Life Cycle Assessment of Waterproof Bags: Method, Hotspots and Why Studies Disagree

The disagreement is in the choices, not the data

A life cycle assessment is not a measurement of a product. It is a model of a product system, and the model has to be told what question it is answering. When a buyer asks for "the environmental impact of this bag", they believe they are asking for a property of the object, like its weight. What they receive instead is the output of a chain of conventions applied to a chain of data, and the conventions carry more weight than any individual number in the inventory. That is why the waterproof bag impact figures quoted across the industry are so widely scattered, and why the scatter is mostly explainable rather than suspicious.

The standards that govern this work — ISO 14040 for principles and framework, and ISO 14044 for requirements and guidelines — are explicit about this. They require the goal and scope to be defined before inventory work begins, they require the functional unit to be stated and justified, and they require a critical review whenever results are disclosed publicly for comparative purposes. Those requirements exist precisely because without them the same product supports any conclusion the commissioner prefers.

The practical consequence for a buyer is that the first question to ask about any LCA is not "what is the number" but "what is the question". A study answering "which material has lower cradle-to-gate impact per kilogram" will not answer "which bag should I buy", and it is entirely normal for the two to give different rankings. Reading an LCA well is mostly a matter of checking that the stated question matches the decision you are making.

It is worth stating what this article is not. It is not a set of impact figures to quote. Any specific figure depends on the electricity grid, the material supplier, the transport mode and the assumptions chosen, and quoting someone else’s figure into your own report is the single most common way LCAs get misused. What transfers is the reasoning, and that is what follows.

The functional unit decides the argument

The functional unit is the quantified performance the study normalises everything to. It is the denominator of every result, and changing it can reverse a ranking without changing a single data point. This is the most consequential and most under-scrutinised line in any LCA of a soft goods product.

The bottom row is the trap. Per-kilogram comparisons are the standard output of material databases and are frequently presented as product results. They are not, because they ignore that a heavier bag may last three times as long and protect three times as much per trip. A 900 g welded backpack that serves nine years has a lower per-year impact than a 400 g coated pack replaced every two, while the per-kilogram figure says the opposite. Both numbers are correct; only one answers the buyer’s question.

For this product category the most defensible unit is impact per year of service, or per litre of protected volume per year where capacities differ, because waterproof bags are durable goods bought to provide a service over time. That requires an assumption about service life, which brings the hardest problem in the whole exercise into the open — and which is why the use phase is treated separately below.

A useful discipline when writing a brief: state the functional unit in one sentence and then check that every alternative being compared can be expressed in it. If an alternative cannot be expressed in the chosen unit — for instance a repair service, which has no natural per-year denominator — the comparison is not valid as framed, and no amount of data will fix that.

System boundary: what is inside and what has been cut

The boundary describes which life cycle stages and which processes are included. Three conventions appear in practice, and the choice between them determines whether a result is comparable to anything else and whether it can support a public claim.

• Cradle-to-gate: raw material extraction through the finished product leaving the factory gate. Excludes distribution, use and end of life. Common in supplier declarations because it is the part a manufacturer can actually measure.
• Cradle-to-grave: adds distribution, retail, use, maintenance and end of life. Required for any whole-product comparison, and dependent on assumptions in every added stage.
• Gate-to-gate: a single process step only, typically assembly. Useful for internal improvement work and useless for product claims, though it is sometimes published as if it were more.
• Cut-off rules: processes below a stated mass, energy or environmental relevance threshold may be excluded, typically 1% of mass or energy, with the excluded total required to stay below about 5%. These rules are legitimate and must be stated, because they are where small but toxic inputs quietly disappear.

For waterproof bags the cut-off rule deserves specific attention, because the substances with the highest per-kilogram impact are present in the smallest quantities: coatings, inks, adhesives, primers and durable water repellent finishes. A study that excludes everything below one per cent of mass can legitimately omit the entire chemical package, which is where a regulator or a non-governmental organisation is most likely to take interest. If the goal includes chemical impact or restricted substance exposure, the cut-off rule has to be set on relevance rather than mass.

Capital goods are the other boundary decision that quietly matters. Whether the welding machines, the factory building and the tooling are included changes results by a small amount for a high-volume product and by a great deal for a short run. Both treatments are defensible under ISO 14044 provided they are stated and applied consistently to every alternative in the comparison. What is not defensible is applying one treatment to option A and the other to option B.

