LCA Engineering common mistakes
LCA Engineering · Understand the traps and how to avoid them.
Treating the functional unit as a physical unit (e.g., '1 kg of product' or '1 m² of flooring') and then comparing systems that deliver different service levels, such as comparing 1 kg of virgin aluminium with 1 kg of recycled aluminium without accounting for the different quality, yield, or service life.
Candidates carry over the reference flow from a familiar dataset or EPD and skip the step of asking what function the studied system actually delivers. The functional unit is quantitative and qualitative and must capture the service provided, not just the mass crossing the factory gate. When the reference flow is confused with the functional unit, the comparison becomes invalid even if the LCI arithmetic is correct.
What to do instead
Before modelling, write the functional unit as a sentence containing: the function (what is delivered), the quantity, the duration or service life, and the quality/performance requirement. Derive the reference flow from the functional unit by scaling the product system, and state explicitly in the goal and scope why the reference flow differs from the functional unit. For comparative studies, verify that both systems meet the same performance requirement and document the equivalence in the scope section.
Cutting off processes on the basis of mass alone (e.g., 'inputs below 1% of mass are excluded') without checking whether the excluded flows carry disproportionate environmental relevance, and without reporting the cut-off criterion, the excluded flows, and a sensitivity check on the result.
Mass-based cut-off rules are easy to apply and are taught as a shortcut, so candidates apply them mechanically. The problem is that small-mass flows can dominate certain impact categories: a catalyst or a process chemical at 0.01% of mass may drive toxicity or eutrophication impacts, and energy carriers can be low in mass but high in global warming potential. Candidates also forget that ISO 14044 requires the cut-off criteria (mass, energy, environmental significance) to be stated and justified.
What to do instead
Apply a combined cut-off on mass, energy, and environmental significance, and state all three criteria in the scope. List every excluded flow with its estimated contribution, and run a sensitivity analysis showing that including the excluded flows does not change the conclusions. Where a flow is small in mass but potentially significant in an impact category, include it and document why.
Using a generic secondary database process as a direct substitute for a specific foreground process without checking its geography, technology, and time representativeness, and without documenting the resulting data quality limitations in the interpretation.
Database processes are convenient and often the only data available, so candidates select the first plausible match by name. Database names hide assumptions: an electricity market process for a different country, a technology from a different decade, or a process with a different system boundary (cradle-to-gate versus gate-to-gate) can shift results substantially. Candidates also skip the data quality assessment because the dataset 'comes from a reputable database'.
What to do instead
For each dataset used, record the reference year, geographic scope, technology level, system boundary, allocation method, and the source. Compare these attributes against the goal and scope, and where they differ, either substitute a better dataset, adapt the dataset (e.g., adjust the electricity mix), or flag the mismatch as a data quality limitation. Use a structured data quality scoring scheme (e.g., pedigree matrix) and carry the result into the interpretation and uncertainty sections.
Choosing an LCIA method at random or defaulting to whichever method the software pre-selects, then reporting a single aggregated score without stating the method version, the impact categories covered, or which categories were excluded, and mixing midpoint and endpoint indicators in the same comparison table.
Software defaults are invisible, and candidates often do not realize that different LCIA methods use different characterization models, different time horizons (e.g., GWP100 versus GWP20), and different sets of categories. Mixing midpoint and endpoint results in one table invites invalid comparisons, and omitting the method version makes results irreproducible. Candidates also present a single score as if it were objective, ignoring the value choices embedded in weighting.
What to do instead
State the LCIA method and version explicitly (e.g., a specific characterization method and version), list all impact categories assessed and any excluded, and keep midpoint and endpoint results in separate tables. Justify the method choice against the goal and scope, and if a single score is presented, disclose the weighting set and treat it as one of several indicators, not the result. Report characterization results before normalization and weighting.
Applying allocation by a single arbitrary rule (usually mass or economic value) without first attempting to avoid allocation by system expansion or subdivision, and without testing the sensitivity of the results to alternative allocation choices.
Allocation is conceptually hard, and candidates reach for the simplest rule. ISO 14044 sets a hierarchy: avoid allocation by subdivision or system expansion, then use physical relationships, then other relationships such as economic value. Candidates also forget that allocation choices can reverse the ranking of options, especially in recycling, co-product, and multi-output systems, and that the chosen rule must be documented and justified.
What to do instead
Work through the ISO hierarchy explicitly: first try subdivision, then system expansion (including substituted products and their burdens), then physical allocation, then economic allocation as a last resort. State the allocation rule, the data used, and the rationale in the scope or inventory section. Run a sensitivity analysis with at least one alternative allocation approach and report whether the conclusions change. For open-loop recycling, state the chosen recycling allocation method and its consequences.
