The 'unknown' part in the Belgian generation mix

[MCvarial]

Belgium always has a generation of 0,5 to 1GW from unknown sources. This generation is rated at 700g/kWh the equivalent to coal, however Belgium does not have a single coal plant in service. The Belgian grid operator Elia defines this unknown part as

'Other': including waste, biomass, blast furnace gas, photovoltaic, etc.

source: http://www.elia.be/en/grid-data/power-generation

These are various generation sources handeled on a state level, for Flanders for example this is 48% biogas and 52% waste recycling. source: http://www.vreg.be/sites/default/files/statistieken/groenestroom/20161104-gsc_2d_uitgereikte_certificaten11.pdf

So my suggestion is to lower the carbon intensity of the unknown part to the equivalent of biomass rather than coal to get a more accurate representation.

[corradio]

Maybe we could say the the unknown part shouldn't change the carbon intensity? Unless you have a solid number I think this is the best option.

[MCvarial]

This seems like the most sensible solution at this point, while it would be possible to get a number for a particular year the generation values are bound to vary year after year.

[corradio]

@brunolajoie what should we do here?

[brunolajoie]

At first sight, I'd rather input a plausible carbon intensity to be consistent with all other countries on the map, rather than "remove" it from c-intensity computation. Are we sure that Elia "Other" correspond to "ENTSOE" other" category that we use for the EMap?

[MCvarial]

The other category of Elia consists of:

• 315MW is blast furnace gas production as baseload
• 719MW of cogeneration
• 363MW of biomass production as baseload
• 248MW of waste burning

[brunolajoie]

Sorry for this basic question @MCvarial, but your MW figure above-mentionned are installed capacities or instantaneous power?

Although I'm aligned on your general idea, I'm not sure to understand precisely what you suggest to change? Could you just explain me a bit more your suggestion? Do we have a detailed enough breakdown on ENTOSE do make the split that you suggest?

[MCvarial]

@brunolajoie

I have made a little spreadsheet comparing the Elia & ElectricityMap generation & installed capacity.

There are multiple discrepancies:

• The wind power reported to ENTSO-E is only the measured wind power, as only 39,64% is monitored the actual generation can be upscaled by dividing the reported number by 0,3964.
• Pumped storage & hydro generation is reported toghetter by Elia while on the map they're seperated but hydro is always stuck at 1MW. Given the vast majority of the installed capacity is pumped storage its probably best to report everything as pumped storage.
• The natural gas generation seems to be higher on the ElectrictyMap than Elia reporting. Which makes me wonder if biomass, cogeneration or blast furnace gas is included. I'd need to confirm that when one of those units trips.
• There's also a large difference between installed capacity reported by Elia & ENTSO-E. Unsure of whats going on here, cogeneration, biomass & mothballed capacity could be included.
• The "other" category of Elia seems to more or less match the "unknown" category of ENTSO-E. Which would mean the reported biomass generation can be substracted from the unknown category. And the carbon intensity of the remaining capacity can be corrected. The first 315MW would be blast furnace gas as baseload and the rest would be waste recycling.

So right now I would do the following:

• Correct the wind generation by dividing it by a factor 0,3964
• Eliminate the water generation category
• Correct the pumped storage installed capacity to 1394MW

I will continue to investigate whats going on with the gas & other category so we can later correct:

• gas generation
• installed gas capacity
• unknown generation
• carbon intensity of unknown generation Historical data would help me a lot with that so I can get multiple data points and wouldn't have to wait 24 hours for Elia to report the actual generation each time.

[brunolajoie]

Thanks a lot @MCvarial please find in this spreadhseet the historical production data derived directly form ENTSOE and displayed on electricity map for belgium for your investigation! Hope it will help you to figure it out

[MCvarial]

Using unit trips I was able to confirm that:

• the 315MW of blast furnace gas generation is included in the gas section on ENTSO-E
• biomass is not included in either the gas category or other category. Meaning:
• The Unknown category is waste recycling & cogeneration.

So now we can:

• Correct the installed gas capacity to 5092MW (5811MW-719MW of cogeneration supported by the fact that the historical peak generation was 5086MW on 22/01/2015)
• Change the name of the Unknown category into cogeneration & waste
• Set the installed capacity to 967MW
• Correct the carbon emissions, the goal of cogeneration units is to produce heat, a bit of that heat is converted into electricity and the "losses" can be used as heat. So you're essentially directly converting heat into electricity. Given burning 1MBtu of natural gas emits 117 pounds of CO2, you're looking at 181g/kWh

[alixunderplatz]

@MCvarial Nice! I think the last point might be controversial. Surely, a co-generation plant actually "avoids" emissions since the exhaust heat is used, which would otherwise be generated by burning fuel for this purpose only. But I'm not sure how far co-gen is already considered in the emission factor used for gas now. It also depends on the total efficiency of (heat output plus electricity output) divided by total fuel input which is definitely <100%, so emissions are for sure higher than 181 g/kWh. Unless you have numbers for both outputs or total fuel efficiency, it is hard to determine.

[HansHyde]

@alixunderplatz, Hello Alix

This may help you, and sorry a bit long.

