Design Optimization of Turbomachinery Rotor Blades using Neural Network Surrogate Models
Description
A sample is made of a design space using Latin hypercube sampling. The geometry for these samples is then generated and evaluated using simulation tools. The result is then used to train a neural network for use as a surrogate model in design optimization.
Detailed example
Current models provide predictions for the margin of safety associated with a rotor blade geometry, enabling faster design optimization than using FEA alone.
AI / analytics pattern
Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relationships.
Automation level / stage
b) Pilot – The use case has been deployed in a limited test or pilot capacity.
Expected benefit
Early work shows R^2 > 0.99 (compared to FEA results) on both test and validation datasets when predicting structural performance of a rotor blade as a function of 6 design variables
Controls / human review
ATO: Yes; PIA: Not published
Data needed
Data consists of Finite Element Analysis (FEA) simulations of rotor blades, generated with ANSYS Mechanical. Solutions were split into separate training, validation, and test datasets.