Predictive Model for Sustainable Exploitation of Geothermal Resources in Africa: The Case of Olkaria Geothermal Field

1Predictive model for sustainable exploitation of geothermal resources 2in Africa: The case of Olkaria geothermal field 3 Zuffi1, C., Fiaschi1, D., Musonye2, X.; Mukhongo2, H.S.; Nafula3 M., Da Silva3 I.P. 41 Department of Industrial Engineering, University of Florence, Italy 52 Kenya Electricity Generating Company, Pension Plaza-Ngara, Nairobi P.O. Box 47936, Kenya 63 Strathmore University, Nairobi, Kenya 7Abstract 8 Geothermal energy represents a crucial resource for a sustainable future, particularly in Africa’s Rift 9 Valley, which holds significant untapped potential. Despite this, high upfront costs and development 10 risks remain major barriers. This study introduces a simplified model calibrated with real data from 11 the Olkaria geothermal field in Kenya. The model enables rapid preliminary assessments of both 12 technical and economic performance, requiring only minimal input data. Additionally, it incorporates 13 a Life Cycle Assessment to evaluate environmental impacts, an aspect rarely explored in African 14 geothermal studies. The research analyses various technological configurations, including Single 15 Flash, Double Flash, and Organic Rankine Cycle systems, aiming to improve efficiency without 16 additional drilling. Findings show that integrating a binary ORC with existing flash systems can boost 17 energy output by up to 20.1%, with only a modest rise in the Levelized Cost of Electricity . Compared 18 to the current Olkaria IV setup, hybrid systems demonstrated lower carbon emissions and reduced 19 material resource use.The results confirm that ORC integration offers the most sustainable pathway 20 for developing high-temperature geothermal resources in the East African Rift. This approach 21 balances energy efficiency, economic feasibility, and environmental impact, providing valuable 22 guidance for future power plant development in regulatory-constrained settings. 23 24 25 26 27 28 29 30 31 32 33 34 35 Keywords: Geothermal energy, Kenya, African development, Renewable Energy, Sustainability 36 power plant 37 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 38 1 Introduction 39 Geothermal energy is considered one of the prospective alternatives for displacing baseload power 40 supplied by fossil fuels to partly offset energy-related greenhouse gas emissions [Sharmin et al., 41 2023]. The African continent, where energy demand is projected to increase tremendously in the next 42 30 years, has vast geothermal resources, mainly located along the African Rift System [Omenda, 43 2015]. However, the high risk associated with the upstream development and the attendant high costs 44 still prove a challenge to tapping into these resources [Shortall et al., 2015]. So far, out of the 15 GW 45 potential in the African Rift System, about 1.02 GW has been developed, with 95% of this developed 46 in Kenya [Omenda et al., 2025]. Surface exploration and exploration drilling to prove geothermal 47 resources are expensive and attract less interest from financiers. Thus, efforts are being put in place 48 by the global geothermal community to develop tools to help conduct preliminary exploration 49 assessments remotely to assess the techno-economic feasibility and boost investor confidence in 50 geothermal projects. 51 Previously, researchers have developed software tools for evaluating the techno-economic feasibility 52 of geothermal projects, considering factors such as resource assessment, plant design, financial 53 modelling, and market dynamics. GETEM is an Excel-based tool used to estimate the Levelized Cost 54 of Electricity (LCOE) for geothermal projects, including hydrothermal and Enhanced Geothermal 55 System (EGS). It allows users to input technical and economic parameters and uses a discounted cash 56 flow method to calculate costs and assess different scenarios [Mine, 2016]. GEOPHIRES is a flexible 57 and user-friendly tool for evaluating geothermal systems’ technical and economic performance, 58 including electricity generation, direct heat use, and combined heat and power (CHP). It simulates all 59 major system components—reservoir, wells, surface plant—and calculates key metrics like LCOE 60 and LCOH using user inputs or default correlations [Beckers and McCabe, 2019]. FGEM is designed 61 to assess flexible geothermal operations' technical and economic feasibility by modeling the 62 interaction between the subsurface, surface plant, and energy markets. It supports the evaluation of 63 different dispatch strategies, including mass flow variation, wellhead pressure modulation, and 64 integration with thermal energy storage (TES) and lithium-ion batteries (LiBs) [Aljubran and Horne, 65 2024]. 66 While these tools provide detailed and thorough analyses, they are mainly useful in later stages after 67 a data collection campaign, because they need a large amount of input data. This includes information 68 about the subsurface, the well, the power plant, and sometimes economic or environmental factors. 69 Collecting this data can be difficult, particularly in areas with limited data availability as witnessed 70 with most of the geothermal areas in Africa. Thus, the question remains whether it is possible to 71 predict the techno-economic performance of a plant with limited available data. This approach could 72 be valuable for assessing geothermal potential on a large scale. Moreover, a significant gap in 73 literature is the inability of these methods to evaluate the potential cascade exploitation of the 74 resource, an issue that remains insufficiently addressed. 75 Another important aspect is including environmental factors to provide a more complete and holistic 76 analysis. In the literature, the life cycle assessment (LCA) methodology is commonly employed to 77 evaluate the environmental impacts of geothermal plants [Parisi et al., 2020; GECO, 2023; Zuffi et 78 al., 2020]. However, despite its widespread use, analyses in the African context remain conspicuously 79 absent [Mukoro et al., 2021]. 80 Therefore, as part of the collaborations established within the European project LEAP-RE [LEAP81 RE 2020], a simplified model has been developed to assess the feasibility of geothermal resource 82 utilization for electricity generation using a minimal set of input parameters. The model is structured 83 in a modular framework, allowing for the evaluation of different technologies individually, as well 84 as their potential integration. This enables a cascading utilization approach, where higher-temperature This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 85 resources are used first, followed by lower-temperature ones, or the enhancement of existing 86 geothermal plants. The model is simulated using data from Kenya’s Olkaria geothermal field, which 87 is in the axial of the central Kenyan Rift. Currently, the Olkaria geothermal field is the second most 88 productive worldwide, surpassed only by the geysers in the USA [Renkens, 2019]. To support the 89 analysis a data collection effort was conducted to perform the first LCA of the a geothermal plant in 90 Africa, addressing a gap in the existing literature [Mukoro et al., 2021]. Finally, the results of the 91 techno-economic and environmental assessments are integrated to provide a decision-making 92 framework for evaluating the sustainability of the proposed solutions. 93 2 Methodology 94 2.1 General Approach 95 To characterize a geothermal resource, several parameters are required [Barbier, 2002; Reed and 96 Anderson, 2011; Di Pippo, 2008]. However, four key parameters are selected in the context of 97 developing a simplified predictive techno-economic model. These parameters are fundamental for 98 calculations and are as follows: 99 Reservoir Temperature (Tgeo,in). It defines the possible applications (electricity 100 production/heat/cold) and also the potential of the selected technology; 101 Mass flow rate (mgeo). It defines the capacity of the energy systems and their components, 102 such