- Email: [email protected]
Geothermal resources in the Asal Region, Republic of Djibouti: An update with emphasis on reservoir engineering studies
Geothermics 39 (2010) 220–227
Contents lists available at ScienceDirect
Geothermics journal homepage: www.elsevier.com/locate/geothermics
Geothermal resources in the Asal Region, Republic of Djibouti: An update with emphasis on reservoir engineering studies Daher E. Houssein a,∗ , Gudni Axelsson b a b
Centre d’Etudes et de Recherche de Djibouti, Route de l’aéroport, CERD, Earth Science, B.P. 486 Djibouti, Djibouti ÍSlenskar Orkurannsóknir (ISOR), 9 Grensasvegi, Reykjavik, 108, Iceland
a r t i c l e
i n f o
Article history: Received 23 July 2009 Accepted 16 June 2010
Keywords: Geothermal Well testing Wellbore simulation Lumped-parameter model Resource assessment Asal Djibouti
a b s t r a c t Three independent geothermal systems have been identified, so far, in the Asal region of the Republic of Djibouti (i.e. Gale le Goma, Fiale and South of Lake). Six deep wells have been drilled in the region, the first two in 1975 and the others in 1987–88. Well A2 was damaged and wells A4 and A5 encountered impermeable yet very hot (340–365 ◦ C) rocks. Wells A1, A2, A3 and A6 produce highly saline (120 g/L TDS) fluids leading to mineral scaling. Well test data indicate that the reservoir might be producing from fractured and porous zones. The estimated permeability-thickness of the deep Gale le Goma reservoir is in the 3–9 darcy-meter range. Lumped-parameter modeling results indicate that well A3 should be operated at about 20 kg/s total flow rate and that injection should be considered to reduce pressure drawdown. The estimated power generation potential of well A3 is 2.5 MWe, and that of all Asal high-temperature hydrothermal systems is between 115 and 329 MWe for a 25-year exploitation period. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Neither fossil fuel resources nor hydroelectric potential exist in the Republic of Djibouti. The lack of these energy resources makes the country heavily dependent on imported oil products, mainly from Saudi Arabia or Dubai. According to data from the stateowned power utility, Electricite de Djibouti (EDD), an estimated 49.7% of the population has access to electricity, 99.5% of which is urban. The average electricity cost was equivalent to US$ 0.3/kWh in 2009, with commercial establishments paying about 40% more than domestic consumers (EDD, 2009). This high cost affects all sectors of society and is considered to be the main barrier to Djibouti’s industrial development. The Republic of Djibouti (surface area 23,000 km2 ) is located in the Horn of Africa (Fig. 1), at the Afar Triple Junction where the Red Sea, Gulf of Aden and East African Rift systems meet (Barberi et al., 1975). Given its geodynamic location, commercial-scale geothermal resources can likely be found in the country. The first geothermal investigations of the Asal area, located in the Afar depression, were undertaken in 1970 and included geological and geochemical studies, as well as geophysical exploration (BRGM, 1973). These studies identified the high-temperature Gale le Goma system, where four deep wells have been drilled (Fig. 2),
∗ Corresponding author. Tel.: +253 35 27 95; fax: +253 35 48 12. E-mail address: [email protected] (D.E. Houssein). 0375-6505/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.geothermics.2010.06.006
but also other promising geothermal areas. This particular geological setting is now under exploration by Reykjavik Energy Invest (REI) with the intention to build a 50-MWe power plant expected to be in operation in 2013. If the area can support and sustain a greater generation capacity, the aim is to expand the plant or build additional ones when there is sufficient demand. The goal is to produce electricity at a lower cost than that generated by burning imported fossil fuels. The main objective of this paper is to add to the work done previously by Aquater (1989) and Daher (2005) for the Asal region, by the following reservoir engineering techniques: • Interpretation (i.e. re-interpretation) of Gale le Goma well A3 well test data using decline curve derivative analysis (Horne, 1995) and numerical modeling (O’Sullivan et al., 2001). The results are then compared to those of conventional log–log, semilog and match point methods used by Aquater (1989) and Daher (2005). Well test modeling software developed by Iceland GeoSurvey (ISOR) (Juliusson et al., 2007) is used to compare estimated hydrological parameters. • The wellbore simulator HOLA (Bjornsson et al., 1993) is used to simulate wellbore conditions (temperature, pressure, etc.) that influence the transport of geothermal fluid from the reservoir to the surface when well A3 is produced. Then the computed productivity index (PI) is compared against the measured injectivity indices (II), defined as the change in injection flow rate divided by the change in stabilized reservoir pressure, derived from injection tests.
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
221
Fig. 1. Map of Djibouti showing the location of the study area. The Afar depression (or triangle) includes parts of Djibouti, Eritrea and Ethiopia.
• Estimation of the size, permeability-thickness product and long-term geothermal production potential of the Asal area using a lumped-parameter model and data from a November 1987 production test performed in the Gale le Goma well A3. The computed reservoir parameters are compared with those obtained from geophysical surveys and conventional well tests.
• Assessment of the Asal geothermal resource using a volumetric method (Muffler and Cataldi, 1978) and parameters derived from resistivity survey data (Arnason et al., 2008), geologic-structural information, results from well A3 production tests and temperature profiles of the six wells drilled in the area. • Estimation of the total accessible energy in the Asal geothermal region based on a Monte Carlo simulation that incorporates
Fig. 2. Map showing the location of the three known geothermal systems (Gale le Goma, Fiale and South of Lake) in the Asal region. The figure also shows the resistivity at 3000 m b.s.l., inferred lineaments in low-resistivity (red lines), seismicity (dark green dots) and geothermal surface manifestations (light green). The two super saline and sealed-off Gale Le Goma and South of Asal Lake systems are indicated, as is the more open, lower salinity Fiale system under Lava Lake (modified from Arnason et al., 2008). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
222
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
uncertainty in the parameters used in the volumetric resource assessment method. The immediate objective of the study presented here is to evaluate the geothermal potential of the Asal region. All the parameters used come from the Gale le Goma system since it is the one that has been investigated till the present time. 2. Geothermal exploration in the Asal Rift 2.1. Geography and geology The Asal Rift is a northwestward extension of the Gulf of AdenTadjura. It extends from the Gulf of Goubet in the SE to Lake Asal in the NW. The lake, which is 155 m below sea level (m b.s.l.; Fig. 2), is 9–10 km wide and is bounded by major normal faults to the SW and NE. The rift due to its unique nature as a landward extension of an oceanic ridge (the Tadjura Ridge) has been studied quite extensively (Barberi et al., 1975; Beyene and Abdelsalam, 2005). The most active part of the Afar depression is the so-called Inner Rift. It is about 3 km wide and is located in the SE part of the main rift (Correia et al., 1985). It hosts most of the recent volcanism, with the last eruption occurring in 1978. The Inner Rift is characterized by intense fracturing and recent lava flows. Its most prominent feature is Lava Lake (sometimes referred to as Fiale), a phreatic crater about 1.3 km wide whose floor is covered by recent basaltic lavas. Based on analysis of drill cuttings and cores from wells drilled in the rift, the stratigraphic column in the area is composed by (from top to bottom) of the Asal series (basalts and hyaloclastites less than 1 My old), the Afar Stratoid series (basalts with some rhyolites 1–4 My old), and the Dala Basalt (basaltic lava flows with some intercalations of rhyolites and trachytes 4–9 My old). 2.2. Wells characteristics In 1975, two deep wells, A1 (1554 m) and A2 (1147 m), were drilled in the SW part of the Asal Rift; i.e. in the “old wellfield” or Gale le Goma area (see Fig. 2). A1 encountered a feed zone at 1137 m depth, while A2 showed no permeability but temperatures above 260 ◦ C (Aquater, 1989). A1 produced about 38 kg/s of a two-phase fluid (20% steam at 6 bar wellhead pressure). Four deep wells were drilled in 1987–88. Wells A3 and A6, located in the same area as A1 and A2, close to a phreatic crater (Fig. 2), encountered the same reservoir as A1 with temperatures of about 280 ◦ C (Daher, 2005). Well A4, drilled about 1.5 km NE of the phreatic crater, showed temperatures close to the boiling curve at elevations lower than 200 m b.s.l. and about 350 ◦ C at bottomhole. However, only low permeabilities were encountered in A4. Downhole temperatures profiles measured in wells A3-A5 are shown in Fig. 3. No wells have been drilled in South of Lake Asal. Based on the resistivity map (Fig. 2), we assume that in this area the same maximum reservoir temperature (∼280 ◦ C) as in Gale le Goma will be encountered. Well A5, drilled in the Inner Rift (or Fiale area) about 700 m west of Lava Lake (Fig. 2), presented sharply increasing temperature below 200 m b.s.l. with a local maximum of about 180 ◦ C at 400 m b.s.l. Below that elevation the rocks have cooled dramatically, down to 60–70 ◦ C around 800 m b.s.l., as compared to temperatures indicated by alteration mineralogy. Below 800 m b.s.l. the temperature rises steeply with depth and reaches about 350 ◦ C at bottomhole. Very low permeabilities were found in A5; higher values are expected in other parts of this area. Even if the permeability is very low in this well (around 0.4 darcy-meter; Daher, 2005), it may be increased by hydraulic stimulation. Since this operation involves
Fig. 3. Temperature profiles of Asal wells A2–A5.
