Consistent units calculator for LS-DYNA, Radioss and Abaqus
Pick a mass, length and time unit and see the force, stress, energy, density and gravity units that go with them. Convert any value between 13 unit systems, from kg-mm-ms and tonne-mm-s to slug-ft-s. Free, no sign-up, runs in your browser.
The third tab checks a material card: it finds which system your density and modulus are really in, from the wave speed and the time step they give. The fourth checks the Week 1 "consistency of units" table of Skill-Lync's LS-DYNA course, cell by cell.
Exact factors from NIST SP 811 · LS-DYNA, Abaqus, Radioss manuals · 325 reports read · Updated 20 Nov 2026
Every derived unit follows from force = mass × acceleration. The table on the right lists all 15 quantities with their size in SI, and steel in that system.
Type numbers like 7.85e-9 or 2.1E5. "In" lists everyday units first, then each system's own unit.
Filled with a common slip: a kg-mm-ms model with E typed in MPa and gravity in m/s². Steel and aluminium are checked against handbook values; for other materials only the wave speed is judged.
Type the numbers from one column of your table (leave blank what you didn't do). Filled with a tonne-mm-s column with the slip the reports make most often in density.
| Quantity | Unit in kg-mm-ms | = in SI |
|---|---|---|
| Mass | kg | 1 kg |
| Length | mm | 0.001 m |
| Time | ms | 0.001 s |
| Force | kg·mm/ | 1 kN |
| Stress, pressure, modulus | kg/ | 1 GPa |
| Energy, work | kg·mm²/ | 1 J |
| Density | kg/ | 1e9 kg/m³ |
| Velocity | mm/ | 1 m/s |
| Acceleration, gravity | mm/ | 1000 m/s² |
| Power | kg·mm²/ | 1 kW |
| Momentum, impulse | kg·mm/ | 1 N·s |
| Spring stiffness, force per length | kg/ | 1e6 N/m |
| Dynamic viscosity | kg/ | 1e6 Pa·s |
| Mass moment of inertia | kg·mm² | 1e-6 kg·m² |
| Strain rate, frequency | 1/ | 1000 1/s |
- Steel in kg-mm-ms: RO 7.85e-6, E 210, gravity 0.009807.
What "consistent units" means
An FE solver is a calculator with no unit keys. LS-DYNA, Radioss, OpenRadioss and Abaqus read plain numbers from the input deck and never ask what they mean. The one thing they assume is Newton's second law with a factor of 1:
One force unit is one mass unit times one length unit per time unit squared. So you choose only three units, for mass, length and time. Everything else follows. Pick kilograms, millimetres and milliseconds, and the force unit is 1 kg × 1 mm/ms² = 1,000 N, a kilonewton. Stress is force per mm², so a kN/mm², which is a GPa. Energy is kN × mm, which is a joule.
The solver never notices if you mix systems. Type steel's density in kg/m³ and its modulus in MPa, and it will run a material whose sound speed is wrong by a factor of thousands. The result is a time step that is far too small or far too big, a model that sags under 1,000 times gravity, or an answer that looks fine and is off by a power of ten.
Consistent units table: 13 systems
The units of force, stress and energy in each system, and what one of each is in SI:
| Mass, length, time | Force | Stress | Energy |
|---|---|---|---|
| kg-m-s | kg·m/s² = 1 N | kg/(m·s²) = 1 Pa | kg·m²/s² = 1 J |
| kg-cm-s | kg·cm/s² = 10 mN | kg/(cm·s²) = 100 Pa | kg·cm²/s² = 100 µJ |
| kg-cm-ms | kg·cm/ms² = 10 kN | kg/(cm·ms²) = 100 MPa | kg·cm²/ms² = 100 J |
| kg-mm-s | kg·mm/s² = 1 mN | kg/(mm·s²) = 1 kPa | kg·mm²/s² = 1 µJ |
| kg-mm-ms | kg·mm/ms² = 1 kN | kg/(mm·ms²) = 1 GPa | kg·mm²/ms² = 1 J |
| g-cm-s | dyne = 10 µN | dyn/cm² = 100 mPa | erg = 100 nJ |
| g-cm-ms | g·cm/ms² = 10 N | g/(cm·ms²) = 100 kPa | g·cm²/ms² = 100 mJ |
| g-cm-us | g·cm/µs² = 10 MN | Mbar = 100 GPa | g·cm²/µs² = 100 kJ |
