S_SESSIONS_BROWSE_TITLE

Dispersed multiphase flows occur frequently in Engineering, and present one of the most complex modelling challenges in CFD. Typically this is done through an Eulerian two-fluid model derived from conditional averaging of the basic Navier-Stokes equations to give separate equation sets for the interpenetrating primary (continuous) and secondary (particulate) phase. In the most advanced formulations this involves solving two complete sets of the NSE, treating the dispersed phase as a continuum. For fluid-particle flows this often includes the solution of the granular temperature equation to account for the particles fluctuation energy. Recent work has indicated that there exist two contributions to the overall fluctuation energy. In this work we present a full model for a Reynolds-Averaged Two Fluid Model (RA-TFM) including continuity, momentum, and turbulence equations derived for both phases, and a granular transport equation, implemented within OpenFOAM. Additionally, a velocity-pressure model and wall boundary conditions are derived in order to account for the near-wall behaviour and are shown to be an effective approach validating against benchmark experimental data. Finally, the primary variables with the RA-TFM are solved in a recently derived block-coupled solution algorithm. The results show substantial improvements in solution time and convergence.

Modern continuous casters, like the new Continuous Caster 23 in Tata Steel, The Netherlands, are equipped with an electromagnetic flow controller that consists of a top and bottom Direct Current (DC) magnetic field and a top Alternating Current (AC) magnetic field. The creation of optimal flow conditions in the mould is not only achieved by braking the fluid (DC field) but also by stirring the flow close to the meniscus (AC field). It is believed that the circulating flow increases the flow velocity close to the solidifying shell in order to remove inclusions and (Argon) bubbles. The stirring improves the homogenisation of the temperature close to the solidifying shell and prevents longitudinal cracks. Optimal flow conditions are contributing to a final product quality with a minimal contamination of inclusions and slag particles. Computational Fluid Dynamics (CFD) can help in finding the optimal AC and DC magnetic field settings given a certain Argon supply, casting width and speed.

In order to model the fluid flow inside continuous casters employed at Tata Steel an OpenFOAM solver has been developed during the last decade. It can handle electromagnetic braking and stirring, solidification and Argon flow. Or, in technical terms, it is a thermal, incompressible MHD solver with a solidification model [1]. The free surface at the meniscus is not included but modelled as a slip boundary condition. Argon is modelled using Lagrangian Particles Tracking (LPT). Turbulence is modelled using Large Eddy Simulation and the turbulent viscosity is calculated with the dynamic Smagorinsky model [2].

During the presentation the physical model and OpenFOAM implementation details will be explained. Some validation cases are presented and real plant simulations are shown.

[1] Z. Liu et al. “Analysis of transient and entrapment of particle in continuous casting mold”, ISIJ Int., Vol. 54 (2014), pp. 2324-2333. [2] D.K. Lilly, “A proposed modification of the Germano subgrid-scale closure method”, Phys. Fluids A, Vol. 4 (1991), pp. 633-635.

Many physical phenomena involved into industrial processes concern a free surface behavior. As an example a flow below meniscus in the continuous casting mold is considered here. Typical flow pattern represents a double roll structure initiated by a liquid steel jet been fed from the submerged entry nozzle ports, which hits the narrow wall of the mold and rebounds in both casting and upward directions. Focusing on the upper part and related upper roll is very important for the continuous casting process, since it characterizes the stability of the free surface being crucial for the slag entrapment preventing and defines amount of the superheat transported to the slag band to constantly melt the flux required for the lubrication between the mold and solidifying shell. Experimental work is expensive, done in harsh conditions and is accompanied by a long list of difficulties. Thereby last decades the computational methods took a leading role.

There are some common approaches to model a free surface, which are typically split to those involving interface tracking with a marker phase or using geometrical reconstruction methods to advect the phase boundary; others so-called Euler-Euler methods are focused on predicting each phase separately not carrying about exact interface position which can be later reconstructed from the phase volume fraction distribution.

On the one hand, the multiphase methods are more accurate, but on the other hand they are highly demanding to the computational resources. Thereby a single phase simulations remain a great alternative if the free surface “deformation” is not dramatic and does not generally define the main flow. However some assumptions concerning immobility of the free surface cause inaccuracies in the momentum transfer, turbulence properties and pressure field modelling results close to the interface.

In the presented study a verification of the single phase approach of the subsurface flow at the interface for the liquid slag / steel melt system in the continuous casting process is verified with the two phase volume-of-fluid (VOF) method using standard MULES and relatively new geometric VOF (iso-advector) interface tracking. The presented study is done for a range of Reynolds numbers and also includes comparison of the turbulence modeling using laminar, Reynolds-averaged Navier-Stokes (RANS) and large eddy simulations (LES) approaches.

The complex multiphysics phenomena of electrohydrodynamic has been studied for more than a century and has led to innovations in a range of sectors; electrospinning for fibre production, drug delivery mechanisms, nanostructure deposition of particles/ion and in application to electrospray thrusters for electric propulsion. In order to analyse the performance of numerical schemes on these electrohydrodynamic problems numerically, a simplified version can be used initially to see whether the numerical method provide accurate results. The setup for a simplified version comprises of an external electric field applied to a droplet which is suspended in another immiscible fluid. The numerical scheme used for this paper to visualize whether the droplet prolate or oblate is based on the Volume-of-Fluid (VOF) method. Since the deformation mainly relies on the electric field strength and the fluid properties, this paper focuses on the effect of permittivity ratio and conductivity ratio on droplet deformation. In addition, the effect of permittivity on the surface charge is examined since they are related to the electric force. Lastly, the numerical results are validated by comparing to the previous analytical and experimental results.

Rogue waves are short-lived large amplitude irregular waves that randomly occur in oceans, which significantly impacts the stability of marine structures. Previous studies have focused on rogue wave loads on fixed structures while the impact on floating structures has barely been understood. Based on OpenFOAM, this paper simulates the motions of a floating structure induced by rogue waves. Code development includes the generation of a rogue wave and its incorporation with the overset mesh technique to allow a large displacement of the structure. Validation of the developed code is first conducted by comparing the motions of a floating body in regular waves. Investigations are then performed on the motions of the same floating body, while induced by rogue waves. The behaviour of the floating body was analysed in detail; for example, the roll motion damping after the wave hit. In addition, the result of the floating body is also compared with its fixed counterpart, and it is found that the loading on a floating body is much smaller than that of the fixed one, which is attributed to a softening effect of the body motions on the incoming wave.

The ocean environment is active, and many natural phenomena impact the character of turbulence, including buoyancy, wind shear, breaking waves, near-free-surface damping, bubbles, and Langmuir circulations. As a result, oceanographic turbulence is generally inhomogeneous, non-stationary, and anisotropic. Simulations in such an environment must also consider the transport of both the temperature and the salinity. In this talk, we will describe the implementation of anisotropic, buoyant turbulence models in OpenFOAM. The implementation includes several distinct stress transport and algebraic models, and separate runtime selectable models of the turbulent transport of temperature and salinity. Additionally, the OpenFOAM implementation of several sources of environmental forcing will be outlined, including solar radiative heating of the ocean mixed layer. These simulation tools are then applied to the replication of a number of laboratory experiments of stratified flows, and to the simulation of an active ocean mixed layer. Finally, the application of the anisotropic turbulence models to engineering problems in the ocean environment will be discussed.

Floating structures are prominently present in coastal and offshore regions, e.g. ships, pontoons, barges and pipelines. Innovative structures such as wave energy converters and complex wind turbine foundations are currently being installed for renewable energy production. These structures need special installation vessels such as a heavy lift jack-up vessel or a heavy lift floating installation vessel. The workability of these vessels depends on the wave, current and wind loading. Not only the forcing is of large importance for the installation of structures but also the response to the environmental loads needs an accurate quantification.

Nowadays, simplified radiation-diffraction models such as linear potential flow solvers based on boundary element methods (BEM) are used to simulate vessels. These models are not capable in resolving nonlinear, viscous and turbulent effects and waves breaking on a vessel. Computational Fluid Dynamics (CFD) is selected to overcome this problem. Simulations of the two-phase flow field are performed by solving the incompressible RANS equations, with a conservation equation for the Volume of Fluid (VoF). This fluid solver is coupled with a motion solver using a partitioned fluid-structure interaction algorithm to obtain a converged solution between the flow field and motion state during every time step of the transient simulation. The mesh motion is organised by either using mesh morphing, sliding meshes or an overset method as implemented in OpenFOAM.

In this work, the installation of a monopile foundation from the offshore heavy lift DP3 installation vessel Orion from the DEME group is studied using OpenFOAM. The mesh motion is performed by using the overset mesh library available in OpenFOAM-v1812. Both waves and an uniform current are generated and absorbed at the boundaries of the computational domain. Numerical wave gauges are used to study the complex wave field amplification in the zone between the vessel and the monopile. Also the motion of the vessel in two degrees of freedom (heave and roll) is tracked and is of interest to define the workability of the vessel for installing monopile foundations.

A function object was implemented into OpenFoam, able to compute fluid structure interaction in one and two degree of freedoms. The function object was coupled with an incompressible flow solver, which is able to compute laminar flows as well as turbulent flows. Applying this function object, we investigated the fluid structure interaction of a low mass ration cylinder (the ratio between the mass and the displaced fluid). Two cases were simulated, the first case was an oscillating cylinder in one degree of freedom (i.e. y direction), and the second was in two degree of freedom (i.e. x and y directions). The shapes of the wake behind the cylinder at different motion positions and time steps were captured. Meanwhile, the vortex shedding frequencies were accurately estimated, which agreed with previous studies. The typical eight trajectory shape of cylinder motions in two degree of freedom was also well captured. Additionally, a discretization uncertainty
analysis was conducted, which verified that our chosen grid and time step size were small enough to obtain reliable simulations.

During the last decades, a lot of effort has been done concerning the appropriate modelling of the complex dynamic nature of floating offshore wind turbines (FOWT). Fully-coupled aero-hydro-servo-elastic engineering tools, of which FAST is the most well-known, have been developed. However, as the hydrodynamic part of these models is primarily based on a potential-flow model, most of these models are not able to evaluate hydrodynamic loads above second-order. As a result, the higher-order nonlinearity, which is distinctive for extreme waves and ultimately important for survival states and fatigue loads, cannot be appropriately accounted for.

Our research aims to fill this gap by studying the dynamic response of a FOWT of the spar-buoy type to extreme waves through OpenFOAM. Therefore, the tower flexibility and mooring, which are important points in assessing the structural integrity of a FOWT subjected to extreme wave events, need to be adequately taken into account.

This paper presents the set-up and first results obtained for this FOWT model. The model utilizes the NavalHydro package for foam-extend with added functionality for mooring and tower flexibility. Instead of utilizing a two-way coupling between a structural model, modelling the wind turbine tower, and the six-degrees-of-freedom rigid body motion solver, modelling the spar-buoy motion, the whole system is solved as one multibody system matrix.

In further work, thorough validation by wave tank experiments will be done and the impact of extreme waves on the overall structural integrity will be tackled.

One of the most interesting properties of a battery is its open circuit potential. From a macroscopic point of view, the electric potential jumps at each electrode-electrolyte interface. Computing these potential jumps is possible using the Nernst equation; the difference of both jumps gives the open circuit potential. The challenge
in modelling three-dimensional potential distributions in electrochemical cells is how to include the internal discontinuities of the electric potential.

The talk will give an overview on simulating electric potential and current distributions in the frame of the finite volume method. Special attention will be paid to modelling internal discontinuities of the electric potential, as they appear at the electrode-electrolyte interface in fuel cells, electrolysers and batteries. The theoretical background will be explained and illustrated with simple examples.

In the present study we will perform computational analyses for a specific wing where shock-boundary layer interaction takes place. We will carry out computations for both 2-D and 3-D flows by using two different compressible Navier-Stokes solvers of OpenFOAM, rhoCentralFoam and sonicFoam along with different turbulence models. After validating the solver by using the experimental results available in the literature, we will conduct a parametric study for the complete wing, where we systematically vary the Mach number and the angle of attack. After obtaining the aerodynamics forces exerting on the wing for the considered cases, we will examine the features of the flows in terms of the locations and strengths of shocks and separations. In order to enhance the aerodynamics performance of the wing, we are planning to put passive flow control techniques into practice such as; placing bump and/or cavity near the interaction point. Then, we are going to provide both qualitative and quantitative comparisons for the flows over the original wing and the one with the proposed passive control surfaces. We will use snappyHexMesh to generate high quality meshes and perform all computations at National High Performance Computing Center, UHEM.

Keywords: Shock wave, boundary layer, airfoil, OpenFOAM, passive flow control

Acknowledgement: This study is funded by TUSAŞ with a project number of TM-2101.

Detailed simulations of nuclear power plants or facilities are required for accomplishing comprehensive operational and safety related analyses of these. System codes have long been employed for this purpose, with the disadvantage of the imposed simplifications as they compute in 1D and lack detailed spatial resolution. On the other hand, it is practically impossible to apply a CFD (Computational Fluid Dynamics) software to numerically predict the behaviour of, e.g., the whole primary circuit of a pressurized water reactor because the required computational resources would be enormous. Combining the advantages of both approaches into a coupled-code can provide the suitable numerical tool for such analyses.

A coupled-code ATHLET-OpenFOAM solver was developed for the investigation of coolant mixing in the primary circuit and in the reactor pressure vessel (RPV) of pressurized water reactors, which is an important topic of reactor safety assessments. Two different coupling implementations, i.e. under-relaxation and a quasi-newton method were utilised and implemented for coupled-code simulations. The newly developed solver was built on the standard OpenFOAM solver buoyantBoussinesqPimpleFoam with the Boussinesq approximation.

