Clamping prismatic LFP cells
What the manufacturers’ specifications require of the fixture that holds large prismatic LFP cells — and where they disagree. A short reference, not a design guide: it restates requirements and does not model what deviating from them costs.
Four numbers in the datasheet
Preload at assembly
The force the fixture applies across the large faces when the module is built. Specifications give a range — 500–5000 N (or 1500–5000 N) for CATL and REPT, 500–3000 N for Hithium, 3000–7000 N for EVE’s MB31 and LF280K — and nearly all of them run their own tests at 300 ± 20 kgf, about 2.9 kN. EVE specifies it at 15–40 % SOC: a charged cell is thicker, so the state of charge at which you tighten changes the force you end up with.
Compression limit
Only EVE states one: no more than 10 000 N for MB31 and LF280K, “otherwise the cell may be damaged”, and 9 kN for LF304, with 7 kN in normal service. The other specifications we found give no upper limit at all.
Swelling force over life
As a cell ages its electrodes thicken and it pushes back on the fixture. In each maker’s own rigid test rig this reaches 25–60 kN by the end of life — several times the preload and, for EVE, several times its own compression limit. Each figure belongs to its rig (plate material, thickness, gap); none is a general maximum.
Room to expand
EVE’s MB31 specification is the only one that says what to do about it: “it is suggested to reserve 2.0 mm ~ 2.5 mm expansion space while grouping the cells”, and the structure has to be designed for the swelling force.
What follows — and what does not
Do not over-tighten at assembly.
Stay inside your cell’s preload range, set at the state of charge the datasheet names. EVE warns that compression past its limit can damage the cell or make it leak.
Design the structure for the swelling force.
A fixture that does not give way has to carry tens of kilonewtons by the end of life — end plates, tie rods and welds included. One that gives way needs that room designed in; EVE suggests 2.0–2.5 mm per group.
Does a rigid clamp destroy the cell?
The datasheets do not say so, and we once said it too quickly ourselves. EVE’s own swelling test holds the cell between 15 mm steel plates with no gap until it reaches 60 % SOH — the cell lives through it. What is unknown is how much life a rigid fixture costs compared with a compliant one. Studies on small pouch cells find that both too little and too much pressure shorten life, but they transfer poorly to large prismatic LFP, and no data we have can put a number on it — so the calculator does not.
Can you skip the preload altogether?
One unsourced comment online says modern CATL and BYD packs use no initial compression. The specifications we found say the opposite: CATL and REPT write that “pack design need to add preload force”, Hithium that the system “requires a certain pre-tightening force”. A 2025 thesis with Volvo Cars — NMC prismatic cells, two per condition, two months of cycling — found that cells without pre-compression faded fastest. What modern packs do use are compliant layers between cells, not zero preload.
What each specification says
As written in the manufacturers’ documents. 300 kgf ≈ 2.94 kN. Swelling forces apply to each document’s own test rig. Specifications get revised — check the revision of yours.
| Cell | Preload | Compression limit | Swelling force | Face, mm | Document |
|---|---|---|---|---|---|
| EVE MB31 · 314 Ah | 3000–7000 N recommended; tests at 300 ± 20 kgf, 15–40 % SOC | ≤ 10 000 N | ≤ 50 kN at 70 % SOH; ≤ 60 kN at 60 % SOH (15 mm steel, no gap) | 173.7 × 204.6 | PBRI-MB31-D06-01 Ver. A manufacturer specification, copy on file suggests reserving 2.0–2.5 mm of expansion space |
| EVE LF280K · 280 Ah | 3000–7000 N recommended; tests at 300 ± 20 kgf, 15–40 % SOC | ≤ 10 000 N | ≤ 3 kN new; ≤ 40 kN at 70 % SOH; ≤ 50 kN at 60 % SOH | 173.7 × 204.6 | PBRI-LF280K-D06-01 Ver. C (2023-06) manufacturer specification, copy hosted by docanpower.com Ver. B (2023-02) said 50 kN; the later revision is quoted |
| EVE LF304 · 304 Ah | 3–5 kN; tests at 300 ± 20 kgf, 30–40 % SOC | 7 kN in service; never above 9 kN | ≤ 3 kN new; ≤ 30 kN at end of life (end of life not defined) | 173.7 × 204.4 | PBRC-LF304-S01-LF Ver. C manufacturer specification, copy hosted by batteryfinds.com |
| CATL · 280 Ah | 500–5000 N for a new cell, ± 200 N; tests at 300 ± 20 kgf | not stated | ≈ 40 000 N at 60 % SOH (15 mm steel plate) | 173.9 × 207.2 | Rev. 1.0, sample stage manufacturer specification, copy hosted by batteryfinds.com height includes the terminal; mass-production values “need to be redefined” |
| CATL · 310 Ah | 1500–5000 N, ± 200 N; tests at 300 ± 20 kgf | not stated | not stated | 171 × 213 | Rev. 1.5 manufacturer specification |
| REPT CB71 · 320 Ah | 1500–5000 N, ± 200 N; tests at 300 ± 20 kgf | not stated | not stated | 174.0 × 204.4 | v01 (2023-02) manufacturer specification, copy hosted by ecoteardown.top the 50 kN in this document is a crush test, not a limit |
| REPT CB75 · 314 Ah | 500–5000 N, ± 200 N; tests at 300 ± 20 kgf between 17 mm steel plates | not stated | not stated | 174.0 × 204.4 | v02 (2024-01) manufacturer specification |
| Hithium · 280 Ah | 500–3000 N for a new cell, ± 200 N; test fixture 3000 ± 200 N | not stated | ≈ 25 000 N at 70 % SOH (15 mm aluminium plate) | 174.70 × 204.47 | HC-A280-SPEC-0037 A/1 (2022-08) manufacturer specification, copy hosted by ecoteardown.top the earlier version says a steel plate |
| Cornex · 280 Ah | no specification with these figures found | ||||
Check your own datasheet
Look for sections called compression force, preload, pre-tightening force, fixture, and swelling or expansion force. Note the state of charge at which the preload is specified and the rig behind the swelling figure. Revisions differ: the February 2023 revision of EVE’s LF280K specification allowed up to 50 kN, the June 2023 revision of the same document 10 000 N. This page quotes the later one.
What this page does not do
It restates what the specifications require and nothing more. It does not say how many years a wrong preload costs: we do not know, and for large prismatic LFP no published study we found does either. The battery life estimator leaves clamping out for the same reason.
Sources
- EVE LF280K · 280 Ah — PBRI-LF280K-D06-01 Ver. C (2023-06)
- EVE LF304 · 304 Ah — PBRC-LF304-S01-LF Ver. C
- CATL · 280 Ah — Rev. 1.0, sample stage
- REPT CB71 · 320 Ah — v01 (2023-02)
- Hithium · 280 Ah — HC-A280-SPEC-0037 A/1 (2022-08)
- Uppsala University MSc thesis with Volvo Cars, 2025 — pre-compression of NMC prismatic cells
- batterydesign.net, LFP cell clamping pressure — including the comment on packs without preload