Past Event! Note: this event has already taken place.
3D Printing, Engineering Design & Data Sciences
November 21, 2016 at 1:30 PM to 3:30 PM
| Location: | 5345 Herzberg Laboratories |
| Audience: | Anyone |
ABSTRACT
Computer controlled machining was developed in 1947-1957 because machining turbine blades and aeroplane wing spars became too complex to be done by people. Computer aided drafting that has mutated into Computer Aided Design (CAD) was developed in the 1950s because the time to produce a military aeroplane was limited by the 10 to 20 years required by 100s of people to draw the 100,000 or more blue prints required. The technology to manufacture VLSI solid state devices such as modern computers and cell phones was developed by the USAF in the 1950s because vacuum tubes were so unreliable that at any given instant more than half of their aeroplanes were grounded because of defective vacuum tubes. Computational mechanics was developed in the 1950s because people could no longer calculate manually the stress in aeroplanes designed to fly near the speed of sound. 3D printing was developed in the late 1970s to make physical models from CAD designs because people were too slow, expensive or not able to make the required physical models. These innovations have brought technology to the stage that it now has the potential to integrate design, manufacturing and in-service behavior into holistic computer models that rapidly optimize every known aspect of the design, manufacturing and operational processes as a virtual process, i.e., before doing anything real.
To use all of these innovations involving computer models in the real world requires verification, validation and uncertainty quantification. In the real world, the required experimental data – sta- tistical evidence is far more important than mathematical proofs. Because of the time and cost of experiments and proprietary data, experimental evidence in the industrial world has been remark- ably rare and uncertainty remarkably high. Turbulent fluid flow and fatigue cracks in aeroplanes are examples of physical phenomena where after more than 100 years of research by the world’s bright- est minds, there are no fundamental theories and none are expected that are not direct functions of parameters measured by experiment, i.e., the theories are essentially simple regression equations.
I will argue that Data Sciences now provides an opportunity to develop the next generation of design technology by correlating experimental data constrained by the laws of physics. The experi- mental data will be acquired using sensor networks that provide multi-scale, multi-physics Big Data sets similar to satellite data acquisition systems. These designs could be called functional designs that could be viewed as optimized functions of a design process that marry design requirements with physics-based models and deep learning with experimental data; i.e., they map the required functionality onto the forms to be manufactured and operated. This will require specialists in Data Sciences, Computer Science and the Design Domains such as Engineering to work together in interdisciplinary teams. Success will depend on people collaborating and cooperating within and between institutions and communities.
ABOUT THE SPEAKER
John Goldak is best known for his research in the computational mechanics of welds. In particular for the development of a heat source model for arc welds that is widely known as the Double Ellipsoid Weld Pool Model. In 2011, The Welding Science and Engineering Conference created the `The Pioneers of Computational Weld Mechanics’ award to honor the contributions of Professor Yukio Ueda, Japan and John Goldak to the development of computational weld mechanics. He is Fellow of the American Welding Society and of the Canadian Welding Association. He is a member of the Canadian Academy of Engineering. He is Founder and President of Goldak Technologies Inc. (GTI), (http://goldak-vrweld.com/ about_vrweld.html), a company dedicated to developing software for design- driven analysis and the optimization of welds and welded structures. GTI was awarded the John S. Hewitt Team Achievement Award by the Canadian Nuclear Society in 2011 as a major player for its computational weld mechanics analysis that contributed to the successful repair of AECL’s NRU reactor in 2009. GTI customers include leading multinationals in the agricultural machinery industry, the Canadian Navy on computational mechanics for weld repair of submarines and leading organizations in the Canadian and USA nuclear power industry.
Contact information: jgoldak@mcro2.Carleton.ca
This seminar is free and open to all. Complimentary coffee, tea and light snacks will be provided beginning at 1:15 p.m. We hope you can join us!
Campus map: http://carleton.ca/campus/map/
Please note that photos or video may be taken at the event which may later be used in print and online media produced by the Institute for Data Science at Carleton University. For any questions or concerns, please contact Jena Lynde-Smith (jena.lyndesmith@carleton.ca).