2nd Online Lightning Talk Mini-workshop

Starting a new forum topic to organize the second workshop. Similar format to the first workshop.

When: Wednesday 15 April 2020 for 2 hours starting as follows (please double check your entry):

Local time UTC offset Location Comment
1500 UTC +0000
0500 HST −1000 Hawaii
0800 PDT −0800 California, Vancouver
1100 EDT −0500 New York
1600 BST +0100 London
1600 IST +0100 Dublin Irish Standard Time
1700 CEST +0200 Berlin, Paris
1700 SAST +0200 Johannesburg
2030 IST +0530 New Delhi Indian Standard Time
2300 CST +0800 Beijing
2300 SGT +0800 Singapore
0000 JST +0900 Tokyo
0100 AEST +1000 Sydney
0300 NZST +1200 Wellington

What: 10 slots for 6 minute talks + 4 minutes questions/comments. Topic is broadly “open models and data for energy modelling”.
Who: Anyone can join to listen (up to 300 participants). We’re limiting it to 10 presentation slots to keep it a reasonable length.
How to suggest a talk: Use the edit button (bottom right) to add your talk to the list.
How to connect: Use Zoom (works for Windozzz, Mac, GNU/Linux, please download, install and test before the workshop ), apologies that it’s not free software, but it works very well for group calls. Technical zoom hosting will be provided by the Energy and Climate Research Network (ECRN) at Dublin City University, with support from the Science Foundation Ireland (SFI) Insight Centre for Data Analytics. Private page with meeting details - unfortunately due to recent incidents of zoom-bombing we have decided to make the link private - you will need to register as a forum participant to get the link and to attend the meeting.
Talk format: Zoom allows you to share your screen with other participants, so you can share your talk slides. We’d appreciate it if you made the slides available beforehand with an open licence like Creative Commons Attribution (CC BY 4.0) .

Schedule (UTC, list of talks below)

1500 Start
1500 Introductions
1510 Talks start
1650 Talks finish
1650 Feedback/Open Forum
1700 Official finish
1730 Final finish

Talks:

Talks listed in order of presentation.

  1. Taco Niet. Reviewing combined modelling Approaches for meeting the UN Sustainable Development Goals The integration of different models through hard- and soft-linking has both advantages and disadvantages but, to date, no review of the strengths and weaknesses of combined models has been published in the literature. SFU’s School of Sustainable Energy Engineering has recently received a grant to review the literature on these combined modelling approaches. The project aims to bring together modellers to discuss the challenges, strengths and weaknesses of combined models and to publish a review article of the findings that will provide guidance to modellers when considering combined modeling approaches.

  2. Daniel Huppmann A common nomenclature for assessing low-carbon transition pathways in Europe The Horizon 2020 project openENTRANCE aims to develop an integrated modelling platform for assessing low-carbon transition pathways in Europe - but integrating multiple frameworks across spatial scales and sectoral dimensions requires a common understanding of terms and definitions. This talk will present ongoing efforts to develop such a nomenclature in an open & transparent manner. Check out the GitHub repository for details.

  3. Zoltan Nagy or Jose Vazquez-Canteli CityLearn: An OpenAI Gym Environment for MultiAgent Reinforcement Learning and Demand Response In brief, CityLearn deals with controlling heat pump operation and battery/thermal storage charge/discharge using RL in multiple buildings (centralized & decetralized) to study building-grid interaction. We have abstracted out the building side so one can focus on algorithm development only, and hence develop benchmarks and comparisons, in particular for multi agent systems. Check out the GitHub Repository for details.

  4. Stefano Moret & Gauthier Limpens. EnergyScope: a novel open-source model for regional energy systems. EnergyScope optimises the investment and operating strategy of a multi-sector multi-vector energy system (including electricity, heating and mobility) for a target year. The linear programming model has an hourly resolution (using typical days) which makes it suitable for the integration of intermittent renewables, and its concise mathematical formulation and computational efficiency are appropriate for quick scenario assessments as well as uncertainty studies. Check out the GitHub Repository and the paper for details. Our slides are available here.

