Talk:Cost of electricity by source: Difference between revisions
ClueBot III (talk | contribs) m Archiving 1 discussion to Talk:Cost of electricity by source/Archive 2. (BOT) |
Sampenrose (talk | contribs) |
||
Line 199: | Line 199: | ||
:We have a summary section already, that's what the lead is. I suggest you add a summary of data to it. [[User:GliderMaven|GliderMaven]] ([[User_talk:GliderMaven|talk]]) 11:10, 3 April 2021 (UTC) |
:We have a summary section already, that's what the lead is. I suggest you add a summary of data to it. [[User:GliderMaven|GliderMaven]] ([[User_talk:GliderMaven|talk]]) 11:10, 3 April 2021 (UTC) |
||
:: Please seek consensus. |
Revision as of 15:00, 3 April 2021
This is the talk page for discussing improvements to the Cost of electricity by source article. This is not a forum for general discussion of the article's subject. |
Article policies
|
Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL |
Archives: Index, 1, 2Auto-archiving period: 83.5 days ![]() |
![]() | This article is of interest to the following WikiProjects: | ||||||||||||||||||||||||||
|
To-do list is empty: remove {{To do}} tag or click on edit to add an item. |
![]() Archives (Index) |
This page is archived by ClueBot III.
|
Can we remove the EIA 1st image?
Hey all, I updated the EIA tables (both of them) to add the new 2020 EIA AEO LCOE data. In the beginning of the EIA section, we have an image which is just a visual representation of the table that immediately follows. I updated the image with the 2020 data as well. But I don't think the image really adds anything to the section since its just a repeat of the table below.
Is anyone opposed to me removing the image?
--Keanwood (talk) 01:17, 10 May 2020 (UTC)
- Ok for me. It would be helpful to have a graphic representation of the numbers (without unnecessarily repeating the numbers) but that figure is of poor quality. --Ita140188 (talk) 01:40, 10 May 2020 (UTC)
Dubious cost of nuclear energy
The first graph of Global studies states the cost of nuclear to be $164/MWh. In France, ARENH forces EDF to sell nuclear power to other electricity companies at €42/MWh. An independent study suggested a new price of €48 to reflect the real cost where EDF itself said it costs €52. How can Lazard be so off, even in its 129-198 USD range? Should we add a warning to users? Uzinagaz (talk) 11:31, 16 March 2021 (UTC)
- I think it's the difference between new build nuclear and old build nuclear. The French fleet was built decades ago, and inflation and asset depreciation has reduced the cost of the electricity in today's prices. A similar thing happens with coal plants. GliderMaven (talk) 14:29, 16 March 2021 (UTC)
- China EPR price tag was 7.5 billion USD, roughly 4,540 USD/kW of installed capacity (Wind is currently at roughly 1,000 USD/kW but has a much lower load factor and lifespan). With a 80% load factor over 60 years, it will produce 694,267,200 MWh of electricity, bringing the cost to 10.8 USD/MWh. Add that to the ARENH revisited price (that already includes deconstruction + spent fuel storage + running costs + maintenance + fuel cost), it's 70 USD/MWh. Granted, it does not include the cost of financing but it's a far stretch from 160. What am I missing? Can we add the "disputed section" tag while we understand this? Uzinagaz (talk) 20:10, 16 March 2021 (UTC)
- No, that's not how Wikipedia works. Editors don't modify an article based on their own research, this is called original research and is not permitted here. Rather, a reliable source has to be cited in order to support the information before it can go in the article. Further down the road comes how to present different sources with a Wikipedia:Neutral point of view. Lklundin (talk) 20:22, 16 March 2021 (UTC)
- You are right to point out I did not make my initial suggestion clear. I am suggesting to remove (or adjust with more information) the figures because:
- 1. The source publication is not peer reviewed (it's written in the source document "source: lazard estimates" - which doesn't give the calculation hypothesis to verify. How is that more reliable than my demonstration?)
