A Compound Disc Spring is a stack of conical Belleville washers arranged in a deliberate mix of series and parallel orientations to deliver a precise force-deflection curve. The conical geometry flexes elastically when loaded along its axis, and stacking direction controls whether deflection or load capacity multiplies. We use compound stacks to maintain bolt preload, absorb shock, and compensate for thermal growth in tight spaces. A 50 mm OD compound stack can hold 40 kN of preload with under 6 mm of working travel — performance no coil spring matches in that envelope.
How the Compound Disc Spring Actually Works
A single Belleville washer is a shallow cone of spring steel. Push down on its rim and the cone flattens, storing energy elastically. The force-deflection curve is non-linear — it rises steeply, plateaus near the flat position, and can even drop if the cone is tall enough. A Compound Disc Spring takes that single washer behaviour and multiplies it by stacking. Same direction (parallel) stacks add force. Alternating direction (series) stacks add deflection. Mix the two and you build a custom curve.
The stacking arithmetic is simple but the tolerances are not. Two washers in parallel give you roughly 2× the force at the same deflection, but only if the contact faces are flat to within 0.02 mm and the guide pin or sleeve clearance sits between 1% and 3% of the inner diameter. Tighter than 1% and you get binding under load — the disc spring assembly locks up before reaching its rated travel. Looser than 3% and the stack walks sideways, contact stresses spike at the rim, and you crack washers at maybe 10% of rated cycle life. Lubrication on the rubbing faces of parallel pairs matters too; dry parallel stacks lose 15-20% of nominal force to friction hysteresis on the unload curve.
The most common failure mode we see in the field is settling — the stack loses preload after the first few thousand cycles because someone specified Group 1 washers (under 1.25 mm thick) for a high-cycle application that needs Group 3 (over 6 mm). The other classic mistake is mixing series and parallel without accounting for the load path. If you stack 4 in series then 2 in parallel, you get 2× force and 4× deflection compared to a single washer, not 8× anything. Get the arithmetic wrong and the bolted joint compensation either bottoms out or never reaches preload.
Key Components
- Conical Disc Washer: The base element — a truncated cone of hardened spring steel, typically 50CrV4 or 1.7102, with hardness around 42-48 HRC. Standard sizes follow DIN 2093, with outer diameters from 8 mm to 250 mm and cone heights tuned for Group 1, 2, or 3 behaviour.
- Guide Pin or Sleeve: Keeps the stack coaxial under load. Pin guides run through the inner diameter; sleeve guides run around the outer diameter. Clearance must be 1-3% of the guide diameter — too tight and the stack binds, too loose and the washers cock and crack at the rim.
- End Plates or Hardened Bushings: Distribute the rim contact force into the surrounding structure. Plates must be flat to 0.02 mm and hardened to at least 55 HRC, otherwise the washer rim plastically indents the seat and the stack loses preload after the first load cycle.
- Lubricant Film: Molybdenum disulphide grease or a dry MoS2 coating on rubbing faces of parallel pairs. Without it the friction hysteresis between pressed faces eats 15-20% of nominal force on the unload stroke and accelerates fretting wear.
- Stack Configuration: The arrangement itself is a component. A 4-in-series stack triples deflection at single-washer force. A 3-in-parallel stack triples force at single-washer deflection. Compound stacks like (2 parallel) × (4 series) multiply both.
Who Uses the Compound Disc Spring
Compound disc springs show up wherever you need high force in a short stroke, or where a bolted joint must hold preload through thermal cycling, vibration, and gasket creep. They handle shock loads that would crush a coil spring and tolerate temperatures from -40°C to over 250°C with the right alloy. The non-linear spring rate also makes them useful where you want a near-constant force across a working range — set the stack height so it operates near the curve's plateau and force barely changes over millimetres of travel.
- Heavy Machinery: Preload springs under the head bolts of Caterpillar 3500-series diesel engine cylinder heads — compensating for gasket creep and thermal expansion across 180°C swings.
- Valve Manufacturing: Live-loaded packing in Fisher control valves at refineries — a stack of compound disc springs maintains stem-seal preload as PTFE packing cold-flows over months of service.
- Press Tooling: Stripper plate return springs in Schuler progressive stamping dies — compound stacks deliver 80 kN in a 30 mm working envelope where a coil spring would need 4× the volume.
- Rail: Buffer stops on EMU rolling stock, including Stadler FLIRT trainsets — compound disc spring assemblies absorb impact energy in coupling collisions without the rebound a rubber buffer produces.
- Wind Energy: Yaw and pitch bearing bolt preload on Vestas V90 turbines — series-parallel stacks under each M36 bolt keep clamp load above 200 kN through 20+ years of vibration.
- Aerospace Ground Support: Tie-down bolt assemblies on missile transport cradles — compound stacks absorb transit shock without losing preload between inspections.
