-
Revealing the effects of laser beam sha** on melt pool behaviour in conduction-mode laser melting
Authors:
Amin Ebrahimi,
Mohammad Sattari,
Aravind Babu,
Arjun Sood,
Gert-willem Römer,
Marcel Hermans
Abstract:
Laser beam sha** offers remarkable possibilities to control and optimise process stability and tailor material properties and structure in laser-based welding and additive manufacturing. However, little is known about the influence of laser beam sha** on the complex melt-pool behaviour, solidified melt-track bead profile and microstructural grain morphology in laser material processing. A simu…
▽ More
Laser beam sha** offers remarkable possibilities to control and optimise process stability and tailor material properties and structure in laser-based welding and additive manufacturing. However, little is known about the influence of laser beam sha** on the complex melt-pool behaviour, solidified melt-track bead profile and microstructural grain morphology in laser material processing. A simulation-based approach is utilised in the present work to study the effects of laser beam intensity profile and angle of incidence on the melt-pool behaviour in conduction-mode laser melting of stainless steel 316L plates. The present high-fidelity physics-based computational model accounts for crucial physical phenomena in laser material processing such as complex laser-matter interaction, solidification and melting, heat and fluid flow dynamics, and free-surface oscillations. Experiments were carried out using different laser beam shapes and the validity of the numerical predictions is demonstrated. The results indicate that for identical processing parameters, resha** the laser beam leads to notable changes in the thermal and fluid flow fields in the melt pool, affecting the melt-track bead profile and solidification microstructure. The columnar-to-equiaxed transition is discussed for different laser-intensity profiles.
△ Less
Submitted 12 November, 2023;
originally announced November 2023.
-
Microfluidic quantum sensing platform for lab-on-a-chip applications
Authors:
Robin D. Allert,
Fleming Bruckmaier,
Nick R. Neuling,
Fabian A. Freire-Moschovitis,
Kristina S. Liu,
Claudia Schrepel,
Philip Schätzle,
Peter Knittel,
Martin Hermans,
Dominik B. Bucher
Abstract:
Lab-on-a-chip (LOC) applications have emerged as invaluable physical and life sciences tools. The advantages stem from advanced system miniaturization, thus, requiring far less sample volume while allowing for complex functionality, increased reproducibility, and high throughput. However, LOC applications necessitate extensive sensor miniaturization to leverage these inherent advantages fully. Ato…
▽ More
Lab-on-a-chip (LOC) applications have emerged as invaluable physical and life sciences tools. The advantages stem from advanced system miniaturization, thus, requiring far less sample volume while allowing for complex functionality, increased reproducibility, and high throughput. However, LOC applications necessitate extensive sensor miniaturization to leverage these inherent advantages fully. Atom-sized quantum sensors are highly promising to bridge this gap and have enabled measurements of temperature, electric and magnetic fields on the nano- to microscale. Nevertheless, the technical complexity of both disciplines has so far impeded an uncompromising combination of LOC systems and quantum sensors. Here, we present a fully integrated microfluidic platform for solid-state spin quantum sensors, such as the nitrogen-vacancy (NV) center in diamond. Our platform fulfills all technical requirements, such as fast spin manipulation, enabling full quantum sensing capabilities, biocompatibility, and easy adaptability to arbitrary channel and chip geometries. To illustrate the vast potential of quantum sensors in LOC systems, we demonstrate various NV center-based sensing modalities for chemical analysis in our microfluidic platform, ranging from paramagnetic ion detection to high-resolution microscale NV-NMR. Consequently, our work opens the door for novel chemical analysis capabilities within LOC devices with applications in electrochemistry, high throughput reaction screening, bioanalytics, organ-on-a-chip, or single-cell studies.
△ Less
Submitted 1 November, 2022; v1 submitted 4 September, 2022;
originally announced September 2022.