Inventory data quality: primary, secondary and the gap between them

The inventory is the list of inputs and outputs — kilograms of film, kilowatt-hours, litres of water, grams of solvent, tonne-kilometres of freight. Where those numbers come from determines how much the result can be trusted, and in this industry the answer is usually a mixture.

Primary data means measured at the specific facility for the specific product: metered electricity per production batch, weighed material consumption per order, actual waste streams. It is expensive to collect, requires cooperation from the supplier, and is the only kind of data that can distinguish one factory from another. Secondary data means taken from a database such as ecoinvent or a national inventory: it represents an average process, often in another country and another decade, and it is entirely appropriate for upstream stages nobody can measure, such as polymer production.

The practical grading that works when reviewing a report: check whether the stages that dominate the result use primary data, and accept secondary data for the long tail. If the hotspot is film production and film production is modelled with an average European dataset while the film is actually made in Asia on a different grid, the result is not wrong so much as uninformative. If the hotspot is assembly electricity and assembly electricity is metered, the study is worth reading.

Data quality assessment is formalised through pedigree matrices and scoring systems that grade each flow on reliability, completeness, temporal correlation, geographical correlation and technological correlation. The UNEP Life Cycle Initiative publishes guidance on applying these, and their use is what separates a defensible study from a spreadsheet. Ask whether a data quality assessment was performed at all; its absence is a stronger signal than any individual figure.

One specific data trap in this category: the bill of materials. Studies frequently model "the fabric" as a single flow, when the real input is a laminate of a woven substrate, an adhesive and a film, each with different upstream burdens. A fabric specification of 300 g/m² tells you nothing about how much of that is polymer, coating and adhesive, and the difference between those three is large. Getting the laminate composition from the material supplier is the single highest-value data request in the whole exercise.

Allocation: the rule that quietly decides recycled content

Most industrial processes produce more than one thing, and the burdens of that process have to be split between them. How that split is done is called allocation, and for recycled content it is not a technicality — it determines whether recycled material appears to have near-zero impact or merely reduced impact.

The two conventions are the cut-off approach and the avoided burden approach, also called system expansion. Under cut-off, recycled material enters the product carrying no burden from its previous life; the recycling process burdens are assigned to the product that becomes waste. Under avoided burden, the product takes on the burdens of collection and reprocessing and receives a credit for the virgin material it displaced. Both are permitted, both are used, and they can differ substantially in the result for a high-recycled-content bag.

This matters for a specific reason in this product category. Recycled polyester and recycled TPU carry a marketing premium, so a brand has a commercial incentive to use the convention that flatters them most. The defence against this is simple: ask which convention was used, and check that the same convention is applied to the virgin alternative. A comparison where the recycled option uses cut-off and the virgin option carries full upstream burden is comparing two different accounting systems, and the resulting percentage means nothing.

There is a second allocation question at end of life that is even more consequential, because it can change the sign of a result. If incineration with energy recovery is credited for the electricity and heat it displaces, and the displaced energy comes from a carbon-intensive grid, incineration can outscore recycling on climate change while losing badly on resource depletion and on every other impact category. That is not a paradox; it is what happens when a multi-category question is answered on one axis. Our article on end-of-life routes for waterproof bags covers what physically happens to these products, which is the input this modelling choice needs.

Impact assessment: the same inventory, different scores

Once the inventory exists, it has to be converted into impact. This is done by characterisation: each emission is multiplied by a factor representing its potency in a given category, and the results are summed. Several established methods do this with different scope and different assumptions, and the choice of method is the third major source of divergence between studies.

• Climate change: the most reported and best harmonised category, expressed as kilograms of carbon dioxide equivalent. Differences between methods here are now relatively small for fossil emissions, larger for land use and biogenic carbon.
• Resource depletion: fossil and mineral scarcity. Highly method-dependent, and the category most likely to move when recycled content is involved.
• Ecotoxicity and human toxicity: large model uncertainty, strongly influenced by whether the chemical package in coatings and inks was included at all.
• Water use: increasingly reported as scarcity-weighted rather than volumetric, which changes rankings between materials substantially.
• Land use and eutrophication: relevant when bio-based or natural fibre components enter the comparison, and capable of reversing a bio-based material’s advantage.
• European Product Environmental Footprint: a prescribed method intended to make comparisons reproducible, which is why it is increasingly requested in European tenders.

The European Commission’s Environmental Footprint reference platform is the most developed attempt to standardise this, and where a buyer wants comparability, specifying an Environmental Footprint compliant method is the most effective single instruction available. Standardising the method removes one of the four sources of disagreement and leaves the other three visible, which is progress.

