Experimental Validation

Since the founding of Pan Computing in 2012, over $7.3M USD has been spent validating the Netfabb Simulation solver predictions by comparing with experimental results for hundreds of design cases/geometries. Below are a few examples meant to highlight the experimental validation methodology.

Powder Bed Fusion


Recoater jam prediction


Description: A component manufactured at the Penn State CIMP3D Additive Manufacturing center experienced a collision with the recoater blade. Netfabb Simulation was used to attempt to identify the cause of failure.


Figure 1.1: Failed experimental build (left) and simulation result at the corresponding build height (right).*

Conclusions: Netfabb Simulation is able to accurately predict the build height at which the recoater jam occurs. The simulation produces a warning for the user at the corresponding layer, warning of a likely collision with the recoater. The cause of the jam is excessive thermal distortion on a shallow overhang feature.

*Figure courtesy of Dr. Tim Simpson (Penn State University)

Distortion compensation


Description: A test geometry manufactured by United Technologies Research Center (UTRC) using Inconel 718 experienced distortion during the manufacturing process. Netfabb Simulation was used to attempt to compensate the geometry prior to manufacturing in order to reduce the net distortion.


Figure 2.1: Original preform (top left), experimental distortion measurement of the manufactured preform showing large distortion (top right), simulated distortion results in close agreement with measured distortion (bottom left), compensated geometry (bottom center), and the experimental distortion results of the manufactured compensated geometry showing reduced distortion.*

Conclusions: Netfabb Simulation can accurately predict both distortion magnitude and trends without requiring any experimental calibration. Netfabb Simulation can also use the calculated distortion to produce a compensated geometry that when manufactured, results in nearly no net distortion.

*Experimental data courtesy of America Makes, GE GRC, and United Technologies Research Center

Part/Powder Interaction


Description: Two simulations are run to investigate the effect of including loose powder in an analysis. The first simulation does not model the loose powder and instead approximates energy lost by conduction into the powder as an artificial heat loss prescribed to the external surfaces of the geometry. The second simulation explicitly models the loose powder and the interaction between the parts and the powder through conduction.


Figure 3.1: Experimental setup of the geometries. The line used to compare the model to the experiment is highlighted in yellow.


Figure 3.2: Thermal simulation results for the simulation case explicitly modeling the loose powder.


Figure 3.3: Simulated distortion results for the two cases compared with the experimental measurements.*

Conclusions: Explicitly modeling the loose powder and the interaction between parts results in much closer agreement with experiment than the case assuming an artificial heat loss on the free surface of the geometries. In order to achieve accurate simulation results for cases such as this, it is necessary to model the full build plate including the loose powder.

*Experimental data courtesy of America Makes, GE GRC, and United Technologies Research Center

Distortion Prediction


Description: A thin wall test geometry manufactured by GE Global Research Center (GEGRC) using Inconel 625 experienced distortion during the manufacturing process. The experimental scan results were used to validated the predictions from Netfabb Simulation. Two simulation are run, one accounting for loose powder and one neglecting the loose powder.


Figure 4.1: Experimental setup (left), simulated distortion with comparison plane show in red (middle), and a comparison of experimental and simulated results (right).*

Conclusions: Netfabb Simulation is capable of predicting the distortion trend and magnitude for thin wall structures experiencing high levels of distortion. Including loose powder in the simulation improves the results.

*Experimental data courtesy of America Makes, GE GRC, and United Technologies Research Center

Support Failure Prediction


Description: A component manufactured at the Penn State CIMP3D Additive Manufacturing center experienced a failure of the support structure. The experimental result was used to validate the support failure capability of Netfabb Simulation.



Figure 5.1: Experimental build showing support failure (left) and simulation results (distortion magnified 5x) showing support failure.*

Conclusions: Netfabb Simulation is able to correctly indicate the region of support failure and simulate the delamination of the part from the support.

*Figure courtesy of Dr. Ed Demeter (Penn State University)

Moving-source Simulation*


Description: An experimental setup was designed to allow for the measurement of in situ temperature and distortion during the Powder Bed Fusion deposition process. The measurement results were used to validate the moving-source model predictions from Netfabb Simulation.


Figure 6.1: Experimental setup used to capture the in situ temperature and distortion. A hollow aluminum box referred to as ‘the Vault’  (top left) is instrumented with a displacement sensor (DVRT) and thermocouples (bottom left). The Vault fits into the build chamber of an EOS or Renishaw printer (top right). The sensors attach to a small, 1 mm thick build plate which distorts during the build (bottom right).


Figure 6.2: In situ distortion measurement trend. When the heat source turns on, the top of the built plate thermally expands, resulting in an upward distortion. When the molten material cools and contracts, the top of the build plate is pulled into compression, resulting in a downward distortion.



Figure 6.3: Calculated temperature and distortion compared with in situ measurements.

Conclusions: Netfabb Simulation is able to accurately calculate temperature and distortion measured during the Powder Bed Fusion process. The final calculated distortion is within 5% error of the measurement.

*Results shown in this section have been peer reviewed an published: Denlinger, Erik R., et al. “Thermomechanical model development and in situ experimental validation of the Laser Powder-Bed Fusion process.” Additive Manufacturing 16 (2017): 73-80.