-
The Effect of Groove Shape on Molten Metal Flow Behaviour in Gas Metal Arc Welding
Authors:
Amin Ebrahimi,
Aravind Babu,
Chris R. Kleijn,
Marcel J. M. Hermans,
Ian M. Richardson
Abstract:
One of the challenges for development, qualification and optimisation of arc welding processes lies in characterising the complex melt-pool behaviour which exhibits highly non-linear responses to variations of process parameters. The present work presents a simulation-based approach to describe the melt-pool behaviour in root-pass gas metal arc welding (GMAW). Three-dimensional numerical simulatio…
▽ More
One of the challenges for development, qualification and optimisation of arc welding processes lies in characterising the complex melt-pool behaviour which exhibits highly non-linear responses to variations of process parameters. The present work presents a simulation-based approach to describe the melt-pool behaviour in root-pass gas metal arc welding (GMAW). Three-dimensional numerical simulations have been performed using an enhanced physics-based computational model to unravel the effect of groove shape on complex unsteady heat and fluid flow in GMAW. The influence of surface deformations on power-density distribution and the forces applied to the molten material was taken into account. Utilising this model, the complex heat and fluid flow in melt pools were visualised and described for different groove shapes. Additionally, experiments were performed to validate the numerical predictions and the robustness of the present computational model is demonstrated. The model can be used to explore physical effects of governing fluid flow and melt-pool stability during gas metal arc root welding.
△ Less
Submitted 13 December, 2021;
originally announced January 2022.
-
The influence of laser characteristics on internal flow behaviour in laser melting of metallic substrates
Authors:
Amin Ebrahimi,
Mohammad Sattari,
Scholte J. L. Bremer,
Martin Luckabauer,
Gert-willem R. B. E. Römer,
Ian M. Richardson,
Chris R. Kleijn,
Marcel J. M. Hermans
Abstract:
The absorptivity of a material is a major uncertainty in numerical simulations of laser welding and additive manufacturing, and its value is often calibrated through trial-and-error exercises. This adversely affects the capability of numerical simulations when predicting the process behaviour and can eventually hinder the exploitation of fully digitised manufacturing processes, which is a goal of…
▽ More
The absorptivity of a material is a major uncertainty in numerical simulations of laser welding and additive manufacturing, and its value is often calibrated through trial-and-error exercises. This adversely affects the capability of numerical simulations when predicting the process behaviour and can eventually hinder the exploitation of fully digitised manufacturing processes, which is a goal of "industry 4.0". In the present work, an enhanced absorption model that takes into account the effects of laser characteristics, incident angle, surface temperature, and material composition is utilised to predict internal heat and fluid flow in laser melting of stainless steel 316L. Employing such an absorption model is physically more realistic than assuming a constant absorptivity and can reduce the costs associated with calibrating an appropriate value. High-fidelity three-dimensional numerical simulations were performed using both variable and constant absorptivity models and the predictions compared with experimental data. The results of the present work unravel the crucial effect of absorptivity on the physics of internal flow in laser material processing. The difference between melt-pool shapes obtained using fibre and CO$_2$ laser sources is explained, and factors affecting the local energy absorption are discussed.
△ Less
Submitted 5 January, 2022;
originally announced January 2022.
-
The Effects of Process Parameters on Melt-pool Oscillatory Behaviour in Gas Tungsten Arc Welding
Authors:
Amin Ebrahimi,
Chris R. Kleijn,
Marcel J. M. Hermans,
Ian M. Richardson
Abstract:
Internal flow behaviour and melt-pool surface oscillations during arc welding are complex and not yet fully understood. In the present work, high-fidelity numerical simulations are employed to describe the effects of welding position, sulphur concentration (60-300 ppm) and travel speed (1.25-5 mm/s) on molten metal flow dynamics in fully-penetrated melt-pools. A wavelet transform is implemented to…
▽ More
Internal flow behaviour and melt-pool surface oscillations during arc welding are complex and not yet fully understood. In the present work, high-fidelity numerical simulations are employed to describe the effects of welding position, sulphur concentration (60-300 ppm) and travel speed (1.25-5 mm/s) on molten metal flow dynamics in fully-penetrated melt-pools. A wavelet transform is implemented to obtain time-resolved frequency spectra of the oscillation signals, which overcomes the shortcomings of the Fourier transform in rendering time resolution of the frequency spectra. Comparing the results of the present numerical calculations with available analytical and experimental datasets, the robustness of the proposed approach in predicting melt-pool oscillations is demonstrated. The results reveal that changes in the surface morphology of the pool resulting from a change in welding position alter the spatial distribution of arc forces and power-density applied to the molten material, and in turn affect flow patterns in the pool. Under similar welding conditions, changing the sulphur concentration affects the Marangoni flow pattern, and increasing the travel speed decreases the size of the pool and increases the offset between top and bottom melt-pool surfaces, affecting the flow structures (vortex formation) on the surface. Variations in the internal flow pattern affect the evolution of melt-pool shape and its surface oscillations.