injecting high-pressure water into the well, and as observed in some enhanced geothermal systems (EGS) projects, it presents a higher likelihood of producing micro-earthquakes that could be felt by the local population and should be carefully monitored (Majer et al., 2007). 2.3. Geophysics In 1988 a group from the National Energy Authority of Iceland (Orkustofnun) performed a central-loop transient electromagnetic (TEM) resistivity survey in the Asal Rift. In November–December 2007, the Iceland Geosurvey (ISOR) conducted TEM and MT (magnetotelluric) resistivity surveys to gather additional data needed to develop a conceptual model of the area that could be used to site future exploration wells. These studies showed that the saline groundwater table drops sharply from about sea level NE of the Gale le Goma area to about the level of Lake Asal in a narrow zone perpendicular to the Rift, just west of Lava Lake. Furthermore, the surveys identified a lowresistivity anomaly under Lava Lake and a local rise in groundwater to above sea level. This was interpreted as indicating a geothermal system under Lava Lake (Fiale area), not hydraulically connected to the Gale le Goma area (Arnason et al., 2008). Considerable microseismic activity was recorded in the Asal Rift in late 2000 and early 2001. A comprehensive study of these events and a review of earlier data showed that seismicity is mainly concentrated in the uppermost 3 km under Lava Lake (Doubre et al., 2007). Also in this area, very hot and ductile rocks appear to be present below a depth of 3 km (Arnason et al., 2008; Doubre et al., 2007). Based on these exploration studies, three separate geothermal fields were identified in the Asal area: Gale le Goma, Fiale (under Lava Lake) and South of Lake (Fig. 2). 2.4. Geochemistry The main results of geochemical studies of water from the Asal area are (Correia et al., 1985; Aquater, 1989; D’Amore et al., 1998): • According to a study performed on samples collected from shallow aquifers in wells Asal 1, 2, 3, 4 and 6, and deep water samples from Asal 3, as well as from samples of Lake Asal, it appears that all aquifers encountered are mainly recharged by sea water (Aquater, 1989).
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
223
Table 1 Results from analysis with different methods of drawdown tests data for A3 (Aquater, 1989; Daher, 2005). Date (1987)
q (tons/h)
q (tons/h)
8 August 24 August 7 September 9 September
225–300 79–130 155–225 300–357
75 51 70 57
Time (h)
11 5 20 2
Semilog
Type curve
kh (Dm)
Skin
kh (Dm)
4.2 14 5.75 14
−5 −5 −5 −5
4.3 13.6 7 15.9
q: flow rate; q: flowrate difference between two successive steps in the production test; kh: permeability-thickness product (transmissivity).
• The Asal geothermal fluids are mixtures of sea water and continental water of meteoric origin with high total dissolved solids (TDS). • The geothermal fluid is not produced by the evaporation of sea water as in the case of the Asal Lake NaCl brine. • The equilibrium temperature calculated from all reactive gaseous species (H2 O, CO2 , H2 , CH4 , CO, N2 , NH3 ), apart from H2 S, is about 260 ◦ C, which is compatible with measured reservoir temperature. • Lake Asal waters are composed of very concentrated sea water due to evaporation, and its CaSO4 content is modified mainly by precipitation. Deep geothermal waters seem to have no contact with Lake Asal waters and their Ca/Mg ratio is extremely different. • The fluids collected in well A5 showed that the water at the center of the rift has a much higher salinity than water near its edges. Due to the adverse brine chemistry — the high TDS (about 116 g/L) that would cause scaling problems in the wells and surface equipment (Virkir-Orkint, 1990) — Reykjavik Energy has diverted its focus from the Gale le Goma system to the Fiale area (Fig. 2). Results of the TEM-MT resistivity surveys carried out in 1988 and 2007, along with earthquake monitoring data, indicate that a separate geothermal system may reside under Lava Lake. Due to the intensive rifting activity in that area and its proximity to the sea, the geothermal fluids at Fiale are expected to be less saline than at Gale le Goma (Doubre et al., 2007). The Gale le Goma site, located at about the same distance from the sea as Fiale, is less active (Doubre et al., 2007) and seems to be a closed reservoir based on the TEM-MT study (Anasson et al., 2008) and well and production test data. 3. Well test analysis Well test data collected during injection and production tests conducted in well A3 by Aquater in August–September 1987 were analyzed using: • Well testing techniques (e.g. log–log, semilog, match point, multirate step) described by Earlougher (1977) and Horne (1995). • Derivative plots to examine the effects of wellbore storage, recharge and barrier boundaries, leakage, delayed gravity response and fracture flow (Bourdet et al., 1983, 1989; Horne, 1995). • Computer code WellTester developed by ISOR (Juliusson et al., 2007) to manipulate and analyze well test (mainly multistep injection or production tests) data and perform regression analysis. 3.1. Previous analyses of A3 well test data Well A3 was drilled in 1987 to a total depth of 1316 m. The well has a 24.4 cm (9–5/8 in.) casing down to 1016 m depth and is completed as an open hole between 1016 and 1316 m, corresponding to the main permeable zones. An upper reservoir is located between 300 and 600 m. Several transient well production tests were con-
Fig. 4. Semilog graph corresponding to the four drawdown tests performed in well A3 in 1987 (Daher, 2005).
ducted in 1987, with pressure being measured at 1075 m depth, near the upper permeable zones of the deep reservoir. Data from four drawdown tests have been used to estimate reservoir parameters (Aquater, 1989). Table 1 summarizes all results of previous analyses of A3 well test data by the conventional type-curve and semilog methods (Aquater, 1989; Daher, 2005). The pressure transient curves corresponding to these four tests are presented on a semilogarithmic graph in Fig. 4. 3.2. Revised analysis of A3 well test data The results from nonlinear regression and derivative plot analysis using the WellTester code are summarized in Table 2; the computed values are best estimates based on nonlinear regression analysis. Only during the 7 September 1987 drawdown test did the flowing pressure stabilize, indicating that a semisteady state had been reached. Results based on that test are, therefore, considered as the most reliable and will be taken as the reference. Some conclusions can be drawn from the results of this analysis: • A dual-porosity reservoir model with infinite radial flow boundary, assuming constant skin and wellbore storage values, best fits the measured data; see Fig. 5. This result is consistent with the fact that the Gale le Goma reservoir is a heterogeneous, naturally fractured reservoir. Note that in the earlier analyses it was assumed Table 2 Results of linear regression and derivative plot methods on 1987 well A3 drawdown test data. Date (1987)
WellTester results Skin
kh (Dm)
PI (L/(s bar))
8 August 24 August 7 September 9 September
−0.5 −3.4 −3.2 −1.8
1.3 9.3 3.2 9.4
23 20 7.2 17
PI: productivity index.
224
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
Fig. 5. Log–log graph showing the fit between model results and selected wellhead pressure data for the 7 September 1987 drawdown test in well A3 n log–log plot. The derivative shows a minimum characteristic of dual-porosity reservoirs.
that the reservoir was a homogenous, porous media system. • The skin effect is negative, indicating a good hydraulic connection between the well and the reservoir and that the wellbore and surrounding reservoir was not damaged during drilling. • Since no injection tests have been performed, we calculated the injectivity index using the WellTester code and parameters measured and derived from the 7 September 1987 production test. The computed injectivity index is close to 7 L/(s bar).