as heat exchangers, turbines, and other mechanical components; 103 Well Depth (Lwell). It defines the depth of the well and thus results in a possible change in 104 temperature at the wellhead and thus at the input to an energy system; 105 Ambient air temperature (Tair). It defines the average temperature of the environment and the 106 related interactions at the condenser and cooling tower of power plants; 107 The predictive model, presented in this work, is divided into two main sections: the production section 108 and the surface section. The production section represents a one-dimensional model of a geothermal 109 well, used to estimate the fluid conditions at the wellhead. The surface section describes the 110 technologies applicable for geothermal energy conversion. This section includes two energy systems: 111 the Flash power plant and the Binary ORC power plant. Each technology is a simplified model, and 112 it is structured in modular blocks, which can be connected in series or in parallel, allowing for the 113 simulation of different plant configurations. 114 The classification of geothermal resources often relies on the Lindal diagram, which outlines the 115 temperature ranges suitable for various applications [Soelaiman, 2016; Gupta and Roy, 2007]. Each 116 thermodynamic block is selected based on a specific temperature classification, ensuring that the 117 energy system model aligns with the thermal characteristics of the geothermal resource, as showed in 118 Figure 1. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 119 120 Figure 1: Geothermal predictive model concept 121 To develop an accurate and fast predictive model, meta-models have been designed. Meta-models 122 are mathematical tools that simplify the evaluation of input-output relationships in complex systems 123 [Palar et al., 2019]. To construct this meta-model, input parameter ranges were selected for each 124 thermodynamic block, showed in Supplementary Material A (SM.A). A homogeneous distribution 125 was created for each parameter, organized into three-dimensional square matrices. The resulting 126 matrix has dimensions of np x np x np. For each set of input matrix values, the three required parameters 127 were selected and evaluated in the simplified models, optimizing the installed power. The obtained 128 results were organized into output matrices and subsequently interpolated using linear 129 multidimensional interpolation [Chan et al., 1997] to generate the meta-model for all output 130 parameters. The value of dimensions of input matrix np is carefully chosen to create a dense input 131 grid as input and set to have a predictive error between the model and the metamodel of a maximum 132 of 0.5 %. 133 The following is a detailed description of each thermodynamic block, and the relative simplified 134 model. 135 2.2 Thermodynamic Blocks 136 2.2.1 Production section – Geothermal well 137 Ther first thermodynamic block represent the geothermal well. It describes the upward flow of 138 geothermal fluid from the deep reservoir to the surface, considering pressure losses and corresponding 139 temperature changes. It represents a simplified version of the Boiling Point Depth (BPD) Model [Di 140 Pippo, 2008; Grant et al., 2011], assuming one-dimensional, vertical, and adiabatic flow within the 141 well. Initial condition: The geothermal fluid starts as a subcooled liquid at the bottom of the well. As 142 the fluid rises, pressure decreases due to both gravity and dynamic flow effects. When the pressure 143 drops to the saturation pressure corresponding to the fluid's temperature, flashing occurs, part of the 144 liquid turns into vapor, and a two-phase flow begins. The depth at which flashing starts is called the 145 boiling depth (E.1). Depending on the relationship between Lf and the total well depth (Lwell): 𝐿 𝑓 = 𝑃 1 ― 𝑃 𝑠𝑎𝑡 ( 𝑇 𝑔𝑒𝑜 ) 𝜌𝑔 + 𝐶 2 𝑚 2 (E.1) This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 146 where P1 is the pressure at the bottom of the well, Psat is the saturation pressure, ρ is the fluid density, 147 g is the gravitational acceleration, and C2 is a coefficient dependent on fluid properties and well 148 geometry. 