Update on open-source energy system modeling in the global south and including Africa
This is a modal window.
Das Video konnte nicht geladen werden, da entweder ein Server- oder Netzwerkfehler auftrat oder das Format nicht unterstützt wird.
Formale Metadaten
Titel |
| |
Serientitel | ||
Anzahl der Teile | 542 | |
Autor | ||
Lizenz | CC-Namensnennung 2.0 Belgien: Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen. | |
Identifikatoren | 10.5446/61445 (DOI) | |
Herausgeber | ||
Erscheinungsjahr | ||
Sprache |
Inhaltliche Metadaten
Fachgebiet | ||
Genre | ||
Abstract |
|
FOSDEM 2023333 / 542
2
5
10
14
15
16
22
24
27
29
31
36
43
48
56
63
74
78
83
87
89
95
96
99
104
106
107
117
119
121
122
125
126
128
130
132
134
135
136
141
143
146
148
152
155
157
159
161
165
166
168
170
173
176
180
181
185
191
194
196
197
198
199
206
207
209
210
211
212
216
219
220
227
228
229
231
232
233
236
250
252
256
258
260
263
264
267
271
273
275
276
278
282
286
292
293
298
299
300
302
312
316
321
322
324
339
341
342
343
344
351
352
354
355
356
357
359
369
370
372
373
376
378
379
380
382
383
387
390
394
395
401
405
406
410
411
413
415
416
421
426
430
437
438
440
441
443
444
445
446
448
449
450
451
458
464
468
472
475
476
479
481
493
494
498
499
502
509
513
516
517
520
522
524
525
531
534
535
537
538
541
00:00
Charakteristisches PolynomMathematische ModellierungSoftware Development KitMathematisches ModellLeistung <Physik>Deskriptive StatistikSystemprogrammierungMathematikSpieltheorieRadikal <Mathematik>Offene MengeDatenfeldPunktSelbst organisierendes SystemGrundraumQuick-SortTwitter <Softwareplattform>SensitivitätsanalyseArithmetische FolgeMengeEvoluteVollständigkeitInformationEntscheidungstheorieMaßerweiterungAuflösung <Mathematik>Globale OptimierungOperations ResearchMultiplikationsoperatorDynamisches SystemTypentheorieDifferenteMinimalgradOpen SourceInformationsspeicherungErneuerungstheorieDienst <Informatik>SoftwarePaarvergleichLineare OptimierungFunktionalanalysisNebenbedingungAnalysisProgrammierparadigmaWeb-SeiteGemeinsamer SpeicherBitrateDatenflussNichtlinearer OperatorARM <Computerarchitektur>HybridrechnerNatürliche ZahlOptimierungsproblemGüte der AnpassungSkriptspracheKugelkappeRahmenproblemSchlüsselverwaltungUngleichungMultiplikationMAPBAYESTeilbarkeitWeb SiteMereologieFrequenzRechter WinkelTesselationFlächeninhaltDreiecksfreier GraphResultanteMatrizenrechnungAggregatzustandComputeranimation
05:54
Mathematisches ModellSystemprogrammierungProzess <Informatik>Mathematische ModellierungIdentitätsverwaltungDesintegration <Mathematik>Thermodynamisches GleichgewichtSensitivitätsanalyseAuflösung <Mathematik>RandwertResiduumGerichtete MengeMotion CapturingWärmestrahlungQuellcodeAnalysisLösungsraumAlgorithmische ProgrammierungDatenanalyseVollständigkeitStandardabweichungOffene MengeFormale SemantikOntologie <Wissensverarbeitung>Derivation <Algebra>CodeLokales MinimumInformationRelation <Informatik>HorizontaleVerkehrsinformationDigitalsignalDatenbankFormale SemantikKanalkapazitätProdukt <Mathematik>CASE <Informatik>ÄhnlichkeitsgeometrieGesetz <Physik>Kategorie <Mathematik>MengeStützpunkt <Mathematik>KonditionszahlVirtuelle MaschineTrajektorie <Kinematik>Prozess <Informatik>InformationZahlenbereichDerivation <Algebra>CodeGeradePerspektiveWasserdampftafelMathematisches ModellPunktNatürliche ZahlResiduumIntegralWärmestrahlungSensitivitätsanalyseRandwertSystemprogrammierungMathematische ModellierungSignifikanztestInformationsqualitätStandardabweichungOptimierungsproblemAnalysisVektorpotenzialBitSoftwareMAPVollständiger VerbandErhaltungssatzProgrammierungMereologieZweiProgrammierumgebungARM <Computerarchitektur>VollständigkeitDifferenteLie-GruppeMobiles InternetQuellcodeResultanteOpen SourceGefangenendilemmaWeb SiteRechenschieberFlächeninhaltBenutzerfreundlichkeitLuenberger-BeobachterMinkowski-MetrikPeer-to-Peer-NetzMultiplikationsoperatorComputeranimationVorlesung/Konferenz
12:48
