Crash box calculator: mean crushing force, energy absorption and the LS-DYNA crash box check
A thin-walled tube hit end-on folds like an accordion at a roughly steady mean crushing force. Give the section, wall thickness, length, material and the mass and speed that hit it. You get that force by up to four published formulas, the fold length, the energy it can absorb, its SEA, whether it stops the mass, and the wall thickness a stroke needs. Free, runs in your browser.
The second tab checks the crash box challenge of Skill-Lync's LS-DYNA course: why the box rebounds instead of folding, what 1.08e5 means in gm-mm-ms, and whether the energy, force or acceleration in your report is consistent. Skill-Lync read 281 public crash box and crush tube reports to build it.
Abramowicz and Jones 1984, 1986 · Wierzbicki and Abramowicz 1983 · Alexander 1960 · Anand et al. 2024 · checked against a separate solution (1,344 checks at build) · Updated 22 Nov 2026
Sizes are mid-surface (the shell's mid-plane, as LS-DYNA meshes it). Material presets carry the yield and UTS each source tested; density, E and ν for steel are the challenge's card values. κ is the share of the length that can crush (0.73 for mild steel square tubes); circular tubes use their own κ.
The defaults are the values most Week 4 reports type: yield 250 MPa, 7.85e-3 g/mm³, E 210 GPa, ν 0.3, 13.8889 mm/ms. MAT_024 without SRC and SRP has no rate effect, and no report Skill-Lync read used them.
| Formula | Static Pm | Source |
|---|---|---|
| Abramowicz and Jones 1984 with the effective crushing distance (κ 0.73); R² 0.95 in Anand et al. | 22.1 kN | Jones, as in Ronchietto 2009 |
| Wierzbicki and Abramowicz 1983 superfolding element, no effective distance: lower | 16.2 kN | Velmurugan and Muralikannan 2009 |
- C/t = 104. Outside the 23.3 to 47.1 of the 42 square-tube tests Anand et al. compared the formulas with: treat the force as an estimate.
- Elastic buckling of the widest wall (150 mm, simply supported, k = 4): 48.6 MPa, below the 250 MPa yield: the walls buckle elastically before they yield.
- Rate effects off: the force is the static one, as a MAT_024 card without SRC and SRP would give.
- Stops in 4.89 mm of the 183 mm effective stroke (3%). That is less than one complete fold (38.4 mm of crush per fold): the end dents, the box does not fold progressively, and an elastic-plastic box rebounds.
| 1.2 mm | 1.5 mm |
|---|
Unit decoder
Check a number from your report
How do you calculate the mean crushing force of a square tube?
Use Abramowicz and Jones's 1984 formula. A square or rectangular tube crushed end-on folds in lobes, and the force swings about a mean. Abramowicz and Jones worked out the plastic work in one fold and divided it by the crush the fold takes up:
σ0 is the flow stress (the yield, or the mean of yield and UTS, as many authors use), t the wall and C the side. For a rectangle, Velmurugan and Muralikannan use the mean side, C = (b + h)/2. The 52.22 already contains κ = 0.73, the effective crushing distance: a fold of length 2H crushes only 73% of 2H before it packs solid. Anand et al. (2024) note that for another κ the force scales by 0.73/κ; they use 0.77 for stainless steel.
