Finding 100 μm Cracks Before a Mining Casting Is Machined
Large mining castings are big, complex in shape and inspected on arrival before being released to machining. A micro-crack or a dimensional deviation that reaches machining can propagate under machining stress and break the part. This project uses CAE simulation to predict stress concentration zones, then scans those zones with an integrated 3D dimensional and crack detection system: key dimensions to within plus or minus 0.01 mm, cracks longer than 100 μm detected, cycle 900 seconds, subject to site acceptance results.
Simulate first, because full-surface high-resolution scanning is too expensive
Scanning the whole surface of a large casting at high resolution costs both time and compute. CAE predicts where machining stress concentrates, the system scans those high-risk zones closely, and micro-crack detection is preserved while the cycle stays acceptable. The simulation is not a formality; it is the mechanism that makes the cycle possible.
0.01 mm and 100 μm are two scales
Dimensional measurement is macroscopic: key dimensions are extracted from calibrated 3D data and compared with drawing tolerances. Crack detection is microscopic: surface micro-topography in stress zones is analysed to find crack features that differ from the rough cast texture. Both use one scan data set but completely different algorithms and thresholds, and no single number describes the capability of the whole system.
Integration removes the second setup
Dimensional scanning of the blank and stress-crack detection share the scan data and are processed separately. After integration, both conclusions stand on the same data, which removes a re-clamp and a second scan and, more importantly, the datum shift that would come between them.
Holding false calls down on a rough blank
Rough cast texture is easily confused with crack morphology. The system analyses micro-topography and uses a crack feature threshold to separate rough texture from a real crack. That threshold is not guesswork: it is calibrated from measured samples in stress zones and improved as samples accumulate.
CAE results also deviate from actual stress distribution, and a poor model means high-risk zones get missed. The project requires calibrating the model against inspection findings, and including casting grade and heat treatment in the simulation. That calibration is long-term work, not a one-off delivery item.
Where 900 seconds sits in the process
The cycle suits the goods-in gate before machining. The casting is large, so the scan path is planned by external form and completed in zones, with dense scanning in key areas to control total time. Conclusions are logged in the data management module, and both releases and blocks are traceable. Speed here is not a virtue in itself: a gate that releases a cracked casting quickly has failed at the only thing it was built to do.
If the premises fail, the scheme fails
It suits incoming and pre-machining inspection of mining castings and large construction machinery castings. The premises are a calibrated simulation model and a scan path that covers the key zones. When the cycle is tight, scanning can be tiered by risk, but the risk of sampling low-risk zones then has to be accepted explicitly rather than assumed away.
Questions before a project is approved
Why CAE first?
Full-surface high-resolution scanning takes too long. Predicting stress zones lets the system scan them closely, balancing crack detection against cycle time.
How is a 100 μm crack found in scan data?
Surface micro-topography is captured in stress zones, crack features are separated from rough texture by morphology, and a threshold controls false calls.
Do dimensions and cracks share one data set?
Yes. Dimensional scanning and stress-crack detection are integrated, share the scan data and process separately, avoiding a second clamp and a second scan.
Will rough surfaces cause crack false calls?
They interfere. Morphology analysis and a crack threshold separate rough texture from crack, and real performance is subject to site acceptance results.