Physics-based Process Research Articles

Development of a multi-layer green-ampt infiltration based modeling framework for hydrologic process simulation and design of various low impact development measures for mitigating urban flooding

ABSTRACT Low Impact Development (LID) measures have proven to be effective in reducing the runoff volume and the peak of runoff at small spatial scales. Assessment of the hydrological impacts of LID measures at river basin scales is currently limited by the absence of adequate physically-based process models that can simulate LID processes at such scales. In this study, an attempt has been made to develop LID modules to simulate the water balance and hydrological processes based on the Multi-Layer Green-Ampt (MLGA) infiltration method, which is an advancement to the existing LID modeling techniques. The performance of the MLGA-based LID modules was benchmarked with the process-intensive numerical tool HYDRUS 1D for about 5000 soil samples across the soil textural classes. The results indicated that the MLGA-based model simulates the hydrological processes in close agreement with HYDRUS for soils with moderate-to-high hydraulic conductivity values, which are amenable for LID implementation.

Investigating the Effects of Mixing Dynamics on Twin-Screw Granule Quality Attributes via the Development of a Physics-Based Process Map.

Twin-screw granulation (TSG) is an emerging continuous wet granulation technique that has not been widely applied in the industry due to a poor mechanistic understanding of the process. This study focuses on improving this mechanistic understanding by analyzing the effects of the mixing dynamics on the granule quality attributes (PSD, content uniformity, and microstructure). Mixing is an important dynamic process that simultaneously occurs along with the granulation rate mechanisms during the wet granulation process. An improved mechanistic understanding was achieved by identifying and quantifying the physically relevant intermediate parameters that affect the mixing dynamics in TSG, and then their effects on the granule attributes were analyzed by investigating their effects on the granulation rate mechanisms. The fill level, granule liquid saturation, extent of nucleation, and powder wettability were found to be the key physically relevant intermediate parameters that affect the mixing inside the twin-screw granulator. An improved geometrical model for the fill level was developed and validated against existing experimental data. Finally, a process map was developed to depict the effects of mixing on the temporal and spatial evolution of the materials inside the twin-screw granulator. This process map illustrates the mechanism of nucleation and the growth of the granules based on the fundamental material properties of the primary powders (solubility and wettability), liquid binders (viscosity), and mixing dynamics present in the system. Furthermore, it was shown that the process map can be used to predict the granule product quality based on the granule growth mechanism.

Solutions to muddy (geoacoustic inversion) problems

Waveguide propagation is a ubiquitous topic in ocean acoustics. This presentation focuses on low-frequency (f < 500 Hz) sound propagation in coastal oceans (water depth shallower than 500 m). These environments are highly dynamical and complex, and act as dispersive waveguides, bounded by the sea-surface and the seafloor. The underwater acoustic field can thus be described by a set of modes that propagate with frequency-dependent speeds. Propagation of transient acoustic signals to a distant receiver (range > 1 km) accrues multi-modal dispersion information about the propagation medium. To extract relevant information about the oceanic environment from the acoustic recording, physics-based processing methods must be used. In this presentation, I will briefly review modal propagation and time-frequency analysis. I will then show how these approaches can be combined into a non-linear signal processing method dedicated to extracting modal information from a single receiver: this information is the foundation of a transdimensional inversion method used to characterize the oceanic environment. This method will be used to perform geoacoustic inversion on the New England Mud Patch. Using several datasets collected under different oceanographic conditions, I will notably demonstrate a successful estimation of the seafloor properties, consistent with geophysical core measurements and other inversion studies, even when the water column is dynamical and mostly unknown.