The ROCOM tests PKLIII T1.1 und 1.2 were used to validate the coupling ATHLET-OpenFOAM solver predictions. In these tests the mixing of the injected coolant was monitored for its complete turnover in the test facility. Therefore, it is necessary to consider both the primary circuit and the RPV. With the coupled ATHLET-OpenFOAM code, the RPV was modelled in 3D by OpenFOAM and the primary circuit in 1D by ATHLET. The results were compared with both, the experiments and the ATHLET stand-alone predictions for which the newly implemented 3D module was used. The simulation results show a good agreement with the experimental data, indicating that the numerical tools can reproduce the mixing processes in the RPV.

Conjugated Heat Transfer (CHT) simulations allow the prediction of complex interactions between fluid and solid mediums. They are applied to a wide range of research and industry domains [1].
Our application is the optimization of heat transfer between heat sinks and a cooling fluid, used to extract heat from server infrastructure.

Adjoint methods allow the optimization of problems with a high number of parameters, using sensitivity information. We frame this optimization problem as a mapping from n input parameters to a (low) m dimensional output functional. Compared to classical approaches of sensitivity generation, e.g. finite differences, a significant improvement in run time can be achieved. For adjoint methods the complexity of the sensitivity generation scales with the output dimension m, instead of the parameter dimension n.

In our presentation, we will show how we adapted our existing discrete adjoint workflow [2] to the OpenFOAM solvers chtMultiRegionFoam and chtMultiRegionSimpleFoam, working towards topology and shape optimization for complex CHT flows. The topology and shape sensitivities are generated efficiently and accurately using algorithmic differntiation. The presented discrete adjoint framework is available as open source from https://stce.rwth-aachen.de/foam.

Referneces:

[1] A. Dorfman. Conjugate problems in convective heat transfer. CRC Press, 2009.

[2] M. Towara, M. Schanen, and U. Naumann. MPI-parallel discrete adjoint OpenFOAM. Procedia Computer Science, 51:19 – 28, 2015. 2015 International Conference on Computational Science.

During the last years GRS has been developing a coupling interface between different OpenFOAM solvers and the system code ATHLET. This coupling allows to split up a flow domain between the two solvers. Then, each solver can be applied to that part of the flow domain, for which it is better suited. The OpenFOAM solver simulates the flow domains where e. g. 3D and turbulent flow phenomena are important. The 1D system code ATHLET efficiently simulates long 1D pipes and the coupling to heat structures like heat exchangers.

The coupling of these codes is numerically challenging, especially if closed loops are to be simulated or if the number of coupling interfaces exceeds two or three. A first implementation to exchange the field values at the coupling interfaces applied under-relaxation to increase numerical stability. In a second step an approach based on a quasi-Newton method was implemented. With this method the response of one simulation code to changes at the interfaces are estimated, which allows to find a common solution faster and in a more stable way.

This new method was applied to simulate the flow in the primary circuit of a liquid sodium cooled fast reactor. ATHLET simulated the flow in the reactor core, in the internals of the intermediate heat exchangers (IHX) and the pumps, the vessel cooling circuit, the secondary and the third cooling circuit, and the flow in the cover gas volume. OpenFOAM was used to simulate the flows in the upper hot and the lower cool plenum. Altogether, in the simulation 17 coupling interfaces were applied. The flow in the primary circuit passes several coupling interfaces in the main flow path between core and upper plenum, upper plenum and IHX, IHX and lower plenum, and lower plenum and pumps (back to the reactor core). This requires special care to get numerically stable solutions. Using the new quasi-Newton method, the flow in the primary circuit was simulated successfully.

Analysing the interaction between a vibrating immersed solid body and its surrounding fluid, also known as fluid-structure interaction (FSI) analysis, constitutes a broad field of research with many applications in science and engineering. Among these applications, the study of the dynamic response of submerged structures in flowing water presents specific challenges related to the non-negligible density of the fluid, which induces a significant added-mass effect. In such systems, mutual effects of the solid on the fluid movement and of the fluid on the solid response must be considered in order to accurately predict the dynamic behavior of the system as a whole. Numerical simulations of these systems hence impose a tight coupling of the fluid and solid simulations through bidirectional transfer of information through the interface between the domains.

This presentation explores the development and validation of an integrated FSI solver based on OpenFOAM fluid and structural analysis capabilities to perform FSI simulations using a fully coupled formulation. In order to validate the developed solver, a coupled validation case is discussed, and results are compared to references data obtained using a commercial solver and results published in literature. The validation test case consists of a plate fixed in a closed box and excited by an initial pressure force. The surrounding fluid damps the displacements of the plate and oscillations disappear. A second test case is also simulated, and results are compared to experimental measurements realized by Andritz. This test case consists of a profiled plate placed in a water tunnel and excited using piezoelectric actuators. Measurements were carried out for several flow conditions and three distinct profiles with varying aspect ratios. Two types of validations are performed, first with experimental data and also with previously computed results using a commercial solver.

Since its inception, OpenFOAM has included solvers for solid mechanics based on the finite volume method, for example, solidDisplacementFoam. The adopted solution procedures were inspired by a number of prior developments within the finite volume computational solid mechanics field. An in-depth review of the finite volume method for solid mechanics can be found in a recent preprint article by Cardiff and Demirdžić. The current work examines a number of solid mechanics cell-centred discretisations and solution algorithms implemented in OpenFOAM, in terms of accuracy, efficiency and robustness, with the aim of providing insight into the relationship between the differing approaches.

A variety of finite volume formulations have been developed for the solution of solid mechanics problems, with differences in terms of discretisation, solution methodology and overall philosophy. Approaches can be classified in a number of ways, for example, based on the: spatial distribution of the primitive variables: cell-centred vs vertex-centred vs staggered- grid; solution algorithm: implicit (solution of a linear system) vs explicit (matrix-free); or stabilisation approach: Rhie-Chow vs Jameson-Schmidt-Turkel vs Godunov methods.

Considering boundary value problems governed by the conservation of linear momentum in Lagrangian integral form, a variety of cell-centred discretisations and related solution methodologies may be employed. For example, formulations may be based on: an unknown displacement vector solved using an implicit segregated or coupled approach; an unknown displacement vector and unknown pressure solved using an implicit segregated or coupled (or partly-coupled) approach; or an unknown displacement vector solved using a fully explicit matrix-free approach.

In the current study, a number of these formulations and their implementations will be examined on a variety of benchmark cases.

Products extruded from a polymer melt are widely used in packaging and infrastructure applications. The key to ensuring product quality is the usage of finely tuned extrusion equipment, including both the feedscrew and the extrusion die.

Recent research has shown that there is a significant thermal interaction between the polymer melt and the extrusion die. In some applications, the high pressure of the extrusion process leads to a geometric deviation in the extrusion die. Both effects can cause product quality to deteriorate, which is why integrative simulation techniques are used to predict them. These techniques link the simulation model of the fluid melt with a simulation model of the solid die.

However, for efficient (automatic) geometric optimization to take place, the effort required to set up a new calculation consisting of multiple geometries and multiple physical models must be reduced. We therefore present a solver set-up that enables quick geometric updates in optimization problems concerning flows in polymer extrusion equipment. This novel method captures both thermal and mechanical interactions between the polymer melt and the metal of the equipment.

Based on an Immersed Boundary approach, the solver calculates temperatures, mechanical stresses etc. on the whole grid, meaning that the distinction between fluid and solid only is made by a variation of the local properties of flow resistance, thermal diffusivity, elastic modulus, etc.. Therefore, creating a new geometry is just a matter of setting up new fields instead of creating a new mesh.

In this work, the solver is benchmarked against both analytical solutions and conventional, non-immersed numerical calculations and found perform well.

A finite volume solver for predicting the linear elastic behavior of an incompressible elastic solid is proposed in [I. Bijelona, I. Demirdzic and S. Muzaferija. A finite volume method for incompressible linear elasticity. Computer Methods in Applied Mechanics and Engineering, 195(44-47):6378-6390, 2006.], where the incompressibility constraint is enforced by employing the hydrostatic pressure as an additional variable and SIMPLE based segregated solution procedure is used to solve resulting set of coupled algebraic equations. In this paper, the above described solver is extended by implementing a block-coupled solution procedure similar to one used for incompressible fluid flow [T. Uroic and H. Jasak. Block-selective algebraic multigrid for implicitly coupled pressure-velocity system. Computers & Fluids, 67:100-110, 2018.]. Although with such approach the substantial improvement of the computational efficiency is obtained, the major problem of finite volume stress analysis related to strong inter-displacement-component coupling is still present.

OpenFOAM provides a rich arsenal of single-physics solvers, while other software projects also offer a wide range of solvers for structural dynamics or heat transfer. Moreover, packages for special applications, such as nuclear reactor safety, hemodynamics, or flood simulations need to integrate 3D flow phenomena into their workflow, which is often built around 1D or 2D models.
The coupling library preCICE for partitioned multi-physics simulations can bring together different solvers, in a minimally invasive way. It allows them to communicate via MPI ports or TCP/IP sockets, it maps the boundary values using advanced methods such as RBF, and it couples them with Interface Quasi-Newton algorithms that accelerate the convergence. Its API is being used in a variety of well-known or in-house solvers, while official, user-ready adapters are provided for open-source packages such as OpenFOAM. The official OpenFOAM adapter supports conjugate heat transfer and fluid-structure interaction out-of-the box, while it was recently extended to also support fluid-fluid coupling.
This talk will present the current status of the coupling library preCICE and its OpenFOAM adapter to the OpenFOAM community, and it will discuss our current research into fluid-fluid coupling and geometric multiscale coupling.

This work is concerned with the study of the flow around a stream tidal turbine andthe wake behavior behind it. Two methods are used in this study, the first is the Multiple Reference Frame (MRF) method and the second is the Actuator Disk Method (ADM). The effects of the tidal turbine on the fluid flow in the radial and in the longitudinal axis are studied, as well as the forces on the turbine blade. Different turbulence intensity conditions are investigated. The computational effort for the two
methods is approximately the same.

Heat transfer is a known challenge in many industrial as well as everyday applications. Commonly the heating or cooling processes are required to be implemented in a compact manner due to various process and/or design constraints. To achieve an augmentation in the heat transfer rate in compact heat exchangers, vortex generators (VGs) such as winglets, wings, ribs or grooves are frequently employed. VGs disrupt the formation of boundary layers by introducing longitudinal as well as transverse vortices and cause enhanced flow mixing, thereby increasing the heat transfer rate at the cost of pressure loss.
In this work a numerical investigation of the influence of winglet configurations in a rectangular duct on the flow and heat transfer characteristics is carried out. The winglets considered in the study are delta and three new winglets namely arrow, delta-cut and X winglet. Nandana and Janoske studied the heat transfer characteristics of a single arrow, delta-cut and X winglet placed in a rectangular duct for laminar flow (Reynolds number of channel flow, Re = 300 − 1200) regime in a previously examined study. Here the study is extended to turbulent regime for Re ranging from 4000 to 12000 and for multiple winglets in the duct. The arrangements of multiple winglets are defined in terms of longitudinal distances between each other. For all the simulations, the flow attack angle (α) and the inclination angle (β) of winglets are chosen to be constant 45° and 90°, respectively. The study focuses on the influence of various aforementioned winglet and flow configurations on the vortex expansions and flow structure as well as heat transfer rate, pressure loss and thermal enhancement factor. The results are compared with the heat transfer rate in a plain rectangular duct without winglets. The RANS k-ω-SST model is employed to calculate turbulence in the flow domain.

At present, the Portuguese footwear industry is considered one of the most important internationalized sectors of the Portuguese economy. Recent developments in the shoe market induced some adaptations on this industrial sector, in order to maintain its reference in the footwear industry, with high technical and innovative solutions, able to offer high‑quality products [1]. To achieve this goal, the Portuguese footwear industry has been investing in research and technological development of new materials, products and processes, to obtain better solutions that result in high end added-value products [2].

Toecaps are one of the most important components in safety footwear, but the currently available ones have a significant contribution to the total part weight, mainly because they are made of steel, due to its high mechanical resistance. This limitation has been motivating efforts to replace steel toecaps by polymeric ones, since they are lighter, insulated and insensitive to magnetic fields. Nevertheless, polymeric toecaps require larger volumes, when compared with their metallic counterparts, which has a negative impact in aesthetics and design.

This work is part of a PhD project, which aims the development of a new composite toecap design with improved mechanical behavior. The toecap design should be defined with the support of the computational modelling open source library, OpenFOAM. In order to accomplish this goal, two different case studies that simulate the laboratory mechanical test conditions will be prepared, one regarding compressive behavior (up to 15 kN) and another a free-falling weight with an impact energy of 200 J [3]. In this work, the free-falling weight case is presented, being the solid mechanics toolbox, developed by P. Cardiff [4-5], the solver selected for modelling purposes. After implementation and validation, different toecap designs will be simulated aiming to identify the geometry that accomplishes the toecap specifications, with the minimum possible amount of material.

References:

[1] APICCAPS, “Statistical Report, Portugal - Footwear, Components and Leather Goods.”, 2017.

[2] CTCP, “Inovação”, 2019 [Online], Available: https://www.ctcp.pt/inovacao/inovacao.asp?op=0&idmp=Ng==&idms=NDU=&lista=todos [Accessed: 17-Jun-2019].

[3] “Personal protective equipment – Safety Footwear”, prEN ISO 20345, 2011.

[4] P. Cardiff et al. (2016), DOI: 10.1002/nme.5345

[5] P. Cardiff et al. (2018), arXiv:1808.10736.

Nuclear power plant operators should conduct in-service testing (IST) to verify the safety functions of safety-related pumps and valves and to monitor the degree of vulnerability over time during reactor operation. The system to which the pump and valve to be tested are installed has various sizes of orifices for flow control and decompression. Rapid flow acceleration and accompanying pressure drop may cause cavitation inside the orifice, which may induce many detrimental effects, such as vibration and damage of the piping system integrity.

