Review the sample deliverable
See the executive decision, controlled inputs, numerical outputs, boundary and next action.
Open sample package →Solved case · inverse synthesis · version 1.0
Required loads in. Three feasible conical-bearing candidates out.
This public case closes the gap between saying that AURA supports inverse design and showing a reproducible, controlled result. The design question starts with required performance and a locked shaft interface—not a preselected bearing geometry.
SOLVED399 FEASIBLE STATESSCREENING ONLYCASE ID AURA-INV-CONE-80R-360A-30D-001
Screening result: within the stated search space, AURA found opposed-conical bearing states satisfying both static criteria. The three published candidates expose different envelope, pressure and flow trade-offs; none is presented as a production release.
1. Requirements and fixed decisions
80 NRequired radial load
360 NRequired axial load
30 mmLocked shaft diameter
| Input | Value | Role in this case |
|---|---|---|
| Support architecture | Opposed conical pair | Axial stiffness projected to the machine axis as Knormal sin²α. |
| Pressure basis | Absolute | Supply candidates are evaluated against pa = 1.0 bar absolute. |
| Radial criterion | Er,req ≤ 0.70 | Normalized radial demand ratio must remain inside the screening limit. |
| Axial criterion | Ez,req ≤ 0.70 | Normalized axial demand ratio must remain inside the screening limit. |
| Operating speed | 20,000 rpm | Recorded as a downstream machine requirement; not used to rank this static inverse search. |
2. Search and ranking
- Generate. The solver evaluates 4,032 combinations across controlled cone angle, length, clearance, supply pressure and feed configuration while preserving the 30 mm shaft interface.
- Reject. Invalid physical/model states are removed before feasibility scoring; 630 combinations were rejected in this run.
- Filter. Both radial and axial screening criteria must pass. The run returned 399 feasible states.
- Rank. The published set deliberately represents three engineering preferences: balanced margin, compact envelope and lower supply pressure.
3. Candidate comparison
| Candidate | Geometry | Supply / flow | Static outputs | Required ratios |
|---|---|---|---|---|
| Balanced margin Top-ranked balanced state | α 60° · L 32.0 mm C 20 µm · Rout 70.43 mm | 5.0 bar abs 2.72 m³/h | Kr 10.38 N/µm Ka 65.17 N/µm Wr,max 145.39 N Wa,max 912.35 N | Er 0.385 Ez 0.276 |
| Compact envelope Smallest published radial envelope | α 50° · L 22.4 mm C 20 µm · Rout 41.70 mm | 8.0 bar abs 4.35 m³/h | Kr 9.50 N/µm Ka 30.62 N/µm Wr,max 133.03 N Wa,max 428.69 N | Er 0.421 Ez 0.588 |
| Lower pressure Lowest published supply pressure | α 60° · L 48.0 mm C 30 µm · Rout 98.14 mm | 3.0 bar abs 2.45 m³/h | Kr 6.48 N/µm Ka 60.54 N/µm Wr,max 136.06 N Wa,max 1,271.27 N | Er 0.412 Ez 0.198 |
Why three answers? Inverse design does not remove engineering judgment. It makes the trade-off explicit. A compact state requires more pressure and flow; the lower-pressure state needs a much larger radial envelope; the balanced state retains the largest minimum screening margin of the published set.
4. Controlled handoff
What can move downstream
Case identity, fixed interface, geometry, absolute-pressure basis, feed architecture, static load capacities, projected stiffnesses, flow estimate and screening criteria can enter the machine-configuration record.
What is still missing
Manufacturing tolerances, frequency-dependent coefficients, thermal and pneumatic transients, complete rotor eigenanalysis, stability margin, uncertainty propagation and case-level experimental correlation.
Decision authority: SCREENING ONLY. The result supports candidate selection for the next engineering stage. It does not authorize geometry freeze, procurement, production speed operation or a universal performance claim. A chosen candidate must be re-evaluated in the actual spindle layout and evidence programme.
5. Reproducibility record
The public JSON record contains inputs, search counts, constraints, all three candidates and the evidence boundary. The solver remains proprietary, but the published result is no longer an unsupported capability statement.
From public evidence to your case
The same structure can be applied to one defined customer decision.
The public cases show the discipline. The feasibility package shows the format in which a customer-specific screen is returned.
Prepare six non-confidential items
Machine, RPM, loads, envelope, gas supply and the decision to be made.
Check project fitStart with the smallest useful answer
Use a Architecture Screen first; move to a pilot only when the decision requires deeper configuration or evidence.
Compare engagements →