Open energy system models
Open energy-system models are energy-system models that are open source.[a] However, some of them may use third-party proprietary software as part of their workflows to input, process, or output data. Preferably, these models use open data, which facilitates open science. Energy-system models are used to explore future energy systems and are often applied to questions involving energy and climate policy. The models themselves vary widely in terms of their type, design, programming, application, scope, level of detail, sophistication, and shortcomings. For many models, some form of mathematical optimization is used to inform the solution process. Energy regulators and system operators in Europe and North America began adopting open energy-system models for planning purposes in the early‑2020s.[1] Open models and open data are increasingly being used by government agencies to guide the develop of net‑zero public policy as well (with examples indicated throughout this article). Companies and engineering consultancies are likewise adopting open models for analysis (again see below). General considerationsOrganizationThe open energy modeling projects listed here fall exclusively within the bottom-up paradigm, in which a model is a relatively literal representation of the underlying system. Several drivers favor the development of open models and open data. There is an increasing interest in making public policy energy models more transparent to improve their acceptance by policymakers and the public.[2] There is also a desire to leverage the benefits that open data and open software development can bring, including reduced duplication of effort, better sharing of ideas and information, improved quality, and wider engagement and adoption.[3] Model development is therefore usually a team effort and constituted as either an academic project, a commercial venture, or a genuinely inclusive community initiative. This article does not cover projects which simply make their source code or spreadsheets available for public download, but which omit a recognized free and open-source software license. The absence of a license agreement creates a state of legal uncertainty whereby potential users cannot know which limitations the owner may want to enforce in the future.[4]: 1 The projects listed here are deemed suitable for inclusion through having pending or published academic literature or by being reported in secondary sources. A 2017 paper lists the benefits of open data and models and discusses the reasons that many projects nonetheless remain closed.[5]: 211–213 The paper makes a number of recommendations for projects wishing to transition to a more open approach.[5]: 214 The authors also conclude that, in terms of openness, energy research has lagged behind other fields, most notably physics, biotechnology, and medicine.[5]: 213–214 GrowthOpen energy-system modeling came of age in the 2010s. Just two projects were cited in a 2011 paper on the topic: OSeMOSYS and TEMOA.[6]: 5861 Balmorel was also active at that time, having been made public in 2001.[b] As of July 2022[update], 31 such undertakings are listed here (with an approximately equal number waiting to be added). Chang et al (2021) survey modeling trends and find the open to closed division about even after reviewing 54 frameworks — although that interpretation is based on project count and not on uptake and use.[7] A 2022 model comparison exercise in Germany reported eight from 40 modeling projects (20%) were open source,[8] these projects also had active communities behind them.[9] Transparency, comprehensibility, and reproducibilityThe use of open energy-system models and open energy data represents one attempt to improve the transparency, comprehensibility, and reproducibility of energy system models, particularly those used to aid public policy development.[2] A 2010 paper concerning energy efficiency modeling argues that "an open peer review process can greatly support model verification and validation, which are essential for model development".[10]: 17 [11] To further honor the process of peer review, researchers argue, in a 2012 paper, that it is essential to place both the source code and datasets under publicly accessible version control so that third-parties can run, verify, and scrutinize specific models.[12] A 2016 paper contends that model-based energy scenario studies, seeking to influence decision-makers in government and industry, must become more comprehensible and more transparent. To these ends, the paper provides a checklist of transparency criteria that should be completed by modelers. The authors however state that they "consider open source approaches to be an extreme case of transparency that does not automatically facilitate the comprehensibility of studies for policy advice."[13]: 4 A one-page opinion piece from 2017 advances the case for using open energy data and modeling to build public trust in policy analysis. The article also argues that scientific journals have a responsibility to require that data and code be submitted alongside text for peer review.[14] And an academic commentary from 2020 argues that distributed development would facilitate a more diverse contributor base and thus improve model quality — a process supported by online platforms and enabled by open data and code.