- 2. A back-of-the-envelope calculation allows to question the validity of the publication's figures
- 3. The section does not state the notes on the source document that give off essential information to correctly interpret the figures
- I'll add to my suggestion to also remove the BNEF statement that is simply plain wrong. You cannot compare intermittent energy costs with controllable energy costs without adding at least storage costs (or controllable energy costs). Using their own figures present there, solar or PV comes out more expensive than anything else. Uzinagaz (talk) 21:15, 16 March 2021 (UTC)
- No, that's not how Wikipedia works. Editors don't modify an article based on their own research, this is called original research and is not permitted here. Rather, a reliable source has to be cited in order to support the information before it can go in the article. Further down the road comes how to present different sources with a Wikipedia:Neutral point of view. Lklundin (talk) 20:22, 16 March 2021 (UTC)
- As the editor who has been updating this page with the Lazard data as it comes out, I'm sympathetic to Uzinagaz's argument that the Lazard estimate is not a good guess at the LCOE of future nuclear SMRs. The EIA guesses they'll be around $70/MWh in 2026: https://www.eia.gov/outlooks/aeo/pdf/electricity_generation.pdf which roughly matches what I've seen elsewhere. But the LCOE studies that this section is based on existing data from new deployments, and we simply have many fewer of those. (Should we count Vogtle's obscene cost overruns?) Nuclear poses challenges to anyone seeking apples-to-apples comparisons. You project a very long lifetime for nuclear plants, which is legitimate, but you don't appear to include a non-0 discount rate, which is IMHO not legit, or a cost of fuel disposal, which is AFAICT impossible: could be negative (fuel is reprocessed) or gigantic (Hanford). Note that the challenges of getting reasonable estimates are not confined to nuclear: IRENA, which unlike BNEF or Lazard is in the business of promoting wind and solar and therefore "biased", has implausibly high estimates for their costs—the opposite of what you would expect from their "bias". Your criticism of BNEF is unjust: they are quite aware of the wide range of costs that can be included or left out, and they do their best. You only cite the omissions whose inclusion would favor nuclear. NuScale and Oklo are working to deliver SMRs at, IIRC, ~$4,000/kw installed; once they are doing that we should have several up to date LCOE values for nuclear under $100. Until then, the best we can do as stewards of this page is to **find good data sources and add them**. I particularly encourage you to put country-specific nuclear data in country sections. In short: cost estimates are complicated, especially for nuclear. As Wiki editors, let's work together to gather and cite the best ones we can find. As a citizen, I like you am rooting for a nuclear renaissance, but that shouldn't affect what I put on this page. Sampenrose (talk) 02:33, 17 March 2021 (UTC)
- Discount rate has little to 0 impact on nuclear electricity cost. Has stated previously, the back-of-the-envelope calculation I provided does not include financing cost that may or may not be relevant depending on how the project is financed (see Cloud200 answer below too). ARENH price includes fuel disposal costs (based on Deep geological repository) so it's already factored in. France has 13% of the world running nuclear reactor by count (58 out of 440) and 15% by installed power (61.4 GW out of 399). If Lazard range does not include this cost, is it representative of reality? I do think nuclear power could help in the climate change challenge (but it doesn't cloud my judgement as going full nuclear is impossible anyway, at least without Breeder reactor), so I spend a lot of time researching it. The figure jumped at me, making me question the validity of it and starting this talk. Now I realize that the problem is not only for nuclear power. The Global studies section does not warn the user of the difficulty to appreciate the cost of electricity globally nor the parameters in play. It also does not warn him of the validity of trying a global approach. At the very least, a global cost estimate should at least propose a range by source to reflect reality and stay as neutral as possible. I favor quality over quantity, if the current sources are too limited, they should not be there. Uzinagaz (talk) 13:13, 17 March 2021 (UTC)
- As the editor who has been updating this page with the Lazard data as it comes out, I'm sympathetic to Uzinagaz's argument that the Lazard estimate is not a good guess at the LCOE of future nuclear SMRs. The EIA guesses they'll be around $70/MWh in 2026: https://www.eia.gov/outlooks/aeo/pdf/electricity_generation.pdf which roughly matches what I've seen elsewhere. But the LCOE studies that this section is based on existing data from new deployments, and we simply have many fewer of those. (Should we count Vogtle's obscene cost overruns?) Nuclear poses challenges to anyone seeking apples-to-apples comparisons. You project a very long lifetime for nuclear plants, which is legitimate, but you don't appear to include a non-0 discount rate, which is IMHO not legit, or a cost of fuel disposal, which is AFAICT impossible: could be negative (fuel is reprocessed) or gigantic (Hanford). Note that the challenges of getting reasonable estimates are not confined to nuclear: IRENA, which unlike BNEF or Lazard is in the business of promoting wind and solar and therefore "biased", has implausibly high estimates for their costs—the opposite of what you would expect from their "bias". Your criticism of BNEF is unjust: they are quite aware of the wide range of costs that can be included or left out, and they do their best. You only cite the omissions whose inclusion would favor nuclear. NuScale and Oklo are working to deliver SMRs at, IIRC, ~$4,000/kw installed; once they are doing that we should have several up to date LCOE values for nuclear under $100. Until then, the best we can do as stewards of this page is to **find good data sources and add them**. I particularly encourage you to put country-specific nuclear data in country sections. In short: cost estimates are complicated, especially for nuclear. As Wiki editors, let's work together to gather and cite the best ones we can find. As a citizen, I like you am rooting for a nuclear renaissance, but that shouldn't affect what I put on this page. Sampenrose (talk) 02:33, 17 March 2021 (UTC)