Currently, there is no Computational Fluid Dynamics (CFD) software licensed from nuclear regulatory agency in Korea, but license applicants use specific CFD software to revise licensing documents for facility improvements and submit analysis results as licensing documents. In case of CFD software currently being used for licensing, modeling and numerical techniques for single-phase flow have reached a certain level, but significant uncertainty may be involved in the analysis results for especially multi-phase flow due to the technical limitations of CFD software. Therefore, from the nuclear regulatory perspective, it is necessary to perform the systematic assessment and prepare the domestic regulatory guideline for checking whether valid CFD software and the numerical modeling is used for IST-related problems. Under these circumstances, CFD simulation for the cavitating flow inside a square-edged orifice was conducted using OpenFOAM v6 and the predicted results were compared to those of commercial CFD software, ANSYS CFX R.18.

Acknowledgement

This work was supported by the Nuclear Safety Research Program through the Korea Foundation Of Nuclear Safety (KOFONS) using the finiancial resource granted by the Nuclear Safety and Security Commision (NSSC) of the Republic of Korea (No. 1805007)

CFD Simulation is difficult for small and medium-sized enterprises(SMEs) to apply to their process. Nevertheless CFD simulation must be an efficient way to reduce the cost of developing SME products and processes. That is why 'High-performance Engineering MOdeling & Simulation(HEMOS)' has been developed. HEMOS is a general-purpose CAE software which can conduct both structural and CFD analyses. Various open-source libraries are used to develop HEMOS. Especially, we used OpenFOAM as a CFD solver. Preprocessor and postprocessor have been also developed on HEMOS. Preprocessor has been developed with OpenCasCade library which is a CAD kernel based on OpenGL. Postprocessor for structural analysis has been developed with self-developed code which is also based on OpenGL, while Paraview is utilized for postprocessing of CFD analysis. CFD part of HEMOS is called HEMOS-Fluid. HEMOS-Fluid has some features for CFD analysis. It is related with some solvers in OpenFOAM which consists of simpleFoam, porousSimpleFoam, pisoFoam and pimpleFoam. In addition, HEMOS-Fluid deals with MRF functions, too. The main purpose of developing HEMOS is to supply very easy and user-friendly CAE software for SMEs engineers. As a result, workflow system has been developed to use it easily for beginners. And the workflow of HEMOS-fluid is designed with standard process of CFD. Few inputs are needed to make a complete case to have a simulation. Most of mandatory inputs do not need to be determined in HEMOS. They are completely automatically determined in HEMOS. HEMOS-Fluid can generate hybrid meshes on arbitrary geometry. HEMOS-Fluid only supports STEP and IGES formats of CAD file. It also provides fluid domain extraction function which can make inner fluid domain or outer fluid domain. Fluid domain extraction algorithms were self-developed based on OpenCasCade functions. HEMOS-Fluid generates OpenFOAM input files and their file structure automatically with user inputs from graphic user interface (GUI). This paper presents how HEMOS-Fluid is designed and how it works by process.

Fine bubble diffusers are often used for aeration systems in large wastewater treatment works. The bubbles created are a source of momentum for mixing and provide oxygen required for the micro-organisms to feed on the organic matter and nitrogen in the wastewater. The energy needed to aerate the system can account for 25-60% of the total energy consumption; therefore, large savings can be gained with an energy efficient aeration system. Biokinetic models have been developed (ASMs, ADMs, etc.) that describe the growth and decay of micro-organisms and their dependence on oxygen in wastewater treatment systems and are a useful tool for understanding the biological processes. However, a critical assumption made by these models is that the reactors are well-mixed such that the parameter concentrations are uniform throughout the reactor. If we reduced the aeration, this would reduce the amount of available oxygen for the reactions and also affect the mixing flow regime. In many cases, the optimal reactor design would be one that uses the least amount of energy to mix while not hindering the desired reactions. By coupling the biokinetic growth models with the hydrodynamics using computational fluid dynamics, we can analyse the impact of the hydrodynamics on the biokinetic growth models in the case of reduced aeration. These enhanced predictions can aid in improved operating conditions of the aeration systems, leading to more efficient processes which reduce the overall cost of the systems.

To achieve this, a multiphase Eulerian-Eulerian approach has been adopted which includes the bubble dynamics and oxygen mass transfer to simulate the hydrodynamics under a range of initial conditions. An experimental large scale transparent aeration reactor has been modelled that is a 2/3 replica of a working pilot-plant. Ultrasonic techniques, dissolved oxygen probes and other experimental methods have been used to take measurements to validate the CFD model which has shown that it can reliably reproduce the hydrodynamics and oxygen distribution over time. The biokinetic models can then be coupled in to the model to investigate the affect of the initial aeration conditions on the biological performance of the pilot-plant reactor.

Since 1886, the Hall-Héroult process developed by Charles Martin Hall and Paul Héroult is the main process for the industrial production of aluminum. Liquid metal instabilities in aluminium cells are a common problem due to high currents (100 – 800 kA) required for production. The induced magnetic field from the current carrying busbars together with high current in liquid metal leads to high Lorentz force. This force results in the movement of the liquid, which can result in a Metal Pad Rolling (MPR) instability. Since the experimental investigation of MPR inside a Hall-Héroult cell is impractical, numerical simulations are essential to examine the fluid flow and magnetic field.

This work focuses on the development of a multi-phase, multi-region solver to predict liquid metal instabilities. The solver is implemented in the open source framework OpenFOAM®. The volume of fluid method with a phase-fraction based reconstruction approach is used to solve the multi-phase system. The magnetic field is implemented using three different methods, the induction equation, Biot-Savart law and the magnetic potential formulation. These three methods were compared in terms of calculation speed and accuracy for a single-phase test case. Additionally, the performance of the magnetic potential formulation is improved by the usage of the Barnes-Hut method, to reduce the original problem from n² to n log(n), where n is the number of computational elements. The solver is validated with an experiment from literature. While the angular frequencies obtained from the simulation are in good agreement with the experiment, further work is needed for the validation of the amplitude.

The OpenFOAM service begin in May at KISTI NURION supercomputer. NURION has a total 570,020 cores consisting of Intel KNL architecture (8305 nodes) and Skylake architecture (132 nodes) in Cray CS500 cluster system. NURION’s theoretical peak(Rpeak) is 25.7PF and Linpack performance(Rmax) is 13.9PF, which marked the 13th place in the list of Supercomputer Top500 in November 2018.

OpenFOAM version 5.x was compiled by comparing optimization options for the installation of KNL compute nodes. GNU C++ compile options of ‘-std=c11 -march=knl -O3 -mavx512f - mavx512pf -mavx512er -mavx512cd -mfma –malign-data=cacheline -finline-functions’ and Intel C++ compile options of ‘-std=c11 -O3 -xMIC-AVX512 -fma -align -finline-functions’ was applied. And some libraries have been included or changed to improve OpenFOAM’s computing performance. The Libhbm library, a wrapper library for handling high bandwidth memory when run in KNL’s flat memory mode, was included. This library was written specifically for running OpenFOAM in the flat memory mode. Smoothers for KNL was also updated. In this treatment, the library libKNL.so was included in the system/controlDict of OpenFOAM which need to be appended in the library lists. We then changed GaussSeidel to GaussSeidelKNL and sysGaussSeidel to symGaussSeidelKNL in system/fvSolution of OpenFOAM.

Parallel execution was tested by the Intel MPI or Open MPI after these installation test. In Open MPI, options of ‘-mca pml cm –mca mtl psm2’ must additionally be included. The combination of the three types (compile options, libraries and MPIs) yielded the best parallel execution time or parallel speed-up depending on the nature of each problem.

Using KISTIFOAM, the pre-processing GUI tool developed as a linux platform, users can create and deliver input files to NURION easily and conveniently. KISTIFOAM will play an important role in increasing the number of OpenFOAM users on the NURION supercomputer.

The Recurrence CFD method of Lichtenegger and Pirker is a method aimed to bridge
the disparity of time scales of certain processes, e.g. a slow chemical reaction
with process times in the order of minutes or even hours, within a bubble column
or an aerated stirred tank with the corresponding two-phase flow dynamics. As the
requirement to correctly resolve the fluid dynamics within the apparatus determines
the simulation time step, simulating the full process becomes unfeasible if the
disparity between the computational effort and the process time of the slower process
becomes increasingly large.

A certain type of flow system does not enter a steady-state and permanently exhibits
transient behaviour, yet these flow systems feature re-occuring flow patterns. For
such a type of flow system the Recurrence CFD method can be applied. First, a
sufficiently long period is simulated, during which the flow fields are recorded.
If this data-base of pre-recorded flow fields is large enough, traversing through
this data-base will create an approximation of the flow system's transient behaviour.

By computing the slow process, e.g. chemical reactions, on this set of pre-computed
flow fields, this process-simulation is greatly accelerated since the computational
effort for the full flow simulation is only expended for building the data-base.

This contribution first discusses the Recurrence CFD method in general, and then
details some technical aspects of the implementation of the Recurrence CFD method,
namely how to deal with data-bases larger than the computer's memory, and how to
deal with turbulence modelling within the frame of Recurrence CFD.

The Recurrence CFD method is a promising candidate to tackle the disparity of
long-time process-simulation and the inherently small time-scales of multi-phase
simulation.

## references

T. Lichtenegger and S. Pirker. Recurrence CFD – a novel approach to simulate multiphase
flows with strongly separated time scales. Chemical Engineering Science, 153:394–410, 2016.

Elementary quantitative benchmark configurations are essential for validation and comparison of interfacial flow solvers. The objective of this paper is to perform a comparative benchmark of three different VoF algorithms implemented in OpenFOAM, namely MULES [2, 6] (solver interFoam), isoAdvector-isoAlpha (solver interIsoFoam) implemented since OpenFOAM+ v1706 by Roenby et al. [4] and isoAdvector-PLIC-RDF, a more recent algorithm from Scheufler and Roenby [5]. The latest is a novel computational interface reconstruction scheme based on the calculation of a reconstructed distance function (RDF). This new scheme has been combined with the interface advection step of the isoAdvector. In this work, we have used the reconstruction method PLIC-RDF with 5 iterations from Scheufler and Roenby [5].
The benchmark is based upon two different test cases. The first case is that of a 3D stagnant bubble under zero gravity. It provides a measure of spurious currents generated by the different VoF methods. The second case is that of a single rising bubble in a quiescent liquid, as originally published in 2D by Hysing et al. [3] and later extended to 3D by Adelsberger et al. [1]. Only the case with density ratio between gas and liquid equal to 1000 is considered in the present work, for both test cases.
Spurious velocities generated in the bubble under zero gravity test case were found to be lower by a factor almost 100 with the PLIC-RDF5 method than with MULES or isoAdvector-isoAlpha. This result applies both to fully structured grids created by the utility blockMesh or with tetrahedral grids.
On the rising bubble test case, bubble centroid, velocity, volume, surface and sphericity as computed by all methods show similar results, which validates PLIC-RDF reconstruction method.
[1] 3D incompressible two-phase flow benchmark computations for rising droplets, 2014
[2] Evaluating the performance of the two-phase flow solver interFoam, 2012
[3] Quantitative benchmark computations of two-dimensional bubble dynamics, 2009
[4] A computational method for sharp interface advection, 2016
[5] Accurate and efficient surface reconstruction from volume fraction data on general meshes, 2019
[6] A new approach to vof-based interface capturing methods for incompressible and compressible flow, 2008.

Mixing is a common industrial operation, whose main objective is to increase the overall interface area between fluids in order to enhance heat and mass transfer. Among the many devices that can be used to accomplish this task, static mixers are widely employed in industrial applications [1]. In the present study, the mixing capability and the flow pattern of a Sulzer static mixer SMX were investigated. The performance of the SMX mixer for gas/liquid dispersions in the laminar flow regime was analyzed using 3D numerical simulations. CFD simulations were based on the VoF algorithms implemented in OpenFOAM, namely interFoam [2] (MULES) and interIsoFoam (isoAdvector-isoAlpha) [3].

The simulations focused on the SMX mixer's mixing behavior when nitrogen bubbles were injected and rose in a stagnant n-heptane liquid. The main purpose of our research was to quantify the flow pattern by collecting data from both CFD and experiments. The complexity of the mixer and the VoF requirements conducted us to use a large number of cells (from 30M to 45M) to perform proper investigations. Parameters such as the injector size and the mixer length were varied in the simulations in order to examine their influence on the mixing performances. CFD was applied to calculate the velocity fields, the bubbles’ sizes and velocities, the pressure drop and the mean residence time.

The combination of these results enabled a detailed overview of the complex phenomena occurring in the SMX under a G/L flow. The numerical results were found in good agreement with the experiments. Besides, the interIsoFoam method turned out to be faster and to lead to better results. The future steps of this study will be to establish a G/L mass transfer model from these results.

[1] Thakur et al., Static Mixers in the Process Industries—A Review, 2003.

[2] Waller, A New Approach to VOF-based Interface Capturing Methods for Incompressible and Compressible Flow, 2008.

[3] Roenby, A Computational Method for Sharp Interface Advection, 2016

Operators of nuclear power plant should conduct in-service test(IST). They check the safety functions of safety-related pump and valves and monitor the degree of vulnerability over time during reactor operation. One of the causes for making poor performance is cavitation that makes vibration and damage of the pump and valves. Because there is no Computational Fluid Dynmics (CFD) software licensed from nulcear regulatory agency in Korea, it is necessary to perform the validation of prediction accuracy for IST-related flow phenomenon like cavitation.

In this paper, the simulations are conducted using OpenFOAM multiphase solver – ‘interphaseChangeFoam’ by implementing an extended cavitation model and ANSYS Fluent multiphase solver. It uses volume of fluid method. The cavitating flow through a NACA0015 hydrofoil at a fixed ange of attack(5°) was assumed to be steady, incompressible, isothermal and multiphase flow. The simulated results were compared with experimental result.

Acknowledgement

This work was supported by the Nuclear Safety Research Program through the Korea Foundation Of Nuclear Safety (KOFONS) using the finiancial resource granted by the Nuclear Safety and Security Commision (NSSC) of the Republic of Korea (No. 1805007)

Numerical wave tanks (NWTs) are an essential tool for wave energy converter (WEC) development. Due to it's opensource nature and wide user base, OpenFOAM is proving to be a good software platform for implementing NWTs for WEC experiments [1, 2]. Indeed, in a recent review of CFD NWTs for
WECs [5], OpenFOAM was the most prominately used CFD software (39% of total).

