Lead Institution:  University of Cambridge
Industry Partners: Rolls-Royce and Sintratec AG
Project Team: Phillip Stanley-Marbell (Cambridge), Robert Hewson, (Imperial College London), Daniela Petrelli (Sheffield Hallam University) and Nick Dulake (Sheffield Hallam University)
Project Duration: 8 months (April 2018 – December 2018)

Research Challenge
Microstructure and composition variations in the materials of manufactured objects dictate their aesthetic and functional properties. Information about millimetre-scale variations in composition and in the rheology of precursor materials during manufacturing a specific part, would provide a unique fingerprint for each instance of a product. This information would enable more accurate estimation of product quality, would enable dynamic adaptation of the manufacturing process to properties detected during the manufacturing process, and more. These new capabilities would in turn enable lower waste and higher quality products, boosting the potential profitability of manufacturing across a broad swathe of manufacturing sectors of value to the UK economy. Recent work has begun to investigate developing machine learning algorithms
to use information on powder bed temperature variations to predict the structural properties of manufactured parts. Despite the potential value of millimetre-level per-part-layer microstructure and chemical composition data, today, no cost-effective and pervasive methods exist for low-cost implementation of characterisation in the manufacturing process, beyond, say, temperature measurements.

Traditional manufacturing processes such as casting and machining of metals, or injection moulding of polymers, make it impossible to know the specific chemical and structural properties within the volume of a manufactured part. Additive manufacturing techniques such as fused deposition modeling (FDM), selective laser annealing (SLA), and selective laser sintering (SLS) create parts one layer at a time, and therefore in principle have the ability to also analyze the structural and chemical composition of a part, layer by layer, as the part is created. This capability is today unexploited. This is a missed opportunity.

Our insight is that we can capitalise on recent advances in sensors and sensor signal processing, to enable a combination of microstructure analysis and chemical composition analysis in real time during the manufacturing process of functional parts created by additive manufacturing. Our research hypothesis is that we can capitalise on recent advances in both infrared laser based proximity sensors as well as spectophtometric systems based on micromirror arrays, to create per-layer structure and composition sensors integrated into an additive manufacturing process. We will test this hypothesis by creating and evaluating a structure-and-composition measurement apparatus, and integrating it into a nylon SLS 3D printer. We will use this apparatus to evaluate the property variations of manufactured components of designs in which knowledge of the millimetre-scale composition variations will be valuable. These designs
will be selected in close collaboration with our industrial partners. In addition to paving the way to the fundamentally new capability of per-part continuous microstructure and composition metrology, the results of this research will enable our vision for integrating computation and sensing additive materials (CSAMs) as a new kind of materials additive applicable to a broad range of manufacturing processes valuable to UK plc.