Tag Archives: innovation

Digital twins could put at risk what it means to be human

Detail from abstract by Zahrah ReshI have written in the past about my research on the development and use of digital twins.  A digital twin is a functional representation in a virtual world of a real world entity that is continually updated with data from the real world [see ‘Fourth industrial revolution’ on July 4th, 2018 and also a short video at https://www.youtube.com/watch?v=iVS-AuSjpOQ].  I am working with others on developing an integrated digital nuclear environment from which digital twins of individual power stations could be spawned in parallel with the manufacture of their physical counterparts [see ‘Enabling or disruptive technology for nuclear engineering’ on January 1st, 2015 and ‘Digitally-enabled regulatory environment for fusion power-plants’ on March 20th, 2019].  A couple of months ago, I wrote about the difficulty of capturing tacit knowledge in digital twins, which is knowledge that is generally not expressed but is retained in the minds of experts and is often essential to developing and operating complex engineering systems [see ‘Tacit hurdle to digital twins’ on August 26th, 2020].  The concept of tapping into someone’s mind to extract tacit knowledge brings us close to thinking about human digital twins which so far have been restricted to computational models of various parts of human anatomy and physiology.  The idea of a digital twin of someone’s mind raises a myriad of philosophical and ethical issues.  Whilst the purpose of a digital twin of the mind of an operator of a complex system might be to better predict and understand human-machine interactions, the opportunity to use the digital twin to advance techniques of personalisation will likely be too tempting to ignore.  Personalisation is the tailoring of the digital world to respond to our personal needs, for instance using predictive algorithms to recommend what book you should read next or to suggest purchases to you.  At the moment, personalisation is driven by data derived from the tracks you make in the digital world as you surf the internet, watch videos and make purchases.  However, in the future, those predictive algorithms could be based on reading your mind, or at least its digital twin.  We worry about loss of privacy at the moment, by which we probably mean the collation of vast amounts of data about our lives by unaccountable organisations, and it worries us because of the potential for manipulation of our lives without us being aware it is happening.  Our free will is endangered by such manipulation but it might be lost entirely to a digital twin of our mind.  To quote the philosopher Michael Lynch, you would be handing over ‘privileged access to your mental states’ and to some extent you would no longer be a unique being.  We are long way from possessing the technology to realise a digital twin of human mind but the possibility is on the horizon.

Source: Richard Waters, They’re watching you, FT Weekend, 24/25 October 2020.

Image: Extract from abstract by Zahrah Resh.

Scattering electrons reveal dislocations in material structure

Figure 9 from Yang et al, 2012. Map of plastic strain around the crack tip (0, 0) based on the full width of half the maximum of the discrete Fourier transforms of BSE images, together with thermoelastic stress analysis data (white line) and estimates of the plastic zone size based on approaches of Dugdale's (green line) and Irwin's (blue line; dimensions in millimetres).

Figure 9 from Yang et al, 2012. Map of plastic strain around the crack tip (0, 0) based on the full width of half the maximum of the discrete Fourier transforms of BSE images, together with thermoelastic stress analysis data (white line) and estimates of the plastic zone size based on approaches of Dugdale’s (green line) and Irwin’s (blue line; dimensions in millimetres).