Dynamic mesh motion (DMM) can prove a challenge during WEC simulation, due to factors such as: (a) large ampitude body motions when a WEC is driven into resonance, with the input waves, by a control system [2], (b) multiple DoF motion by a floating WEC, and/or (c) multi-body WEC with different sections oscillating closely to each other. Fortunately, the overset method has recently become available in OpenFOAM , which possesses the potential to eliminate many barriers for the challenging DMM in WEC simulations.

The extended abstract will present results from a number of different case studies, considering difficult to handle DMM scenarios which can be encountered in WEC NWT experiments. Preliminary investigations of utilising the overset method, for single body, point absorber-type WECs, reported in [3,4], are extended to consider larger amplitude motions. Additionally, simulations of a hinged raft-type WEC, are presented, involving two connected bodies, which pitch around a mutual hinge, whilst heaving and surging in extreme waves.

[1] Davidson et al. Implementation of an openfoam NWT for numerical wave tank experiwave energy experiments. In 11th European Wave and Tidal Energy Conference, 2015.

[2] Davidson et al. Evaluation of energy maximising control systems for WECs using OpenFOAM. In OpenFOAM-Selected papers of the 11th Workshop, Springer, 2018.

[3] Windt et al, Performance assessment of the overset grid method for NWT experiments in the openfoam environment. In 37th International Conference on Ocean, Offshore and Arctic Engineering, 2018.

[4] Windt et al. On the importance of advanced mesh motion methods for WEC experiments in cfd-based NWTs. In VIII International Conference on Computational Methods in Marine Engineering, 2019.

[5] Windt et al, High-fidelity numerical modelling of ocean wave energy systems: A review of CFD-based NWTs. Renewable and Sustainable Energy Reviews, 2018.

This paper describes recent updates in the Naval Hydro Pack. The Naval Hydro Pack is an OpenFOAM library for broad range of computational marine hydrodynamics problems based on foam-extend (see Vukcevic et. al. 2018 for additional details). After successfully resolving the issue with spurious air velocities using the Ghost Fluid Method (GFM, see Vukcevic et. al. 2017), a problem had surfaced related to over–prediction of eddy viscosity near the interface. The over–prediction has been successfully resolved by implementing the turbulence models that are sensitive to free surface flows related to gravity waves (see Larsen and Fuhrman 2018). This enabled us to accurately predict the added resistance of ships with forward speed, even for short wave lengths.

The majority of the last year’s development has been related to handling multiple floating bodies in an efficient and versatile way. The Naval Hydro Pack now offer run–time selectable mesh deformation strategies that are suitable for all kinds of body representations using standard body–fitted mesh, Overset Mesh (OM), Immersed Boundary (IB), or their combination. This, for example, enables a simulation of a self–propelled ship with active rudders, where the body–fitted Overset Mesh (OM) is used for the ship and the IB is used for the rudders. The results of such a simulation will be presented in the full paper. Substantial developments have been made both in OM and IB methods to allow efficient and robust parallel implementations, now tested on hundreds of cores.

In addition, a solver that takes into account compressibility of the air has been developed for sloshing and slamming effects. The implementation based on the extended GFM allows large Courant numbers to be used, resulting in only a minor computational overhead compared to our incompressible solvers.

This paper presents a developed new coupled method with HOS and CFD in overset grid. Our in-house CFD solver naoe-FOAM-SJTU, with the help of overset grid, can simulate large amplitude motion of structures. When considering wave structure interaction, irregular wave is essential for analyzing the motion response and impact influence of offshore structure. However, with large cost and time consuming, a long-time wave evolution in CFD is difficult to achieve. Combined a potential theory HOS with CFD and validated the coupled method in overset gird, this paper provides a better method to solve wave structure interaction in an efficient and accurate way.

The paper discusses calm water and added resistance simulation of a tanker ship model using OpenFOAM. Initially, calm water simulations are performed for a KVLCC2 model, for varying Froude numbers, with sinkage and trim free motion, after verification and validation study. The uncertainty estimation for the verification study is discussed extensively, including studies with multiple uncertainty estimation methods. The target of the uncertainty study is to provide a relative comparison among the popular uncertainty estimation methods used in CFD simulations and also discuss their advantages and limitations. Next, head wave simulations are attempted with a heave and pitch free motion, and the results are compared with experimental data. The paper comments on the findings from the uncertainty study and concludes that the OpenFOAM solver is well capable of predicting ship resistance in sea conditions.

Extrusion blow molding is one of the most popular processes for manufacturing hollow parts made of thermoplastics. The few numerical simulation codes developed for the extrusion blow molding process, do not allow the modelling of all the relevant phases comprised in the process. The authors of this work are developing a set of computational tools aiming the numerical modelling of all the relevant phases of the extrusion blow molding process, namely: parison extrusion, mold clamping, parison inflation and cooling, employing the OpenFOAM computational library.

This works presents the set of modelling tools developed to simulate both the clamping and inflation phases of the extrusion blow molding process. During clamping, the parison is placed between the two halves of the mold, which are then clamped together, deforming the parison and forming the mold cavity. Along the inflation, pressure is applied through air blown into the clamped parison, inflating it until it touches the mold cavity surface.

The implemented model comprises an explicit calculation of the parison’s deformation, which is defined as a finite area mesh in OpenFOAM, and allows the use of two constitutive models: linear Elastic and multimode K-BKZ integral viscoelastic. The results obtained with the code allow the prediction of the final parison’s thickness distribution, thus being and useful tool to support design activities related to this transformation process.

Lost cores made from salt (sodium chloride) are used in high pressure die
casting. Core failure can be a problem due to the high dynamical forces and/or
weaknesses of the salt core. In this work the solidification and shrinkage of
salt cores is simulated to obtain the residual stresses developed during the
process. The simulation is performed in two stages: Simulation of the
solidification process and simulation of the residual stresses. The latter,
employs the residual temperature concept proposed by Tropsa. In this concept the
temperature gradient is recorded when a cell solidifies. This gradient field is
then used to calculate the residual temperature by integrating the temperature
gradient along the solidification path. The residual stresses are then
calculated by solving the thermo-elastic problem with existing solvers.
In this presentation, the methodology and implementation will be outlined and
some initial validation results will be shown.
Vlado Tropsa: Predicting Residual Stresses due to Solidification in Cast Plastic
Plates, PhD Imperial College, 2001

In this work, benchmark non-Newtonian fluid flow simulations are performed for a few canonical configurations, using various viscoelastic constitutive equations. The 4:1 plane contraction, and cross-slot geometries are chosen as representative flow configurations. ViscoelasticFluidFoam solver from the foam extend library of OpenFOAM is used to solve the governing equations for various models. A detailed analysis of flow quantities, for a set of given boundary and initial conditions is performed for all the models. The sensitivity of flow quantities of interest to FENE-P, PTT-Linear, Oldroyd-B, XPP-SE and Carreau Yasuda constitituve equations is presented for both the benchmark geometries.

URANS simulations have been conducted with Foam-Extend-4.0 on a low specific speed centrifugal pump (RP26, n_q=26 min^-1) and a single blade pump (1BP, n_q=45 min^-1). 1BPs are characterized by highly unsteady flow fields even at design conditions. Low specific speed centrifugal pumps also show this flow behavior, especially in part load operation. It is aimed to assess the accuracy of the scale adaptive k-ω-SST-SAS model for these types of pump flows. Results are compared to state-of-the-art k-ω-SST simulations, spatially and temporally high resolved flow measurement data at the impeller outlet of the RP26 and in-house pressure and flow rate measurements of the 1BP.

Due to the occurrence of unphysical pressure oscillations (UPO), the rotor-stator interface algorithm is investigated. On rotor-stator-interfaces with non-planar shape - for radial pumps a cylindrical shaped plane - flux imbalances are detected, which are triggered by the misalignment of neighboring faces. A conformally meshed interface avoids the flux imbalances and is used to compare the flow fields of simulations with and without UPOs. It is verified, that the flow field and turbulent variables are not significantly affected by unbalanced interface fluxes and no unphysical oscillations occur for these fields.

For URANS simulations with the SST model, ensemble averaged mean velocity as well as head are in good agreement to the measurement data for both off-design and design conditions. The turbulence field, containing the standard deviation of the ensemble averaged velocity and the modelled turbulent kinetic energy (TKE), is significantly overestimated in part load operation, because of a high amount of TKE at the volute tongue.
The SST-SAS model is in a similar good agreement for the ensemble averaged mean velocity and head. In contrast to the SST model, reasonable values of TKE are calculated resulting in a significantly improved agreement to measured turbulence in partial load. Thus, a clear advantage of scale adaptive turbulence modeling to predict turbulent quantities in highly unsteady pump flows is shown.

Turbulent gas mixing in a buoyant flow is an important physical mechanism in many applications ranging from industrial gaseous emissions into atmosphere to nuclear reactor containment flows. With relevance to nuclear reactor safety, erosion of a stratified light gas layer is often studied due to the possibility of hydrogen layer formation and flammability in the containment as it occurred at the Fukushima Daiichi accidents. Experiments were performed in MiniPanda facility (ETH Zürich, Switzerland) to study the erosion of an initially present helium layer by an impinging hot vertical air jet with different momentum.

The “containmentFoam” solver is developed at Forschungszentrum Jülich (presented in another paper) to study the gas distribution and pressurization of reactor containment during an accident and assess possible safety measures. Basically, it bases on the URANS equations closed by k-omega SST model. The original k-Omega SST model in OpenFOAM is improved by adding turbulence production and dissipation terms due to density gradient.

In the present paper, above improvements are validated on basis of the three different MiniPanda test cases, two in the complex regime of nearly equal buoyancy driven mixing and momentum driven mixing (Froude numbers ~ 0.7, 0.8, injection Reynolds numbers ~ 2350, 1250, mass Peclet numbers ~ 530, 280), while the another one in dominated momentum driven mixing regime (Froude number ~ 1.3, injection Reynolds number ~ 4400, mass Peclet number ~ 990). It is found that the buoyancy turbulence terms lead to a considerably slower erosion of stratified layer due to damping of turbulent kinetic energy at the interaction zone of jet and stratified layer and must be considered for a successful validation.

The standard implementation of pressure velocity coupling in OpenFOAM makes use of a modified Rhie-Chow momentum interpolation method. Since the original Rhie-Chow interpolation method is time step size dependent and a checkerboard pressure field might be obtained for small time steps an additional correction term is already implemented in OpenFOAM in order to prevent these two deficiencies. This correction term is known to add numerical dissipation to the algorithm and therefore undesirably affects the accuracy of LES results. Recently the implementation of this correction term was modified in the latest OpenFOAM version 6. In our study we investigate the influence of both correction terms (OpenFOAM 6 implementation and former implementation) on LES results of turbulent channel flow at Re_τ = 180 and Re_τ = 500 in comparison with results where the correction term is excluded. LES are performed without and with a sub-grid scale model (WALE) in order to evaluate the amount of undesired dissipation added by the correction term and of desired dissipation added by the sub-grid scale model. In addition, we examine how different grid resolutions in streamwise and spanwise direction affect the LES results and analyze the behavior of the correction term. Simulation results are assessed based on time-averaged profiles of velocity, root mean square of fluctuating velocities, dissipation of turbulent kinetic energy, as well as energy spectra of streamwise velocity and are compared with DNS reference data.
Our study shows a remarkable deterioration of the results for the higher Reynolds number when the correction term is used emphasizing its dissipative character. Especially the new implementation of the correction term in OpenFOAM 6 affects the LES results even more. For the lower Reynolds number, the influence is much less pronounced. The grid study shows a decreasing influence of the correction term at higher grid resolutions. Comparing the results obtained from simulations with and without a sub-grid scale model, we can show that the added dissipation of the correction term dominates the dissipation of the sub-grid scale model and that the sub-grid scale model does not generally improve the LES channel flow results.

This work presents a novel Large Eddy Simulation approach with a direct resolution of the subgrid motion of fine concentrated vortices. The method is based on combination of a grid based and grid free computational vortex particle (VPM) methods. The large scale flow structures are simulated on the grid whereas the concentrated structures are modeled using VPM. Due to this combination the advantages of both methods are strengthened whereas the disadvantage are diminished. The procedure of separation of small concentrated vortices from the large scale ones is based on LES filtering idea.

Due to the flow instability, which is an inherent property of turbulent flows, small scales are generated inside the velocity field represented on the grid. As soon as a typical size of any fine vortex structure becomes comparable with the grid size or less than it the structure disappears due to artificial or numerical viscosity.

The purpose of this new method is to protect such vortices from the artificial damping inherent to grid based techniques. For that, the vortices are detected in small scale vorticity field and converted to vortex particles (using an extended Lagrangian library in OpenFOAM) which keep their identity and are transported according to viscous flow equations written for vorticity-velocity variables. use of decomposition of the velocity and vorticity fields into the grid based and grid free components u = u^g + u^vand ω = ω^g + ω^v and their substitution into the Navier Stokes equation results in two coupled transport equations. The flow dynamics is governed by these two coupled transport equations taking two-way interaction between large and fine structure into account.

Validation and verification was performed for a well tried benchmark test of the decaying homogeneous isotropic turbulence and free turbulent jet flow. The energy back-scattering and anisotropy of small scale motions are observed using this method. High accuracy result at the resolution much less that the pure grid technique is also acheived. VπLES goes into DNS when ∆ → 0 and is automatically switched off at Re → 0.

