Scope 3 Category 3: Separate Fuel Supply, Purchased Energy and Network Losses
Original original synthetic illustrative energy-accounting desk scene. Bright illustrative energy-accounting workspace with a fuel record, utility bill, physical meter and clean industrial energy context; no real measurements or certification symbol. Spatial grouping, branches, proportions or directional paths carry the claim with concise labels.
Diagram text description
A company's gas bill and electricity meter can support several emissions calculations without representing the same emissions twice. Burning gas in its boiler belongs in Scope 1; producing and transporting that gas belongs in Scope 3 Category 3. Purchased electricity has a similar split between generation, upstream fuel supply and the energy lost before delivery. The quantity may come from one ledger, but each calculation needs its own factor boundary.
The GHG Protocol organises Category 3 into four activities. Fuel and energy end users generally examine Activities A, B and C. Utilities and energy retailers also examine purchased energy sold onward under Activity D. Begin by identifying which role the company plays, then separate the components before calculating them.
The examples below show how that separation works, why a loss-rate denominator matters and how to reconcile a small site's result. All numerical factors in the worked examples are fictional teaching inputs, not factors to use in an inventory. The method is useful because it turns an ambiguous life-cycle total into identifiable quantities and stages.
What belongs in Scope 3 Category 3?
The reporting company's purchased and consumed fuel and energy provide the starting population. Category 3 relates to production and delivery stages outside its direct fuel combustion and outside the generation emissions counted in Scope 2. The GHG Protocol identifies four activities, labelled A through D, each with a different applicability test and potentially different emission-factor components. A business should map those activities before selecting a calculation formula.
For a manufacturer burning natural gas in its own boiler, upstream extraction, processing and movement of that gas are an Activity A question. Its boiler combustion remains Scope 1. For an office using purchased grid electricity, upstream fuel supply to the generator is Activity B; the generation emissions assigned to the office's purchased electricity remain Scope 2. The energy generated but lost in transmission and distribution is an Activity C question. Activity D is different again: it concerns power purchased by a utility or energy retailer and then sold to end users.
The activity names are more useful than a single total because they reveal the missing data. Purchased fuel type and quantity, energy consumption by region, supplier information, loss rates and factor coverage are not interchangeable. A full life-cycle energy factor may include combustion and losses, while a factor labelled “upstream” may exclude one or both. Document the chosen component before multiplying any quantity.
For an end user, consumed fuel leads to Activity A, while consumed purchased electricity leads to B and C. Activity D concerns purchased power resold to end users by energy retailers; it is discussed separately below. The rails distinguish activity types rather than adding their quantities.
Diagram text description
Activity A: upstream production of fuel the company uses
Activity A covers extraction, production and transportation of fuels the reporting company consumes. Examples in the GHG Protocol include coal mining, gasoline refining, natural-gas transmission and biofuel production. The activity applies to fuel end users. It does not count combustion in company-owned or controlled equipment a second time; that direct combustion is Scope 1. Follow a staged method rather than applying one generic life-cycle number.
Reconcile the fuel population. Start with the fuel quantities and types already used in the Scope 1 inventory, then compare them with purchasing and site records. Record the year, unit, supplier or region and any estimated quantities. This establishes which fuel the company consumed without bringing combustion emissions into Category 3.
Identify the upstream factor component. The official chapter permits supplier-specific or average-data methods. Supplier-specific information can describe extraction, production and transport associated with the fuel purchased; an average factor estimates those stages per suitable consumption unit. Check the source's fuel type, technical boundary, geography and vintage. Neither “supplier-specific” nor “life cycle” is a guarantee that the component fits.
Exclude and document combustion. If a life-cycle factor includes both upstream and combustion stages, separate them before combining with the Scope 1 result. The official method discusses subtracting a combustion component where included, using factors with comparable time, technology and geography. Do not subtract two incompatible factors merely to create a precise-looking residual. Keep source documents and the treatment of any biomass or biofuel memo item visible for specialist review.
Follow the fuel from extraction, processing and delivery into combustion, then useful energy. The first stage represents upstream Category 3 Activity A. Combustion is a separate Scope 1 component for the end user. The final useful-energy label describes the output, not an additional inventory category.
