Sandipan Mishra, a controls systems expert and assistant professor in the Department of Mechanical, Aerospace, and Nuclear Engineering at Rensselaer Polytechnic Institute, has an important goal – to help AM ‘learn’, so to speak, by developing advanced sensing and control algorithms that can improve additive manufacturing technologies, including 3D printing.
Mishra’s research, titled “Multi-objective Learning Control Strategies for Additive Manufacturing,” involves creating smarter control systems that will use sensor measurements to help 3D printers learn and adapt as they are operating. Many AM technologies work by applying or printing thin layers of materials on top of one another – literally constructing the object from the ground up, one layer at a time. Mishra is seeking to create and design a feedback system that will enable an additive manufacturing system to make small, iterative refinements in the midst of a printing job. In theory, the system will be able to continually assess the progress of a print job, and then automatically make necessary adjustments to ensure the finished good will have specific pre-determined properties or geometries.
“Despite its tremendous potential, additive manufacturing is hampered by poor process reliability and throughput,” Professor Mishra explained to Metal Powder Report. “The systems simply are not precise or robust enough to be scaled up and used for commercial, mass-manufactured products.”
Currently, most AM processes operate without any sensor measurement for correction during the process – known as “open-loop” control. As Mishra explains: “We are developing sensors that can detect the ‘state’ of the part in real-time, then feed this information back into the system to correct for errors, such as voids, or compensate for gradual drifts in parameters, and variable operating conditions.”
The ‘feedback control law’ uses process models in conjunction with sensor readings. Since most AM processes are layer-based naturally, it is possible use information from previous layers to correct mistakes by adjustments to the current layer/future layers to be laid down.
Predicting properties
What parameters will the research focus on? Geometry measurements such as layer-height are naturally important, according to Professor Mishra. “However, in addition, depending on the process, temperature measurements can provide key insights into the state of the process and can further predict mechanical properties of the final metal part – such as residual stress, strength and microstructure,” he explained. “IR cameras can be used for these temperature measurements as well. We are also investigating high-speed (real-time) measurements of droplet size in jet-based printing, using intelligent sensing techniques that do not require cameras (which are typically slow).”
In metal AM processes, residual stresses and porosity are key issues. These are caused by poor thermal gradients. “One approach to reduce this will be through path planning,” Professor Mishra explained. “We are developing optimisation algorithms that can reduce thermal gradients by intelligent choice of paths. In jet-based polymer applications, geometry is a key issue. This is often because process variations cause changing jet-droplet sizes. By using real-time sensors and feedback control, we will aim to reduce droplet size variability. By using layer-to-layer control, we will correct mistakes made in earlier layers.”
What are the main issues affecting AM at the moment? “Achieving profitable production-scale throughput and process/part reliability is the main barrier to success,” Mishra added. “Parts need to be made fast and within tolerances (for both geometric as well as material properties).”
How does the choice of materials affect this? “Materials play a key role in the choice of process parameters, as well as the feasibility of using a specific process for a given material. Materials need to be developed that are suitable for the typical types of AM processes that are currently available. Conversely, new processes need to be developed that are suitable for the materials, too.”
The future
AM has good potential to be commercially viable, Professor Mishra says. “Different AM processes have potential in different sectors. Metal AM processes have tremendous potential in industrial applications such as the aerospace industry and the automotive industry, while polymer/liquid-based processes can help in flexible electronics, consumer products such as smart-phones and LED technologies.”
Overall, while admitting it has process issues, Professor Mishra is very optimistic about the future of AM. “The AM industry is set to explode over the next few years, starting at the hobby ‘consumer’ level all the way through to large-scale industrial consumers such as GE, automotive and aerospace.”
Alongside this growth, however, new regulations will have to be put in place. “As these processes become more widespread, in addition to process/equipment development, a very critical issue is the establishment of standards – something which has yet to be explored,” Mishra explained. “The nature of manufacturing will change dramatically, with new supply chains needing to be put in place. As in the semiconductor manufacturing industry, I believe that there will be a separation of ‘design houses’ and ‘fab houses’. There will be large ‘fab houses’ being established, which will have a large array of expensive state-of-the-art AM machines that ‘design houses’ will utilise for manufacture and delivery.”
Among other prospects, Professor Mishra highlighted organ printing applications as a very exciting future technology. (Disclosure: He is currently working on the design and automation of a synthetic organ printer.)
While Mishra believes traditional manufacturing will still retain hold over large-scale part manufacture, in the case of mass-customised expensive parts, AM may overtake traditional manufacturing alternatives. “Custom jewellery is already being made in Europe,” he observed. “The construction industry has been an AM process for a long time, and I believe there are some very interesting new opportunities for AM in large-scale construction.”
Biography
Professor Mishra joined the Rensselaer faculty in 2010, after serving for two years as a postdoctoral researcher at the University of Illinois at Urbana-Champaign. He earned his bachelor’s degree in mechanical engineering from the Indian Institute of Technology Madras, and his doctoral degree in mechanical engineering from the University of California, Berkeley.
Among his achievements, Professor Mishra won a prestigious five-year, US$400,000 Faculty Early Career Development Award (CAREER) from the National Science Foundation (NSF) to continue the research. The CAREER Award is given to faculty members at the beginning of their academic careers and is one of NSF’s most competitive awards, placing emphasis on high-quality research and novel education initiatives.
In addition to the CAREER Award, Professor Mishra receives research funding from the NSF Civil, Mechanical and Manufacturing Innovation Division to investigate ways of using image sensors for high-speed adaptive optics systems. He is leading an NSF Sustainable Energy Pathways research project at Rensselaer, in collaboration with the University of Wisconsin, to support research into advanced controls for energy-efficient heating, ventilation, and air-conditioning systems.
For additional information on Mishra’s research at Rensselaer, go here.