4. A3 wellbore simulation Well A3 was discharged into a silencer that also acts as a steam/water separator at atmospheric pressure. The flow of the liquid phase was measured using the weir box method and that of the steam by measuring critical lip pressure before it was discharged into the air (Grant et al., 1982). The lip pressure method is not quite as accurate as the separator method, but is commonly used because a minimum of hardware and instrumentation is required to obtain quite good results. Wellbore simulator HOLA (Bjornsson et al., 1993) is used to simulate wellbore parameters (e.g. temperature, enthalpy, pressure). The temperature and pressure profiles in A3 measured on 11 November 1987 during a well discharge test were used in the HOLA analysis. It was assumed that: (1) there was no heat exchange between the well and the surrounding formation; (2) there was only one main feed zone at about 1075 m depth; and (3) the feedzone temperature was 263.6 ◦ C, corresponding to the one measured at that depth (see Fig. 3). After a lengthy process of trial-and-error and least-squares fitting by HOLA, a satisfactory agreement between computed and measured data was achieved (see Fig. 6). The modeling results are summarized in Table 3. The computed total flow rate for the design wellhead pressure of 22.5 bar g is 45 kg/s. This value is close to the one obtained during the August–September 1987 drawdown tests (i.e. 50 kg/s at 22 bar g). The calculated productivity index value for the 11 November 1987 production test is 2.8 L/(s bar). According to the HOLA simulation, the steam-flow rate was 5 kg/s (at 22.5 bar g). Assuming that the conversion rate is 2 kg/s of steam per MWe (Grant et al., 1982), the power generation potential of well A3 is about 2.5 MWe.
Fig. 6. Comparison between computed and measured temperature and pressure profiles in well A3 when producing at a rate of 45 kg/s (162 tons/h).
Table 3 HOLA simulation results for well A3 based on 11 November 1987 production test. Parameter
Value
Wellhead pressure (bar g) Wellhead temperature (◦ C) Wellhead dryness (%) Wellhead enthalpy (kJ/kg) Wellhead total flow (kg/s) Number of feed zones Depth of feed zone (m) Bottomhole enthalpy (kJ/kg) Productivity index (L/(s bar))
22.5 218.4 10.93 1140 45 1 ∼1075 1150.5 2.8
5. Reservoir response simulation Reliable models of geothermal systems are essential for predicting changes resulting from fluid production and/or injection operations and to estimate their production capacity under different reservoir management plans. O’Sullivan et al. (2001) presented the state-of-art advances and emerging trends in geothermal reservoir simulation. Even though that review was done almost 10 years ago, it is still relevant.
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
225
Table 4 Estimated reservoir properties for the Asal geothermal reservoir based on a lumped parameter model, well A3 data (Figs. 6 and 7) and assuming a confined system and 2D radial flow. Model type
Reservoir volume (km3 )
Area (km2 )
Permeabilitythickness (Dm)
Closed two-tank system Open two-tank system
10.5 9.7
4.2 3.9
13 11.6
Fig. 7. Wellhead pressure and mass extraction rates for well A3 during an August–November 1987 test.
A lumped-parameter model of the Gale le Goma system was used to simulate the observed pressure decline (drawdown) in well A3 during production testing. This type of model consists of a few capacitors or tanks that are connected by resistors as described by Axelsson (1989). The computer program LUMPFIT, which employs a nonlinear, iterative, least-squares procedure, has been used successfully to simulate data on pressure changes caused by mass extraction and/or injection in geothermal systems. The code uses an inverse method that requires much less time and operator intervention than direct or forward modeling (Axelsson and Arason, 1992; Axelsson et al., 2005). The production rates and reservoir pressures for the AugustNovember 1987 well tests were used as input in the LUMPFIT program (Fig. 7). The comparison between observed and calculated reservoir pressures is presented in Fig. 8. The best fits are obtained when using two-tank closed and open models. Both yield similarly acceptable matches, providing a correlation coefficient (R) of 96.5% and a standard deviation of 0.14 bars. The lumped-parameter models were used to estimate the overall average reservoir properties of the Gale le Goma geothermal field assuming that the reservoir was liquid-dominated and that storage was governed mainly by liquid and formation compressibility. Table 4 shows the modeling results based on a radial flow (2D) model and assuming a 2500-m reservoir thickness (h). Based on the volumes estimated by the two lumped-parameter models (about 10 km3 ) the reservoir would extend over an area of about 4 km2 when assuming a confined system; this surface area is comparable to the one indicated by the 2007–2008 MT and TEM surveys (Arnason et al., 2008). The estimated reservoir
Fig. 9. Comparison of wellhead pressures predicted in well A3 by a two-tank, open, lumped-parameter model for three assumed production rates.
permeability-thickness is about 12 darcy-meter, which is somewhat higher than the value obtained from well test analyses (Tables 1 and 2). The main objective of modeling geothermal systems is to assess their production potential. The maximum allowable drawdown in the field in question determines the maximum potential of the system. The lumped models were used to predict reservoir pressure changes under different rates of fluid production. During the August–November 1987 A3 production test, the average mass extraction rate was 50 kg/s, and the wellhead pressure decreased from about 20.5 to about 18 bar (Fig. 7). Three production scenarios (20, 30 and 50 kg/s; Fig. 9) were used to simulate reservoir response to production during a 1-year period. For commercial development only the 50 kg/s case would be realistic, but the two lumped-parameter models estimate that the reservoir drawdown would be between about 14 and 16 bars after 1 year. These pressure drops are much too high and injection should be considered to compensate for the low natural recharge. 6. Volumetric resource assessment
Fig. 8. Graph showing the match between wellhead pressure measured in well A3 and those computed using lumped-parameter (two-tank closed and open) models.
Different methods used to assess geothermal resources are described by Muffler and Christiansen (1978) and Muffler and Cataldi (1978). These authors also discuss the theoretical background, assumptions and development of the equations used in these models. The most commonly used, particularly during the initial phases of a geothermal project mainly because of lack of sufficient data, is the volumetric method. Due to the simplifying assumption made, the volumetric method tends to be much less accurate than the numerical models. We apply the volumetric method to the Asal area to get an initial capacity estimate. The purpose of our assessment is to estimate the electricity generation capacity for the region assuming a 25-year power plant life. Because most of the parameters assumed in the calculations are associated with considerable uncertainties, we use a Monte Carlo technique.
226
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
Table 5 Most likely values and probability distributions for the patameters used to estimate power generation capacities in the different Asal geothermal fields. Input parameter
Units
Assumed distribution
Best guess
Area
km2
Triangular
4 Gale 4 Fiale 3South
Thickness Reference temperature (Tref)
◦
m C
Triangular Constant
1500 40
Reservoir temperature (Tres )
◦
Triangular
280 Gale 280 Fiale 280 South
Time period (t) Recovery factor (Rf ) Fluid density (fluid ) Fluid heat capacity (ˇfluid ) Rock density (rock ) Rock heat capacity (ˇrock ) Reservoir porosity () Conversion efficiency () Load factor (L)
Years % kg/m3 J/(kg ◦ C) kg/m3 J/(kg ◦ C) % % %
Constant Triangular Constant Constant Constant Triangular Constant Constant Constant
25 0.1 890 4800 2870 900 10 40 95
C
6.1. Parameters used in the model Based on the resistivity surveys conducted by ISOR (Arnason et al., 2008), the median estimated surface areas of the three geothermal systems, Gale le Goma, Fiale and South of Lake Asal, are 4, 4 and 3 km2 , respectively. From the temperatures profiles (Fig. 3) and the MT and TEM data (Arnason et al., 2008), the thickness of these reservoirs is assumed to vary between 1500 and 2000 m. In the calculations, a rejection (condenser) temperature (Tref ) of 40 ◦ C was selected (MIT, 2006). The reservoir temperature (Tres ) was assumed to be homogenous, 280 ◦ C being the most likely, 265 ◦ C the minimum, 290 ◦ C the maximum for both Gale le Goma and South and 350 ◦ C the maximum for Fiale. This temperature range was based on the temperature profiles of wells A6, A3 and A4 drilled in the Gale le Goma area (Fig. 3). Because of the high salinity of the fluid in the deep Asal reservoir, our calculations assume a fluid average density (fluid ) of 890 kg/m3 (Aquater, 1989) and a heat capacity (ˇfluid ) of 4800 J/(kg ◦ C). The deep Asal reservoirs are considered to be composed mostly of metamorphic rocks (Aquater, 1989); therefore, the value for the heat capacity of the rock, ˇrock , is chosen to be between 0.8 and 1 kJ/kg ◦ C, the values measured experimentally by Vosteen and Schellschmidt (2003). The porosity ϕ of the basalt rocks in the Asal area is of the order of 10% (Aquater, 1989). The average density of the rock (rock ) is set at 2870 kg/m3 on the basis of the gravity data collected by BRGM in 1979 (Aquater, 1989). The geothermal recovery factor Rf , which represents the ratio of extracted thermal energy (measured at the wellhead) to the total thermal energy contained in a given subsurface volume of rock and water (Muffler and Cataldi, 1978), is assumed to be between 0.05 and 0.2, with 0.1 being the best guess. The electrical conversion efficiency () is taken to be 40%, like in the case of the Wairakei power plant (Thain and White, 1993). A 95% load factor L is used, based on 18 days of preventive maintenance per year. All the input parameters used in the volumetric assessment are listed in Table 5. 6.2. Results and discussion The results of the volumetric assessment of the electricity generation potential of the three geothermal fields in the Asal region are summarized in Table 6. These estimated generation capacities are only preliminary estimates since they are based on parameters with considerable uncertainties (in the South Asal Lake area, no
Minimum
Maximum
2 2 2
6 6 8
1000
2000
265 265 265
290 350 290
0.05
800
0.2
1000
Table 6 Estimated power generation capacity of the three Asal geothermal fields based on the volumetric method and Monte Carlo calculations. Geothermal field
Estimated capacity (MWe) Minimum
Gale le Goma Fiale South of Asal Lake Total
Median
Maximum
37 42 36
62 71 65
99 116 114
115
198
329
Minimum, median and maximum are the P90, P50, and P10 values. Here Pxx indicates that there is a xx probability that the power generation capacity is at least the estimated value.
