Worked Example: Cell-Level DFMEA

Separator Failure Modes and Internal Short Circuit

This is a worked example accompanying Failure Mode and Effects Analysis (FMEA) in Battery Design. It walks through a single DFMEA line at cell level in full, from failure mode to recommended action, using the separator as the subject.

The separator is a good first example because it is thin, mechanically weak relative to the forces around it, and sits directly between the two electrodes. A localised failure in the separator does not stay localised: it becomes an internal short circuit, and internal short circuits are one of the more direct paths to thermal runaway. A DFMEA line on the separator is therefore a genuinely high-severity entry, which makes it a useful illustration of how Severity, Occurrence, and Detection interact.

Item and function

Separator, jelly-roll or stacked-electrode cell. Function: physically isolates anode and cathode while allowing ionic transport through the electrolyte-filled pores.

Failure mode

Separator develops a localised pinhole or tear during winding or stacking.

This is deliberately specific. “Separator fails” is not a usable DFMEA line because it gives no traction on cause or control. A pinhole from mechanical damage during winding is a distinct failure mode from separator shrinkage under heat, or from dendrite growth puncturing the separator during cycling; each of those would be its own line with its own cause and controls, even though they share a similar effect.

Effect

At the point of the defect, anode and cathode make direct electrical contact. This produces a localised internal short circuit, with rapid local heating at the contact point. Depending on the contact resistance and current available, this can self-limit and show up only as an anomalous self-discharge or capacity fade, or it can propagate into full thermal runaway of the cell.

Severity: 9. The upper end of the scale is reserved for effects with direct safety consequence to the end user, and an uncontained cell-level thermal event qualifies. It is not scored 10 only because a single-cell internal short does not automatically guarantee propagation to neighbouring cells or a hazardous outcome at pack level; that escalation is itself a separate System FMEA line (see below).

Cause

Particle contamination on the winding or stacking line: a hard particle (metal fragment, agglomerated catalyst material, or process debris) becomes embedded between electrode and separator during winding, and punctures the separator either immediately under winding tension or later under cell compression and cycling.

Other plausible causes for the same failure mode, worth listing as separate cause rows against the same failure mode and effect, include: separator supplied with an existing coating defect from the separator manufacturer, and mechanical damage from a misaligned winding roller.

Occurrence: 3. Contamination-driven separator punctures are a known, monitored failure mode in cell manufacturing rather than a theoretical one, but a mature production line with cleanroom controls and incoming separator inspection should see this at low frequency. A newer line, or one without established particle-count controls, would score this higher.

Current controls

In-line vision or thickness inspection of the separator immediately before winding, to catch particles or defects before they are built into the cell
HiPot testing post-assembly, which applies a voltage across the electrode stack and checks for breakdown, catching a meaningful fraction of separator defects before the cell ships
Incoming inspection of separator rolls against the supplier’s defect-density specification

Detection: 4. HiPot testing is a genuinely effective detection control for this failure mode, since a punctured separator often shows reduced dielectric strength even if it has not yet caused a full short. Detection is not scored lower because HiPot does not catch every sub-critical defect, particularly a very small puncture that only becomes a short under mechanical compression or cycling stress that is not present during the test itself.

RPN and Action Priority

RPN = 9 × 3 × 4 = 108.

Under Action Priority scoring, Severity 9 with Occurrence 3 and Detection 4 bands as High. This is the point of using Action Priority rather than RPN alone: a Severity-9 failure mode stays High priority even though the RPN of 108 might look moderate compared with a lower-severity, higher-occurrence line elsewhere on the same worksheet that produces a similar or higher RPN. Severity should not be allowed to get diluted by good Occurrence and Detection scores when the effect is this serious.

Recommended action

Add an automated optical defect-detection camera on the winding line, positioned to catch particle contamination before the separator is wound into the stack, rather than relying on HiPot to catch it after the fact
Tighten the cleanroom particle-count specification for the winding area
Consider a post-formation Electrochemical Impedance Spectroscopy check as a secondary detection layer for cells that pass HiPot but show anomalous early-life impedance behaviour

After implementing the camera and tightened particle spec, Occurrence would be expected to drop (not Severity, which is a property of the failure mode itself and does not change based on how well it is controlled). A revised Occurrence of 2 with Detection unchanged at 4 would reduce RPN to 72, but more importantly would not change the Action Priority banding, since Severity 9 still dominates. This is a useful reminder that RPN reduction and genuine risk reduction do not always move together, which is exactly the weakness Action Priority scoring is designed to correct.

Where this connects

This single DFMEA line sits underneath a broader System FMEA question: if a cell does go into internal short and thermal runaway, does it propagate to neighbouring cells? That question is addressed separately in the pack-level and System FMEA worked examples, since propagation depends on module architecture, thermal propagation barriers, and venting design rather than on anything specific to the separator itself.

Download the FMEA Worksheet Template to work through this and your own examples.