissolved silica obsewed for the down hole fluid sampled in 1998 is only due to the sampling conditions. Discrepancies are obsewed between the concentrations of dissolved sulphate, magnesium, strontium, lithium, barium, aluminium and trace metal ions such as iron, manganese, copper, zinc, nickel, zinc, cobalt, cadmium, silver, lead anaiysed on samples collected at surface and at bottom hole (Table 8). The scaie deposits collected in 1993 at a depth of around 322 m into well BO-2 and constituted of polymetallic sulphide, calcite and amorphous silica, suggest that these species are trapped in this type of minerals which would precipitate during the rise of the deep fluid. The assumption of brine contamination during the sampling by the KUSTER sampler or by dissolution of mineral deposits collected with the fluid sample seems to be less relevant. Using the BODP composition at 242 C and for both phzs values, water is largely oversaturated with respect to minerais such as chalcopyrite (CuFeSî), chalcocite (Cu&), covellite (CuS), cadmium sulphide (CdS), sphaierite (ZnS), millerite (NiS) and vaesite (NiS?). It is under-saturated with respect to aiabandite (MnS). For a ph2s value of 10-'.'~ bar, water is saturated with respect to galena (PbS). For a ph2s value of 10-'.~ bar, water is under-saturated with respect to this mineral. Even if thermodynamic data for some of these minerais are inaccurate or questionable, it can be noticed that for most of them, an increase in ph and a decrease in temperature favour their precipitation. Consequently, their formation with carbonate minerals after CO? degassing (increase in ph) is not surprising. in conclusion, a careful down hole fluid sampling in both wells is absolutely necessafy to accurately determine alkaiinity and the dissolved sulphide concentration. As the COz partial pressure is now relatively well known, a better knowledge of these parameters will allow to obtain a more precise determination of ph in the deep geothermai brine than presently (ph between 5,O and 5.8) and an higher confidence in the Saturation inde caiculations between the geothermal fluid and many minerals (especiaily sulphide and carbonate minerals). BRGM Report R 40646

71 Conclusions In conclusion, the multiple tracer test performed during the stimulation operation has shown and confirmed that no evident hydraulic connection eists between wells BO-4 and BO-2 even when a large volume of sea water is injected from BO-4 (about m3). The neighbouring thermal springs concerned by a chemical monitoring indicate no direct hydraulic connection with BO-4 fissures network. Indeed, no tracer restitution can clearly be related to the injection of the organic compounds (sodium naphtionate, sodium benzoate and isophtalic acid) in well BO-4. Both production tests (RPT and ER) designed to evaluate the modifications in BO-4 discharge (percentage of separated phase and chemical characteristics) due to the stimulation operation were limited to single flow rate tests. The well was not completely open neither during RPT or EPT. For these reasons, the <<dry >> and << wet >> behaviour of the well noticed during the RPT cannot be argued by geochemical monitoring. The results obtained on brine, steam condensate.and non condensable gases are similar in terms of phase proportions and chemical, isotopic composition in both RPT and EPT discharged fluid. The improvement of production by thermal stimulation had no significant consequences on chemical parameters of the total fluid discharged. It can be deduced that either the improvement of production was homogeneous dong the production zones or the stimulated production level or fracture delivered a fluid whose characteristics are close to those of the total discharge of the well (miing of the different feeding zones). According to the results obtained from natural tracers (chloride, magnesium, sulphate, deuterium and oygen-18 particularly) and organic compounds used during the stimulation operation, it can be obsewed that injected sea water is never found at high proportions in the discharged fluid. Chemical composition of both RPT and EPT samples is always little influenced by sea water. The percentage of remaining sea water can be estimated to 0.2% in discharged fluids at the end of RFT and EFT. As a result, it can be concluded that sea water is rapidly and completely mied with reservoir geothennal fluid. This can be eplained by either a relatively significant circulation rate or a high water-rock ratio, which is in agreement with the isotopic oygen-18 signature of deep geothennal fluid. The monitoring has allowed to better characterise the chemical and isotopic compositions of the deep geothermal fluid from well BO-4 which were badly known before the two production tests. It has contributed to check and compare the previous results obtained for well BO-2. It has shown that the deep BO-4 fluid is very close to BO-2 one on a chemical, isotopic and physical point of view. The total proportions in weight of each phase of the fluid are similar for both wells. Some inaccuracies subsist on the values of key parameters such as alkalinity, ph, dissolved aluminium, iron, BRGM Report R 40646

