Publications
123 results found
Hammond J, Pietropaoli M, Montomoli F, 2022, Robust data-driven turbulence closures for improved heat transfer prediction in complex geometries, INTERNATIONAL JOURNAL OF HEAT AND FLUID FLOW, Vol: 98, ISSN: 0142-727X
Hammond J, Montomoli F, Pietropaoli M, et al., 2022, Machine Learning for the Development of Data-Driven Turbulence Closures in Coolant Systems, JOURNAL OF TURBOMACHINERY-TRANSACTIONS OF THE ASME, Vol: 144, ISSN: 0889-504X
- Author Web Link
- Cite
- Citations: 3
Pepper N, Montomoli F, Sharma S, 2022, A Non-Parametric Histogram Interpolation Method for Design Space Exploration, JOURNAL OF MECHANICAL DESIGN, Vol: 144, ISSN: 1050-0472
Pepper N, Crespo L, Montomoli F, 2022, Adaptive learning for reliability analysis using Support Vector Machines, RELIABILITY ENGINEERING & SYSTEM SAFETY, Vol: 226, ISSN: 0951-8320
- Author Web Link
- Cite
- Citations: 1
Hammond J, Pepper N, Montomoli F, et al., 2022, Machine Learning Methods in CFD for Turbomachinery: A Review, INTERNATIONAL JOURNAL OF TURBOMACHINERY PROPULSION AND POWER, Vol: 7
- Author Web Link
- Cite
- Citations: 6
Hammond J, Marioni YF, Montomoli F, 2022, Error Quantification for the Assessment of Data-Driven Turbulence Models, FLOW TURBULENCE AND COMBUSTION, Vol: 109, Pages: 1-26, ISSN: 1386-6184
- Author Web Link
- Cite
- Citations: 1
Cassinelli A, Mateo Gabín A, Montomoli F, et al., 2022, Reynolds sensitivity of the wake passing effect on a LPT cascade using spectral/hp element methods, International Journal of Turbomachinery, Propulsion and Power, Vol: 7, Pages: 8-8, ISSN: 2504-186X
Reynolds-Averaged Navier–Stokes (RANS) methods continue to be the backbone of CFD-based design; however, the recent development of high-order unstructured solvers and meshing algorithms, combined with the lowering cost of HPC infrastructures, has the potential to allow for the introduction of high-fidelity simulations in the design loop, taking the role of a virtual wind tunnel. Extensive validation and verification is required over a broad design space. This is challenging for a number of reasons, including the range of operating conditions, the complexity of industrial geometries and their relative motion. A representative industrial low pressure turbine (LPT) cascade subject to wake passing interactions is analysed, adopting the incompressible Navier–Stokes solver implemented in the spectral/hp element framework Nektar++. The bar passing effect is modelled by leveraging a spectral-element/Fourier Smoothed Profile Method. The Reynolds sensitivity is analysed, focusing in detail on the dynamics of the separation bubble on the suction surface as well as the mean flow properties, wake profiles and loss estimations. The main findings are compared with experimental data, showing agreement in the prediction of wake traverses and losses across the entire range of flow regimes, the latter within 5% of the experimental measurements.
Hammond J, Pietropaoli M, Montomoli F, 2022, Topology optimisation of turbulent flow using data-driven modelling, STRUCTURAL AND MULTIDISCIPLINARY OPTIMIZATION, Vol: 65, ISSN: 1615-147X
- Author Web Link
- Cite
- Citations: 3
Bisio V, Montomoli F, Rossin S, et al., 2022, Predictions and uncertainty quantification of the loading induced by deflagration events on surrounding structures, Process Safety and Environmental Protection, Vol: 158, Pages: 445-460, ISSN: 0957-5820
The threat of accidental hydrocarbon explosions is of major concern to industrial operations; in particular, there is a need for design tools to assess and quantify the effects of potential deflagration events. Here we present a design methodology based on analytical models that allow assessing the loading and structural response of objects exposed to pressure waves generated by deflagration events. The models allow determining: i) the importance of Fluid-Structure Interaction (FSI) effects; ii) the transient pressure histories on box-like or circular cylindrical objects, including the effects of pressure clearing; iii) the dynamic response of structural components that can be idealised as fully clamped beams. We illustrate by three case studies the complete design methodology and validate the analytical models by comparing their predictions to those of detailed CFD and FE simulations. We employ the validated analytical models to perform Monte Carlo analyses to quantify, for box-like structures, how the uncertainty in input design variables propagates through to the expected maximum force and impulse. We present this information in the form of non-dimensional uncertainty maps.