Formale SpracheProgrammierungMathematikCodeDokumentenserverSoftwareSpeicherabzugMathematische ModellierungSuite <Programmpaket>Formale GrammatikKontextbezogenes SystemMAPAnalysisVektorpotenzialSelbst organisierendes SystemKollaboration <Informatik>QuellcodeSelbstrepräsentationAuswahlaxiomMathematisches ModellSoftwarewartungInformationSensitivitätsanalyseInformationProzess <Informatik>DatenstrukturMomentenproblemOpen SourceFormale SpracheMereologieBoolesche AlgebraFaserbündelGlobale OptimierungGruppenoperationAnalytische MengeGrundraumCodeSelbst organisierendes SystemHyperbelverfahrenMathematisches ModellBitSoftwareentwicklerDokumentenserverProjektive EbeneMathematische ModellierungGebäude <Mathematik>SoftwarewartungRechenschieberPerspektiveSelbstrepräsentationRahmenproblemSoftwareWärmeübergangMengeAnalysisKontextbezogenes SystemMathematikEinfach zusammenhängender RaumSicherungskopieForcingTopologieOrtsoperatorVerkehrsinformationSpezielle unitäre GruppeARM <Computerarchitektur>Einfache GenauigkeitZellularer AutomatReelle ZahlPlastikkarteMAPFlächeninhaltSchlüsselverwaltungOrdnung <Mathematik>Gewicht <Ausgleichsrechnung>MultiplikationsoperatorSichtenkonzeptNeunOrdinalzahlRichtungDigitalisierungInnere EnergieComputeranimation
19:43
SoftwarewartungCodeInformationSensitivitätsanalyseQuellcodeRegulator <Mathematik>Schnitt <Mathematik>ARM <Computerarchitektur>SoundverarbeitungComputerspielGeradeMultiplikationsoperatorMinkowski-MetrikOffene MengeMathematisches ModellKategorie <Mathematik>Virtuelle MaschineMathematikPunktRichtungReelle ZahlUnrundheitZusammenhängender GraphMAPFokalpunktInformationGruppenoperationDatenbankSelbst organisierendes SystemDatenfeldFormale SpracheSichtenkonzeptVerkehrsinformationGrundsätze ordnungsmäßiger DatenverarbeitungHorizontalePublic-domain-SoftwareProgrammfehlerOpen SourceSpieltheorieWort <Informatik>AppletGewicht <Ausgleichsrechnung>Quick-SortRandomisierungSystemplattformGesetz <Mathematik>EinsWeg <Topologie>Computeranimation
26:45
PunktwolkeFlussdiagramm
Transkript: Englisch(automatisch erzeugt)
00:07
Okay, my name is Robbie Morrison and I'm here to talk about energy system modeling. I want to take you right up to the stratosphere. A couple of things on my background, I won't go through all this, but I started climate
00:25
campaigning 33 years ago. I started high resolution national energy system modeling 23 years ago, sorry 28 years ago, and I started open source energy system modeling 20 years ago.
00:42
So I was right at the beginning of those trends pretty much. I want to talk briefly about the Open Energy Modeling Initiative, which started about eight years ago, and it's an informal collection of modelers, and we now have about 1,000 people
01:05
involved. The bulk of them are early-stage full-time researchers, and that gives you an idea of how much sort of interest there is in this open side. There is an entire parallel universe doing closed modeling that we don't have much contact
01:22
with in the power companies, in the World Bank, in the multilateral organizations. So I'm only going to talk about the open source side. And the final point up here is that the, I just clicked this to gas that, the final point is that this whole field has flipped in the last year, radically.
01:46
I get contacted by corporations and economists and so forth now, which would never have happened two years ago. So this is a complete game change. I'm not going to talk very much about energy system modeling, but if you want an
02:03
introduction I recommend this YouTube, which is made with my partner in a car park, and it's descriptive and it's quite good. This is a quick schematic showing what these models can capture. This just happens to be one that I pulled up that's hybrid with agent-based modeling
02:23
in it. But you see a lot of the entities, if you like, that were being discussed in the previous talks, but brought together in a collective. So we have households, and we have market operators, and we have lines companies, and
02:40
we have markets, and we have AC power flow, and we have a lot of kit in the system, hydro systems, storage, gas turbine sets, and so forth, and a whole lot of external characteristics coming in through weather conditions, interest rates, and so forth.