It is almost impossible to manufacture metal components that are flawless.  Every flaw or imperfection in a metallic component is a potential site for the initiation of a crack that could lead to the failure of the component [see ‘Alan Arnold Griffith’ on April 26th, 2017].  Hence, engineers are very interested in understanding the mechanisms of crack initiation and propagation so that these processes can be prevented or, at least, inhibited.  It is relatively easy to achieve these outcomes by not applying loads that would supply the energy to drive failure processes; however, the very purpose of a metal component is often to carry load and hence a compromise must be reached.  The deep understanding of crack initiation and propagation, required for an effective and safe compromise, needs detailed measurements of evolution of the crack and of its advancing front or tip [depending whether you are thinking in three- or two-dimensions].  When a metal is subjected to repeated cycles of loading, then a crack can grow incrementally with each load cycle; and in these conditions a small volume of material, just ahead of the crack and into which the crack is about to grow, has an important role in determining the rate of crack growth.  The sharp geometry of the crack tip causes localisation of the applied load in the material ahead of the crack thus raising the stress sufficiently high to cause permanent deformation in the material on the macroscale.  The region of permanent deformation is known as the crack tip plastic zone.  The permanent deformation induces disruptions in the regular packing of the metal atoms or crystal lattice, which are known as dislocations and continued cyclic loading causes the dislocations to move and congregate around the crack tip.  Ultimately, dislocations combine to form voids in the material and then voids coalesce to form the next extension of the crack.  In reality, it is an oversimplification to refer to a crack tip because there is a continuous transition from a definite crack to definitely no crack via a network of loosely connected voids, unconnected voids, aggregated dislocations almost forming a void, to a progressively more dispersed crowd of dislocations and finally virgin or undamaged material.  If you know where to look on a polished metal surface then you could probably see a crack about 1 mm in length and, with aid of an optical microscope, you could probably see the larger voids forming in the material ahead of the crack especially when a load is applied to open the crack.  However, dislocations are very small, of the order tens of nanometres in steel, and hence not visible in an optical microscope because they are smaller than the wavelength of light.  When dislocations congregate in the plastic zone ahead of the crack, they disturb the surface of the metal and causing a change its texture which can be detected in the pattern produced by electrons bouncing off the surface.  At Michigan State University about ten years ago, using backscattered electron (BSE) images produced in a scanning electron microscope (SEM), we demonstrated that the change in texture could be measured and quantified by evaluating the frequency content of the images using a discrete Fourier transform (DFT).  We collected 225 square images arranged in a chessboard pattern covering a 2.8 mm by 2.8 mm square around a 5 mm long crack in a titanium specimen which allowed us to map the plastic zone associated with the crack tip (figure 9 from Yang et al, 2012).  The length of the side of each image was 115 microns and 345 pixels so that we had 3 pixels per micron which was sufficient to resolve the texture changes in the metal surface due to dislocation density.  The images are from our paper published in the Proceedings of the Royal Society and the one below (figure 4 from Yang et al, 2012) shows four BSE images along the top at increasing distances from the crack tip moving from left to right.  The middle row shows the corresponding results from the discrete Fourier transform that illustrate the decreasing frequency content of the images moving from left to right, i.e. with distance from the crack.  The graphs in the bottom row show the profile through the centre of the DFTs.  The grain structure in the metal can be seen in the BSE images and looks like crazy paving on a garden path or patio.  Each grain has a particular and continuous crystal lattice orientation which causes the electrons to scatter differently from it compared to its neighbour.  We have used the technique to verify measurements of the extent of the crack tip plastic zone made using thermoelastic stress analysis (TSA) and then used TSA to study ‘Crack tip plasticity in reactor steels’ [see post on March 13th, 2019].

Figure 4 from Yang et al, 2012. (a) Backscattered electron images at increasing distance from crack from left to right; (b) their corresponding discrete Fourier transforms (DFTs) and (c) a horizontal line profile across the centre of each DFT.

Figure 4 from Yang et al, 2012. (a) Backscattered electron images at increasing distance from crack from left to right; (b) their corresponding discrete Fourier transforms (DFTs) and (c) a horizontal line profile across the centre of each DFT.

Reference: Yang, Y., Crimp, M., Tomlinson, R.A., Patterson, E.A., 2012, Quantitative measurement of plastic strain field at a fatigue crack tip, Proc. R. Soc. A., 468(2144):2399-2415.

Lacking creativity

detail tl from abstract painting by Zahrah RI feel that I am moving to the next level of experience with online meetings but I am unsure that it will address the slow down in productivity and a loss of creativity being reported by most leaders of research groups to whom I have spoken recently.  About a month ago, we organised an ‘Away Day’ for all staff in the School of Engineering with plenary presentations, breakout groups and a Q&A session.  Of course, the restrictions induced by the pandemic meant that we were only ‘away’ in the sense of putting aside our usual work routine and it only lasted for half a day because we felt a whole day in an online conference would be counter productive; nevertheless, the feedback was positive from the slightly more than one hundred staff who participated.  On a smaller scale, we have experimented with randomly allocating members of my research team to breakout sessions during research group meetings in an attempt to give everyone a chance to contribute and to stimulate those serendipitous conversations that lead to breakthroughs, or least alternative solutions to explore.  We have also invited external speakers to join our group meetings – last month we had a talk from a researcher in Canada.  We are trying to recreate the environment in which new ideas bubble to the surface during casual conversations at conferences or visits to laboratories; however, I doubt we are succeeding.  The importance of those conversations to creativity and innovation in science is highlighted by the story of how Emmanuelle Charpentier and Jennifer Doudna met for the first time at a conference in Puerto Rico.   While wandering around San Juan on a warm Caribbean evening in 2011 discussing the way bacteria protect themselves against viruses by chopping up the DNA of the virus, they realised that it could be turned into molecular scissors for cutting and editing the genes of any living creature.  They went home after the conference to their labs in Umea University, Sweden and UC Berkeley respectively and collaborated round the clock to implement their idea for which they won this year’s Nobel Prize for Chemistry.  Maybe the story is apocryphal; however, based on my own experience of conversations on the fringes of scientific meetings, they are more productive than the meeting itself and their loss is a significant casualty of the COVID-19 pandemic.  There are people who point to the reduction in the carbon footprint of science research caused by the cancellation of conferences and who argue that, in order to contribute to UN Goals for Sustainable Development, we should not return to gatherings of researchers in locations around the world.  I agree that we should consider our carbon footprint more carefully when once again we can travel to scientific meetings; however, I think the innovations required to achieve the UN Goals will emerge very slowly, or perhaps not all, if researchers are limited to meeting online only.

