Service Robot Fleet Emissions: Scope 2 Arithmetic and ESRS Datapoints
At a glance: A robot fleet is a small emissions source with a disproportionate reporting cost, because the numbers have to survive assurance. The question facility teams actually get asked is narrow: how many tonnes of CO2e does the fleet add, which scope does that sit in, and what evidence supports it. This guide does the arithmetic with worked examples, sets out the location-based and market-based methods and the double-count trap between them, lists the ESRS datapoints a fleet has to feed, and describes the monthly pipeline that makes the annual number defensible.
Why the Robot Line Is Bigger Than It Looks
At the level of a single robot, fleet emissions are small: a cleaning robot consuming 3 MWh a year is responsible for roughly 0.8 to 1.2 tonnes of CO2e on a typical European grid factor, and less on a low-carbon one. The reason the line still matters is not its size but its behaviour. It is an electricity load that doubles when a fleet doubles, it is often the fastest-growing load in a building, and it is metered, which means the number can be evidenced to a precision that heating and travel data rarely can.
That combination makes the fleet an attractive line for a sustainability team under pressure to show measured rather than estimated reductions. It also makes it vulnerable: a fleet line computed with a nameplate wattage and a rule-of-thumb utilisation will not survive an auditor's question about the source, and the correction in the following year looks like a restatement.
The good news is that a fleet is one of the few Scope 2 sources where the operator controls the measurement completely. Everything below follows from that, and the wider reporting architecture the fleet figure feeds into is set out in the ESG reporting framework guide.
Scope 1, 2 and 3: Where Robots Land
The reporting framework is set by the GHG Protocol, and the fleet's emissions split across all three scopes depending on who owns what and where the boundary is drawn. The table below is the mapping an operator should use, with the ownership dependency called out, because the same robot can sit in different scopes for a purchase and a RaaS arrangement.
| Source | Scope | Ownership condition | Typical size for a 20-robot fleet |
|---|---|---|---|
| Electricity consumed by the fleet | Scope 2 (purchased electricity) | Whatever the power purchase arrangement, this is purchased electricity | 15 to 25 tCO2e per year on a mid-carbon grid |
| Refrigerant from robot cooling, where fitted | Scope 1 | Only where the operator owns and operates the equipment | Usually negligible; some cleaning robots use sealed systems with no recharge |
| Fuel in an on-site service vehicle supporting the fleet | Scope 1 | Operator-owned or controlled vehicle | Depends on service model, not on robot count |
| Embedded emissions of the robots (manufacture) | Scope 3, category 1 (purchased goods) | Purchase or capitalised lease | 2 to 5 tCO2e per robot, dominated by the battery |
| Transport of robots to site | Scope 3, category 4 (upstream transport) | Any structure | 0.2 to 1 tCO2e per unit, sea freight dependent |
| Battery replacement and end-of-life | Scope 3, category 1 or 5 | Depends on who owns the pack | 0.5 to 1.5 tCO2e per pack replaced |
| Business travel of service engineers | Scope 3, category 6 | Any structure | Site-visit dependent, often larger than expected |
Two boundary decisions do most of the work. Under a purchase, the embedded emissions land in the buyer's Scope 3 category 1 in the year of purchase, which creates a spike every refresh cycle. Under a RaaS or operating lease, the same emissions sit with the provider, and the client reports only the electricity, which flattens the client's line substantially. Neither is more honest than the other, but the choice must be stated, because a reader comparing two operators without knowing the structure will draw the wrong conclusion.
The Arithmetic: kWh to tCO2e
Scope 2 emissions are electricity multiplied by an emission factor. That is the whole calculation, and everything that goes wrong with it happens at the two inputs. Consumption must come from metered data attributable to the fleet, and the factor must match the geography and the method.
A worked example makes the order of magnitude concrete. Take a 20-robot cleaning fleet with a measured annual consumption of 48 MWh at the dock sub-meter. Apply a location-based grid factor of 0.25 kgCO2e per kWh, which is representative of a mid-carbon European grid in 2026.