Diagram text description
Activity B: upstream production of purchased electricity and heat
Activity B concerns the upstream extraction, production and transport of fuels used to generate electricity, steam, heating or cooling that the reporting company purchases and consumes. The combustion at the generator that produces the purchased energy is not the same component: for the energy end user it is accounted for in Scope 2. The GHG Protocol calls for disaggregated factors so the upstream stage can be placed in Category 3 without repeating generation emissions.
Use the quantities from the Scope 2 energy record as a starting point. Break purchased electricity, steam, heating and cooling down by supplier, grid region or country where the factor requires it. A utility-specific upstream factor can be appropriate when its boundary and period match the purchased energy. Otherwise, a representative grid, country or regional factor may be used under the official average-data method. A generic world-average factor applied to a diverse site portfolio without explanation can conceal meaningful supply differences.
Inspect what “life cycle” means in any supplier file. It may include fuel extraction, electricity generation, line losses or even other stages. The Category 3 chapter specifically warns that the upstream factor for purchased electricity must exclude generation combustion already represented in Scope 2. It also says transmission and distribution losses should be subtracted from an upstream factor if already included, because Activity C treats them separately. The calculation should retain a component map, not just a spreadsheet cell with a factor name.
The left stage covers upstream production of fuels used to generate purchased electricity. The right stage is generation associated with the purchased electricity. Activity C treats network losses separately. The arrow follows the physical energy chain; it does not add another reporting category.
Diagram text description
Activity C: turn network losses into an explicit quantity
Activity C concerns energy generated but lost in transmission and distribution before reaching the end user. Its emissions include upstream fuel supply and generation for that lost energy. Purchased electricity measured at the customer's meter is a different quantity. Establish how a published loss rate defines its denominator before combining the rate with meter data.
Consider a fictional network delivering 95 MWh to a site. If losses are 5% of energy entering the network, the incoming quantity is 95 ÷ (1 − 0.05) = 100 MWh. The lost quantity is therefore 5 MWh. Multiplying the delivered 95 MWh by 5% would instead give 4.75 MWh: that calculation treats the rate as a percentage of delivered energy, which is a different definition. The physical split is 100 entering, 95 delivered and 5 lost.
For the same physical example, losses expressed relative to delivered energy would be 5 ÷ 95 = 5.263%. Multiplying 95 MWh by that rate gives the same 5 MWh loss. This is a denominator demonstration, not an instruction to alter every published factor. A supplier or national dataset may already provide a loss-emissions factor per delivered kWh, incorporating its own conversion. In that case, use the specified activity basis rather than adding another loss adjustment.
With an illustrative life-cycle factor of 0.40 tonnes CO2e per MWh of lost electricity, Activity C is 5 × 0.40 = 2.00 tonnes CO2e. The 95 MWh delivered to the customer still has its own Scope 2 generation and Category 3 upstream calculations. Keep those three components separate. The relevant grid, year, factor stages and rate definition determine whether this structure fits a real dataset; the teaching values do not establish a representative grid intensity.
The fictional network input is 100 MWh. It splits into 95 MWh delivered and 5 MWh lost. The loss is 5% of input, or about 5.263% of delivered energy. The diagram illustrates the denominator distinction; it is not a universal network loss rate.
Diagram text description
Activity D: purchased power resold by energy retailers
Activity D covers the generation of electricity, steam, heating or cooling purchased by the reporting company and sold to end users. The GHG Protocol says this is particularly relevant to utility companies and energy retailers purchasing wholesale power, including from independent generators, for resale. It includes generation and upstream stages associated with the purchased-and-resold energy. It is not an automatic additional line for every business that buys electricity to run its own operations.
This distinction is easy to miss when the same corporate group both consumes energy and sells some onward. Separate energy used by the reporting company's facilities from purchased energy resold to customers. Identify the reporting entity and the transactions, then document whether the company is an end user for Activities A to C, an energy retailer for Activity D, or both for different volumes. The official chapter provides supplier-specific and grid-average options for this branch, tied to source and quantity of the energy purchased for resale.
Do not copy a utility's reporting example into an ordinary manufacturer's inventory. Equally, a company that sells excess power into a grid may need a specific applicability review rather than assuming Activity D never arises. Record the sale arrangement, power source, metered quantities and factor boundary; take a disputed electricity-market classification to a suitably qualified reviewer. The article is an accounting workflow, not a determination of a particular contract's regulatory status.
