This section describes the key enablers required for successful long-term energy policy planning. Each section contains a self-assessment guiding the user to identify the priority areas for further development. The lists of questions are nowhere near exhaustive, but high number of either ‘YES’ or ‘NO’ answers provide indications of the status quo.

Developing scenario-building capabilities is a long-term undertaking that requires commitment and constant training. National ownership and capacity building are key principles for developing a robust strategic energy planning process. The Energy Technology Systems Analysis Program (ETSAP) is one of the IEA’s longest-running Technology Collaboration Programmes. In its experience, dedicated institutes and universities can offer the commitment necessary to develop modelling capabilities. but in all cases a clear mandate and continuous support from the government are necessary.

Governments also play a vital role in ensuring that external support for energy planning aims to maximise the long-term benefits for the country. While criteria for donors have been developed, it is up to policymakers to ensure the principles are respected.

To avoid optimising the energy system at only the sectoral level, policymakers need to obtain a holistic overview of the entire system and all potential pathways. However, given the complexity of modern systems, it is not feasible to conduct such analysis manually. Energy system modelling thus allows policymakers, analysts and statisticians to use scenario analysis data to assess the impacts of various alternatives, which can in turn be used to inform and elaborate policies and targets.

Once a country acknowledges the need to improve its long-term energy planning process and capacities, the first question should NOT be “what tool should we use?”. Instead, the foremost order of business should be to appoint a dedicated entity to co‑ordinate the policy planning process and take stock of existing national energy modelling capacities.

Initial resource allocation should also take the skill development of individual experts into account. Furthermore, it should be recognised at the highest political level that developing scenario-building capacity is a long-term task requiring commitment and constant training.

Many of the differences between “continuous” and “discrete” planning are linked to the funding of activities. In practice, a co‑ordinator of long-term energy planning activities, as well as energy modelling experts, should be sustainably funded (i.e. staff salaries and IT infrastructure costs should mainly be covered by the core state budget, instead of through international grants or project funding).

In this context, analytical capacity refers to the ability of policymakers and other officials to develop, understand and use scenario insights to inform policy and investment decisions. Developers and users of the scenarios are rarely the same people, therefore these capacities can be developed independently. For a country only establishing a scenario-based energy policy planning process, focus should be on the ‘user base’, i.e. strengthening their understanding of the research questions, scenario development, as well as the overall policy planning process elements.

At the highest levels of national and international energy policy-making, countries’ ambitions and obligations can both be reduced to single-figure targets (e.g. “The share of renewable electricity will be 50% by 2030.”). Announcing such targets is easy, but for it to hold any credibility it must become official national policy, which requires a great deal of analysis – particularly to create the scenarios to be researched during policy development.

The capacity to conduct this type of analysis often resides either within the government (internal capacity) or another institute (external capacity). The latter can be further divided into public (national) or private (either national or international). The International Renewable Energy Agency (IRENA) has summarised the pros and cons of both alternatives.

ETSAP has surveyed where energy analysis capacity resides within its community member countries. The vast majority (90%) of respondents indicated either a national technical institute or a university, with only one of the 20 countries surveyed indicating that a government department possesses this capacity.

Indeed, experience has shown that a hybrid approach in which technical capacity exists in a national technical institution that collaborates closely with the government tends to yield good results. In this setup, the capacity is protected from potential political turmoil and constant short-term assignments. There is also consensus among energy planners that, even with dedicated staff, it takes several years to develop a national energy model from scratch.

Possessing the right data for quantitative analysis is crucial for its relevance. Countries are therefore encouraged to systematically collect energy statistics, adhering to international methodologies and standards, particularly the International Recommendations for Energy Statistics (IRES). When data gaps are detected, working with key data providers to narrow those omissions is recommended. Implementing a Long-Term Energy Policy Planning Process for Azerbaijan discusses how to develop new surveys to increase energy data coverage. Rarely all data needed for the energy system analysis is readily available. Therefore, improving data at the national level should be seen as a continuous process paralleling energy planning.

To this end, the IEA recently developed guidelines to improve national energy demand data and energy efficiency indicators, as well as a framework for statistical capacity to produce national energy data (forthcoming). As we therefore do not offer detailed advice on improving the quality of energy statistics here, users applying this roadmap in a national context are advised to consult the data-related documents if they encounter important data gaps.

However, it is worth highlighting one of the key points of these publications: the need to collaborate with key data providers. The entity responsible for producing national energy statistics (e.g. a national statistical office) should be involved in policy planning as both a data provider and the authority tracking (quantifying) the impact of the policies through the established indicators.

