Big Data for Assessment and Enhancement of Casting Processes (ACRC Funded)

Big Data

Problem Statement:

Porosity, caused by solidification shrinkage or gas bubbles trapped within the casting during solidification, are inevitable features in metal castings. When a cast part contains porosity beyond allowed levels, the casting is considered to contain defects and it leads to a rejection of the product. Statistics show that in die castings, by which 2/3 of Al castings are produced in North America, 35% of the casting defects are directly caused by porosity, and up to 65% of the defects are porosity related issues. In many cases, porosity features are located internally in casting parts, and are not exposed until the castings are machined. At this stage, the cast part has accumulated a significant amount of operation cost, and producers have to absorb higher rejection cost if the casting is scrapped.

In our modern plants, we capture much data. This includes molten metal preparation details, processing data, simulation data, part geometry data (CAD files), NDE data, defect location data, etc. However, it is a lost opportunity if we do not fuse the data into a portal that gives us the knowledge base about the casting process and a framework for continuous learning. The premise being that without continuous learning one cannot continually be at the leading edge.