[15] State projectsState-sponsored open source projects in any domain are a relatively new phenomena. As of 2017[update], the European Commission now supports several open source energy system modeling projects to aid the transition to a low-carbon energy system for Europe. The Dispa-SET project (below) is modeling the European electricity system and hosts its codebase on GitHub. The MEDEAS project, which will design and implement a new open source energy-economy model for Europe, held its kick-off meeting in February 2016.[16]: 6 [17] As of February 2017[update], the project had yet to publish any source code. The established OSeMOSYS project (below) is developing a multi-sector energy model for Europe with Commission funding to support stakeholder outreach.[18] The flagship JRC-EU-TIMES model however remains closed source.[19] The United States NEMS national model is available but nonetheless difficult to use. NEMS does not classify as an open source project in the accepted sense.[14] A 2021 research call from the European Union Horizon Europe scientific research funding program expressly sought energy system models that are open source.[20] SurveysA survey completed in 2021 investigated the degree to which open energy-system modeling frameworks support flexibility options, broken down by supply, demand, storage, sector coupled, and network response. Of the frameworks surveyed, none supported all types, which suggests that the soft coupling of complementary frameworks could provide more holistic assessments of flexibility. Even so, most candidates opt for perfect foresight and do not natively admit probabilistic actions or explicit behavioral responses.[21] Electricity sector modelsOpen electricity sector models are confined to just the electricity sector. These models invariably have a temporal resolution of one hour or less. Some models concentrate on the engineering characteristics of the system, including a good representation of high-voltage transmission networks and AC power flow. Others models depict electricity spot markets and are known as dispatch models. While other models embed autonomous agents to capture, for instance, bidding decisions using techniques from bounded rationality. The ability to handle variable renewable energy, transmission systems, and grid storage are becoming important considerations.
AMIRIS
AMIRIS is the open Agent-based Market model for the Investigation of Renewable and Integrated energy Systems. The AMIRIS simulation framework was first developed by the German Aerospace Center (DLR) in 2008 and later released as an open source project in 2021.[22] AMIRIS enables researchers to address questions regarding future energy markets, their market design, and energy-related policy instruments.[23] In particular, AMIRIS is able to capture market effects that may arise from the integration of renewable energy sources and flexibility options by considering the strategies and behaviors of the various energy market actors present. For instance, those behaviors can be influenced by the prevailing political framework and by external uncertainties.[24] AMIRIS may also uncover complex effects that may emerge from the inter‑dependencies of the energy market participants.[25]  The embedded market clearing algorithm computes electricity prices based on the bids of prototyped market actors. These bids may not only reflect the marginal cost of electricity production but also the limited information available to the actors and related uncertainties. But also the bidding can be strategic as an attempt to game official support instruments or exploit market power opportunities. Actors in AMIRIS are represented as agents that can be roughly divided into six classes: power plant operators, traders, market operators, policy providers, demand agents, and storage facility operators. In the model, power plant operators provide generation capacities to traders, but do not participate directly in markets. Instead, they supply traders who conduct the marketing and deploy bidding strategies on the operators behalf. Marketplaces serve as trading platforms and calculate market clearing. Policy providers define the regulatory framework which then may impact on the decisions of the other agents. Demand agents request energy directly at the electricity market. Finally, flexibility providers, such as storage operators, use forecasts to determine bidding patterns to match their particular objectives, for instance, projected profit maximization. AMIRIS is based on the open Framework for distributed Agent-based Modelling of Energy systems or FAME.[26][27] AMIRIS can simulate large‑scale agent systems in acceptable timeframes. For instance, the simulation of one year at hourly resolution may take as little as one minute on a contemporary desktop computer. The researchers at DLR also have access to high-performance computing facilities. Breakthrough Energy Model
The Breakthrough Energy Model is a production cost model with capacity expansion algorithms and heuristics, originally designed to explore the generation and transmission expansion needs to meet U.S. states' clean energy goals. The data management occurs within Python and the DCOPF optimization problem is created via Julia. The Breakthrough Energy Model is being developed by the Breakthrough Energy Sciences team. The open data underlying the model builds upon the synthetic test cases developed by researchers at Texas A&M University.[28][29][30] The Breakthrough Energy Model initially explored the generation and transmission expansion necessary to meet clean energy goals in 2030 via the building of a Macro Grid.[31] Ongoing work adds and expands modules to the model (e.g. electrification of buildings and transportation) to provide a framework for testing numerous scenario combinations. Development of and integration with other open-source data sets is in progress for modeling countries and regions beyond the United States. The model was applied subsequently the 2021 Texas power crisis, in which winter power outages resulted in hundreds of deaths and billions of dollars in economic losses.[32]: 1 DIETER