- In all economic models of power generation there are two critical factors that can make energy very expensive or very cheap, both on paper and in reality: Discount rate and cost of financing. Incidentally, in most nuclear power economic models (including LAZARD), for nuclear these parameters are always somehow set to the worst-case scenario. Discount rate is set to a very high value (8-12%), basically meaning that the nuclear power plant loses economic value after 10-20 years, which is utterly stupid if look at nuclear power plants providing the same amount of energy after 60 years of operations. The second factor, cost of financing, is more realistic - cost of private financing of nuclear power plant is very expensive and takes up to 60% of its price. Just to repeat: up to 60% of some nuclear power plants is the loan interest. In countries where power infrastructure is publicly funded, this cost if obviously negligible, as you have noticed already in case of China. All that is explained pretty well in these two articles:
- I suppose the first one could be used as a WP:RS, but I'm also pretty sure this criticism of LAZARD methodology is also present in literature. Also, nothing prevents you from adding ARENH or any other actual nuclear electricity pricing data to the article if they can be referred to WP:RS and they are not used to criticise LAZARD directly ("look, these prices are lower!") but any thinking reader can compare these two and draw their own conclusions. Cloud200 (talk) 22:20, 16 March 2021 (UTC)
- Your first link does question the methodology validity of Lazard for nuclear costs. If this part is questionnable, how could we trust the rest? It is my understanding that a source can be quoted when it is reliable. My point is that it might not be and I'm asking for other's opinion. Uzinagaz (talk) 13:13, 17 March 2021 (UTC)
- Because LCOE is what it is - it doesn't claim to describe the actual price of energy from existing or new power plants. It's an financial tool for investment evaluation, pretty much like ROI, it doesn't care about CO2 or anything else but describes a particular parameter from investor's perspective. The fact that it's widely misinterpreted as "cost of energy" is another question and initially renewable energy supporters protested against it just as much as nuclear, but then LCOE started to show renewables as cheap so they calmed down. The article probably should have big fat warning on the top about misinterpreting the LCOE formula. Cloud200 (talk) 14:07, 17 March 2021 (UTC)
- Your first link does question the methodology validity of Lazard for nuclear costs. If this part is questionnable, how could we trust the rest? It is my understanding that a source can be quoted when it is reliable. My point is that it might not be and I'm asking for other's opinion. Uzinagaz (talk) 13:13, 17 March 2021 (UTC)
Proposed changes
Based on the review on the article and partially on the above discussion I would like to propose the following changes:
- Move the Per-unit cost metrics above the Global studies section to prevent casual readers from jumping into conclusions before they understand what they are looking at
- Add a bold note in the Global studies section referring to the LCOE article for disclaimers about the caveats of LCOE formula
- Remove the chart in the beginning of Global studies section as it serves as the best self-proof of the confusion caused by the concept of LCOE: averaging estimates from studies based on radically different assumptions (capacity factor, discount rate) is not only WP:OR but also ridicules the authority of Wikipedia as reputable source.
Cloud200 (talk) 18:17, 19 March 2021 (UTC)
- I am blocking consensus on these proposals, noting "ridicules the authority of Wikipedia" as not in scope for editing decisions. Sampenrose (talk) 02:47, 20 March 2021 (UTC)
- I tried removing the graph with the averages, but I was reverted by Sampenrose. I don't want to start reverting reverts, but this is a very clear case of WP:OR. In particular, it is definitely not a routing calculation (WP:CALC) and as this talk page discussion proves, there is no consensus for it. So I would be grateful if another user would take a look and fix this. --Ita140188 (talk) 02:19, 20 March 2021 (UTC)
- Hi Ita140188 Regarding "There is no consensus for it", I translate that as "I made a unilateral decision to remove sourced material from another user and got reverted." Regarding "this is a very clear case of WP:OR", you have not made that case. I have reviewed WP:OR and do not see it. The bar graph presents the table in graphic form, as is common on Wikipedia. If you would be happier with an X-axis that had 1 value per source instead of all N values divided by N, I am perfectly happy to accept such an edit. Sampenrose (talk) 02:44, 20 March 2021 (UTC)
- Update: I said that the chart removal was made "without opening a Talk discussion"; that was incorrect. Someone else in effect opened that discussion. Sampenrose (talk) 02:49, 20 March 2021 (UTC)
- Thanks for your replies. As for the consensus, WP:CALC is about consensus to add non-trivial calculations, not the other way around. Anyway, I would be completely ok if the graph had a bar for each source, instead of one bar for the average. The OR problem is the average here, not the graph. I can do the change if you want. --Ita140188 (talk) 06:49, 20 March 2021 (UTC)
