Industry
Cloud-native CAE (Computer-Aided Engineering)
Company
Luminary Cloud
Project Setup & Geometry Validation
From Stuck Uploads to Clear Guidance: Designing Geometry Issue Reporting
Before, engineers faced opaque “upload failed” messages, long delays, and no clear way to diagnose geometry issues — leaving projects stalled before simulations could even start.
Role
Lead Product Designer, acting PM
Time
3 months (concept to launch)
Team
1 designer (me), 2 engineers (platform, 3D viz), 2 physicists (geometry, mesh)
Problem Space
Wait time too long: Users got stuck at the upload dialog with no visibility.
Opaque errors: Geometry issues were either hidden or surfaced as random technical codes.
Result: Projects often stalled before users could even run their first simulation, blocking adoption.
Research & Diagnosis
Analyzed error logs across hundreds of failed uploads.
Conducted user interviews with CFD and mechanical engineers to uncover pain points.
Mapped geometry failure patterns (gaps, overlaps, CAD/CFD mismatches, unsupported features).
Key insight: Most issues weren’t solver bugs — they were geometry readiness problems that users couldn’t detect or fix without clear feedback.
Design Process
Design Goals
Transparency: Show upload progress and surface geometry issues early.
Actionability: Provide clear, plain-language error messages and guidance.
Efficiency: Reduce idle wait time by enabling parallelizable setup tasks.
Resilience: Improve error handling for partial imports and large files.
Explorations
Collaborated with computational geometry engineers to define which errors could be automatically diagnosed.
Explored inline error reporting within the 3D viewer — highlighting problematic regions directly on models.
Designed stepwise project setup to allow work on naming, units, or boundary conditions while geometry processed in the background.
User Testing
Ran usability sessions with CFD engineers (primary) and mechanical engineers (target).
Tested large STEP/STL files, multi-part assemblies, and edge cases like missing units.
Findings:
Visual highlights + plain language beat raw error logs.
Users valued the ability to continue setup tasks while geometry uploaded.
Guidance on how to fix issues before re-upload was essential.
Iterations & Solution
Introduced a Geometry Validation Panel with:
List of detected issues (e.g., “Non-manifold edge at propeller blade tip”).
3D highlights showing the exact problem areas.
Suggested next steps or documentation links.
Designed progress indicators and parallel setup flows to reduce perceived wait times.
Improved error handling to allow partial imports rather than total failure.
Preserved backward compatibility with legacy workflows while layering new UI.
Impact / Results
Reduced average geometry upload failures by surfacing errors earlier.
Shortened time to first simulation by enabling users to resolve issues upfront.
Improved user trust by making error handling visible, predictable, and recoverable.
Helped unlock adoption with mechanical engineers new to CFD workflows.
Reflection / Learnings
Even in highly technical products, clarity > cleverness. Engineers preferred plain language and actionable feedback over technical codes.
Designing under constraints meant prioritizing quick wins with high visibility (error handling, visual highlights) while planning longer-term automation.
Future opportunity: auto-repair suggestions (healing small gaps, auto-scaling units) to reduce the need for external CAD cleanup.
Problem Space
Wait time too long: Users got stuck at the upload dialog with no visibility.
Opaque errors: Geometry issues were either hidden or surfaced as random technical codes.
Result: Projects often stalled before users could even run their first simulation, blocking adoption.
Research & Diagnosis
Analyzed error logs across hundreds of failed uploads.
Conducted user interviews with CFD and mechanical engineers to uncover pain points.
Mapped geometry failure patterns (gaps, overlaps, CAD/CFD mismatches, unsupported features).
Key insight: Most issues weren’t solver bugs — they were geometry readiness problems that users couldn’t detect or fix without clear feedback.
Design Process
Design Goals
Transparency: Show upload progress and surface geometry issues early.
Actionability: Provide clear, plain-language error messages and guidance.
Efficiency: Reduce idle wait time by enabling parallelizable setup tasks.
Resilience: Improve error handling for partial imports and large files.
Explorations
Collaborated with computational geometry engineers to define which errors could be automatically diagnosed.