In-situ qualification and physics-based process design for aerosol jet printing via spatially correlated light scattering measurements

Aerosol jet printing is a contactless, digital, and additive technique broadly used for manufacturing flexible, hybrid, and conformal electronics. However, both intra-batch and batch-to-batch variability have hindered widespread industry adoption and scaling to production volumes. Recently, light scattering measurements have emerged as a tool to measure aerosol volume fraction – a key parameter determining deposition rate – and have proven an effective feedback source for closed-loop control on timescales ranging from tens of minutes to hours. While this is a promising capability to mitigate long-term process drift, it lacks the temporal resolution to validate print quality for complex and precise electronic circuits. Here, real-time process monitoring with <1 s resolution is demonstrated for in-situ qualification of aerosol jet printing. To begin, the correlation between light scattering and aerosol deposition is validated at 500 ms time intervals, allowing deposition rate to be mapped to positional coordinates and providing a new data stream to drive quality control assessments. This allows optical scattering to be connected to electrical properties including line and sheet resistance, resulting in predictions nominally within 10% of measured values for sets of printed devices. Finally, the ability to spatially correlate print defects during fabrication is demonstrated as a foundation for a print repair framework, successfully repairing prints with randomly induced defects to achieve resistance within 5% of a control set.

Inverse Problems for Physics-Based Process Models

We describe and compare two formulations of inverse problems for a physics-based process model in the context of uncertainty and random variability: the Bayesian inverse problem and the stochastic inverse problem. We describe the foundations of the two problems in order to create a context for interpreting the applicability and solutions of inverse problems important for scientific and engineering inference. We conclude by comparing them to statistical approaches to related problems, including Bayesian calibration of computer models.

Applications and prospects of machine learning for aerosol jet printing: A review

Aerosol Jet Printing (AJP) is an additive manufacturing process that deposits ink-like materials suspended as an aerosol mist. AJP creates three-dimensional (3D) functional structures onto flat or conformal surfaces in complex shapes without the aid of additional tooling, enabling the manufacturing of extremely fine electrical interconnects with freeform structures. Due to the novelty and complexity of AJP, physical understanding is rather limited, hindering physics-based process modeling and analysis. Fortunately, the data resources from AJP applications, e.g., 3D Computer-Aided-Design data, Standard Triangle Language files, in-situ images of part, and nozzle motion records, provide an unparalleled opportunity for developing data-driven, Machine Learning (ML) methods to characterize AJP processes, support process control, and facilitate product improvement. To thoroughly identify the newfound opportunities, this study reviews state-of-the-art ML methods used in AJP applications, investigates open issues in AJP, and outlooks future development of ML-based research topics for AJP. It sheds light on how to maximize the value of ML on AJP data to develop scalable, generalizable decision-making methods. More future works along the direction will be motivated.

Machine learning-based identification of interpretable process-structure linkages in metal additive manufacturing

The use of data-driven methods for metal additive manufacturing (AM) is currently gaining importance as indicated by the increasing number of scientific literature in this field. Incorporation of data-driven methods has the potential to eliminate current bottlenecks in microstructure design given the diverse and complex nature of microstructures in additively manufactured metals. So far, coupling of existing simulation methods, e.g. physics-based process and microstructure models, to simulate AM microstructures with desired morphological characteristics requires extensive computational resources, high computation times and therefore, allows no scalable output. The extension of experimental- and simulation-based approaches by machine learning (ML) algorithms enables fast and computationally efficient predictions. However, the underlying architecture of ML algorithms often does not allow domain experts to interpret how predictions of the model were made and which features are responsible to what extent. This is why ML models are often referred to as black- box models. In this study, we present a data-driven framework based on physics-based simulation data to reveal explainable process-(micro)structure (P-S) linkages for metal AM. We provide an open-source dataset of 960 unique 3D microstructures created by simulation of powder bed fusion in metal AM. We employed the stochastic parallel particle kinetic simulator (SPPARKS) that is based on the kinetic Monte Carlo (kMC) method as an exemplary AM microstructure generator. Selected ML regression algorithms aim to predict 3D chord length distributions (CLDs), as a morphology descriptor, depending on the associated process parameter combinations. Various dimension reduction algorithms are applied for computationally efficient use of the data space. The proposed methodology allows (i) microstructure predictions under given processing conditions and (ii) to navigate experts in the process parameter space to achieve target microstructures. In this context, SHAP (SHapley Additive exPlanations) values are used to decipher the contribution of individual process parameters to the microstructure evolution. In particular, SHAP values calculated in this study unfold the width of the melt pool and the heat-affected zone as dominant features on the model output. We provide open-access to the used dataset and methods for the scientific community to gain experience with the proposed approach.

