Machine Learning for Turbomachinery

We create Machine Learning systems using small but high-fidelity datasets

machine-learning-for-turbomachinery v1

 

geometry-parameterization

Geometry Parameterization

The complete, but efficient, description of the 3D blade shape in as few parameters as possible. This means doing away with the traditional ‘blade angle distribution’ paradigm and instead using blade loading and a 3D Inverse Design approach.

high-fidelity-performance-data

High Fidelity Performance Data

TURBOdesign1 provides a seamless link to high-fidelity CAE simulation with a number of leading commercial systems. Meshing, pre-processing, solving and post-processing are all automatically handled within the TURBOdesign1 environment. environment.

Reactive-Response-Surface-with-Machine-Learning1

Computational Efficiency

ADT’s optimizer engine - RRS+CAE - dramatically reduces the number of high-fidelity simulation runs required to accurately explore the turbomachinery design space. Reducing cost and time by orders of magnitude, and making full 3D, multi-point, multi-objective blade shape optimization a reality on standard desktop hardware.
 

ML-expert-systems

Machine Learning

Building on the core pillars of powerful geometry parameterization, integrated high fidelity performance data, and computational efficiency, ADT is bringing unprecedented Machine Learning methods to Turbomachinery Blade Design.

machine-learning-case-study-centrifugal-pump

Machine Learning for Centrifugal Pump Design

Our blog demonstrates how machine learning and 3D inverse design create optimized pumps in just hours, leading to massive efficiency gains.

Discover how it works
machine-learning-case-study-axial-fans

Machine Learning for Axial Fan Design

Learn how ADT's 3D Inverse Design and Reactive Response Surface combine to create an efficient machine learning tool for optimizing axial fan blades. efficiency and reduced noise with a multi-point process.

 

Get the full story

.


Ready To Accelerate your Engineering Process?

.