The global and national energy systems techno-economic (GNESTE) database: Cost and performance data for electricity generation and storage technologies

Power sector and energy systems models are widely used to explore the impacts of demographic, socio-economic or policy changes on the cost and emissions of electricity generation. Technology cost and performance data are essential inputs to such models. Despite the ubiquity and importance of these parameters, there is no standardised database which collates the variety of values from across the literature, so modellers must collect them independently each time they populate or update model inputs, leading to duplicated efforts and inconsistencies which can profoundly influence model results. Technology cost and performance varies between countries, regions and over time, meaning that data must be country- or region-specific and frequently updated. Values also vary widely between sources, so obtaining a broad consensus view is critical. Here, we present a database which collates historical, current, and future cost and performance data and assumptions for the six most prominent electricity generation technologies; coal, gas, hydroelectric, nuclear, solar photovoltaic (PV) and wind power, which together accounted for over 92 % of installed generation capacity in 2022. In addition, we provide the same data for utility-scale battery energy storage systems (BESS), regarded as critical to the integration of variable renewables such as wind and solar PV. The data are global in scope but with regional and national specificity, covers the years 2015 through to 2050, and span 5518 datapoints from 56 sources. The database enables modellers to select and justify model input data and provides a benchmark for comparing assumptions and projections to other sources across the literature to validate model inputs and outputs. It is designed to be easily updated with new sources of data, ensuring its utility, comprehensiveness, and broad applicability in future.


a b s t r a c t
Power sector and energy systems models are widely used to explore the impacts of demographic, socio-economic or policy changes on the cost and emissions of electricity generation.Technology cost and performance data are essential inputs to such models.Despite the ubiquity and importance of these parameters, there is no standardised database which collates the variety of values from across the literature, so modellers must collect them independently each time they populate or update model inputs, leading to duplicated effort s and inconsistencies which can profoundly influence model results.Technology cost and performance varies between countries, regions and over time, meaning that data must be country-or region-specific and frequently updated.Values also vary widely between sources, so obtaining a broad consensus view is critical.Here, we present a database which collates historical, current, and future cost and performance data and assumptions for the six most prominent

Value of the Data
• The database is both comprehensive and open-source.It can be used to select and justify model inputs helping to overcome issues with data inaccessibility which are a considerable barrier to developing and calibrating energy and power systems models, particularly in developing nations.• The spatial and temporal coverage and country-level breakdown make the database applicable to a wide range of models covering different geographic regions and time horizons.• All recordings use a consistent structure, units, and currencies allowing different sources to be compared quickly, and provide a means of validating both model inputs and outputs.• The parameters recorded in the database are of broad utility across many types of model and are therefore in high demand among the energy modelling community.• Example applications include calculating levelized costs of electricity generation, finding costeffective decarbonisation pathways, and optimising power sector investment and operation.

Background
Global temperatures are rising which is having an unprecedented impact on the global energy system and human society [ 1 ].Accelerating global decarbonisation effort s is essential if the world is to limit further warming [ 2 ].Power sector and energy systems models are widely used to explore the impact of demographic, socio-economic and policy changes on the cost and emissions of electricity generation.Technology cost and performance data are essential inputs to these models.Cost and performance vary by region and over time, meaning that data must be region-specific and constantly updated, and are typically represented using average values based on available data, so they also vary by source.Despite their ubiquity and importance, there is no standardised database which collates technology cost and performance estimates from across the literature, so modellers must collect them independently each time they populate or update model inputs, leading to duplicated efforts and inconsistencies between studies.There are many influential sources which cover multiple technologies from the IEA [ 3 , 4 ], IRENA [ 5 ], EIA [ 6 ], NREL [ 7 ], CSIRO [ 8 ], Danish Energy Agency [ 9 ], DESNZ [ 10 ], and Lazard [ 11 ], among others.However, these exist in a variety of formats, using different units, customs and currencies, which we have harmonised in the GNESTE database.
Here, we present an open-access database of cost and performance data from the open literature covering seven key power generation and storage technologies: • Battery energy storage systems (BESS): which are the fastest growing form of power system flexibility and will be critical to integrating large shares of variable renewable energy [ 13 ].
The database was assembled for use with the OSeMOSYS framework [ 14 ], but it is equally applicable to other energy and power system models.It aims to provide an accessible, useable and extendable resource for modellers, policymakers, and other stakeholders worldwide.This database streamlines the process of model setup and calibration, reducing the need to duplicate effort s when comparing or validating model inputs and outputs.

