Validation
Nineteen cases with known answers run through the same pipeline your studies use, and every one lands inside its tolerance.
25 of 25 within tolerance
Each bar runs from 0 to the case’s tolerance, so a shorter bar means a closer answer.
| Case | Known answer | Wobbl | Error | Tolerance | Result | Confidence |
|---|---|---|---|---|---|---|
| Textbook formulas | ||||||
| Cantilever tip deflectionA steel bar fixed at one end with a weight at the tip, and how far the tip drops | 0.8016 mm | 0.7956 mm | 0.74% | 3% | Pass | High |
| Cantilever bending stress at mid spanThe same steel bar, and the bending stress halfway along it | 30 MPa | 29.94 MPa | 0.21% | 3% | Pass | High |
| Plate with a hole in tensionA flat plate with a hole in the middle, pulled in tension | 30.35 MPa | 29.89 MPa | 1.54% | 4% | Pass | High |
| Thick walled pressure vesselA thick walled tube holding very high pressure inside | 231.3 MPa | 232.3 MPa | 0.43% | 3% | Pass | High |
| Round shaft in torsionA round steel shaft held at one end and twisted at the other | 55.13 MPa | 55.6 MPa | 0.85% | 3% | Pass | High |
| Cantilever under its own weightA one meter steel bar, fixed at one end, sagging under its own weight | 1.443 mm | 1.441 mm | 0.16% | 3% | Pass | High |
| Ball dropped 300 m without air | 7.822 s | 7.822 s | 0.00% | 0.1% | Pass | High |
| Granite sphere dropped 300 m through air | 56.7 m/s | 56.7 m/s | 0.00% | 0.1% | Pass | High |
| Ball thrown at 30 m/s and 40 deg without air | 90.38 m | 90.38 m | 0.00% | 0.1% | Pass | High |
| Vertical rocket at constant thrust in a vacuum | 412.8 m/s | 412.8 m/s | 0.00% | 0.1% | Pass | High |
| Vertical rocket at constant thrust in a vacuum | 757.7 m | 757.7 m | 0.00% | 0.1% | Pass | High |
| Payload under a 2 m parachute | 4.122 m/s | 4.122 m/s | 0.00% | 0.1% | Pass | High |
| Industry benchmarks | ||||||
| NAFEMS LE1 elliptic membraneAn oval ring pulled outward around its outer edge, a standard test for stress solvers | 92.7 MPa | 92.92 MPa | 0.23% | 4% | Pass | High |
| NAFEMS LE10 thick plate under pressureA thick oval plate pressed on its top face, a standard test for 3D stress | -5.38 MPa | -5.412 MPa | 0.59% | 5% | Pass | Medium |
| Real parts from the bake off | ||||||
| Cast motor mount with a belt pull through four boltsA cast aluminum motor mount pulled by a conveyor belt through four bolts | 124.3 MPa | 123.7 MPa | 0.49% | 5% | Pass | Medium |
| ABS enclosure shell pressed near a cornerThe molded plastic shell of a device, pressed with 250 N near one corner | 93 MPa | 92.98 MPa | 0.02% | 10% | Pass | Medium |
| PA12 truss drone arm at full thrustA printed drone arm with truss windows, at full motor thrust and torque | 11.5 MPa | 11.49 MPa | 0.11% | 8% | Pass | High |
| PLA shelf bracket, 0.5 mm inside cornerA printed shelf bracket with a sharp inside corner, resting against a wall | 78 MPa | 80.48 MPa | 3.18% | 8% | Pass | High |
| PLA shelf bracket, 6 mm filletThe same shelf bracket with its inside corner rounded to 6 mm | 25.3 MPa | 26.38 MPa | 4.26% | 5% | Pass | Medium |
| Heat | ||||||
| Conduction through a slab into convectionAn aluminum bar held at 100 C at one end and cooled by air at the other | 97.67 C | 97.67 C | 0.00% | 1% | Pass | High |
| Aluminum cube cooling in airA small aluminum cube cooling down in air for ten minutes | 58.01 C | 58.27 C | 0.44% | 1% | Pass | High |
| Aluminum bar whose end jumps to 120 CAn aluminum bar at 20 C whose end is suddenly held at 120 C, read 10 mm in after 2 s | 52.68 C | 52.23 C | 0.85% | 2% | Pass | High |
| Circuits | ||||||
| Red LED on 5 V through 150 ohmA red LED on 5 V through a 150 ohm resistor | 20 mA | 20 mA | 0.00% | 0.5% | Pass | High |
| RC chargingA capacitor charging through a resistor | 3.161 V | 3.161 V | 0.00% | 0.2% | Pass | High |
| Buck converter output rippleThe output ripple of a 12 V to 5 V buck converter | 7.949 mV | 8.561 mV | 7.70% | 15% | Pass | High |
Every number above comes from one full run of wobbl validate, which took 29 minutes on a 2018
six core laptop, except the bar whose end jumps to 120 C, which joined in 0.1.1 and comes from a
run of its own. Each case whose reference is exact for the modeled setup lands within 2%. The
real parts are compared with reference models that were refined until they stopped changing,
and their tolerances follow how settled those references were.
Medium confidence in this table is Wobbl naming what a number leans on, not a miss. On NAFEMS LE10 the 2 mm band that stands in for the benchmark’s line support makes a peak far above the answer, which is excluded and named. On the cast motor mount the verdict flips if the servo is treated as rigid instead of a bolt group. On the enclosure shell large deflection theory lowers the peak by 11%, so the linear answer is the conservative one, and the reference is linear too. On the filleted bracket the bonded screw bore reads nearly twice the real peak. The buck converter is compared with the ideal waveform formula, which leaves out the switch and diode drops that ngspice models, so it gets a wider tolerance.
You can run every case yourself with wobbl validate.