Skip to main content
远算科技Yuansuan Technology
Enter GEWU
GEWUNEXT-GENERATION WORKBENCH ENHANCED BY ENGINEERING AI

Don't let great designs lose to compute limits.

Explore more design possibilities. Keep proven methods in use. Connect established CAE and native computation, build Engineering Apps and train specialized AI.

Discuss my design taskSee design exploration
Concept illustration of GEWU wheel simulation and design space exploration
13° wheel-impact analysis
See GEWU explore and validate design alternatives
VALIDATION BOTTLENECKS

What is holding your designs back?

  • Model preparation taking too long?How much time is left to improve the design?

  • Too many parameter combinations?Are only a few designs getting validated?

  • The results are ready.Does judgment still depend on a few experts?

  • Your tasks are ready to run.Still waiting for a CAE license?

  • More jobs need to run at once.Is your HPC cluster holding them in a queue?

  • Another design to evaluate.The same manual steps all over again?

PRODUCT INTRODUCTION

More design options. More confident choices.

Explore designs with models, load cases and constraints, and retain validated methods and review criteria.

GEWU· Engineering AI WorkbenchHOW IT WORKS
GEWU · Model · Explore the design space · Validate & generate AppsModel an engineering problem, explore the design space across variables, operating conditions, and constraints in parallel, validate better solutions, and package validated models, methods, and criteria as reusable Engineering Apps.Calculation modesEngineering modelObjects & modelsMaterials & conditionsGoals & criteriaIntelligent meshingLess model preparationCompatible Mode｜Industry gold standardCommercial CAE · existing licensesUse your current tools and workflowsNative Mode｜White-box computationPhysical assumptionsNumerical methodsSolver processValidation evidenceInspect · reproduce · improveDesign ExplorationDesign-space explorationVariables · conditions · constraintsCase 1Case 2Case 3Geometry parametersMaterial parametersOperating conditionsConstraint boundaryCandidate regionsExcluded regionsValidate resultsStructural response · normalized illustrationComputedReferenceTolerance bandCriteria metScope: validated materials and load casesEngineering Apps & AIModelMethodCriteriaPackage as AppParametric modelvalidation AppValidated versionGeometryParametersMaterialLibraryLoadLoad casesResponseReadyVerdictPassedReportViewParameters readyRunValidation feedbackRefine modelImprove methodAdjust constraints
  1. 01

    Create the engineering model

    GEWU · Create the engineering modelReuse existing CAE models Trace and repeat native solvesEngineering modelObjects & modelsMaterials & conditionsGoals & criteriaIntelligent meshingLess model preparationCompatible Mode｜Industry gold standardCommercial CAE · existing licensesUse your current tools and workflowsNative Mode｜White-box computationPhysicalassumptionsNumericalmethodsSolverprocessValidationevidenceInspect · reproduce · improve
  2. 02

    Explore the full design space

    GEWU · Explore the full design spaceExplore parameters, load cases, and constraints Screen designs with AI and high-fidelity checksDesign-space explorationVariables · conditions · constraintsCase 1Case 2Case 3Geometry parametersMaterial parametersOperating conditionsConstraint boundaryCandidate regionsExcluded regions
  3. 03

    Validate results, generate Engineering Apps

    GEWU · Validate results, generate Engineering AppsCheck responses, errors, and scope Package workflows and criteria for team reuseValidate resultsStructural response · normalized illustrationComputedReferenceTolerance bandCriteria metScope: validated materials and load casesModelMethodCriteriaPackage as AppParametric modelvalidation AppValidated versionGeometryParametersMaterialLibraryLoadLoad casesResponseReadyVerdictPassedReportViewParameters readyRun
    Validation feedback · Back to modeling & explorationRefine model · Improve method · Adjust constraints
ModelExplore the design spaceValidate & generate Apps

Four core GEWU capabilities

  1. Compatible & native computation

    Keep existing CAE Choose native computation

  2. Intelligent design exploration

    HPC and AI sampling Focus on key design options

  3. Build Engineering Apps

    Reuse validated methods Focus on design changes

  4. Train specialized AI

    Capture expert reasoning Support reviews of similar tasks

REAL TASK

Make design trade-offs with evidence

View the 13° wheel-impact validation case
13° wheel-impact analysisActual product interfaceReading guide
Left | Native ModeRight | Compatible Mode
GEWU · Compare solver results

Check load cases, units and color-scale ranges before comparing displacement.