03:03
So that sort of gives you the broad picture. If you want to look at the models that exist, this Wikipedia page is worthwhile. It's about half complete, and it covers the various models. Some are directed specifically to the energy sector, but increasingly they're sector coupled
03:24
and they come into the whole energy system. The basic paradigm is operations research. So the underlying model produces a set of constraints in a sparse matrix, has a goal
03:43
function which is normally minimum aggregate cost, and feeds that all into a solver and returns a result. The way that the analysis proceeds is by so-called comparative analysis of scenarios. So you pick a base scenario, a reference scenario, and then you propose different scenarios
04:08
that you want to explore with nuclear, without nuclear, and so on, and so on. These are the high resolution. They have a lot of detail in them, so they have the plant and the network and so forth
04:24
in them. A lot of external circumstances, weather, demand for energy services and so forth. They are contiguous time, which is really important nowadays because with renewables and storage you can't kind of do typical periods, you actually have to work your way
04:41
through the entire system as it evolves. That evolution might be out for 30 years, out to 2050. There's a degree of different types of foresight. Sometimes it's perfect foresight, so we know everything about the future. Other times it's stepwise, so we do recursive dynamics.
05:07
Not up here, technological progress is included. One factor, multi-factor, for example, the uptake of a particular technology like solar PV, the model will internally reduce the costs for that particular technology as
05:27
it's taken up and evolves through time. The optimization is usually mixed integer linear programming. Anything else more exotic runs into performance issues.
05:41
Conceptual extensions include embedded decision taking using agency, multi-criteria optimization, some assessment of co-benefits such as urban air quality, sensitivity to the framing of the problem, the role of uncertainty and the exploration of near optimal solutions.
06:04
This is system modeling. All systems have kind of natural systems and problems, if you like, together have natural boundaries. If you want to model Europe, or you want to model an energy system in Germany, you
06:22
probably want to go to the boundaries of Europe, for example, because that's the kind of a natural point. The methods naturally seek technical synergies. That's one of the advantages of using these systems. The least cost approach will pick up the synergies and get them working.
06:46
Future climate change is normally included, projected future climate change. These models may exhibit undue sensitivity to both data quality and to system resolution, so they're not without issues that have to be explored by modelers.
07:08
They started off with energy systems, electricity systems coupled into district heating and the gas and so forth, but they're increasingly branching out into land usage, water use,
07:25
the industrial sector, when you were looking at things like hydrogen, ammonia, thermal integration and steel production. Carbon capture is included now, outside of the energy system, so residual emissions
07:40
from cement and from agriculture are now being included in these models. E-mobility also, and we've had some talks about vehicle charging, but this is to look at the whole picture and not just the perspective of the householder or even the lines company. Co-benefits beyond climate change mitigation I mentioned.
08:04
What isn't in the models is there is no embedded economy. If you want to do that, then you have to go to process-based integrated assessment models, which are widely used by the IPCC, in which case you have a lot more of an economic
08:24
take on the system. The model started off being open source, but there are good reasons why we want to look beyond the open source, and the first reason is to go to open science.
08:43
So we want genuinely open data, and we want it under communal curation. We want full transparency, and we want it, as modellers, we want an engaged overarching community so that we can compare and contribute and support each other.
09:04
The goal in my take is that we should be looking at public policy analysis, which is based on peer production, on commons-based peer production. The reason I say that, and I think there was a talk earlier this morning from the
09:24
European Commission, people like the European Commission do not have the capacity to explore the solution space, and I will add, nor do they have the creativity required. That's not a criticism, that's just an observation.
09:40
So we really want a massive effort in exploring what our future could be out to 2050, the kind of trajectories and pathways and requirements that are needed. Some potential for public engagement, but very few examples to date when these models are
10:01
used for more specific projects. Our biggest Achilles heel is complete and coherent data for public interest analysis. We are not data scientists, we are desperate to have data which is complete and coherent.
