Energy system analytics and good governance-U4RIA goals of Energy Modelling for Policy Support


 Energy modelling is the process of using mathematical models to develop abstractions and then seek insights into future energy systems. It can be an abstract academic activity. Or, it can insert threads that influence our development. We argue therefore, that energy modelling that provides policy support (EMoPS) should not only be grounded in rigorous analytics, but also in good governance principles. As, together with other policy actions, it should be accountable. Almost all aspects of society and much of its impact on the environment are influenced by our use of energy. In this context, EMoPS can inspire, motivate, calibrate, and ‘post assess’ energy policy. But, such modeling is often undertaken by too few analysts under time and resource pressure. Building on the advances of ‘class leaders’, we propose that EMoPS should reach for practical goals — including engagement and accountability with the communities it involves, and those it will later affect. (We use the term Ubuntu, meaning ‘I am because you are’ to capture this interdependency). We argue that Ubuntu, together with retrievability, repeatability, reconstructability, interoperability and auditability (U4RIA) of EMoPS should be used to signal the beginnings of a new default practice. We demonstrate how the U4RIA principles can contribute in practice using recent modelling of aspirational energy futures by Costa Rica as a case study. This modelling effort includes community involvement and interfaces and integrates stakeholder involvement. It leaves a trail that allows for its auditing and accountability, while building capacity and sustainable institutional memory.


Introduction
Energy Modelling for Policy Support (EMoPS) is more than simply an analytical activity -it affects and involves communities. Including those involved in real time in the EMoPS proces, and those outside of the EMoPS process, who may be affected by its outcomes at some later time.
Communities in focus include those involved in the analytical and policy formulation work itself, subsequent (energy dependent) service users, those locally and internationally who are impacted by the uptake of energy system development, as well as project developers and nanciers. Energy is not an end in and of itself, but a means to supply -in the context of nancial and non-monetary costs -energy services. Energy demands are thus driven by other sectors and determine wider production possibilities.
With its policy impact EMoPS, which can in uence energy system con guration and service costsin uences key aspects of societal development, and its constituencies. For these reasons, EMoPS should be held accountable through adherence to good governance as well as rigorous scienti c practice.
The signi cance of integrated modelling EMoPS is in part an analytical activity that, via mathematical abstractions programmed into computer models, projects internally consistent scenarios of energy system development. As an academic activity, it requires scienti c rigour. Its analytical components include links with scienti c bodies of knowledge and analysts often interact with specialists from other domains.
EMoPS scenarios (one of several modelling artifacts 1,2 noted in the supplementary online material (SOM)) typically provide directly, or with additional analysis, explicit quantitative pictures of: technology and infrastructure con gurations; energy security and trade levels; direct and secondary market implications and investment provisions; emissions trajectories; revenue and public subsidy requirements; and life-style implications. Scenarios enable a 'thought-experiment space' that allows one to run through various futures that may be driven by new policies and evaluate the effect of sets of policies. One common practice is to develop aspirational end-points for those scenarios and then model to investigate normative scenarios of how one might get there -a technique known as backcasting. 3 Aspects of scenarios have relevance to large sets of communities. They can also be translated into a 'language' that is understood by each community -though they are often domain speci c. Scenarios can also be loaded with tacit judgments that are not necessarily made explicit. 4 No existing guidelines govern the identi cation of the relevant communities, the identi cation of the scenario elements and aspirations of importance to those communities, their active translation and appropriate interface of that community with an EMoPS process. Neither are the costs nor bene ts of potential guidelines articulated, should they indeed be adopted. This despite widespread, often tacit agreement of the value of scenarios for those communities. 5 The communities affected Aside from those that will later be potentially affected, all speci c communities associated with the EMoPS ecosystem of activities and scenarios produced need support for the health of the activity. These may include: the energy modelling team and broader energy modelling community; energy policy analysts that bridge the policy process with EMoPS; a (formal or informal) coordination group that enables strategic intelligence and manages information ows with key external communities; a coordination group of a broader participation process; and, the funders that support the EMoPS process.
The latter might arise from dedicated government budgets or external support agencies.
From EMoPS scenarios, the implications and 'underpinning goals' of different futures are articulated, and their implications for affected communities sketched ( Table 1). The communities that are represented (or can be represented in those sketches) include various policy organs; direct energy industry suppliers and indirect (energy) service users; civil society and international actors. Here, internationals are at least divided into business, development, environment, and energy(-system) trading partners. Consider for example, multinational energy companies, the UNFCCC, countries with electricity and gas interconnectors, respectively as examples. Four critical observations arise 1. Opaque analysis: even when deeply domain speci c, despite a growth of efforts to combat this, the modelling process often remains a 'black box'. 6 That means that it is often impossible to develop sensible knowledge management within a national modelling team. When active knowledge management is missing due to opacity, it is not feasible to create an ecosystem with institutions and research centres that could increase national competence and capacity. That results in continuous wasted effort. In developing country contexts, wasted resources and negative consequences are reported. 7 2. Community cooperation: the level of involvement of socio-economic actors is necessarily constrained as resource, time, and other factors create impractical overheads. The bene ts of competent cooperation and co-creation are well known, including the potential for increased accountability 8 and ownership. 9 But various levels of cooperation with the modelling team are possible. A stylised representation is given by in gure 1. 10 Yet many EMoPS activities are without appropriately de ned analytical and organisational work ows. They have no explicit interface for community participation. As such even 'light' cooperation can be di cult.
3. Downstream matters: the impact of the resulting energy policy from EMoPS is not trivial. And the continuum from EMoPS to implementation is often broken. This can range from the translation of energy model results into energy policy to energy policy that is incoherent and contradictory to policy in other areas. A common example of the former is the implementation of non-market mechanisms to proxy the output of a market model, creating inaccuracies. A common example of the latter is the dependence on indirect fossil fuel taxes for national development revenue, while efforts are made to reduce fossil fuel consumption for reasons of GHG mitigation.