The 52.22 is not a fitted number. It is the minimum, over the fold's rolling radius b and half-length H, of the fold energy Anand et al. write out (their eq. 17):
Skill-Lync's verifier minimises it numerically and gets 38.11 at b/t = 0.715 (C/t)1/3 and H/t = 0.989 (C/t)2/3: 52.22 × 0.73 = 38.12, b 0.72 and H 0.99, as quoted. The calculator also gives the two other square-tube formulas, for comparison:
Wierzbicki and Abramowicz's 1983 superfolding element has no effective crushing distance, so it comes out 27% under Abramowicz and Jones. Magee and Thornton's is a fit to tests. The fold length and stroke follow:
Mild steel is stronger when it is crushed fast. The Cowper-Symonds law raises the flow stress with the strain rate, and Abramowicz and Jones estimate the mean rate in the folds from the impact speed V:
Two sets of mild steel constants are in use: D = 40.4 s⁻¹, q = 5, the traditional pair Jones lists, and D = 6,844 s⁻¹, q = 3.91, which Abramowicz and Jones used for their tubes. They differ a lot: at 13.9 m/s on the Week 4 box the first raises the force 1.98×, the second 1.26×. Jones lists D = 6,500 s⁻¹, q = 4 for aluminium alloy, but Hsu and Jones's own tests on 6063-T6 found it almost rate-insensitive, so the aluminium presets start with rate effects off.
What formula do you use for a circular crush tube?
For ring-shaped (concertina) folds, use Abramowicz and Jones's 1986 closed form, as Ly et al. (2015) write it, with its own effective crushing distance:
It is a fit to the minimum of a fold-energy expression Anand et al. give (their eq. 13). Skill-Lync's verifier minimised that expression for D/t from 20 to 100 and fitted a line through the minima: 20.79 √(D/t) + 11.90, against the published 20.79 and 11.90. The minimum also fixes the fold half-length:
The calculator lists three more circular formulas: Alexander's 1960 original, Wierzbicki et al.'s 1992 two-fold model, and the cube-root form Velmurugan and Muralikannan quote. In Anand et al.'s comparison with 44 tests and their own simulations, the Abramowicz and Jones model fitted best (R² 0.98) and Alexander's ran low (R² 0.73).
The formulas assume concertina folds. In Anand et al.'s dataset no tube with D/t above 51.5 folded in pure concertina, and every tube above 100 folded in the diamond mode, which these formulas don't model.
How close are the formulas to tests?
Between −9% and +20% on the square and rectangular steel tubes here, and up to 30% off on circular ones. The table puts the headline formulas against published tests, each with the flow stress its source used. The first row is a published worked number: Ronchietto et al. (2009) compute 34.53 kN for their 50 × 50 × 1.6 mm tubes at σ0 = 328 MPa, and the calculator gives 34.53 kN.
| Test | Section | σ0 used | Predicted | Measured | Diff. |
|---|---|---|---|---|---|
| Ronchietto et al. 2009, quasi-static, single tubes 400-700 mm | 50 × 50 × 1.6 | 328 MPa the paper's static yield | 34.53 kN | 28.87 kN | +20% |
| Velmurugan and Muralikannan 2009, quasi-static | 60 × 60 × 1.6 | 314.12 MPa yield | 35.14 kN | 38.30 kN | −8% |
| Velmurugan and Muralikannan 2009, quasi-static | 70 × 50 × 1.6 | 292.99 MPa yield | 32.78 kN | 35.55 kN | −8% |
| Abramowicz and Jones 1984 mild steel tubes (Anand et al. 2024 dataset rows 37-38) | 49.3 × 49.3 × 1.63 | 296.5 MPa (σy + σu)/2 | 32.04 kN | 35.28 kN | −9% |
| Zhang and Zhang, Q235 (Anand et al. 2024 dataset rows 11-13) | 40 × 40 × 1.2 | 270.9 MPa (σy + σu)/2 | 16.39 kN | 16.11 kN | +2% |
| Abramowicz and Jones 1984 steel tube, concertina (Anand et al. 2024 dataset row 1) | D 55.88 × 1.2 | 279 MPa (σy + σu)/2 | 22.96 kN | 32.60 kN | −30% |
| Guillow et al. 2001 AA6060-T5, concertina (Anand et al. 2024 dataset row 33) | D 48.4 × 1.35 | 196.3 MPa (σy + σu)/2 | 18.41 kN | 17.50 kN | +5% |
| Velmurugan and Muralikannan 2009, quasi-static (75 mm as given) | D 75 × 1.6 | 266.94 MPa yield | 39.16 kN | 53.80 kN | −27% |
Square and rectangular: Abramowicz and Jones 1984, κ 0.73. Circular: Abramowicz and Jones 1986. Measured: the mean of the quasi-static tests named. Anand et al. tabulate the tubes' mean radius; D is twice it.