A caution about single-score results. Aggregating categories into one number requires weighting, and weighting is a value judgement about how much climate matters relative to toxicity relative to water. There is no technical answer, which is why single scores should never be published without the underlying category results and the weighting set used.

Hotspots: where the burden actually concentrates

For all the methodological complexity, the empirical finding across studies of durable soft goods is stable and useful: upstream material production dominates, and everything downstream is smaller than intuition suggests. The ranges below reflect that pattern with the confidence attached, which is the part that matters for deciding what to work on.

The action implication is blunt and worth internalising: for this product, material specification and weight reduction are the levers, and factory electricity is not. That is counter-intuitive because the factory is the visible part of the supply chain, and because renewable electricity at the assembly site is easy to buy and easy to communicate. Both are worth doing, but a buyer who spends their effort on the 8% and ignores the 65% has optimised the part of the system they could see.

This is also why material weight work pays disproportionately. Our guide to lightweight construction and weight optimisation describes where mass can be removed without losing function, and the durability trade-off is the constraint: removing mass that shortens service life moves the use-phase denominator in the wrong direction and can make the per-year result worse. Weight reduction is only a win if the product still lasts.

The use phase nobody models honestly

Use phase is where LCAs of durable goods lose credibility, because it is modelled with assumptions rather than data and the assumptions are frequently chosen without evidence. For a waterproof bag the candidate use-phase burdens are washing, drying, user transport of the bag, and repair. Of these, washing and service life dominate, and both are guesses.

The arithmetic is worth doing once. A warm machine wash followed by tumble drying carries an energy burden in the range of a fraction of a kilogram to over a kilogram of carbon dioxide equivalent per cycle depending on the grid and the dryer. If a study assumes the bag is washed six times over its life and the real number is zero, the use phase figure is overstated by a large multiple; if the real number is twenty, it is understated by the same logic. Since the use phase is small relative to material production, this rarely reverses a ranking — but it is quoted often enough to be worth checking, and it is the number most likely to be challenged.

Service life is the more serious issue, because it is the denominator of the most decision-useful functional unit and it is almost never measured. Nobody in this industry has good field data on how long bags actually last; there is warranty return data, which is biased toward early failure, and there is anecdote. The honest treatment is to run the comparison at two or three service life values and report the range, which is exactly what a sensitivity analysis should do. A study that reports one service life number with no range has buried its largest assumption.

There is a commercial dimension here too. Extending service life is the most effective impact reduction available for a durable good, because it divides the whole upstream burden by a larger number of years. That makes repairability, spare parts availability and honest durability testing environmental interventions, and our article on predicting durability from accelerated ageing is the technical basis for claiming any of it.

End-of-life modelling and the credit question

End of life is a small number with a large argument attached, and the argument is about credit. When a product is recycled, the question is whether the product system receives credit for the material it returned and for the virgin material avoided. When it is incinerated with energy recovery, the question is whether it receives credit for the energy displaced. How these credits are calculated can move a result enough to change the ranking between two disposal routes, and it is the most common source of implausible conclusions.

The guidance that keeps this honest is to model what actually happens rather than what is desired. If the real destination for a laminated waterproof shell in the target market is residual waste or energy recovery, then modelling closed-loop recycling because it is the aspiration produces a result that describes a different world. Regional variation is large, so a study covering several markets should use several end-of-life profiles rather than one average.

The second discipline is to report burden and credit separately before netting them. A net end-of-life figure of minus 0.4 kg tells you nothing about whether that came from avoided virgin polymer or from displaced grid electricity. Separating them lets a reader apply their own judgement, and it prevents the most common misuse of LCA results in marketing: treating an accounting credit as a physical reduction that has already happened.

Finally, end of life is where the physical reality described in the recycling literature and the modelling conventions most often part company. A study that credits recycling at a 90% material recovery rate for a welded multi-polymer laminate is modelling a process that does not exist at commercial scale. Sanity-check the recovery rate against what a reprocessor would actually accept, and the result will usually be more defensible and less favourable.

Why two published studies disagree: five specific causes

When two credible studies reach opposite conclusions about the same product, the cause is almost always one of the following five. Running through them in order takes ten minutes and resolves most apparent contradictions.

The diagnostic procedure is mechanical. Put both studies side by side and write down, for each: the functional unit, the boundary, the allocation rule, the service life and the electricity dataset. If any two of those differ, the studies are not comparable and neither contradicts the other — they are answering different questions. If all five match and the results still diverge, then there is a genuine data or modelling disagreement worth investigating, and that is the point at which a critical review becomes necessary.

