Search

Purpose: Simulation-based digital twins represent an effort to provide high-accuracy real-time insights into operational physical processes. However, the computation time of many multi-physical simulation models is far from real-time. It might even exceed sensible time frames to produce sufficient data for training data-driven reduced-order models. This study presents TwinLab, a framework for data-efficient, yet accurate training of neural-ODE type reduced-order models with only two data sets. Design/methodology/approach: Correlations between test errors of reduced-order models and distinct features of corresponding training data are investigated. Having found the single best data sets for training, a second data set is sought with the help of similarity and error measures to enrich the training process effectively. Findings: Adding a suitable second training data set in the training process reduces the test error by up to 49% compared to the best base reduced-order model trained only with one data set. Such a second training data set should at least yield a good reduced-order model on its own and exhibit higher levels of dissimilarity to the base training data set regarding the respective excitation signal. Moreover, the base reduced-order model should have elevated test errors on the second data set. The relative error of the time series ranges from 0.18% to 0.49%. Prediction speed-ups of up to a factor of 36,000 are observed. Originality: The proposed computational framework facilitates the automated, data-efficient extraction of non-intrusive reduced-order models for digital twins from existing simulation models, independent of the simulation software., Comment: Accepted version of the revised manuscript published in Engineering Computations

As humans advance toward a higher level of artificial intelligence, it is always at the cost of escalating computational resource consumption, which requires developing novel solutions to meet the exponential growth of AI computing demand. Neuromorphic hardware takes inspiration from how the brain processes information and promises energy-efficient computing of AI workloads. Despite its potential, neuromorphic hardware has not found its way into commercial AI data centers. In this article, we try to analyze the underlying reasons for this and derive requirements and guidelines to promote neuromorphic systems for efficient and sustainable cloud computing: We first review currently available neuromorphic hardware systems and collect examples where neuromorphic solutions excel conventional AI processing on CPUs and GPUs. Next, we identify applications, models and algorithms which are commonly deployed in AI data centers as further directions for neuromorphic algorithms research. Last, we derive requirements and best practices for the hardware and software integration of neuromorphic systems into data centers. With this article, we hope to increase awareness of the challenges of integrating neuromorphic hardware into data centers and to guide the community to enable sustainable and energy-efficient AI at scale., Comment: 11 pages, 2 figures, presented as poster at NICE 2024, 2nd version with updated author list and minor updates

This work demonstrates the use of the Digital Twin methodology to predict the water concentration and temperature of chicken meat. It marks a milestone on the path to autonomous cooking devices that do not only control device temperatures but culinary food quality markers as well. A custom water transport equation is coupled to the energy equation. The transport equations are implemented in ANSYS Fluent 2019R2 via User Defined Function (UDF) code written in C. The model is in good agreement with experiments provided by project partners. Thermal fluid-structure interaction simulations of pan-frying are performed to obtain realistic heat transfer coefficients. They indicate that the coupling of food transport equations to the surrounding heat transfer mechanisms, such as radiation and natural convection, seems promising for future research. Co-simulation of the process is not feasible during operation in the field, so reduced-order models (ROM) are introduced. An evaluation of ROM toolkits on the ANSYS platform reveals that linear time-invariant (LTI) models are unsuitable for cooking applications. In contrast, the recently launched Dynamic ROM Builder predicts the core temperatures with significantly low errors (factor ten below the model and discretization errors of the full-order model). Two examples demonstrate the usage of a Digital Twin controlling the core temperature of chicken fillets. The PI closed-loop control system remains insensitive to errors induced by the Dynamic ROM evaluation., Comment: Accepted version of the manuscript published in CADFEM ANSYS Simulation Conference 2019. Received a best paper award

One possible way of making thermal processing controllable is to gather real-time information on the product's current state. Often, sensory equipment cannot capture all relevant information easily or at all. Digital Twins close this gap with virtual probes in real-time simulations, synchronized with the process. This paper proposes a physics-based, data-driven Digital Twin framework for autonomous food processing. We suggest a lean Digital Twin concept that is executable at the device level, entailing minimal computational load, data storage, and sensor data requirements. This study focuses on a parsimonious experimental design for training non-intrusive reduced-order models (ROMs) of a thermal process. A correlation ($R=-0.76$) between a high standard deviation of the surface temperatures in the training data and a low root mean square error in ROM testing enables efficient selection of training data. The mean test root mean square error of the best ROM is less than 1 Kelvin (0.2 % mean average percentage error) on representative test sets. Simulation speed-ups of Sp $\approx$ 1.8E4 allow on-device model predictive control. The proposed Digital Twin framework is designed to be applicable within the industry. Typically, non-intrusive reduced-order modeling is required as soon as the modeling of the process is performed in software, where root-level access to the solver is not provided, such as commercial simulation software. The data-driven training of the reduced-order model is achieved with only one data set, as correlations are utilized to predict the training success a priori., Comment: Accepted version of manuscript published in Innovative Food Science and Emerging Technologies

This work introduces the concept of an autonomous cooking process based on Digital Twin method- ology. It proposes a hybrid approach of physics-based full order simulations followed by a data-driven system identification process with low errors. It makes faster-than-real-time simulations of Digital Twins feasible on a device level, without the need for cloud or high-performance computing. The concept is universally applicable to various physical processes., Comment: Accepted version of manuscript published in Proceedings of WCCM-ECCOMAS 2020

In this numerical study, an original approach to simulate non-isothermal viscoelastic fluid flows at high Weissenberg numbers is presented. Stable computations over a wide range of Weissenberg numbers are assured by using the root conformation approach in a finite volume framework on general unstructured meshes. The numerical stabilization framework is extended to consider thermo-rheological properties in Oldroyd-B type viscoelastic fluids. The temperature dependence of the viscoelastic fluid is modeled with the time-temperature superposition principle. Both Arrhenius and WLF shift factors can be chosen, depending on the flow characteristics. The internal energy balance takes into account both energy and entropy elasticity. Partitioning is achieved by a constant split factor. An analytical solution of the balance equations in planar channel flow is derived to verify the results of the main field variables and to estimate the numerical error. The more complex entry flow of a polyisobutylene-based polymer solution in an axisymmetric 4:1 contraction is studied and compared to experimental data from the literature. We demonstrate the stability of the method in the experimentally relevant range of high Weissenberg numbers. The results at different imposed wall temperatures, as well as Weissenberg numbers, are found to be in good agreement with experimental data. Furthermore, the division between energy and entropy elasticity is investigated in detail with regard to the experimental setup., Comment: 25 pages, 13 figures, 2 tables