The widest square-tube miss is Ronchietto's own: their prediction, made with the static yield, is +20% on their tubes. On circular tubes the formula is +5% on Guillow's aluminium tube and −27% to −30% on the two steel tubes. Treat any one formula as an estimate within that spread, and confirm with a test or a simulation.
The fold length checks out too: Ronchietto's 700 mm tube crushed 505 mm in 23 lobes, 22.0 mm a lobe, and 2Hκ is 22.9 mm. Dropped on by 340 kg, their 300 mm single tubes give a dynamic check:
| Impact speed | Measured Pmd | D 6,844, q 3.91 | D 40.4, q 5 |
|---|---|---|---|
| 7.99 m/s | 45.97 kN | 44.5 kN (−3%) | 70.9 kN (+54%) |
| 7.25 m/s | 40.81 kN | 44.2 kN (+8%) | 70.2 kN (+72%) |
| 6.49 m/s | 40.96 kN | 44.0 kN (+7%) | 69.5 kN (+70%) |
| 5.76 m/s | 42.24 kN | 43.7 kN (+3%) | 68.6 kN (+62%) |
With Abramowicz and Jones's constants the dynamic force is within 9% at every speed. Jones's traditional D 40.4, q 5 doubles the static force and overshoots by more than half: for mild steel crush tubes, use 6,844 and 3.91, or your own test data.
Why does the Week 4 crash box rebound instead of folding?
Because it carries about a tenth of the energy one fold needs. The challenge sends a 150 × 100 mm steel box, 250 mm long, into a rigid wall at 50 km/h on its own mass, first elastic, then MAT_024 at 1.2 mm and 1.5 mm. The brief says the run should either rebound completely or buckle and fold into self-contact. Only the first is possible:
| 1.2 mm | 1.5 mm | |
|---|---|---|
| Mass (solver) | 1,120 g | 1,400 g |
| Kinetic energy | 108 J = 1.08e+5 N·mm | 135 J = 1.35e+5 N·mm |
| Mean crush force (250 MPa, no rate effect) | 22.1 kN | 32.1 kN |
| Crush that absorbs all of it | 4.89 mm | 4.21 mm |
| Crush one fold takes up (2Hκ) | 38.4 mm | 41.4 mm |
| Share of one fold | 13% | 10% |
| Mass that folds it over the whole 182.5 mm stroke | 41.8 kg | 60.7 kg |
4.89 mm is about an eighth of one fold. The leading edge dents, the walls store the rest elastically, and the box springs back off the wall: case (1). The thicker box is stronger, so it dents even less (4.21 mm), though it carries 25% more energy. To see progressive folding, case (2), add mass behind the box: about 42 kg at 50 km/h folds the 1.2 mm box over its effective stroke, and about 9 kg completes one fold. At the 15 km/h of the RCAR low-speed structural test (40% overlap, rigid barrier at 10°), the same box stops about 460 kg.
Two more estimates explain what the reports see. When a bar hits a rigid wall, the end stops at once and a stress wave runs back along it:
For the card's steel, c = 5,172 m/s and at 13.9 m/s σ = 564 MPa, above the yields most reports typed (250 to 355 MPa) and the same at both thicknesses. So the impact end yields on first contact in both runs, which is why the peak stresses come out close. And the walls are slender, b/t = 125 at 1.2 mm. Elastic buckling of a long plate simply supported on its edges:
gives 48.6 MPa for the 150 mm wall and 109 MPa for the 100 mm wall at 1.2 mm (75.9 and 171 MPa at 1.5 mm): the walls buckle elastically long before they yield. These are estimates with the formula shown, not a substitute for the run.
