Styxis
Company
Knowledge

A battery recycling rate does not establish lithium recovery

Whole-battery recycling efficiency and recovery of a particular material use different denominators.

Different denominators behind a percentage

How much of a battery is recycled and how much of its lithium is recovered are different questions. Percentages do not make their targets interchangeable. Truthound can check the category combinations and calculation contract in a prepared table. It should not be presented as producing or certifying a physical recovery outcome.

The assessment did not raise the targets

The European Commission's September 11, 2026 assessment concluded that existing battery recycling-efficiency and material-recovery targets remained appropriate. It did not raise them. The official explanation distinguishes chemistry-specific weight-based efficiency from recovery of cobalt, copper, lead, nickel and lithium across different target years.

Separate battery mass from material mass

An illustrative table should keep whole-battery input mass separate from the input and recovered mass of a particular material. Do not copy a whole-battery percentage into a lithium row. Establish the process boundary, which material was analysed and what remained unmeasured before comparing the ratios.

Choose the question before the indicator

One analytical path prioritises battery-level operating efficiency. Another examines recovery of a strategically important material. Neither is universally superior. A question about process losses and a question about a particular raw-material supply need different outputs, even when both begin with the same operating register.

Allowed combinations in Truthound

Truthound Core Python SDK 3.3.4, operating on a prepared Polars table, provides ColumnPairInSetValidator for supplied allowed pairs of two fields. It can check an operator-defined material-and-metric association. ExpressionValidator can apply a caller-defined Polars row expression after units and denominators are established. Chemistry and target year need additional explicit rules; one two-field check does not resolve every classification.

When to keep the existing predicate

Pandera 0.35.0 pandas user predicates are a reasonable alternative for existing denominator calculations. Preserve a function that already distinguishes unmeasured material, denominators of 0 and different target years. Compare it with declared Truthound checks on the same illustrative inputs. Two paths that merely accept percentages between 0 and 100 are not necessarily checking the same meaning.

Zero and unmeasured change the story

Keep a recovered mass of 0 distinct from an absent measurement. Replacing unmeasured data with 0 can manufacture poor recovery; excluding those rows can make the ratio look better. Record the inclusion policy and the responsible owner's assessment of what could actually be measured.

A target year is not an operating result

A target year also does not mean the operating result for that year has arrived. Separate a preparatory comparison with a future criterion from assessment of a completed reporting period. Retaining source period and criterion version prevents an unchanged policy target from being mistaken for improved plant performance.

Make the two indicators visible

Show whole-battery efficiency and material recovery alongside each other. If one passes while the other lacks evidence, that difference identifies the next investigation. Explaining which resource returned under which boundary is more useful for a supply-chain decision than compressing everything into one recycling percentage.

Conditions

Prepare verified input and recovered masses with chemistry, material and target-year definitions.

Alternatives

Existing Pandera 0.35.0 pandas predicates can maintain separate denominator rules for the indicators.

Limitations

  • A valid row does not certify physical recovery or regulatory compliance.

Evidence

Event checked
Subject checked
Modified