COMPATIBLE & NATIVE COMPUTATION

Build on existing investments. Open, transparent computation methods.

When commercial CAE licenses are limited, assess native computation as another validation path for the task.

Compatible ModeIndustry gold standard

Use existing CAE models, solvers and commercial licenses.

Operating conditions

Parallel runs depend on both license capacity and compute resources.

View calculation workflow

Connect your existing solving workflow to design exploration, organize runs and compare results.

  1. Existing model
  2. Commercial CAE
  3. Result validation

Native ModeWhite-box computation

Trace physical assumptions, numerical methods and the solving process.

Conditions of use

Within supported physics and load cases, compare reference results and confirm error tolerances.

View calculation workflow

Inspect physical assumptions, numerical methods and the solving process to reproduce and improve the computation.

  1. Define the physicsMaterials · loads · boundaries
  2. Choose the methodMesh · algorithm · equations
  3. Inspect the solveIterations · logs · convergence
  4. Validate the resultError · criteria · scope

Both modes share design exploration, Engineering Apps and engineering AI.

Inspect the computation

Wheel impact | Reviewing the computation

Start with the model inputs. Inspect the basis of the calculation and connect the run and validation records needed to reproduce it.

Actual product interface · Model & INP editor

What should a review trace?

  1. 01
    Physical assumptions
    Materials · loads · constraints

    Check material, contact and boundary definitions in the input file against the engineering problem.

  2. 02
    Numerical method
    Element type · discretization · solver settings

    Inspect element and solver settings. Record the method and version used.

  3. 03
    Solving process
    Run configuration · logs · convergence

    Review the run logs and retain the inputs and configuration needed to reproduce the calculation.

  4. 04
    Validation evidence
    References · error · scope

    Compare results under the same conditions. Record criteria, deviations and conditions of use.

INTELLIGENT DESIGN EXPLORATION

Explore more designs within your budget

Combine elastic HPC, AI sampling and high-fidelity review to select candidate designs.

Exploration example
Prioritize promising directions · Exploration exampleAI uses samples and uncertainty to suggest points for engineer confirmation.m / kgt / mmσ
Computed samplesProposed next batchPrediction uncertainty
BUILD ENGINEERING APPS · REUSE CALCULATION WORKFLOWS

Proven methods, ready to reuse

Update design inputs and reuse validated calculation steps and review criteria.

13° wheel-impact analysisApp interface illustration

Response and criteria

Displacement
Peak and time history
Error and scope
Check acceptance rules
Result record
Linked task and version
Next similar task

Update the inputs. Reuse the validated workflow and evaluation rules.

Acceptance rules
Response metrics · Error tolerance
Method version
v1.0

Scope: reviewed materials and impact conditions

App and report examples

Engineering App results

Interface illustration

Response metrics and acceptance rules.

Validation report structure

Report template

Task, results and review records.

TRAIN SPECIALIZED AI · CAPTURE EXPERT JUDGMENT

Expert knowledge for more engineering decisions

Learn from engineer demonstrations, corrections and evaluation so judgment can support more similar tasks. Engineers confirm key decisions.

Wheel local-response review · Training illustration

  1. AI first assessment

    Select a region from contour peaks

  2. Engineer correction

    Add load paths and time responses

  3. Review & evaluation

    Compare reference results and confirm scope

View an AI training example
Example training task

Decide which local responses need review

A
Region of interest
B
Compare time responses
  1. 01

    Engineer demonstration

    Mark regions along the impact load path and record why they matter.

  2. 02

    Correct the judgment

    Initial AI judgmentChoose review locations using only the contour color peak.

    Engineer correctionUse adjacent regions and multiple timesteps to support the judgment.

  3. Review & evaluate

    Compare references and acceptance rules. Confirm task scope and conditions of use.

PRODUCT COLLABORATION

Engineering expertise for the whole team

View product relationships
GEWU

Build Apps and specialized AI with version and validation records.

LUBAN

Combine Engineering Apps into pipelines for repeatable team use.

MOZI

Invoke Apps and specialized AI toward an engineering goal and follow execution.

START WITH A REAL TASK

Start validating your next great design here

Start with your design goal and current bottleneck. Define the first validation path together.

  • The bottleneck worth solving first
  • Initial validation goals and criteria
  • Reusable outcomes to build
Discuss my design taskEnter GEWU