In this work a novel machine learning method is presented to infer algebraic stress models for the closure of RANS equations directly from high-fidelity LES or DNS data. Commonly these models are derived from Reynolds-stress models as a projection onto a set of tensor polynomials. Instead, the new learning algorithm builds the models by identifying relevant functions from a library of candidate functions. A library of candidate functions is constructed by applying a set of simple mathematical operations to the invariants of the strain-rate and rotation-rate tensor as raw inputs and combining the resulting functions to new candidates. By using sparsity-promoting regression techniques based on regularised least squares regression, e.g. elastic net regression, new models are found as a sparse combination of these candidates. By being data-driven the method relaxes assumptions commonly made in the process of model development. In order to extract useful information from the high-fidelity data we use a novel k-corrective-frozen-RANS tool implemented in OpenFOAM: The ω-equation of the k-ω SST model is passively solved given the data and the resulting ω is used to compute the residual of the k-equation. Both variables are used to correct the underlying linear eddy viscosity model and serve as a target for the machine learning. Models are learnt for a set of canonical flow cases, i.e. periodic hills (Re=10595), converging-diverging channel (Re=12600) and curved backward-facing step (Re=13700). The predictions of the discovered models are significantly improved over the k-ω SST. The assessment will focus on the training error and the predictive performance of the learnt models for new test cases - unseen during the learning phase. Finally, the applicability of the method for rapid and robust model development for industrially relevant flow cases will be discussed.

This training session covers parametric modelling techniques within CAESES along with its capabilities for process automation and shape optimization. A marine propeller will be investigated in steady-state, open-water condition using cfMesh for automated grid generation and a simpleFoam solver. A design space exploration allows convenient automation of studies on domain size and grid resolution and provides valuable insight into the propulsive characteristics of different geometry variants. Based on a Response Surface Model a subsequent optimization will be demonstrated in the search for maximum open-water efficiency. Attendees should bring a basic understanding of OpenFOAM as well as a fundamental knowledge of surrogate modelling techniques. No hands-on required.

Fuel cells convert hydrogen into electricity. Emitting only water as a reaction product, these devices are extremely environmentally-friendly. Conventional polymer electrolyte fuel cells (PEMFC) require an active ventilation system for air supply and water management. Air-breathing PEMFC work passively consuming air from the ambient. Renouncing valves, gas channels and pumps, such cells can be built amazingly small.

During operation, water vapour is generated as a reaction product in a PEMFC. Especially at high current density, the vapour condenses into liquid water. Blocking the reaction zones, liquid water is one of the main reasons for low cell voltage and high internal resistance. Water management is especially important for air-breathing cells, as there is no active air flow available that could remove the water. Liquid water must be transported from the reaction side by means of evaporation and natural convection.

In the first part of the talk, the theory of the fuel cell model is presented. Details on its implementation in the open source library, openFuelCell, will be given. Secondly, numerical results on modelling an air-breathing fuel cell operated at CIEMAT will be shown and discussed.

A flexible modeling approach for the accurate approximation of convection‐dominated reactive species boundary layers is introduced. A substitute problem is solved numerically and analyzed employing statistical methods. The numerical data is then used to train a machine learning model which can be used to approximate the reactive mass transfer locally if a direct resolution of the concentration boundary layer is unfeasible. Compared to previous modeling approaches, the machine learning model replaces the analytical solution of a simplified substitute problem, which makes it applicable to more complicated and general settings. The machine learning workflow is implemented based on OpenFOAM for data-generation, scikit-learn for data-processing and analysis, and PyTorch to learn an approximate representation of the data. One advantage of PyTorch models is that they can be easily deployed in C++ codes like OpenFOAM. The mass transfer from rising bubbles at high Schmidt numbers is a challenging problem and suitable to validate the data-driven model. Model errors are reported for various bubble shapes and dynamic conditions.

During a severe reactor accident like Fukushima, large quantities of steam, non-condensable gases(e.g. H2 and CO2), and fission products are released into the reactor containment. CFD models are used to predict gas mixing inside the containment, its pressurization and appropriate counter measures. In general, thermal radiation modeling is ignored due to low temperature differences. However, recently it has been found that thermal radiation phenomenon plays a significant role in buoyancy driven flows and gas mixing processes even at low temperature differences and levels. This is because of the concentration of the participating gas medium (steam) becomes significantly high ( average 0 to 60vol % and locally 100vol % possible) under accidental conditions.
Now OpenFOAM provides three kinds of solver for the radaitive transfer equation. The viewFactor model in OpenFOAM-6 is applicable only for surface to surface radiation, while the P1 model uses the simplified diffusion equation valid only for optically thick media. Even though the Discrete Ordinate Method (DOM) is widely used in CFD, this method suffers from false scattering and ray effect, which can cause unphysical simulation results. Moreover, the DOM method cannot handle the scattering on particles or droplets due to the limited angular discretization used in the model. In this paper, a Monte Carlo solver including non-gray gas models is developed and coupled to energy equation of the containmentFOAM solver, developed at Forschungszentrum Juelich for the analysis of gas mixing and pressurization of the reactor containment.
The Monte Carlo method has the ability to model the non-gray gas spectral properties, particle scattering effects and interaction with surfaces. Besides gray gas model, three non-gray gas models: the Weighted Sum of Gray Gases(WSGG) model, the Statistical Narrow Band(SNB) model and the Line-by-Line(LBL) model are compared for CO2-H2O gas mixtures. A 2D analytical benchmark case using grey gas model is computed to verify the fvMCM solver. On this basis, the same case is repeated with different concentrations of CO2 and H2O. It is concluded that the SNB model is reasonably good for computing radiation heat transfer in CO2-H2O gas mixtures.

Fluid-particle transport systems present a significant practical relevance, in several engineering applications, such as oil sands mining and polymer processing. In several cases it is essential to consider that the fluid, in which the particles are dispersed, has underlying viscoelastic characteristics. For this aim, a novel numerical algorithm was implemented on an open-source finite-volume viscoelastic fluid flow solver coupled with an immersed boundary method, by extending the open-source computational fluid dynamics library CFDEMcoupling. The code is able to perform fully-resolved simulations, wherein all flow scales, associated with the particle motion, are resolved. Additionally, the formulation employed exploits the log-conformation tensor approach, to avoid high Weissenberg number issues. The accuracy of the algorithm was evaluated by studying several benchmark flows, namely: (i) the sedimentation of a sphere in a bounded domain; (ii) rotation of a sphere in simple shear flow; (iii) the cross-stream migration of a neutrally buoyant sphere in a steady Poiseuille flow. In each case, a comparison of the results obtained with the newly developed code with data reported in the literature is performed, in order to assess the code accuracy and robustness. Finally, the capability of the code to solve a physical challenging problem is illustrated by studying the interactions and flow-induced alignment of three spheres in a wall-bounded shear flow. The role of the fluid rheology and finite gap size on both the approach rate and pathways of the solid particles are described.

Standard design procedures and simulation tools for marine structures are aimed primarily for use by the offshore oil and gas. Mooring system restoring forces acting on floating offshore structures are obtained from a quasi-static mooring model alone or a coupled analysis based on potential flow solvers that do not always consider nonlinear mooring-induced phenomena or interactions between fluid and structure and the associated viscous effects. This paper presents the predictions of hydrodynamic damping of moored offshore structures by using the coupled tool between a dynamic mooring model and a Reynolds-averaged Navier-Stokes solver. The hydrodynamic damping coefficients were obtained by using a least-square algorithm to fit the time trace of decay tests.

We analyzed a moored floating cylinder and validated our motion, mooring force and hydrodynamic damping predictions against experimental measurements. The mooring system consisted of three catenary chains. The analyzed response comprised the decaying oscillating body motions, the associated linear and quadratic damping characteristics, and the associated tensile forces in mooring chains. It indicated that for some cases like pitch and surge, the potential-based solver cannot accurately predict the oscillatory motions of the cylinder at the natural frequencies, nor the hydrodynamic damping coefficients, which is caused by the lack of accurate prediction of the viscous effects by simplified methods adopted for the potential solver. By contrast, predicted cylinder motions, mooring forces and hydrodynamic damping coefficients by the coupled RANS solver generally correlated favorably to comparable experimental data, thereby confirming that the implementation of our coupling tool enabled the reliably prediction of the system’s response and mooring forces as well as hydrodynamic damping in various sea conditions.

An efficient hybrid numerical method was developed in OpenFOAM to study wave-induced ship response coupled with sloshing-induced compressible fluid response in partially filled tanks of a 138,000m³ liquefied natural gas carrier at zero speed in regular head and beam waves. A nonlinear boundary element Rankine source code, solving the incompressible incident flow field, accounted for the nonlinear stationary forward speed problem and the weakly nonlinear seakeeping problem. A Reynolds-averaged Navier-Stokes equations solver, computing the compressible sloshing flow in tanks, considered the associated discontinuous numerical domains of coupled simulations. Grid sensitivity studies were performed. Comparative available experimental model test measurements of ship response and free surface sloshing elevations in tanks validated the numerical method. Further investigations regarding the influence of wave steepness and tank fill levels on sloshing were performed. Results demonstrated that liquid sloshing in tanks had a large influence on ship surge and roll responses but hardly effected ship heave motions.

Northern European countries such as Belgium, are characterised by a very typical coastal defence system: a hard dike with a promenade and an almost continuous line of high-rise buildings, fronted by a mildly sloped and very shallow foreshore. The presence of this shallow foreshore influences wave transformation from offshore to nearshore and the interaction with coastal structures significantly. However in literature, very few studies are available on wave overtopping and wave impact on coastal structures for shallow foreshores. Furthermore, the effect of sediment mobilisation and transport on the total energy budget is still not completely understood. Current research, performed in the framework of the CREST research project, aims at developing a reliable and accurate numerical tool for the assessment of wave overtopping over the dike crest for the specific case of a very shallow and morphologically dynamic foreshore.

Numerical simulations are carried out for the case as described above. The simulations are performed using the interFoam solver coupled with a sediment transport module. The turbulent flow field is simulated with the k-Omega SST model. Wave generation and absorption are implemented using the IHFOAM toolbox.Hence, two interfaces are present: the water-air interface which is captured by the Volume of Fluid (VOF) method and the water-sediment interface which is captured with a moving mesh approach. Sediment transport is implemented through traditional bed load and suspended load formulae. The resulting bed level change is based on the continuity of sediment and is calculated with the Exner equation. Arbitrary mesh interfaces (AMIs) are implemented to preserve mesh quality when simulating both moving (shallow foreshore) and non-moving regions (dike, promenade, buildings). First static simulations are performed and the numerical results are compared with experimental data. These results give insight into the accuracy of the simulations and the abitlity of OpenFOAM to correctly predict overtopping rates. Next, moving bed simulations are performed to study the effect of the developed scour hole in front of the dike on wave overtopping rates and velocites on the dike and promenade. The results of these simulations will be presented at the conference.

Computational Fluid Dynamics (CFD) modelling of floating structures like wave energy converters (WEC), moored ships, floating offshore wind turbines (FOWT), or movable platforms represents one of the biggest challenges for offshore and coastal engineering applications, and, indeed, a considerable amount of research has been conducted in the fields of renewable energy, naval engineering and oil-gas industry.

In a CFD mesh-based solver, the quality of wave-structure interaction is strongly dependent on the robustness of the mesh motion technique. Wave and currents often leads to very large displacements of movable structures, thus, a strong tool for moving grids is needed.

Although deforming grid approaches are available, several instabilities can be observed when dealing with complex structures (real geometries), therefore, the Overset Mesh technique (OpenFOAM-v1806) is appearing as the most suitable technique for marine applications.
In this work, first, we show the validation of wave interaction with a floating structure using the Overset mesh. Then, we extend the analysis to complex floating structures (single and multi-bodies). A newly implementing technique for sharp interface advection called isoAdvector (Roenby et al., 2016 [1]) has been coupled with the Overset mesh. The performance of the isoAdvector is analysed when simulating waves travelling trough fixed and moving overset grids. To deal with the over production of turbulence levels, a stabilized turbulence model (Larsen and Fuhrman 2018 [2]) has been included in the model. Last updates of IHFOAM (Higuera et al., 2013 [3]) suite are also presented. The numerical implementations have been developed in OpenFOAM-v1812 (ESI-group [4]).

Several simulations have been performed (currents and waves interaction) with single or multi floating bodies.

References
[1] J. Roenby, H. Bredmose, H. Jasak, A computational method for sharp interface advection,
Royal Society Open Sience 3 (11) (2016) 160405.
[2] B. E. Larsen, D. R. Fuhrman, On the over-production of turbulence beneath surface
waves in reynolds-averaged navier-stokes models, Journal of Fluid Mechanics 853 (2018) 419–
460.
[3] P. Higuera, J. L. Lara, I. J. Losada, Realistic wave generation and active wave absorption for navier-stokes models: Application to openfoam R , Coastal Engineering 71 (2013)
102–118.
[4] ESI-Group, Openfoam the open source cfd toolbox. (https://www.openfoam.com/)

Detailed chemistry in FOAM can take up a substantial amount of time in the
solution process. A potential avenue to accelerate the time to solution is to
solve the species and energy equations together in a block-coupled manner. The
work follows the spirit of the work of Di Domenico, while the required
sensitivities (Jacobian) of the chemical sources terms are obtained by numerical
differentiation. The block-coupled species and energy transport has then been
implemented into two solvers: a laminar flame solver and an artificially
thickened flame solvers. The solvers have been validated for 1D test cases as
well as a complex 3D validation case.
This presentation covers an outline of the implementation, validation results
as well as performance comparisons for different time steps.
Massimiliano Di Domenico: Numerical Simulations of Soot Formation in Turbulent
Flows. PhD Thesis Universität Stuttgart, VT-Forschungsbericht 2008-01

Abstract
The main motivation for this study is to assess the applicability of OpenFOAM to simulate flow inside safety-related pumps in the nuclear power plant. As a preliminary study, steady-state Reynolds averaged Navier-Stokes simulations with a two-equation turbulence model have been conducted by using OpenFOAM on tip leakage flow inside a linear cascade under the design condition, which significantly affects tip leakage vortex, and secondary flow, and rapid pressure change near tip clearance. The pressure and velocity distributions on the both sides of the cascade have been compared with experimental data and the predicted results of a commercial code, ANSYS CFX R18.1. As a result, there has been good agreement between the results of OpenFOAM simulation, and the results of the other comparisons in terms of the pressure and velocity distributions.