Sakuma M, Pepper N, Warnakulasuriya S, et al., 2022, Multi-fidelity uncertainty quantification of high Reynolds number turbulent flow around a rectangular 5:1 cylinder, 3rd International Conference on Uncertainty Quantification in Computational Sciences and Engineering, Publisher: Techno Press, Pages: 127-136, ISSN: 1226-6116
This work shows the application of Multi-fidelity Uncertainty Quantification to WindEngineering problems. As test case a rectangular shape is used, with a fillet radius, in orderto represent the geometrical variations that can affect buildings or other bluff bodies. Therectangular cylinder used has a chord-to-thickness ratio 5:1. This rectangular shape is animportant basic shape for wind engineering tasks, e.g. in case of buildings or other bluff bodiesexposed to the flow. Moreover it is well investigated and documented.Coarse and fine meshes are used as low and high fidelity models respectively. To performCFD simulations, the stabilized finite element methods are used in both the high and low fidelitymodel with a CFD code developed by TUM and the International Center for Numerical Methodsin Engineering. The underlying UQ framework is based on a Sparse Arbitrary Moment BasedAlgorithm (SAMBA) developed at ICL. In the formulation the number of simulations is reducedusing a Smolyak sparsity model.The multi-fidelity extension, with application to wind engineering problems is discussed andpresented in this work. The overall goal of such formulation is to gain an accuracy of mixed lowhigh fidelity simulations comparable to the ones obtained with only high fidelity simulations, ata fraction of the computational cost.
Raske N, Gonzalez OA, Furino S, et al., 2022, THERMAL MANAGEMENT FOR ELECTRIFICATION IN AIRCRAFT ENGINES: OPTIMIZATION OF COOLANT SYSTEM
This work shows the application of fluid Topology Optimization for the thermal management of electric parts in aircraft engines. There is a growing trend in electrification of current engines, but this requires higher power to be dissipated. Standard coolant systems based on serpentines are not effective enough to remove the heat produced by these systems. In this work a new solution is shown which has a higher heat dissipation, lower pressure loss and a reduction in the mass of the heat exchanger. The optimization is carried out with the commercial solver TOffee. Manufacturing constraints allow the part to be manufacturable using two different methods: additive manufacturing and milling with diffusion bonding. Overall, the optimized geometry, a nearly 2D coldplate with extruded channels, shows higher heat dissipation and a reduction of weight of 38.5% if compared with the baseline heat exchanger. The overall pressure drop has been also reduced by 65%.
Lahooti M, Vivarelli G, Montomoli F, et al., 2022, Under-resolved direct numerical simulation of naca0012 at stall
In this work a high-order spectral-h/p element solver is employed to efficiently but accurately resolve the flow field around the NACA0012 aerofoil. In particular, the conditions considered are a Reynolds number of 150000 and three angles of attack, namely 9◦, 10◦ and 12◦. This particular study aims at providing the necessary preliminary insight into the flow dynamics of the turbulent transition at near-post stall with very well resolved Large Eddy Simulation range if not Direct Numerical Simulation. Therefore, the first step consists of determining the mesh resolution required and the spanwise length. Our results repeatedly demonstrate the possible existence of a large-scale low-frequency 2D dominant flow structure over the span where. It was found that employing a single chord length in z is not sufficient to capture it. Concerning the flow behaviour, it is observed that a laminar separation bubble forms at the aerofoil leading edge. This tends to move upstream, shorten in length and increase in height as the angle of attack is increased. Mild three-dimensional behaviour is seen right from the beginning of the aerofoil suction surface with turbulent transition occurring just after the reattachment point. In particular, this is seen to happen sooner with higher aerofoil inclination. Finally, our results indicate an interaction of large-scale structures with the boundary layer.
Vivarelli G, Isler JA, Montomoli F, et al., 2022, HIGH-ORDER SPECTRAL/HP COMPRESSIBLE AND INCOMPRESSIBLE COMPARISON OF TRANSITIONAL BOUNDARY-LAYERS SUBJECT TO A REALISTIC PRESSURE GRADIENT AND HIGH REYNOLDS NUMBER
Within the literature, there are limited high-order results concerning large Reynolds number flows under the influence of strong adverse pressure gradients, mainly due to the computational expense involved. The main advantage in employing high-order methodologies over standard second-order finite-volume solvers, relates to their ability to increase accuracy with a significantly lower number of degrees of freedom. In theory, this would permit Direct Numerical Simulation sort of analysis. Yet, there is still a significant computational cost involved. For this reason, an efficient approach to analyse such flows by means of a Nektar++ high-order Implicit Large Eddy Simulation is proposed. The flow conditions considered in this case cause a separation bubble to form with consequent turbulent transition. In particular, Tollmien-Schlichting instabilities trigger Kelvin-Helmholtz behaviour, which in turn cause the turbulent transition. The bulk of the study will be carried out with the incompressible flow solver, as it is assumed that compressibility effects are negligible within the boundary layer. An initial 2D analysis will be conducted to determine the necessary spatial resolution and whether it is possible to consider a subset of the overall simulation domain to reduce the computational expense. Once this will have been established, the 3D results will be achieved by Fourier expansion in the cross-flow direction. These results will prove the cost-effectiveness of the methodology, that could be used within an industrial setting with a limited turn-around time. Additionally, a comparison between the results achieved by means of the Nektar++ compressible flow solver in 2D and 3D will be provided, to assess any differences that may be present.