72 sulphide and some trace metal concentrations which will have to be suppressed by additional works (careful down hole samplings). The use of a scale inhibitor (IDOS 130) during al1 the stimulation operation has allowed to prevent anhydrite scale deposits predicted by modelling and observed during the RPT. These deposits probably occurred during the first interference test in March 1998 whereas only 100 m3 of sea water had been injected. Apart from the objective of the stimulation operation, the geochemical monitoring has shown that direct escapes of deep geothermal brine, mied with very small variable proportions of sea water after vaporisation and cooling, outcome from the littoral spring BO-3 Beach. The fumaroles situated near this thermal spring are probably the outcome of the steam resulting from the vaporisation of the deep geothermal fluid. During the humid period (August - October), an additional contribution of superficial fresh water was observed in this spring. The geochemical monitoring of the thermal spring Tuyau, situated near the geothermal power plant, essentially constituted of fresh water (> 90%) with low amounts of deep geothermal brine, suggests an 'additional and increasing contribution of stem condensate. Presently, this contribution could have reached a stationary state. According to Cl concentration and SD and S180 values, it would represent about 45% of steam condensate. The evolution of such a contribution would have to be rapidly monitored to optimise the eploitation conditions of the geothermal fluid. Acknowledgements: the authors are grateful to the staff of the Geothermal Power Plant for site facilities. BRGM Report R 40646

73 References Abou Akar A., Matray J.M. et Brach M. (1992) - Etude géochimique du fluide géothermal du puits BO-2 (centrale EDF) et des sources thermales de la région de Bouillante (Guadeloupe). Rap. BRGM R IRG SGN 92,41 p. Adams M.C., Benoit W.R., Bodvarsson G.S. and Moore J.N. (1989) - The Diie Valley, Nevada tracer test. Trans. Geothem. Resour. Coun., 13, pp Adams M.C. and Davis J. (1991) - Kinetics of fluorescein decay and its application as a geothermal tracer. Geothemics, 20, no 112, pp Adams M.C., Moore J.N., Fabry L.G. and Ahn J.H. (1992) - Thermal stabilities of aromatic acids as geothermal tracers. Geothemics, 21, no 3, pp Adams M.C., Moore J.N., Benoit W.R., Doughty Ch. and Bodvarsson G.S. (1993) - Chemical tracer test at the Diie Valley geothermal field, Nevada. Geothermal reservoir technology, research program, report prepared for U.S. Department of Energy by University of Utah Research Institute, Salt Lake City. Arnason B. (1977) - The hydrogen-water isotope thermometer applied to geothermal areas in Iceland. Geothemics, 5, pp D'Arnore F. and Gianelli G. (1984) - Mineral assemblages and oygen and sulphur fugacities in natural water-rock interaction processes. Geochim. Cosmochim. Acta, 49, pp D'Amore F. and Panichi C. (1980) - Evaiuation of deep temperatures of hydrothermal systems by a new gas-geothermometer. Geochim. Cosmochim. Acta, 44, pp D'Arnore F. and Truesdell A.H. (1985) - Calculation of geothermal reservoir temperatures and steam fraction from gas composition. Geothermal Resources Transactions, vol. 9, Part 1, pp D'Amore F. and Truesdell A.H. (1988) - A review of solubilities and equilibrium constants of gaseous species of geothermal interest. Sciences Géologiques, Strasbourg, 40 p. Demians d'archimbaud J. et Munier-Jolain J.P. (1975) - Les progrès de l'eploration géothermique à Bouillante en Guadeloupe. «Tiré à part», pp Ferronsky V. and Polyakov V.A. (1982) - Environmental isotopes in the hydrosphere. A Wiley-Interscience Publication, 466 p. BRGM Report R

