What is a life cycle inventory (LCI)?

The life cycle inventory (LCI) is the second phase of an LCA and the most concrete: counting everything that enters and leaves the system under study. How many kilowatt-hours of electricity, kilograms of raw material and cubic metres of water went in; how many kilograms of CO₂, grams of SO₂ and kilograms of waste came out.

What is an elementary flow?

The inventory’s unit is the ‘elementary flow’: matter or energy taken directly from nature or released directly into it. Iron ore extracted from the ground is an elementary flow; methane released to air is an elementary flow. A steel sheet bought from a supplier is not — it is a product flow from the technosphere, and it is opened up through its own chain until only elementary flows remain.

Unit processes and the chain

The system is built from unit processes: the smallest modelling unit, each with its own inputs and outputs. A cement plant’s kiln is a unit process; coal mining is another; the truck carrying the coal is a third. Linked together, these processes reveal the entire chain behind the product.

In theory the chain is infinite — the electricity of the factory that made the truck’s tyre, the steel of the plant that made that electricity. Cut-off rules decide where to stop.

Foreground and background data

You do not measure the whole inventory yourself, and you do not need to. Processes under your control — your own site, your own recipe, your direct suppliers — are the foreground and should be modelled with primary data. Generic inputs such as grid electricity, commodity chemicals and standard transport are the background and come from a database.

ecoinvent is the most widely used database for that background, covering over twenty thousand processes. But background data is no substitute for foreground data: replace your site’s actual energy consumption with a sector average and you are no longer modelling your product but an average one.

How inventory quality is measured

• Temporal representativeness — which year the data covers, and whether production has changed since
• Geographic representativeness — whether German data represents Turkish production
• Technological representativeness — whether it is the same product by the same production route
• Completeness — how many sites the average covers, and whether the sample suffices
• Precision — whether it is measured, calculated or estimated

These five dimensions are scored in a framework called the pedigree matrix and reported as a data quality rating (DQR). If you are preparing an EPD under EN 15804, that scoring is mandatory.