Earth Notes: A Note On Variations in UK Grid Electricity CO2 Intensity with Time

What is "the" CO2 intensity (emissions) from consuming 1kWh from the UK's grid?

Work in progress!

The first thing to understand is that CO2 emissions or intensity per kWh generated for (and thus drawn from) the UK's grid is not constant; it varies by time of day and by season because of fuel mix and transmission losses and transmission patterns/routes (which means that it also matters slightly where on the grid you draw from).

Also, with significant penetration of intermittent renewables such as wind (~5% of peak winter demand 2008/2009 with a target of maybe 50% of peak) then the hour-by-hour intensity is even more volatile and hard to know.

There are several different 'average' values quoted and used too, eg: the UK government's 'aspirational' marginal cost of about 0.43kgCO2/kWh, the DEFRA 3-year rolling average figure (0.52kgCO2/kWh as of 2008/2009), 0.422kgCO2/kWh for grid-supplied electricity and 0.568kgCO2/kWh grid-displaced electricity to compute emissions for UK building regulations (2006, part L).

Incidentally, a working assumption of mine has been that grid CO2 intensity will peak when demand does because everything will be working flat out, including all the 'dirtiest' (most CO2-intense) generators. Clare Hanmer of the Carbon Trust did some work that suggests this is not necessarily so, and also that the diurnal/daily swing in intensity is larger in summer than in winter, given some educated guesses, etc. Part of my reason for this note is to test her interesting assertions! I'm borrowing some of her numbers and assumptions shamelessly.

Data Source: ELEXON

For the GB system, (ie the UK excluding Northern Ireland) ELEXON provides real-time and historical data on a number of grid parameters through the bmreports.com site, including generation by fuel type of plant connected to the high voltage transmission system. This does not by any means represent all generation, and may in particular under-represent wind and other distributed generation that connects at lower voltages and even within the distribution networks.

As at 2009/06/19 the BWEA UKWED database shows 3625MW of operational UK wind capacity and the BWEA's Jan Matthiesen kindly told me that 215MW of that is in Northern Ireland, ie not in the GB grid. At the same time the Elexon GB metered wind capacity is 1426MW, ie ~60% of the UK's operational capacity is embedded and not visible in Elexon's stats, so wind already is a significantly 'distributed' source.

ELEXON has very helpfully licensed me their data to conduct this study and related work.

Sample 1: Late March 2009

For the first 24-hour period that I examined from early March 2009, and ignoring the effects of pumped storage and interconnectors entirely, I saw ~20% variation in intensity in the 'high-voltage' connected generation, and an average for a consumer (including 9% loss in transmission/distribution) of well above the government's 'aspirational' marginal-generation figure and slightly above the 3-year DEFRA rolling-average. The intensity was fairly flat (and high) until about 8pm (after peak demand), and dropped to a low around midnight and rose again to 'high' around 8am, so showing a fairly strong first-order correlation of intensity with demand. There may have been more coal and less nuclear in the mix than typical.

Sample 2: Night of 2009/04/12

This 24-hour period during a bank-holiday weekend shows a 24% variance in kgCO2/kWh; intensity is relatively low because there is relatively little coal in the mix (<20%).

Effective grid carbon intensity for a domestic user is currently 327gCO2/kWh including transmission and distribution losses of 9%.

Latest available grid generation carbon intensity (ignoring transmission/distribution losses) is approximately 300gCO2/kWh at Mon Apr 13 00:55:00 BST 2009 over 26467MW of generation, with a rolling average over 24h of 350gCO2/kWh.

Minimum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 299gCO2/kWh at Mon Apr 13 00:45:00 BST 2009.

Maximum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 389gCO2/kWh at Sun Apr 12 22:05:00 BST 2009.

Average/mean grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 350gCO2/kWh over the sample data set, with an effective end-user intensity including transmission and distribution losses of 382gCO2/kWh.

Overall generation intensity (kgCO2/kWh) computed using the following fuel intensities (other fuels/sources are ignored): CCGT=0.36 COAL=0.91 INTFR=0.09 NPSHYD=0.0 NUCLEAR=0.0 OCGT=0.48 OIL=0.61 OTHER=0.61 WIND=0.0.

Sample 3: Evening of 2009/04/15

For some reason there seems to have been an enormous (~40%) variation min to max; hour-by-hour variation is ~30%. (Overnight intensity then dropped to 289gCO2/kWh making the mean 381gCO2/kWh and variability a whopping 44%!)

Effective grid carbon intensity for a domestic user is currently 425gCO2/kWh including transmission and distribution losses of 9%.

Latest available grid generation carbon intensity (ignoring transmission/distribution losses) is approximately 390gCO2/kWh at Wed Apr 15 21:40:00 BST 2009 over 37690MW of generation, with a rolling average over 24h of 393gCO2/kWh.

Minimum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 310gCO2/kWh at Wed Apr 15 00:15:00 BST 2009.

Maximum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 508gCO2/kWh at Wed Apr 15 20:45:00 BST 2009.

Average/mean grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 393gCO2/kWh over the sample data set, with an effective end-user intensity including transmission and distribution losses of 428gCO2/kWh.