74 Foumier R.O. (1979) - A revised equation for the NaIK geothermometer. Geoth. Res. Council Trans., 3, pp Fournier R.O. and Potter R.W. (1979) - Magnesium correction to the Na-K-Ca chemical geothermometer. Geochim. Cosmochim. Acta, 43, pp Foumier R.O. and Rowe J.J. (1966) - Estimation of underground temperatures from the silica content of water from hot Springs and wet-stem wells. Amer. J. Sci., 264, pp Foumier R.O. and Tmesdell A.H. (1973) - An empirical Na-K-Ca geothermometer for natural waters. Geochim. Cosmochim. Acta, 37, pp Giggenbach W.F. (1980) - Geothermal gas equilibria. Geochim. Cosmochim. Acta, 44, pp Giggenbach W.F. (1988) - Geothermal solute equilibria. Denvation of Na-K-Mg-Ca geoindicators. Geochim. Cosmochim. Acta, 52, pp Giggenbach W.F. and Goguel R.L. (1989) - Collection and analysis of geothermal and volcanic water and gas discharges. Fourth Edition. Report CD2401, Chemistry Division, DSIR, New Zealand. Herbrich B. (1996) - Mesures dans le puits de Bouillante BO-4, Injections de traceurs géochimiques. Rap. 96 CFG 57, 11 p. Marini L. (1987) - Geochemistry of North Ghoubbat-Asal Region. Doc. Geothermica Italiana, 28 p. Michard G. (1979) - Géothemomètres chimiques. Bull. BRGM (2) ili, n02, pp Michard G. (1989) - Equilibres chimiques dans les eau naturelles. Publisud, 357 p. Nicholson K. (1993) - Geothermal fluids. Chemistry and Eploration Techniques. Springer, 261 p. Sanjuan B. and Brach M. (1997) - Etude hydrogéochimique du champ géothermique de Bouillante (Guadeloupe). Rap. BRGM R39880,84 p. Sanjuan B. (1998) - Champ géothermique de Bouillante. Rapport d'état d'avancement des travau en géochimie. Rap. BRGM N2592,29 p. Sanjuan B., Lasne E., Brach M., Vaute L. and Bellon J.F. (1999) - Champ géothermique de Bouillante (Guadeloupe): 1. Test de traçage chimique entre les forages BO-4 et BO-2 (mars-juin 1998). Rap. BRGM à paraître. BRGM Report R 40646

75 Traineau H., Sanjuan B., Beaufort D., Brach M., Castaing C., Correia H., Genter A. and Herbrich B. (1997) - The Bouillante geothermal field (F.W.I.) revisited: New data on the fractured geothermal reservoir in light of a future stimulation eperiment in a low productive well. Proceedings, Twenty-Second Workshop on Geothermal Reservoir Engineering, Stanford University, California, January 27-29,7 p. Wolery T.J. (1995a) - EQ3NR, A computer program for geochemical aqueous speciation solubility. Calculations : Theoretical manual, User's guide and related documentation (version 7.2b). Lawrence Livemzore National Laboratory, 246 p. Wolery T.J. (1995b) - EQ6, A computer program for reaction path modeling of aqueous geochemical systems : Theoretical, User's guide and related documentation (version 7.2b). Lawrence Livemzore National Laboratory, 337 p. BRGM Report R 40646

76 APPENDI 1 Characteristics of the scale inhibitor IDOS130 BRGM Report R 40646

77 -ph: 3,s-4,s. - Viscosté : d s à 2S0C. - Densiré : Point de gel : -lo C. -Po& d'ébullition : 100 C. - inkhnudle. - Soluble dans l'eau m toutes proportions. APPLICATIONS L'IDOS 130 est spécialement concu pour controler la formatik des dépôta de carbonate. LTDOS 130 est un copolymere qui donne d'ecellents résultats en application par squeezes. MODE D'EMPLOI LlDOS 130 a une très bonne stabiitd pour une application A haute température (>150 C). LTDOS 130 s'utilise B des concentrations qui vont de 5 à 30 ppm. LWOS 130 n'est pas un produit dangererrv mais if est tout de même préférable lors de sa manipulation de porter des gants et des Iunenes. Cer idormaions ne wmt ni r~im SPS mrutiw me gmmiie. Ella dicouirnt d'+nicco mnnvcr et sant &nu& à titre indicatifpu guider I'utiliSPtM dans JCT prcmiho applicati- I -ma ' --.. mw-. a. =venue ~ean-jouras - 2. Pelrollére ISOU GARGENVILLE -FRANCE TB,. 33 (0)) W -Fo 53 (0)) Mal. ini0arep.b. nm. //w i~.>ucn~io br31molofsre3mm7324w,.g~- rep.h BRGM Report R 40646