Source:

Clive Cookson, A dynamic Nobel duo with natural chemistry, FT Weekend, 10/11 October 2020.

Image: Extract from abstract by Zahrah Resh.

Poleidoscope (=polariscope + kaleidoscope)

A section from a photoelastic model of turbine disc with a single blade viewed in polarised light to reveal the stress distribution.Last month I wrote about the tedium of collecting data 35 years ago without digital instrumentation and how it led me to work on automation and digitalisation in experimental mechanics [see ‘35 years later and still working on a PhD thesis‘ on September 16th, 2020].  Thirty years ago, one of the leading methods for determining stresses in components was photoelasticity, which uses polarised light to generate fringe patterns in transparent components or models that correspond to the distribution of stress.  The photoelastic fringes can be analysed in a polariscope, of which the basic principles are explained in a note at the end of this post.  During my PhD, I took hundreds of black and white photographs in a polariscope using sheets of 4×5 film, which came in boxes of 25 sheets that you can still buy, and then scanned these negatives using a microdensitometer to digitise the position of the fringes.  About 15 years after my PhD, together with my collaborators, I patented the poleidoscope which is a combination of a polariscope and a kaleidoscope [US patents 6441972 & 5978087] that removes all of that tedium.  It uses the concept of the multi-faceted lens in a child’s kaleidoscope to create several polariscopes within a compound lens attached to a digital camera.  Each polariscope has different polarising elements such that photoelastic fringes are phase-shifted between the set of images generated by the multi-faceted lens.  The phase-shifted fringe patterns can be digitally processed to yield maps of stress much faster and more reliably than any other method.  Photoelastic stress analysis is no longer popular in mainstream engineering or experimental mechanics due to the simplicity and power of digital image correlation [see ‘256 shades of grey‘ on January 22nd, 2014]; however, the poleidoscope has found a market as an inspection device that provides real-time information on residual stresses in glass sheets and silicon wafers during their production.  In 2003, I took study leave for the summer to work with Jon Lesniak at Glass Photonics in Madison, Wisconsin on the commercialisation of the poleidoscope.  Subsequently, Glass Photonics have  sold more than 250 instruments worldwide.

For more information on the poleidoscope see: Lesniak JR, Zhang SJ & Patterson EA, The design and evaluation of the poleidoscope: a novel digital polariscope, Experimental Mechanics, 44(2):128-135, 2004

Note on the Basic principles of photoelasticity: At any point in a loaded component there is a stress acting in every direction. The directions in which the stresses have the maximum and minimum values for the point are known as principal directions. The corresponding stresses are known as maximum and minimum principal stresses. When polarised light enters a loaded transparent component, it is split into two beams. Both beams travel along the same path, but each vibrates along a principal direction and travels at a speed proportional to the associated principal stress. Consequently, the light emerges as two beams vibrating out of phase with one another which when combined produce an interference pattern.   The polarised light is produced by the polariser in the polariscope and the analyser performs the combination. The interference pattern is observed in the polariscope, and the fringes are contours of principal stress difference which are known as isochromatics. When plane polarised light is used black fringes known as isoclinics are superimposed on the isochromatic pattern. Isoclinics indicate points at which the principal directions are aligned to the polarising axes of the polariser and analyser.

Image: a section from a photoelastic model of turbine disc with a single blade viewed in polarised light to reveal the stress distribution.