Additive Manufacturing could be transformational – delivering lighter, stronger parts and novel product capabilities. But it is a major challenge to design AM materials and repeatable AM processes, with high reliance on costly, time-consuming experiment and prototyping. A data-driven approach with machine learning can help you to understand how material and process parameters drive AM part performance and to focus your experimental resources effectively.
The Alchemite machine learning software builds models from real-world sparse, noisy, AM data, where other machine learning methods fail. You can gap-fill and get much more from your data. Understand critical property/process relationships. Decide which changes in material or processing will give the best results, for example, to minimise defects. And design more focused testing programs to help deliver projects faster, with up to 90% fewer experiments.
Project MEDAL - AMRC, Boeing, GE Additive, and Constellium
In collaboration with the Advanced Manufacturing Research Centre (AMRC) and Boeing, Intellegens applied the Alchemite™ deep learning software to design new AM parameter sets for laser powder bed fusion (LPBF) and test them for nickel base alloys across two experimental cycles. Results from the second cycle showed good agreement with the predictions. The model was then used to develop process parameters for the new additive-specific Aheadd® CP1 powder from Constellium in combination with the 400W M2 machine from GE Additive. The project team was able to move from the new powder to final parameters in just two builds, while applying no expert statistical knowledge, dramatically speeding up AM process parameter optimisation.
Process optimisation with Lawrence Livermore National Laboratory
At an Intellegens webinar, Gabe Guss from Lawrence Livermore National Laboratory showed how Alchemite™ machine learning can be applied to predict and optimise print parameters for additively-manufactured parts. Alchemite™ gave results that compared well with alternative analytical approaches, with the added benefits of ease-of-use and the potential to rapidly model new scenarios with many parameters.
Designing a material for direct laser deposition
The process to ‘print’ materials is often poorly understood and subject to significant variation based on the exact processing parameters and conditions. In our example, limited data was available on a direct laser deposition method, consisting of only ten sets of processing variables. This dataset was not enough for traditional machine learning techniques to be able to predict the properties of a wider family of processing variables. We optimized the process using historical welding data and the available sparse direct laser deposition dataset, potentially saving years of research.
News and partnerships
Feature Article - Metal AM Magazine
Intellegens featured in Metal AM Magazine, discussing applications of machine learning to additive manufacturing. The article explores some of the challenges of applying machine learning in AM projects, how these are being overcome, and the resulting benefits.
Intellegens AM solution wins awards
The Intellegens additive manufacturing solution was a winner in the Innovation Awards at the 2021 AM Tech Forum event, organised by the American Society of Mechanical Engineers (ASME). Intellegens scooped two of the awards – Best in Class Software, and the Startup Award. You can now view a recording of the award-winning presentation.
Ansys - Integration with materials data management
Intellegens partners with engineering simulation leader Ansys to integrate Alchemite™ technology into Granta MI, the leading materials data management system for AM applications. For Granta MI user organisations, this provides an additional way to access Alchemite™ capabilities from within their existing workfows.
Alchemite™ for AM Package
This package provides the Alchemite™ Analytics browser-based software and the option of the Alchemite™ Engine API for integration of the Alchemite™ method with in-house workflows and tools. Alchemite™ enables you to understand how factors such as material composition, part geometry, orientation relative to the AM machine, atomization and laser parameters, and machine output signals impact properties such as porosity, defect area, toughness, elongation, hardness, and cost. You can:
- Auto-generate and refine models that identify key property / process relationships
- Design optimal powder properties and machine parameters to achieve target outcomes
- Monitor production data to enable failure analysis and maintain quality control
- Gap-fill and validate data to improve data quality
- Apply adaptive Design of Experiments (DoE) to focus testing, saving time and expense