149 If Lf ≥ Lwell, the fluid reaches the surface as a liquid—natural flow occurs with minimal energy 150 loss. 151 If Lf < Lwell, the fluid rises in two-phase flow, which introduces additional pressure losses due 152 to hydrostatic pressure, friction, and phase slip (relative velocity between phases). 153 The SM.B shows and describes in more detail all the fundamental equations and parameters 154 2.2.2 Surface section – flash and binary power plant 155 The Flash system model is divided into two main blocks, representing the two fundamental phases of 156 the thermodynamic cycle in a geothermal power plant: 157 Steam production and separation: This block covers the initial part of the cycle, from the 158 wellhead to the turbine. The geothermal fluid passes through an expansion valve and is then 159 separated into a vapor phase and a liquid phase. The separator regulates the pressure to ensure 160 the proper operation of the turbine. The amount of separated steam is managed to maximize 161 energy production without dropping below the minimum pressure required by the condenser. 162 Within the separator, pressure must remain between the pressure upstream of the valve and 163 the condenser pressure. To ensure this condition, a control parameter called xrel is introduced 164 (E.2). It represents the ratio between the actual vapor fraction and the maximum allowable 165 vapor fraction. This parameter helps identify the operating conditions that maximize turbine 166 output without compromising system pressure balance. 𝑥 𝑟𝑒𝑙 = 𝑥 𝑠𝑒𝑝 𝑥 𝑚𝑎𝑥 = 𝑥 𝑠𝑒𝑝 𝑥(ℎ 1 ′ ; 𝑃 𝑐𝑜𝑛𝑑 ) (E.2) 167 Cooling and recirculation: After passing through the turbine, the steam is condensed by a 168 cooling system consisting of a condenser, two pumps, and a cooling tower. Part of the 169 cooled water is reinjected into the system, while the remaining part is used as cooling fluid 170 in the condenser. The system ensures that operating temperatures remain within safe limits, 171 adapting to the external ambient air temperature. 172 A key feature of the model is its modular structure, which allows the output from the first block to 173 be used as input for additional thermodynamic cycles, such as in double flash systems. 174 175 The ORC binary thermodynamic block includes a heat exchanger (MHE), turbine, condenser, and 176 pump, with an additional fictitious separator used only for metamodel development. Unlike flash 177 systems, the ORC uses a secondary working fluid, allowing flexibility in fluid choice based on 178 thermodynamic and environmental performance. Selected fluids (R1233zd(E), R1234ze(Z), 179 isopentane) were chosen for their subcritical behavior and low Global Warming Potential (GWP). 180 The model determines optimal working fluid pressure and mass flow through an iterative process. 181 Heat exchange is governed by the geothermal temperature difference (ΔTgeo), adjusted to prevent 182 supercritical conditions. If the working fluid becomes superheated, it all enters the turbine; if not, 183 only the vapor portion does. The entire cycle is solved respecting mass and energy balances, ensuring 184 realistic sizing and compatibility with market-available heat exchangers. 185 The extended description and the equations of flash and binary blocks is reported in the SM.C. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 186 2.3 Economic modeling 187 The economic evaluation of the system is carried out in two stages. The first stage involves 188 performing a cost analysis of the individual plant components, utilizing thermo-economic correlations 189 as those proposed in Turton (2008) [Turton, 2008]. Key parameters necessary for these correlations, 190 such as the surface area of the condenser and evaporator, the flow rates managed by the compressor, 191 and the volume of the separator, are derived from the thermodynamic blocks. For the separator, the 192 cost estimation employs the thermo-economic