Analysis of burr formation in finish machining of nickel-based superalloy with worn tools using micro-scale in-situ techniques

The formation of burrs is among the most significant factors affecting quality and productivity in machining. Burrs are a negative byproduct of machining processes that are difficult to avoid because of a limited understanding of the complex burr formation mechanisms in relation to cutting conditions, including both process parameters and tool condition. Thus, the objective of this work was to characterize burr formation under finish machining conditions via a high-speed, high-resolution in-situ experimental method. Various parameters pertaining to burr geometry such as height, thickness, and initial negative shear angle were measured both during and after cutting. Results showed that varying the conditions of uncut chip thickness, tool-wear, and cutting speed all have a significant effect on burr formation, although certain burr metrics were found to be insensitive with respect to different process conditions because the difference was statistically insignificant. This study provides new insights into the relationships between the workpiece material's microstructure, machining parameters, and tool condition on both crack formation and propagation/plasticity during burr formation. Using digital image correlation (DIC) and a physics-based process model not previously utilized for burr formation analysis, the displacement and corresponding flow stress were calculated at the exit burr root location. This novel semi-analytical approach revealed that the normalized stress at the exit burr root was approximately equal to the flow stress for a variety of different conditions, indicating the potential for model-based prediction of burr formation mechanics. Finally, this study investigates factors that influence fracture evolution during exit burr formation. It was found that negative exit burrs are a direct result of high strain rate and high uncut chip thickness, which was expected, but also a microstructural size effect and a tool-wear effect, neither of which have been previously reported. By harnessing ultra-high-speed imaging and advanced optical microscopy techniques, this manuscript deals with the fundamentals of burr formation, including new insights into material response at the grain-scale to the loads imposed with both sharp and worn tools.

This Looks Like That There: Interpretable Neural Networks for Image Tasks When Location Matters

Abstract We develop and demonstrate a new interpretable deep learning model specifically designed for image analysis in Earth system science applications. The neural network is designed to be inherently interpretable, rather than explained via post hoc methods. This is achieved by training the network to identify parts of training images that act as prototypes for correctly classifying unseen images. The new network architecture extends the interpretable prototype architecture of a previous study in computer science to incorporate absolute location. This is useful for Earth system science where images are typically the result of physics-based processes, and the information is often geolocated. Although the network is constrained to only learn via similarities to a small number of learned prototypes, it can be trained to exhibit only a minimal reduction in accuracy relative to noninterpretable architectures. We apply the new model to two Earth science use cases: a synthetic dataset that loosely represents atmospheric high and low pressure systems, and atmospheric reanalysis fields to identify the state of tropical convective activity associated with the Madden–Julian oscillation. In both cases, we demonstrate that considering absolute location greatly improves testing accuracies when compared with a location-agnostic method. Furthermore, the network architecture identifies specific historical dates that capture multivariate, prototypical behavior of tropical climate variability. Significance Statement Machine learning models are incredibly powerful predictors but are often opaque “black boxes.” The how-and-why the model makes its predictions is inscrutable—the model is not interpretable. We introduce a new machine learning model specifically designed for image analysis in Earth system science applications. The model is designed to be inherently interpretable and extends previous work in computer science to incorporate location information. This is important because images in Earth system science are typically the result of physics-based processes, and the information is often map based. We demonstrate its use for two Earth science use cases and show that the interpretable network exhibits only a small reduction in accuracy relative to black-box models.

Rainfall Runoff Balance Enhanced Model Applied to Tropical Hydrology

The integrative and comprehensive analysis considering the spatial and temporal representation of the hydrological process, such as the distribution of rainfall, land cover and land use, is a challenge for the water resources management. In tropical areas, energy availability throughout the year defines the rainfall distribution and evapotranspiration rate according to vegetation heterogeneity. To quantify water balance in tropical areas including these heterogeneities in the soil-vegetation-atmosphere relationship, we developed a fully distributed hydrological model called the Rainfall Runoff Balance Enhanced Model (RUBEM). The model was developed under a physics-based process structure, using remote sensing data to represent soil-water balance patterns, such as evapotranspiration, interception, baseflow, lateral flow, recharge, and runoff. The calibration procedure was based on nine global parameters. RUBEM could represent the spatio-temporal heterogeneities (soil, land use and land cover (LULC), topography, vegetation, and climate) in three basins in a tropical area. The results showed good adherence between the processes governing the soil-vegetation-atmosphere relationship according to the humidity indicator and the runoff coefficient. Overall, RUBEM can be used to help improve the management and planning of integrated water resources under climate, land use, and land cover changes in tropical regions.