10:20
If it's dirty, it's a problem. If the semantics behind the data collection is somewhat inconsistent, it's a problem. If the information is missing, it's a problem. This may not be an issue for our data scientists using statistical techniques or
10:40
machine learning, but it is for us. One issue that doesn't get much airplay are data standards, and quite a lot of the data standards in this area, especially in the electricity sector, are proprietary. They come under so-called FRAND, we heard about that, fair, reasonable and
11:01
non-discriminatory conditions. The problem is if the data standards are legally incumbent, then the code bases that reflect that and the data sets that comply with it could become derivative works under
11:21
intellectual property law and we are in trouble. So we want basically CC by 4.0 or something similar on the data standards. I'll skip the last bit on data sets actually. And I'll skip this slide, but I just want to point out that the situation in
11:43
Europe is pretty awful on a number of levels. Go to the US and you'll find a much friendlier environment for this kind of public interest information. Okay, second part of my talk is about the global south, and the question is why
12:01
is someone who's white, male and old standing here talking about the global south? And my short answer is I'm from Aotearoa, New Zealand, and New Zealand became bicultural over my lifetime, and I saw that process and contributed to it. I had radio programs in the early 90s on sustainability and conservation on
12:24
tribal radio, on eevee radio, and so forth. I went to land occupations. I organized joint meetings with tribes, Hui they're called, and they take place on Marae. So that's kind of my backstory about why I can talk about this, I think.
12:42
This is a map of Africa with the high voltage network present, and you will see that there is very little structure there. South Africa, a little more. David is going to talk a little more about this, so I won't.
13:03
This is another example of a model called osmosis. This is in Africa, and these are the cumulative trades out for the next 30 years. So this is the kind of thing that the models are starting to look at.
13:20
There are two overarching projects in this area, the osmosis global project. Osmosis is written in a high-level mathematical programming language called MathProg, and the second one is pipes and meets earth, which is written in Python, and you'll hear a little bit more about. One of the interesting things I thought, I looked up,
13:41
Software Heritage collects the forks for a particular code base, and lists 135 fork repositories for osmosis and 308 fork repositories for pipes. So that gives you an idea of how the open source world works
14:02
when people will fork the project. These aren't hostile forks, I presume, and use them for their own work, and hopefully contribute their contributions back upstream. There's clear activity now in Central America, Costa Rica, South America,
14:22
countries like Brazil, India, and surrounding regions, South Africa, Sub-Saharan Africa, and most of this is in the context of academic work, we have no connection, or very little connection, crossing over with the multilateral development organisations and so forth.
14:43
So this is the parallel universe I mentioned. How are we doing for time? How am I doing for time? Sorry. Ten minutes, okay. One of the issues that we face, I think,
15:01
is interacting with official agencies, because we are relatively informal and relatively self-directed. And we also are a competition against, you know, the agencies like the International Atomic Energy Agency, or IRENA, or whoever are doing their own analysis.
15:22
And I quite like this quote from Oliver Geddon, everyday politics is therefore dominated not by evidence-based policymaking, but by attempts at policy-based evidence-making, and that's exactly what we want to avoid. I talked to the incumbent NGOs about using our kind of analysis,
15:42
and they weren't very interested, but I feel quite encouraged now, because there are a new set of foundation-backed think tanks who are actually very keen on this kind of stuff. And I'm sorry I can't mention too many names, because I was ill for the two weeks prior to this talk, and I didn't get consent to talk about them. But a couple are climate analytics and transition zero.
16:04
Some official agencies are starting to talk about open sourcing their stuff, but they're not doing it in a particularly robust way, in my opinion. And this is a problem. They will either open WASH, or they will do what's called throw their code over the wall,
16:21
which is put it on GitHub, but there's no attempt to develop it. There are no issues listed, and whether it even runs is open to question. In regards working with the global south, and I had about 10 interviews with researchers in the global south
16:42
to try and find out what scopes and issues, the unstructured interviews, but it was kind of interesting. So the clear benefits of open source projects, of course, few cost barriers, with the caveat that the commercial solvers can be expensive.
17:04
One license for Gurobi might equal three full-time researchers in India, for example. There's a soft technology transfer. It's bi-directional. It's lightweight.
17:21
All the software projects, bundle associated communities, and this is, I think, really a useful part. And the work is transparent. It can be studied and challenged, which I think is really important. There are some cross-cultural considerations, I think, that are necessary to explore.
17:43
And I talk about this in Aotearoa, New Zealand, becoming bi-cultural. But indigenous languages bundle different concepts. And they're quite noticeably different. Sovereignty is an issue. It's really easy to transgress sovereignty
18:02
without realising it. There's a question of representation. The projects are all pretty much white and male in the global north at the moment.
18:20
And the next question really is also a matter to be traversed, is that the framing of the models and the problems from a global north perspective may not be very appropriate to the circumstances in the global south. Final slide.