Definition of units
The database covers nine parameters needed to model the generating costs of electricity generation and storage which are defined in Table 1 .All cost parameters are given in 2023 US Dollars.
For BESS and hydroelectric power (specifically pumped hydro storage), the efficiency variable instead refers to the round-trip efficiency of charging and discharging, net of plant selfconsumption.Costs for BESS can be measured relative to total energy storage capacity instead of maximum power output, so entries for Capital Cost are measured in both USD/kW and USD/kWh, while Fixed O&M and Total O&M are measured in both USD/kW/yr and USD/kWh/yr depending on which metric was used by each source.It is possible to convert between these using the Energy:Power Ratio of the storage system (kWh/kW, or simply hours), which is given in the database.
The GNESTE database includes historical data from 2015 to 2023 and projections for the years 2024, 2025, 2030, 2040 and 2050.Data were collected from reports, academic articles, webpages and databases of national and international organizations.In adherence to U4RIA1 guidelines, the data are retrievable, reusable, repeatable, reconstructable, interoperable, and auditable.

Definition of technologies
The GNESTE database covers seven technologies which are collectively divided into 33 categories, which are presented in Table 2 .Full definitions of each category can be found in the Metadata of the GNESTE database.

Summary of values for coal power
Table 3 compiles the sources used for each variable and how many values were collated.Figs. 1 and 2 present an excerpt of the values exhibited across the literature for key parameters for the recent period (2020 to 2024), alongside projections for the near (2030) and far fu-Table 3 The number of datapoints collected and sources used for each parameter in the database for coal power.

Parameter
Number  ture (2050).Fig. 1 presents the range of fixed operating and capital costs for different categories of coal-fired power stations, whilst Fig. 2 presents values of capital costs for coal (exc.Lignite) (an aggregate of 'Coal', 'Coal -Supercritical', 'Coal -Ultrasupercritical', and 'Unspecified'), both regional and projected.
The median values for the recent period (2020 to 2024), aggregated across all world regions and all technology sub-types was 2235 USD/kW for capex, 53.5 USD/kW/yr for fixed O&M, 5.3 USD/MWh for variable O&M, 6.4 USD/MWh for fuel price, 39.5 % for efficiency, 5.25 years for construction time, and 40 years for operating life.There were no data for total O&M within these years.

Summary of values for gas power
Table 4 compiles the sources used for each variable and how many values were collated.Figs. 3 and 4 present an excerpt of the values exhibited across the literature for key parameters for the recent period (2020 to 2024), alongside projections for the near future (2030) and far future (2050).Fig. 3 presents the range of fixed operating and capital costs for the different categories of gas-fired power stations for 2020-2024 whilst Fig. 4 presents values of capital costs for power plants using closed-cycle gas turbine (the 'CCGT' category), both regional and projected.

Table 4
The number of datapoints collected and sources used for each parameter in the database for gas power.

Parameter
Number The median values for the recent period (2020 to 2024), aggregated across all world regions and all technology sub-types was 1040 USD/kW for capex, 17 USD/kW/yr for fixed O&M, 4.0 USD/MWh for variable O&M, 5.0 USD/MWh for total O&M, 11.8 USD/MWh for fuel price, 45 % for efficiency, 1.5 years for construction time, and 25 years for operating life.

Summary of values for hydroelectric power
Table 5 compiles the sources used for each variable and how many values were collated.

Table 5
The number of datapoints collected and sources used for each parameter in the database for hydroelectric power.