Start with a reconciled fuel-and-energy activity record
Establish the reporting population. Start with the Scope 1 fuel inventory and Scope 2 purchased-energy ledger for the same organisational boundary and year. Reconcile fuel volume, fuel type, electricity, steam, heating and cooling to bills, meters or purchasing records. Sites, suppliers and grid regions should be identified where they affect the available emission factors. Retain adjustments such as estimated meter periods or corrected invoice quantities rather than silently substituting a new total.
Assign each relevant component once. Mark fuel-combustion emissions as Scope 1 and the generation emissions of purchased energy as Scope 2. Then create separate Category 3 records for upstream purchased fuels, upstream purchased electricity or other energy, and T&D losses. Create an Activity D record only where the reporting company bought energy for resale. This map should make an item visible even when data are missing; “not calculated” is an exception to resolve, not a zero.
Reconcile units and ownership. A litre of vehicle fuel, a cubic metre of gas and a kilowatt-hour of purchased electricity need different activity data and factors.
Keep consumed energy apart from energy purchased for resale, and check whether a supplier's upstream number already covers transport or losses. The output of this step is an activity-by-activity population with source records, not yet a claim of accurate emissions. It gives factor selection a controlled starting point.
Choose an upstream factor that excludes the wrong component
The official Category 3 chapter describes supplier-specific and average-data approaches for the upstream fuel and energy activities. A supplier-specific number can be useful when the provider states which extraction, processing and transport stages it includes, the fuel or generation technology, geography, time period, unit and allocation. An average factor can be defensible where individual supplier data are unavailable, provided it represents the relevant fuel, energy carrier and region. Choose the method based on actual source fit and materiality, not on the marketing label “primary data.”
For Activity A, the upstream component must exclude the combustion already in Scope 1. For Activity B, it must exclude electricity or heat generation combustion already in Scope 2. Activity C needs the relevant life-cycle factor and loss rate without recapturing a loss already embedded in another component. Activity D uses factor coverage for purchased power sold onward, which can include generation as well as upstream stages. The same factor must not be pasted into all four rows simply because each row mentions energy.
An upstream-only factor may match Activity A or B after its unit and activity match are checked. A combined factor containing combustion or generation needs a documented stage breakdown before use here. The two rows are alternative factor situations, not steps in one calculation.
Diagram text description
When a supplier gives only an all-in factor, ask for the underlying components or seek a documented secondary factor that matches the desired component. If a defensible split is not possible, retain the uncertainty and escalate the choice. A detailed but opaque supplier factor is not necessarily more appropriate than a transparent representative factor. Keep the source file and version, and record what the factor does not cover as carefully as what it includes.
A component map is especially useful when two supplier files use similar names. One may contain upstream production only; another may combine that stage with combustion or generation. Mark the included stages before selecting a column for the calculation. If the source cannot be separated, obtain a documented component rather than subtracting an unrelated factor merely because the units match.
A worked site calculation: three components, one energy ledger
Take a fictional site that consumes 10,000 litres of fuel and receives 95 MWh of electricity during one year. Assume that the fuel factor is 0.50 kg CO2e per litre for upstream production and delivery only. Assume a separate electricity upstream factor of 0.050 tonnes CO2e per delivered MWh, excluding generation and network losses. These are invented values chosen to make the units visible.
Table 1: scroll horizontally if needed.
| Category 3 component | Activity and factor | Illustrative result |
|---|---|---|
| A: upstream purchased fuel | 10,000 litres × 0.50 kg CO2e/litre | 5,000 kg = 5.00 tonnes CO2e |
| B: upstream purchased electricity | 95 MWh × 0.050 tonnes CO2e/MWh | 4.75 tonnes CO2e |
| C: lost electricity | 5 MWh lost × 0.40 tonnes CO2e/MWh | 2.00 tonnes CO2e |
| A + B + C | Separate non-overlapping components | 11.75 tonnes CO2e |
The Activity C quantity follows the preceding 5%-of-input network example. Activity D is not applicable to this site's assumed operations because it buys no energy for resale. Its fuel combustion and purchased-electricity generation remain in Scope 1 and Scope 2 respectively, outside this Category 3 subtotal. The table does not calculate those two scopes.