18:41
Challenges, just overarching challenges. Most of these won't be very surprising. Code maintenance is always a challenge. Maintainers, support for maintainers. Building a suitable knowledge commons is going to be a real challenge. For instance, the International Energy Agency
19:01
only sells its data under non-disclosure. We don't get hold of that, although it's collected from our national governments. The European Union is focused on data commodification through its single digital market. The scientific institutions are unnecessarily protective.
19:21
I talked about cross-cultural issues. We need to find new ways of interacting with official agencies to get any of this information into the policy process. And I'll just conclude with a quotation from an East German playwright, Heine Müller, that optimism is just a lack of information.
19:43
Okay, that's it. Thank you. Yeah, any questions? Can you speak up a little too, if you ask questions?
20:01
You said the European Union has some issues with open data. I know that the European Space Agency has really strong footprint on doing all this, or is it Sentinel data stuff, maximum open to drive a new economy? So has this not spread to the other agencies yet?
20:20
No, the ones I'm going to mention, and I will mention some names, the Mera data for future climate is under a bespoke license. The YASA data on scenarios going forward, also under a bespoke license, and so on.
20:41
So a lot of the Horizon 2020 projects are also problematic. The stuff under statutory reporting is also legally encumbered. So I can't, for the life of me, understand why, but some of it's technically encumbered.
21:00
So for example, the transparency platform run by ENSO-E is legally encumbered. The EEX data from the European Energy Exchange also, and also technically encumbered. You can't scrape it, you can't cut and paste it off the website.
21:21
It's not very deep protection. And we've complained, my friends, to ASA, the regulator, and they say it's compliant. Sorry, yeah? So now I'm just publishing air quality data
21:42
under random crappy licenses where they can do stupid things retrospectively. We dissuade them to publish all under a UK government license. Another case- where it's the open government license, UK 3.0. Yeah. And the other decent experience was with Elexon, UK, bouncing where I, well,
22:01
I think I'm their only official licensee, but other than the fact that they tell me to retract everything, I can use it all the way over, which is quite nice. So, you know, it can happen sometimes. I just want to comment on licensing. The one that, really the only license that works is CC by 4.0. If you go to the open government license,
22:21
UK 3.0, you will find it's not interoperable with Creative Commons. And so you end up with legal data silos. But still that's what it is. I know all the licenses are written by lawyers. I can assure you that,
22:40
and the lawyers all know what they're doing. I don't think you saw some of the things they were publishing. Okay, okay, okay. There's a question out there, or? No. Yeah, yeah, yeah.
23:02
Remind went open, that's from PIC, went to one of the high GPL licenses. I filed a bug report on that because the GPL licenses have a clause on them. Remember when Java was proprietary?
23:22
And you have to have an open language for a GPL license. They use GAMS, which is not an open language. And I filed a bug report. And I know personally the lawyer who responded, who said it was okay. Now look, I'm not an open source lawyer.
23:40
I didn't write the textbook. But that was where that discussion went. Have you seen any new funding come into this particular field to open things up more? Because all I know is that in December, I know that the Creative Commons group, they've started to hire new roles in this specific role because they landed like a small millions of euros grant
24:03
for this. But beyond that, I don't know if there's, if you know any other groups starting to do stuff in this field. Um, the overarching, okay, okay, yeah, finished. Thank you. Oh, sorry. The question was funding for specifically for open source.
24:24
And the sort of short answer is, hang on, the short answer is that the funding, I'm talking about Germany, let's say, has been quite good for modelling in general. And it hasn't been specifically directed to open source. The high level organisation,
24:41
the Open Energy Modelling Initiative, hasn't needed resources as yet. But what will happen going forward, I don't know. But the funders are interested in the kind of open science component of what we do. That's quite clear. And I presume that the next rounds of funding
25:00
will start looking for real open source projects to be, to be for support. Yeah. Has to change. And how can we push for the change so that we get these open data? So what are the levers we have to pull?
25:24
Um, in a particular, well, the question was, how, what levers are needed to come to genuinely open data. It depends on the jurisdiction. In the US, it's quite good. Federal, work by federal employees is public domain.
25:42
And there's been enough copyright, litigation around copyright, that most of the stuff isn't actually covered, protected by copyright. They don't have a database directive. Tracking back to Europe, the only solution I can see is CC by 4.0 as a policy,
26:02
which is lightweight, doesn't require legislative change and so forth. But it does require the European Union to get out of the data commodification. And I didn't mention it, but there's a thing called the Data Producers Act, which is still live, which might come back into the Data Act,
26:22
the proposed Data Act. And that would have been a complete travesty for us because that would mean all this machine generated data would now have its own intellectual property. And I couldn't think of anything worse. Okay, yep, thank you everyone.