Acknowledgement

This work was supported by the Nuclear Safety Research Program through the Korea Foundation Of Nuclear Safety (KOFONS) using the financial resource granted by the Nuclear Safety and Security Commission (NSSC) of the Republic of Korea (No. 1805007).

We have studied how to use machine learning techniques to predict fluid dynamic performance. In the field of traditional CFD, the analysis has been performed by generating a unique mesh system for the object and solving a linear system of equation numerically. Thus, generating a mesh system for analysis has always been a bottleneck of the CFD technique, and various techniques including meshless method have been developed to remedy this problem. Another approach is to apply a machine learning technique to predict the performance of an object due to surrounding fluid flow. Unlike traditional CFD, which obtains a solution, machine learning can be understood as a process of obtaining the coefficient matrix of the system of equations. At this time, the size of the system should be kept constant. However, if the grid system is changed according to the shape, the size changes as well. To tackle this problem, we used the cut-cell method of the immersed boundary method. A predictive model is derived by learning a list of volumes included in the shape as input values and the drag, lift, or velocity components in x, y, and z directions of each cell as output values. The detailed results will be discussed in the presentation.

Considering a floating plate on water surface, its effect of attenuating incoming waves is currently unclear. Especially, when the floating plate is deformable, its bottom boundary induces a complex effect on the transmission fluid. This study investigates how much can wave energy go through the floating plate, particularly, the role played by the elasticity of the plate. From a rigid condition, young modulus of the plate is reducing gradually and the transmitted wave energy is measured. The fluid behaviours underneath the floating plate are also observed to analyse the reason for the attenuation.

To simulate the above problem, a multiphase hydroelastic solver has been developed based on OpenFOAM, with the Navier-Stokes equations describing the fluid domain, the St. Venant Kirchhoff solid model governing the plate deformation and a coupling scheme to achieve the fluid-structure interaction. Thereby, fully coupling has been achieved between wave motion and the elastic deformation of the floating plate. In addition, the new solver has been validated by experiments.

This work can be useful for studying wave transmission over a rigid/elastic floating body, for example, the attenuation of ocean waves along an ice floe, which is of special research interest due to the climate change in the Arctic.

Modern propellant grains used in solid rocket motors (SRM) are characterized by complex three-dimensional (3D) geometries developed from primitive two-dimensional shapes. The grain area profile coupled with a propellant burn rate generates the mass flow to deliver the required thrust. In most cases, several design iterations are required to match a grain geometry to a specified thrust profile and to provide this thrust with low-pressure loss and in a manner that does not endanger the integrity of the solid rocket motor.

Typically, expensive and coupled Computational Fluid Dynamics (CFD) simulations are required to characterize erosion in the motor port, pressure drops across the port, and the acoustic characteristics of the motor port during operation and particularly the transient phases during start-up and blow-down of the motor. Complex commercial CFD tools exist to perform the required simulations but these tools are expensive, commercial and obfuscated.

For these applications, the CFD method must compute the flow in a time-dependent domain, and the method must track the propellant grain surface, the velocity of which depends on the burning rate and local flow properties. The coupled problem is complex and the primary contributor to this complexity is the deformation required in the computational for arbitrarily complex 3D geometries.

The aim of this paper is to describe a coupled 3D simulation capability for the design and analysis of SRM’s with a novel burn surface treatment based on geometric immersed boundary conditions implemented in OpenFOAM®.

The methodology followed in constructing the solver for the flow-field is described where the governing equations, model equations, and boundary conditions are listed, and their use is motivated. Dynamic mesh handling and specifically the geometric immersed boundary condition is described well as its coupling with the flow solver. Results from the method described here is then evaluated against a realistic SRM configuration and compared against experimental data. Good agreement was found between the simulation data and the experimental data and a typical simulation requires only a few hours to solve. The method can easily extend to contain any number of governing equations into the same time integration step.

Among the operating devices using wave energy conversion technology, the Oscillating Wave Surge Converter (OWSC) stands out as an efficient and stable solution. These devices are composed by a buoyant flap that rotates around a hinge fixed close to the seabed. The prediction of the hydrodynamic loads they will withstand over their service life is an ongoing challenge. One of the hydrodynamic phenomena arisen when the device is operating under rough sea states is wave slamming. This phenomenon is particularly critical, since the resulting high peak-pressures in short periods of time can have negative effects on the structure. These pressure loads are influenced by the relative velocity and the contact angle between the flap and the incident wave. The aim of this study is to predict the peak pressure loads obtained during wave slamming using numerical methods. To achieve this, a two-dimensional model of the OWSC within a Numerical Wave Tank (NWT) is developed in OpenFOAM. The model setup is based on an existent experimental test, but it is worth noting that the approximation of the experimental wave was made using the Cnoidal wave theory. The NWT used a static boundary generator at the inlet patch combined with boundary conditions based on active wave absorption in both ends. The OWSC is simplified as a rigid body-fitted flap located within the NWT. Furtherly, the discretisation of the mesh surrounding the body is handled by an overlapping mesh and is independent of the one used for the NWT, allowing the high-amplitude rotations of the OWSC. The proposed numerical model accurately predicted the peak pressure loads acting on the face of the flap. In addition, the results of the angle of rotation show good agreement with the ones obtained experimentally, as it is not completely the case of the angular velocity of the OWSC, where the numerical model is not able to capture the negative peak velocities. In this case, the two-dimensional model using overlapping mesh is good to predict the slam pressure loads acting on the flap, however, further work needs to be done to capture the peak angular velocities.

In this work a deterministic thin film lubrication model implemented inside the foam-extend framework is presented.

Depending on a local film thickness between two lubricated rough surfaces in relative motion four lubrication regimes are distinguished [1]: thick film hydrodynamic, thin film hydrodynamic, mixed and boundary regime. In both hydrodynamic regimes surfaces are completely separated by the lubricant, i.e. there is no contact between surface asperities. In the thin film regime surface roughness can significantly influence the lubricant flow, while in the thick film regime smooth surfaces are a valid assumption. In the mixed regime contact pressure is shared between asperities in contact and lubricant, while in the boundary regime almost the whole contact pressure is carried by the asperities in contact. The implemented model is capable of calculating all four regimes with continuous transitions between them. The lubricant flow is modeled using the Reynolds equation [2], a two–dimensional partial differential equation governing the lubricant pressure between two surfaces in relative motion. The equation is discretised using the Finite Area Method, a two–dimensional counterpart of the Finite Volume Method, over a curved computational surface mesh. The asperity contact between two rough surfaces is calculated using a deterministic elastic–perfectly plastic contact model based on the work done by Stanley and Kato [3] and Sahlin et al. [4, 5] Since temperature increase can significantly influence transport properties of the lubricant, especially viscosity, the lubricant film temperature is calculated using a two–dimensional energy equation with an assumption of parabolic temperature profile across the film thickness [6]. The implemented model considers temperature and pressure dependence of density and viscosity, as well as non–Newtonian effects. The model is tested using two point–contact test cases, simulating a ball–on–disc experiment. The first test case [7] uses Turbo T9 oil and is purely hydrodynamic. In the second case [8] Turbo T68 oil is used, and both hydrodynamic and mixed regimes are considered.

Drilling is essential for the recovery of sub-surface energy sources, such as oil, gas and geothermal energy.
It typically accounts for large shares of the project costs. For optimal drilling operations it is required to
achieve an efficient transport of cuttings from the drill-bit to the surface. As drilling often reaches several
thousand meters below the surface, in-situ measurements are very challenging. Therefore, limited field
knowledge about the underlying phenomena exists and many investigations rely on simplified laboratory
setups and detailed software-based simulations.
Beside technical challenges drilling projects are always very costly, e.g. in case of deep geothermal wells
only the drilling costs account for at least 50% of the total project costs. Large shares of these costs are
caused by unexpected interruptions of the drilling process caused by damages to underground equipment.
These can demand days for pulling out and tripping in again the drill string. Particular importance in
these fatigue processes are lateral vibrations of the drill string. Previous work already analyzed different
factors influencing the damping of drill string vibrations, e.g. drilling mud, fluid flows, rotation.
The presented paper addresses the challenges of the interactive effects in a new approach by combining
simulation tools for fluid solid interaction and particle transport. While fluid solid interaction is con-
sidered using the fluid solid interaction toolkit published for foam-extend-4.0, LIGGGHTS is utilized
for the simulation of the particle flows. Previously, the cfdem project already released a library for
coupling LIGGGHTS with different flavours of OpenFOAM. However, foam-extend in version 4.0 was
not included in the upstream code. In a first step the presented work ported the coupling library to
foam-extend-4.0 before combining the solvers from both particle simulation and fluid solid interaction in
one unique single solver. During the scope of this work it became apparent that this integration needs to
go beyond the existing code and also provide means to transfer deformations and forces of the particles
acting on the solid between LIGGGHTS and the fluid solid interaction toolkit. The paper closes with a
brief simulation study which demonstrates the correct execution of the new proposed solver.

The immersed boundary method enables performing CFD simulations around objects without making it necessary to generate time-consuming, body conformal grids. The ability to execute large-scale computer simulations qualifies to explore the possibility of coupling the phase fraction based interface tracking Volume of Fluid (VoF) method within this context. To this end, we propose a combination of the new geometric VoF solver, interFlow (based on isoAdvector) with an in-house developed discrete forcing, direct, immersed boundary method, namely, Arbitrary Body Simulation - FOAM (ABSFoam). The coupled solver demonstrates the ability to simulate static and dynamic contact angles on the surface of volumetrically resolved moving particles of any shape. An arbitrary amount of modeling complexity can be added to enable highly accurate force calculations on the surface of the particles. The two-way coupling between fluid field and particles can then enable a better resolution of the flow field around the particles. In this work, we present the test cases that will be used to validate experiments which record a droplet (We ~ O(10)) falling on a non-constrained particle. Initial results from the simulations show that a surprisingly small amount of energy is required to move a particle from rest, which is also confirmed through experiments.
Future work will focus on the development of more precise modeling, specifically, on the surface of the substrate enabling interactions with roughness models, force calculations in the nano-region as well as inquiring the ability to scale the solver for performing multi-particle simulations.

In this paper, a density-based steady solver is developed using the pseudo time-stepping and local time-stepping based on the transient solver rhoCentralFoam in the famous CFD framework OpenFOAM. The key technologies and detailed implementation are introduced. A typical benchmark test case, inviscid flow over the NACA 0012 airfoil, is used to investigate the performance of the newly implemented solver. Simulation results on a HPC cluster show that the steady solver is much more efficient than the transient solver, and a speedup of about 4 is obtained for the test case with a total number of 22473 cells.

Portuguese shoe industry can be regarded as one of the most important sectors of the country’s economy. This sector exports 95% of its production and has been growing steadily over the last years [1]. In fact, a truly amazing transformation has gained momentum in the past two decades [2], where Portuguese companies passed form a labor-intensive industry to a highly modern competitive industry. This framework and the ever increasing need to innovate, motivated some Portuguese companies, universities and research centers to join efforts to promote advances in R&D for this industry giving birth to the cooperation project, FAMEST.

The FAMEST project comprises several subprojects aiming the development of innovative products for the shoe industry, which requires know-how in several engineering areas, from thermal and structural analysis to design and manufacturing. Among those subprojects, is the development of toecaps, whose functional requirement is to protect the work’s feet from static and impact loads.

Shoe toecaps are usually built from steel or aluminum, but in recent years plastic/composite materials have been gaining momentum due to their light weight, electric and thermal insulant properties, and non-magnetic characteristics, among others.

Computational modeling can be viewed as a tool to support design and manufacturing activities, which could contribute to increase the added value of a product and/or the optimization of a process. One of the goals of the supra mentioned project is to disseminate the use of simulation tools, based on the OpenFOAM® computational library, among the Portuguese shoe industry. As such, in this work, the goal is to resort to computational modelling tools to study the quasi-static compression loading of a toecap, aiming to improve its performance. For this purpose, the solids4Foam [3] toolbox was used.

References:

[1]- https://www.apiccaps.pt/news/calcado-em-numeros/3276.html, accessed on the 5/5/2019

[2] - https://www.apiccaps.pt/getfilev2/?f=/2018-facts--numbers.pdf&idf=MzEzMg== , accessed on the 5/5/2019

[3]- P. Cardiff et al., An open-source finite volume toolbox for solid mechanics and fluid-solid interaction simulations, 2018

Coolants are frequently used in machining in order to combat the high thermo-mechanical loads experienced near the cutting edges of the tool. Managing the temperatures at the cutting edge has many benefits including an extended tool life and improved surface integrity of the work piece. Significant increase in demand for high strength and temperature resistant alloys from the aeronautical and automotive industries has only increased the need for efficiently applied coolants and improved management of the cutting temperatures when working with these hard-to-cut materials.

Due to the hostile environment near the cutting edges in machining operations, simulations can be invaluable in determining what physical processes are at work. The primary objective of this study is to understand the flow structure and heat transfer involved when cooling with conventional oil-water emulsion coolants. To this end, a method which combines a volume of fluid approach with a modified conjugate heat transfer solver tailored to take into account thermal properties of multiple fluids is implemented and validated using OpenFOAM.

The distribution of conventional coolant on both a machining tool and a simplified geometry is modelled and a study into the heat transfer for conventional coolant cases presented.

Pressure-based algorithm is applied as a major analysis method in most commercial analysis packages, and is widely used in research and industry. In pressure-based CFD, the pressure–velocity coupling algorithm is used to convert a continuity equation to a pressure equation. The Rhie–Chow interpolation scheme is one of the most widely used pressure–velocity coupling methods[1]. The segregated algorithm with Rhie-Chow interpolation scheme is widely used because it is not demanding of computation resources, and is highly efficient for the analysis of an incompressible region with little change in the density.