- The chart is WP:OR because it applies an additional aggregation to data that cannot be aggregated in such way. Each of these LCOE estimates are calculated on different methodology and you just can't go and average them as it's comparing apples an oranges. Presenting it as three separate bars for each publication as Sampenrose would be absolutely acceptable as it would be indeed just graphical presentation of the table - it's the average that is a problem here. Cloud200 (talk) 09:21, 20 March 2021 (UTC)
- "Each of these LCOE estimates are calculated on different methodology and you just can't go and average them as it's comparing apples an oranges." The entire point of LCoE is that it makes apples-to-apples comparisons. That's what "levelized" means. That said, we have consensus on a path forward for the graphic, so the disagreement is moot. Sampenrose (talk) 15:33, 20 March 2021 (UTC)
Something like this, I just can't figure out how to add Lazard, IRENA etc to the legend:
![]() | Graphs are unavailable due to technical issues. There is more info on Phabricator and on MediaWiki.org. |
Thanks, I corrected the values and added a legend (the order is decided by the graph apparently):
![]() | Graphs are unavailable due to technical issues. There is more info on Phabricator and on MediaWiki.org. |
--Ita140188 (talk) 10:12, 20 March 2021 (UTC)
- Edit: I fixed the order by putting Lazard first, which has values for all technologies. --Ita140188 (talk) 10:18, 20 March 2021 (UTC)
- I also oppose putting the note there and am going to remove it. Sampenrose (talk) 15:33, 20 March 2021 (UTC)
- Ita140188 thank you so much for revising the graphic! The work involved can be tedious; I appreciate the effort and good will you invested. Sampenrose (talk) 15:33, 20 March 2021 (UTC)
- Ok, I thought nobody opposed the other changes proposed. What do we now then? I believe the current presentation of LCOE as "price of energy" is still misleading. Returning to the comment by Sampenrose above, levelization just at one parameter (financial) but ignores capacity factor and assumes an arbitrary discount rate as discussed in the publications linked above. Cloud200 (talk) 15:14, 23 March 2021 (UTC)
- I'm very sympathetic to the fact that you used the Talk page to float your ideas and took silence for consent. If you read back a couple years, that's how I started. I'll try to pay closer attention and engage constructively here; my apologies for lagging. Regarding the substance of your point: 1. criticisms of LCoE exist. 2. LCoE includes capacity factor: it's measure of energy, not power. 3. To my knowledge, the most respected global survey of energy prices are the BNEF, IRENA, and Lazard surveys that use LCoE. If there is a better survey, please cite its data! If there is a methodology with comparable uptake (not one that you like per your arguments on this page, one that is as recognized as LCoE), please cite that! 4. I have no idea why averaging discount rates is illegitimate; it seems to me obviously better than not averaging them. 5. Your above "LCOE ... doesn't claim to describe the actual price of energy from existing or new power plants. It's an financial tool for investment evaluation, pretty much like ROI, it doesn't care about CO2 or anything else but describes a particular parameter from investor's perspective" does not reflect how BNEF/IRENA/Lazard present LCoE or are received by their readership. A more accurate statement would be along the lines of "determining average costs is by construction an exercise in collapsing a diverse set of data points into a single number that does not reflect their full range. LCoE focuses on factors X, Y, and Z, making simplifications X1, Y1, and Z1. It does not include factors A, B, and C." You can check the three LCoE-based studies by evaluating their common assertion that solar and wind are the cheapest sources of energy. If it is true, you would expect other sources to report that solar and wind are market leaders because of low prices. Which, in fact, they do.Sampenrose (talk) 03:44, 24 March 2021 (UTC)
- Ok, I thought nobody opposed the other changes proposed. What do we now then? I believe the current presentation of LCOE as "price of energy" is still misleading. Returning to the comment by Sampenrose above, levelization just at one parameter (financial) but ignores capacity factor and assumes an arbitrary discount rate as discussed in the publications linked above. Cloud200 (talk) 15:14, 23 March 2021 (UTC)
- Ita140188 thank you so much for revising the graphic! The work involved can be tedious; I appreciate the effort and good will you invested. Sampenrose (talk) 15:33, 20 March 2021 (UTC)
- Sampenrose Some examples of inconsistency between LCOE models and actual energy prices were already given by Uzinagaz. This is precisely what drawn me into this discussion earlier and after doing some research on my own I did see that there is indeed a problem of LCOE being widely misinterpreted.
- Discount rate is an arbitrary parameter, assumed by the author of the model. It's not a physical parameter, like capacity factor, neither economic, like price of concrete. If you are an investor interested in a particular project and you estimate LCOE for that specific project, you know what the discount rate on your investment is. If you are Lazard, you basically come up with a number that is taken out of a hat.
- Why this is important? Because, as demonstrated in the OECD NEA report[1], if you assume 3% discount rate, residential PV LCOE is $100-200 but if you assume 10% it suddenly skyrockets to $150-350. For nuclear power they jump from $30-60 range to $50-130 range etc, and the ranges are becoming much wider too. Any model that produces such massively different outputs based on a single assumed parameter is prone to be wrong more often than right.
- I did check Irena 2019 study, they assume discount rate of 7.5%. Why? No idea. They just take this value but don't explain why.
