Abstract: We present a superparameterization of the ECMWF global weather forecasting model OpenIFS with a local, cloud-resolving model. Superparameterization is a multiscale modeling approach used in atmospheric science in which conventional parameterizations of small-scale processes are replaced by local high-resolution models that resolve these processes. Here, we use the Dutch Atmospheric Large Eddy Simulation model (DALES) as the local model. Within a selected region, our setup nests DALES instances within model columns of the global model OpenIFS. This is done so that the global model parameterizations of boundary layer turbulence, cloud physics and convection processes are replaced with tendencies derived from the vertical profiles of the local model. The local models are in turn forced towards the corresponding vertical profiles of the global model, making the model coupling bidirectional. We consistently combine the sequential physics scheme of OpenIFS with the Grabowski superparameterization scheme and achieve concurrent execution of the independent DALES models on separate CPUs. The superparameterized region can be chosen to match the available compute resources, and we have implemented mean-state acceleration to speed up the LES time stepping. The coupling of the components has been implemented in a Python software layer using the OMUSE multi-scale physics framework. As a result, our setup yields a cloud-resolving weather model that displays emergent mesoscale cloud organization and has the potential to improve the representation of clouds and convection processes in OpenIFS. It allows us to study the interaction of boundary layer physics with the large scale dynamics, to assess cloud and convection parameterization in the ECMWF model, and eventually to improve our understanding of cloud feedback in climate models. [Regional superparameterization in a Global Circulation Model using Large Eddy Simulations, Fredrik Jansson, Gijs van den Oord, Inti Pelupessy, Johanna H. Grönqvist, A. Pier Siebesma, Daan Crommelin, Under review (2018)]

Abstract: Transient molecular-continuum coupled flow simulations often suffer from high thermal noise, created by fluctuating hydrodynamics within the molecular dynamics (MD) simulation. Multi-instance MD computations are an approach to extract smooth flow field quantities on rather short time scales, but they require a huge amount of computational resources. Filtering particle data using signal processing methods to reduce numerical noise can significantly reduce the number of instances necessary. This leads to improved stability and reduced computational cost in the molecular-continuum setting. We extend the Macro-Micro-Coupling tool (MaMiCo) - a software to couple arbitrary continuum and MD solvers - by a new parallel interface for universal MD data analytics and post-processing, especially for noise reduction. It is designed modularly and compatible with multi-instance sampling. We present a Proper Orthogonal Decomposition (POD) implementation of the interface, capable of massively parallel noise filtering. The resulting coupled simulation is validated using a three-dimensional Couette flow scenario. We quantify the denoising, conduct performance benchmarks and scaling tests on a supercomputing platform. We thus demonstrate that the new interface enables massively parallel data analytics and post-processing in conjunction with any MD solver coupled to MaMiCo.

Abstract: Computational models have been widely used to predict the efficacy of surgical interventions in response to Peripheral Occlusive Diseases. However, most of them lack of a multiscale description of the development of the dis-ease, which it is our hypothesis being the key to develop an effective predictive model. Accordingly, in this work we present a multiscale computational framework that simulates the generation of atherosclerotic arterial occlusions. Starting from a healthy artery in homeostatic conditions, the perturbation of specific cellular and extracellular dynamics led to the development of the pathology, with the final output being a diseased artery. The presented model was developed on an idealized portion of a Superficial Femoral Artery (SFA), where an Agent-Based Model (ABM), locally replicating the plaque development, was coupled to Computational Fluid Dynamics (CFD) simulations that define the Wall Shear Stress (WSS) profile at the lumen interface. The ABM was qualitatively validated on histological images and a preliminary analysis on the coupling method was conducted. Once optimized the coupling method, the presented model can serve as a predictive platform to improve the outcome of surgical interventions such as angioplasty and stent deployment.

Abstract: Flow phenomena in porous media are relevant in many industrial applications including fabric filters, gas diffusion membranes, and biomedical implants. For instance, nonwoven membranes can be used as filtration media with tailored permeability range and controllable pore size distribution. However, predicting the structure-property relations that arise from specific porous microstructures remains a challenging task. Theoretical approaches have been limited to simple geometries and can often only predict the general trend of experimental data. Computer simulations are a cost-effective way of validating semi-empirical relations and predicting the precise relations between macroscopic transport properties and microscopic pore structure. To this end, multiscale simulation techniques have proven particularly successful in solving numerically the coupled partial differential equations for the complex boundary conditions in porous media. In this talk, I will present simulations of multiphase flow in fibrous porous media based on a multiphase lattice Boltzmann model for water droplets in oil. We study the effect of fibrous structures and their surface properties on the coalescence behavior of water droplets. We will discuss how the insights can be used to design optimized materials for diesel fuel filters and other filtration devices.

Abstract: In this paper, a parametric study has been conducted to evaluate the effects of frequency and duration of the short burst pulses during pulsed radiofrequency ablation (RFA) in treating chronic pain. Affecting the brain and nervous system, this disease remains one of the major challenges in neuroscience and clinical practice. A two-dimensional axisymmetric RFA model has been developed in which a single needle radiofrequency electrode has been inserted. A finite-element-based coupled thermo-electric analysis has been carried out utilizing the simplified Maxwell’s equations and the Pennes bioheat transfer equation to compute the electric field and temperature distributions within the computational domain. Comparative studies have been carried out between the continuous and pulsed RFA to highlight the significance of pulsed RFA in chronic pain treatment. The frequencies and durations of short burst RF pulses have been varied from 1 Hz to 10 Hz and from 10 ms to 50 ms, respectively. Such values are most commonly applied in clinical practices for mitigation of chronic pain. By reporting such critical input characteristics as temperature distributions for different frequencies and durations of the RF pulses, this computational study aims at providing the first-hand accurate quantitative information to the clinicians on possible consequences in those cases where these characteristics are varied during the pulsed RFA procedure. The results demonstrate that the efficacy of pulsed RFA is significantly dependent on the duration and frequency of the RF pulses.