wells have been drilled), particularly in the size of the reservoirs and in the assumed recovery factor. At the present time, the Fiale field is under development. A 50-MWe geothermal power plant might be installed there in the next few years. This size of plant might be realistic as the mean power capacity for the Fiale field is estimated to be 71 MWe and the 10–90% confidence interval of the estimate is 42–116 MWe (Table 6). 7. Conclusions The main results of our reservoir engineering analysis of data from the Asal geothermal systems are: • The three geothermal system identified so far in the Asal area, Gale le Goma, Fiale and South of Lake Asal, do not seem be interconnected based on the MT and TEM geophysical surveys performed by Isor at the end of 2007. • The Gale le Goma deep reservoir is heterogeneous and best simulated using a dual-porosity model. Considering the geology of the Asal area, the other two geothermal systems are expected to be of the same type. • The estimated power production capacity of Gale le Goma well A3 is 2.5 MWe. Its main feed zone is at about 1075 m depth. • To reduce reservoir pressure drawdown, injection should be considered. • Due to the high salinity of the geothermal fluid the possibility of mineral scaling in the wells and surface equipment should be investigated. • Using a volumetric method, the combined electricity generation potential of the three Asal geothermal systems is estimated to be between 115 and 329 MWe (Table 6). As more data on these
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
areas become available, numerical studies should be performed to better determine electricity generation capacity of the Asal region. Acknowledgements Thanks to Reykjavik Energy Invest for allowing us to use data from the geophysical survey made in the Asal Geothermal System. Our appreciation goes also to ISOR for permitting us to use the @RISK software, Dr. Jalludin Mohamed, Director of the Centre d’Etudes et de Recherche de Djibouti for giving us access to Asal data, Mr. Knutur Arnason for his critical advice on geophysical data interpretation and to Dr. Sabodh K. Garg for his help with calculations using volumetric method. References Aquater, 1989. Djibouti Geothermal Exploration Project Republic of Djibouti: Final Report. Aquater SpA, Lorenzo in Campo, Pesaro, Italy, 159 pp. Arnason, K., Eysteinsson, H., Vilhjalmsson, A.M., 2008. The Asal Geothermal Field, Djibouti, ISOR Report 2008/019. ISOR Geophysical Surface Exploration, 70 pp. Axelsson, G., 1989. Simulation of pressure response data from geothermal reservoir by lumped parameter model. In: Proceedings of the 14th Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, CA, USA, pp. 257–263. Axelsson, G., Arason, Th., 1992. LUMPFIT, Automated Simulation of Pressure Changes in Hydrological Reservoir, User’s Guide Version 3.1. Orkustofnun Report, Reykjavík, Iceland, 32 pp. Axelsson, G., Björnsson, G., Quijano, J.E., 2005. Reliability of lumped parameter modelling of pressure changes in geothermal reservoirs. In: Proceedings of the 2005 World Geothermal Congress, Antalya, Turkey, April 24–29, 2005, paper 1179, 8 pp. Barberi, F., Ferrara, G., Santacroce, R., Varet, J., 1975. Structural evolution of the Afar triple junction. In: Pilger, A., Rösler, A. (Eds.), Afar Depression of Ethiopia, vol. 1. Schweizerbart, Stuttgart, Germany, pp. 38–54. Beyene, A., Abdelsalam, M.G., 2005. Tectonics of the afar depression: a review and synthesis. Journal of African Earth Sciences 41, 41–59. Bjornsson, G., Arason, P., Bodvarsson, G.S., 1993. The Wellbore Simulator HOLA. Version 3.1. User’s Guide. Orkustofnun, Reykjavík, Iceland, 36 pp. Bourdet, D., Whittle, T.M., Douglas, A.A., Pirard, Y.M., 1983 May. A new set of type curves simplifies well test analysis. World Oil, 95–106. Bourdet, D., Ayoub, J.A., Pirard, Y.M., 1989. Use of pressure derivative in well-test interpretation. SPE Formation Evaluation 4 (20), 293–302.
227
BRGM, 1973. Etude géophysique par sondages Melos et sondages électriques de la région du lac Asal (TFAI). Report. Bureau de Recherches Géologiques et Minières 73SGN010GTH, Orléans, France, 18 pp. Correia, H., Fouillac, C., Gerard, A., Varet, J., 1985. The Asal geothermal field, Republic of Djibouti. Geothermal Resources Council Transaction 9, 513–519. D’Amore, F., Giusti, D., Abdallah, A., 1998. Geochemistry of the high-salinity geothermal field of Asal, Republic of Djibouti, Africa. Geothermics 27, 197– 210. Daher, E.H., 2005. Analysis of geothermal well test data from the Asal rift area, Republic of Djibouti. In: GTP Yearbook 2005, Report No. 6. United Nations University Geothermal Training Program, Reykjavik, Iceland, 21 pp. Doubre, C., Manighetti, I., Dorbath, L., Dorbath, C., Bertil, D., Delmond, J.C., 2007. Crustal structure and magmato-tectonic processes in an active rift (AsalGhoubbet, Afar, EastAfrica). 2. Insights from the 23-year recording of seismicity since the last rifting event. Journal of Geophysical Research 112, B05406, doi:10.1029/2006JB004333. EDD, 2009. Electricity Cost Reduction Assessment & Energy Access Sector Strategy. Unpublished report ESMAP 097/06. Energy Sector Management Assistance Program, Djibouti, 158 pp. Earlougher, R.C., Jr., 1977. Advances in well test analysis. Society of Petroleum Engineers Monograph Series 5, Dallas, TX, USA, 264 pp. Grant, M.A., Donaldson, I.G., Bixley, P.F., 1982. Geothermal Reservoir Engineering. Academic Press Ltd., New York, NY, USA, 369 pp. Horne, R.N., 1995. Modern Well Test Analysis—A Computer-Aided Approach, second edition. Petroway, Inc., Palo Alto, CA, USA, 257 pp. Juliusson, E., Gretarsson, G.J., Jonsson, P., 2007. Well Tester 1.0b, User’s Guide. ISOR, Reykjavík, Iceland, 26 pp. Majer, E.L., Baria, R., Stark, M., Oates, S., Bommer, J., Smith, B., Asanuma, H., 2007. Induced seismicity associated with enhanced geothermal systems. Geothermics 36, 185–222. MIT, 2006. The Future of Geothermal Energy: Impact of Enhanced Geothermal Systems [EGS] on the United States in the 21st Century. Massachusetts Institute of Technology, Cambridge, MA, USA, Available on World Wide Web at: http://www1.eere.energy.gov/geothermal/future geothermal.html. Muffler, L.J.P., Cataldi, R., 1978. Methods for regional assessment of geothermal resources. Geothermics 7, 53–89. Muffler, L.J.P., Christiansen, R.L., 1978. Geothermal resource assessment of the United States. Pure and Applied Geophysics 117, 160–171. O’Sullivan, M.J., Pruess, K., Lippmann, M.J., 2001. State of the art of geothermal reservoir simulation. Geothermics 30, 395–429. Thain, I.A., White, B.R., 1993. Wairakei power development efficiency studies. In: Proceedings of the 15th New Zealand Geothermal Workshop, Auckland University, Auckland, New Zealand, pp. 41–45. Virkir-Orkint, 1990. Djibouti. Geothermal Scaling and Corrosion Study. Report. Virkir-Orkint, Reykjavik, Iceland, 109 pp. Vosteen, H.D., Schellschmidt, R., 2003. Physics and Chemistry of the Earth 28, 499–509.