- BNEF, Irena and Lazard are private companies, working for investment banks. They are not NGOs, public research institutions or non-profit organisations. A for-profit company that comes up with a proprietary methodology that has a massive impact on projected "price of a technology" should fire a warning light due to possible conflict of interest.
- And the LCOE models actually wildly differ when compared to actual prices of energy. One ARENH example[2] has been discussed above - that was $58, which is well in the middle of the 3% discount rate range but well below the median for 10% range. Another example is in Sweden[3], article in Swedish but the key messages are:
- In public debate people actually do use BNEF estimates as actual expected price for energy from that source. In that case, arguing that Swedish estimates for nuclear plant are "too low", because BNEF says they should be much higher.
- Another example is OL3 power plant in Finland, where the actual investor calculated LCOE is way lower:
Based on the above production targets and the current operating and capital cost expectations, TVO’s average long-term production cost target for OL1, OL2 and OL3 EPR combined is expected to be approximately 30 € / MWh. OL1 and OL2 together produce 14–15 TWh / year at a cost of 20 EUR / MWh, which means that TVO's own calculation (corresponding to LCOE) for OL3 is around 42 EUR / MWh, in line with the simple calculation from public data above. So what does Bloomberg New Energy Finance (BNEF) state for the results of its LCOE calculation for OL3? A range has been reached that ranges from a minimum value of the equivalent of 190 EUR / MWh up to a staggering 375 EUR / MWh. Two things about this are strange, partly that you do not seem to have a clue where you think the figure will actually land, for a project that has been completed for a long time and will go into commercial operation at the beginning of next year, and partly that the cost estimates are up to 900% higher than the owner's easily verifiable public figures from annual reports.
- The author makes a very good point about BNEF and other proprietary models:
These so-called actual costs are said to differ from the assumed costs that have been used as input values in the studies. The criticism is quite diffuse as it never questions any input value (for example capital cost, operating cost, operating length, etc.). Instead, the criticism consists of a comparison of the “LCOE value” (electricity production cost per kWh) that one would get if it is calculated with these input values, and a set of “LCOE values” which is thus stated by a specific company (BNEF). (...) In my opinion, it is deeply inappropriate, as academics in the media, to trumpet "what new nuclear power costs" by repeating values from a publicly secret LCOE estimate of a private company, performed with unclear input values and with unclear methods, based on a selection of a handful of projects (out of more than 100 projects in the last 20 years). The figures from BNEF have a certain shock value, as they stand out by being dramatically higher than corresponding estimates from virtually all other similar calculations and estimates, carried out by, for example, Energiforsk, the Swedish Energy Agency, International Energy Agency (IEA), European Commission, OECD and the U.S. Energy Information Administration (EIA). Could it really be that all other organizations are completely wrong and only BNEF is right about this?
The he goes into a in-depth analysis:
BNEF states its own input values, without reference to sources, with values that are often incorrect. Sometimes the input values seem to correspond fairly well with reality, sometimes not at all. In the case of Finland (OL3), for example, a cost of capital has been assumed for the customer that is up to 230% above the actual value. For the cost item “fixed operating costs”, a value has been assumed that is up to 300% higher than the fixed operating costs for Nordic nuclear power today. Illustrative of the quality of the data collection performed for the analysis is that in their calculations they also stated an incorrect size of the reactor, 1400 MW instead of 1600 MW. BNEF has not been able to provide any explanation for any of these discrepancies.
And finally:
The figures from BNEF can therefore not be interpreted as, or referred to, as the “actual cost of new nuclear power” and should absolutely never have been presented in such a way in the Swedish debate.
I don't want to copy-paste the whole article about BNEF here, but it contains much more examples of distortion of actual costs by BNEF. Therefore, presenting on Wikipedia results of proprietary models based on undisclosed parameters by private companies who have vital interests in specific valuation of specific companies and doing so without a word of caution is simply irresponsible. Cloud200 (talk) 09:48, 24 March 2021 (UTC)
- Thanks so much for the thoughtful reply. It will take me a day or two to get through your links; I'll try to respond in depth by this weekend. Sampenrose (talk) 15:52, 24 March 2021 (UTC)
- Thank you for the interesting discussion and points raised. If I can give my view, I think the LCOE has three main problems:
- It is heavily reliant on financial assumptions, especially for technologies where capital costs are dominant.
- For technologies like nuclear power, where costs are heavily influenced by political decisions, quality of the supply chain, and ultimately vary widely by country and even single project, it is also difficult (if not impossible) to interpret a single number of even a range.
- The LCOE always assumes a marginal penetration for non dispatchable technologies. In an extreme example, you cannot just replace 100% of your gas generation with solar because its LCOE is lower, you would end up with a non functioning energy system.
