Module End Plate Stiffness Optimization under Dynamic Non Linear Battery Swelling Forces
Optimizing module end plate stiffness between 100 and 130 kN/mm balances non-linear cell swelling forces to prevent pore crushing and interface delamination.

Trace
Prismatic lithium-ion cells swell through two separate mechanisms over their service life: reversible breathing as lithium intercalates during normal cycling, and irreversible volume growth driven by solid electrolyte interphase buildup, outgassing, and transition metal migration into the lattice. In a 100 Ah nickel-manganese-cobalt cell, reversible thickness expansion typically runs between 1.2% and 2.5% from 0% to 100% state of charge. Irreversible growth adds a further 4.0% to 8.0% across 1,500 full equivalent cycles at 25 degrees Celsius.
The resulting force profile is non-linear, determined by how the jelly roll, aluminum can walls, compressible polyurethane or aerogel buffers, and the module end plate deflect together under load.
Sizing an end plate without accounting for stack compliance leads directly to premature capacity fade. An excessively rigid plate can drive internal stack pressure past 0.80 MPa at end-of-life full charge, crushing electrode pore volume below the 20% needed for electrolyte percolation. That brings on lithium plating, sharp impedance gains, and rapid loss of cyclable lithium.
If the plate deflects too much, on the other hand, stack pressure can fall below 0.05 MPa at beginning-of-life discharge, letting the graphite coating shear away from the copper current collector under cyclic movement. For commercial automotive packs, the workable operating window remains confined between 0.15 MPa and 0.45 MPa over the life of the pack.
A module end plate deflecting more than 0.40 millimeters across a 12-cell prismatic stack permits low-pressure interfacial delamination early in life, whereas deflections below 0.10 millimeters trigger pore collapse during end-of-life charging.
Dynamic loads complicate the steady-state picture. Hard discharge pulses create thermal gradients across the cell, with the core running 12 to 18 Kelvin hotter than the can exterior and driving localized force spikes before the heat evens out. Standard automotive shock inputs ~ 25g to 50g vertical and longitudinal events ~ then add momentary inertial loads to an already stressed structure.
Modules that look acceptable on paper can shear tie-rod anchor welds or bow past enclosure seal limits during physical pack validation. Keeping these loads manageable requires tuning foam non-linearity, transient mechanics, and end plate geometry together rather than treating each in isolation.

Stack
The net force reaching an end plate is the cumulative expansion of the cell cores minus whatever displacement the compliant interlayer pads absorb. In typical 12 to 24 cell series modules, sleeved or laminated prismatic cans alternate with compressible cushions, and the end plate functions as a structural beam tied back by side straps, bottom runners, or bolted tie rods.

Nonlinear Force Displacement Response
The load-deflection curve of a prismatic stack departs sharply from Hooke’s law. Below 0.10 MPa, surface asperities on the cans and the open pores of the jelly roll produce a relatively soft initial contact stiffness. Push past that, however, and the cells collapse the void volume in the cushion pads, driving the material into a densification regime where tangent modulus jumps by two orders of magnitude.
Stack mechanics can be captured by coupling free cell expansion against the compression characteristics of the pads. The normal force F over end plate area A follows:
F = A E_pad(epsilon) ( (N delta_cell – delta_endplate) / (N t_pad_0) )
Here N is the number of cells in the group, delta_cell is unconstrained expansion per cell at a given charge and health state, t_pad_0 is nominal pad thickness before compression, delta_endplate is deflection at the end plate centerline, and E_pad(epsilon) is the strain-dependent tangent modulus of the pad. Because E_pad climbs steeply once foam strain passes 40%, underestimating delta_cell forces the assembly onto the stiff part of the densification curve and rapidly spikes mechanical load onto the plate.
| Chemistry Designation | Nominal Cell Capacity (Ah) | Reversible Breathing (Percent) | Irreversible EOL Swelling (Percent) | Beginning Target Pressure (MPa) | Peak Allowed EOL Pressure (MPa) |
|---|---|---|---|---|---|
| LiFePO4 (LFP) | 105 | 0.8 to 1.4 | 6.0 to 11.0 | 0.08 to 0.15 | 0.50 |
| LiFePO4 (LFP) | 280 | 1.1 to 1.8 | 8.0 to 14.0 | 0.10 to 0.20 | 0.60 |
| NMC 622 | 65 | 1.5 to 2.8 | 3.5 to 6.5 | 0.15 to 0.25 | 0.45 |
| NMC 811 | 100 | 2.0 to 3.4 | 4.5 to 8.5 | 0.15 to 0.30 | 0.50 |
| NMC/Silicon-Graphite | 78 | 3.5 to 6.0 | 7.0 to 15.0 | 0.20 to 0.35 | 0.75 |
LFP and high-nickel chemistries impose very different mechanical demands on the retention hardware. High-nickel NMC breathes noticeably during single cycles as lithium shifts between hosts, whereas LFP breathes little cycle to cycle but shows heavy irreversible volume growth over long cycling due to two-phase transition dynamics and steady SEI growth. Silicon-blended graphite anodes push breathing displacement higher still ~ up to 6% in a single cycle ~ which strains standard packaging envelopes and demands significantly higher compliance reserves throughout the retention structure.