Desalinating a real hyper-saline pre-treated produced water via direct-heat vacuum membrane distillation

Membrane distillation (MD) is an emerging thermal desalination technology capable of desalinating waters of any salinity. During typical MD processes, the saline feedwater is heated and acts as the thermal energy carrier; however, temperature polarization (as well as thermal energy loss) contributes to low distillate fluxes, low single-pass water recovery and poor thermal efficiency. An alternative approach is to integrate an extra thermal energy carrier as part of the membrane and/or module assembly, which can channel externally provided heat directly to the membrane-feedwater interface and/or along the feed channel length. This direct-heat delivery has been demonstrated to increase single-pass water recovery and enhance the overall thermal efficiency. We developed a bench-scale direct-heated vacuum MD (DHVMD) process to desalinate pre-treated oil and gas “produced water” with an initial total dissolved solids of 115,500 ppm at a feed temperature ranging between 24 and 32 °C. We evaluated both water flux and specific energy consumption (SEC) as a function of water recovery. The system achieved a 50% water recovery without significant scaling, with an average flux >6 kg m–2 hr–1 and a SEC as low as 2,530 kJ kg−1. The major species of mineral scales (i.e., NaCl, CaSO4, and SrSO4) that limited the water recovery to 68% were modeled in terms of thermodynamics and identified by scanning electron microscopy and energy-dispersive X-ray spectroscopy. In addition, we further developed and employed a physics-based process model to estimate temperature, salinity, water transport and energy flows for full-scale vacuum MD and DHVMD modules. Model results show that a direct-heat input rate of 3,600 W can increase single-pass water recovery from 2.1% to 3.1% while lowering the thermal SEC from 7,800 kJ kg−1 to 6,517 kJ kg−1 in an unoptimized module. Finally, the scaling up potential of DHVMD process is briefly discussed.

Optimisation of manufacturing process parameters for variable component geometries using reinforcement learning

Tailoring manufacturing processes to optimum part quality often requires numerous resource-intensive trial experiments in practice. Physics-based process simulations in combination with general-purpose optimisation algorithms allow for an a priori process optimisation and help concentrate costly trials on the most promising variants. However, considerable computation times are a significant barrier, especially for iterative optimisation. Surrogate-based optimisation often helps reduce the computational effort but surrogate models are typically case-specific and cannot adapt to different manufacturing situations. Consequently, even minor problem variations e.g. geometry adaptions invalidate the surrogate and require resampling of data and retraining of the surrogate. Reinforcement Learning aims at inferring optimal actions in variable situations. In this work, it is used to train a neural network to estimate optimal process parameters (“actions”) for variable component geometries (“situations”). The use case is fabric forming in which pressure pads are positioned to optimise the material intake. After training, the network is found to give meaningful parameter estimations even for new geometries not considered during training. Thus, it extracts reusable information from generic process samples and successfully applies it to new, non-generic components. Since data is reused rather than resampled, the approach is deemed a promising option for lean part and process development.

Uncertainty analysis of curing-induced dimensional variability of composite structures utilizing physics-guided Gaussian process meta-modeling

Composite manufacturing process involves a suite of complex and inter-related procedures that span across multiple physics domains and scales. A variety of uncertainties that inevitably exist may degrade the quality of composites produced. Quantitatively characterizing the effect of uncertainties in such process hence becomes critically important. In this study, we establish a systematic framework for uncertainty analysis of composite manufacturing process. A finite element processing model has been developed to characterize the multi-physics and multi-scale nature of composite manufacturing process, based upon which a Gaussian process (GP) meta-model has been synthesized for efficient uncertainty quantification. By leveraging the well-trained GP meta-model, an importance ranking analysis of uncertainties was then carried out using a series of metrics, i.e., Pearson coefficient, Sobol index and Shapley Additive exPlanations (SHAP). Since the physics-based processing model involves several simplifying assumptions and empirical relations, modeling errors were also considered in the uncertainty analysis. Comprehensive case studies, which aim at elucidating the causes of composite spring-in angles, were conducted to examine the feasibility and validity of this new framework. Specifically, the mean error of GP predictions is smaller than 2%, and the uncertainty importance ranking can be obtained with high confidence in both cases with and without modeling errors.