Contents lists available at ScienceDirect
Geothermics journal homepage: www.elsevier.com/locate/geothermics
Geothermal resources in the Asal Region, Republic of Djibouti: An update with emphasis on reservoir engineering studies Daher E. Houssein a,∗ , Gudni Axelsson b a b
Centre d’Etudes et de Recherche de Djibouti, Route de l’aéroport, CERD, Earth Science, B.P. 486 Djibouti, Djibouti ÍSlenskar Orkurannsóknir (ISOR), 9 Grensasvegi, Reykjavik, 108, Iceland
a r t i c l e
i n f o
Article history: Received 23 July 2009 Accepted 16 June 2010
Keywords: Geothermal Well testing Wellbore simulation Lumped-parameter model Resource assessment Asal Djibouti
a b s t r a c t Three independent geothermal systems have been identified, so far, in the Asal region of the Republic of Djibouti (i.e. Gale le Goma, Fiale and South of Lake). Six deep wells have been drilled in the region, the first two in 1975 and the others in 1987–88. Well A2 was damaged and wells A4 and A5 encountered impermeable yet very hot (340–365 ◦ C) rocks. Wells A1, A2, A3 and A6 produce highly saline (120 g/L TDS) fluids leading to mineral scaling. Well test data indicate that the reservoir might be producing from fractured and porous zones. The estimated permeability-thickness of the deep Gale le Goma reservoir is in the 3–9 darcy-meter range. Lumped-parameter modeling results indicate that well A3 should be operated at about 20 kg/s total flow rate and that injection should be considered to reduce pressure drawdown. The estimated power generation potential of well A3 is 2.5 MWe, and that of all Asal high-temperature hydrothermal systems is between 115 and 329 MWe for a 25-year exploitation period. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Neither fossil fuel resources nor hydroelectric potential exist in the Republic of Djibouti. The lack of these energy resources makes the country heavily dependent on imported oil products, mainly from Saudi Arabia or Dubai. According to data from the stateowned power utility, Electricite de Djibouti (EDD), an estimated 49.7% of the population has access to electricity, 99.5% of which is urban. The average electricity cost was equivalent to US$ 0.3/kWh in 2009, with commercial establishments paying about 40% more than domestic consumers (EDD, 2009). This high cost affects all sectors of society and is considered to be the main barrier to Djibouti’s industrial development. The Republic of Djibouti (surface area 23,000 km2 ) is located in the Horn of Africa (Fig. 1), at the Afar Triple Junction where the Red Sea, Gulf of Aden and East African Rift systems meet (Barberi et al., 1975). Given its geodynamic location, commercial-scale geothermal resources can likely be found in the country. The first geothermal investigations of the Asal area, located in the Afar depression, were undertaken in 1970 and included geological and geochemical studies, as well as geophysical exploration (BRGM, 1973). These studies identified the high-temperature Gale le Goma system, where four deep wells have been drilled (Fig. 2),
∗ Corresponding author. Tel.: +253 35 27 95; fax: +253 35 48 12. E-mail address: [email protected] (D.E. Houssein). 0375-6505/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.geothermics.2010.06.006
but also other promising geothermal areas. This particular geological setting is now under exploration by Reykjavik Energy Invest (REI) with the intention to build a 50-MWe power plant expected to be in operation in 2013. If the area can support and sustain a greater generation capacity, the aim is to expand the plant or build additional ones when there is sufficient demand. The goal is to produce electricity at a lower cost than that generated by burning imported fossil fuels. The main objective of this paper is to add to the work done previously by Aquater (1989) and Daher (2005) for the Asal region, by the following reservoir engineering techniques: • Interpretation (i.e. re-interpretation) of Gale le Goma well A3 well test data using decline curve derivative analysis (Horne, 1995) and numerical modeling (O’Sullivan et al., 2001). The results are then compared to those of conventional log–log, semilog and match point methods used by Aquater (1989) and Daher (2005). Well test modeling software developed by Iceland GeoSurvey (ISOR) (Juliusson et al., 2007) is used to compare estimated hydrological parameters. • The wellbore simulator HOLA (Bjornsson et al., 1993) is used to simulate wellbore conditions (temperature, pressure, etc.) that influence the transport of geothermal fluid from the reservoir to the surface when well A3 is produced. Then the computed productivity index (PI) is compared against the measured injectivity indices (II), defined as the change in injection flow rate divided by the change in stabilized reservoir pressure, derived from injection tests.
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
221
Fig. 1. Map of Djibouti showing the location of the study area. The Afar depression (or triangle) includes parts of Djibouti, Eritrea and Ethiopia.
• Estimation of the size, permeability-thickness product and long-term geothermal production potential of the Asal area using a lumped-parameter model and data from a November 1987 production test performed in the Gale le Goma well A3. The computed reservoir parameters are compared with those obtained from geophysical surveys and conventional well tests.
• Assessment of the Asal geothermal resource using a volumetric method (Muffler and Cataldi, 1978) and parameters derived from resistivity survey data (Arnason et al., 2008), geologic-structural information, results from well A3 production tests and temperature profiles of the six wells drilled in the area. • Estimation of the total accessible energy in the Asal geothermal region based on a Monte Carlo simulation that incorporates
Fig. 2. Map showing the location of the three known geothermal systems (Gale le Goma, Fiale and South of Lake) in the Asal region. The figure also shows the resistivity at 3000 m b.s.l., inferred lineaments in low-resistivity (red lines), seismicity (dark green dots) and geothermal surface manifestations (light green). The two super saline and sealed-off Gale Le Goma and South of Asal Lake systems are indicated, as is the more open, lower salinity Fiale system under Lava Lake (modified from Arnason et al., 2008). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
222
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
uncertainty in the parameters used in the volumetric resource assessment method. The immediate objective of the study presented here is to evaluate the geothermal potential of the Asal region. All the parameters used come from the Gale le Goma system since it is the one that has been investigated till the present time. 2. Geothermal exploration in the Asal Rift 2.1. Geography and geology The Asal Rift is a northwestward extension of the Gulf of AdenTadjura. It extends from the Gulf of Goubet in the SE to Lake Asal in the NW. The lake, which is 155 m below sea level (m b.s.l.; Fig. 2), is 9–10 km wide and is bounded by major normal faults to the SW and NE. The rift due to its unique nature as a landward extension of an oceanic ridge (the Tadjura Ridge) has been studied quite extensively (Barberi et al., 1975; Beyene and Abdelsalam, 2005). The most active part of the Afar depression is the so-called Inner Rift. It is about 3 km wide and is located in the SE part of the main rift (Correia et al., 1985). It hosts most of the recent volcanism, with the last eruption occurring in 1978. The Inner Rift is characterized by intense fracturing and recent lava flows. Its most prominent feature is Lava Lake (sometimes referred to as Fiale), a phreatic crater about 1.3 km wide whose floor is covered by recent basaltic lavas. Based on analysis of drill cuttings and cores from wells drilled in the rift, the stratigraphic column in the area is composed by (from top to bottom) of the Asal series (basalts and hyaloclastites less than 1 My old), the Afar Stratoid series (basalts with some rhyolites 1–4 My old), and the Dala Basalt (basaltic lava flows with some intercalations of rhyolites and trachytes 4–9 My old). 2.2. Wells characteristics In 1975, two deep wells, A1 (1554 m) and A2 (1147 m), were drilled in the SW part of the Asal Rift; i.e. in the “old wellfield” or Gale le Goma area (see Fig. 2). A1 encountered a feed zone at 1137 m depth, while A2 showed no permeability but temperatures above 260 ◦ C (Aquater, 1989). A1 produced about 38 kg/s of a two-phase fluid (20% steam at 6 bar wellhead pressure). Four deep wells were drilled in 1987–88. Wells A3 and A6, located in the same area as A1 and A2, close to a phreatic crater (Fig. 2), encountered the same reservoir as A1 with temperatures of about 280 ◦ C (Daher, 2005). Well A4, drilled about 1.5 km NE of the phreatic crater, showed temperatures close to the boiling curve at elevations lower than 200 m b.s.l. and about 350 ◦ C at bottomhole. However, only low permeabilities were encountered in A4. Downhole temperatures profiles measured in wells A3-A5 are shown in Fig. 3. No wells have been drilled in South of Lake Asal. Based on the resistivity map (Fig. 2), we assume that in this area the same maximum reservoir temperature (∼280 ◦ C) as in Gale le Goma will be encountered. Well A5, drilled in the Inner Rift (or Fiale area) about 700 m west of Lava Lake (Fig. 2), presented sharply increasing temperature below 200 m b.s.l. with a local maximum of about 180 ◦ C at 400 m b.s.l. Below that elevation the rocks have cooled dramatically, down to 60–70 ◦ C around 800 m b.s.l., as compared to temperatures indicated by alteration mineralogy. Below 800 m b.s.l. the temperature rises steeply with depth and reaches about 350 ◦ C at bottomhole. Very low permeabilities were found in A5; higher values are expected in other parts of this area. Even if the permeability is very low in this well (around 0.4 darcy-meter; Daher, 2005), it may be increased by hydraulic stimulation. Since this operation involves
Fig. 3. Temperature profiles of Asal wells A2–A5.
