Towards a Model-Agnostic Online Platform for Structured Input, Scenario Management, and Results Analysis

  • genre: break‑out‑group
  • title: Towards a Model-Agnostic Online Platform for Structured Input, Scenario Management, and Results Analysis
  • presenter: @akanudia
  • description:

I’d like to propose a breakout session at the upcoming Openmod workshop at KTH (March 25–26, 2025) to explore a structured, model-agnostic approach to input creation, scenario specification, and results analysis.

Many open models (e.g., OSeMOSYS, PyPSA, Calliope) handle input data in different ways, and scenario analysis often requires manual scripting. I’d like to demonstrate a flexible, rule-based workflow that allows users to browse, run, and compare the same dataset across multiple models (TIMES and OSeMOSYS, to start with – others can be added).

Veda has supported TIMES models from the beginning and Veda online has emerged as a powerful companion recently. It has made it possible to engage domain experts and stakeholders directly in the modeling process. What I am demonstrating here is a generalized version of the same platform.

Session Outline (Demonstrating Two Workflows):

Workflow #1: From Veda Templates to Multi-Model Execution

  1. Start with Veda templates from a GitHub repository.
  2. After syncing the files, choose to browse the input as either OSeMOSYS or TIMES.
  3. Run the model in either framework.
  4. Compare outputs between the two models.
  5. If the same repo is shared with multiple users, they can view model results collaboratively.

Workflow #2: Starting from OSeMOSYS CSV Files

  1. Start with a folder of OSeMOSYS CSV input files.
  2. Add a configuration file for orienting the interface.
  3. Then, follow the same steps as Workflow #1—browse, run, and compare.

Key Discussion Points:

  • How a model-agnostic platform can improve accessibility and transparency in Openmod.
  • How structured input processing, synthetic input views, and scenario layering can enhance the user experience.
  • Whether Openmod members see value in structured, rule-based scenario management vs. script-based approaches.

Participants will be able to use their own files, or my samples, with these workflows during the session.

Looking forward to your views on this.

PS: I plan to have the first tier of this platform completely free for all users. I will have to impose some restrictions on resource utilization because I cover the cloud infrastructure costs myself.

  • background:
    Some links:

https://veda-documentation.readthedocs.io/en/latest/
Veda online is already free for pedagogical use: Veda online is free for students — Veda Online documentation

2 Likes

Just for fun, my crack at that same general problem in 2002 using UML:

  • Morrison, Robbie (6 September 2002). The energy-services supply systems (ESSS) common model initiative — Release A. Berlin, Germany: Technical University of Berlin. doi:10.5281/zenodo.6619604. Previously published on a now defunct website from that era.
1 Like

Thanks for the link Robbie. Yours is much more general and organized. I am simply trying to improve a feasible solution. One step at a time.

London‑based TransitionZero have a data workflow that supplies OSeMOSYS and PyPSA‑based models and other tooling from the same database. Perhaps you will have an opportunity to discuss TZ’s approach with staff at Stockholm? More here in the TZ mission statement — which also briefly covers efforts to automate scenario build and analysis practices (and note the print to PDF button):

  • Gray, Matt (23 January 2025). Our secret master plan. TransitionZero. London, United Kingdom. Access date 2025-02-28.
1 Like

Thanks for the link to TZ’s master plan… very interesting!

I’ve been developing KINESYS, a flexible global energy system modeling framework, with a vision quite similar to that of TZ - enabling rapid, collaborative scenario analysis without requiring any coding expertise or familiarity with TIMES. However, the route so far has been different:

  • While TZ is building an open platform with external funding, I have been working largely on my own as a modeller, supported by a small team of programmers.
  • Without external funding, I have relied on a service-based model, charging clients to sustain development.

That said, I’m interested in exploring other routes that could have a bigger impact - for example, replacing proprietary data sources with open ones and moving toward an open system that can serve a broader community.

KINESYS is already in action - see this success story on how it’s enabling rapid and collaborative modeling. For more details, check out the documentation.

KiNESYS instances are deployed on the same online platform that I am extending to support OSeMOSYS.

Hello,

At Reiner Lemoine Institut we developped a platform to create and simulate energy systems with oemof: open-plan-tool · GitHub (deployed online with open-plan.rl-institut.de). Over the years many people interested by the GUI contacted me and it led me to think to have a model agnotic platform is something we should push for in a larger group such as open-mod. As I wanted to propose this as a do-a-thon I came across your do-a-thon and I think I will just join yours :slight_smile: Looking forward!

Hi @akanudia,
we have been developing an open community database and methodology for scenario comparison for some years. It is based on the Open Energy Ontology (OEO) and a customized Knowledge Graph.
We are currently improving the workflow and will be implementing more functions as part of the NFDI4Energy project.

I would be happy to discuss the approaches and solutions, unfortunately I will not be able to travel to Stockholm for the openmod this year.

Scenario Bundles and comparison tool:
https://openenergyplatform.org/scenario-bundles/main

Documentation:

After looking closer into the proposed workshop, I think this should not be part of an openmod.
The presented veda tool is a commercial product and has no open-source license.
Providing a free academic version has no benefit for the openmod community.
Please discuss @willu47 @robbie.morrison @tom_brown

1 Like

@ludwig.huelk invited me to respond. For context, I am not involved in the organization of the Stockholm workshop, nor will I be attending. I do provide technical support in relation to the use of this forum.

I recently spoke to @akanudia by video on 27 February 2025 during his registration with the openmod forum. @ludwig.huelk is correct that Veda is closed source and operates under a freemium model where basic use is free (as in beer) and supported use requires payment. All users agree to non‑disclosure. This is to make the project financial viable. The development team is located in India. In contrast, most projects within the openmod community are variously supported via higher education, public good science, and/or philanthropic funding. With some project teams occasionally commissioned to provide paid analysis.