One more thing to know. 1,120 g at 7.85e-3 g/mm³ and 1.2 mm means 1,18,896 mm² of shell, 4.9% less than 2 × (150 + 100) × 250 = 1,25,000 mm². The real mesh is a little smaller than the one set of dimensions a report gives. The mean crush force goes with C1/3, so that is about 2% on the force. With the nominal area the 1.2 mm box would weigh 1,178 g.
What did 231 Week 4 reports find?
Skill-Lync read all 281 public crash box and crush tube reports on 22 Nov 2026: 231 titled "Week 4 - Crash Box Simulation", 14 on the aluminium crush tube project and the rest on other crash box and crush tube projects. 38 of the Week 4 reports are mostly screenshots (under 2,500 characters of text), so every count below is a floor.
- Rebound: 94 reports describe the box rebounding or bouncing back, outside the brief's own sentence (which 53 quote). Example: report.
- Velocity typed: 13.88 mm/ms in 78, 13.89 in 34, 13.888 or 13.889 in 40; one typed 4.166, which is 15 km/h, not 50 (report). 13.88 is 0.06% slow: 0.1% less energy, harmless.
- Yield on the MAT_024 card: 250 MPa in 22, 345 or 355 in 19, 310 in 13, 350 in 8. None of the 231 mentions SRC, SRP or Cowper-Symonds in its text: the runs are rate-insensitive.
- The kinetic energy quoted: 34 reports write it in the text; 8 put a unit on it that doesn't fit. 1.08e5 written as joules (report), 110 kJ (report), 0.09E+6 kJ for internal energy (report), 0.1075 MJ (report), 1.08E+05 N/mm, which is not an energy (report). All of them are 108 J.
- Acceleration: 201 reports plot it, 20 give the unit in the text, as mm/ms². 1 mm/ms² is 1,000 m/s², 102 g: a node peak of 135 mm/ms² is 13,800 g.
- Termination time: 2 ms in 19, 1 ms in 17, 3 ms in 17, 5 ms in 13, 10 ms in 7. At 13.9 m/s the box covers 13.9 mm a millisecond: a 1 ms run sees the impact only if the wall starts closer than that, and a few milliseconds of contact are needed to see the rebound.
11 reports share one sentence giving the total energy as 0.11E+6 and 0.14E+6 N-mm: those units are right, 110 J and 140 J (rounded).
What are the LS-DYNA units for energy, force and stress in gm-mm-ms?
N·mm (one millijoule), newtons and MPa. LS-DYNA has no built-in units: the numbers you type fix them. In gm-mm-ms:
| Quantity | gm-mm-ms unit | Equals | kg-mm-ms | tonne-mm-s |
|---|---|---|---|---|
| Energy | g·mm²/ms² | 1 N·mm = 1 mJ | 1 J | 1 mJ |
| Force | g·mm/ms² | 1 N | 1 kN | 1 N |
| Stress | g/(mm·ms²) | 1 MPa | 1 GPa | 1 MPa |
| Acceleration | mm/ms² | 1,000 m/s² = 102 g | 1,000 m/s² | 0.001 m/s² |
| Velocity | mm/ms | 1 m/s | 1 m/s | 0.001 m/s |
| Density | g/mm³ | 10⁶ kg/m³ (7.85e-3 = 7,850) | 10⁹ kg/m³ | 10¹² kg/m³ |
So the 1.08e5 the solver prints for the 1.2 mm box is 108 J. The decoder in the second tab converts any value from the 13 systems of the consistent units calculator, which uses the same definitions.
How do you size a crash box for an impact?