Current/latest fuel mix: CCGT@17709MW COAL@8889MW INTFR@1572MW INTIRL@0MW NPSHYD@269MW NUCLEAR@8581MW OCGT@0MW OIL@0MW OTHER@0MW PS@199MW WIND@471MW.

Overall generation intensity (kgCO2/kWh) computed using the following fuel intensities (other fuels/sources are ignored): CCGT=0.36 COAL=0.91 INTFR=0.09 NPSHYD=0.0 NUCLEAR=0.0 OCGT=0.48 OIL=0.61 OTHER=0.61 WIND=0.0.

Sample 4: Evening of 2009/06/29

As of this sample there had been a potentional time-shifting saving in the last 24 hours of 35%.

Effective grid carbon intensity for a domestic user is currently 460gCO2/kWh including transmission and distribution losses of 9%.

Latest available grid generation carbon intensity (ignoring transmission/distribution losses) is approximately 422gCO2/kWh at Mon Jun 29 20:55:00 BST 2009 over 35535MW of generation, with a rolling average over 24h of 422gCO2/kWh.

Minimum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 324gCO2/kWh at Mon Jun 29 03:10:00 BST 2009.

Maximum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 494gCO2/kWh at Mon Jun 29 14:10:00 BST 2009.

Average/mean grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 422gCO2/kWh over the sample data set, with an effective end-user intensity including transmission and distribution losses of 460gCO2/kWh.

Hours that are basically green, but in which there is draw-down from grid-connected storage with its attendant energy losses and also suggesting that little or no excess non-dispatchable generation is available, ie that are marginally green, are shaded olive.

Current/latest fuel mix at Mon Jun 29 20:55:00 BST 2009: CCGT@16377MW COAL@9820MW INTFR@248MW INTIRL@0MW NPSHYD@135MW NUCLEAR@8539MW OCGT@0MW OIL@0MW OTHER@0MW PS@294MW WIND@122MW.

Current draw-down from storage is 294MW.

Overall generation intensity (kgCO2/kWh) computed using the following fuel intensities (other fuels/sources are ignored): CCGT=0.36 COAL=0.91 INTFR=0.09 NPSHYD=0.0 NUCLEAR=0.0 OCGT=0.48 OIL=0.61 OTHER=0.61 WIND=0.0.

Key to fuel types/names:

CCGT

Combined-Cycle Gas Turbine

INTFR

French Interconnector

INTIRL

Irish Interconnector

NPSHYD

Non-Pumped-Storage Hydro

OCGT

Open-Cycle Gas Turbine

PS

Pumped Storage Hydro

The intensity curve is unusually flat with coal apparently doing load-following (only ~7% min to max, and within a couple of days was down to ~4%), probably because of the gas shortage in the cold weather (with a gas demand of 441m^3/4.9TWh or ~200GW average over the day, causing National Grid to issue a Gas Balancing Alert (GBA) at 13:10 for Gas Day 04/01/2010, and demand went higher still later in the week) and a lack of (nuclear, ie low-carbon) electricity imports from France. There is also a big change in usage given the first two working days back since Christmas.

Effective grid carbon intensity for a domestic user is currently 583gCO2/kWh including transmission and distribution losses of 9%.

Latest available grid generation carbon intensity (ignoring transmission/distribution losses) is approximately 535gCO2/kWh at Tue Jan 05 20:10:00 UTC 2010 over 54357MW of generation, with a rolling average over 24h of 533gCO2/kWh.

Minimum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 513gCO2/kWh at Tue Jan 05 05:00:00 UTC 2010.

Maximum grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 549gCO2/kWh at Mon Jan 04 22:05:00 UTC 2010.

Average/mean grid generation carbon intensity (ignoring transmission/distribution losses) was approximately 533gCO2/kWh over the sample data set, with an effective end-user intensity including transmission and distribution losses of 581gCO2/kWh.

Current/latest fuel mix at Tue Jan 05 20:10:00 UTC 2010: CCGT@21678MW COAL@22760MW INTFR@0MW INTIRL@0MW NPSHYD@219MW NUCLEAR@7761MW OCGT@0MW OIL@0MW OTHER@0MW PS@1076MW WIND@863MW.

Current draw-down from storage is 1076MW.

Generation by fuel category:

Overall generation intensity (kgCO2/kWh) computed using the following fuel intensities (other fuels/sources are ignored): CCGT=0.36 COAL=0.91 INTFR=0.09 INTIRL=0.7 NPSHYD=0.0 NUCLEAR=0.0 OCGT=0.48 OIL=0.61 OTHER=0.61 WIND=0.0.

Key to fuel codes:

CCGT

Combined-Cycle Gas Turbine

INTFR

French Interconnector

INTIRL

Irish Interconnector

NPSHYD

Non-Pumped-Storage Hydro

OCGT

Open-Cycle Gas Turbine

PS

Pumped Storage Hydro

The fuel-mix (FUELINST) data is only available from Nov 2008, so here is an initial analysis for the whole of 2009. (I intend to update this periodically with new data.) This is generation intensity, ie ignoring transmission/distribution losses.