injecting high-pressure water into the well, and as observed in some enhanced geothermal systems (EGS) projects, it presents a higher likelihood of producing micro-earthquakes that could be felt by the local population and should be carefully monitored (Majer et al., 2007). 2.3. Geophysics In 1988 a group from the National Energy Authority of Iceland (Orkustofnun) performed a central-loop transient electromagnetic (TEM) resistivity survey in the Asal Rift. In November–December 2007, the Iceland Geosurvey (ISOR) conducted TEM and MT (magnetotelluric) resistivity surveys to gather additional data needed to develop a conceptual model of the area that could be used to site future exploration wells. These studies showed that the saline groundwater table drops sharply from about sea level NE of the Gale le Goma area to about the level of Lake Asal in a narrow zone perpendicular to the Rift, just west of Lava Lake. Furthermore, the surveys identified a lowresistivity anomaly under Lava Lake and a local rise in groundwater to above sea level. This was interpreted as indicating a geothermal system under Lava Lake (Fiale area), not hydraulically connected to the Gale le Goma area (Arnason et al., 2008). Considerable microseismic activity was recorded in the Asal Rift in late 2000 and early 2001. A comprehensive study of these events and a review of earlier data showed that seismicity is mainly concentrated in the uppermost 3 km under Lava Lake (Doubre et al., 2007). Also in this area, very hot and ductile rocks appear to be present below a depth of 3 km (Arnason et al., 2008; Doubre et al., 2007). Based on these exploration studies, three separate geothermal fields were identified in the Asal area: Gale le Goma, Fiale (under Lava Lake) and South of Lake (Fig. 2). 2.4. Geochemistry The main results of geochemical studies of water from the Asal area are (Correia et al., 1985; Aquater, 1989; D’Amore et al., 1998): • According to a study performed on samples collected from shallow aquifers in wells Asal 1, 2, 3, 4 and 6, and deep water samples from Asal 3, as well as from samples of Lake Asal, it appears that all aquifers encountered are mainly recharged by sea water (Aquater, 1989).
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
223
Table 1 Results from analysis with different methods of drawdown tests data for A3 (Aquater, 1989; Daher, 2005). Date (1987)
q (tons/h)
q (tons/h)
8 August 24 August 7 September 9 September
225–300 79–130 155–225 300–357
75 51 70 57
Time (h)
11 5 20 2
Semilog
Type curve
kh (Dm)
Skin
kh (Dm)
4.2 14 5.75 14
−5 −5 −5 −5
4.3 13.6 7 15.9
q: flow rate; q: flowrate difference between two successive steps in the production test; kh: permeability-thickness product (transmissivity).
• The Asal geothermal fluids are mixtures of sea water and continental water of meteoric origin with high total dissolved solids (TDS). • The geothermal fluid is not produced by the evaporation of sea water as in the case of the Asal Lake NaCl brine. • The equilibrium temperature calculated from all reactive gaseous species (H2 O, CO2 , H2 , CH4 , CO, N2 , NH3 ), apart from H2 S, is about 260 ◦ C, which is compatible with measured reservoir temperature. • Lake Asal waters are composed of very concentrated sea water due to evaporation, and its CaSO4 content is modified mainly by precipitation. Deep geothermal waters seem to have no contact with Lake Asal waters and their Ca/Mg ratio is extremely different. • The fluids collected in well A5 showed that the water at the center of the rift has a much higher salinity than water near its edges. Due to the adverse brine chemistry — the high TDS (about 116 g/L) that would cause scaling problems in the wells and surface equipment (Virkir-Orkint, 1990) — Reykjavik Energy has diverted its focus from the Gale le Goma system to the Fiale area (Fig. 2). Results of the TEM-MT resistivity surveys carried out in 1988 and 2007, along with earthquake monitoring data, indicate that a separate geothermal system may reside under Lava Lake. Due to the intensive rifting activity in that area and its proximity to the sea, the geothermal fluids at Fiale are expected to be less saline than at Gale le Goma (Doubre et al., 2007). The Gale le Goma site, located at about the same distance from the sea as Fiale, is less active (Doubre et al., 2007) and seems to be a closed reservoir based on the TEM-MT study (Anasson et al., 2008) and well and production test data. 3. Well test analysis Well test data collected during injection and production tests conducted in well A3 by Aquater in August–September 1987 were analyzed using: • Well testing techniques (e.g. log–log, semilog, match point, multirate step) described by Earlougher (1977) and Horne (1995). • Derivative plots to examine the effects of wellbore storage, recharge and barrier boundaries, leakage, delayed gravity response and fracture flow (Bourdet et al., 1983, 1989; Horne, 1995). • Computer code WellTester developed by ISOR (Juliusson et al., 2007) to manipulate and analyze well test (mainly multistep injection or production tests) data and perform regression analysis. 3.1. Previous analyses of A3 well test data Well A3 was drilled in 1987 to a total depth of 1316 m. The well has a 24.4 cm (9–5/8 in.) casing down to 1016 m depth and is completed as an open hole between 1016 and 1316 m, corresponding to the main permeable zones. An upper reservoir is located between 300 and 600 m. Several transient well production tests were con-
Fig. 4. Semilog graph corresponding to the four drawdown tests performed in well A3 in 1987 (Daher, 2005).
ducted in 1987, with pressure being measured at 1075 m depth, near the upper permeable zones of the deep reservoir. Data from four drawdown tests have been used to estimate reservoir parameters (Aquater, 1989). Table 1 summarizes all results of previous analyses of A3 well test data by the conventional type-curve and semilog methods (Aquater, 1989; Daher, 2005). The pressure transient curves corresponding to these four tests are presented on a semilogarithmic graph in Fig. 4. 3.2. Revised analysis of A3 well test data The results from nonlinear regression and derivative plot analysis using the WellTester code are summarized in Table 2; the computed values are best estimates based on nonlinear regression analysis. Only during the 7 September 1987 drawdown test did the flowing pressure stabilize, indicating that a semisteady state had been reached. Results based on that test are, therefore, considered as the most reliable and will be taken as the reference. Some conclusions can be drawn from the results of this analysis: • A dual-porosity reservoir model with infinite radial flow boundary, assuming constant skin and wellbore storage values, best fits the measured data; see Fig. 5. This result is consistent with the fact that the Gale le Goma reservoir is a heterogeneous, naturally fractured reservoir. Note that in the earlier analyses it was assumed Table 2 Results of linear regression and derivative plot methods on 1987 well A3 drawdown test data. Date (1987)
WellTester results Skin
kh (Dm)
PI (L/(s bar))
8 August 24 August 7 September 9 September
−0.5 −3.4 −3.2 −1.8
1.3 9.3 3.2 9.4
23 20 7.2 17
PI: productivity index.
224
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
Fig. 5. Log–log graph showing the fit between model results and selected wellhead pressure data for the 7 September 1987 drawdown test in well A3 n log–log plot. The derivative shows a minimum characteristic of dual-porosity reservoirs.
that the reservoir was a homogenous, porous media system. • The skin effect is negative, indicating a good hydraulic connection between the well and the reservoir and that the wellbore and surrounding reservoir was not damaged during drilling. • Since no injection tests have been performed, we calculated the injectivity index using the WellTester code and parameters measured and derived from the 7 September 1987 production test. The computed injectivity index is close to 7 L/(s bar).