| g-mm-s | g·mm/s² = 1 µN | g/(mm·s²) = 1 Pa | g·mm²/s² = 1 nJ |
| g-mm-ms | g·mm/ms² = 1 N | g/(mm·ms²) = 1 MPa | g·mm²/ms² = 1 mJ |
| t-mm-s | t·mm/s² = 1 N | t/(mm·s²) = 1 MPa | t·mm²/s² = 1 mJ |
| slug-ft-s | lbf = 4.448 N | lbf/ft² (psf) = 47.88 Pa | ft·lbf = 1.356 J |
| lbfs2in-in-s | lbf = 4.448 N | psi = 6.895 kPa | in·lbf = 113 mJ |
Steel (7,850 kg/m³, 210 GPa, as in the LS-DYNA keyword manual's own example), the 56.33 km/h (35 mph) NCAP full-frontal barrier speed and standard gravity (9.80665 m/s²) written in each system. The LS-DYNA support site's table uses 7,830 kg/m³ and 207 GPa, so its numbers differ in the third digit:
| System | RO, steel | E, steel | 56.33 km/h | Gravity |
|---|---|---|---|---|
| kg-m-s | 7850 | 2.1e11 | 15.65 | 9.807 |
| kg-cm-s | 0.00785 | 2.1e9 | 1565 | 980.7 |
| kg-cm-ms | 0.00785 | 2100 | 1.565 | 9.807e-4 |
| kg-mm-s | 7.85e-6 | 2.1e8 | 15650 | 9807 |
| kg-mm-ms | 7.85e-6 | 210 | 15.65 | 0.009807 |
| g-cm-s | 7.85 | 2.1e12 | 1565 | 980.7 |
| g-cm-ms | 7.85 | 2.1e6 | 1.565 | 9.807e-4 |
| g-cm-us | 7.85 | 2.1 | 0.001565 | 9.81e-10 |
| g-mm-s | 0.00785 | 2.1e11 | 15650 | 9807 |
| g-mm-ms | 0.00785 | 210000 | 15.65 | 0.009807 |
| t-mm-s | 7.85e-9 | 210000 | 15650 | 9807 |
| slug-ft-s | 15.2 | 4.386e9 | 51.34 | 32.17 |
| lbfs2in-in-s | 7.35e-4 | 3.046e7 | 616 | 386.1 |
lbf·s²/in is the mass that 1 lbf accelerates at 1 in/s² (175.1 kg); a slug is the mass 1 lbf accelerates at 1 ft/s² (14.59 kg). Both are exact from 1 lb = 0.45359237 kg and gn = 9.80665 m/s².
How to convert any quantity yourself
Write the quantity's dimensions as powers of mass, length and time: density is M¹L⁻³T⁰, stress is M¹L⁻¹T⁻², acceleration is L¹T⁻². Then divide the SI value by the size of each new unit raised to that power. With M, L and T the sizes of the new units in kg, m and s:
For steel's density in tonne-mm-s: M = 1,000 kg, L = 0.001 m, the powers are 1 and −3. So 7,850 / (1000 × 0.001⁻³) = 7,850 / 10¹² = 7.85 × 10⁻⁹ t/mm³. The modulus, M¹L⁻¹T⁻² with T = 1 s, gives 210 × 10⁹ / (1000 × 1000) = 210,000, which is MPa. In kg-mm-ms (M = 1, L = T = 0.001) the same modulus is 210 × 10⁹ / (0.001⁻¹ × 0.001⁻²) = 210 GPa.
The slip to watch for is the time unit in a squared term. A millisecond is 10⁻³ s, but a millisecond squared is 10⁻⁶ s². That is why gravity in kg-mm-ms is 9.81 × 10³ mm / 10⁶ ms² = 0.00981, not 9.81. The LS-DYNA manual's own *LOAD_BODY_Z example uses exactly that: a constant curve of 1 with scale factor 0.00981 "in units of mm/ms²".
The 30-second check: wave speed and time step
An explicit solver takes the time step from the time a stress wave needs to cross the smallest element. For a shell (the LS-DYNA theory manual's form):
Ls is the element's characteristic length and TSSFAC is the scale factor on *CONTROL_TIMESTEP: 0.9 by default, lower in some contact and blast set-ups. For solids the sound speed is √(E(1−ν)/((1+ν)(1−2ν)ρ)) and the step is about L/c, a little less with bulk viscosity. Compute c from your own RO, E and PR in your own units, then turn it back into m/s. Steel must give about 5,422 m/s and aluminium about 5,394 m/s. A 5 mm steel shell then gets Δt = 0.9 × 5 mm / c = 0.83 µs, which is 8.3e-4 ms in a kg-mm-ms deck and 8.3e-7 s in a tonne-mm-s deck.
Each common slip gives a signature factor. E typed in MPa in a kg-mm-ms model makes c 31.6 times too high (√1000), so the time step is 31.6 times too small and the run takes 31.6 times longer. Density left in kg/m³ in a tonne-mm-s model makes c a million times too small. The third tab does this arithmetic and tells you which system each number really belongs to.