DIETER stands for Dispatch and Investment Evaluation Tool with Endogenous Renewables. DIETER is a dispatch and investment model. It was first used to study the role of power storage and other flexibility options in a future greenfield setting with high shares of renewable generation. DIETER is being developed at the German Institute for Economic Research (DIW), Berlin, Germany. The codebase and datasets for Germany can be downloaded from the project website. The basic model is fully described in a DIW working paper and a journal article.[33][34] DIETER is written in GAMS and was developed using the CPLEX commercial solver. DIETER is framed as a pure linear (no integer variables) cost minimization problem. In the initial formulation, the decision variables include the investment in and dispatch of generation, storage, and DSM capacities in the German wholesale and balancing electricity markets. Later model extensions include vehicle-to-grid interactions and prosumage of solar electricity.[35][36] The first study using DIETER examines the power storage requirements for renewables uptake ranging from 60% to 100%. Under the baseline scenario of 80% (the lower bound German government target for 2050), grid storage requirements remain moderate and other options on both the supply side and demand side offer flexibility at low cost. Nonetheless, storage plays an important role in the provision of reserves. Storage becomes more pronounced under higher shares of renewables, but strongly depends on the costs and availability of other flexibility options, particularly biomass availability.[37] Dispa-SET
Under development at the European Commission's Joint Research Centre (JRC), Petten, the Netherlands, Dispa-SET is a unit commitment and dispatch model intended primarily for Europe. It is written in Python (with Pyomo) and GAMS and uses Python for data processing. A valid GAMS license is required. The model is formulated as a mixed integer problem and JRC uses the proprietary CPLEX sover although open source libraries may also be deployed. Technical descriptions are available for versions 2.0 [38] and 2.1.[39] Dispa-SET is hosted on GitHub, together with a trial dataset, and third-party contributions are encouraged. The codebase has been tested on Windows, macOS, and Linux. Online documentation is available.[40] The SET in the project name refers to the European Strategic Energy Technology Plan (SET-Plan), which seeks to make Europe a leader in energy technologies that can fulfill future (2020 and 2050) energy and climate targets. Energy system modeling, in various forms, is central to this European Commission initiative.[41]  The model power system is managed by a single operator with full knowledge of the economic and technical characteristics of the generation units, the loads at each node, and the heavily simplified transmission network. Demand is deemed fully inelastic. The system is subject to intra-period and inter-period unit commitment constraints (the latter covering nuclear and thermal generation for the most part) and operated under economic dispatch.[39]: 4 Hourly data is used and the simulation horizon is normally one year. But to ensure the model remains tractable, two day rolling horizon optimization is employed. The model advances in steps of one day, optimizing the next 48 hours ahead but retaining results for just the first 24 hours.[39]: 14–15 Two related publications describe the role and representation of flexibility measures within power systems facing ever greater shares of variable renewable energy (VRE).[42][43] These flexibility measures comprise: dispatchable generation (with constraints on efficiency, ramp rate, part load, and up and down times), conventional storage (predominantly pumped-storage hydro), cross-border interconnectors, demand side management, renewables curtailment, last resort load shedding, and nascent power-to-X solutions (with X being gas, heat, or mobility). The modeler can set a target for renewables and place caps on CO2 and other pollutants.[39] Planned extensions to the software include support for simplified AC power flow [c] (transmission is currently treated as a transportation problem), new constraints (like cooling water supply), stochastic scenarios, and the inclusion of markets for ancillary services.[40] Dispa-SET has been or is being applied to case studies in Belgium, Bolivia, Greece, Ireland, and the Netherlands. A 2014 Belgium study investigates what if scenarios for different mixes of nuclear generation, combined cycle gas turbine (CCGT) plant, and VRE and finds that the CCGT plants are subject to more aggressive cycling as renewable generation penetrates.[45] A 2020 study investigated the collective impact of future climatic conditions on 34 European power systems, including potential variations in solar, wind, and hydro‑power output and electricity demand under various projected meteorological scenarios for the European continent.[46] Dispa-SET has been applied in Africa with soft linking to the LISFLOOD model to examine water‑energy nexus problems in the context of a changing climate.[47] EMLab-Generation