4. A3 wellbore simulation Well A3 was discharged into a silencer that also acts as a steam/water separator at atmospheric pressure. The flow of the liquid phase was measured using the weir box method and that of the steam by measuring critical lip pressure before it was discharged into the air (Grant et al., 1982). The lip pressure method is not quite as accurate as the separator method, but is commonly used because a minimum of hardware and instrumentation is required to obtain quite good results. Wellbore simulator HOLA (Bjornsson et al., 1993) is used to simulate wellbore parameters (e.g. temperature, enthalpy, pressure). The temperature and pressure profiles in A3 measured on 11 November 1987 during a well discharge test were used in the HOLA analysis. It was assumed that: (1) there was no heat exchange between the well and the surrounding formation; (2) there was only one main feed zone at about 1075 m depth; and (3) the feedzone temperature was 263.6 ◦ C, corresponding to the one measured at that depth (see Fig. 3). After a lengthy process of trial-and-error and least-squares fitting by HOLA, a satisfactory agreement between computed and measured data was achieved (see Fig. 6). The modeling results are summarized in Table 3. The computed total flow rate for the design wellhead pressure of 22.5 bar g is 45 kg/s. This value is close to the one obtained during the August–September 1987 drawdown tests (i.e. 50 kg/s at 22 bar g). The calculated productivity index value for the 11 November 1987 production test is 2.8 L/(s bar). According to the HOLA simulation, the steam-flow rate was 5 kg/s (at 22.5 bar g). Assuming that the conversion rate is 2 kg/s of steam per MWe (Grant et al., 1982), the power generation potential of well A3 is about 2.5 MWe.
Fig. 6. Comparison between computed and measured temperature and pressure profiles in well A3 when producing at a rate of 45 kg/s (162 tons/h).
Table 3 HOLA simulation results for well A3 based on 11 November 1987 production test. Parameter
Value
Wellhead pressure (bar g) Wellhead temperature (◦ C) Wellhead dryness (%) Wellhead enthalpy (kJ/kg) Wellhead total flow (kg/s) Number of feed zones Depth of feed zone (m) Bottomhole enthalpy (kJ/kg) Productivity index (L/(s bar))
22.5 218.4 10.93 1140 45 1 ∼1075 1150.5 2.8
5. Reservoir response simulation Reliable models of geothermal systems are essential for predicting changes resulting from fluid production and/or injection operations and to estimate their production capacity under different reservoir management plans. O’Sullivan et al. (2001) presented the state-of-art advances and emerging trends in geothermal reservoir simulation. Even though that review was done almost 10 years ago, it is still relevant.
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
225
Table 4 Estimated reservoir properties for the Asal geothermal reservoir based on a lumped parameter model, well A3 data (Figs. 6 and 7) and assuming a confined system and 2D radial flow. Model type
Reservoir volume (km3 )
Area (km2 )
Permeabilitythickness (Dm)
Closed two-tank system Open two-tank system
10.5 9.7
4.2 3.9
13 11.6
Fig. 7. Wellhead pressure and mass extraction rates for well A3 during an August–November 1987 test.
A lumped-parameter model of the Gale le Goma system was used to simulate the observed pressure decline (drawdown) in well A3 during production testing. This type of model consists of a few capacitors or tanks that are connected by resistors as described by Axelsson (1989). The computer program LUMPFIT, which employs a nonlinear, iterative, least-squares procedure, has been used successfully to simulate data on pressure changes caused by mass extraction and/or injection in geothermal systems. The code uses an inverse method that requires much less time and operator intervention than direct or forward modeling (Axelsson and Arason, 1992; Axelsson et al., 2005). The production rates and reservoir pressures for the AugustNovember 1987 well tests were used as input in the LUMPFIT program (Fig. 7). The comparison between observed and calculated reservoir pressures is presented in Fig. 8. The best fits are obtained when using two-tank closed and open models. Both yield similarly acceptable matches, providing a correlation coefficient (R) of 96.5% and a standard deviation of 0.14 bars. The lumped-parameter models were used to estimate the overall average reservoir properties of the Gale le Goma geothermal field assuming that the reservoir was liquid-dominated and that storage was governed mainly by liquid and formation compressibility. Table 4 shows the modeling results based on a radial flow (2D) model and assuming a 2500-m reservoir thickness (h). Based on the volumes estimated by the two lumped-parameter models (about 10 km3 ) the reservoir would extend over an area of about 4 km2 when assuming a confined system; this surface area is comparable to the one indicated by the 2007–2008 MT and TEM surveys (Arnason et al., 2008). The estimated reservoir
Fig. 9. Comparison of wellhead pressures predicted in well A3 by a two-tank, open, lumped-parameter model for three assumed production rates.
permeability-thickness is about 12 darcy-meter, which is somewhat higher than the value obtained from well test analyses (Tables 1 and 2). The main objective of modeling geothermal systems is to assess their production potential. The maximum allowable drawdown in the field in question determines the maximum potential of the system. The lumped models were used to predict reservoir pressure changes under different rates of fluid production. During the August–November 1987 A3 production test, the average mass extraction rate was 50 kg/s, and the wellhead pressure decreased from about 20.5 to about 18 bar (Fig. 7). Three production scenarios (20, 30 and 50 kg/s; Fig. 9) were used to simulate reservoir response to production during a 1-year period. For commercial development only the 50 kg/s case would be realistic, but the two lumped-parameter models estimate that the reservoir drawdown would be between about 14 and 16 bars after 1 year. These pressure drops are much too high and injection should be considered to compensate for the low natural recharge. 6. Volumetric resource assessment
Fig. 8. Graph showing the match between wellhead pressure measured in well A3 and those computed using lumped-parameter (two-tank closed and open) models.
Different methods used to assess geothermal resources are described by Muffler and Christiansen (1978) and Muffler and Cataldi (1978). These authors also discuss the theoretical background, assumptions and development of the equations used in these models. The most commonly used, particularly during the initial phases of a geothermal project mainly because of lack of sufficient data, is the volumetric method. Due to the simplifying assumption made, the volumetric method tends to be much less accurate than the numerical models. We apply the volumetric method to the Asal area to get an initial capacity estimate. The purpose of our assessment is to estimate the electricity generation capacity for the region assuming a 25-year power plant life. Because most of the parameters assumed in the calculations are associated with considerable uncertainties, we use a Monte Carlo technique.
226
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
Table 5 Most likely values and probability distributions for the patameters used to estimate power generation capacities in the different Asal geothermal fields. Input parameter
Units
Assumed distribution
Best guess
Area
km2
Triangular
4 Gale 4 Fiale 3South
Thickness Reference temperature (Tref)
◦
m C
Triangular Constant
1500 40
Reservoir temperature (Tres )
◦
Triangular
280 Gale 280 Fiale 280 South
Time period (t) Recovery factor (Rf ) Fluid density (fluid ) Fluid heat capacity (ˇfluid ) Rock density (rock ) Rock heat capacity (ˇrock ) Reservoir porosity () Conversion efficiency () Load factor (L)
Years % kg/m3 J/(kg ◦ C) kg/m3 J/(kg ◦ C) % % %
Constant Triangular Constant Constant Constant Triangular Constant Constant Constant
25 0.1 890 4800 2870 900 10 40 95
C
6.1. Parameters used in the model Based on the resistivity surveys conducted by ISOR (Arnason et al., 2008), the median estimated surface areas of the three geothermal systems, Gale le Goma, Fiale and South of Lake Asal, are 4, 4 and 3 km2 , respectively. From the temperatures profiles (Fig. 3) and the MT and TEM data (Arnason et al., 2008), the thickness of these reservoirs is assumed to vary between 1500 and 2000 m. In the calculations, a rejection (condenser) temperature (Tref ) of 40 ◦ C was selected (MIT, 2006). The reservoir temperature (Tres ) was assumed to be homogenous, 280 ◦ C being the most likely, 265 ◦ C the minimum, 290 ◦ C the maximum for both Gale le Goma and South and 350 ◦ C the maximum for Fiale. This temperature range was based on the temperature profiles of wells A6, A3 and A4 drilled in the Gale le Goma area (Fig. 3). Because of the high salinity of the fluid in the deep Asal reservoir, our calculations assume a fluid average density (fluid ) of 890 kg/m3 (Aquater, 1989) and a heat capacity (ˇfluid ) of 4800 J/(kg ◦ C). The deep Asal reservoirs are considered to be composed mostly of metamorphic rocks (Aquater, 1989); therefore, the value for the heat capacity of the rock, ˇrock , is chosen to be between 0.8 and 1 kJ/kg ◦ C, the values measured experimentally by Vosteen and Schellschmidt (2003). The porosity ϕ of the basalt rocks in the Asal area is of the order of 10% (Aquater, 1989). The average density of the rock (rock ) is set at 2870 kg/m3 on the basis of the gravity data collected by BRGM in 1979 (Aquater, 1989). The geothermal recovery factor Rf , which represents the ratio of extracted thermal energy (measured at the wellhead) to the total thermal energy contained in a given subsurface volume of rock and water (Muffler and Cataldi, 1978), is assumed to be between 0.05 and 0.2, with 0.1 being the best guess. The electrical conversion efficiency () is taken to be 40%, like in the case of the Wairakei power plant (Thain and White, 1993). A 95% load factor L is used, based on 18 days of preventive maintenance per year. All the input parameters used in the volumetric assessment are listed in Table 5. 6.2. Results and discussion The results of the volumetric assessment of the electricity generation potential of the three geothermal fields in the Asal region are summarized in Table 6. These estimated generation capacities are only preliminary estimates since they are based on parameters with considerable uncertainties (in the South Asal Lake area, no
Minimum
Maximum
2 2 2
6 6 8
1000
2000
265 265 265
290 350 290
0.05
800
0.2
1000
Table 6 Estimated power generation capacity of the three Asal geothermal fields based on the volumetric method and Monte Carlo calculations. Geothermal field
Estimated capacity (MWe) Minimum
Gale le Goma Fiale South of Asal Lake Total
Median
Maximum
37 42 36
62 71 65
99 116 114
115
198
329
Minimum, median and maximum are the P90, P50, and P10 values. Here Pxx indicates that there is a xx probability that the power generation capacity is at least the estimated value.