Make the mean force times the stroke at least the kinetic energy, then check the deceleration it gives. The calculator works the impact as a constant mean force:
and the energy capacity, the specific energy absorption on the tube's own mass (p is the perimeter), and the wall that stops the mass within the stroke you choose (square and rectangular in closed form; circular by bisection):
Real crushing starts with a peak well above the mean, before the first fold forms. Velmurugan and Muralikannan measured peak/mean ratios of 2.75 (square), 3.25 (rectangular) and 2 (circular) on annealed steel tubes. The force chart draws the peak at that ratio and labels it indicative. Triggers (beads, holes, a tapered end) bring it down and are designed in simulation.
Long tubes can bend as a whole instead of folding. Ronchietto et al.'s 50 × 50 × 1.6 mm tubes folded quasi-statically up to L/C = 15 and bent at 16; dropped on at 8 m/s, they still folded at 18. The calculator warns above L/C 15. It also flags sections outside the C/t of 23.3 to 47.1 that Anand et al.'s 42 square-tube tests cover. The Week 4 box, at C/t = 104, is far outside it.
What this calculator does not cover
- The first peak force. No open source gives a formula Skill-Lync could check, so the chart only marks it at the measured ratios above.
- Guillow et al.'s empirical circular formula (Fav/Mp = 72.3 (D/t)0.32): left out because no open source Skill-Lync could read gives their definition of Mp.
- A global-bending criterion. Only Ronchietto's test lengths are used, as a warning.
- Foam-filled, multi-cell, tapered or triggered tubes, and oblique impact. The formulas are for empty, straight, single-cell tubes hit end-on.
Setting up a crash box run is the easy part; knowing what energy, force and deformation it should give before you press run is what crash teams hire for. The crash box is a project in Skill-Lync's LS-DYNA and crashworthiness courses.
Frequently asked questions
What is the formula for the mean crushing force of a square tube?
Abramowicz and Jones: Pm = 13.06 σ0 t^5/3 C^1/3, from Pm/M0 = 52.22 (C/t)^1/3 with M0 = σ0t²/4 and an effective crushing distance of 73%. For a rectangle, C = (b + h)/2. A 50 × 50 × 1.6 mm steel tube at 328 MPa gives 34.5 kN, the value Ronchietto et al. publish; their tubes measured 28.9 kN.
How much energy can a crash box absorb?
Mean crushing force times the effective stroke, about 73% of the length for a square steel tube. Strain rate raises the force for mild steel. The Week 4 box (150 × 100 × 1.2 mm, 250 mm long, 250 MPa) holds about 4,036 J statically, 37 times the 108 J it carries in at 50 km/h.
Why does the crash box rebound in the LS-DYNA Week 4 challenge?
It carries too little energy to fold. On its own 1120 g at 50 km/h it brings 108 J; at a mean crush force of 22.1 kN that stops it in 4.89 mm, an eighth of one fold (38.4 mm). The end dents and the elastic energy throws it back. Add mass to see folding.
Is 1.08e5 in LS-DYNA joules or kilojoules?
Neither. In the gm-mm-ms system the energy unit is g·mm²/ms², which is one N·mm, one millijoule. 1.08e5 of them is 108 J, the kinetic energy of 1.12 kg at 50 km/h. In kg-mm-ms the energy unit is kN·mm, one joule, and in tonne-mm-s it is N·mm again.
What is the specific energy absorption (SEA) of a crash box?
The energy absorbed over the effective stroke divided by the tube's own mass, in kJ/kg. Velmurugan and Muralikannan measured about 13 kJ/kg for annealed 60 × 60 × 1.6 mm steel square tubes and 18 kJ/kg for 75 × 1.6 mm circular tubes of the same material.
What is the effective crushing distance of a tube?
The share of the length a tube can crush before the folds pack solid. Abramowicz and Jones use 73% for mild steel square tubes, and Anand et al. use 77% for stainless steel. For circular tubes it is 0.86 − 0.568√(t/D): about 78% for a tube with D/t of 50.