Marioni YF, Cassinelli A, Adami P, et al., 2022, DEVELOPMENT OF MACHINE-LEARNT TURBULENCE CLOSURES FOR WAKE MIXING PREDICTIONS IN LOW-PRESSURE TURBINES
In this work, a DNS – Machine Learning (ML) framework is developed for low-pressure turbine (LPT) profiles to inform turbulence closures in Reynolds-Averaged Navier–Stokes (RANS) calculations. This is done by training the coefficients of Explicit Algebraic Reynolds Stress Models (EARSM) with shallow artificial neural networks (ANN) as a function of input flow features. DNS data are generated with the incompressible Navier–Stokes solver in Nektar++ and validated against experiments. All calculations include moving bars upstream of the profile to capture the effect of incoming wakes. The resulting formulations are then implemented in the Rolls-Royce solver HYDRA and tested a posteriori. The aim is to improve mixing predictions in LPT wakes, compared to the baseline model, Wilcox’s k− ω SST, in terms of velocity profiles, turbulent kinetic energy (TKE) production and mixing losses. LPT calculations are run at Reynolds numbers spanning from ≈ 80k to ≈ 300k, to cover the range of aircraft engine applications. Models for the low and high Reynolds datasets are trained separately and a method is developed to merge the two together. The resulting model is tested on an intermediate Reynolds case. This process is followed for two computational domains: one starting downstream of the profile trailing edge and one including the last portion of the profile. Finally, the developed closures are tested on the entire profile, to confirm the validity of the improvements when the additional effect of transition is included in the simulation. This work explains the methodology used to develop ML-driven closures and shows how it is possible to combine models trained on different datasets.
Frey Marioni Y, de Toledo Ortiz EA, Cassinelli A, et al., 2021, A Machine Learning Approach to Improve Turbulence Modelling from DNS Data Using Neural Networks, INTERNATIONAL JOURNAL OF TURBOMACHINERY PROPULSION AND POWER, Vol: 6
Pepper N, Gaymann A, Montomoli F, et al., 2021, Local bi-fidelity field approximation with knowledge based neural networks for computational fluid dynamics, Scientific Reports, Vol: 11, Pages: 1-11, ISSN: 2045-2322
This work presents a machine learning based method for bi-fidelity modelling. The method, a Knowledge Based Neural Network (KBaNN), performs a local, additive correction to the outputs of a coarse computational model and can be used to emulate either experimental data or the output of a more accurate, but expensive, computational model. An advantage of the method is that it can scale easily with the number of input and output features. This allows bi-fidelity modelling approaches to be applied to a wide variety of problems, for instance in the bi-fidelity modelling of fields. We demonstrate this aspect in this work through an application to Computational Fluid Dynamics, in which local corrections to a velocity field are performed by the KBaNN to account for mesh effects. KBaNNs were trained to make corrections to the free-stream velocity field and the boundary layer. They were trained on a limited data-set consisting of simple two-dimensional flows. The KBaNNs were then tested on a flow over a more complex geometry, a NACA 2412 airfoil. It was demonstrated that the KBaNNs were still able to provide a local correction to the velocity field which improved its accuracy. The ability of the KBaNNs to generalise to flows around new geometries that share similar physics is encouraging. Through knowledge based neural networks it may be possible to develop a system for bi-fidelity, computer based design which uses data from past simulations to inform its predictions.