Part-optimized forming by spatially distributed vaporizing foil actuators

Electrically vaporizing foil actuators are employed as an innovative high speed sheet metal forming technology, which has the potential to lower tool costs. To reduce experimental try-outs, a predictive physics-based process design procedure is developed for the first time. It consists of a mathematical optimization utilizing numerical forming simulations followed by analytical computations for the forming-impulse generation through the rapid Joule heating of the foils. The proposed method is demonstrated for an exemplary steel sheet part. The resulting process design provides a part-specific impulse distribution, corresponding parallel actuator geometries, and the pulse generator’s charging energy, so that all process parameters are available before the first experiment. The experimental validation is then performed for the example part. Formed parts indicate that the introduced method yields a good starting point for actual testing, as it only requires adjustments in the form of a minor charging energy augmentation. This was expectable due to the conservative nature of the underlying modeling. The part geometry obtained with the most suitable charging energy is finally compared to the target geometry.

Physics-Based Simulations of Chip Flow over Micro-Textured Cutting Tool in Orthogonal Cutting of Alloy Steel

Physics-based process simulations have the potential to allow virtual process design and the development of digital twins for smart machining applications. This paper presents 3D cutting simulations using the finite element method (FEM) and investigates the physical state variables that are fundamental to the reduction in cutting forces, friction, and tool wear when micro-textured cutting tools are employed. For this goal, textured cemented carbide cutting tool inserts are designed, fabricated, and tested in the orthogonal dry cutting of a nickel-chromium-molybdenum alloy steel. Cutting forces and friction coefficients are compared against the non-textured tool, revealing the effects of texture parameters. Chip flow over the textured tool surface and process variables at the chip-tool contact are investigated and compared. The results reveal the fundamental sources of such improvements. Archard’s wear rate as a composition of process variables is utilized to compare experimental and simulated wear on the textured cutting tools. The effects of texture and cutting conditions on tool wear and adhesion characteristics are further discussed on the simulation results with experimental comparisons. It was found that the results obtained from these simulations provide further fundamental insights about the micro-textured cutting tools.

Experimental analysis of tilt angle-dependent dynamic properties of a 5-axis milling center

Free-formed surfaces, e.g. of dies, can be finished using 5-axis milling processes. Due to the varying tool-engagement situations and the dynamic behavior of the tool-spindle-machine system, the avoidance of chatter vibrations is challenging for these processes. Since the dynamic behavior depends on the position of the axes, the process stability can be improved by choosing beneficial tilt angles along the NC path. In this paper, a geometric physically-based process simulation was used to investigate the dependence of the calculated stability limit on the tilt angle and the speed-dependent parameters of the force model.

Modeling the Effect of Fabrication Process on Grain Boundary Formation in Nb/Al-AlO$_x$/Nb Josephson Junction Circuit

By modeling the propagation of a seed layer with various crystal orientations, this study explores the influence of process variations on grain formation with the help of a physics-based process simulator. Grain boundaries allow easy diffusion of foreign atoms through the lattice, which causes Al to move inside the Nb bottom electrode layer and more interestingly, O to penetrate through the Al layer during oxidation and create a barrier with non-uniform thickness. In addition to thickness variations, the grain structure exhibited by Nb and Al can cause significant suppression of supercurrent at the boundaries depending on the degree of lattice mismatch, impurity deposition, etc. This work details the process simulation of grain boundary formation and aims to provide geometrical models that may be used in the simulation of device performance to account for process-induced variations.

