The plant data LCA teams actually need
Plants juggle throughput, uptime, and safety. Then a request for “primary data for the EPD” lands and everything slows. Here’s the short, practical list LCA practitioners truly need from manufacturing sites, plus which teams usually hold it and when a small, PCR‑acceptable assumption is fine so projects dont stall.


Start with one clear reference year
Pick a 12‑month window and stick to it. LCAs and EPDs are built on a single reference year so meters, invoices, and production logs line up cleanly. For brand‑new lines, a shorter prospective window can work if it is transparently flagged and replaced once a full year is available.
The standard plant data pack
This is the core set most verifiers expect for A1–A3. Keep it site specific and annual.
- Annual electricity use by meter or site (kWh) and procurement type (grid, on‑site, certificates).
- Fuels used on site by type and quantity (natural gas, LPG, fuel oil, biomass). Include backup boilers.
- Water in and wastewater out by source and discharge route. Note any treatment chemicals.
- Annual production volumes for each product family and co‑products. Declare units.
- Process losses, rework, trim, and off‑spec material as percentages or mass.
- Packaging and consumables used in manufacturing, not just outbound packaging.
- Hazardous and non‑hazardous waste by stream and destination.
- Process gases and refrigerants, including leak make‑up.
If a data point is missing, document the gap and an interim assumption. The reviewer’s first check is whether every line item is traceable to a reasonable source.

Win A $50 Amazon Gift Card in One Click!
Enter weekly raffle in one click • Help us get to know our readers and improve!
Who usually holds the numbers
Utilities or maintenance own meters and boiler logs. Finance or ERP administrators hold production and yield. EHS tracks waste, wastewater, and chemicals. Procurement has energy and material invoices. Quality owns scrap and rework rates. Warehouse or logistics knows packaging. Name the owners early so nothing bounces.
Process losses and workable assumptions
Exact scrap rates are best, yet many PCRs allow a small, conservative default when meters cannot separate trim from product. A practical ceiling often used in construction PCR work is 2 percent for undifferentiated process loss, provided the assumption is justified and verified against the rulebook for your product. There is no universal default across programs, so confirm in your applicable Part A and Part B.
Some PCRs even prescribe category‑specific figures. Structural steel fabrication uses an industry overage rate of 7.71 percent for material required to ship one metric ton of finished steel, set by the sector’s PCR and survey data (AISC, 2025) (AISC, 2025). Use these published anchors when they exist and disclose any departures.
Electricity and fuels move the needle
Grid power is often the single biggest lever in A3. The U.S. national average total output emissions rate was 823.1 lb CO2 per MWh in 2022, with wide variation by subregion, per the latest EPA eGRID update published in 2025 (EPA eGRID, 2025) (EPA eGRID, 2025). That variance is why practitioners ask for exact kWh by site and the procurement details. Small metering errors can ripple into noticeable GWP shifts.
Water and wastewater without the maze
Report total intake by source and total discharge by destination, then call out on‑site treatment. Where precise flows are pooled, triangulate with pump hours, tank volumes, or utility bills and state the method. Nationally, the last full USGS estimate put self‑supplied industrial withdrawals at about 14.8 billion gallons per day, roughly five percent of total withdrawals, which underscores why regulators watch these numbers even if your process is not water‑intensive (USGS, 2018) (USGS, 2018).
Keep your rulebook current
Data acceptance depends on the PCR and its version. The main construction PCR at the International EPD System moved to version 2.0.x in 2025 and is scheduled to remain valid until 2030‑04‑07, with transition guidance published by the program operator (EPD International, 2025) (EPD International, 2025). Matching your assumptions to the rule in force saves days in verification.
Make verification painless
Attach evidence to every figure. For energy, include invoices and meter screenshots. For fuels, keep delivery tickets and boiler logs. For scrap, export the yield report and note how regrind is handled. Put all assumptions in one short memo with sources and any calculation steps. Reviewers reward clarity.
Why this matters commercially
Specifiers often face modelled penalties when a product lacks a product‑specific, third‑party verified EPD. Supplying clean, site‑level data gets credible declarations out faster so teams can pursue projects without last‑minute compliance detours. That speed cushions margins because bids compete on performance, not just price.
The one‑page handoff
Write down the standard list, the data owners, the reference year, and any PCR‑allowed defaults you intend to use. Share it before the first data pull. When everyone sees the same checklist, EPD work stops feeling like paperwork and starts feeling like a production run you can ship on time.
Frequently Asked Questions
Which plant data points are most often missing in first submissions for LCAs and EPDs?
Scrap and rework percentages, packaging used in manufacturing, and refrigerant top‑ups. These are rarely in the ERP by default, so flag them early.
Is a 2% process‑loss assumption always accepted in EPDs?
No. Some PCRs allow a small default loss when exact metering is impractical, but limits and documentation requirements vary. Treat 2% as a cautious placeholder only after checking your PCR language.
Who should own electricity data if multiple tenants share one meter?
Facilities should own the primary meter, with a sub‑metering plan or allocation method agreed in writing with other tenants. Document basis and apply it consistently across the reference year.
How do we handle a new production line with only three months of data?
Use a prospective dataset clearly labeled with the period and assumptions, then commit to update once a full 12 months exist. Many verifiers accept this when disclosed and justified.