EMLab-Generation is an agent-based model covering two interconnected electricity markets – be they two adjoining countries or two groups of countries. The software is being developed at the Energy Modelling Lab, Delft University of Technology, Delft, the Netherlands. A factsheet is available.[48] And software documentation is available.[49] EMLab-Generation is written in Java. EMLab-Generation simulates the actions of power companies investing in generation capacity and uses this to explore the long-term effects of various energy and climate protection policies. These policies may target renewable generation, CO2 emissions, security of supply, and/or energy affordability. The power companies are the main agents: they bid into power markets and they invest based on the net present value (NPV) of prospective power plant projects. They can adopt a variety of technologies, using scenarios from the 2011 IEA World Energy Outlook.[50] The agent-based methodology enables different sets of assumptions to be tested, such as the heterogeneity of actors, the consequences of imperfect expectations, and the behavior of investors outside of ideal conditions. EMLab-Generation offers a new way of modeling the effects of public policy on electricity markets. It can provide insights into actor and system behaviors over time – including such things as investment cycles, abatement cycles, delayed responses, and the effects of uncertainty and risk on investment decisions. A 2014 study using EMLab-Generation investigates the effects of introducing floor and ceiling prices for CO2 under the EU ETS. And in particular, their influence on the dynamic investment pathway of two interlinked electricity markets (loosely Great Britain and Central Western Europe). The study finds a common, moderate CO2 auction reserve price results in a more continuous decarbonisation pathway and reduces CO2 price volatility. Adding a ceiling price can shield consumers from extreme price shocks. Such price restrictions should not lead to an overshoot of emissions targets in the long-run.[51] EMMA
EMMA is the European Electricity Market Model. It is a techno-economic model covering the integrated Northwestern European power system. EMMA is being developed by the energy economics consultancy Neon Neue Energieökonomik, Berlin, Germany. The source code and datasets can be downloaded from the project website. A manual is available.[52] EMMA is written in GAMS and uses the CPLEX commercial solver. EMMA models electricity dispatch and investment, minimizing the total cost with respect to investment, generation, and trades between market areas. In economic terms, EMMA classifies as a partial equilibrium model of the wholesale electricity market with a focus on the supply-side. EMMA identifies short-term or long-term optima (or equilibria) and estimates the corresponding capacity mix, hourly prices, dispatch, and cross-border trading. Technically, EMMA is a pure linear program (no integer variables) with about two million non-zero variables. As of 2016[update], the model covers Belgium, France, Germany, the Netherlands, and Poland and supports conventional generation, renewable generation, and cogeneration.[52][53] EMMA has been used to study the economic effects of the increasing penetration of variable renewable energy (VRE), specifically solar power and wind power, in the Northwestern European power system. A 2013 study finds that increasing VRE shares will depress prices and, as a consequence, the competitive large-scale deployment of renewable generation will be more difficult to accomplish than many anticipate.[54] A 2015 study estimates the welfare-optimal market share for wind and solar power. For wind, this is 20%, three-fold more than at present.[55] An independent 2015 study reviews the EMMA model and comments on the high assumed specific costs for renewable investments.[33]: 6 GENESYS
GENESYS stands for Genetic Optimisation of a European Energy Supply System. The software is being developed jointly by the Institute of Power Systems and Power Economics (IAEW) and the Institute for Power Electronics and Electrical Drives (ISEA), both of RWTH Aachen University, Aachen, Germany. The project maintains a website where potential users can request access to the codebase and the dataset for the 2050 base scenario only.[56] Detailed descriptions of the software are available.[57][58] GENESYS is written in C++ and uses Boost libraries, the MySQL relational database, the Qt 4 application framework, and optionally the CPLEX solver. The GENESYS simulation tool is designed to optimize a future EUMENA (Europe, Middle East, and North Africa) power system and assumes a high share of renewable generation. It is able to find an economically optimal distribution of generator, storage, and transmission capacities within a 21 region EUMENA. It allows for the optimization of this energy system in combination with an evolutionary method. The optimization is based on a covariance matrix adaptation evolution strategy (CMA-ES), while the operation is simulated as a hierarchical set-up of system elements which balance the load between the various regions at minimum cost using the network simplex algorithm. GENESYS ships with a set of input time series and a set of parameters for the year 2050, which the user can modify. A future EUMENA energy supply system with a high share of renewable energy sources (RES) will need a strongly interconnected energy transport grid and significant energy storage capacities. GENESYS was used to dimension the storage and transmission between the 21 different regions. Under the assumption of 100% self-supply, about 2500 GW of RES in total and a storage capacity of about 240000 GWh are needed, corresponding to 6% of the annual energy demand, and a HVDC transmission grid of 375000 GW·km. The combined cost estimate for generation, storage, and transmission, excluding distribution, is 6.87 ¢/kWh.