correlation provided by Mosaffa et al. (2017). The 193 specific components and the corresponding equations used for each are detailed in SM.D. Economic 194 estimations for geothermal wells typically rely on linear, exponential, or logarithmic correlations. In 195 this study, we adopted the correlation tailored for African applications as proposed by Shamoushaki 196 et al. (2021). The cost of the well can greatly affect the total cost of the plant, both in terms of initial 197 investment and LCOE [Shamoushaki et al., 2021, Karimi and Mansouri, 2018], and therefore special 198 attention is put on the number of well and their depth. The number of wells in the analysis depends 199 on the specific case. If known, it can be directly input into the economic model. Otherwise, two 200 options are available: assume one production and one reinjection well, or estimate the well count 201 using an empirical correlation based on plant capacity, as derived from literature data see equation 202 and reference in the SM.D. 203 With the economic model described above, the initial investment costs and total operation and 204 maintenance (O&M) costs are evaluated to determine the Levelized Cost of Electricity (LCOE). The 205 LCOE represents the full cycle costs (both fixed and variable) of a generation technology per unit of 206 energy, enabling comparisons across different technologies irrespective of their size, cost structure, 207 or lifespan [Ueckerdt et al., 2013]. In the context of renewable energy systems, the value of produced 208 energy is influenced by the resource availability and by the intermittent nature of the energy source. 209 Consequently, the LCOE is intrinsically linked to the variability patterns that define the energy 210 generation profile [Tran and Smith, 2018]. This parameter is particularly important in this study, as 211 it serves as a significant economic indicator [de Simon-Martin et al., 2022]. By consolidating these 212 costs into a single economic indicator, LCOE effectively quantifies the cost of electricity unitary 213 production cost ($/kWh), as discussed by de Simón-Martín et al. (2022). 𝐿𝐶𝑂𝐸 = 𝐶 𝑇𝐶𝐼 + 𝐶 𝑊𝐷 + ∑ 𝑛 𝑖 = 1 𝐶 𝑝𝑟 (1 + 𝑟) 𝑖 ∑ 𝑛 𝑖 = 1 𝐸 𝑖 (1 + 𝑟) 𝑖 (E.3) 214 The equation’s parameters in E.3 are the following: CTCI is the total capital investment cost; Cpr for 215 power plant is the Total Production Cost (CTPC) [Shamoushaki et al., 2022] for other application is 216 the annual operating and maintenance cost (CO&M) [de Simon-Martin et al., 2022]; Ei is the annual 217 electricity produced; r is the discount rate; n is the lifetime of the energy system. For the calculation 218 of CTCI, the method and parameters used in this work were taken from Karimi & Mansouri, (2018), 219 Shamoushaki et al., (2022). For the evaluation of LCOE with the integration of geothermal wells 220 Hackstein & Madlener (2021) have been adopted. While for the evaluation of CTPC, Shamoushaki et 221 al. (2022) was followed. For CO&M, literature reference values were taken [Beckers et al., 2021]. The 222 discount rate was set for each application at 7% ([Beckers et al., 2021] and the lifetime varies between 223 power plant and other application is fixed to 30 years [Shamoushaki et al.,2021; Beckers et al., 2021; 224 Geoenvi 2023]. All correlations used and the general expression of the thermo-economic model are 225 shown in Table 1. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 226 Table 1: Economic correlations for Thermo-economic model Cost type Relation Total Cycle Cost 𝑇𝐶 𝐵 = 𝑖 ― 𝑡ℎ 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝑖 = 1𝑠𝑡 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝐶 𝑖 Cost of site preparation Cost of service facilities Allocated cost 𝐶 𝑠𝑖𝑡𝑒 = 0.05 ∗ 𝑇𝐶 𝐵 𝐶 𝑠𝑒𝑟𝑣 = 0.05 ∗ 𝑇𝐶 𝐵 𝐶 𝑎𝑙𝑙𝑜𝑐 = 0 Total Direct Permanent Investment Cost of contingencies and contractor’s fee 𝐶 𝐷𝑃𝐼 = 𝑇𝐶 𝐵 + 𝐶 𝑠𝑖𝑡𝑒 + 𝐶 𝑠𝑒𝑟𝑣 + 𝐶 𝑎𝑙𝑙𝑜𝑐 𝐶 𝑐𝑜𝑛𝑡 = 0.18 ∗ 𝐶 𝐷𝑃𝐼 Total depreciable capital Cost of plant startup 𝐶 𝑇𝐷𝐶 = 𝐶 𝐷𝑃𝐼 + 𝐶 𝑐𝑜𝑛𝑡 𝐶 𝑠𝑡𝑎𝑟𝑡𝑢𝑝 = 0.1 ∗ 𝐶 𝑇𝐷𝐶 Permanent investment Working capital 𝐶 