Appropriate answerability: accountability for related national development nancing (as it relates to
EMoPS and its output) is lacking. Money might be directly wasted 7 as there may be no simple knowledge management process in the analytical team. As its insights might not be scrutinizable, the resulting (much larger indirect) investments and the broader development they drive are at risk as a result. (That is not to say that an 'exemplary' EMoPS will produce scenarios that perfectly 'predict' future needs. Although it might be built on the best available information, it is still imperfect knowledge. 12 However, if its outputs and process cannot be scrutinized and built upon, it is not possible to know if due diligence was applied.)

Interacting With Affected Communities
The information exchange interfaces between analytical modelers and potentially affected communities are steadily improving in most cases.
Firstly there is an acknowledgment of the need for an interface. 13 Existing 'interfaces' are overwhelmingly community or sub-community speci c. 14 At national and global level there is a move to improve communication with the nance and economic sectors, with two notable examples being: The Network for Greening the Financial System (NGFS) explorer 15 provides a tailored view for the nancial sector of a limited set of global integrated assessment models' quanti cation of investments required to transform the energy system.
The 2050 platform 16 uses a simple spreadsheet 'dashboard', to be used across modelling tools and country studies, with a common 'language' for cross-country comparisons -characterising a limited set of transparent technical economic transformations and serving as an aggregator in a bottom-up approach, where the global vision emerges as a composite of sectoral and national pathways.
We discern four general forms of interface, represented by: The Climate Calculator by the UK Department of Energy and Climate Change (DECC) 17 provides a systematic calibration of sub-sector policy ambition and a set of xed reports. The approach used a simple accounting model by default, but allows for the introduction of other approaches. 18 Fully open source, it is deployed by at least 24 national governments. 19 Gami cation to reach a broader audience, are typically designed to interactively communicate the relationships between key drivers -their interaction and the resulting outcomes can instill intuition and help dispel myths. 20 Ad hoc dashboards are popular, these might range from standardised static graphics (typically covering capacity and production metrics) to Sankey charts to dynamic systems with varying levels of interactivity.
The subsequent dialogue and discussion between client and modeller. Although unless carefully minuted, this can increase opacity when bilateral and private.
The design of these interfaces are often driven by the terms of reference of the study in question, which may be in the form of 'long term energy scenario' procurement, 21 or within the tradition of the modellers, which will have elements of speci c requirements. A development of importance is the databasing of results in a manner that the interfaces discussed above can be developed using standard web APIswith one community resource being the Open Energy Platform (OEP). 22 While interfaces are being developed, simply having an interface might not be su cient. It may require that the EMoPS process actively interface with the target communities as part of its analytical or organisational work ow for effective stakeholder engagement.