78 Geochemical monitoring results obtained during the multitracer test associated to the stimulation operation: - from well BO-2, - from springs SI, S2 and S3. BRGM Report R 40646

79 Well Sampling and on site rneasurements Napht. contamination 7 R = rinsing water

80 Srmpli IdinUflcaüon Number Date 8098.F F2-343 BWB.F F :W 18/08/98 22:CQ O:W :W C+ F.pH2 A- F Sampllnn Wpe ISO Res. Nepht NF. NA NF. NA NF. NA R R R R Na.Banz. NF. NA R R Napht. ~gn c 1 cl < 1 < 1 Fluor. pgn Analyses on site Ca AIL. meli mmen CI men HS mgn NH, mgli 510, mg11 ph Physlcoshemldiy Eh T Cond.(ZVC) mv "C mslcm Lab anmlyses Iso.Ac. p@ Benz. pgli Mg mell COmmEnb R = low quanlity of water sampled dunng rinsing anly

81

82 Ssrnpli ldsntlncatlon Number Date B098F F &F2-436 BO98F llW98 20:W 01M9M 02:W 01109I98 08:W 011W198 14:W CI F-pH2 A- F Sam~lln!aiype Ras. Napht, NF. NA ISO NF, NA NF. NA Na-Benz. NF. NA Mpht. ug11 C 1 < 1.; 1.; 1 Fluor. vgll Analyses on rlte Ca AIL. mgil meq.4 CI mgil HS mgil 0.52 NH, mgfl SI02 mgfl 594 ph 7.04 Physleoshemlstry Eh T Cond.(ZO'C) mv 'C mslcm Lab. ~nalvses lao. Ac. vgf1 Benz. g c 10 Ma mgfl 1.0 Commenh

83 O s m 03 a.@. O 2 O> Sampla Identlflcatlon Number Date 0098-F F F F F F F F F2-488 B098F F F :W :W :W :W :W :W :W :W 231(W198 08:W :W :W :W CI A. F-PH~ F sampllng type ISO Re*. N~pht. Na-Ben.. NF. NA NF, NA NF, NA NF. NA Y Y Y Napht. pgli c l ci C 1 < 1 Fluor. Ilan Analyses on rite Ca 1\11, mali me4.n CI mgli HS mgn 0.85 NH4 man 510~ man 642 ph 8.61 Phydcochemistry Eh T mv 'C Cond.(20mC) mslcm 34.6 Leb. analyser 1io.A~. Mg mgil Benz. vgn < 10 < 10 c 10 c 10 < 10 c 10 c Commenh Fmm thls date, geochemical monilaring of sodium benmate only