Compiling an energy balance according to international recommendations is a good starting point for elaborating an energy system model, with the minimum viable dataset items being:

Energy supply data:

• energy production
• energy imports and exports
• energy stocks.

Energy demand data:

• energy transformation (e.g. electricity and heat, oil refining)
• consumption by economic sector (industry, transport, residential, services, agriculture, etc.)
• non-commercial energy consumption (e.g. fuelwood).

However, country circumstances must always be considered as certain energy flows may be unique to the country context and thus not included in the generic energy statistics guidelines. Also, data requirements are not limited to basic energy information, as a range of other inputs – sometimes interconnected adding to the complexity – may also be necessary:

• macroeconomic drivers (population, GDP)
• economic activity data (e.g. steel production) and end-use intensities
• historical stock (e.g. electrical capacity), cost and performance (e.g. process efficiencies) of technologies
• policies and regulations (quantified for the models)
• socio-economic drivers
• price data (fuel prices, end-user prices, CO₂ price).
  

Energy data have three key uses in the energy policy planning process:

Calibrating the used-energy model. Model assumptions are validated to ensure that similar results are produced as for the year for which the latest data are available. Without this step, the model would be disconnected from reality. Therefore, calibration to the initial period is one of the key tasks when setting up any model. The more disaggregated data are on the demand side, the more accurate the modelling results can be.

Users of bottom-up energy models should interact with the energy data providers in case notable discrepancies are observed to understand whether they are caused by limitations in energy data collection or the model itself. Ideally, energy data collection and modelling support each other’s activities.

Scenario drafting. Without accurate information on the current energy landscape, it is not possible to quantify current key performance indicators (KPIs). In other words, there are no benchmarks against which to set future targets (e.g. the share of renewables in final energy consumption).

Tracking progress. Part of developing long-term energy policies is to define a set of KPIs that can be used to monitor the impact of those policies. Implementing a Long-Term Energy Policy Planning Process for Azerbaijan (pp. 33-34) discusses this subject in detail.

It is important to acknowledge that no model or tool can improve the quality of input data. Even a well-designed model fed poor or incomplete information will not produce useful results. Conversely, uncertainty of the modelling results can be reduced if the input data is of high quality (in terms of accuracy, coverage, etc.).

With improved data availability, certain assumptions, non-country-specific or low-quality data can be replaced with primary data to reduce the uncertainty of model outputs. To quantify the overall uncertainty of a model’s results – primarily a function of the input information – energy system analyses are exposed to sensitivity tests. Through this testing, it is possible to identify which input values have the greatest impact on the results and what the overall confidence interval of the results is.

Nevertheless, limited data availability should not dissuade one from undertaking long-term energy planning. While improving the data situation will take time, it must begin somewhere. It is best to start from modest ambitions, develop an efficient methodology, identify key data gaps and establish a co‑ordination framework – and then gradually develop and improve work in these areas.

Investment decisions cannot be tested in the real world, as they involve large capital requirements. In addition, certain choices may have irreversible impacts e.g. on the energy system or environment. Scenarios have therefore become the de facto approach to depict possible outcomes of development alternatives when evaluating climate and energy policy measures. It is useful to distinguish between the scenario development process (including actors involved and models used) and the use of scenarios as a tool in policy planning.

The Intergovernmental Panel on Climate Change (IPCC) defines a scenario as a “coherent, internally consistent and plausible description of a possible future state of the world. It is not a forecast; rather, each scenario is one alternative image of how the future can unfold.” It also recommends creating a set of scenarios to reflect the range of uncertainty in projections.

Based on its stocktaking exercise, IRENA has formulated a series of key recommendations for policymakers to maximise the benefits of the scenario approach:

Strengthen scenario development by

• Establishing a strong governance structure through a participatory scenario development process and co‑ordination among all stakeholders.
• Expanding the boundaries of scenarios to include social impacts of the energy transition and innovation in the energy sector.

Improve scenario use by

• Clarifying the purpose of scenario building internally.
• Communicating the results transparently and effectively (see also Enabler 6).

Particularly because of the general global push for net zero emissions, countries must prepare scenarios that not only take national priorities into account, but also international commitments, such as nationally determined contributions (NDCs) to reduce GHG emissions. These must be considered in government policies as well as in investment deals with companies.

