Netfabb Simulation Analysis Time Benchmark

BLOG

By Michael Gouge

Introduction

Previous Blog Posts have shown the usefulness of Mesh Adaptivity and the effect of Mesh Settings on simulation within the context of Mesh Convergence. Here we will extend that discussion to show the Simulation times that should be expected for a range of additive geometries using Netfabb Simulation. These parts include a simple bracket, a large impeller, and a generatively designed automotive bracket, each shown below:

Figure 1 Simple Bracket

Figure 2 Impeller

Figure 3 Automotive Bracket

The STL files for these geometries are available to download here in the Bench Mark Model Files.zip so that this study can be replicated by our users.

Simulation Settings

All 3 geometries are simulated using the following simulation settings:

  • PRM File:Generic Inconel 625 settings: 250 W 800 mm/s 40 micron layer thickness
  • Build Plate: Inconel 625, 25 mm thick
  • Boundary Conditions:
    • No build plate heating
    • Estimated Heat Loss Thermal Boundary conditions, Mechanically fixed build plate
  • Plasticity: Off
  • Heat Treatment: None
  • Automotive Upright Support Structure Volume Fraction: 0.01

Two sets of mesh settings will be shown for each model, one coarse, one fine. The coarse setting is the coarsest mesh which shows convergence while the fine settings are set to better capture the curvature of the test parts.

Mesh settings

The mesh settings and resulting mesh size for the 3 example geometries at both mesh settings are given below:

ModelMesh DetailLayers Per ElementCoarsening GenerationsFinal Node CountFinal Element Count
Simple BracketCoarse2514416523134
ImpellerCoarse16117914190208
Automotive BracketCoarse132698385380187
Simple BracketFine1817586141422
ImpellerFine81681877329974
Automotive BracketFine921637118951987

The Resulting meshes can be seen below:

Figure 4 Simple Bracket Coarse MeshFigure 5 Simple Bracket Fine Mesh
Figure 6 Impeller Coarse MeshFigure 7 Impeller Fine Mesh
Figure 8 Automotive Bracket Coarse MeshFigure 9 Automotive Fine Mesh

Simulation Results

The magnitude of displacement results for the 3 examples are given in the following figures. The result increment is after manufacture and the part has cooled down to room temperature. The mesh edges are turned off for the Automotive upright so as not to cover the simulation results.

Figure 10 Simple Bracket Coarse Displacement Results

Figure 11 Simple Bracket Fine Displacement Results

Figure 12 Impeller Coarse Displacement Results

Figure 13 Impeller Fine Displacement Results

Figure 14 Automotive Bracket Coarse Displacement Results

Figure 15 Automotive Bracket Fine Displacement Results

In the table below simulation times are given, along with an estimate of the time it would take to additively manufacture each geometry, and a ratio of time to simulate vs the estimated time to build each part. Build time estimates are calculated by the solver from the volume of the part and the processing conditions. These estimates do not include any lead in time prior to the start of manufacturing or post processing cooldown time. All simulations were completed on a 14 core Intel Xeon 2.2GhZ processor.

Computational Times

ModelMesh DetailPeak RAM UsageSimulation TimeEstimated Build TimePercentage of Simulation to Build Time
Simple BracketCoarse0.98 GB2 min 13 s9 hours 43 min<1%
ImpellerCoarse3.76 GB8 min15 hours 12 min<1%
Automotive BracketCoarse10.7 GB38 min 34 s1 day 5 hours2%
Simple BracketFine1.86 GB5 min 13 s9 hours 43 min<1%
ImpellerFine16.2 GB47 min 50 s15 hours 12 min5%
Automotive BracketFine42.7 GB3 hours 1 min1 day 5 hours10%


Conclusions

This exercise has shown Netfabb Simulation users the expected run times for 3 benchmark geometries using 2 mesh settings each. This should assist current and potential users in creating an expectation of simulation speeds on their machines. It has been shown that even for a large complex geometry using very fine mesh settings 10 different simulations can be completed in the same time as it would take to print the actual part. This computational speed allows users to attempt various orientation, support, and even geometrical choices before committing to printing finalized parts.

Ready to MAKE?

Related Articles

Add comment

No comments yet

    Ready to MAKE?

    © 2021 Autodesk Inc. All rights reserved