Pepper N, Montomoli F, Sharma S, 2021, Identification of missing input distributions with an inverse multi-modal polynomial chaos approach based on scarce data, Probabilistic Engineering Mechanics, ISSN: 0266-8920
Pepper N, Montomoli F, sharma S, 2021, Data fusion for Uncertainty Quantification with Non-intrusive Polynomial Chaos, Computer Methods in Applied Mechanics and Engineering, Vol: 374, ISSN: 0045-7825
This work presents a framework for updating an estimate of a probability distribution, arising from an uncertainty propagation using Non-intrusive Polynomial Chaos (NIPC), with scarce experimental measurements of a Quantity of Interest (QoI). In recent years much work has been directed towards developing methods of combining models of different accuracies in order to propagate uncertainty, but the problem of improving uncertainty propagations by considering evidence from both computational models and experiments has received less attention.The framework described here uses the Maximum Entropy Principle (MEP) to find an updated, least biased estimate of a probability distribution by maximising the entropy between the original and updated estimates. A constrained optimisation is performed to find the coefficients of a Polynomial Chaos Expansion (PCE) that minimise the Kullback–Leibler (KL) divergence between estimates, while ensuring that the new estimate conforms to constraints imposed by the available experimental measurements of the QoI. In this work a novel constraint is used, based upon the Dvoretzky–Kiefer–Wolfowitz inequality and the Massart bound (DKWM), as opposed to the more commonly used moment-based constraints. Such a constraint allows scarce experimental data to be used in informing the updated estimate of the probability distribution.
Montomoli F, Antorkas S, Pietropaoli M, et al., 2021, Towards digital design of gas turbines, JOURNAL OF THE GLOBAL POWER AND PROPULSION SOCIETY
- Author Web Link
- Cite
- Citations: 1
Friso R, Oliani S, Casari N, et al., 2021, Towards a machine learning based design for fouling of an axial turbine vane
Computational fluid dynamics (CFD) is increasingly used during the design phase of turbomachinery. Reducing the cost of such computations is one of the major challenges in the industrial field. The multi-physics phenomena and the multidisciplinary interactions needed for the design of the engine components are difficult to be faced with the classical design methods. In addition, the interest of manufacturers and operators of turbomachines in the increase in performance when degradation processes occur is growing. In the aeronautical sector degradation is among the most critical issues, as it can lead to the in-flight shutdown of the engine. In a bid to tackle these big problems, new design methods based on approximation techniques have been developed. These techniques are called surrogate models and are currently the most used design methods for the aerodynamic design of aircraft engines. In this work, the assessment of a surrogate surface built for the purpose of optimizing an HPT vane in degrading conditions is performed. Using machine learning and statistical techniques, a sensitivity analysis is conducted in order to reduce the problem dimensions. The results of the sensitivity analysis are used for a study of the surrogate, with the purpose of obtaining design guidelines when deterioration effects are considered in the design phase. The main outcome of this study is a map, that outlines the best design zone defined by the combination of the most influential parameters.
Cassinelli A, Mateo A, Montomoli F, et al., 2021, REYNOLDS SENSITIVITY OF THE WAKE PASSING EFFECT ON A LPT CASCADE USING SPECTRAL/HP ELEMENT METHODS
RANS methods will continue to be the backbone of CFD-based design, but the recent development of high-order unstructured solvers and meshing algorithms, combined with the lowering cost of HPC infrastructures, has the potential to allow for the introduction of high-fidelity simulations in the design loop, taking the role role of a virtual wind tunnel. A representative industrial low pressure turbine (LPT) cascade subject to wake passing interactions is analysed, adopting the incompressible Navier-Stokes solver implemented in the spectral/hp element framework Nektar++. The bar passing effect is modelled by leveraging a spectral-element/Fourier Smoothed Profile Method. The Reynolds sensitivity is analysed, focusing in detail on the dynamics of the separation bubble on the suction surface as well as mean flow properties, wake profiles and loss estimations. The main findings are compared with experimental data, showing remarkable agreement in the prediction of wake traverses and losses across the entire range of flow regimes.
Fadiga E, Casari N, Suman A, et al., 2021, Design considerations and numerical simulations of variable thickness scroll geometries, Pages: 1626-1636
Among the suitable expanders for small scale Organic Rankine Cycles (ORCs), scrolls are one of the most frequent solutions. The main advantages related to such machines are cost-effectiveness, high efficiency, silent operation, small size and high reliability. The great majority of scroll machines available on the market is designed using a spiral profile generated by the involute of a circle, resulting in a constant wall thickness. The main consequence of this approach is that an increase in the volume ratio has to be obtained with an increase in the scroll length, resulting in a machine augmented size. Currently, the state of the art in scroll expanders for small size ORC is to revert a compressor in its operation. Variable thickness geometries offer an alternative way to modify the expansion ratio of the machine, with no modifications on the scroll length. In this work, the authors have investigated a scroll geometry derived from a curvature radius parametrized with involute angle, firstly introduced by Gravesen and Henriksen (2001). In the beginning, the influence of the parameterization on the volume ratio has been deeply analyzed. At a later stage, a noteworthy geometry has been simulated using an extension of the OpenFOAM C++ toolbox for Computational Fluid Dynamics.