△ Less
Submitted 6 April, 2021;
originally announced April 2021.
-
Random pattern and frequency generation using a photonic reservoir computer with output feedback
Authors:
Piotr Antonik,
Michiel Hermans,
Marc Haelterman,
Serge Massar
Abstract:
Reservoir computing is a bio-inspired computing paradigm for processing time dependent signals. The performance of its analogue implementations matches other digital algorithms on a series of benchmark tasks. Their potential can be further increased by feeding the output signal back into the reservoir, which would allow to apply the algorithm to time series generation. This requires, in principle,…
▽ More
Reservoir computing is a bio-inspired computing paradigm for processing time dependent signals. The performance of its analogue implementations matches other digital algorithms on a series of benchmark tasks. Their potential can be further increased by feeding the output signal back into the reservoir, which would allow to apply the algorithm to time series generation. This requires, in principle, implementing a sufficiently fast readout layer for real-time output computation. Here we achieve this with a digital output layer driven by a FPGA chip. We demonstrate the first opto-electronic reservoir computer with output feedback and test it on two examples of time series generation tasks: frequency and random pattern generation. We obtain very good results on the first task, similar to idealised numerical simulations. The performance on the second one, however, suffers from the experimental noise. We illustrate this point with a detailed investigation of the consequences of noise on the performance of a physical reservoir computer with output feedback. Our work thus opens new possible applications for analogue reservoir computing and brings new insights on the impact of noise on the output feedback.
△ Less
Submitted 19 December, 2020;
originally announced December 2020.
-
Fluid drag reduction by magnetic confinement
Authors:
Arvind Arun Dev,
Peter Dunne,
Thomas M. Hermans,
Bernard Doudin
Abstract:
The frictional forces of a viscous liquid flow are a major energy loss issue and severely limit microfluidics practical use. Reducing this drag by more than a few tens of percent remain illusive. Here, we show how cylindrical liquid-in-liquid flow leads to drag reduction of 60-99% for sub mm and mm sized channels, irrespective of whether the viscosity of the transported liquid is larger or smaller…
▽ More
The frictional forces of a viscous liquid flow are a major energy loss issue and severely limit microfluidics practical use. Reducing this drag by more than a few tens of percent remain illusive. Here, we show how cylindrical liquid-in-liquid flow leads to drag reduction of 60-99% for sub mm and mm sized channels, irrespective of whether the viscosity of the transported liquid is larger or smaller than that of the encapsulating one. In contrast to lubrication or sheath flow, we do not require the continuous flow of the encapsulating lubricant, here made up of a ferrofluid held in place by magnetic forces. In a laminar flow model with appropriate boundary conditions, we introduce a modified Reynolds number with a scaling that depends on geometrical factors and viscosity ratio of the two liquids. It explains our whole range of data and reveal the key design parameters for optimizing the drag reduction values. Our results therefore open the route to microfluidics designs with pressure gradients possibly reduced by orders of magnitudes.
△ Less
Submitted 1 October, 2021; v1 submitted 26 June, 2020;
originally announced June 2020.
-
Helium Ion Microscopy for Reduced Spin Orbit Torque Switching Currents
Authors:
Peter Dunne,
Ciaran Fowley,
Gregor Hlawacek,
**u Kurian,
Gwenaël Atcheson,
Silviu Colis,
Niclas Teichert,
Bohdan Kundys,
M. Venkatesan,
Jürgen Lindner,
Alina Maria Deac,
Thomas M. Hermans,
J. M. D. Coey,
Bernard Doudin
Abstract:
Spin orbit torque driven switching is a favourable way to manipulate nanoscale magnetic objects for both memory and wireless communication devices. The critical current required to switch from one magnetic state to another depends on the geometry and the intrinsic properties of the materials used, which are difficult to control locally. Here we demonstrate how focused helium ion beam irradiation c…
▽ More
Spin orbit torque driven switching is a favourable way to manipulate nanoscale magnetic objects for both memory and wireless communication devices. The critical current required to switch from one magnetic state to another depends on the geometry and the intrinsic properties of the materials used, which are difficult to control locally. Here we demonstrate how focused helium ion beam irradiation can modulate the local magnetic anisotropy of a Co thin film at the microscopic scale. Real-time in-situ characterisation using the anomalous Hall effect showed up to an order of magnitude reduction of the magnetic anisotropy under irradiation, and using this, multi-level switching is demonstrated. The result is that spin-switching current densities, down to 800 kA cm$^{-2}$, can be achieved on predetermined areas of the film, without the need for lithography. The ability to vary critical currents spatially has implications not only for storage elements, but also neuromorphic and probabilistic computing.