  5. Oleg Lugovoy. USENSYS development & Open Decarbonization updates.

    • United States Energy System (USENSYS) is capacity expansion model with primary focus on renewables and energy transition. USENSYS is an open source capacity expansion model, based on energyRt package for R. The current state of the model covers electric power sector and has 49 regions (48 lower states and Washington DC), and two time-resolution versions:
      – renewables balancing version with 1 year and 8760 hours, 49 regions;
      – electric power system transition version with 1-300 sub-annual slices and 50-100 years of horizon.
    • Open Decarbonization – an open energy modeling initiative with the purpose to develop a knowledge platform of sharable tools for reproducible decarbonization analyses, accelerate dissemination and implementation of the tools, and develop low carbon energy future scenarios available for free public use and open discussion.
      The project aims to connect modelers/researchers who has developed models and decarbonization scenarios with modelers/researchers who want to recalibrate the models for another country/region/sector, reproduce and build on the analysis, contribute low decarbonization scenarios to the public domain.
  6. Diederik Coppitters Uncertainty quantification and robust design optimization framework for hybrid renewable energy systems. The framework aims for computationally-efficient robust design optimization of hybrid renewable energy systems, including the quantification of uncertainty related to lack of data (epistemic uncertainty) and uncertainty related to natural variation (aleatory uncertainty). The framework provides designs with optimized average techno-economic performance, as well as designs which are least-sensitive to real-world uncertainties. Additionally, the dominating stochastic input parameters on the performance variation are provided through sensitivity indices, while the effect of epistemic uncertainty on these sensitivity indices indicates their confidence level.

  7. Kamaria Kuling Comparison of Two Different Equations for Modelling Energy Storage with OSeMOSYS A comparison of different methods of modelling energy storage that have been used in the open source capacity expansion modelling system OSeMOSYS. Variable renewable energy sources (VREs) such as wind and solar provide a low carbon alternative to meet our energy demands, but one drawback of such technologies is their dependence on weather cycles. Energy storage is one solution to allow energy demands to be met while using VREs. Consequently, it is paramount when making decisions regarding future energy use and incorporating more VREs into our energy infrastructure to be able to include storage in an energy model, and to have trustworthy and reliable methods to do so. This presentation will compare two storage equations formulated by Welsh et al. and Niet for OSeMOSYS and their effect on model outputs and performance.

  8. Leonard Göke. anyMOD - A framework for energy system modelling with high levels of renewables and sectoral integration. anyMOD.jl provides a framework to generate large-scale energy system models, that account for close interdependencies between sectors and the variable nature of wind and solar. Drawing on basic concepts of graph theory, this is achieved by two novel features. Frist, the level of temporal and spatial detail can be varied by energy carrier. Second, context-dependent substitution of energy carriers can be modelled. Check out the repository here.

  9. Mihir Desu. Decarbonizing the Energy System: The impact and cost of using fragmented grid planning models to guide investment decisions. Converting opaque and black box capacity expansion and production cost models used for resource investment planning decisions into open and transparent tools.

  10. Adriaan Hilbers. Renewable test case power system models
    We discuss a set of simple renewable power system models for benchmarking exercises for time series and optimisation methods. In many fields, standard benchmarks exist; notable examples are MNIST or CIFAR in Computer Vision and the Lorenz 63 system in Dynamical Systems. Test models used in power system research tend to differ per investigation, with each paper using a different (often not open-source) model. This repository provides a few simple test models to fill this gap. The models can be run “off-the-shelf”, containing pre-determined topologies, technologies and time series data. All that needs to be specified is the subset of time series data to use and a number of switches (e.g. integer or ramping constraints, whether to allow unmet demand) that ensure the model can contain most features seen in more complicated systems. These models are not modelling frameworks like OseMOSYS or Calliope (which can be used to create arbitrary power system models, but are not models themselves). The models are built and can run in Python. Documentation and examples can be found in the Github repository.

Format for each talk

  • The speaker will be moderated into the meeting by the timekeeper
  • The speaker can share their slides or screen by hovering over the main screen Zoom and clicking “Share” and choosing what they would like to share. Please remember to unmute your microphone and introduce yourself briefly.
  • They can speak for 6 minutes. All other participants will be muted by the moderators during this time.
  • After 5 minutes they will received a “1 minute warning” on audio from the timekeeper
  • After 6 minutes they will be asked to stop talking by the timekeeper
  • If they are still speaking after 6.5 minutes, the timekeeper will mute them (sorry).
  • To ask questions to the speaker, write your question in the public “Chat”.
  • The question moderator will group similar questions and ask the questioner to put their question over audio to the speaker.
  • After 9.5 minutes the timekeeper will ask everyone to start wrapping up and prepare for the next speaker.
  • If you have further questions for the speaker, please contact them privately or use the open forum at the end of the session.

Timekeepers/Moderators

@bmcm will do Zoom hosting duties (managing who’s (un)muted etc).
@dnock (Destenie Nock), @johannes.hampp and @tniet (Taco Niet) will co-host, time keep & moderate the questions after each talk:

Recording

We’d like to record the entire session and, for people who give their consent, make the recordings available after the session under a Creative Commons Attribution (CC BY 4.0) to those who were not able to make the meeting. We will try to make a registration page when you log into the meeting where you can give your consent or not. We will not publish anything without the consent of those being recorded. You may withdraw your consent afterwards as well. We will respect the wishes of anyone who asks to delete the recording of them during times when they were talking. We remind all users that the meeting was advertised on a public forum and we cannot stop any participants recording the meeting with screen capture. Please contact @robbie.morrison with questions about the recording.