- Hydraulic Power: Relief valve poppet return springs in Bosch Rexroth high-pressure stacks operating at 350 bar — disc spring assembly cracks open repeatably because the spring rate is steep and predictable.
The Formula Behind the Compound Disc Spring
The compound stack force at a given deflection follows directly from how you arrange the washers. The single-washer force-deflection curve from DIN 2092 sets the baseline. Then the stacking rule scales force by the number in parallel and scales deflection by the number in series. At the low end of the typical operating range — say 25% of single-washer flat deflection — the stack behaves almost linearly and force is easy to predict. Push to 75% deflection and you're climbing the steep part of the curve where small position errors produce big force changes. The sweet spot for most preload applications sits between 60% and 75% of flat — high enough to give meaningful working force, low enough to avoid the plateau where force becomes insensitive to position.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Fstack | Total force produced by the compound stack at a given total deflection | N | lbf |
| np | Number of washers stacked in parallel (same orientation) | — | — |
| ns | Number of washer pairs stacked in series (alternating orientation) | — | — |
| Fsingle(s) | Force of a single washer at deflection s, from DIN 2092 curve | N | lbf |
| s | Deflection of a single washer in the stack | mm | in |
| sstack | Total deflection of the compound stack | mm | in |
Compound Disc Spring Interactive Calculator
Vary the number of series groups and parallel washers to see the stack force, travel, washer count, and spring-rate multipliers.
Equation Used
For a compound Belleville washer stack, washers nested in the same direction act in parallel and multiply load capacity by p. Alternating groups act in series and multiply working deflection by s. The effective spring-rate multiplier is therefore p/s relative to one washer.
- Ideal Belleville washers with identical force-deflection behavior.
- Parallel washers share load equally and add force.
- Series groups add deflection at the same load.
- Friction, settling, guide clearance, and washer tolerances are not included.
Worked Example: Compound Disc Spring in an injection-mould ejector return assembly
You are designing the ejector-plate return spring set for a 350-tonne Engel Victory injection moulding machine. The ejector plate retracts 8 mm per cycle and must apply 12 kN of return force at full retraction to overcome ejector-pin friction and seat the plate against its stops. You have selected DIN 2093 disc washers with OD 50 mm, ID 25.4 mm, thickness 3 mm, single-washer flat deflection s<sub>0</sub> = 1.15 mm, and single-washer force at flat F<sub>0</sub> = 7,650 N. You need to design a compound stack that delivers 12 kN at 8 mm total deflection without exceeding 75% of flat on any individual washer.
Given
- Ftarget = 12,000 N
- stotal = 8 mm
- F0 = 7,650 N
- s0 = 1.15 mm
- Max single-washer deflection = 75% of s<sub>0</sub> = 0.86 mm
Solution
Step 1 — figure out how many washers in series we need to reach 8 mm total without exceeding 0.86 mm per washer:
So each washer in the series chain deflects s = 8 / 10 = 0.80 mm, which is 70% of flat. That sits squarely in the design sweet spot.
Step 2 — at 70% of flat the single-washer force from a DIN 2092 Group 2 curve runs about 85% of F0:
Step 3 — at nominal stacking, find how many in parallel to hit 12 kN:
So the stack is (2 parallel) × (10 series) = 20 washers total, delivering Fstack = 2 × 6,500 = 13 kN at 8 mm. That gives an 8% margin over the 12 kN target.
Now check the operating-range behaviour. At the low end, when the ejector has only retracted 2 mm (25% of working stroke), each washer sees 0.20 mm of deflection — about 17% of flat. Force at that point drops to roughly 0.30 × 7,650 × 2 = 4,600 N. The plate barely starts moving. At the nominal 8 mm full retraction, you get the 13 kN computed above. Push the system to 9 mm by accident — say a worn limit stop — and each washer hits 0.90 mm, which is 78% of flat. You're now climbing the steep part of the curve and force jumps to roughly 14.5 kN per parallel pair × 2 = 29 kN. That spike will fatigue the ejector pins and crack the stripper plate inside a few thousand cycles.
Result
The compound disc spring stack is 2 in parallel × 10 in series, 20 washers total, delivering 13 kN at 8 mm working deflection — comfortably above the 12 kN target. In practice this means the ejector plate snaps back smartly against its stops without bouncing, and the spring rate stays predictable because every washer operates near 70% of flat. Across the working range, return force climbs from about 4.6 kN at 2 mm of stroke to 13 kN at 8 mm, with a sharp non-linear spike if travel ever exceeds 8 mm. If you measure return force well below 13 kN, three failure modes lead the list: (1) the parallel pairs were assembled dry instead of with MoS2 grease, costing 15-20% to friction hysteresis; (2) the guide pin clearance sits below 1% of the ID, binding the stack before full deflection; or (3) the end plates were specified at HRC 40 instead of HRC 55+, letting the washer rims indent the seats and absorb deflection that should have gone into spring force.