It is also worth checking who commissioned each study and whether a critical review was performed. Environmental Product Declaration programmes and ISO 14044 both require review for public comparative claims, and a reviewed study carries a materially higher burden of credibility than an unreviewed one. This is not about doubting integrity; it is about the fact that unreviewed work is not required to survive hostile reading, and much of it does not.

Uncertainty, sensitivity and how far a result can move

Every LCA result has an uncertainty range, and a result presented without one should be treated as incomplete. Uncertainty enters through data variability, model assumptions and methodological choices, and it is not uniform across the inventory: some flows are known within a few per cent, others within a factor of two.

• Parameter uncertainty: variability and measurement error in individual flows, commonly quantified by Monte Carlo simulation over distributions rather than point values.
• Scenario uncertainty: choices such as transport mode, service life and end-of-life route, which are not distributions but discrete alternatives and are handled by scenario analysis.
• Model uncertainty: the structure of the impact method itself, which cannot be quantified and is handled by running more than one method.
• Completeness uncertainty: flows excluded by cut-off rules, which is why the cut-off statement matters for interpretation as well as for transparency.

A good sensitivity analysis answers the question a decision-maker actually has: which assumption, if wrong, would change my decision. Practically, that means varying the top three contributors and the service life, and reporting the outcome as a range with the switch-over point identified. If a comparison between material A and material B flips when service life moves from five to six years, that is the most valuable sentence in the report, and it is the one most often omitted because it makes the recommendation conditional.

For buyers, the takeaway is to stop asking for a number and start asking for the range and the switch-over point. A supplier who says "material A is better provided the bag lasts more than four years" has told you something actionable. A supplier who says "material A is twenty per cent better" has told you something that will not survive a change of assumptions.

A short checklist for reading a report

The following questions can be worked through in about fifteen minutes and will identify most of what is wrong with a report before any numbers are examined. They are ordered so that the disqualifying ones come first.

• Was the study critically reviewed, and by whom? Public comparative claims require it under ISO 14044.
• Is the functional unit stated, and is it the unit relevant to the decision you are making?
• Is the boundary stated, and does it include every stage you care about, with cut-offs declared?
• Does the study use primary data for the stages that dominate the result, and is the source and age of secondary data given?
• Which allocation convention was used, and is it applied identically to every alternative?
• Is the service life assumption stated, justified, and tested across a range?
• Are burden and credit at end of life reported separately before netting?
• Is uncertainty or sensitivity analysis reported, with the switch-over point identified?
• Are all impact categories reported individually, or only a single score with undisclosed weighting?

Applied consistently, this checklist changes the commercial dynamic of buying an LCA, because suppliers who know these questions will be asked produce better work in anticipation. It also protects against the most expensive failure mode in the field: commissioning a study, receiving a number, making a material change on the basis of it, and discovering later that the number depended on an assumption nobody examined. The material substitution itself is usually the costly part, not the study.

Where a buyer wants a number for external communication, the safest route is a product-specific Environmental Product Declaration under an established programme, because the programme rules fix the method, the review and the disclosure format. That removes the freedom that produces disagreement, at the cost of flexibility. For internal decisions, a well-documented internal study is usually sufficient and much cheaper.

What an assessment is worth, and what it is not

The value of a life cycle assessment for this product is not the headline number. It is the hotspot identification, because hotspots tell you where engineering effort produces environmental return, and for waterproof bags that answer is consistent enough to act on without a bespoke study: material production dominates, weight and material choice are the levers, and service life is the denominator that can undo both.

The second value is comparative discipline. A study forces a bill of materials to be written down with real quantities, forces assumptions to be stated in public, and forces the chemical package to be named rather than waved at. Organisations that complete one usually discover that their specification was vaguer than they believed, and that is worth something independently of any impact figure.

What an assessment cannot do is settle a marketing argument, because the answer depends on choices the reader can legitimately make differently. Any public claim built on a single LCA result should be phrased to survive that: state the functional unit, state the boundary, and do not present a modelling credit as a physical outcome. Claims that do this are rarely challenged; claims that do not are the reason sustainability communication in this category has a credibility problem.

If you want the practical version of this applied to a specific style, the most useful input you can send is a bill of materials with real quantities and a realistic service life estimate. Review the production programme to see what documentation exists at each stage, then send the specification and we will tell you which lines of it carry the most weight. Minimum order quantity is 500 pieces per style, samples in 6–10 working days, bulk production in 35–50 days, quoted FOB Xiamen.

Frequently Asked Questions

Q1. Why do two life cycle assessments of the same bag give different answers?