Rib
Flat plates of uniform thickness are an inefficient way to contain swelling. Pinned or clamped along its vertical edges and loaded with uniform pressure, a simple flat plate bows into a shallow bowl, concentrating deflection at the geometric center. That deflection creates an uneven pressure field across the terminal cell face: edge perimeters see sharp stress concentrations while contact pressure drops off across the middle.

Topology Optimization and Section Modulus Distribution
Preventing localized electrode damage requires an end plate section that maintains relatively uniform pressure over the broad face of the cell. Mechanically, this means placing the section moment of inertia where bending moments peak rather than distributing material uniformly.
Production end plates generally replace solid slabs with pressure-die-cast aluminum or multi-void extrusions using targeted ribbing. Moving metal away from the neutral axis provides deep webs where midspan bending is highest, with sections tapering down toward the strap anchors. Cast alloys such as EN AC-43400 (AlSi10Mg(Fe)) and extruded EN AW-6060 T6 provide good strength-to-weight performance while retaining sufficient elongation during crash events.
Designers typically weigh three structural topologies for these ribs:
- Orthogonal grillage patterns stiffen the plate in both horizontal and vertical directions, checking edge bulging from the jelly roll and maintaining even contact across the cell insulation.
- Diagonal isogrid webbing spreads shear loads cleanly into top and bottom mounting bosses, helping resist twist when the pack experiences chassis torsion or multi-axis vibration.
- Centrally crowned variable-thickness extrusions concentrate wall thickness at the centerline and taper outward toward the side rails, placing material where the primary bending moment requires it.
Matching the end plate’s section modulus to the parabolic moment distribution prevents contact pressure loss and delamination across the center of the cell.

FEA Contact Formulations and Gap Management
Validating these topologies calls for non-linear finite element contact models that handle gap opening and interface slip. As swell forces rise, outward plate deflection lifts load from the cell center and pushes it toward the perimeter shoulders near the strap mounts. Modeling this with linear elastic assumptions or bonded surfaces misses that redistribution and can underpredict local peak stresses on the end cell by up to 45%.
Contact definitions generally apply penalty-based formulation with Coulomb friction coefficients between 0.15 and 0.25 for polyimide-wrapped cans resting against anodized aluminum. Foam buffers require hyperelastic or crushable foam representations ~ such as Ogden or Fu-Chang formulations ~ calibrated against physical test curves across the relevant strain rates.
Dedicated dies for cast or extruded ribs require substantial tooling investment upfront. Whether those tooling costs make financial sense depends on annual production volumes and the packaging weight targets set for the pack.

Anchor
End plates cannot hold swell forces alone; reaction loads pass directly into side straps, structural side plates, or weld seams in the bottom tray. In practice, the joint linking the end plate to these tension members is the most common point of mechanical failure during life cycling and shock testing.