Parameter
Number  Figs. 5 and 6 present an excerpt of the values exhibited across the literature for key parameters for the recent period (2020 to 2024), alongside projections for the near future (2030) and far future (2050).Fig. 5 presents the range of fixed operating and capital costs for the reservoir and run of river (RoR) categories of hydroelectric power stations for 2020 to 2024 whilst Fig. 6 presents values of capital costs for reservoir-based hydroelectric power stations, both regional and projected.
The median values for the recent period (2020 to 2024), aggregated across all world regions and all technology sub-types was 6407 USD/kW for capex, 92.3 USD/kW/yr for fixed O&M, 0.4 USD/MWh for variable O&M, 22.0 USD/MWh for total O&M, 4 years for construction time, and 50 years for operating life.

Summary of values for nuclear power
Table 6 compiles the sources used for each variable and how many values were collated.Fig. 7 and Fig. 8 present an excerpt of the values exhibited across the literature for key parameter in each world region for the recent period (2020 to 2024), alongside projections for the

Table 6
The number of datapoints collected and sources used for each parameter in the database for nuclear power.

Parameter
Number near future (2030) and far future (2050).Fig. 7 presents the range of fixed operating and capital costs for different categories of nuclear power for 2020 to 2024, whilst 8 presents values of capital costs for Generation III technologies (the aggregate of 'PWR', 'LWR', 'PHWR' 'VVER'), both regional and projected.
The median values for the recent period (2020 to 2024), aggregated across all world regions and all technology sub-types was 7350 USD/kW for capex, 131.5 USD/kW/yr for fixed O&M, 3.1 USD/MWh for variable O&M, 31.2USD/MWh for total O&M, 6.90 USD/MWh for fuel price, 38 % for efficiency, 7.4 years for construction time, and 40 years for operating life.

Summary of values for solar PV
Table 7 compiles the sources used for each variable and how many values were collated.Figs. 9 -11 present an excerpt of the values exhibited across the literature for key parameters for the recent period (2020 to 2024), alongside projections for the near future (2030) and far future (2050).

Table 7
The number of datapoints collected and sources used for each parameter in the database for solar PV power.

Parameter
Number   and financing costs for large-scale fixed solar PV (the aggregate of 'Fixed Axis' and 'Unspecified' categories).
The median values for the recent period (2020 to 2024), aggregated across all world regions and all technology sub-types was 975 USD/kW for capex, 21.8 USD/kW/yr for fixed O&M, 13.9 USD/MWh for total O&M, 2 years for construction time, 30 years for operating life and 6 % for cost of capital.There were no data on variable O&M, although this is typically considered to be zero.

Summary of values for wind power
Table 8 compiles the sources used for each variable and how many values were collated.Figs. 12 -14 present an excerpt of the values exhibited across the literature for key parameters for the recent period (2020 to 2024), alongside projections for the near future (2030) and far future (2050).Fig. 12 presents the range of fixed operating and capital costs for onshore and

Table 8
The number of datapoints collected and sources used for each parameter in the database, for wind power.

Parameter
Number offshore wind farms for 2020 to 2024 while Figs. 13 and 14 present values of capital costs and financing costs for onshore wind farms.
The median values for the recent period (2020 to 2024), aggregated across all world regions and all technology sub-types was 1750 USD/kW for capex, 32.6 USD/kW/yr for fixed O&M, 3.6 USD/MWh for variable O&M, 30.9 USD/MWh for total O&M, 3 years for construction time, 25 years for operating life and 5.6 % for cost of capital.

Summary of values for batteries
Table 9 compiles the sources used for each variable and how many values were collated.Figs. 15 and 16 present an excerpt of the values exhibited across the literature for key parameters for the recent period (2020 to 2024), alongside projections for the near future (2030) and far future (2050).Fig. 15 presents the range of fixed operating and capital costs for lithium-ion batteries (an aggregate of the 'Lithium-ion', 'Lithium-ion NMC' and 'Lithium-ion LFP' categories) for 2020 to 2024, whilst Fig. 16 presents values of capital costs for lithium-ion battery energy storage systems, both regional and projected.
The median values for the recent period (2020 to 2024), aggregated across all world regions and all technology sub-types was 402 USD/kWh for capex, 60.8 USD/kW/yr for fixed O&M, 0.6 USD/MWh for variable O&M, 85 % for efficiency, 1 years for construction time, and 15 years for operating life [ 6 , 11 , 54 , 59 ].

