Where weld quality data fits in the battery passport, and where it does not

Engineers who run welding cells in battery module plants often ask the same question once the passport project reaches their desk: do we need to put our weld quality data into the battery passport? The short answer is no. The longer answer is that weld data still belongs in the passport's orbit, just not inside the mandatory record.

This post separates what Regulation (EU) 2023/1542 requires from what is useful, and suggests where per-weld data should live so that it is available when someone needs it.

What the battery passport actually requires

The passport content is defined in Annex XIII, which I walked through in detail in the EU battery passport 2027 guide. It has four parts: public information on the battery model, model information for persons with a legitimate interest, test reports for authorities only, and data on the individual battery for persons with a legitimate interest.

Read the list with a welding engineer's eyes and you find nothing about joints. Part 1 asks for capacity, voltages, power capability, expected lifetime, round trip efficiency and "internal battery cell and pack resistance". Part 2 asks for detailed composition, spare parts and dismantling information, including "amount of cells used and layout". Part 4 asks for performance and durability values, state of health, the battery's status and data resulting from use.

The only manufacturing facts in the passport are where and when the battery was made (via Annex VI Part A) and who made it. Weld seam geometry, spot weld energy, inspection results and rework history are not mentioned anywhere.

Where welding still touches passport fields

Welding has an indirect effect on several passport values. The table maps the relevant Annex XIII items to the joint quality questions behind them.

The pattern is consistent. The passport records outcomes at battery level. It does not record the causes at joint level.

Why resistance values do not stand in for weld quality

It is tempting to argue that the pack resistance in the passport already "covers" weld quality. The research on tab joints does not support that.

In a parametric study of spot welding 18650 cells to nickel-plated steel connectors, Masomtob and colleagues measured both electrical resistance and shear strength of the welded contacts. Because the resistance differences between their weld contacts were minimal, they used mechanical strength as the deciding criterion for weld quality.

The picture is not uniform. Kaufmann et al. found resistance differences of less than 13 % across their green-laser copper-to-aluminium parameter sets, yet still report a correlation between electrical resistance and tensile strength, and suggest that a non-destructive resistance measurement may allow joint strength to be predicted. Resistance can therefore help as a screen, but with such a small spread it needs a precise measurement, and it is a single value per joint, not a direct look at fusion or porosity.

A weak weld can still pass a coarse resistance check and still be the joint that cracks after thermal cycling and vibration. Mechanical integrity, fusion depth and porosity are what decide how long the joint lasts, and none of them show up in a pack resistance value. I covered the defect side in laser welding of battery cells: typical defects and inline inspection methods and the spot welding side in resistance spot welding of battery tabs.

Three places weld quality data can live

If weld data is not mandatory but is useful, where should it go? There are three realistic options. They are not exclusive.

Option 1: internal record keyed to the passport identifier

The passport's unique identifier, which Article 77(3) requires to comply with ISO/IEC 15459 parts 1 to 6 or equivalent standards, is a natural primary key. If every module record in the plant uses it, a weld inspection result can be found years later from nothing more than the QR code on a returned battery.

This option needs no change to the passport and no decision about who outside the company may see the data. It is the right default for most manufacturers.

Option 2: voluntary additions to the passport

According to the VDE announcement of DIN DKE SPEC 99100, the specification covers both the data attributes the regulation requires and voluntary additions. A manufacturer could in principle publish a summary weld quality attribute, for example "all cell connections inspected inline, no rework", as a voluntary statement.

Be careful with this. Anything in the passport must be accurate and kept up to date under Article 77(4), and a voluntary claim becomes a claim you must be able to defend. A summary statement is usually wiser than raw per-weld data.

Option 3: restricted data for remanufacturers and recyclers

Remanufacturers and second-life operators care about which modules can be reused. Per-weld quality could help them decide, for instance, which modules had rework. The regulation's legitimate-interest mechanism is designed for this kind of audience, but who counts as a person with a legitimate interest, and how far they may download and re-use the data, is left to Commission implementing acts. Article 77(9) set 18 August 2026 as the deadline for adopting them; the deadline passed without the act being adopted, and the Commission's timeline of May 2026 places it in the fourth quarter of 2026.

Even once those rules exist, they govern the mandatory passport content. Sharing per-weld data with downstream partners is a bilateral, contract-based matter, not a passport function.

What a useful per-weld record contains

A weld record that will still be useful in five years needs more than a pass or fail flag. Here is a minimal checklist:

• the module or pack identifier, identical to the passport's unique identifier or reliably mapped to it
• the joint position within the module (cell number, tab, busbar)
• the process parameters actually applied, not just the recipe name
• the inspection result and, if available, a severity grade rather than a binary verdict
• the raw or reduced sensor data the grade was based on, or a reference to where it is stored
• the time stamp and the station, plus the version of the inspection model or threshold set
• any rework, with its own result

The structure of such records, from the sensor to the module, is the subject of a separate post on per-weld traceability.

Integrity matters even for internal data

Article 78 requires that "data authentication, reliability and integrity shall be ensured" for the passport. Internal weld records are not covered by that clause. But the moment they are used to defend a warranty decision, or to argue that a field failure was not a manufacturing defect, the same questions arise: who wrote this value, when, with which software version, and has it been changed since?

Weld data that cannot answer those questions is weak evidence. It is worth designing the record so that it can.

Where NeuroControls fits

Our work is on the sensing side: recording what happens during each weld, with grading by models our engineers train on your data. Our recording software stores data with a versioned manifest, per-segment SHA-256 hashes and an append-only journal, and exports to open formats such as JSONL and CSV, so per-weld records can be keyed to whatever identifier your MES uses. The measurement chain comes from our research on laser and spot welding inspection, described on our research projects page.

For an overview of where we apply multimodal sensing in cell and module production, see our battery production page. How recorded data is stored, hashed and exported is covered under data and integration.