@ludwig.huelk is not the only person in this community to point out that Veda is not open source (although I cannot be more specific here).

Clearly, Veda does not fall under the rubric of the openmod. My view is that the closed nature of Veda is there to facilitate development and not to provide commercial returns. Perhaps @akanudia can comment on this?

It is important to recognize that the openmod does not endorse projects but rather offers a social umbrella. So the question is essentially whether the proposed breakout group is suitable for this upcoming workshop. Under established practices, that is solely a matter for the workshop organizers.

I would like to hear from @akanudia about future plans. It should be remembered that the development of the MARKAL and later TIMES frameworks were funded by governments as an international consortium. Back in the early‑1990s, the annual fee was USD $25k. The TIMES framework is now published under a copy‑left license, while a GAMS subscription is required to run the software, and, despite some attempts and exceptions, national datasets remain largely private.

If this is a journey towards openness for the broader TIMES community, I feel more relaxed about one breakout group. But that requires some assessment of intent — always difficult when the open‑source license is absent. Others in this community should chip in their views too. HTH, R.

Dear Robbie,

My unchanging driver for the past 30 years has been to build, use, and maintain models more efficiently. I started out with a set of scripts to manage input data and analyze results. At the time, ETSAP was a group of energy and modeling experts - not programmers. I evolved my scripts into Veda to support that group, enabling them to work with MARKAL and TIMES without needing to code.

Over the last five years, my focus has broadened beyond non-programmers to making models accessible to non-modelers as well. This shift led to Veda evolving into Veda Online (VO).

However, the whole energy system modeling domain has always been too small to sustain a dedicated software product. Veda tools are the result of 25 years of work by a team of 6 - 8 talented programmers, and even today, the software business barely breaks even. Till a few years ago, Veda was heavily subsidized by my modeling consulting business.

My Intentions Behind This Initiative

  1. Technical Thrill: I want to push the boundaries of what’s possible: supporting multiple models from the same structured input data and making it as easy to compare model outputs as it is to compare scenarios.
  2. Engaging Beyond TIMES: I see value in interacting with a broader community beyond the TIMES ecosystem and learning from open-source tools and processes.
  3. Lowering Barriers for Non-Programmers: Many people interested in energy system modeling don’t have programming skills. I want to create an option for them to engage with open-source models in a structured way.

General Observations

In my opinion, it would be sub-optimal for the open-source world and proprietary tools to operate in isolation. We still have a long way to go in translating modeling into real-world impact, and a more inclusive ecosystem - where different approaches interact - can only help.

xl2TIMES is a good example of this interaction. It has already started parsing Veda syntax using Python, which is a great step. But a full UI requires several additional components beyond syntax parsing. If nothing else, a fully functional VO supporting multiple models might help shape efforts like xl2TIMES into complete UI solutions.

Future Plans

I intend to keep a version of this platform entirely free for all open-source modelers. However, since I cover the cloud costs myself, I will need to set some resource limits on free access to ensure sustainability. I believe that scaling up options will appear if there is value in the platform.

I look forward to hearing what you and others think about this.

@akanudia The context you provided is valuable and will no doubt be useful to the organizers of the Stockholm workshop. And your reply is largely in line with our earlier discussion. I would also like to note my previous review of the TIMES ecosystem posted on this forum two years back — and to say that that review was written as an outsider. My video meeting with @akanudia two weeks ago was the first time I had had personal contact with the TIMES community. Thanks again for your thoughts. R

Thanks @ludwig.huelk for the intervention, and @robbie.morrison and @akanudia for the respectful and thoughtful discussion.

In short, our take as organisers of this year’s openmod is not to exclude anyone who wants to engage with the open-source community. We are happy to include Amit’s breakout into the programme and provide the space and opportunity to engage with the openmod community.

A broader reflection, perhaps with a short anecdote, if you will permit me!

Between 2009 and 2012, I worked as a Research Associate at the nascent UCL Energy Institute on the UK MARKAL model and the TIAM-UCL model. I attended several ETSAP meetings. In one meeting, I presented some insights which had used the internal workings of the Microsoft Access database that sat behind VEDA, much to the horror of some attendees! I had “hacked” the database, by guessing the password to it (or perhaps I found the password in the code). Needless to say, the database connection was secured in later iterations of the software. So the ETSAP group were well aware of the “benefits” of a community that probed and experimented with code, although this conflicted with intellectual property and commercial interests, which were designed into the way that IEA-ETSAP operated.

Anyhow, as @robbie.morrison points out, it is the funding model which often influences whether the code or data is open or closed. I do not think it is useful for openmod to exclude individuals or groups because they have worked within the incentive structure they find themselves. Of course, openmod began and continues with a spirit which rejects the status quo and strongly promotes openness of data, models and source code. I think this is good and should continue. But we should also recognise that achieving “full openness” is also difficult in many different contexts. Perhaps we can discuss this further in Stockholm.

I will also point out that within the ETSAP members, there exists a huge amount of energy system modelling experience, knowledge and skill. openmod members could learn from these individuals (including Amit), just as ETSAP could learn from the practices and experiences of openmod. Let’s start this exchange with the hope that it continues, and we can break down a few more silo walls…

It would be great to get Juha (Spine Toolbox developer [open-source]), Vedran (OSeMOSYS User Interface developer [open-source]), Charlie Heap (LEAP developer [freemium]), Transition Zero (Scenario builder [open-source, with paid commercial offering?]) and more in the room at the same time, for example, to discuss the challenges and concepts of user-interfaces for energy system models.

2 Likes