In spite of these advantages, there are limitations associated with the application of this analysis algorithm to a compressible flow with large variations in density, the flow inside a combustor with sudden changes in pressure, and the flow of an expanding high-pressure gas. Various studies have been conducted to with the goal of overcoming these limitations.

In this paper, a pressure-based computational fluid dynamics algorithm for the numerical analysis of internal and external compressible flow is developed. To this end, an improved pressure-based segregated and coupled algorithms was developed. Especially, the coupled algorithm was yield by combining the pressure–velocity coupled algorithm and the pressure–enthalpy coupled algorithm. In addition, the Kurganov–Tadmor flux splitting scheme, which was studied by Kim[2] and Kraposhin et al. [3], was applied to a developed pressure-based solvers.

To confirm the analytical ability of the developed segregated solvers, the hyper tube problems was calculated. A hyper tube system is a cylindrical train that moves at high speed through a pipe with a low density. To confirm the analytical ability of the developed coupled solvers, a variety of Mach number flow problems were performed using the developed solver.

References

[1] J. Patankar, Numerical heat transfer and fluid flow, CRC Press, USA, 1980.

[2] T.W. Kim, J.H. Gill, Development of Improved Pressure-Based Solver Algorithm for Compressible Flow, KSCFE, 22.4 (2017) 28.

[3] M. Kraposhin, A. Bovtrikova, S. Strijhak, Adaptation of Kurganov-Tadmor Numerical Scheme For Applying in Combination With the PISO Method in Numerical Simulation of Flows in a Wide Range of Mach Numbers, Proc. Comp. Sci. 66 (2015) 43–52.

In the context of the design and assessment of the performance characteristics of pico-hydraulic turbines for the recovery of energy in water distribution networks, an usage of open source tools is explored. An already designed turbine is chosen to be modeled, meshed and simulated. Two types of reduced domains are explored: a pure axisymmetric domain and a combined 2D-axisymmetric one. With hybrid meshing capabilities in mind, three open source meshing software are explored: Salome 9 mesh module and a combined usage of Gmsh 4.3 and Engrid-master. Taking into account low Reynolds number near wall turbulence modeling, dense and thick prismatic boundary layers are made on blades. Simulations are carried out with foam-extend 4.1, which in turns was selected for its implementation of the mixing plane interface, necessary for the explored reduced domains. Global integral performance variables and convergence characteristics are compared with results obtained with a common commercial simulation software in a similar setup. The explored mesh codes exhibit problems in robustness, being the most robust the Gmsh code, with the drawback that it doesn't owns the capability of generate hybrid meshes with prismatic boundary layers en 3D domains. On the other extreme, Engrid code is capable of generating high quality prismatic boundary layers but its relatively buggy and unstable. Simulation results shows that explored open source tools achieve similar global integral variables (with discrepancies of maximum ~5% and oscillations amplitudes about ~10%), but longer simulation times and poorer convergence characteristics. Improvements to the mesh and simulation codes are proposed.

Liquid metal batteries (LMBs) are promising candidates for electrical energy storage. An LMB is a concentration cell made of three liquid layers, stably stratified by density. A molten salt acts as an electrolyte between two liquid metal electrodes. The simple chemistry and geometry, the liquid nature of the active layers and the presence of multi-physics phenomena have made the LMB an intriguing candidate for continuum mechanics investigations. Simultaneous transport of charge, heat, mass and momentum takes place in each liquid layer together with chemical reactions. The interfaces between layers are the places in which electrochemical reactions occur along with interfacial transport phenomena.
In our work we investigate heat and mass transport in LMBs with openFOAM libraries using a multi-region approach. We assign to each layer a numerical region, and we design a procedure able to ensure the physical coupling between the different transport mechanisms through an iterative procedure. The heat and mass transfer equations are solved on a global mesh and in the positive electrode region respectively. Then we solve the Navier-Stokes equations in each fluid region. Appropriate boundary condition were designed to ensure consistent transport at the interfaces between different regions. Thanks to this procedure we can compute temperature and concentration distributions and the corresponding thermal and compositional convection. Therefore, we can investigate the interaction of different mechanisms and can give a prediction of the fluid flow in the interior of an LMB. The numerical procedure is introduced as well as the first results. Furthermore, the modeling limitations and the future developments are discussed.

The support of computational modelling tools is a wise way to save resources in numerous industries, including polymer and composites processing. This framework has been allowing to replace the traditional entirely experimental based trial-and-error design approaches, by exploratory studies done with the support of computational modelling tools, with clear advantages for the final products obtained. In this work we illustrate the numerical based design process with two cases studies of the profile extrusion industry, aiming to: (i) balance the outlet flow of a co-extrusion die and (ii) improve the design of an extruded profile reinforced with continuous fibres. The work was done with the support of two different OpenFOAM solvers, interFoam, to design of the co-extrusion die, and solids4Foam, for the mechanical design of the composite profile. These studies are being undertaken in a broader R&D project leaded by SOPREFA, which integrates also the University of Minho, University of Aveiro and the Portuguese Footwear Technological Centre (CTCP). SOPREFA is a well-known Portuguese pool cover and deck manufacturer, and with this project aims to increase its competitiveness, by improving the performance of two currently available products.

Acknowledgements

This work is co-funded by the European Regional Development Fund (ERDF) through the PORTUGAL 2020 Programme under the project SMARTPOOL COVER AND DECK (n° 017656). The authors would like to acknowledge the Search-ON2: Revitalization of HPC infrastructure of UMinho” (NORTE-07-0162-FEDER-000086), co-funded by the North Portugal Regional Operational Programme (ON.2 – O Novo Norte), under the National Strategic Reference Framework (NSRF), through the European Regional Development Fund (ERDF).

One effective way of combining in a synergic way the properties of individual polymers is the manufacturing of multi-layered products, whose performance of is directly linked to the layers uniformity. Polymer multilayer co-extrusion is a manufacturing process wherein two or more polymers are fed to a common extrusion die to form a multi-layered product. The usual manufacturing approach starts by co-extruding two layers which are duplicated in one Interfacial Surface Generator Module (ISGM). The multiplication of the layers is achieved by dividing the flow of the two inlet layers in two individual streams, deforming and overlapping the two individual streams, which are subsequently joined to reach a four layer structure. This work comprises a computational based study aiming to improve the knowledge related to the design of ISGM, which was done with the support of the multiphase flow solvers from OpenFOAM computational library. Each ISGM comprises several geometric transformations, which can be made simultaneously or sequentially, producing effects on the velocity fields that impact the uniformity of the layers. The simultaneous combination of geometrical transformations, allows reducing both the device length and, in general, the total pressure drop. However, when specific effects are combined, the layers become unevenly distributed. To better understand the system behaviour, several configurations for the ISGM were studied, aiming to identify the geometry details that promote a non-uniform layer distribution. The results obtained showed the simultaneous combination of specific geometrical transformations that have a negative impact on the uniformity of the polymer layer thickness distribution, at the device outlet. Thus, configurations that promote this type of flow distribution should be avoided.

Suspensions consisting of solid particles dispersed in a liquid often exhibit complex flow properties that can not be adequately modeled with a simple Newtonian approach. The presently investigated suspension, which serves as a suspension or slurry electrode for zinc-air flow batteries, consists of a liquid electrolyte and microscopic zinc particles. To stabilize the suspension, i.e. prevent sedimentation of the metallic particles, the liquid electrolyte is gelled with a polymeric binder (polyacrylic acid, PAA). The binder induces significant shear-thinning properties of the liquid electrolyte phase accompanied by pronounced apparent slip at smooth walls. The addition of particles to the electrolyte further modifies the flow behavior and distinct thixotropic properties are observed.
Here we present a macroscopic, homogeneous approach to simulate the complex suspension flow based on the single-phase Navier-Stokes equations. The time-dependent rheological behavior of the bulk fluid is modeled with a structural kinetics approach, where the local history- and shear-dependent state of the material structure is described with a transported dimensionless parameter. The wall-fluid interaction, i.e. the wall slip, is modeled with a modified non-linear Navier slip law. Instead of trying to individually measure the isolated rheology and wall-slip parameters with a series of tailored experiments, measured velocity profiles were used as a reference to calibrate the models. Due to the opacity of the fluid, an ultrasound flow mapping technique (Ultrasound Imaging Velocimetry) was applied to measure two-dimensional flow fields in a reference experiment. The numerical model in combination with the obtained parameters reproduce the flow characteristics reasonably well for different flow conditions, especially the extended stagnation zones observed for the thixotropic suspension. The presented numerical models could consequently be used to investigate comparable fluid flows for relevant applications and identify critical flow situations. For example, in flow batteries with suspension electrodes stagnation zones must be avoided to prevent blockage of the flow channels due to build-up of solid residue from reaction products, e.g. ZnO in zinc-air flow batteries. Generally it is desirable to avoid such zero-flow zones in piping and fluidic equipment for complex (multiphase) fluids, which increase pressure drops and can possibly lead to sedimentation and ultimately blockages.

The cavitating flow through an axisymmetric nozzle was simulated using an Euler-Euler, an Euler-Lagrange, and a multi-scale Euler-Lagrange method. The Euler-Lagrange approach and the hybrid multi-scale Euler-Lagrange approach were implemented into OpenFOAM. In the Lagrangian methods, interactions between vapour bubbles and the liquid phase were considered in form of a two-way coupling. Main components of the Lagrangian approach - bubble dynamics and bubble motions - were verified and validated. The multi-scale Euler-Lagrange method captured large vapour volumes on an Eulerian frame, while small vapour volumes were treated as spherical Lagrangian bubbles. Procedures to transform vapour volumes between the Eulerian and the Lagrangian frame were verified. Numerical results of the flow through an axisymmetric nozzle were investigated in terms of spatial discretisation for the Euler-Euler approach and in terms of cavitation nuclei distribution for both Lagrangian based approaches. While the Euler-Euler method used information from the flow solution to assess cavitation-induced erosion, the multi-scale method predicted erosion based on collapse dynamics of Lagrangian bubbles. Numerical erosion predictions were compared to experimentally measured erosion patterns.

Cavitation in injection systems can lead to erosion and failure. However, cavitation is desired for an enhanced fuel jet breakup into small droplets leading to a more homogeneous combustion. We investigate both cavitation erosion and primary breakup in Diesel injectors with two different numerical methods. For both methods the needle movement corresponding to one injection cycle is prescribed as moving wall boundary condition.

The first method is tailored to the detection of erosion sensitive wall zones. It is a density-based, compressible flow solver with a non-viscous flux formulation for cavitating flow. Time integration is performed with a four stage low storage explicit Runge-Kutta scheme. This combination enables the simulation of wave dynamics and resolution of shock waves, which are crucial for erosion. Cavitation is modeled using a barotropic law. Since cavitation is inertia driven, no turbulence model is utilized. By applying erosion indicators at the walls of the injector we have been able to predict partially the experimentally found erosion prone regions for two different ship injector designs.

For the simulation of primary breakup the focus is on accurate calculation of turbulent structures with an LES, which is not possible with the upwind-based method for cavitating flow discussed above. Thus, we use a pressure-based, incompressible flow solver for this purpose. The three phases liquid and vaporous fuel as well as air are modeled with the Volume of Fluid (VOF) method. Phase change due to cavitation is taken into account by the mass transfer model of Sauer and Schnerr. Besides the in-nozzle flow the spray breakup is investigated up to 7 mm downstream of one nozzle hole. Primary breakup of the jet is assessed by detection of Euler ligaments and statistical evaluation of their properties. The LES is performed with the WALE model. The simulation method is validated by comparison with µLIF experimental results in the near nozzle region. A qualitative agreement is found regarding the structure and spray cone angle of the jet. At low lift cloud cavitation occurs, while at high lift string cavitation is observed. The cavitation types are in correspondence with distinct vortex structures and turbulence intensities.

The increasing number of satellites being launched to very low Earth orbits, (VLEO) below 450km, has highlighted a lack of existing knowledge about the influence of atmospheric drag in these conditions. Above 450km, drag forces are often neglected, due to the low density of the atmosphere. Below this, ignoring these forces can lead to detrimental effects on mission performance.

The atmosphere in VLEO is rarefied, and continuum flow conditions do not apply. Free molecular flow is assumed. Historically, a drag coefficient value of 2.2 was often used for compact satellites, but more recently, particle methods such as direct simulation Monte Carlo (DSMC) have been employed to give more accurate values. dsmcFoam is an open-source, validated code, which has been used in the past to perform these simulations. It is a proven method for drag analysis. However, it suffers from drawbacks related to computational load and time requirements.

Analytical methods offer an alternative to DSMC, for example through tools such as ADBSat. This program treats each body as being made up of individual panels, for which the drag force can be calculated using analytical methods. The cumulative total of all the panels produces the total body drag coefficient. Although far less time-consuming, this method suffers from some limitations due to its inability to compute multiple particle reflections, which are yet to be thoroughly characterised.

dsmcFoam represents a known benchmark against which to compare ADBSat. This paper compares equivalent test cases from both programs, and uses the results to identify scenarios, satellite geometries, and flight conditions for which there is a clear benefit of using one method over the other. Further validation will be sought from previous experimental results.

The benefits and applications of both dsmcFoam and ADBSat to the field of VLEO satellite flight will be explored. ADBSat is expected to become invaluable for real-time drag coefficient estimation, for which DSMC is prohibitively time-intensive. The DSMC method may be better for in-depth analysis of more complex spacecraft geometries. Finally, the influence that the choice of software will have on the process of designing satellite missions in VLEO will be discussed.