Tension Strap Joint Mechanics
Under static swelling, side straps see largely pure tension, but road vibration and structural flex in the pack introduce in-plane shear and peel moments at the joints. Three fastening methods dominate current module production: laser welding, blind rivets, and threaded fasteners.
| Joint Architecture | Tensile Capacity per Joint (kN) | Assembly Takt Time (s) | Tooling NRE Burden (USD) | Service and Rework Capability | Failure Mode Under Cyclic Load |
|---|---|---|---|---|---|
| Wobble Laser Seam Weld | 22 to 38 | 1.5 to 3.0 | 120,000 to 250,000 | Destructive only | Heat-affected zone fatigue cracking |
| Flow Drill Screws (M5/M6) | 14 to 24 | 4.0 to 6.5 | 45,000 to 80,000 | Thread extraction permitted | Thread stripping and torque relaxation |
| Self-Piercing Rivets (SPR) | 12 to 19 | 2.0 to 3.5 | 60,000 to 110,000 | Destructive only | Button pull-out under shear fatigue |
| Structural Bolts with Thread Inserts | 28 to 45 | 8.0 to 14.0 | 15,000 to 35,000 | Fully serviceable | Bolt preload loss from seating creep |
Laser welding offers the fastest cycle time on automated lines, but it demands fit-up gaps tighter than 0.08 millimeters across the joint. Normal variations in cell dimensions or pad thickness can spread that gap open, producing spatter, weld porosity, or poor penetration that fails under fatigue. Threaded fasteners avoid putting heat near cell insulation, but they slow down assembly and require tight thread-locking controls so road harmonics cannot back them out.

Side Strap Buckling and Dynamic Elastic Recovery
As cells charge and discharge under aggressive use, side straps stretch and relax cyclically. Any vertical asymmetry in end plate stiffness or offset between the plate’s neutral axis and the strap load line creates an eccentric moment. That eccentricity can buckle thin side straps out of plane, particularly when vertical chassis accelerations coincide with full cell charge.
If a strap buckles outward by even 1.5 millimeters, its tensile membrane stiffness drops abruptly and the end plate tilts. That rotation unloads compression across the upper portion of the stack while pinching bottom cell corners, risking ruptured wrap insulation and ground faults to the tray.
Proposals to thin out side straps for weight and cost savings often rely on beginning-of-life swell forces staying well below yield. That logic overlooks low-cycle fatigue, which governs joint endurance after hundreds of breathing cycles under sustained load.

Drift
Dimensional stability in a live module depends on viscoelastic relaxation and creep running across several components at once. Over a ten to fifteen year design life, the mechanical equilibrium between end plate flexure and cushion elasticity slowly drifts.

Viscoelastic Relaxation of Inter-Cell Cushions
Open-cell polyurethane and silicone foams relax measurably under sustained compression. Held at a constant 30% strain, microcellular polyurethane sheds 20% to 35% of its return force within 1,000 hours at 40 degrees Celsius. Ceramic aerogel blankets offer higher thermal and flame resistance, but they take a progressive permanent set and can shed particulate under repeated mechanical cycling.
When a compressible pad loses 30% of its force through relaxation over 2,000 hours, unabsorbed cell expansion shifts directly into the structural end plate.
If pads relax too quickly, discharged stack pressure drops toward zero, allowing internal layers to separate during cold operation. On the other hand, pads with high mechanical hysteresis cannot compress fast enough during rapid charging, leaving the end plate to absorb transient force spikes with almost no cushioning.

Creep Deformation in Structural Polymers and Die Castings
Lightweighting efforts often lead teams to consider glass-filled polymers such as 30% to 50% reinforced PPS or PA66-GF50. These materials show good short-term tensile numbers on a standard test sheet, but their resistance to long-term creep under sustained load and elevated temperature is poor.
- Initial instantaneous elastic deformation develops during initial module banding, producing 0.15 to 0.25 millimeters of deflection at midspan.
- Primary viscoelastic settling occurs over the first 72 hours after assembly, dropping strap tension or bolt torque by 8% to 12% as surface high spots bed in.
- Secondary steady-state thermal creep builds across thousands of operating hours at pack temperatures between 35 and 55 degrees Celsius, causing the center span to bow outward permanently.
- Tertiary localized tertiary creep rupture starts at sharp rib transitions, creating microcracks that can fail under dynamic road shock.
Die-cast aluminum holds its geometry far better than polymers over time, though bolted joints in aluminum still suffer relaxation. Embedment and local yielding under bolt heads gradually bleed off clamp preload unless drawings specify hardened steel washers and controlled-thread fasteners.
Creep testing must run at peak operating conditions ~ generally 55 to 60 degrees Celsius for traction applications. Ambient room-temperature test data hides the steep rise in creep rate that occurs once polymers approach their glass-transition thresholds in hot packs.