△ Less
Submitted 14 September, 2020; v1 submitted 15 May, 2020;
originally announced May 2020.
-
A Mechanically Tunable Quantum Dot in a Graphene Break Junction
Authors:
Sabina Caneva,
Matthijs D. Hermans,
Martin Lee,
Amador Garcia-Fuente,
Kenji Watanabe,
Takashi Taniguchi,
Cees Dekker,
Jaime Ferrer,
Herre S. J. van der Zant,
Pascal Gehring
Abstract:
Graphene quantum dots (QDs) are intensively studied as platforms for the next generation of quantum electronic devices. Fine tuning of the transport properties in monolayer graphene QDs, in particular with respect to the independent modulation of the tunnel barrier transparencies, remains challenging and is typically addressed using electrostatic gating. We investigate charge transport in back-gat…
▽ More
Graphene quantum dots (QDs) are intensively studied as platforms for the next generation of quantum electronic devices. Fine tuning of the transport properties in monolayer graphene QDs, in particular with respect to the independent modulation of the tunnel barrier transparencies, remains challenging and is typically addressed using electrostatic gating. We investigate charge transport in back-gated graphene mechanical break junctions and reveal Coulomb blockade physics characteristic of a single, high-quality QD when a nanogap is opened in a graphene constriction. By mechanically controlling the distance across the newly-formed graphene nanogap, we achieve reversible tunability of the tunnel coupling to the drain electrode by five orders of magnitude, while kee** the source-QD tunnel coupling constant. These findings indicate that the tunnel coupling asymmetry can be significantly modulated with a mechanical tuning knob and has important implications for the development of future graphene-based devices, including energy converters and quantum calorimeters.
△ Less
Submitted 10 February, 2020;
originally announced February 2020.
-
Neutron imaging of liquid-liquid systems containing paramagnetic salt solutions
Authors:
Tim A. Butcher,
G. J. M. Formon,
P. Dunne,
T. M. Hermans,
F. Ott,
L. Noirez,
J. M. D. Coey
Abstract:
The method of neutron imaging was adopted to map the concentration evolution of aqueous paramagnetic Gd(NO3)3 solutions. Magnetic manipulation of the paramagnetic liquid within a miscible nonmagnetic liquid is possible by countering density-difference driven convection. The formation of salt fingers caused by double-diffusive convection in a liquid-liquid system of Gd(NO3)3 and Y(NO3)3 solutions c…
▽ More
The method of neutron imaging was adopted to map the concentration evolution of aqueous paramagnetic Gd(NO3)3 solutions. Magnetic manipulation of the paramagnetic liquid within a miscible nonmagnetic liquid is possible by countering density-difference driven convection. The formation of salt fingers caused by double-diffusive convection in a liquid-liquid system of Gd(NO3)3 and Y(NO3)3 solutions can be prevented by the magnetic field gradient force.
△ Less
Submitted 13 January, 2020;
originally announced January 2020.
-
A Differentiable Physics Engine for Deep Learning in Robotics
Authors:
Jonas Degrave,
Michiel Hermans,
Joni Dambre,
Francis wyffels
Abstract:
An important field in robotics is the optimization of controllers. Currently, robots are often treated as a black box in this optimization process, which is the reason why derivative-free optimization methods such as evolutionary algorithms or reinforcement learning are omnipresent. When gradient-based methods are used, models are kept small or rely on finite difference approximations for the Jaco…
▽ More
An important field in robotics is the optimization of controllers. Currently, robots are often treated as a black box in this optimization process, which is the reason why derivative-free optimization methods such as evolutionary algorithms or reinforcement learning are omnipresent. When gradient-based methods are used, models are kept small or rely on finite difference approximations for the Jacobian. This method quickly grows expensive with increasing numbers of parameters, such as found in deep learning. We propose the implementation of a modern physics engine, which can differentiate control parameters. This engine is implemented for both CPU and GPU. Firstly, this paper shows how such an engine speeds up the optimization process, even for small problems. Furthermore, it explains why this is an alternative approach to deep Q-learning, for using deep learning in robotics. Finally, we argue that this is a big step for deep learning in robotics, as it opens up new possibilities to optimize robots, both in hardware and software.