wells have been drilled), particularly in the size of the reservoirs and in the assumed recovery factor. At the present time, the Fiale field is under development. A 50-MWe geothermal power plant might be installed there in the next few years. This size of plant might be realistic as the mean power capacity for the Fiale field is estimated to be 71 MWe and the 10–90% confidence interval of the estimate is 42–116 MWe (Table 6). 7. Conclusions The main results of our reservoir engineering analysis of data from the Asal geothermal systems are: • The three geothermal system identified so far in the Asal area, Gale le Goma, Fiale and South of Lake Asal, do not seem be interconnected based on the MT and TEM geophysical surveys performed by Isor at the end of 2007. • The Gale le Goma deep reservoir is heterogeneous and best simulated using a dual-porosity model. Considering the geology of the Asal area, the other two geothermal systems are expected to be of the same type. • The estimated power production capacity of Gale le Goma well A3 is 2.5 MWe. Its main feed zone is at about 1075 m depth. • To reduce reservoir pressure drawdown, injection should be considered. • Due to the high salinity of the geothermal fluid the possibility of mineral scaling in the wells and surface equipment should be investigated. • Using a volumetric method, the combined electricity generation potential of the three Asal geothermal systems is estimated to be between 115 and 329 MWe (Table 6). As more data on these
D.E. Houssein, G. Axelsson / Geothermics 39 (2010) 220–227
areas become available, numerical studies should be performed to better determine electricity generation capacity of the Asal region. Acknowledgements Thanks to Reykjavik Energy Invest for allowing us to use data from the geophysical survey made in the Asal Geothermal System. Our appreciation goes also to ISOR for permitting us to use the @RISK software, Dr. Jalludin Mohamed, Director of the Centre d’Etudes et de Recherche de Djibouti for giving us access to Asal data, Mr. Knutur Arnason for his critical advice on geophysical data interpretation and to Dr. Sabodh K. Garg for his help with calculations using volumetric method. References Aquater, 1989. Djibouti Geothermal Exploration Project Republic of Djibouti: Final Report. Aquater SpA, Lorenzo in Campo, Pesaro, Italy, 159 pp. Arnason, K., Eysteinsson, H., Vilhjalmsson, A.M., 2008. The Asal Geothermal Field, Djibouti, ISOR Report 2008/019. ISOR Geophysical Surface Exploration, 70 pp. Axelsson, G., 1989. Simulation of pressure response data from geothermal reservoir by lumped parameter model. In: Proceedings of the 14th Workshop on Geothermal Reservoir Engineering, Stanford University, Stanford, CA, USA, pp. 257–263. Axelsson, G., Arason, Th., 1992. LUMPFIT, Automated Simulation of Pressure Changes in Hydrological Reservoir, User’s Guide Version 3.1. Orkustofnun Report, Reykjavík, Iceland, 32 pp. Axelsson, G., Björnsson, G., Quijano, J.E., 2005. Reliability of lumped parameter modelling of pressure changes in geothermal reservoirs. In: Proceedings of the 2005 World Geothermal Congress, Antalya, Turkey, April 24–29, 2005, paper 1179, 8 pp. Barberi, F., Ferrara, G., Santacroce, R., Varet, J., 1975. Structural evolution of the Afar triple junction. In: Pilger, A., Rösler, A. (Eds.), Afar Depression of Ethiopia, vol. 1. Schweizerbart, Stuttgart, Germany, pp. 38–54. Beyene, A., Abdelsalam, M.G., 2005. Tectonics of the afar depression: a review and synthesis. Journal of African Earth Sciences 41, 41–59. Bjornsson, G., Arason, P., Bodvarsson, G.S., 1993. The Wellbore Simulator HOLA. Version 3.1. User’s Guide. Orkustofnun, Reykjavík, Iceland, 36 pp. Bourdet, D., Whittle, T.M., Douglas, A.A., Pirard, Y.M., 1983 May. A new set of type curves simplifies well test analysis. World Oil, 95–106. Bourdet, D., Ayoub, J.A., Pirard, Y.M., 1989. Use of pressure derivative in well-test interpretation. SPE Formation Evaluation 4 (20), 293–302.
227
BRGM, 1973. Etude géophysique par sondages Melos et sondages électriques de la région du lac Asal (TFAI). Report. Bureau de Recherches Géologiques et Minières 73SGN010GTH, Orléans, France, 18 pp. Correia, H., Fouillac, C., Gerard, A., Varet, J., 1985. The Asal geothermal field, Republic of Djibouti. Geothermal Resources Council Transaction 9, 513–519. D’Amore, F., Giusti, D., Abdallah, A., 1998. Geochemistry of the high-salinity geothermal field of Asal, Republic of Djibouti, Africa. Geothermics 27, 197– 210. Daher, E.H., 2005. Analysis of geothermal well test data from the Asal rift area, Republic of Djibouti. In: GTP Yearbook 2005, Report No. 6. United Nations University Geothermal Training Program, Reykjavik, Iceland, 21 pp. Doubre, C., Manighetti, I., Dorbath, L., Dorbath, C., Bertil, D., Delmond, J.C., 2007. Crustal structure and magmato-tectonic processes in an active rift (AsalGhoubbet, Afar, EastAfrica). 2. Insights from the 23-year recording of seismicity since the last rifting event. Journal of Geophysical Research 112, B05406, doi:10.1029/2006JB004333. EDD, 2009. Electricity Cost Reduction Assessment & Energy Access Sector Strategy. Unpublished report ESMAP 097/06. Energy Sector Management Assistance Program, Djibouti, 158 pp. Earlougher, R.C., Jr., 1977. Advances in well test analysis. Society of Petroleum Engineers Monograph Series 5, Dallas, TX, USA, 264 pp. Grant, M.A., Donaldson, I.G., Bixley, P.F., 1982. Geothermal Reservoir Engineering. Academic Press Ltd., New York, NY, USA, 369 pp. Horne, R.N., 1995. Modern Well Test Analysis—A Computer-Aided Approach, second edition. Petroway, Inc., Palo Alto, CA, USA, 257 pp. Juliusson, E., Gretarsson, G.J., Jonsson, P., 2007. Well Tester 1.0b, User’s Guide. ISOR, Reykjavík, Iceland, 26 pp. Majer, E.L., Baria, R., Stark, M., Oates, S., Bommer, J., Smith, B., Asanuma, H., 2007. Induced seismicity associated with enhanced geothermal systems. Geothermics 36, 185–222. MIT, 2006. The Future of Geothermal Energy: Impact of Enhanced Geothermal Systems [EGS] on the United States in the 21st Century. Massachusetts Institute of Technology, Cambridge, MA, USA, Available on World Wide Web at: http://www1.eere.energy.gov/geothermal/future geothermal.html. Muffler, L.J.P., Cataldi, R., 1978. Methods for regional assessment of geothermal resources. Geothermics 7, 53–89. Muffler, L.J.P., Christiansen, R.L., 1978. Geothermal resource assessment of the United States. Pure and Applied Geophysics 117, 160–171. O’Sullivan, M.J., Pruess, K., Lippmann, M.J., 2001. State of the art of geothermal reservoir simulation. Geothermics 30, 395–429. Thain, I.A., White, B.R., 1993. Wairakei power development efficiency studies. In: Proceedings of the 15th New Zealand Geothermal Workshop, Auckland University, Auckland, New Zealand, pp. 41–45. Virkir-Orkint, 1990. Djibouti. Geothermal Scaling and Corrosion Study. Report. Virkir-Orkint, Reykjavik, Iceland, 109 pp. Vosteen, H.D., Schellschmidt, R., 2003. Physics and Chemistry of the Earth 28, 499–509.