The emissions are not "avoided" per se, but our efficiency of CO2 emissions per usable (consumed) energy increases, albeit marginally, with district or process heat recovered from a generator.

See this document... note the formulas starting on pg 71 Energy Transformation Efficiencies - https://www.ems.psu.edu/~radovic/Chapter4.pdf

Note that by the time the primary energy (coal, oil, nat gas, etc) reaches an electric generator, it has already gone through 3 steps with an efficiency & "carbon cost" of their own. For wind or PV, these 3 steps would not need to be included as the fuel (wind or sunshine) arrives for free.

I used this to calculate a (reasonable) carbon cost of natural gas delivered to a power plant of 22kG per MMBTU. For coal, it works out to 41kg per MMBTU delivered.

Egas = = [Eextraction] [Eprocessing] [Etransport] [Epower plant] [Etransmission] [Eelectric heater] = (0.73) (0.97) (0.95) (0.35) (0.90) (1.00) = 0.21 (21%) <-- this means 79% of all energy is lost.

There are 6 steps here. Note that our LCAfuel ### gCO2eq/kW values are attempting to capture a value which includes the first 4 steps (plus additional for plant infrastructure, construction & decommissioning, etc).

For your example above, you are entering step 5 with 2 choices;

1) the electricity goes off to experience 2 more steps, 1) losses in transmission & distribution AND 2) losses in conversion at the end user.

or

2) the ejected heat has to go through 3 more steps; 1) efficiency of capture at the electric generator via a HRSG (in a combined cycle NG plant), or a heat exchanger for an ICE (reciprocating engine), 2) transmission & distribution, 3) conversion at the end user.

So, in reality, we can calculate energy or carbon numbers with a relatively high degree of accuracy, if we keep in mind the values we calculate for each of the 6 or 7 steps is a weighted average proportional to its own reasonable assumptions.

For example, in my calculation above, for Eextraction(gas) I used a 50/50 mix of oil & gas used to extract it, giving me a Carbon Cost of 66 kgCO2 per MMBTU at an efficiency of 73%, this means 73% moves forward, 27% is lost forever. 17.84kg CO2 emissions per MMBTU is our Carbon Cost of extracting natural gas.

By the same calculations... 1.56 kg CO2 per MMBTU for Eprocessing 2.61 kg CO2 per MMBTU for Etransport

Total "Carbon Cost" of Natural Gas - 22.0 kg CO2 per MMBTU of natural gas delivered to our Power Plant (step 4).

PS... as Github is limited on what formats can be used & files types shared, you can reach me via info in my Github profile if you want & I can send you the "calculator" if you want to play around with it.

[HansHyde]

Hi @alixunderplatz ...

Olivier & Bruno, (@corradio @brunolajoie)

I don't want to insult anyone's intelligence (through "lost in translation" discussions or Github/Slack functionality limits) any more, making assumptions one-way or the other.

So can you confirm the following is fundamentally accurate regarding the ElectricityMap methodology and the important assumptions which need to be made by the user/viewer of EM's visual displays in order to use the data in realtime?

Thanks...

If this is fundamentally correct, then I have a follow-up distinction for you to verify regarding what the "Carbon Intensity in the Last 24 hours" graph implies; 1) for a realtime viewer, and 2) what is available for a "Pro" user that can access all the datapoints available in 1 hour increments. It will be very short! :D

[corradio]

I'm not sure this is the best place to ask/answer the question, but the quick answer is that the electricity unit displayed on the Electricity Map is power (W). Either the data received is already power (in that case we assume the value to be constant until the next data point arrives), or the data received is energy during a time interval, and then we divide energy by time to obtain the mean power during that interval.

Carbon emission factors for each power plant type are given as gCO2eq/kWh (grams of co2 equivalent per energy unit). The real-time power flows are solved to figure out where power comes from, enabling us to assign co2eq emissions from the viewpoint of the consumer (more in the README). Note that this is not in accordance to point 3, where you look at generation mix (we look at consumption mix).

I think we're in agreement although nothing is "dropped" as you say, and there's no need to introduce new units (MW_it). Furthermore, we're not even dealing with evenly spaced time intervals so there's not 12 sub-calculations.

Important: all electricity map units are rates (units / time), i.e. an instantaneous snapshot.

To keep this issue clean, I'd prefer if you asked your Pro question at pro@electricitymap.org where it belongs :)

--

@MCvarial thank you the analysis. We unfortunately can't change the name of the category (we'd need ENTSOE to do that), however we could assign a "waste/cogeneration" emission factor based on the LCA values we have specifically for the "unknown" category, highlighting our assumptions. The rationale for keeping the unknown category is to force ENTSOE and the data provider to update its data, and also to be compliant about future changes (they might change the unknown category without letting us know..). CO2 factors are not put in our database so we can change them easily. Changing category mappings requires going through the full database, so I'd refrain from doing that.

Olivier

[MCvarial]

@alixunderplatz

Cogeneration units in Belgium are not powerplants that recover some of the dissipated heat for heating purposes. There are gas & waste recycling units that do this but those aren't in the cogeneration category.