Related on Skill-Lync
- Consistent units calculator: kg-mm-ms, gm-mm-ms, tonne-mm-s and 10 more systems
- Explicit time step calculator: the LS-DYNA time step for the shell mesh
- True stress-strain calculator: the MAT_024 curve from test data
- Energy absorption in a car crash: crumple zones explained with numbers
- FAQs in crashworthiness analysis
- Student project: Week 4 crash box simulation
- LS-DYNA courses compared
- All free engineering tools
Sources
- F. Ronchietto, S. Chung Kim Yuen, G. N. Nurick (University of Cape Town), "Response of axially stacked square tubes to axial impact loads", Latin American Journal of Solids and Structures 6 (2009) 413-440: Jones's Pm/M0 = 52.22 (C/H)1/3 and the dynamic form with 0.33V/C; 50 × 50 × 1.6 mm, 34.53 kN predicted; Table 1 quasi-static tests (lobes, global bending at 800 mm); Table 6 drop tests, 340 kg.
- R. Velmurugan, R. Muralikannan (IIT Madras), "Energy absorption characteristics of annealed steel tubes of various cross sections in static and dynamic loading", LAJSS 6 (2009) 385-412: C = (c + d)/2, 38.27 and 52.22 forms, 86.14 circular form, κ 0.77, 0.33V/c, D 6844 s⁻¹ and q 3.91; Tables 1-2 (yield, UTS, mean loads, SEA); peak/mean ratios.
- S. Anand, R. Alderliesten, S. G. P. Castro, "Crashworthiness in preliminary design: mean crushing force prediction for closed-section thin-walled metallic structures", International Journal of Impact Engineering 188 (2024) 104946 (open access, TU Delft repository): eqs. 12-17, κ 0.73 for mild steel and the 0.73/κ scaling, flow-stress definitions, Appendices A-B (42 square and 44 circular tests, materials), R² of each model.
- H. A. Ly, V. N. V. Nguyen, T. Thai-Quang, H. H. Nguyen, ASEAN Engineering Journal A 5(2) (2015): Abramowicz and Jones's circular mean load with κ = 0.86 − 0.568√(t/2R) and the 0.25V/(Rκ) strain rate, D 6844, q 3.91.
- A. A. N. Aljawi, "Axial crushing of square steel tubes", 6th Saudi Engineering Conference, KFUPM (2002): Pm = 9.56 σ0 t5/3W1/3 (Wierzbicki and Abramowicz 1983) and 13.06 (Abramowicz and Jones).
- N. S. Marshall, MSc thesis, University of Cape Town (1995), §2.3: D 40.4 s⁻¹, q 5 for mild steel ("traditionally") and the D 6844 s⁻¹, q 3.91 Abramowicz and Jones used. S. S. Hsu, N. Jones, LAJSS 1 (2004) 277-296: D 6500 s⁻¹, q 4 for aluminium alloy as listed in Jones's Structural Impact, and their 6063-T6 tests.
- MIT OpenCourseWare 2.080J Structural Mechanics, Lecture 11, Buckling of plates and sections (Fall 2013), eqs. 11.15 and 11.51: plate buckling, k = 4, σ = 4π²E/(12(1 − ν²)) (t/b)².
- Ansys LS-DYNA Verification Manual, VM-LSDYNA-IMPACT-004: speed of sound from E and ρ (5,189 mm/ms for E 210 GPa, ρ 7.8e-6 kg/mm³) and the first stress wave from the impedance, 0.3035 GPa.
- RCAR, Low-speed structural crash test protocol, Issue 2.5 (Sept 2024): 15 km/h, 40% overlap, rigid barrier at 10°. F. Pled, W. Yan, C. Wen, ACAM 2007: aluminium tube 2,700 kg/m³, E 68.9 GPa, ν 0.33.
- Skill-Lync, all 281 public crash box and crush tube reports, read 22 Nov 2026: velocities, yields, energies and units quoted. scripts/verify/crash_box.py re-derives the coefficients by minimising the fold energies, re-computes every case and the inverse by root-finding; the page checks itself against it at build time.