△ Less
Submitted 24 November, 2018; v1 submitted 5 November, 2016;
originally announced November 2016.
-
Embodiment of Learning in Electro-Optical Signal Processors
Authors:
Michiel Hermans,
Piotr Antonik,
Marc Haelterman,
Serge Massar
Abstract:
Delay-coupled electro-optical systems have received much attention for their dynamical properties and their potential use in signal processing. In particular it has recently been demonstrated, using the artificial intelligence algorithm known as reservoir computing, that photonic implementations of such systems solve complex tasks such as speech recognition. Here we show how the backpropagation al…
▽ More
Delay-coupled electro-optical systems have received much attention for their dynamical properties and their potential use in signal processing. In particular it has recently been demonstrated, using the artificial intelligence algorithm known as reservoir computing, that photonic implementations of such systems solve complex tasks such as speech recognition. Here we show how the backpropagation algorithm can be physically implemented on the same electro-optical delay-coupled architecture used for computation with only minor changes to the original design. We find that, compared when the backpropagation algorithm is not used, the error rate of the resulting computing device, evaluated on three benchmark tasks, decreases considerably. This demonstrates that electro-optical analog computers can embody a large part of their own training process, allowing them to be applied to new, more difficult tasks.
△ Less
Submitted 27 October, 2016; v1 submitted 19 October, 2016;
originally announced October 2016.
-
Online Training of an Opto-Electronic Reservoir Computer Applied to Real-Time Channel Equalisation
Authors:
Piotr Antonik,
François Duport,
Michiel Hermans,
Anteo Smerieri,
Marc Haelterman,
Serge Massar
Abstract:
Reservoir Computing is a bio-inspired computing paradigm for processing time dependent signals. The performance of its analogue implementation are comparable to other state of the art algorithms for tasks such as speech recognition or chaotic time series prediction, but these are often constrained by the offline training methods commonly employed. Here we investigated the online learning approach…
▽ More
Reservoir Computing is a bio-inspired computing paradigm for processing time dependent signals. The performance of its analogue implementation are comparable to other state of the art algorithms for tasks such as speech recognition or chaotic time series prediction, but these are often constrained by the offline training methods commonly employed. Here we investigated the online learning approach by training an opto-electronic reservoir computer using a simple gradient descent algorithm, programmed on an FPGA chip. Our system was applied to wireless communications, a quickly growing domain with an increasing demand for fast analogue devices to equalise the nonlinear distorted channels. We report error rates up to two orders of magnitude lower than previous implementations on this task. We show that our system is particularly well-suited for realistic channel equalisation by testing it on a drifting and a switching channels and obtaining good performances
△ Less
Submitted 19 October, 2016;
originally announced October 2016.
-
Towards Trainable Media: Using Waves for Neural Network-Style Training
Authors:
Michiel Hermans,
Thomas Van Vaerenbergh
Abstract:
In this paper we study the concept of using the interaction between waves and a trainable medium in order to construct a matrix-vector multiplier. In particular we study such a device in the context of the backpropagation algorithm, which is commonly used for training neural networks. Here, the weights of the connections between neurons are trained by multiplying a `forward' signal with a backward…
▽ More
In this paper we study the concept of using the interaction between waves and a trainable medium in order to construct a matrix-vector multiplier. In particular we study such a device in the context of the backpropagation algorithm, which is commonly used for training neural networks. Here, the weights of the connections between neurons are trained by multiplying a `forward' signal with a backwards propagating `error' signal. We show that this concept can be extended to trainable media, where the gradient for the local wave number is given by multiplying signal waves and error waves. We provide a numerical example of such a system with waves traveling freely in a trainable medium, and we discuss a potential way to build such a device in an integrated photonics chip.
△ Less
Submitted 30 September, 2015;
originally announced October 2015.