Reckoning
Balancing end plate stiffness means bridging FEA structural output and electrochemical boundaries. The task balances conflicting demands: trimming part mass and packaged volume while holding stack contact pressure inside the narrow window cells need to survive.

Worked Optimization Construction: 16-Cell Prismatic Module
The trade-offs are visible in a sizing study for a 16-cell series module using NMC 811 prismatic cans. The comparison tests three end plate options: a solid billet plate, a cast ribbed structure, and a thin stamped steel shell.
The evaluation uses the following operational baseline:
- Cell broad face geometry measures 148.0 mm wide by 91.0 mm high, giving an active contact area A of 13,468 mm2.
- Unconstrained swelling strain combines 1.8% reversible breathing at 100% SOC with 5.5% irreversible growth at 80% capacity retention, totaling 23.36 mm of unconstrained stack growth across 16 cells.
- Inter-cell cushion configuration uses 17 microcellular foam pads at 1.50 mm nominal thickness each, with a non-linear response where tangent modulus climbs from 1.2 MPa initially to 18.5 MPa near full densification.
- Maximum allowable end plate centerline deflection is held between 0.12 mm and 0.35 mm across an operating range of minus 20 to plus 55 degrees Celsius.
| Design Architecture Metric | Case A: Solid Billet Plate (Al 6061-T6) | Case B: Cast Ribbed Plate (AlSi10Mg) | Case C: Compliant Stamping (Steel HC340LA) |
|---|---|---|---|
| End Plate Thickness Profile (mm) | 18.0 (Uniform) | 4.5 base wall, 14.0 central ribs | 2.5 (Uniform) |
| Mass per End Plate Pair (kg) | 2.62 | 1.34 | 1.08 |
| End Plate Centerline Stiffness (kN/mm) | 185.0 | 115.0 | 38.0 |
| Centerline Deflection at EOL (mm) | 0.09 | 0.21 | 0.68 |
| EOL Peak Stack Force (kN) | 18.8 | 12.4 | 6.2 |
| EOL Average Stack Pressure (MPa) | 0.61 | 0.38 | 0.14 |
| Internal Cell Delamination Risk | Zero | Zero | High at center |
| Active Material Pore Crushing Risk | Severe at corners | Zero | Zero |
| Tooling Investment NRE (USD) | 5,000 (CNC) | 85,000 (Die Cast) | 45,000 (Stamping) |
The comparison shows why mid-range stiffness delivers the best cell life. Case A is far too rigid. With the plate yielding just 0.09 millimeters, the cells absorb their own swelling internally.
Total load rises to 18.8 kN, which pushes average interface pressure to 0.61 MPa. That exceeds the 0.50 MPa limit for NMC 811, fracturing secondary particles, breaking binder bonds, and driving sharp impedance growth after 800 cycles.
Case C goes too far in the other direction. The 2.5 mm steel stamping deflects 0.68 millimeters under load, holding end-of-life stack pressure to 0.14 MPa and preventing particle crushing, but introducing two critical faults. At low charge states, the deflection is large enough to allow interfacial delamination in the center of the electrode.
At the same time, because the plate bows at midspan while staying pinned at the straps, pressure distribution is heavily skewed: outer edges carry 0.40 MPa while the center drops below 0.02 MPa. That mismatch concentrates current density along the margins, accelerating peripheral aging and triggering local lithium plating.
Case B finds the middle ground. Ribbing provides an effective centerline stiffness of 115 kN/mm, allowing 0.21 millimeters of deflection. That keeps end-of-life pressure at 0.38 MPa, right within the target 0.15 to 0.45 MPa electrochemical envelope.
It also trims 49% of the mass compared to the solid billet plate, offsetting the casting tooling investment through better pack gravimetric density and lower warranty risk.
End plate design effectively sets the structural warranty line for a pack. Accepting unverified module hardware that lacks testing under combined vibration, high state of charge, and elevated temperature leaves the buyer holding field failure liability. Validating end-of-life retention demands multi-axis shaker profiles inside environmental chambers at peak operating temperatures, with explicit compliance targets written into the production supply agreement.
Engineering and quality teams protect pack programs by tying receiving inspection criteria to mechanical deflection limits, ensuring production lots match the stiffness and pad damping curves agreed upon during design sign-off.





