Defensible Assumptions When Primary EPD Data Is Missing
Missing plant meters. Unknown fastener specs. A supplier who replies next week, not today. EPD work still has to ship. The trick is using modeled or secondary data in a way a verifier can test quickly and a specifier can trust. Here is the playbook that keeps schedules tight while keeping claims defensible, not wishful.
Start with what “defensible” really means
A defensible assumption is transparent, time‑stamped, and tied to a credible source. It is also conservative enough that a small error does not flip the story. Think of it like using a calculator in an exam. The teacher does not care that you used one. They care that you showed your work and keyed in the right formula.
Where generalization is usually acceptable
Minor bill‑of‑materials items like generic fasteners, labels, or packaging spacers can use representative datasets when their total mass and impact are small compared to the product core. So can identical or near‑identical components sourced from multiple approved vendors, if the spec and process are the same. Note the rule of thumb, then keep the door open for a later swap once purchase orders or batch tickets arrive.
Where you should not generalize
High‑impact inputs need evidence, not vibes. Aluminum is a textbook case. Primary aluminum averages about 16 t CO2e per tonne globally while recycled aluminum is roughly 0.5 t CO2e per tonne. That is a swing of more than thirty times, which means you must anchor claims in mill certs or recycled content declarations (International Aluminium Institute, 2024) (IAI, 2024). Electric motors and drives are another trigger category because duty cycle and efficiency class shift A3 energy models meaningfully.
Modeling plant energy without perfect meters
If sub‑metering is missing, build a bottom‑up model from nameplate power, load factors, and run hours. Then map electricity to the right grid. Carbon intensity can vary by an order of magnitude across U.S. regions. Recent EPA data show Washington around 0.06 kg CO2 per kWh and Kentucky near 0.79 kg CO2 per kWh for delivered electricity. Choosing the wrong grid is not a rounding error, it rewrites A3 results (EPA eGRID, 2024) (EPA eGRID, 2024).
Supplier data that is good enough to defend
Ask for documents that survive an audit trail. Specification sheets with alloy grades and coatings. VOC content disclosures. Recycled content letters tied to delivery dates or heat numbers. If only generic info arrives, mark it as provisional and set a review date. That small flag tells a verifier you knew the limits and planned the fix.
Want to navigate EPD creation challenges more effectively?
Follow us on LinkedIn for insights that help you make defensible assumptions and stay competitive in your projects.
Separate measured from modeled everywhere
Keep two columns in your working LCA model. One for measured plant data. One for modeled or library values. Tag each row with the evidence source and the person who entered it. If a number changes, keep the old value in a change log with a short reason. This is boring in the best way and it saves days during verification.
Practical guardrails the team can live with
Adopt a short assumption note that always answers five questions.
- What is the assumption and the exact value used
- Why it was needed and which PCR rule it maps to
- The source and year of the dataset
- Who approved it internally
- When it will be revisited or replaced
Use the same template for every product family so nothing slips in teh rush.
Validate the big rocks fast
Run one sensitivity screen before you fall in love with the numbers. Nudge electricity by plus or minus 20 percent. Swap a key material between two realistic suppliers. If the rank order of contributors changes, you need better evidence. If results barely move, the assumption is likely fit for purpose until renewal.
Make the verifier’s job easy
Label any generalization directly in the EPD background report and call out the path to primary data. Include the filename of the supporting document in parentheses right after the assumption. Verifiers do not need more words. They need less hunting.
Keep the schedule, keep the trust
Defensible assumptions are not shortcuts. They are scaffolding that lets a team move while the last measurements arrive. Use them where risk is low. Replace them where impact is high. And always show your work. That is how EPD projects stay fast without becoming fragile, and how numbers earn their place in specs and bids.
Two numbers worth bookmarking
Aluminum’s GHG spread between primary and recycled production is large, which is why recycled content letters matter so much in bill of materials reviews (International Aluminium Institute, 2024) (IAI, 2024). U.S. grid intensity differences can easily overshadow small modeling tweaks, so plant location mapping should be one of the first checks, not the last (EPA eGRID, 2024) (EPA eGRID, 2024).
Frequently Asked Questions
How should manufacturers document assumptions in EPD models to avoid verifier pushback
Create a standard note for every assumption that records the value used, the reason it was needed, the matching PCR rule, the source and year, the approver, and the planned revisit date. Keep measured and modeled data in separate columns and maintain a change log that preserves superseded values.
When is it acceptable to generalize across suppliers in an EPD
When components are functionally identical and produced to the same specification and process. Keep evidence like spec sheets and approved vendor lists. Mark the entry as provisional until you receive supplier‑specific declarations or mill certificates for high‑impact inputs.
Which inputs usually require primary evidence rather than secondary data
High‑impact materials such as aluminum, steel, cement binders, resins, and electric motor systems. These can dominate A1 to A3 impacts. For aluminum in particular, the gap between primary and recycled routes is more than thirtyfold on average, so recycled content must be documented with supplier declarations (IAI, 2024).
About the Author
Walker Ryan is a climate-tech entrepreneur focused on driving industrial decarbonization through better data. As the founder and CEO of Parq, he helps manufacturers generate high-quality, third-party–verified carbon disclosures at scale—accelerating a traditionally slow and expensive process. Before starting Parq, Walker led over $200 million in sustainability-focused investments as VP of Strategy & Growth at ReStream Solutions, following earlier experience in investment banking at Deutsche Bank. He brings a rare mix of capital markets expertise and hands-on sustainability knowledge to tackling the infrastructure of industrial emissions.