Rules

Since we may be more than 50 people, we have to enforce some rules to respect everyone’s time and attention.

  • We will keep military time and discipline.
  • Download, install and test Zoom before the workshop .
  • Use video if you can.
  • Use a stable internet connection.
  • Don’t talk unless invited to by one of the moderators.
  • Use a headset if you’re talking.
  • When you’re not talking, mute your microphone.
  • If you call with a hurricane in the background, we will mute you.
  • Ask questions in the public “Chat”.
4 Likes

We should correct the translation of UTC to individual time zones because of daylight savings starting this weekend.

Hi @tom_brown - times updated. Thanks for catching that!

Is registration needed as a participant?

@MohamedH The logistics have mostly been discussed on the openmod mailing list.

Yes, you will need to register as a participant in the mini‑workshop. You will need to supply your name and email address. And you will also be asked to optionally consent to having any contributions you make being distributed under a Creative Commons attribution license.

The URL for the mini‑workshop will be advised in this topic 24 hours prior to the Zoom event. That is an attempt to make it more difficult for random people to search for the URL and then prank the meeting.

1 Like

@olugovoy - I see that you’ve suggested two talks for this session. In the interests of giving everyone a chance to present, can you either choose one talk or combine your two talks into one?

Thanks,
Taco…

@tniet Hey Taco, I’ve signed up for the last slot. If this doesn’t work, let me know!

@tniet - sure! I will squeeze to one.

@olugovoy - Thanks! ‘See’ you on Wednesday.

My I file one request? Can we use other tools as Zoom for future events? There are some serious security discussion around Zoom. We are therefore not allowed to use Zoom and are excluded from such events. I think this will be similar for other organisations.

Sorry that you are excluded by the use of Zoom. You are the first person to complain about this.

The problem is that there is no other good service that is platform-independent and scales to 150+ participants on five continents. If you know of one, please suggest one. My experience with Jitsi the last few weeks has not been good (sound and video problems, long time lags for screen sharing). DFN has had scaling problems and is anyway not friendly to non-Germans. Skype you can forget.

Zoom are also addressing security concerns so perhaps this issue will resolve itself.

Hi @carstenhoyer The downsides of Zoom are widely discussed, for instance Paul (2 April 2020) on vendor abuse and security and Oliver (2020) on administrator access. Recent articles on alternatives to Zoom include Farough (3 April 2020) and Paul (9 April 2020).

The most direct open source equivalent is Jitsi as already pointed out. But Jitsi does not scale to the numbers we are encountering, namely around 200 participants. Indeed Paul (9 April 2020) suggests an upper bound of 75 participants and ideally less than 35. Jami does offer some video-conferencing functionality but is more akin to Skype.

The are other free‑to‑use proprietary products aside from Zoom, including BlueJeans (Paul 9 April 2020).

So I think we are stuck with Zoom? DCU kindly host our mini‑workshops with an educational license covering a maximum of 300 participants. Fortunately we are not approaching that constraint at the moment.

I should note that privacy is not such an issue in our use case but collecting personal information and information abuse are.

Note that the LibrePlanet 2020 conference hosted by the Free Software Foundation ran remotely and entirely on free software this year (Farough 14 March 2020). But that event was largely a streaming exercise.

@tom_brown I would have complained about Zoom too had I not been complicit in its usage. :neutral_face:

References

Farough, Greg (14 March 2020). LibrePlanet day 1: Can free software carry an entire online conference? Yes, it can!. Free Software Foundation (FSF). Boston, Massachusetts, USA.

Farough, Greg (3 April 2020). Better than Zoom: try these free software tools for staying in touch. Free Software Foundation (FSF). Boston, Massachusetts, USA.

HowtoForge (2018). How to create your own video conference server using Jitsi meet on Ubuntu 18.04 LTS. HowtoForge.

Oliver, Lindsay (19 March 2020). What you should know about online tools during the COVID-19 crisis. Electronic Frontier Foundation. San Francisco, California, USA.

Paul, Kari (2 April 2020). “‘Zoom is malware’: why experts worry about the video conferencing platform”. The Guardian. London, United Kingdom. ISSN 0261-3077.

Paul, Kari (9 April 2020). “Worried about zoom’s privacy problems? A guide to your video-conferencing options”. The Guardian. London, United Kingdom. ISSN 0261-3077.