Pepper N, Crespo L, Montomoli F, 2021, Adaptive Learning for Reliability Analysis using Support Vector Machines, Pages: 242-249
A novel algorithm is presented for adaptive learning of an unknown function that separates two regions of a domain. In the context of reliability analysis these two regions represent the failure domain, where a set of constraints or requirements are violated, and a safe domain where they are satisfied. The Limit State Function (LSF) separates these two regions. Evaluating the constraints for a given parameter point requires the evaluation of a computational model that may well be expensive. For this reason we wish to construct a meta-model that can estimate the LSF as accurately as possible, using only a limited amount of training data. This work presents an adaptive strategy employing a Support Vector Machine (SVM) as a meta-model to provide a semi-algebraic approximation of the LSF. We describe an optimization process that is used to select informative parameter points to add to training data at each iteration to improve the accuracy of this approximation. A formulation is introduced for bounding the predictions of the meta-model; in this way we seek to incorporate this aspect of Gaussian Process Models (GPMs) within a SVM meta-model. Finally, we apply our algorithm to two benchmark test cases, demonstrating performance that is comparable with, if not superior, to a standard technique for reliability analysis that employs GPMs.
Voet L, Ahlfeld R, Gaymann A, et al., 2021, A hybrid approach combining DNS and RANS simulations to quantify uncertainties in turbulence modelling, Applied Mathematical Modelling: simulation and computation for engineering and environmental systems, Vol: 89, Pages: 885-906, ISSN: 0307-904X
Uncertainty quantification (UQ) has recently become an important part of the design process of countless engineering applications. However, up to now in computational fluid dynamics (CFD) the errors introduced by the turbulent viscosity models in Reynolds-Averaged Navier Stokes (RANS) models have often been neglected in UQ studies. Although Direct Numerical Simulations (DNS) are physically correct, obtaining a large enough set of DNS data for UQ studies is currently computationally intractable. UQ based only on RANS simulations or on DNS thus leads to physical and statistical inaccuracies in the output probability distribution functions (PDF). Therefore, three hybrid methods combining both RANS simulations and DNS to perform non-intrusive UQ are suggested in this work. Low-fidelity RANS simulations and high-fidelity DNS are combined to give an approximation of an output PDF using the advantages of both data sets: the physical accuracy via the DNS and the statistical accuracy via the RANS simulations. The hybrid methods are applied to the flow over 2D periodically arranged hills. It is shown that the Gaussian CoKriging (GCK) method is the best hybrid method and that a non-intrusive hybrid UQ approach combining both DNS and RANS simulations is possible, with both physically more accurate and statistically better PDF.
Tagarielli V, Gauch H, Bisio V, et al., 2020, Predictions of the transient loading exerted on circular cylinders by arbitrary pressure waves in air, Journal of Fluid Mechanics, Vol: 901, ISSN: 0022-1120
This study investigates the transient loading exerted on rigid circular cylinders by impinging pressure waves of arbitrary shape, amplitude, and time duration. Numerical calculations are used to predict the transient flow around the cylinder for wide ranges of geometric and loading parameters. An analytical model is developed to predict the transient loading history on the cylinder and this is found in good agreement with the results of the numerical calculations. Both models are used to identify and explore the different loading regimes, and to construct non16 dimensional maps to allow direct application of the findings of this study to the design of structures exposed to the threat of pressure wave loading.
Cavazzini G, Giacomel F, Ardizzon G, et al., 2020, CFD-based optimization of scroll compressor design and uncertainty quantification of the performance under geometrical variations, ENERGY, Vol: 209, ISSN: 0360-5442
Hammond J, Montomoli F, Pietropaoli M, et al., 2020, MACHINE LEARNING FOR THE DEVELOPMENT OF DATA DRIVEN TURBULENCE CLOSURES IN COOLANT SYSTEMS, ASME Turbo Expo
Pietropaoli M, Gaymann A, Montomoli F, 2020, Three-Dimensional Fluid Topology Optimization and Validation of a Heat Exchanger With Turbulent Flow, ASME IGTI 2020
Pepper N, Montomoli F, Giacomel F, et al., 2020, Uncertainty Quantification and Missing Data for Turbomachinery With Probabilistic Equivalence and Arbitrary Polynomial Chaos, Applied to Scroll Compressors, ASME IGTI 2020
Antorkas S, Montomoli F, Massini M, 2020, Topology Optimization of Gas Turbine Blades for Additive Manufacturing, GPPS 2020
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.