Why the cathode determines everything
The difference sits in the cathode. In LiFePO4 (LFP) the iron phosphate is bound in an olivine structure. The phosphorus–oxygen bond is strong; under heat the lattice releases barely any oxygen. In Li-NMC (lithium nickel manganese cobalt) there is a layered oxide that releases oxygen when overheated. This oxygen is the fire accelerant inside the cell.
The most important safety difference follows directly: the thermal-runaway onset — the temperature at which the exothermic chain reaction becomes self-sustaining — is around 250 °C for LFP, and about 150 to 210 °C for NMC depending on the cell type. NMC starts the chain reaction earlier, releases more energy and brings its own oxygen. That is not a marketing statement, that is cell chemistry.
The numbers the design hangs on
The price for the stability of LFP is energy density. Roughly:
| Parameter | LiFePO4 | Li-NMC |
|---|---|---|
| Gravimetric (Wh/kg) | ~90–120 | ~150–220 |
| Volumetric (Wh/L) | ~220–330 | ~350–550 |
| Nominal cell voltage | 3.2 V | 3.6–3.7 V |
| Cycles to 80 % SoH | ~3,000–6,000 | ~1,000–2,500 |
| Thermal-runaway onset | ~250 °C | ~150–210 °C |
Two consequences for practice. First: at the same weight NMC delivers roughly 50 to 80 % more energy, and at the same volume often twice as much. Where the chassis is fixed, volumetric density (Wh/L) wins, not gravimetric. Second: under daily full cycling LFP lasts many times longer. At 500 full cycles per year, that separates the replacement interval in years, not months.
The lower cell voltage of LFP (3.2 V instead of 3.6 V) also means: for the same system voltage you need more cells in series. A 48 V system is around 15 LFP cells versus 13 NMC cells. That costs space and one more balancing line in the BMS.
Two projects, two decisions
In the cable crane project with MM Forsttechnik the case was clear early. Energy recuperation in steep terrain — the descending load feeds energy back into the pack under braking — produces many partial cycles per working day and recurring charge currents. Add to that a deployment site that, in a thermal event, is neither quickly reachable nor extinguishable. Here cycle life over years and an onset that keeps the maximum distance from any real operating temperature counted. It became LFP, designed for high recuperation currents rather than for minimum weight.
Conversely with the compact AGV: installation space predefined, the vehicle has to fit under an existing load platform, and the required runtime simply could not be accommodated in the volume without higher volumetric density. There it went to Li-NMC — with a correspondingly stricter thermal concept and tighter BMS limits, because the onset sits closer to operation. In both cases the chemistry followed the constraint, not the wish.
Decision rule
LiFePO4 when:
- ✓ safety in the fault case dominates (hard to reach, not extinguishable, close to people)
- ✓ more than ~500 full cycles/year or a target service life of 8+ years
- ✓ operation in non-air-conditioned areas with high ambient temperature
- ✓ weight and volume are uncritical
Li-NMC when:
- ✓ installation space or weight are the hard limit — check Wh/L, not just Wh/kg
- ✓ a moderate cycle count with high required energy in a small volume
- ✓ a robust thermal management is provided to compensate for the lower onset
The quickest test: calculate your energy demand against the available volume. If LFP fits in, the discussion is usually over — the safety and service-life reserve comes for free. If the volume gets tight, NMC is the way, but then with the thermal concept the lower onset demands.
Once you have your deployment profile on the table — cycles per day, ambient temperature, installation space, charge window — the chemistry can be reliably locked down in a single conversation. This is exactly the design we are happy to discuss against your real numbers.