Auditability and good governance
Accountability is "the fact of being responsible for what you do and able to give a satisfactory reason for it, or the degree to which this happens". 23 At a minimum, a critical component of EMoPS accountability is auditability. As the EMoPS process directly uses and indirectly in uences taxpayer money -this is an important rst step. (Indeed the initial trial of U4RIA guidelines focuses on ensuring accountability to the community funding the analytical activity in question.) To be substantially sustainable, the EMoPS process requires basic knowledge management in the energy modelling team. With all EMoPS elements available for uptake when staff change, new modelling efforts, or modellers, can build on the old. Accountability can go further, to deliberate multi-sector, multi-stakeholder/community modelling cooperation. That cooperation might be unilateral or be part of joint action, with associated burdens, bene ts, and required interfacing. Auditability provides a minimum level of accountability and requires that documentation be part of the modelling process.
To move to practical action, an initial set of guidelines have been developed and are being trialled. These guidelines include a template to be annexed to terms of reference (ToR) of EMoPS projects. The introduction of the guidelines and ToR annex trial is to overcome short term incentives that mitigate against some forms of accountability-by-default when EMoPS is undertaken in the short term. (Some incentives and the draft ToR annex are noted in SOM7.) The guidelines are labelled 'U4RIA'.
U4RIA stands for Ubuntu, Retrievability, Rusability, Repeatability, Reconstructability, Interoperability and Auditability. Where Ubuntu is derived from the nguni word relating to our shared humanity.

Analytic And Governance Principles
U4RIA extends previous 'best practice' 24 and advances in transparency 25 by placing an emphasis on active mapping of affected communities, identi cation of common information andits translation, and the EMoPS process interface with those communities. Analytically it applies aspects of FAIR principles 26 .
Bookended by Ubuntu and Auditability, the U4RIA goals provide a set of guidelines and best practices for EMoPS. Speci cally,U4RIA introduces guidelines to promote the following: Ubuntu: requires identifying those to whom EMoPS should be accountable and to what degree. This will include: their role in the organisational work ow, determining the EMoPS output of relevance, and its translation and interface requirements in relation to EMoPS digital work ows.
Retrievability: functional retrievability is necessary. Noting that even published data are often neither easily ndable nor accessible.
Reusability: we note that though elements of EMoPS might be reusable, common licensing constraints (noted in SOM6) make this di cult.
Repeatability: though elements of EMoPS might be retrievable, lack of an explicit and user-friendly digital work ow, make repeatability functionally di cult.
Reconstructability: extends 'repeatability' to include the instructions for how to rebuild the EMoPS elements, such as input data, model relations and resulting scenarios. This is done with appropriate 'digital' metadata, as well as appropriately documented digital and organisational work ows. Organisational work ows are required to reconstruct stakeholder input (noting that this may result in difference with different circumstance) Interoperability: allows for scenario outputs to be (1) tested by other models or approaches and (2) their compatibility to sub-sector or broader integration with other modelling for policy support. An example of sub-sector integration includes technical operations required to integrate high levels of renewable energy technologies (RET) into a power system. An example of broader integration would include assessing potential tax revenue changes associated with (subsidised) EVs replacing (more heavily taxed) fossil-fueled cars.
Auditability: is essential as EMoPS is part of the policy process it needs to be held accountable to good governance principles. Those include direct accountability to the internal or external funder of EMoPS, as well as to society more broadly.
Selected bene ts of Retrievability, Reusability, Repeatability, Reconstructability, Interoperability and Auditability delineated by some key affected communities are detailed in SOM5.