84 SI spring (BO-3 Beach) - Sampling and on site measurements Sample Number 8098-SI SI3 809SS S S S SI SI SI SI SI SI SI &S SI S S S SI ESI B SI SI SI SI SI SI SI S SI S SI SI SI-42 identiiication Date 01104(98 11:W M104/98 10:lO 03/04/98 11: :OO :30 O :W : : : :40 14/04/98 11: :10 16/04/98 11:20 17/04/98 11:lO 18/04/98 11: :lO :45 211C419811:OO 22/04/98 11:OO 23/04/98 11: :20 27/04/98 1l:W :30 04\05/98 13:W 07/05/98 12: : : :W 27/08/ : :W 02/09/98 08: : :30 07/09/98 09:lO 09/09/98 09: :W 43/09/ S:W O O:W O :05 C+ F-pH2 Sampling type A- F ISO NF, NA Tracera NF, NA Fluo. Napht. pgll pgll cl <1 <1 cl SI 4.9 <1 <1 <1 <1 5.3 cl cl cl el 5.5 C l c 1 el < 1 cl c 1 < 1 cl C l c 1 c 1 c 1 cl < Analyses CI mgfi KM IlW on site Ca tngli Alk. m8q.n S102 mdi % ph Physicochemistry Eh,,. T T, mv 'C 96.0 'C O O O O Cond. mslcm Lab. analyses Benz. lso. Ac. pgfi pd1 C 10 < 10 C 10 < 10 < 10 cl0 c 10 CIO c cl0 CIO c 10 CIO c 10 C 10 c 10 < 10 < 10 <IO CIO Cl0 < 10 <IO C 10 CIO <IO <50 <50 <50 ~ 5 0 c50 c50 Commente rough sea 1 possible miing with seawater rough sea 1 possible miing with seawater rough sea + rainwaterl possible miing Twat. = 76 C Twat. = 82'C ralnwater 1 possible miing (Tviat. = 72%) Twat. = 79% Twat. = 82% Twat. = 84'C Twat. = 83'C emergence obsetved at 101'C (Twat. = 81'C) Twat. = 76'C low tide, observed emergence (Twat. = 84%) low Ude, obsetved emergence M t. = 89.C) Twat. = 81 C Twat. = 83% Twat. = 83'C veiy low tide. new obsetved emergenws veiy lowtide (Twat. = 77%) Twat. = 79% Twat. = 71% Twat. = 80 C Twat. = 81'C Twat. 82% (low Ede) Twat. = 81'C (low tlde) Twat. = 78.C (low tide) Twat. = 81'C (low tide) Twat. = 73% (sea at 50 cm from spring) Twat. = 86% (sea at 50 cm from spring) Twat = 86% (sea at 1 m from sprina) Twat. = 91% (sea at 1 m from spring) high tide (sampling performed in an upper side) high tlde (sampling perlonned in an upperslde)

85 S2 spring (Ravine Blanche) - Sampling and on site measurements

86 S3 spring (Tuyau) - Sampling and on site measurements simple Numbsr û3.4 B098S S û3-10 BM BOS8û Bo9BS3-21 B098S S S M S3.33 Identlnciüon Dib B098S :00~ :50 07R : :48 MlC8/ : /199808:35 17/06/1998 W: H998 10:20 24m998 14:OO : /199808: : : /199809:30 24/07/ : :M 31107/199808:55 05/08/ :55 12/08/ :W 25108/ :m : :M 31108/ :35 02/09/ :20 04/09/1998 W:25 07/09/ :W 091Wll998 10:15 11/09/ : :W MH :W 07HOl :00 28HO :20 Snmpllng type CI A- ISO F-pH2 F NF. NA Y Tracers NF. NA ' FIuo <1 < 1 c i < 1 c 1 c 1 < 1 c i Anahes on site W M W HO M) W ph Phyrhochernlstry Lab. analyses Eh,. T Ti*, Cond. Bena lso. Ac. mv 'C 'C mslcm pgll I@ O TT '10 < < 10 c <IO C Commene flow rate = 2.5 llmn ph-meter breakdown approimative flow rate = 2.4 llmn (flow rate = 2.7 Vmn) 1 strong raln approimative Oow rate = 2.6 Vmn approimative Oow rate = 3.75 Umn / Flow rate and temperature increase for water approimative Rwrate = 3.75 llmn approimative Ow rate = 3.75 llmn min during the nlghtl approimalii Am rate ;. 2.8 llmln appro*matie Ow rate = 3.75 llmn approimative lïw rate = 3.75 llmn approimaliielïw rate = 3.75 llmn apprcmmaii Ow rate = 3.75 llmn approimalive Oow rate = 3.6 llmn (strong min the day before) approimative Oow nite = 3.3 Vmn (Row rate decrease in spite olstrono rains) approimative Ow rate = 3.75 llmn approimative flw rate = 3.5 llmn approfimativa flw rate = 3.75 Vmn