Explored inline error reporting within the 3D viewer — highlighting problematic regions directly on models.
Designed stepwise project setup to allow work on naming, units, or boundary conditions while geometry processed in the background.
User Testing
Ran usability sessions with CFD engineers (primary) and mechanical engineers (target).
Tested large STEP/STL files, multi-part assemblies, and edge cases like missing units.
Findings:
Visual highlights + plain language beat raw error logs.
Users valued the ability to continue setup tasks while geometry uploaded.
Guidance on how to fix issues before re-upload was essential.
Iterations & Solution
Introduced a Geometry Validation Panel with:
List of detected issues (e.g., “Non-manifold edge at propeller blade tip”).
3D highlights showing the exact problem areas.
Suggested next steps or documentation links.
Designed progress indicators and parallel setup flows to reduce perceived wait times.
Improved error handling to allow partial imports rather than total failure.
Preserved backward compatibility with legacy workflows while layering new UI.
Impact / Results
Reduced average geometry upload failures by surfacing errors earlier.
Shortened time to first simulation by enabling users to resolve issues upfront.
Improved user trust by making error handling visible, predictable, and recoverable.
Helped unlock adoption with mechanical engineers new to CFD workflows.
Reflection / Learnings
Even in highly technical products, clarity > cleverness. Engineers preferred plain language and actionable feedback over technical codes.
Designing under constraints meant prioritizing quick wins with high visibility (error handling, visual highlights) while planning longer-term automation.
Future opportunity: auto-repair suggestions (healing small gaps, auto-scaling units) to reduce the need for external CAD cleanup.
Problem Space
Wait time too long: Users got stuck at the upload dialog with no visibility.
Opaque errors: Geometry issues were either hidden or surfaced as random technical codes.
Result: Projects often stalled before users could even run their first simulation, blocking adoption.
Research & Diagnosis
Analyzed error logs across hundreds of failed uploads.
Conducted user interviews with CFD and mechanical engineers to uncover pain points.
Mapped geometry failure patterns (gaps, overlaps, CAD/CFD mismatches, unsupported features).
Key insight: Most issues weren’t solver bugs — they were geometry readiness problems that users couldn’t detect or fix without clear feedback.
Design Process
Design Goals
Transparency: Show upload progress and surface geometry issues early.
Actionability: Provide clear, plain-language error messages and guidance.
Efficiency: Reduce idle wait time by enabling parallelizable setup tasks.
Resilience: Improve error handling for partial imports and large files.
Explorations
Collaborated with computational geometry engineers to define which errors could be automatically diagnosed.
Explored inline error reporting within the 3D viewer — highlighting problematic regions directly on models.
Designed stepwise project setup to allow work on naming, units, or boundary conditions while geometry processed in the background.
User Testing
Ran usability sessions with CFD engineers (primary) and mechanical engineers (target).
Tested large STEP/STL files, multi-part assemblies, and edge cases like missing units.
Findings:
Visual highlights + plain language beat raw error logs.
Users valued the ability to continue setup tasks while geometry uploaded.
Guidance on how to fix issues before re-upload was essential.
Iterations & Solution
Introduced a Geometry Validation Panel with:
List of detected issues (e.g., “Non-manifold edge at propeller blade tip”).
3D highlights showing the exact problem areas.
Suggested next steps or documentation links.
Designed progress indicators and parallel setup flows to reduce perceived wait times.
Improved error handling to allow partial imports rather than total failure.
Preserved backward compatibility with legacy workflows while layering new UI.
Impact / Results
Reduced average geometry upload failures by surfacing errors earlier.
Shortened time to first simulation by enabling users to resolve issues upfront.
Improved user trust by making error handling visible, predictable, and recoverable.
Helped unlock adoption with mechanical engineers new to CFD workflows.
Reflection / Learnings
Even in highly technical products, clarity > cleverness. Engineers preferred plain language and actionable feedback over technical codes.
Designing under constraints meant prioritizing quick wins with high visibility (error handling, visual highlights) while planning longer-term automation.
Future opportunity: auto-repair suggestions (healing small gaps, auto-scaling units) to reduce the need for external CAD cleanup.