Scenarios can be divided into three main categories depending on their approach. It is worth noting that the mechanics of energy system modelling are the same regardless of the selected scenario type.

Often, the scenarios developed include both a reference case (often colloquially called business-as-usual) assuming that current trends and policies remain in effect unchanged, and an ambitious (even idealistic) case that demonstrates at what cost such goals would be met. The other scenarios then set between these two extremes. Most importantly, developing scenarios for a country’s energy future should ideally be done in consultation with a range of experts (i.e. in climate, the environment and economics). Furthermore, scenario development entails more than just modelling: building and maintaining an organisation’s capacities is a challenge that requires time and continuous training to adapt to the evolving modelling environment.

Exploratory (possible) scenario:

• addresses the question, “What could happen?”
• example research query: “We have pledged to develop 500 MW of renewable electricity capacity by 2040. How could this impact the energy system?”
• example scenario: IEA Stated Policies Scenario (STEPS)

Normative (target-oriented):

• addresses the question, “How can a specific target be reached?”
• example research query: “We have pledged to become a net-zero emitter by 2050. How can this be achieved at lowest cost?”
• example scenario: IEA Net Zero Emissions by 2050 Scenario (NZE)

Predictive (probable) scenarios:

• addresses the question, “What will happen on the condition of some specified events ?”

Often, the scenarios developed include both a reference case (often colloquially called business-as-usual) assuming that current trends and policies remain in effect unchanged, and an ambitious (even idealistic) case that demonstrates at what cost such goals would be met. The other scenarios then set between these two extremes. Most importantly, developing scenarios for a country’s energy future should ideally be done in consultation with a range of experts (i.e. in climate, the environment and economics). Furthermore, scenario development entails more than just modelling: building and maintaining an organisation’s capacities is a challenge that requires time and continuous training to adapt to the evolving modelling environment.

Key takeaways have emerged from lessons and interviews related to the Finnish scenario development process:

Rather than suggesting that energy modelling should not be done, this famous quote often attributed to George Box implies that no model can completely and accurately mimic reality. This underlines the importance of rigorous preparation of scenarios and their underlying data inputs.

Energy sector models are analytical tools used to analyse the performance and dynamics of energy systems over time. The primary purpose of energy modelling is to provide evidence for policymakers on the impacts of different policy options to national energy security, economic development and environmental aspects. As many energy system models are effective illustrators of technology-rich, least-cost energy system pathways, they have been used extensively (initially in the 1970s and 1980s) to explore least-cost options to reinforce energy security and, more recently (particularly since 2000), to transition to a low-carbon future.

For conducting quantitative analyses, energy systems are presented in organised model structures involving inputs/data, equations and outputs. This enables analysts to compare different system configurations without the cost of building them. Ideally, the results will make it easier to design an energy system, but will also account for local/national resources, energy demand, policies and other constraints.

Building an energy system model requires several key components: a model generator; a solver; interfaces for handling data and results; and a detailed database. Energy system models also require key external inputs such as energy supply and demand data, as well as policy objectives (scenarios). While this may sound complicated, good results can already be achieved with relatively simple spreadsheet tools. Thus, a country just initiating development of a national energy system model may begin its journey with this approach. Once the policy planning processes have been established, it may become relevant to develop more complex capacity and models.

In other words, a model translates a real-world energy system into a set of mathematical equations. These equations then rely on input data (see Enabler 3) to begin the computations, but also use it to formulate objectives to be reached at the end of the modelling exercise.

While energy modelling should not be considered an ultimate policy objective or a means to formulate absolute policy decisions, it is a useful tool for informing discussions and drafting legislation. Furthermore, given the complexity and global nature of today’s energy systems, it is also the only option for managing countless system variables.

Nevertheless, energy modelling should not be seen as an isolated mechanical data input-output activity. The several steps it encompasses both before and after the actual computational modelling exercise aim to minimise uncertainty and maximise the usefulness of the results.

The first steps of the modelling process show that having high-quality initial data is crucial. When information is not available, assumptions and estimates must be made, but quantifying certain real-life phenomena (e.g. socio-economic behaviour) is extremely complex, if not impossible. Furthermore, the technical constraints, efficiencies and costs of new technologies, as well as the timing of their commercial introduction, are highly speculative.

The uncertainty that these approximations inevitably introduce into the model outputs must be understood and potentially quantified through multiple scenarios or sensitivity analyses. Through these examinations, it is possible to gauge a model’s usefulness for describing the potential future, given a set of boundary conditions and/or ultimate targets. Indeed, models should demonstrate robust steps that need to be taken in the immediate future while ensuring that the chosen path will not be regretted later.

