We are pulling together COVID-19 related resources, patents and funding.

Innovation

In a manufacturing setting, the traditional approach to design optimization of a new product involves a lot of experimental testing, evaluating prototypes, and going through multiple design iterations.

As the volume and complexity of data increases, it is extremely challenging for engineers to make sense of all the multi-dimensional information and make sound decisions in a timely manner. This uncertainty increases the number of costly experimental test campaigns, lengthens development timescales, and raises the cost of development.

In an effort to combat these limitations, industry increasingly relies on high-fidelity computer models as virtual representations of real-world devices. High-fidelity modeling represents an improvement over costly physical development and testing, but, importantly, it remains time-consuming.

Argonne National Laboratory is augmenting high-fidelity modeling with machine learning to dramatically accelerate the design optimization process while maintaining the reliability of the data.


Technology Advancement

A job that might take hours using high-fidelity modeling alone takes milliseconds when the modeling is augmented by machine learning.

Impact

Argonne recently worked with a global petroleum and natural gas company to optimize a diesel engine to run on a gasoline-like fuel. Using high-fidelity modeling, the company’s development time took months. By tapping into Argonne’s supercomputing resources and machine learning expertise, the company was able to reduce the development time to days, with the same quality of result.

Files

Machine learning techniques can help companies reduce design time from months to days and slash product development costs.

Machine Learning Helping Optimize Engineering Design and Process

Argonne National Laboratory |
Department of Energy DOE Vehicle Technology Office (Washington DC)
|
Parallel Works Inc. (Chicago IL)
|
Convergent Science Inc. (Madison WI)
Publication Date
Sep 1, 2019
Agreement Type