Intelligent approach based on FEM simulations and soft computing techniques for filling system design optimisation in sand casting processes

This paper reports an intelligent approach for modeling and optimisation of filling system design (FSD) in the case of sand casting process of aluminium alloy. In order to achieve this purpose, physics-based process modeling using finite element method (FEM) has been integrated with artificial neural networks (ANN) and genetic algorithm (GA) soft computing techniques. A three-dimensional FE model of the studied process has been developed and validated, using experimental literature data, to predict two melt flow behaviour (MFB) indexes named ingate velocity and jet high. Two feed-forward back-propagation ANN-based process models were developed and optimised to establish the relationship between the FSD input parameters and each studied MFB index. Both ANN models were trained, tested and tuned by using database generated from FE computations. It was found that both ANN models could independently predict, with a high accuracy, the values of the ingate velocity and the jet high for training and test data. The developed ANN models were coupled with an evolutionary GA to select the optimal FSD for each one. The validity of the found solutions was tested by comparing ANN-GA prediction with FE computation for both studied MFB indexes. It was found that error between predicted and simulated values does not exceed 5.61% and 6.31% respectively for the ingate velocity and the jet high, which proves that the proposed approach is reliable and robust for FSD optimisation.

Uniform Crystal Formation and Electrical Variability Reduction in Hafnium-Oxide-Based Ferroelectric Memory by Thermal Engineering

In this paper, we achieved excellent variation control, endurance enhancement, and leakage reduction in zirconium (Zr)-doped hafnium oxide (Hf1–xZrxO2) based ferroelectric films by the germination of large ferroelectric grains through extending the duration of rapid thermal annealing without increasing the temperature beyond 700 °C. The pivotal point of this work is to reduce electrical variations in ferroelectric capacitors, which we attribute to the random variation in ferroelectric and dielectric phases. The motivation for this research originally stemmed from Johnson–Mehl–Avrami–Kolmogorov’s theory of nucleation, which predicts that a sufficient time is required for forming any particular phase during phase transformation. Instead of increasing the temperature beyond 700 °C, extending the duration of rapid thermal annealing allows uniform crystal to form, which agrees with the theory. A 10 nm thick Hf1–xZrxO2 film (x = 0.2) has been fabricated. The duration and temperature of rapid thermal annealing (RTA) under a nitrogen (N2) atmosphere at 1 atmospheric pressure (atm) is varied to observe its effect on crystal formation and the electrical properties of the devices. It was observed that very low temperature and short (30 s) duration annealing at 1 atm pressure cannot infuse ferroelectricity in Hf1–xZrxO2. Structural analysis clearly showed the formation of large and uniform ferroelectric domains with negligible impact on surface roughness upon extending the annealing duration up to 180 s. The device-to-device variation in terms of standard deviation of coercive voltage and peak capacitance are reduced from 400 to 17 mV and from 20 to 4 fF/cm2, respectively, by increasing the RTA duration at 700 °C. The devices have also displayed endurance above 1 million cycles and leakage current density below 10 pA/μm2. The simple physics-based process discussed here reduces the thermal budget required for Hf1–xZrxO2, mitigates the randomness in the domain distribution, and infuses deterministic switching. This improvement paves the way for implementing Hf1–xZrxO2-based, deeply scaled ferroelectric devices for memory and steep slope device applications.

Development and validation of a robust integrated thermal power plant model for load loss analysis and identification

The development of deep learning methodologies for the analysis of thermal power plant load losses requires a combination of real plant data and data derived from fundamental physics-based process models. For this purpose, a robust integrated power plant thermofluid process model of a complete +600MW coal-fired power plant was developed within the Flownex Simulation Environment. It consists of standard and compound components, combined with specially developed scripts to ensure complete energy balance, specifically on the two-phase tank components. This enables simulation of the complete plant operation to determine power output as a function of any given set of internal and external operational variables, boundary conditions and component states. The model was validated against real plant design and acceptance test data. In order to demonstrate the ability of the model it was used to evaluate the plant performance related to three specific load loss inducing scenarios. The results show that a combination of mechanical faults, process anomalies and operational phenomena can be analysed. This provides the basis for generating model-based performance data that can be combined with real plant data to facilitate the development of deep learning analytics tools for load loss fault diagnosis and root cause analysis, as well as fault propagation and load loss forecasting.