𝑇𝑃𝐼 = 𝐶 𝑇𝐷𝐶 + 𝐶 𝑠𝑡𝑎𝑟𝑡𝑢𝑝 𝐶 𝑊𝐶 = 0 Total Capital Investment 𝐶 𝑇𝐶𝐼 = 𝐶 𝑇𝑃𝐼 + 𝐶 𝑊𝐶 Cost of wages and benefits Cost of salaries and benefits Cost of materials and services Cost of maintenance overhead 𝐶 𝑊𝐵 = 0.035 ∗ 𝐶 𝑇𝐷𝐶 𝐶 𝑆𝐵 = 0.035 ∗ 𝐶 𝑇𝐷𝐶 𝐶 𝑇𝑃𝐼 = 𝐶 𝑊𝐵 𝐶 𝑀𝑂 = 0.05 ∗ 𝐶 𝑊𝐵 Direct Manufacturing costs Fixed Manufacturing Cost 𝐶 𝐷𝑀𝐶 = 𝐶 𝑀𝑂 + 𝐶 𝑇𝑃𝐼 + 𝐶 𝑆𝐵 + 𝐶 𝑊𝐵 𝐶 𝐹𝐼𝑋 = 0.02 ∗ 𝐶 𝑇𝐷𝐶 Total annual cost of manufacture 𝐶 𝐶𝑂𝑀 = 𝐶 𝐷𝑀𝐶 + 𝐶 𝐹𝐼𝑋 General Expense 𝐶 𝐺𝐸 = 0 Total production cost 𝐶 𝑝𝑟 = 𝐶 𝐶𝑂𝑀 + 𝐶 𝐺𝐸 227 228 2.4 Case studies 229 2.4.1 Olkaria Geothermal Field 230 The case study to which the model developed in this work is applied in the Olkaria geothermal field, 231 focusing on data from wells that feed steam to Olkaria IV. The power plant is a single flash with an 232 installed capacity of 150 MWe, and it has two turbines, each with a rated capacity of 75 MWe. The 233 net output capacity is 140 MWe and exploits a liquid-dominated resource hosted in Trachyte 234 formation between 1.2 to 3 km below the surface [Musonye, 2015]. The plant is served by 21 wells: 235 18 production wells and 3 re-injections wells. Each turbine is fed by a steam line with an intake flow 236 rate of 135kg/s, 180°C temperature, and pressure between 9-12 bar. Several scenarios were analyzed, 237 which are briefly described below and discussed in more detail in the following section: 238 1) Scenario Base case: This scenario represents the case study of Olkaria IV, thus using the data 239 regarding the wells, a Single Flash system was calculated 240 2) Scenario Double Flash (FII): In this scenario, another flash is added to the results obtained for 241 a single flash by connecting two thermodynamic blocks 242 3) Scenario Triple Flash (FIII): In this scenario, an additional thermodynamic block is added to 243 the FII case to simulate a third level of power generation 244 4) Single Flash and Binary cycle (FI - ORC) scenario: here an ORC-type thermodynamic block 245 is added to the basecase scenario 246 5) Double Flash and Binary cycle (FI - ORC) scenario: Here, however, a thermodynamic block 247 is added to the FII scenario ORC This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 248 2.4.2 Life Cycle Assessment 249 Data collection was carried out to calculate the environmental impact of a sustainability analysis. The 250 approach followed was Life Cycle Assessment (LCA), per the framework established by ISO 14040 251 and ISO 14044. The objective is to determine the environmental impact of Olkaria and identify the 252 most significant environmental indicators and compare it to other systems. The approach used for this 253 analysis is cradle-to-gate type. The system boundaries encompass all processes related to the 254 construction, operation, and maintenance phases. The end-of-life phase is not considered due to the 255 lack of a disposal program. The functional unit selected is the electrical energy produced by the 256 powerplant in kWh, considering a useful 30-year lifetime and a capacity factor of 0.94, corresponding 257 to 8234 hours/year of operation. 258 2.5 Environmental model 259 The processes for the construction phase which were modeled in this settlement are the geothermal 260 wells and wellheads; the building hosting the powerplant, and the collection pipelines. The average 261 geothermal wells depth is 3000 m each. The collection pipelines are divided into 28 km of production 262 and 5 km of re-injection pipe. The machinery and power plant have not been modelled with primary 263 data from KenGen because of lack of data. Consequently, machinery and plant data were adopted 264 from the model proposed by Karlsdottir et al. (2015). For the scenario analysis involving the ORC 265 plant, the reference inventory for the machinery was taken from Frick et al., 2010. 266 The operation phase consumes 40.4 m3 of water per day for plant operation and approximately 126 267 MWh of electricity for heat distribution. Of this electricity, 16% is taken from the national grid and 268 84% from plant production. Direct emissions to the atmosphere are mainly CO2, H2S, H2 and CH4. 