- So obviously these numbers must be taken with a huge grain of salt, and this should be made very clear in this article. --Ita140188 (talk) 03:02, 25 March 2021 (UTC)
- Thank you for the interesting discussion and points raised. If I can give my view, I think the LCOE has three main problems:
- After looking in vain through Cloud200's references for data in the cost of energy that is at all comparable to BNEF, IRENA, or Lazard's, I continue to believe that the broadly recognized studies that actually exist are the ones we should use. While there are some good points made by several of you regarding LCoE, there are also incorrect and debatable ones. In the end, I cannot support removing the best data sources we have. Let's improve this page by:
- Cleaning up the many outdated sections
- Tracking down other reputable cost surveys.
- If we can find new cost surveys of comparable quality that use a non-LCoE methodology, let's add them. If they don't exist, isn't that evidence that LCoE is the state of the art? Sampenrose (talk) 00:40, 28 March 2021 (UTC)
- After looking in vain through Cloud200's references for data in the cost of energy that is at all comparable to BNEF, IRENA, or Lazard's, I continue to believe that the broadly recognized studies that actually exist are the ones we should use. While there are some good points made by several of you regarding LCoE, there are also incorrect and debatable ones. In the end, I cannot support removing the best data sources we have. Let's improve this page by:
- It's not about state of the art or not: I agree that LCOE is most widely used way to compare energy costs in a generic way (without specific context). What I'm saying is that for many technologies, and especially renewables and nuclear, a measure of cost without context has little value to understand how these technologies compare in a real grid. For the case of nuclear, it's because of the uncertainty given the few projects currently existing, and for the case of renewables because they are not dispatchable. In the case of non-dispatchable sources, the actual grid costs depends a lot on the current penetration of the technology. For very high penetrations, grid costs (such as seasonal storage) are completely dominant, so the cost of generation alone becomes almost meaningless. --Ita140188 (talk) 03:08, 28 March 2021 (UTC)
- "I agree that LCOE is most widely used way to compare energy costs in a generic way (without specific context). What I'm saying is that for many technologies, and especially renewables and nuclear, a measure of cost without context has little value to understand how these technologies compare in a real grid." I believe you have conceded the point. Sampenrose (talk) 01:47, 29 March 2021 (UTC)
- @Sampenrose: Not sure what you are talking about. We are not having a debate competition. My point is that there should be some form of note to clarify that these numbers are not absolute and have limitations, and may even be meaningless in certain contexts (which is the original point of this discussion).--Ita140188 (talk) 02:09, 29 March 2021 (UTC)
- Since LCoE is the standard way of measuring the cost, we should not be labeling it as limited or meaningless. Sampenrose (talk) 05:02, 29 March 2021 (UTC)
- Not sure if I am not being clear. What I mean is: LCOE is used to give one number (or range) independent of context. However, in many cases, context (eg. regulatory environment, current penetration of non-dispatchable generation, etc.) is extremely significant in determining the actual cost of technologies. So LCOE (or similar universal context-free cost measures) should be taken with a huge grain of salt when comparing costs, especially of non-dispatchable technologies. --Ita140188 (talk) 05:49, 29 March 2021 (UTC)
- Sampenrose Sorry, but this is not how this discussion should look like. You have happily dismissed five pages of very specific and sourced criticism of how LCOE is presented here by a simple statement that some arguments "incorrect and debatable" WP:WEASEL without even pointing out a single one. Also, nobody ever proposed "removing" the three LCOE sources and this is something you made up now. This discussion started with a proposal to clearly annotate the existing LCOE data to avoid confusing people who then run around and claim "nuclear will cost $198 because this is what Lazard said". Cloud200 (talk) 17:02, 28 March 2021 (UTC)
- Sampenrose Further to the NEA LCOE data linked above, which you simply ignored (I can't believe you classified it as "incorrect or debatable"), there's also 2018 IPCC LCOE estimates, which clearly explains the limitations of the methodology (requires knowledge of capital flows, ignores costs of balancing variable energy sources etc) and presents LCOE in four scenarios, with 5% and 10% cost of capital, carbon tax and capacity factor - page 1331. I can see no other way to make this article reliable and science based than 1) clearly explain limitations of LCOE methodology as suggested from the beginning, 2) include NEA and IPCC LCOE estimates in addition to the three existing private reports. Cc Ita140188 [4] Cloud200 (talk) 18:09, 28 March 2021 (UTC)
- OK, some specifics.
- "Discount rate is an arbitrary parameter, assumed by the author of the model. It's not a physical parameter, like capacity factor, neither economic, like price of concrete. If you are an investor interested in a particular project and you estimate LCOE for that specific project, you know what the discount rate on your investment is. If you are Lazard, you basically come up with a number that is taken out of a hat." Capacity factor is NOT a physical parameter. It's a statistical measure of a physical parameter (watts) over some time period. When the period is in the past, it can be extremely precise. When the period is in the future, it is an estimate, just like discount rate. Discount rate is an estimate of a real phenomenon: that people value present returns more than future returns. Lazard is, according to Wikipedia "the world's largest independent investment bank". You have accused them of gross incompetence, if not actual fiduciary negligence. Such assertions do not belong on Wikipedia. If they have been made by reputable sources, please cite them.