Dear all -

Just to confirm that the second 2nd Openmod Online Lightning Talk Mini-workshop is due to kick off today at 15:00 UTC (just under two hours time). You do need to do a zoom registration for the event: check the details (including the zoom link) on this private openmod forum page. You do need to already be a registered openmod forum user to access this page. Scroll down to find the message with the unobfuscated registration URL (only posted as of yesterday).

Kind regards - @bmcm

1 Like

Regarding video-conferencing software, BigBlueButton apparently scales better than Jitsi, given a sufficiently well resourced hosting environment. Setting up the server is, by all accounts, not trivial. BigBlueButton scales to 50 active participants for meetings and to 20 for panel discussions with passive audiences in the hundreds. It seems this video-conferencing software is the one to watch. Insiders think BigBlueButton will be adopted by FSF going forward.

1 Like

The ACM recently published an extensive report (Lopes et al 2020) on virtual conferencing.

The ACM suggest Zoom + Slack for conferences with fewer than 200 participants. For the record, the free and open source equivalent of Slack is Mattermost and that software works perfectly well.

References

ACM (2020). Virtual conferences: concrete advice to turn your event into a state-of-the-art online gathering. Association for Computing Machinery (ACM). New York, USA. Landing page for report.

Lopes, Crista Videira, Jeanna Matthews, Benjamin Pierce, Blair MacIntyre, Gary Olson, Rob Lindeman, Francois Guimbretiere, Srinivasan Keshav, and Vicki Hanson (13 April 2020). Virtual conferences: a guide to best practices — Version 1.1. New York, USA: Association for Computing Machinery (ACM).

1 Like

Workshop attendance

Participation rates for the second virtual workshop are overall about half that of the first workshop. Compare here. One explanation could be the need to register with this forum to gain access. Going forward, if we do decide to relax this gate‑keeping requirement, then we should probably lock down the participants and only take questions by chat entry instead of allowing live responses and exchanges.

  • unique participants : 81
  • attended for 20 minutes or longer : 81
  • cumulative non‑overlapping participation : 142 hours
  • registration TLDs : at  be  ca  ch  com  de  dk  earth  edu  es  fi  fr  ie  kz  pl  ru  se  uk

Participation bar graph:

openmod-mini-workshop-02

Technical notes: Multiple potentially overlapping sessions by the same participant are now aggregated using first join and last leave timestamps. All sessions, including those less than one minute in duration, are plotted. Deduplication is based on email addresses rather than stated names because repeat attempts to attend sometimes resulted in truncated names being entered. Tools: python, pandas, matplotlib.

DAT (no copyright), SVG (CC‑BY‑4.0 license), python3 (ISC license) files (no personal data):

Edit: the python script to create these plots is now on GitHub:

@robbie.morrison Thanks for sharing this! One question: as briefly discussed at the end of the meeting, will the recording be made available?

@stefanomoret The short answer is yes.

But it’s a little more complicated. Distribution of the video content is difficult when the presentations contained lack Creative Commons CC‑BY‑4.0 licenses. In addition, the audience contributions present in each video need to be checked against the registration consents file from Zoom — a few participants elected “no” so it cannot assumed everybody can be automatically included.

The best solution, in any case, is that the presentations be licensed under Creative Commons CC‑BY‑4.0.

If this is not possible, the alternatives are legally fraught. I have asked open source lawyers about the general copyright embedded in webinar recordings and that question is far from straightforward. In Europe, the GDPR personal data protection law applies and the videos clearly contain personally identifiable information (PII). And in Germany, personality rights (KunstUrhG) apply, which also requires consent to distribute images or recordings of private individuals. The two latter doctrines would require seeking new consent in the absence of a CC‑BY‑4.0 license in the video.

So again, the best solution is to get CC‑BY‑4.0 licenses on the presentations. Otherwise potentially I have dozens of emails to send and monitor to obtain appropriate permissions.

Also worth reminding everyone that the “open” in Open Energy Modelling Initiative refers to code, data, and content licensing. This community is not a general energy analysis community — for that there are other alternatives, including EMF, EMP, EnR, WEC, and Forschungsnetzwerke Energie.

Individuals can essentially upload whatever they like to social media but the workshop organizers are honor bound to respect both individual rights and intellectual property rights.

Finally to note that the openmod is looking for someone to run a Nextcloud instance. If hosting fees are an issue, these can easily be covered by the community — there are offers already.

For completeness, here is a screenshot of the optional consent question upon registration. This consent is sought anew for each mini‑workshop. The first two workshops provided the same question, but the wording can be modified for future workshops.

consent.screenshot-20200326t220617z