269 Six make-up wells are expected to be built in 25 years to maintain the geothermal steam flow rate at 270 the required level. In addition, lubricant oil is consumed for the machinery, while Sodium Carbonate 271 is used for anticorrosion and anti-scaling of the pipes. All inventory data are reported in 272 Supplementary Materials. The results shown in this paper refer to the single score. The contribution 273 analysis of the main indicator analyzed is reported in the SM.F. 274 3 Results 275 3.1 Validation techno-economic 276 3.2 Results metamodels 277 The thermodynamic model is validated by comparing several power plants analyzed in the literature, 278 and this is reported in the Supplementary Materials (SM.G). The general results of the metamodels 279 are reported in the following paragraph. 280 3.2.1 Thermodynamic metamodels – Power Production 281 Figure 2 compares three geothermal technologies, by analyzing installed power as a function of flow 282 rate and geothermal fluid temperature, with an external temperature fixed at 30°C. The meta-models 283 estimate power output, highlighting significant differences in operational temperature ranges. The 284 Single Flash system achieves high power output at geothermal temperatures above 250°C, where the 285 Binary ORC is ineffective. The Double Flash outperforms the Single Flash, particularly between 286 100°C and 150°C, increasing power by 6-7%, and above 200°C, where gains exceed 20%. However, 287 in high ambient temperatures, rising condensation temperatures reduce efficiency, limiting power 288 gains to 0.5-1%, making the system less viable. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 289 Comparing Figure 2A and Figure 2B, the Double Flash expands the installed power area beyond 50 290 MW and shows steeper power growth with increasing flow rate (mgeo) and geothermal temperature 291 (Tgeo). The Binary ORC, within its operational range, enables slightly larger plants than the flash 292 systems at the same geothermal conditions. For example, at 200°C and 400 kg/s, the ORC reaches 293 20.9 MW, Single Flash 20.1 MW, and Double Flash 23.6 MW. For lower flow rates (<200 kg/s) at 294 maximum Tgeo, installed capacities remain moderate: Single Flash (~37 MW), Double Flash (~46 295 MW, +24%), ORC (~12 MW). 296 A key distinction lies in operational temperature ranges: ORC functions between 80-200°C, while 297 flash systems extend from 100°C to >350°C, allowing greater energy exploitation at high 298 temperatures. While power output is similar when operating at the same temperature, the flash 299 systems leverage higher temperature resources for greater efficiency. Prediction uncertainty varies, 300 being lowest for Single Flash, higher for ORC, and highest for Double Flash. These findings clarify 301 each technology’s capacity and optimal conditions based on flow rate and temperature parameters. 302 303 304 305 Figure 2: Metamodel surface representing the power installed for Single (A), Double Flash (B) and 306 ORC Binary (C) 307 3.2.2 Economic metamodels – Power Production 308 This section presents the results of the economic metamodels. Validating the model requires a similar 309 approach to the thermodynamic model but with actual LCOE values for specific cases, which are 310 rarely provided alongside necessary thermodynamic and economic parameters in literature. Thus, the 311 model’s predictions can only be compared to known reference values. 312 A major contributor to total plant cost and LCOE is the cost of wells, primarily influenced by their 313 number and depth. To assess this impact, four scenarios with well depths of 1000 m, 2000 m, 3000 314 m, and 4000 m were analyzed. Figure 3 illustrates the LCOE (c$/kWh) as a function of geothermal 315 resource flow rate (1-80 kg/s) and temperature, with reference values for photovoltaic (8.7 c$/kWh, C B A This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 499 Musonye, X. (2015). Sub-surface petrochemistry, stratigraphy and hydrothermal alteration of the domes area. 500 Olkaria geothermal field, Kenya.