- "The LCOE always assumes a marginal penetration for non dispatchable technologies. ... you would end up with a non functioning energy system" You are correct that LCoE is not a good measure of what it does not purport to measure. The measures I am aware of that incorporate whole-grid costs such as https://www.edf.org/sites/default/files/documents/SB100%20clean%20firm%20power%20report%20plus%20SI_clean.pdf (from an energy expert, it cites a Lazard levelized cost of storage study approvingly) and several others find that we can add variable renewables to our grid at ever-declining prices for years, but that we will have to supplement them. They are not arguments that LCoE is bogus or that nuclear is cheaper than renewables, because it isn't and it isn't, respectively.
- I could keep going, but I believe I have made my point.
- If you are mad that LCoE finds that the cost of nuclear is high, there is a simple solution that would work for me: add a section on cost of nuclear that cites reputable sources. They should be easy to find. The core issue we'll still disagree on is whether a hypothetical watt of electricity delivered in 2065 should be valued in 2021 the same as a watt of electricity delivered in 2021. That's silly and incorrect, but it's not illegitimate. If you can find a source that asserts it does (and you probably can), it's entirely appropriate to post that to this page. It's just not appropriate to remove or mark with warnings reputable sources which assert the (serious and correct) opposite--not because I say they are serious and correct, but because they are the most reputable sources we have, in the literal sense of "reputable".
- "Discount rate is an estimate of a real phenomenon" - this is precisely the point of the criticism of LCOE expressed in NEA, IPCC and other articles linked here. Discount rate is different for each project. Each investor values their money differently. What Lazard does is to take a number of arbitrarily selected discount rates from undisclosed projects, and assume that value as "model discount rate". This lack of transparency, which leads to cost estimates which are absurdly inflated as compared to actual costs is precisely at the base of criticism of Lazard. Cloud200 (talk) 06:59, 29 March 2021 (UTC)
- "You have accused them of gross incompetence, if not actual fiduciary negligence" - not me. NEA, IPCC and the guy from Sweden, who pointed out that actual cost of the investment in nuclear power plant in Finland is much lower than Lazard estimate. Cloud200 (talk) 06:59, 29 March 2021 (UTC)
- "a hypothetical watt of electricity delivered in 2065 should be valued in 2021" - in this specific example we are talking about existing projects with existing price tags, in Finland and France, not hypothetical ones. Cloud200 (talk) 06:59, 29 March 2021 (UTC)
- OK, some specifics.
Additions of old studies, discount rate
My thanks to Cloud200 for tracking down and adding the NEA and IPCC studies from 2018, and for highlighting discount rate (DR). A couple thoughts:
- If we want to apply DR, then a single table should not show prices calculated from different DRs as if they were the same.
- If a study offers 3%, 7%, and 10% DRs, then show all three somehow when representing that study.
- 2018 studies and 2021 studies (BNEF) do not deserve equal placement on sources which have seen significant deflation in the intervening period. The two 2018 studies show much higher prices for wind and solar because they cost much more in 2017 than they did in 2020, when the respective data was gathered. The 2018 studies were presumably accurate at the time, but are not now, and should not be used **for that purpose**. If they are the best sources for other purposes, by all means let's use them for those purposes.
From previous discussions, I believe that Cloud200 is concerned that the more recent studies are overstating the cost of nuclear generation by applying a discount rate which is too high. Let's present the reputable information they have added in a way that does justice to it without doing injustice to the other reputable sources on the page. I have some ideas about how to do so, but perhaps Cloud200 would like to suggest some first. Sampenrose (talk) 03:12, 1 April 2021 (UTC)
- What you seem to be getting from these studies, is that cost of PV and wind has dropped 3x in just two years from 2018 to 2020, but at the same time cost of nuclear has increased 3x times. Is that what you are saying? Cloud200 (talk) 10:54, 1 April 2021 (UTC)
- I'm saying the opposite, actually: the 3x difference is an artifact of comparing sets of studies with different methodologies, and we should fix that **while still respecting the value of both sets of sources**. Here is BNEF in 2018: https://assets.bbhub.io/professional/sites/24/2018/02/Sustainable-Energy-in-America-2018-Factbook.pdf. On slides 40-41 you can see that it put solar at about $54/MWh and wind at about $60, vs $39 and $41 this year. (If you track down older Lazard and IRENA studies you'll find similar shifts; you could even find them from the history of this page.) You have made quality additions to this page, for which I repeat my thanks. Can you suggest a presentation of those additions in a way that doesn't create the 3x artifact (and addresses the points I raised above)?