Table 9
The number of datapoints collected and sources used for each parameter in the database for BESS.

Parameter
Number

Demonstration of calculating LCOE
To give an example application, the GNESTE database can be used to calculate the Levelized Cost of Electricity (LCOE), a widely-used metric for comparing the economic efficiency of different generating technologies.Adapting the IEA's formula [ 3 ] to use our variable names, LCOE can be calculated via: Where t is the year, and both the numerator and denominator sums run from t = 0 to LIFETIME .The denominator contains the specific energy output (in MWh per kW capacity): where CF is the capacity factor, or utilization of the technology, which is user-defined based on the specific application, location and market.Total operations & maintenance cost (in $/kW/yr) is given by the OPEX_T variable, or can be calculated as: The annual cost of fuel input (in $/kW/yr) is given by: And finally, if relevant, the annual cost of carbon emissions (in $/kW/yr) is given by: Where FUEL_CI is 344.5 kgCO 2 /MWh for coal [ 62 ] or 364.2 kgCO 2 /MWh for lignite [ 63 ], 204.8 kgCO 2 /MWh for natural gas [ 62 ], and zero for hydro, nuclear, solar, and wind.CARBON_PRICE is the cost of emitting a tonne of CO 2 , which is user-defined based on the market and scenario considered.
To give a simple demonstration of how the GNESTE database can be used for technoeconomic calculations, Eqs. ( 6) -( 11) calculate the LCOE of each electricity generating technology in turn.Each calculation is based on the median values from the GNESTE database listed in the previous sections, assuming all capital expense is occurred in a single year (at the start of the project), all other variables remain constant (in real terms) over the project lifetime, and there is no price on carbon emissions ( CARBON_PRICE = 0 USD/tCO 2 ).These examples could be made more accurate by including details such as the build time (meaning energy production does not   (low-case, excluding CCS) [ 11 ], which is notably higher as Lazard estimates capex to be 60 % higher than found here.
For hydroelectric power generation (averaged across all types): using example values of CF = 50 % and WACC = 10 %.For nuclear power (averaged across all types): using example values of CF = 90 % and WACC = 10 %.This compares to Lazard's estimate of 141-221 USD/MWh for new-build nuclear [ 11 ], which is notably higher as Lazard estimates capex to be 15-85 % higher than found here.For fixed-axis solar PV: (10) using an example value of CF = 15 %, and CAPEX = 954 USD/kW (the median of 'Fixed Axis' and 'Unspecified' categories).This compares to Lazard's high-case estimate of 96 USD/MWh for utility-scale solar PV, and low-case estimate of 49 USD/MWh for commercial and industrial solar PV [ 11 ].
For onshore wind: (11) using an example value of CF = 30 %.This compares to Lazard's estimates which range from 24 to 75 USD/MWh for onshore wind [ 11 ].

Demonstration of calculating LCOS
For battery storage, the GNESTE database can instead be used to calculate the Levelized Cost of Storage (LCOS), a widely-used metric for comparing the economic efficiency of different storage technologies.Adapting Schmidt's formula [ 13 ] to use our variable names, LCOS can be calculated via: Where the cost of charging the battery (in $/kW/yr) is given by:

CHARGING = E LE CT RI CI T Y _P RI CE EF F I CI ENCY
• E NE RGY (13) And ELECTRICITY_PRICE is the volume-weighted average price for electricity used to charge the storage system, which is user-defined based on the market and scenario considered.
For lithium-ion energy storage systems with 4-h duration:

Experimental Design, Materials and Methods
The data set was collated by reviewing reports, websites, and datasets from international and national organisations, and peer-reviewed journal papers.The search for additional sources was conducted until we reached a saturation of information.Only primary sources were used, and care was taken to ensure that data points were not duplicated across sources.Sources were prioritised according to the robustness of their methodologies and representativeness of data.For example, surveys of actual installed costs were prioritised over modelled estimates.All data entries were reviewed by each member of the research team for quality assurance.Raw data were converted to the standardised units shown in Table 1 , with all currencies converted to 2023 US Dollars using the source country's GDP deflator as a measure of general inflation to convert into 2023 local currency [ 64 ], then using the 2023 year-average market exchange rate to convert to US Dollars [ 65 ].All currency conversion rates are listed within the metadata that accompanies the database.