Picking the tools

Compiling an exhaustive review of all available modelling tools is not feasible given the sheer number of options. Before discussing individual tools, it is essential to understand that common to each one is the need to have experts operating, maintaining and updating them. Countries at the initial stages of establishing an in-house energy planning team should start small, for example with a simple spreadsheet tool instead of a complex energy system optimisation model.

The IEA publication Implementing a Long-Term Energy Policy Planning Process for Azerbaijan reviews some commonly used tools and touches upon grouping the tools from a technical point of view (optimisation vs simulation, bottom-up vs top-down). From the perspective of capacity building, it is important to distinguish between commercial (proprietary) and open-source (non-proprietary) tools.

Commercial tools come at a cost for users, but often offer a wider support base, are updated regularly and provide professional training courses. In many cases, developing countries and academic institutions may purchase the licence at a discounted rate. Nevertheless, annual licence costs are often a barrier to the adoption of modern modelling tools.

To address the cost issue, more and more tools are being made available for free. Generally, these tools rely on user-community maintenance and peer support, including through capacity-building events.

A country’s choice of model should be based on its distinct national needs. While models have been tailored to specific situations in the past, this solution depends solely on a country’s internal capacity and thus poses a higher risk of corporate memory loss than more widely used applications. Tools (often more than just one) should be chosen based on several considerations:

• Is the tool able to provide insights into the main research questions?
• What are the tool’s primary limitations?
• Is the tool widely in use, i.e. is there a support community?
• What are the running costs of the tool (e.g. for licences)?
• Does domestic capacity exist or must it be developed?
• How difficult is it to develop relevant IT skills?

Regardless of the tool, according to ETSAP a key success factor is to have a committed team working almost full-time on the modelling activity during the initial phase when capacity is being developed. Such a team becomes all the more important once the activity expands to cover several parallel models specialising in different aspects of the economy.

A general criticism of energy system models is that their structural complexity and numerous input parameters and variables render them a “black box” for non-modellers who lack specialised programming knowledge. It should also be emphasised that the results of the modelling must be interpreted and refined by experts to become useful inputs for policy-making. Thus, such capacity should also be developed (see Enabler 2).

Multiple scenario analyses consistent with different policies produce wide uncertainty ranges in the results, whereas policymakers often prefer to see a discrete number of outcomes instead of a range. While academic literature embraces uncertainty, the public perceives a wide range of outcomes as an indication that scenarios do not offer much more than evidence that almost anything is possible.

Since a model’s accuracy – and thus its usefulness – inevitably depends upon the number of details and input fed into it, it is necessary to find other means of sharing results than asking non-experts to explore the model or the raw results themselves. Once a modelling exercise has yielded robust outputs, it is important that policymakers communicate this information in a concise and transparent manner.

For full transparency, the final deliverable (e.g. a report or presentation) should be accompanied by relevant documentation to allow interested parties to examine the underlying assumptions and data sources used for the modelling exercise.

Furthermore, it is in the best interests of the modelling team to properly document their work so that the model (and the results) can withstand both national and international scrutiny. Ideally, the scenario development phase should already be clearly documented and publicly available.

The UK Energy Research Centre defines three levels of model transparency:

1. Open-description models – provide a concise methodological summary, and outline documentation and links to outputs and applications.
2. Open-access models – give a user group access and shared responsibility for model development, plus offer full documentation and data sets.
3. Open-source models – are fully transparent and accessible, available for any user to download and use.

However, experience shows that energy modellers – whether in academic, government or consulting environments – often struggle to make their models open and accessible to key stakeholders. Ensuring a model’s transparency is time-consuming, unglamorous, rarely prioritised by funders, and undervalued in terms of a modeller’s career progression. Nevertheless, transparency is crucial to ensure that the implications and limitations of a model’s insights can be fully understood.

A medium-term aim is to make all policy-orientated energy modelling more transparent (at least level 1 – open description), while all publicly funded energy modelling should reach a higher transparency threshold (at least level 2 – open access). Modellers can update their entries as efforts are made to increase the transparency of their models.

Because countries can find it difficult to build on their energy plans, especially when planning activities are supported by external assistance (e.g. donor-funded consultancies), the U4RIA standards were developed to regulate the documentation of data and data sources, methodologies applied, and assumptions used in energy system planning studies and projects.