Compound Disc Spring vs Alternatives
Compound disc springs are not the only way to deliver short-stroke high force. The two real alternatives in industrial practice are heavy-duty coil compression springs and elastomer (urethane) springs. Each wins on different axes — pick based on stroke, force density, temperature, and how predictable the force-deflection curve needs to be.
| Property | Compound Disc Spring | Heavy-Duty Coil Spring | Urethane Spring |
|---|---|---|---|
| Force density (force per unit volume) | Very high — 40 kN in 50 mm OD × 30 mm height | Low — same force needs 4× volume | Medium — 60% of disc spring density |
| Working stroke | Short — typically 5-15 mm per stack | Long — 50 mm+ readily achievable | Medium — 10-25 mm typical |
| Force-deflection curve | Non-linear, tunable via stacking | Linear and predictable | Highly non-linear, hard to predict above 40% strain |
| Operating temperature range | -40°C to 250°C with standard steel; 400°C with Inconel | -40°C to 200°C | -30°C to 80°C — degrades fast above |
| Cycle life at rated load | 2 × 10⁶ cycles for Group 2 at 75% deflection | 10⁷ cycles at 50% rated load | 5 × 10⁵ cycles before compression set |
| Cost per assembled stack | Medium — $30-150 for typical industrial stack | Low — $10-50 | Low — $5-30 |
| Sensitivity to misalignment | High — needs 1-3% guide clearance | Low — tolerates side load | Medium |
| Best application fit | Bolt preload, valve packing, press tooling | Long-stroke return springs, machine balancers | Vibration isolation, low-cycle bumpers |
Frequently Asked Questions About Compound Disc Spring
The most likely culprit is friction hysteresis on the parallel pairs. When two washers are stacked face-to-face in parallel, their rubbing surfaces drag against each other as the stack flexes. Dry steel-on-steel costs you 15-20% per parallel pair on the unload curve, and you've got 4 in parallel — the loss compounds.
Pull the stack, clean the faces, apply a thin film of MoS2 grease (Molykote G-n Plus or equivalent) between every parallel-paired face. Do not lubricate the rim contacts on series transitions — those need to grip. Reassemble and retest. If you're still 10%+ low, check whether your end plates are dishing under load; soft seats absorb deflection that never reaches the spring.
Work the deflection requirement first. Divide the total stroke you need by 0.75 × s0 (the safe per-washer deflection limit) and round up — that's your ns. Then work the force requirement. Divide your target force by the single-washer force at that operating deflection and round up — that's your np.
Doing it in the other order traps you. If you size parallel first, you might end up forcing each washer to deflect past 80% of flat to make the stroke, and the force-deflection curve goes non-linear and unpredictable above 75%. Stroke first, then force.
This is classic settling, and it almost always traces to either washer group selection or seat hardness. If you specified Group 1 washers (under 1.25 mm thick) for a sustained-load preload application, they relax under stress because the thin section creeps at elevated temperature. Switch to Group 2 or Group 3 with the appropriate thickness for sustained load.
Second cause: end plate hardness below 55 HRC. The washer rim concentrates contact stress at maybe 1,500 MPa. Soft seats plastically deform under that, and the stack effectively shortens — preload drops. Hardened, ground bushings at 58-62 HRC fix it.
Don't. DIN 2093 sets dimensional tolerances but the force-deflection curve depends on heat treatment, surface finish, and edge condition — and those vary from supplier to supplier even within spec. Mixing brands inside a single parallel pair is the worst case because the stiffer washer takes more than its share of load and yields early.
Series stacks tolerate mixing slightly better than parallel because deflections add rather than forces, but you'll still see a force-deflection curve that doesn't match either supplier's published data. Buy the entire stack from one batch, one manufacturer.
Compound disc springs win when you need passive, maintenance-free preload compensation that survives 20+ years of vibration and thermal cycling. Hydraulic nuts give you precise initial preload but they don't compensate for gasket creep, bolt relaxation, or thermal growth — once the nut is locked, the joint is on its own.
Rule of thumb: if the joint sees less than 50°C of temperature swing and minimal vibration, hydraulic tensioning alone is fine. If it sees significant thermal cycling, vibration, or known creep in the clamped material (composite flanges, soft gaskets), put compound disc springs under the nuts. Vestas, Siemens Gamesa, and most major turbine OEMs do exactly that on yaw and pitch bearing bolts.
You're operating across the curve plateau. Belleville washers with a height-to-thickness ratio above about 1.4 produce a force-deflection curve that flattens — and can even drop — between roughly 50% and 100% of flat deflection. If your design crosses that region, force barely changes with position and the stack feels mushy or dead.
That's a feature for constant-force applications like valve packing, but a bug for return springs or preload. Switch to a lower h/t ratio washer (Group 2 with h/t around 0.75) and the curve stays monotonic and crisp across the full working range.
References & Further Reading
- Wikipedia contributors. Belleville washer. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