Limitations
Data were more widely available for OECD nations, particularly in Europe and North America, and the larger BRICS countries, notably India and China, which are the focus of many assessments.Substantial data gaps exist among primary sources, and thus in our database, for Africa,

Fig. 1 .
Fig. 1.Range of a) capital costs and b) fixed operating costs for coal-fired power stations, by category in 2020 to 2024.The lower and upper bounds of each shaded box represent the lower and upper quartile of the data (25th and 75th percentile respectively), with the central line representing the median.Whiskers represent the range of values falling within 1.5 times the inter-quartile range, and outliers are shown with diamonds.The 25th, 50th and 75th percentile values are written to the left of each bar.

Fig. 2 .
Fig. 2. Range of operating costs for coal-fired power stations in a) 2020 to 2024 for each world region, where available, and b) global projections for 2030 and 2050.

Fig. 3 .Fig. 4 .
Fig. 3. Range of a) capital costs and b) fixed operating costs for gas-fired power stations in 2020 to 2024, by category.

Fig. 5 .
Fig. 5. Range of a) capital costs and b) fixed operating costs for hydroelectric power stations in 2020 to 2024, for reservoir and run-of-river (RoR).
Fig. 9 presents the range of capital costs and fixed operating costs for the different categories of solar farms for 2020 to 2024 while Figs. 10 and 11 present values of capital costs

Fig. 6 .
Fig. 6.Range of operating costs for hydroelectric power stations in a) 2020 to 2024 for each world region, where available, and b) global projections for 2030 and 2050.

Fig. 7 .
Fig. 7. Range of a) capital costs and b) fixed operating costs for nuclear power stations in 2020 to 2024, for each category.

Fig. 8 .
Fig. 8. Range of operating costs for nuclear power stations in a) 2020 to 2024 for each world region, where available, and b) global projections for 2030 and 2050.

Fig. 9 .
Fig. 9. Range of a) capital costs and b) fixed operating costs in 2020 to 2024 for solar PV generation, by category.

Fig. 10 .
Fig. 10.Range of capital costs for solar farms in a) 2020 to 2024 for each world region, where available, and b) global projections for 2030 and 2050.

Fig. 11 .
Fig. 11.Range of financing costs for solar farms in each world region, where available, in 2020 to 2024.

Fig. 12 .
Fig. 12. Range of a) capital costs and b) fixed operating costs, for onshore and offshore wind farms in 2020 to 2024.

Fig. 13 .
Fig. 13.Range of capital costs for onshore wind farms in a) 2020 to 2024 for each world region, where available, and b) global projections for 2030 and 2050.

Fig. 14 .
Fig. 14.Range of financing costs for onshore wind farms in each world region, where available, in 2020 to 2024.

Fig. 15 .
Fig. 15.Range of a) capital costs and b) fixed operating costs in 2020 to 2024, for lithium-ion and vanadium redox-flow battery energy storage systems.There was no data available on fixed operating costs for vanadium redox-flow batteries between 2020 and 2024.

Fig. 16 .
Fig. 16.Range of capital costs for lithium-ion battery energy storage systems in a) 2020 to 2024 for each world region, where available, and b) global projections for 2030 and 2050.

•
Coal and natural gas: which supply 35 % and 23 % of global electricity respectively, but must both be rapidly phased down to meet global decarbonisation objectives [ 12 ].Fossil-fuelled plants are considered both as conventional unabated plants, and equipped with carbon capture and storage (CCS).• Hydroelectric and nuclear power: which are the two largest sources of low-carbon energy, supplying 15 % and 9 % of global electricity respectively [ 12 ].• Solar PV: which supplies 5 % of global electricity, and has grown ten-fold in the decade to 2022 [ 12 ].
• Wind energy: which supplies 7 % of global electricity, and has grown three-fold in the decade to 2022 [ 12 ].

Table 1
Description of the cost and performance parameters collated in the database.

Table 2
Summary of the categories of each technology collated in the database.