Case Study Costa Rica
Since 2018, Costa Rica has run a public engagement process to determine the preferred trajectory of its energy system (and its entire economy) with the goal of reaching net-zero emissions by 2050, all discussed in its National Decarbonization Plan. 27 It was this application that rst gave rise to the U4RIA principles.
Its modelling effort to develop an o cial EMoPS to support national energy policy includes: community involvement that builds human capacity and sustainable institutional memory and leaves a trail that allows for its auditing and accountability plus; interfaces and integration of stakeholders that pushed for U4RIA of the EMoPS process.

Community involvement
U4RIA is applied to the elements and work ows to allow for the sustainability of the EMoPS effort and its accountability to constituent communities.
Firstly the modelling team was developed and capacity built. Effort was made to ensure that local capacity was developed in a partnership between government and academia, with policy needs feeding research agendas. And with the academics plugging into broader modelling communities 28,29 that included academic partners, directors and engineers from energy-related organizations, NGOs, development banks, and civil society.
The need to produce an EMoPS tool suitable for multiple experts enables auditing and accountability. It also embraced the co-creation of common databases stored today here. 30 Stakeholder involvement Broader stakeholders were mapped by potential impact and included universities, government departments, electricity and oil companies, NGOs, and development banks. 27 And methods for communication were developed. These included the use of the 2050 template, custom presentations and workshops to allow for stakeholder appropriate translation. 31 An organisational work ow (Fig. 3) was developed to ensure that the effort was collaborative scenarios were developed in response to stakeholder feedback (ibid).
Input data, the model code, and digital work ow scripts are easily retrieved from GitHub 32 archived with a speci c work ow 33 allowing for retrieval, reuse and repeatability. In particular it was reused to develop an o ce national decarbonisation investment plan 34 . There is limited descriptive meta-data other than the date stamping, which should be improved. A description of the tool covering its original version is available. 35 Further, a publication describing the process has been peer reviewed was peer reviewed. 31 Interoperability was achieved by clear data and model (meta data) descriptions. Those include -with (semi and fully) automated data exchange -the link of the EMoPS to the GEM7, a transfer estimation module between actors, 36 and on-going effort to link the EMoPS with a high resolution power ow simulation of the electricity system. The second allowed for studies with scal options in line with decarbonization goals housed in the nance ministry with development banks partnership, and subsequent loans. 37 The combination allows for aspects of an audit of the organisational and analytical work ow. This has resulted in subsequent work to build on that effort that can be used to support national commitments and multi-donor bank development loans.
Parallel efforts are undertaken in Cyprus -developed together with IRENA, 38 and used for European Union reporting obligations. 39 However, as con dentiality constraints on certain data prevented some U4RIA goals from being achieved, an open and public equivalent of key data was developed. 40 Comparable efforts across global, regional and non-governmental domains are being developed to support the transition to U4RIA focused EMoPS. Notable examples include the Open Energy Outlook project covering the United States 41,42 and similar initiatives being undertaken broadly within the openmod 25 community.

Conclusion And Next Steps
We argue that U4RIA development and adoption is a step toward making energy modelling for policy support more accountable. As there are limited applications of good governance and FAIR scienti c principles, to EMoPS, it is necessary that U4RIA is trialed and developed. Though experience gained in Costa Rica, the beginnings of 'how' this might be done are suggested. Meanwhile a working document was presented (see SOM7) that is being trialled and re ned. It is based on existing World Bank Group standards, and extended by the German Corporation for International Cooperation (GIZ) for application. A round table process to capture lessons and revisions for this document is currently underway.
The bene ts of moving toward a more inclusive and fully open analysis of climate and energy policies include better quality analysis, better knowledge retention, enhanced public accountability, and potentially improved public acceptance.
Declarations improve the e cacy of the science policy interface and provide draft terms of reference' for good practice and, in particular, when procuring energy modelling services for policy support from third-party organizations.
Apart from support in kind provided by the employers of the authors we also acknowledge funding from he Energy and Economic Growth (EEG) and Climate Compatible Growth Program (#CCG) of the UK's Foreign Development and Commonwealth O ce (FCDO). The views expressed in this paper do not necessarily re ect the UK government's o cial policies.
Con ict of interest: There are no con icts of interest. Figure 1 Modelling with stakeholders. Source10 Energy development and its impact. Source11