"For example, if you were sintering titanium powder to make an aircraft part, you can detect if the laser has caused the alloy to overheat as it's spent too long in a particular location, as the way the metal will solidify will be different if it was heated to a different temperature. It's always those cooling stresses in any part which generate weak points or points of failure."
General Electric is using sensors and big data technology to detect temperature anomalies while parts are being made. The company is using the "in-process inspection technology" for its microscopic weld pool manufacturing that's moving at hundreds of millimetres per second. Even a slight diversion could result in a massive defect. According to GE, it can turn "terabytes of raw data to megabytes of useful information", increase production speeds by 25 per cent and reduce post-build inspection time by 25 per cent.
Enterprise use of 3D printing means CIOs and their IT organisations will need to be deal with even more data: Both data that can drive improvements to production processes and the data necessary to enable production in the first place. For example, Basiliere says CIOs will have to securely store and make available to multiple users of computer-aided design (CAD) files.
"IT may not be doing the design or physically creating the item, but they do have to provide that back bone support for all the data that is being used, supporting the CAD software," he says.
"Also, enabling the printers to run on the network, and ensure the network is secure from people exporting those proprietary designs or hacking into the company and seeking those designs.
"Then you are going to have to store it, because the theory now is that we are enabling long tail manufacturing. By doing so, it means you need to have the CAD file securely archived somewhere so that when there's a need later on for a one off or short-term production of that item it can be looked up and retrieved and printed."
Bender says data logged during the manufacturing process could potentially be quite large depending on how the types of data collected, sampling frequency and the number of parts made. However, he adds that models for 3D printing are simpler than those used for traditional injection moulding, and therefore require less disk space to store.
"In traditional injection moulding, mass manufacture of complex shapes, everything has to have on the sides of it a taper or a slope that allows the part to fall out of the mould after it has been made. That actually makes your model a lot more mathematically complex and makes for larger data storage requirements.
Sign up for CIO Asia eNewsletters.