(MSc). 501 Omenda, P., & Simiyu, S. (2015). Country update report for Kenya 2010-2014. In Proceedings World 502 Geothermal Congress (pp. 19-25). 503 Omenda, P., Ofwona, C., & Mangi, P. (2025). Kenya: The Most Successful Geothermal Development in 504 Africa. In Geothermal Power Generation (pp. 863-891). Elsevier Science Ltd. https://doi.org/10.1016/B978505 0-443-24750-7.00022-1 506 Palar, P. S., Liem, R. P., Zuhal, L. R., & Shimoyama, K. (2019, July). On the use of surrogate models in 507 engineering design optimization and exploration: The key issues. In Proceedings of the Genetic and 508 Evolutionary Computation Conference Companion (pp. 1592-1602). 509 https://doi.org/10.1145/3319619.3326813 510 Parisi, M. L., Douziech, M., Tosti, L., Pérez-López, P., Mendecka, B., Ulgiati, S., ... & Blanc, I. (2020). 511 Definition of LCA guidelines in the geothermal sector to enhance result comparability. Energies, 13(14), 3534. 512 https://doi.org/10.3390/en13143534 513 R. Di Pippo, 2008. Geothermal power plants. Elsevier. https://doi.org/10.1016/B978-0-7506-8620-4.X5001-1 514 Renkens, I. (2019). The impact of renewable energy projects on indigenous communities in Kenya. IWGIA: 515 Denmark, December. 516 Shamoushaki, M., Fiaschi, D., Manfrida, G., & Talluri, L. (2022). Energy, exergy, economic and 517 environmental (4E) analyses of a geothermal power plant with NCGs reinjection. Energy, 244, 122678. 518 https://doi.org/10.1016/j.energy.2021.122678 519 Shamoushaki, M., Fiaschi, D., Manfrida, G., Niknam, P. H., & Talluri, L. (2021). Feasibility study and 520 economic analysis of geothermal well drilling. International Journal of Environmental Studies, 78(6), 1022521 1036. https://doi.org/10.1080/00207233.2021.1905309 522 Sharmin, T., Khan, N. R., Akram, M. S., & Ehsan, M. M. (2023). A state-of-the-art review on 523 geothermal energy extraction, utilization, and improvement strategies: conventional, hybridized, 524 and enhanced geothermal systems. International Journal of Thermofluids, 18, 100323. 525 https://doi.org/10.1016/j.ijft.2023.100323 526 Shortall, R., Davidsdottir, B., & Axelsson, G. (2015). Geothermal energy for sustainable 527 development: A review of sustainability impacts and assessment frameworks. Renewable and 528 sustainable energy reviews, 44, 391-406. https://doi.org/10.1016/j.rser.2014.12.020 529 Sigfússon, B., & Uihlein, A. (2015). 2014 JRC geothermal energy status report. 530 Soelaiman, T. A. F. (2016). Geothermal energy. Electric Renewable Energy Systems, 114-139. 531 Tran, T. T., & Smith, A. D. (2018). Incorporating performance-based global sensitivity and uncertainty 532 analysis into LCOE calculations for emerging renewable energy technologies. Applied energy, 216, 157-171. 533 https://doi.org/10.1016/j.apenergy.2018.02.024 534 Trinomics, B. V. (2020). Final Report Cost of Energy (LCOE). Energy costs, taxes and the impact of 535 government interventions https://energy.ec.europa.eu/system/files/2020 536 10/final_report_levelised_costs_0.pdf 537 Turton, R., Bailie, R. C., Whiting, W. B., & Shaeiwitz, J. A. (2008). Analysis, synthesis and design of chemical 538 processes. Pearson Education. 539 Ueckerdt, F., Hirth, L., Luderer, G., & Edenhofer, O. (2013). System LCOE: What are the costs of variable 540 renewables?. Energy, 63, 61-75. https://doi.org/10.1016/j.energy.2013.10.072 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed 541 Zuffi, C., Manfrida, G., Asdrubali, F., & Talluri, L. (2022). Life cycle assessment of geothermal power plants: 542 A comparison with other energy conversion technologies. Geothermics, 104, 102434. 543 https://doi.org/10.1016/j.geothermics.2022.102434 544 Zuffi, C., Ungar, P., Fiaschi, D., Manfrida, G., & Batini, F. (2024). Qualtra Geothermal Power Plant: Life 545 Cycle, Exergo-Economic, and Exergo-Environmental Preliminary Assessment. Sustainability, 16(11), 4622. 546 https://doi.org/10.3390/su16114622 547 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=5334709 Preprint not peer reviewed