48,000 kWh x 0.25 kgCO2e per kWh = 12,000 kgCO2e = 12.0 tCO2e
Now the same fleet under a market-based method, on a contract that is 60 percent backed by renewable energy certificates at a residual mix factor of 0.55 kgCO2e per kWh for the uncovered portion and a supplier-specific factor of 0.02 for the covered portion:
48,000 x 0.4 x 0.55 + 48,000 x 0.6 x 0.02 = 10,560 + 576 = 11,136 kgCO2e = 11.1 tCO2e
The gap between the two methods on this fleet is under a tonne, because the grid factor is already moderate. On a coal-heavy grid with a location factor near 0.8 the same fleet produces 38.4 tCO2e location-based against a market-based figure that can fall below 5, and the method choice then dominates the entire result. That is why the method has to be declared alongside the number, not buried in a note.
| Input | Where it comes from | Audit expectation |
|---|---|---|
| Fleet consumption (kWh) | Dock sub-meter or per-dock energy field, 15-minute interval | 12 months of interval data with no gaps; estimated intervals flagged |
| Location-based factor | Published grid factor for the country and year | Named source, publication year, and the version used |
| Market-based factor | Supplier-specific factor plus residual mix for uncovered volume | Certificate evidence, retirement record, and the contract period it matches |
| Meter-to-invoice reconciliation | Utility bills for the same period | Explained variance between sub-meter total and billed site total |
Location-Based Versus Market-Based, and the Double-Count Trap
The GHG Protocol's Scope 2 guidance requires both a location-based and a market-based figure, reported side by side. The location-based figure uses the average emission factor of the grid where the electricity is consumed. The market-based figure reflects the instruments the reporting entity has purchased: supplier-specific tariffs, renewable energy certificates, guarantees of origin, power purchase agreements.
The trap is that both figures can be arithmetically correct and still be double-counted at portfolio level. If the building's landlord purchases renewable certificates for the whole site and the tenant operating the robot fleet also applies certificates for its own load, the same megawatt-hours are claimed twice. The same thing happens when a RaaS provider reports the fleet's electricity under its own market-based figure while the client reports the consumption under its location-based figure for the same site.
Three rules prevent it. First, decide at contract stage who reports the fleet's electricity: the site's electricity account holder, or the party that holds the certificates. Second, make the RaaS contract state explicitly whether the provider or the client counts the fleet's consumption, and put the same statement in the sustainability annexes of both reports. Third, hold the certificate retirement record centrally, so a certificate can only be claimed once, with a named reporting entity attached to it.
The double-count question is also a cost question. Certificates are bought for a purpose, and paying twice for the same claim is the most common waste in a Scope 2 programme. The interaction with the site's electricity cost is covered in the electricity cost guide, and the demand line that sits beside the energy line in the demand charge and peak-load guide.
What ESRS Actually Asks For
Under the Corporate Sustainability Reporting Directive, the European Sustainability Reporting Standards define the datapoints. A robot fleet touches a small but specific set, and mapping them once removes a scramble every reporting cycle. The table below is the set a fleet operator or a client reporting a fleet should expect to populate, with the practical source.
| Datapoint | ESRS reference | Where the fleet number comes from |
|---|---|---|
| Gross Scope 1 GHG emissions | E1-6, paragraph 44 | Only where the operator owns combustion sources; usually nil for a fleet |
| Gross Scope 2 location-based emissions | E1-6, paragraph 46 | Metered kWh times published grid factor |
| Gross Scope 2 market-based emissions | E1-6, paragraph 47 | Metered kWh times contractual instruments |
| Scope 3 category 1 purchased goods | E1-6, paragraph 51 | Embedded emissions per robot from the supplier or a spend-based estimate |
| Energy consumption total and share from renewables | E1-5 | Dock sub-meter total, split by contractual instrument |
| Energy intensity per net revenue | E1-5 | Fleet kWh divided by the reporting entity's net revenue |
| Transition plan actions and capex | E1-1, E1-3 | Fleet electrification and charging measures, with cost |
| Energy consumption outside the value chain | E1-5 | Only where the fleet operates on client sites and is not on the operator's meter |
The last row is the one fleets get wrong. A cleaning contractor whose robots plug into the client's building supply is consuming electricity that is not on the contractor's own meter. That consumption belongs in the contractor's "energy consumption outside the value chain" disclosure, not in its Scope 2, and the client may separately report it as its own Scope 2. Both disclosures are correct if the boundary is stated; only one of them is correct if the boundary is not.