Which system to use
- kg, mm, ms: force in kN, stress in GPa, energy in joules. Steel RO = 7.85e-6, E = 210, gravity 0.00981. Time steps print near 10⁻³ ms and energies in J, which suits crash and impact work.
- tonne, mm, s (Abaqus "SI (mm)"): force in N, stress in MPa, energy in mJ. Data-sheet stresses go in as printed, but density is a tiny 7.85e-9 and time steps print near 10⁻⁷ s. Common for implicit and static work.
- g, mm, ms: also N and MPa, with the millisecond of crash work. Steel RO = 7.85e-3, E = 210,000. Energies print in mJ.
- kg, m, s (SI): no conversion at all, but meshes are in metres, so a 5 mm element is 0.005 and CAD imports usually need scaling.
- Avoid kg, mm, s: force in millinewtons and the steel modulus as 2.1e8. It is consistent but easy to misread.
Whichever you choose, every number in the deck has to follow it. That includes load curves (their time axis too), contact and spring stiffnesses, initial velocities, gravity and material curves.
What 325 Week 1 reports got wrong
Skill-Lync's LS-DYNA course opens with this exercise: write 1 kg, 1 m, 1 s, 1 N, 1 Pa, 1 J, 1 kg/m³, 210 GPa, 56.33 km/h and 9.8 m/s² in 12 systems. We fetched all 325 public reports (2019 to 2025) and checked a random 59 against the exact values above, eight high-risk cells each. 43 had a readable table; the rest were attachments, images or cut off.
- Most cells are right, most reports are not. 252 of 299 readable cells (84%) were correct, but 22 of the 36 reports with six or more readable cells (61%) had at least one wrong.
- The US system is the weak spot. The slug-ft-s modulus was wrong in 14 of 36 reports, 8 of them by a lost 10⁹ ("4.368" for 4.386 × 10⁹). Of the 22 right ones, 8 read 4.368 × 10⁹: within 0.4% only because the stress factor was rounded to 0.0208 (it is 0.020885).
- The microsecond row breaks in different ways. 12 of 40 got the g-cm-µs modulus or velocity wrong: two inverted the time factor (E = 2.1 × 10²⁴), two gave the factor for 1 km/h instead of 56.33 km/h.
- The tonne slip hits two cells at once. Taking 1 kg as 0.1 or 10⁻⁴ tonne puts the tonne-mm-s density at 10⁻¹⁰ or 10⁻¹³ and the modulus off by the same factor. 31 of 36 had the density (10⁻¹²) right.
- Gravity is rarely the problem in the table: 37 of 39 kg-mm-ms gravity cells were right (71% of reports started from 9.8, 21% from 9.81). The usual failure there is a conversion factor typed where the value belongs.
Some tables lose exponents when the page is flattened, so a few "lost 10⁹" cases may be display damage rather than the student's slip, and several reports share copied rows, so the counts overstate independent mistakes.
Units the table doesn't show
- Strain-rate parameters are 1/time. A Cowper-Symonds C given in 1/s is divided by 1,000 for a millisecond model.
- Load curves carry units on both axes. A velocity-time curve built in m/s against seconds must become mm/ms against ms (the velocity number stays the same, the time axis is multiplied by 1,000).
- Temperature isn't one of the three base units. Thermal conductivity and specific heat need the system's power, energy, mass and length units plus your temperature unit; expansion coefficients (1/K) don't change.
- Energies in the output are in the system's energy unit: J in kg-mm-ms, mJ in tonne-mm-s and g-mm-ms. The same crash shows an internal energy 1,000 times bigger in tonne-mm-s.
- Strain, Poisson's ratio and friction coefficients have no units and never change.
What this calculator does not cover
- Unit systems that use a temperature unit (thermal and coupled analyses): it converts the mechanical part only.
- Rotational quantities in degrees: solvers take some angles in degrees and some in radians, depending on the keyword. Check the manual entry.
- Solver unit settings. Radioss /BEGIN declares an input and a work unit system and the Starter converts between them, but not every keyword uses those units; this tool assumes one consistent set typed throughout.
- The material values are handbook round numbers (steel 7,850 kg/m³ and 210 GPa, aluminium 2,700 kg/m³ and 70 GPa). Real alloys differ by a few per cent, which the card check tolerates (3%).
Skill-Lync's LS-DYNA and crashworthiness courses go from unit systems and material cards to full-vehicle crash models, the work CAE teams at OEMs and suppliers hire for.
- LS-DYNA for Structural Mechanics/FEA (12 weeks; this unit table is its first challenge)
- Crashworthiness Analysis using HyperMesh and Radioss
- Post Graduate Program in CAE
Frequently asked questions
What are consistent units in LS-DYNA?