The most useful review is to follow the dimensions through the arithmetic. A factor in kilograms needs conversion before addition to values in tonnes. A delivered-MWh upstream factor cannot be used as though it described network input. A combined generation-and-upstream factor cannot replace the upstream-only factor in row B without changing what the row measures.
Now change only the upstream fuel factor from 0.50 to 0.60 kg per litre. Activity A becomes 6.00 tonnes and the Category 3 subtotal becomes 12.75 tonnes, an increase of 1.00 tonne. Fuel consumption has not increased. That difference is a factor change, so it should not be described as a deterioration in the site's operating efficiency. Keeping activity and factor revisions separate makes that distinction straightforward.
Horizontal bar lengths are proportional to 5.00 tonnes upstream fuel, 4.75 tonnes upstream purchased electricity and 2.00 tonnes associated with lost energy. Their sum is 11.75 tonnes CO2e. Scope 1 combustion and Scope 2 generation are outside this subtotal; the factors are fictional.
Diagram text description
Category 3 evidence checklist for a reviewer
A reviewer should be able to move from the Category 3 total to each energy activity and then to its supporting meter, bill, supplier statement and factor source. The following checks do not certify the number; they expose the important classification and data decisions before a report is approved.
Table 2: scroll horizontally if needed.
| Review item | Question to answer | Evidence to retain |
|---|---|---|
| Activity | Is this upstream fuel, upstream purchased energy, a T&D loss or power bought for resale? | Activity map and reporting-entity decision |
| Boundary | Which combustion or generation amount is already in Scope 1 or 2? | Scope 1/2 inventory cross-reference |
| Quantity | Does the fuel or energy amount cover the same year, site and unit as the factor? | Bill, meter or fuel ledger and conversion |
| Factor | Which stages, geography, technology and period does it represent? | Original factor file and component map |
| Losses | Are T&D losses already embedded, and what rate was selected? | Supplier or grid loss source and adjustment |
| Review | What remains estimated or unresolved? | Exception owner, version and reviewer decision |
One material energy record should not appear as both an upstream purchased-energy component and an unqualified “all-in” Scope 2 add-on. A useful final check is to trace a sample of high-energy sites through every component and the relevant Scope 1/2 cross-reference. If a component has no source or no clear reporting boundary, keep it flagged for method review rather than hiding it in a residual.
Use the reconciliation to choose the next improvement
Once the components reconcile, look at which uncertainty could change a decision. In the fictional site, upstream electricity contributes 4.75 tonnes. If the team has only a broad regional upstream factor, obtaining a better-matched supplier or grid dataset could matter more than polishing a small fuel record already supported by complete invoices. Materiality helps direct the next data request; it does not make a missing activity disappear.
Try two simple sensitivity checks. A 10% increase in the upstream electricity factor changes row B by 0.475 tonnes. A 10% increase in the upstream fuel factor changes row A by 0.500 tonnes. Both are appreciable compared with the 2.00-tonne loss component, but neither calculation identifies the actual uncertainty of those factors. They show how responsive the subtotal is to a defined change. Use real source-quality information to decide which scenario is plausible.
The energy team can then divide the work. Site owners resolve quantities and missing periods. The person maintaining emission factors checks units, geography and included stages. The inventory owner decides how a changed method affects comparative reporting. A useful next meeting has a concrete question, such as whether the available electricity factor excludes network losses, rather than a request to “improve the carbon data.”
Use those responsibilities to close the highest-priority exception first. Record the changed input and compare its effect with the sensitivity result. The next site can then reuse the resolved method while retaining its own quantities and factor match.
Conclusion: make each component explainable
Category 3 becomes manageable when each energy quantity has a defined job. Use the four-activity framework to separate upstream fuel, upstream purchased energy, network losses and energy bought for resale. Start with the same energy population used elsewhere in the inventory, but use factors for the additional stages being measured.
The worked site reaches 11.75 tonnes CO2e only under its stated fictional factors and loss assumptions. Its value is the calculation structure: every subtotal has a quantity, denominator and boundary that another person can follow. Apply that structure to one real site next. A resolved factor boundary or corrected meter period is a more useful improvement than adding decimal places to an unexplained total.
Company carbon accounting