Assurance: What the Auditor Will Ask to See
Limited assurance on a sustainability statement is now the norm for larger entities, and an auditor's tests on a fleet line are predictable. Expect a sample of months traced from the report back to the sub-meter and the utility bill, a check that the factor version matches the reporting year, a reconciliation of the fleet total against the site total, and a test of the certificate retirement records against the market-based figure.
Four artefacts make that pass quickly. A meter register listing every dock circuit, its meter identifier and its commissioning date. A consumption workbook with one row per interval or per month, with estimated values flagged and the estimation method stated. A factor register naming the source and version for each factor used, with the reporting year it applies to. And a certificate register with retirement dates, volumes and the reporting entity each certificate is assigned to.
The most common assurance finding on a fleet line is not a wrong number but an unexplained gap: a month where the sub-meter was offline and the value was interpolated without a note. Flagging that at the time costs nothing; discovering it during assurance costs a restatement.
Building the Monthly Data Pipeline
The difference between a fleet emissions figure that takes three weeks to produce and one that takes an afternoon is whether the data is collected monthly or annually. Five steps make it routine.
Step one, meter the docks. A sub-meter per charge-dock group, read at 15-minute intervals, exported monthly to a single store. If docks share a panel, meter the panel and subtract known loads, and record the subtraction method once so it does not change silently.
Step two, export the fleet software data monthly. Per-dock energy, charging events and robot status, into the same store. This is the dataset that lets a month be attributed to a specific robot, site or site area, which is what a client-level report requires.
Step three, apply the factors. Keep the factor register versioned, because grid factors are published annually and a mid-year update must not silently rewrite prior months. The peak-load and demand-charge guide explains why the same interval data is also the basis for the cost side of the report.
Step four, reconcile monthly. Compare the fleet sub-meter total to the site's billed consumption every month rather than annually. A monthly variance outside a stated tolerance is a meter or a mapping error, and finding it in month two is cheap.
Step five, report quarterly internally. A quarterly internal figure forces the pipeline to stay alive and gives the sustainability team a trend to explain before the annual cycle begins. The building-level integration for that reporting is covered in the occupancy sensor and BMS integration guide, and the broader reporting architecture in the ESG reporting framework guide.
One more design choice pays for itself: expose the same measured kWh in the fleet management platform that the operations team already uses. The fleet management platform is where charging schedules and duty cycles are set, so the sustainability number and the operational number then come from one source rather than two spreadsheets that drift apart.
Frequently Asked Questions
Are service robots Scope 2 or Scope 3 emissions?
The electricity a robot consumes is Scope 2 for whichever entity purchases that electricity. The emissions embedded in manufacturing the robot are Scope 3 category 1 for the buyer, or sit with the provider under RaaS. The same robot therefore appears in different scopes for different reporting entities, which is why the boundary has to be stated.
How many tonnes of CO2e does one cleaning robot produce per year?
At 3 MWh per year and a mid-carbon grid factor of 0.25 kgCO2e per kWh, one robot produces about 0.75 tCO2e a year. A 20-robot fleet at a higher duty cycle, around 48 MWh total, produces roughly 12 tCO2e location-based. The grid factor, not the robot, is the dominant variable.
Does charging robots off-peak reduce reported emissions?
Not on a location-based figure, because the annual grid factor does not change with the hour. On a market-based figure it can, if the tariff itself is time-differentiated and the off-peak product carries a lower supplier-specific factor. The operational saving from shifting load is real, but it lands on the cost line rather than the emissions line, which is why the energy and route optimization guide treats scheduling as a cost measure and metering as the emissions measure.
Do renewable certificates for the building cover the robots?
Only if the certificates are assigned to the consumption the robots represent and are not already claimed by another party at the same site. That allocation should be written into the lease or the service contract; otherwise the same certificates can be claimed by the landlord and the tenant, which is a double count.
What if the robots plug into the client's power, not ours?
Then the consumption is not on the operator's meter and should be disclosed as energy consumption outside the value chain rather than as the operator's Scope 2. The client may report the same electricity in its own Scope 2. Recording both, with the boundary stated, is the correct treatment.
Get a Quote
Tell us about your space and workload, we reply within 24 hours.