Confusing the various 'carbon footprint' boundaries and reporting a cradle-to-gate partial result as if it were a full product carbon footprint, or omitting the treatment of biogenic carbon, land-use change, and the removal/offset distinction from the reported figure.
Carbon footprinting standards and guidance distinguish cradle-to-gate, cradle-to-grave, and cradle-to-cradle scopes, and require specific treatment of biogenic carbon, direct and indirect land-use change, and the separation of removals from emission reductions. Candidates often take a gate-to-gate inventory number, label it a product carbon footprint, and omit the boundary statement. They also conflate offsets with reductions, which is a reporting error rather than a modelling preference.
What to do instead
State the system boundary explicitly in the report title and scope (e.g., 'cradle-to-gate, excluding use and end-of-life'). Apply the standard's rules for biogenic carbon and land-use change and disclose them, keep fossil and biogenic emissions in separate lines, and report removals separately from emission reductions. Never net offsets against the footprint without a clear, separately labelled disclosure. Cross-check that the declared boundary matches the declared use of the result.
Presenting LCI/LCIA results as point values without any uncertainty analysis, and when a sensitivity analysis is requested, varying only one parameter at a time and calling that a complete uncertainty assessment.
Software outputs look precise, and candidates treat the calculated number as the answer. One-at-a-time sensitivity shows local response but misses parameter interaction and the combined effect of data uncertainty. Candidates also fail to distinguish scenario, sensitivity, and uncertainty analysis, and do not connect data quality scoring to quantitative uncertainty ranges, which ISO 14044 expects in the interpretation phase.
What to do instead
Distinguish the three analyses in the report: scenario analysis (alternative system configurations), sensitivity analysis (contribution of individual parameters and choices), and uncertainty analysis (combined data uncertainty, e.g., Monte Carlo simulation or pedigree-based ranges). Report results as ranges or distributions where feasible, identify the parameters that dominate variance, and state explicitly whether the conclusions are robust across the tested variations. Link data quality scores to the uncertainty ranges used.
Describing an EPD as a certification of environmental superiority or as a comparative label, and confusing the roles of the PCR, the program operator, and the verification body; likewise treating ISO 14025 (EPD) and ISO 14067 (carbon footprint) as interchangeable.
EPDs are widespread in procurement, and candidates absorb marketing language about them. In fact an EPD reports declared environmental performance against a PCR under a program operator, with independent verification of the declaration, and it is not by itself a comparative assertion between competing products. ISO 14025 governs Type III environmental declarations, while ISO 14067 governs quantification and communication of a product carbon footprint; they serve different purposes and have different requirements.
What to do instead
State the roles precisely: the PCR defines the product-category rules, the program operator administers the scheme, and the verifier checks the LCA and the declaration against the PCR and the relevant standards. Describe the EPD as a verified, PCR-based disclosure of declared performance, not a ranking or a superiority claim. Keep the standards distinct in your answer: cite ISO 14025 for Type III declarations and ISO 14067 for product carbon footprint quantification and communication, and note where a carbon footprint may be embedded in an EPD.
Describing circular economy or eco-design strategies in generic terms ('design for recycling', 'increase recyclability') without quantifying the effect in the LCA model, and without accounting for the quality loss and actual end-of-life fate of the material in the studied region.
Circular economy vocabulary is familiar from policy discourse, and candidates reproduce slogans instead of modelling. A design change that nominally improves recyclability may not reduce impacts if collection rates are low, if the recycled material displaces virgin material only partially, or if downcycling reduces quality. Candidates also ignore that the end-of-life scenario is a modelling choice that must be stated and tested.
What to do instead
Translate each circular or eco-design measure into a concrete change in the inventory: material quantity, energy demand, lifetime, collection rate, recycling yield, and substituted product. Model the actual end-of-life fate for the relevant region and state the recycling allocation approach. Quantify the change in impact results relative to the baseline, and test alternative end-of-life scenarios in a sensitivity analysis. Report trade-offs across impact categories rather than a single improvement claim.
Conducting or interpreting a comparative assertion intended for public disclosure without a critical review, and treating an internal review or a single reviewer's check as satisfying the ISO 14044 critical review requirement.
Candidates underestimate when critical review is mandatory and what it must cover. ISO 14044 requires critical review for comparative assertions intended to be disclosed to the public, and specifies the review type (internal expert, external expert, or panel of interested parties) and its scope, including the goal and scope, inventory, impact assessment, and interpretation. A verification of an EPD is a related but distinct activity, and an internal check by a colleague does not meet the requirement for public comparative claims.
What to do instead
Identify in the goal and scope whether the study supports a comparative assertion intended for public disclosure; if so, plan a critical review by an external expert or a panel of interested parties, covering method, data, and conclusions. Document the reviewer's credentials, the review statement, and how comments were addressed. Distinguish this from EPD verification under a program operator, and state which applies to the deliverable.