Because they make different choices before collecting data. The functional unit, system boundary, allocation convention, service life assumption and electricity dataset can each change the ranking, and differences in these are far more consequential than differences in data.

Q2. What is a functional unit?

The quantified performance that every input and output is normalised to, such as one bag, one year of service, or one litre of protected volume per year. Changing it can reverse a comparison without changing any data.

Q3. Is a per-kilogram comparison useful?

Only for screening materials, never for choosing products. Per-kilogram figures ignore service life and capacity, so a heavier bag that lasts three times longer can look worse per kilogram and be better per year.

Q4. What is the difference between cradle-to-gate and cradle-to-grave?

Cradle-to-gate covers extraction through the factory gate and excludes distribution, use and end of life. Cradle-to-grave includes all stages and is required for whole-product comparisons, but depends on assumptions in every added stage.

Q5. What are cut-off rules?

Thresholds allowing small flows to be excluded, commonly one per cent of mass or energy with a five per cent total limit. They are legitimate if stated, and dangerous in this product because coatings and inks are small by mass and significant by impact.

Q6. What is the difference between primary and secondary data?

Primary data is measured at the specific facility for the specific product. Secondary data comes from databases representing average processes, sometimes in other regions or decades. The test is whether the dominant stages use primary data.

Q7. What is allocation?

The rule for splitting the burdens of a process between its co-products, and for deciding whether recycled material enters with no burden or with reprocessing burdens and an avoided-burden credit. It dominates results for high-recycled-content products.

Q8. Which allocation convention is correct?

Both cut-off and avoided burden are permitted. What matters is that one convention is applied consistently to every alternative in the comparison, and that the choice is stated so the reader can interpret the result.

Q9. Where is the biggest impact in a waterproof bag?

Raw material production, typically 55 to 80 per cent of cradle-to-gate impact. Factory assembly is usually 5 to 15 per cent, which means material choice and weight matter more than factory electricity despite being less visible.

Q10. How important is the use phase?

Small in magnitude but large in uncertainty, because washing frequency and service life are assumptions rather than measurements. Service life matters most of all, since it is the denominator of any per-year result.

Q11. Why does service life matter so much?

Because dividing the total burden by the number of years of service is what makes a durable product look good. Doubling assumed life halves the annual figure, so an unstated service life assumption makes any per-year result uninterpretable.

Q12. Why can incineration score better than recycling?

When incineration with energy recovery is credited for displacing carbon-intensive grid electricity. On climate change alone that credit can outweigh recycling, while on resource depletion and other categories recycling still wins. Single-axis comparisons hide this.

Q13. How much uncertainty is normal?

Large. Parameter uncertainty is commonly quantified by Monte Carlo simulation, while scenario choices such as transport mode and service life are handled by scenario analysis. A result without a stated range should be treated as incomplete.

Q14. What is a sensitivity analysis for?

To identify which assumption, if wrong, would change the decision. The most valuable output is the switch-over point, such as the service life at which one material overtakes another.

Q15. Should I ask for a single score?

No. Single scores require weighting across impact categories, which is a value judgement. Ask for individual category results, or for an Environmental Footprint compliant study if comparability matters more than flexibility.

Q16. What is an Environmental Product Declaration?

A verified declaration under a programme that fixes the method, the review requirement and the disclosure format. It removes much of the freedom that causes studies to disagree, at the cost of flexibility.

Q17. Is a bespoke assessment worth commissioning?

For hotspot identification and specification discipline, yes. For a headline marketing number, usually not, because the number depends on choices a critic can legitimately make differently. State the method if you publish.

People Also Ask

What is a life cycle assessment?

A standardised model of every input and output across a product’s life, from raw material extraction through manufacture, use and disposal, converted into environmental impact categories.

What is a functional unit in LCA?

The performance measure everything is normalised to, such as one year of service. Choosing it differently can reverse a comparison without changing any data.

Why do LCA results conflict?

Because studies differ in functional unit, boundary, allocation rule, service life assumption and electricity dataset, and those choices outweigh differences in data.

Which stage dominates a bag’s impact?

Raw material production, typically 55 to 80 per cent of cradle-to-gate impact, which makes material choice and weight the main levers.

What is allocation in LCA?

The rule for splitting burdens between co-products, and for deciding whether recycled input enters the product burden-free or with reprocessing burdens and a credit.

Does a longer lasting bag always have lower impact?

Per year of service, yes. Dividing total burden by more years is usually the largest reduction available for a durable product such as a waterproof bag.