LS-DYNA, Radioss and Abaqus store plain numbers and never ask for units. They only assume that force = mass × acceleration holds with factor 1 in whatever units you typed. So once you pick a mass, a length and a time unit, the force, stress, energy and density units are fixed. In kg, mm, ms the force unit is the kN, stress is GPa and energy is the joule. Steel there is RO = 7.85e-6 and E = 210.
What are the units of stress and force in tonne-mm-s?
Force is the newton (1 t × 1 mm/s² = 1 N), stress is MPa (N/mm²) and energy is the millijoule (N·mm). Density has to be in t/mm³: steel at 7,850 kg/m³ is 7.85e-9. Gravity is 9,807 mm/s². This is the "SI (mm)" set in the Abaqus documentation, and you can type data-sheet values in MPa straight in.
How do I convert density from kg/m³ to kg-mm-ms or tonne-mm-s?
Divide by the system's density unit. 1 kg/mm³ is 10⁹ kg/m³, so steel's 7,850 kg/m³ is 7.85e-6 kg/mm³ in kg-mm-ms. 1 t/mm³ is 10¹² kg/m³, so in t-mm-s it is 7.85e-9. Time doesn't appear in density, so g-mm-ms gives 7.85e-3 g/mm³ and g-mm-s gives the same number.
What value of gravity do I use in kg-mm-ms?
9.80665 m/s² is 9,806.65 mm/s², and a millisecond squared is 10⁻⁶ s², so g = 0.00981 mm/ms² (9.81e-3). Typing 9.81 or 9810 in a kg-mm-ms model makes gravity 1,000 or 10⁶ times too strong. In t-mm-s and g-mm-s it is 9,807 mm/s², and in g-cm-µs it is 9.81e-10 cm/µs².
How do I check whether my LS-DYNA units are consistent?
Work out the sound speed from the material card in your own units, c = √(E/ρ(1−ν²)), and turn it back into m/s. Steel must give about 5,422 m/s and aluminium about 5,394 m/s. If you get a factor of 10, 31.6, 1,000 or 10⁶ off, the density and the modulus are in different systems. A second check: a 5 mm steel shell should give an explicit time step near 0.83 µs.
Which unit system is best for crash simulation?
Any consistent set gives the same answer; the choice is about readable numbers. kg-mm-ms keeps the time step near 10⁻³ ms, forces in kN, stresses in GPa and energies in joules. t-mm-s and g-mm-ms give newtons and MPa, which match material data sheets, but energies print in mJ, 1,000 times bigger numbers. Avoid kg-mm-s: force comes out in millinewtons and the steel modulus as 2.1e8.
Related on Skill-Lync
- LS-DYNA courses compared and LS-DYNA and crash project ideas
- All FEA courses compared
- Explicit time step calculator (time step, cycles and DT2MS in kg-mm-ms, tonne-mm-s or kg-m-s)
- Crash box calculator: what 1.08e5 means in gm-mm-ms, and the crash box challenge's energy, force and folds
- Stiffness calculator: beam, column and spring stiffness, element stiffness matrices
- Beam natural frequency calculator
- Yield strength calculator
- All free engineering tools
Sources
- LS-DYNA Keyword User's Manual Vol. I R14 (Ansys): "Units" (force unit = mass unit × acceleration unit; steel 7.85E-06 and 210.0 in mm, ms, kg, kN), *CONTROL_TIMESTEP (TSSFAC), *LOAD_BODY (gravity example, 0.00981 mm/ms²). Vol. II: *MAT_ELASTIC and *MAT_PIECEWISE_LINEAR_PLASTICITY fields RO, E, PR.
- LS-DYNA Theory Manual R14, ch. 25 "Time Step Control": shell Δt = Ls/c with c = √(E/ρ(1−ν²)), beam √(E/ρ), solid sound speed and bulk viscosity term.
- LS-DYNA support: "Definition of a consistent system of units" (15-system table, steel 7.83e3 kg/m³, 207 GPa).
- Abaqus documentation, "Conventions: units": "Abaqus has no built-in system of units"; SI, SI (mm), US (ft) and US (inch) columns.
- Altair Radioss: /BEGIN (input and work unit systems; "Radioss does not have a built-in unit system").
- NIST SP 811, Appendix B: ft, in, gn, lbf, slug and psi factors; the pound is the 1959 international pound, 0.45359237 kg exactly.
- 49 CFR 571.208 (FMVSS 208) and NHTSA NCAP test reports: 56.3 km/h (35 mph) full frontal rigid barrier.
- 325 public Skill-Lync student reports titled "consistency of units", read 20 Nov 2026 (audit sample described above).