[57] A 2016 study looked at the relationship between storage and transmission capacity under high shares of renewable energy sources (RES) in an EUMENA power system. It found that, up to a certain extent, transmission capacity and storage capacity can substitute for each other. For a transition to a fully renewable energy system by 2050, major structural changes are required. The results indicate the optimal allocation of photovoltaics and wind power, the resulting demand for storage capacities of different technologies (battery, pumped hydro, and hydrogen storage) and the capacity of the transmission grid.[58] NEMO
NEMO, the National Electricity Market Optimiser, is a chronological dispatch model for testing and optimizing different portfolios of conventional and renewable electricity generation technologies. It applies solely to the Australian National Electricity Market (NEM), which, despite its name, is limited to east and south Australia. NEMO has been in development at the Centre for Energy and Environmental Markets (CEEM), University of New South Wales (UNSW), Sydney, Australia since 2011.[59] The project maintains a small website and runs an email list. NEMO is written in Python. NEMO itself is described in two publications.[60]: sec 2 [61]: sec 2 The data sources are also noted.[60]: sec 3 Optimizations are carried out using a single-objective evaluation function, with penalties. The solution space of generator capacities is searched using the CMA-ES (covariance matrix adaptation evolution strategy) algorithm. The timestep is arbitrary but one hour is normally employed. NEMO has been used to explore generation options for the year 2030 under a variety of renewable energy (RE) and abated fossil fuel technology scenarios. A 2012 study investigates the feasibility of a fully renewable system using concentrated solar power (CSP) with thermal storage, windfarms, photovoltaics, existing hydroelectricity, and biofuelled gas turbines. A number of potential systems, which also meet NEM reliability criteria, are identified. The principal challenge is servicing peak demand on winter evenings following overcast days and periods of low wind.[60] A 2014 study investigates three scenarios using coal-fired thermal generation with carbon capture and storage (CCS) and gas-fired gas turbines with and without capture. These scenarios are compared to the 2012 analysis using fully renewable generation. The study finds that "only under a few, and seemingly unlikely, combinations of costs can any of the fossil fuel scenarios compete economically with 100% renewable electricity in a carbon constrained world".[62]: 196 A 2016 study evaluates the incremental costs of increasing renewable energy shares under a range of greenhouse gas caps and carbon prices. The study finds that incremental costs increase linearly from zero to 80% RE and then escalate moderately. The study concludes that this cost escalation is not a sufficient reason to avoid renewables targets of 100%.[61] OnSSET
OnSSET is the OpeN Source Spatial Electrification Toolkit. OnSSET is being developed by the division of Energy Systems, KTH Royal Institute of Technology, Stockholm, Sweden. The software is used to examine areas not served by grid-based electricity and identify the technology options and investment requirements that will provide least-cost access to electricity services. OnSSET is designed to support the United Nations' SDG 7: the provision of affordable, reliable, sustainable, and modern energy for all. The toolkit is known as OnSSET and was released on 26 November 2016. OnSSET does not ship with data, but suitable datasets are available from energydata.info. The project maintains a website and runs a mailing list.[63][64][65]  OnSSET can estimate, analyze, and visualize the most cost-effective electrification access options, be they conventional grid, mini-grid, or stand-alone.[66] The toolkit supports a range of conventional and renewable energy technologies, including photovoltaics, wind turbines, and small hydro generation. As of 2017[update], bioenergy and hybrid technologies, such as wind-diesel, are being added. OnSSET utilizes energy and geographic information, the latter may include settlement size and location, existing and planned transmission and generation infrastructure, economic activity, renewable energy resources, roading networks, and nighttime lighting needs. The GIS information can be supported using the proprietary ArcGIS package or an open source equivalent such as GRASS or QGIS.[67] OnSSET has been applied to microgrids using a three‑tier analysis starting with settlement archetypes.[68] OnSSET has been used for case studies in Afghanistan,[69] Bolivia,[68][70] Cameroon, | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||