- It would certainly make sense to indicate the year of data set (as opposed to publication year) but my point is that the differences in prices between NEA, IPPC and the proprietary studies are not caused only by actual change in price over time but also by different methodologies, as demonstrated by the absurd $163 Lazard price for nuclear energy, which started this whole series of changes weeks ago. If we take Lazard price in 2020 at face value, we would need to assume price for nuclear has increased threefold from 2018 as compared to IPCC and NEA - which is of course not true, as it's more likely that Lazard just cherry picked the most delayed and most expensive project for their estimate, while IPCC and NEA clearly took a broader and more representative sample. Cloud200 (talk) 11:37, 2 April 2021 (UTC)
- I'm saying the opposite, actually: the 3x difference is an artifact of comparing sets of studies with different methodologies, and we should fix that **while still respecting the value of both sets of sources**. Here is BNEF in 2018: https://assets.bbhub.io/professional/sites/24/2018/02/Sustainable-Energy-in-America-2018-Factbook.pdf. On slides 40-41 you can see that it put solar at about $54/MWh and wind at about $60, vs $39 and $41 this year. (If you track down older Lazard and IRENA studies you'll find similar shifts; you could even find them from the history of this page.) You have made quality additions to this page, for which I repeat my thanks. Can you suggest a presentation of those additions in a way that doesn't create the 3x artifact (and addresses the points I raised above)?
- Conversely, do you think that Nuclear looks better if you only include the lowest discount rate in the article? If not, why are you only doing that? Given that discount rate is an external factor, that can vary globally over time, it doesn't seem reasonable to do that. You seem to have been basically cherry picking the data to favour nuclear power. GliderMaven (talk) 12:16, 2 April 2021 (UTC)
- "You seem to have been basically cherry picking the data to favour nuclear power." I agree that **since the source shows multiple DRs** then citations of the article should too (point #2 above). But Cloud200 has explained that they think the NEA article is a better source for nuclear, and I agree with the (implicitly included) softer claim that it is a reputable source, so let's figure out the best way to present it. Regarding Cloud200's specific reasoning:
- "my point is that the differences in prices between NEA, IPPC and the proprietary studies are not caused only by actual change in price over time but also by different methodologies ..." -> Agreed, thank you!
- "...as demonstrated by the absurd $163 Lazard price for nuclear energy, which started this whole series of changes weeks ago. ..." Your evaluation of "absurd" may or may not be correct, but it is not appropriate here. Lazard is a solid source, you don't get to throw it out because you disagree with it any more than someone else should get to throw out the quality sources you've added.
- "...If we take Lazard price in 2020 at face value, we would need to assume price for nuclear has increased threefold from 2018 as compared to IPCC and NEA..." Agreed (again) on "as compared to IPCC and NEA" but not "from 2018". Lazard 2018 did not find a 3x lower cost than Lazard 2020.
- "...which is of course not true..." Agreed, thank you!
- "...as it's more likely that Lazard just cherry picked the most delayed and most expensive project for their estimate, while IPCC and NEA clearly took a broader and more representative sample." Per "absurd" above, "more likely" and "clearly" are your personal evaluations. And they are perfectly reasonable evaluations, but one editor's evaluations don't get to overrule general Wikipedia editing principles.
- Here's where I think we are:
- Cloud200 has made the case for explicitly showing the DR behind specific cost estimates. This feels to me like a genuine improvement to the page that all editors should support, but please speak up if there is disagreement.
- Cloud200 has found new reputable sources which complement the existing sources. This also feels like a genuine improvement to the page that all editors should support, but please speak up if there is disagreement.
- The specific edits which introduced DR and the new sources could use some tweaks for reasons given above, and since Cloud200 did the hard work of adding improvements to the page, I'd like to give them first cut at addressing the issues that GliderMaven and I have raised.
- My thanks to all involved for working through these concerns. Sampenrose (talk) 15:19, 2 April 2021 (UTC)
- "You seem to have been basically cherry picking the data to favour nuclear power." I agree that **since the source shows multiple DRs** then citations of the article should too (point #2 above). But Cloud200 has explained that they think the NEA article is a better source for nuclear, and I agree with the (implicitly included) softer claim that it is a reputable source, so let's figure out the best way to present it. Regarding Cloud200's specific reasoning:
- I think that's broadly the right direction, but the details matter. Let's see what Cloud200 proposes. Sampenrose (talk) 18:07, 2 April 2021 (UTC)
Reverted rearrangement of major sections
GliderMaven (talk · contribs) moved the Global Studies section, the contents of which are under current discussion, without seeking consensus. I have reverted that edit until we have had a chance to discuss it. My opinion is that many readers who come to "Cost of electricity by source" will want to know values rather than a long discussion of methodology, and that therefore values belong at the top. I welcome other perspectives.
- It's completely unreadable and completely inconsistent, so I've reverted you. It was just awful. GliderMaven (talk) 11:07, 3 April 2021 (UTC)
- We have a summary section already, that's what the lead is. I suggest you add a summary of data to it. GliderMaven (talk) 11:10, 3 April 2021 (UTC)
- Please seek consensus.