Note how carbon intensity is higher during the day, during the working week, and in the colder winter months, all where demand is highest; ie there is a fairly clear correlation between demand and intensity.

Note also how the mean variability (ie the maximum available CO2 savings from load-shifting) during each day varies from a low of 8% mid-winter to more than 4 times that in summer/autumn. My assumption is that nuclear (baseload) and renewables (zero carbon) are effectively used preferentially in the grid, but demand above their generation is satisfied by increasingly carbon-intense fuels, firstly gas (eg relatively-clean CCGT) all the way up to coal. In winter the zero/low carbon fuels make only a relatively small dent, but moving towards summer at night they cover a large chunk of demand and bring carbon intensity down further than is possible in winter.

It seems to be a good idea from an intensity point of view as well as an infrastructure-sizing and cost point of view to avoid running deferrable load at times of peak demand. If you can delay major loads (such as the dishwasher or washing machine at home) until late in the evening or the small hours after midnight you will probably significantly reduce your carbon footprint.

Note, however, that if the only reduction is a virtual one because of relatively fixed non-dispatchable zero-carbon generation (nuclear), then in order to achieve real footprint/emission reductions, and not just make everyone else's consumption higher/browner, you may have to defer load until other zero-carbon sources do increase output eg because the wind is blowing stronger or the sun shining brightly, rather just when zero/low carbon sources happen passively to form a higher proportion of extant demand. Having said that, if the demand curve becomes flatter though behaviour changes, then it will be possible to have (for example) more nuclear in the mix, and reduce the inherent waste of energy in using (eg pumped) storage, which would reduce long-term carbon footprint through requiring less fossil-fuel peak-demand support. This area requires more analysis.

Input data runs from Thu Jan 01 00:00:00 GMT 2009 to Thu Dec 31 23:55:00 GMT 2009.

Data Analysed By Hour-of-Day (GMT)

Data Analysed By Week/Weekend

Data Analysed By Month

Data Analysed By Year

Automation Test Footnote

As of 2009/05/10 I have one hosted Web server drop to reduced power mode unless it can detect that the status is 'green' over HTTP, and also our dishwasher controlled via X10 and the HEYU software to run when grid status is 'green' with status updated hourly to ensure that a 'quick' wash (<1h) cannot be interrupted.

What Emissions Can One Household Save?

Let us first assume that on average each household could defer 1kWh from the day/evening to overnight each day, which is about one warm/hot wash load for example. Many families will be running more than one a day, for example, so only need to defer some of their normal activity.

That would be 365kWh/household deferred from near maximum to near minimum intensity if it can be timed roughly right.

(The government (eg in DUKES) reports there to be ~25M UK households, so that would be ~9TWh of shifted load ie about 8%-ish of domestic demand, which seems within the bounds of pausibility.)

According to my partial data and analysis above it looks as though there's at least 100g/day (~20%) of CO2 to be saved per shifted kWh over the year as at 2009 (higher including summer months).

That suggests >30kg CO2 emissions per year saved by this for each load timeshifted per day.

That is the same saving as turning off our heating entirely for half a week in the middle of 2009's very severe winter.

Given a target that Europeans (well, everyone on the planet, but India and China are there more or less in 2009) should be aiming for of (say) total 2tCO2 per person per year that's maybe 2% of the total allowance so quite significant, with the remaining portion looking like 10%+ of that target 2t/y.

I take from this incidentally that we can probably keep our (efficient) dishwashers and washing machines and such mod cons and meet that target.

News

Sources/Links

• ELEXON: "The Balancing and Settlement Code Company".
• National Grid: Electricity demand - Last 7 Days.
• Estimating marginal CO2 emissions rates for national electricity systems, A.D. Hawkes (doi:10.1016/j.enpol.2010.05.053).
• British Wind Energy Association's UK Wind Energy Database (BWEA UKWED).
• Part L of the Building Regulations, Proposed New Editions of the Approved Documents L: Consultation (2008).
• Digest of UK Energy Statistics 2007 (DUKES) DBERR.
• UK Air Quality Archive: Atmospheric Emissions reports including GHG inventory.
• Carbon emissions factors and calorific values as used in the UK Greenhouse Gas National Inventory Report (2006) National Atmospheric Emissions Inventory.
• Guidelines to DEFRA's Greenhouse Gas Conversion Factors for Company Reporting DEFRA 2007.
• Carbon Emissions from CHP by AEA Energy & Environment for the 2007 NAEI / UK Greenhouse Gas Inventory.
• DBERR Fuel Mix Disclosure Table 2007/2008 (excluding 1.09 transmission and distribution loss factor (not to be applied to embedded generation)):
  

  Energy Source	g/kWh
  Coal	910
  Natural Gas	360
  Nuclear	0
  Renewables	0
  Other	610
  Overall average	480

  
  The 'natural gas' figure is assumed to be mainly CCGT, and thus excluding OCGT used heavily in balancing, ie frequency-response mode. Relative kgCO2/kWh OCGT:CCGT estimated to be 1.33:1, though possibly closer to 1.2:1.