-
Photonic Delay Systems as Machine Learning Implementations
Authors:
Michiel Hermans,
Miguel Soriano,
Joni Dambre,
Peter Bienstman,
Ingo Fischer
Abstract:
Nonlinear photonic delay systems present interesting implementation platforms for machine learning models. They can be extremely fast, offer great degrees of parallelism and potentially consume far less power than digital processors. So far they have been successfully employed for signal processing using the Reservoir Computing paradigm. In this paper we show that their range of applicability can…
▽ More
Nonlinear photonic delay systems present interesting implementation platforms for machine learning models. They can be extremely fast, offer great degrees of parallelism and potentially consume far less power than digital processors. So far they have been successfully employed for signal processing using the Reservoir Computing paradigm. In this paper we show that their range of applicability can be greatly extended if we use gradient descent with backpropagation through time on a model of the system to optimize the input encoding of such systems. We perform physical experiments that demonstrate that the obtained input encodings work well in reality, and we show that optimized systems perform significantly better than the common Reservoir Computing approach. The results presented here demonstrate that common gradient descent techniques from machine learning may well be applicable on physical neuro-inspired analog computers.
△ Less
Submitted 12 January, 2015;
originally announced January 2015.
-
Trainable and Dynamic Computing: Error Backpropagation through Physical Media
Authors:
Michiel Hermans,
Michaël Burm,
Joni Dambre,
Peter Bienstman
Abstract:
Machine learning algorithms, and more in particular neural networks, arguably experience a revolution in terms of performance. Currently, the best systems we have for speech recognition, computer vision and similar problems are based on neural networks, trained using the half-century old backpropagation algorithm. Despite the fact that neural networks are a form of analog computers, they are still…
▽ More
Machine learning algorithms, and more in particular neural networks, arguably experience a revolution in terms of performance. Currently, the best systems we have for speech recognition, computer vision and similar problems are based on neural networks, trained using the half-century old backpropagation algorithm. Despite the fact that neural networks are a form of analog computers, they are still implemented digitally for reasons of convenience and availability. In this paper we demonstrate how we can design physical linear dynamic systems with non-linear feedback as a generic platform for dynamic, neuro-inspired analog computing. We show that a crucial advantage of this setup is that the error backpropagation can be performed physically as well, which greatly speeds up the optimisation process. As we show in this paper, using one experimentally validated and one conceptual example, such systems may be the key to providing a relatively straightforward mechanism for constructing highly scalable, fully dynamic analog computers.
△ Less
Submitted 24 July, 2014;
originally announced July 2014.
-
Memristor models for machine learning
Authors:
Juan Pablo Carbajal,
Joni Dambre,
Michiel Hermans,
Benjamin Schrauwen
Abstract:
In the quest for alternatives to traditional CMOS, it is being suggested that digital computing efficiency and power can be improved by matching the precision to the application. Many applications do not need the high precision that is being used today. In particular, large gains in area- and power efficiency could be achieved by dedicated analog realizations of approximate computing engines. In t…
▽ More
In the quest for alternatives to traditional CMOS, it is being suggested that digital computing efficiency and power can be improved by matching the precision to the application. Many applications do not need the high precision that is being used today. In particular, large gains in area- and power efficiency could be achieved by dedicated analog realizations of approximate computing engines. In this work, we explore the use of memristor networks for analog approximate computation, based on a machine learning framework called reservoir computing. Most experimental investigations on the dynamics of memristors focus on their nonvolatile behavior. Hence, the volatility that is present in the developed technologies is usually unwanted and it is not included in simulation models. In contrast, in reservoir computing, volatility is not only desirable but necessary. Therefore, in this work, we propose two different ways to incorporate it into memristor simulation models. The first is an extension of Strukov's model and the second is an equivalent Wiener model approximation. We analyze and compare the dynamical properties of these models and discuss their implications for the memory and the nonlinear processing capacity of memristor networks. Our results indicate that device variability, increasingly causing problems in traditional computer design, is an asset in the context of reservoir computing. We conclude that, although both models could lead to useful memristor based reservoir computing systems, their computational performance will differ. Therefore, experimental modeling research is required for the development of accurate volatile memristor models.
△ Less
Submitted 14 July, 2014; v1 submitted 9 June, 2014;
originally announced June 2014.