Cogeneration is the replacement of heating boilers in the industry by gas turbines or gas motors that drive a generator where the waste heat is the same as the heat production of the old heater unit. In other words any heat loss used to be there with the classic heater unit too, so its attributed to the heating proces rather than the electricity generating proces.

[HansHyde]

@corradio

Great, we are in 100% understanding of each other. Rates = instantaneous time. 12 was an example were the realtime data frequency is every 5 minutes.

I would only add, a displayed Rate, can easily be confused with a quantity (rate x time). I don't pay for a Rate, I pay for a unit of electricity (rate x time). If the Rate is 10gCO2/kWh and I consume 1 kWh, then I have "consumed" 10g x 3600seconds = 36,000 grams of CO2 or 36kg of CO2.

I don't need to ask the follow-up question, I already know the answer.

You mention the rate is from the perspective of the Consumer. The data is obtained from the perspective of the Generator, hence you would have to include losses from Delivery (Transmission & Distribution) to the Consumer. To keep it simple, 4-7% is industry standard.

Is this included within the Rate displayed for the perspective of the Consumer? IPCC rates are from the perspective of the Electric Generator.

I can't find any indication that LCA Rates beyond generation is included --> pg 539

Figure 7�6 | Comparative lifecycle greenhouse gas emissions from electricity supplied by commercially available technologies (fossil fuels, renewable, and nuclear power)

IPCC 2014 Climate Synthesis - https://www.ipcc.ch/pdf/assessment-report/ar5/wg3/ipcc_wg3_ar5_chapter7.pdf

[alixunderplatz]

There are many different methods of assigning emissions to co-generated heat and electricity. I don't know how many of you know the German language, but this study: https://www.ffe.de/download/wissen/334_Allokationsmethoden_CO2/ET_Allokationsmethoden_CO2.pdf comes to the conclusion, there is no ultimate method to calculate the related emissions fro co-generation and there is no distinct solution to it. it's up to you which method you use :))) But this needs to be done consistently for everything similar, of course.

Here is the maths for the methods:

Here are the results for 1000 Wh energy content of gas:

[brunolajoie]

@MCvarial let's summarize: (if I missed something let me know)

• 🆗 Updating capacities (hydro, gas, unknown, biomass) as you suggested
• 🆗 Update biomass and unknown CO2 intensity. What's your suggested value exactly and how do you compute it? Let's try to stick to a linear combination of IPCC's median values for each fuel type if possible
• 🚫 Change wind production output with your factor 1/0,3964: We need to stick to raw parser data. We will relay your finding to ENTOSE so they become aware of the discrepancy in a separate, cleaner issue!

-- Parenthesis (@HansHyde,

• Yes we want to include a fixed T&D loss per country to refine consumer-side emissions estimations. We wanna find something that is dependent on the country. lets keep this discussion in a separate issue.
• I'm not sure to get your calculus here "If the Rate is 10gCO2/kWh and I consume 1 kWh, then I have "consumed" 10g x 3600seconds = 36,000 grams of CO2 or 36kg of CO2." I would rather say that you have consumed 10gCO2.

[MCvarial]

@brunolajoie

The IPCC doesn't provide carbon intensities for cogeneration (75% of the installed capacity) & waste burning (25% of the capacity). We know cogeneration is much lower than the current 700g. And its lower than the 490g of CCGT generation. On the lowest end its 181g/kWh if you assume the electricity generation is the byproduct of heat production which is what all these cogeneration units do.

If you stick to the reported ENTSO-E generation for wind then the installed capacity for wind should be changed to 1039MW which is the share of capacity thats monitored.

[brunolajoie]

@MCvarial: clear. Let's wait a bit to refine carbon-intensity of co-gen vs. non-co-gen and assume its normal gen. I'm currently trying to find country-specific LCA values in issue #237.

Hello @Dalius-ENTSO-E ! I'd like to inform you about a discrepancy we found between Belgium ENTSOE TP production data and Elia data. As discovered by @MCvarial, and summarized in this spreadsheet, we see two main issues: 1/ pumped hydro generation seems to be under-reported in ENTSOE 2/ Only a third of BE wind capacity seems to be monitored Just wanted to let you know this finding, and ask for a possible insight on your side. Is Elia your primary data source too?

[Dalius-ENTSO-E]

Elia is checking installed capacities. Regarding wind on-shore, we see very low values since ~December 12. Elia is also informed and I hope, that they will update data very soon.

[corradio]

@brunolajoie @MCvarial what is the status here?

[brunolajoie]

If I understand it well, there are two main issues:

• Wind reported value seem to be about a third of actual wind output. I don't know what we can do about that except wait for data stream to be improved. (Find below historical wind output for BE that we have in our database: mean 463MW, max 2155MW)

• Unknown values actually refer to co-generation from waste. We could change defaultCO2intensity of unknown to that of biomass, OR change the parser category mapping from "unknown" to "biomass"

Then we'll need to adjust capacities installed accordingly, but that's a minor issue.

What's your view on that @MCvarial @alixunderplatz ?

[corradio]

Closing due to inactivity