Multi element wing tips consisting of cascaded airfoils are known to have a positive impact on the induced drag. Due to various limitations the principle is less common in technical applications but can be seen from many bird species especially in soaring flight. In the present study a cascaded wing tip model inspired by the wing tip of the white stork has been investigated in order to understand the complex flow in which the individual elements interfere with each other. The flow, which has a relatively low Reynolds number, features laminar and turbulent regions, massive separation and reattachment. The local inclination of the individual elements is varying in spanwise direction. Thus the regions of separation strongly depend on the angle of attack. Therefore the general flow structure changes between several states.
In order to understand the sensitivity to modeling approaches, in particular with respect to turbulence modeling, results from simulations with different RANS turbulence models will be compared for steady and unsteady simulations at several angles of attack. Since the flow encounters transition from laminar to turbulent state while passing the airfoils the a transition prediction model has been taken into account. However, in most cases this seems irrelevant since the transition occurs within separated shear layers and not in attached wall boundary layers, where the transition model is acting.
Finally, the results will be compared qualitatively with wind tunnel measurements, which have been performed on a mounted stork wing. These comparisons cannot be made quantitatively due to differences between the mounted stork and the idealized and simplified CFD model. However, the comparison of velocity distributions in the wake flow allows for a basic validation of the CFD simulation setup.

Shear flows are a common feature in both natural and engineering environments. In the presence of certain statically-stable density stratifications, shear flow will become unstable and develop into Kelvin-Helmholtz instabilities which break and become turbulent. Results of direct numerical simulations of shear instabilities will be presented in this talk. Time-resolved and time-averaged flow fields and domain energy spectra will be compared across a number of different flow Reynolds numbers and initial Richardson number combinations to examine the DNS capabilities of OpenFOAM. Topics covered will include the generation of initial turbulence including active and passive scalar variance within the domain as well as turbulence advected into the domain through inlet boundary conditions.

In the context of external aerodynamics and vehicle simulation using Computational Fluid Dynamics (CFD), it is particularly challenging to predict the flow around a tire due to the high level of geometric complexity and the interaction of the rotating tire with the stationary ground. Though there are various methods that attempt to find a solution to this problem in the context of CFD analysis. In the present work, a new generalized moving reference frame (GRF) method is used for capturing the effect of the lateral grooves of the tire together with an AMI approach for the simulation of the rim. The method tackles many issues present in the conventional MRF approach which can lead to a slower solution convergence and, most importantly, allows for the use of non-circular reference frames. This property makes application to complex geometries much more tractable. The method can also automatically detect cells that should be included in the generalised reference frame without requiring extensive user input. This is in contrast with conventional MRF methods, where the explicit specification of a rotationally symmetric volume in the form of a closed surface description is typically required.
In the case of the isolated wheel study, the ventilation moments between the sliding mesh and the rotating wall approach are nearly equivalent when dealing with a slick tire geometry (as expected). This is not the case for a detailed tire geometry containing lateral grooves. In the grooved case, the rotating wall approach fails to predict the ventilation moment produced by the detailed tire’s features. In contrast, the GRF method managed to capture the rotational effects much more accurately, albeit with some deviation from the ideal solution represented by the SM results. The most compelling evidence for the GRF method’s utility comes from the full vehicle comparison. Using GRFs to account for the grooves results in a correct prediction of the direction of the trend in 𝛥𝐶𝐷 and a respectable magnitude error of just over one count (0.0012) or 30%. This is substantially better than the RW approach which gets both the direction and magnitude of the trend wrong.

Computational fluid dynamics (CFD) has the prediction capability to influence decisions early in the design process. In order to assess the state-of-the-art of CFD and its prediction capability in the medical devices industry, the U.S. Food and Drug Administration (FDA) proposed two benchmark models for validation. CFD submissions from 23 participants were also made available [1]. The focus of this work is to explore hybrid and multi-block structured meshing strategies and their impact on solution accuracy for one of the benchmark cases, a centrifugal blood pump. Designed to operate across a large range of flow conditions, the blood pump is geometrically simple, consisting of four blades attached to a rotor base and shaft.

The meshing strategies considered include hybrid viscous unstructured and multi-block structured meshes. Highlighting meshing best practices, an emphasis is placed on mesh quality characteristics such as included angles, skewness, and orthogonality.

For the hybrid viscous unstructured mesh, an advancing normal technique starting from a quad-dominant surface mesh generates the required near-wall resolution. During the viscous extrusion process, elements are subject to quality checks allowing the front to stop locally if necessary to improve overall cell quality. The final front of the viscous extrusion transitions to an isotropic tetrahedral off-body mesh, completing the computational domain surrounding the rotor. Using a structured O-H topology at the interface between the rotor housing and pipes, path-based algebraic extrusions create the volume mesh within the inlet and outlet pipes.

To create the multi-block structured mesh, spatial dimensioning criteria used in the unstructured process served as guidelines for the surface and volumetric blocking topology development. Each block of the point-matched structured grid used transfinite interpolation for initialization and elliptic smoothing where required.

Caelus v8.04 [2], a derivative of OpenFOAM, was used to perform the simulations. In order to compare results with Malinauskas [1], a steady-state moving reference frame (MRF) simulation was run using the realizable κ-ε turbulence model. Shown below is the two-dimensional profile of the velocity magnitude |U_xy| at a pump condition of 6 L/min and 3500rpm.

Modelling of filter applications is characterized by the coverage of a wide range of length scales, from single fibres and single particles up to complete filter houses in plant engineering.

For the simulation of filtration applications on a macroscopic scale, the OpenFOAM toolbox already provides a set of useful numerical models, for instance porous zones or porous baffles to realize the flow resistance of filter media.

We developed a number of suitable OpenFOAM solvers and appropriate numerical models to meet the requirements of an extended application range.

The numerical solvers and models we implemented so far cover pure depth filtration, cake filtration of the combination of both. For the depth filtration fixed 3D cell zones can be used, with models for deposition probability and penetration depth. The cake filtration is based on dynamically growing 3D cell zones. Continuing towards the macroscopic scales the filter media are modelled based on thin-shell cyclic patches, which have much lower requirements on the geometrical resolution of the filter elements, and which store additional parameters like filter media thickness or deposited mass on a cell face level.

As laws of resistance the Darcy- and Forchheimer models are implemented, for the mesoscale application the Ergun equation and its variants.

The focus on the Lagrangian particle tracking as the key element of the dynamic filtration modelling requires an improved parallel processing of the steady state particle tracking algorithm. Additional stall detection methods for the particle movement prevent deadlocks. Local adaptive time stepping methods improve the accuracy of the particle trajectories.

The numerous numerical models, solvers and utilities shall be combined into a comprehensive framework for the modelling of filtration applications.

The benefit for the user is in particular the combinability of the different models for his specific application. With a unified case structure and solver control the error rate can be reduced, as well as the training time, while the scope of applications is extended.

For our developers the software maintenance and the support will be simplified. For the extension of existing models and the addition of new models the framework will provide suitable templates.

In this study we investigate shock wave-boundary layer interactions of high enthalpy Mach 7 reacting flow on a double wedge model with fixed fore- and varied aft-angles. Both 2-D and 3-D results were conducted for thermally equilibrium conditions. Viscosity of each specie is calculated by the formulation offered by Blottner and the mixture viscosity is found out by Wilke’s formula. Thermophysical properties of each specie were determined by JANAF thermodynamic package. Backward and forward reaction rates are calculated from Arrhenius Law. For high enthalpies, we modeled air with 5 species including 19 chain reactions. Numerical computations in this study were performed by a newly coded open-source Navier-Stokes solver, which was developed within the framework of OpenFOAM. Validation processes of newly- developed code for reacting and non-reacting two dimensional cases were completed. We compared our results with numerical and experimental results available in the literature. The solver is tested for thermally non-equilibrium cases. Numerical Schlieren images of our 2-D and 3-D results were compared. Surface heat flux distributions that are available in the literature were used to validate our solver.

The main process in wastewater treatment usually consists in a biological treatment for organic matter removal, but, if additionally, nutrient removal is required, a more advanced treatment process is needed. However, conventional configurations for Biological Nutrient Removal (BNR) entail complex treatment systems, which imply a significant increase in the volume needed and energy consumption compared to only organic matter removal processes.

In that context, the Group of Environmental Engineering of the University of Cantabria (Spain) has patented several biological reactors with the aim of fulfilling the mentioned challenges. Concretely, AnoxAn (Tejero et al. 2010) unifies the anaerobic and anoxic zones of the conventional BNR configurations in a single reactor. However, the multi-environmental behaviour (hydraulic separation must be maintained between anaerobic and anoxic zones) and singular elements configuration (mixing devices and baffles) in AnoxAn entail a complex hydrodynamic operation that must be deeply studied.

At this aim, a 3D Computational Fluid Dynamics (CFD) based model is built in OpenFOAM® (v1812) and successfully validated using three different Residential Time Distribution (RTD) experimental tests (pulse and step tracer tests). Turbulent flow is simulated with the standard k-ε model and tracer transport is reproduced by means of a passive scalar transport equation.

Besides, specific modelling approaches for singular elements are validated, achieving significant computational cost savings. Concretely, the baffle between anoxic and clarification zone is modelled as a one phase porous media and the impeller with a plane disk approach that reproduces the three main velocities in rotational motion (axial, radial and tangential).

A two-stage numerical resolution is implemented: (i) First, hydrodynamic steady state solution is reached using SIMPLE algorithm and, then (ii) transient resolution for tracer transport is performed using PIMPLE algorithm. The transient solution is essential for having high spatial flow and tracer concentration resolution at every time of each RTD experiment simulated.

With the results obtained with the transient simulations, an exhaustive hydrodynamic analysis is performed. As a result, dead volumes and short circuiting zones are tracked, quantified and studied. Finally, singular elements influence in the hydraulic performance of AnoxAn is also analyzed.

Adjoint optimisation methods are in use for a wide spectrum of applications and industries. One of the emergent bottlenecks in their integration into ever more complex engineering processes is the lack of functionality for automatic robust design; the ability to create a shape that is the optimal compromise across multiple operating points and flow conditions.

ENGYS has further developed its HELYX-Adjoint solver tool box to be able to address this bottleneck. For a given geometry multiple operating points are considered. This includes different boundary conditions, e.g. flow rate and even varying numbers of active inlet and outlets. All operating points are given weights and the primal and adjoint solutions for all are then combined into a single solution. Thus, a valid robust design optimisation process can be followed.

The system extensions required for multi-point also allows for the optimisation of different separate geometries in the same design space in a simultaneous process. An additional optimisation objective to avoid overlap of different designs was added. It is now possible to for example optimise different disconnected ducts occupying the same design space taking into account the above-mentioned robust design approach as well, hence creating the best routing of all ducts with respect to the optimisation objectives.

The new functionality will be demonstrated in detail on a simple testcase and more complex applications will be presented.

In the nuclear industry, SFP(Spent Fuel Pool) is liquid-filled rectangular tank with spent fuel assemblies inside, without top cover. The liquid, water, is partially filled within the tank and limits rising temperature induced by spent fuel assemblies as a coolant. In case of seismic circumstance, SFP could be affected by sloshing phenomena and sloshed liquid could overflow over side walls. The amount of overflowed liquid should be provided into the tank by additional facilities. Whereas, in case of severe circumstance such as power failure, devastating secondary accident could occur. In order to prevent these accidents, the equipment or the facility which is not affected by power failure should be developed. Therefore, estimate of the amount of overflowed liquid and behavior of the overflowed free surface need to be investigated. In this study, the amount of overflowed liquid and behavior of the overflowed free surface in the liquid filled rectangular tank under seismic excitation are investigated by numerical simulation and experiments.

Applications involving porous media are increasing day by day and new ways to model porous materials are also being developed. While the traditional porous applications have largely remained confined to low speeds through porous matrix, lately, high speed applications like transpiration cooling for hot structures are gaining attention. Thus, need for numerical solvers that can treat porous media realistically is growing.

Current treatment of flow in porous materials in OpenFOAM involves an approach to add viscous and inertial resistance terms in the Navier–Stokes momentum equation in the form of an external source where porous regions are defined and solving the equations either explicitely or implicitly. It also treats the porous flow with the assumption of local thermal equilibrium (LTE). Additionally, the porosity term is neglected for the steady state solvers like rhoSimpleFoam or rhoPorousSimpleFoam since the governing equations do not involve the porosity influenced temporal derivative. Thus a new solver is being developed in order to have a more realistic simulation of the flow through porous media. This new porous solver is aimed at consideration of transient compressible flow as well as local thermal non-equilibrium (LTNE) for porous geometry. Based on the Darcy-Forchheimer formulation, it solves the modified momentum and energy equations instead of adding an external source to classical momentum equations. This solver, therefore, is intended for purely porous geometries involving only one kind of porous material. While thermophysical model inherent to OpenFOAM framework is standard way to include species properties, the solver currently uses a custom dictionary for this purpose. A couple of test cases are also presented for comparison with existing solvers mentioned above. Currently the solver is in its basic stages and there is a lot of scope to improve it. Efforts are ongoing to use thermophysicalmodel class as well as create a more robust code.

Design optimisation using Computational Fluid Dynamics (CFD) often requires extremising multiple (and often conflicting) objectives simultaneously. In such cases, usually there is no unique solution, but a range of solutions trading off between the objectives. The set of solutions optimally trading off the objectives are known as the Pareto set, and in practice only an approximation of the set may be achieved. Multi-Objective Evolutionary Algorithms (MOEAs) are known to perform well in estimating the optimal Pareto set. However, these often require thousands of function evaluations, which is impractical for computationally expensive evaluations. An alternative approach is Multi-Objective Bayesian Optimisation (MOBO), which has been proved to be an effective approach with limited budget on function evaluations [2]. In the present work, a recently developed MOBO approach [3] will be utilised to optimise a tube-bundle heat exchanger design using CFD to evaluate its performance. For this purpose, an open-source Python-based framework has been developed [1] to automate the optimisation procedure utilising OpenFOAM-4.x (automated using PyFoam) to perform the CFD evaluations. It should be noted that for a more realistic industrial application, constraints would be applied to account for manufacturability limitations. To allow for a comprehensive investigation into the efficacy of the proposed framework, and the performance of MOBO in engineering design, a number of heat exchanger design scenarios (cases), typically observed in industry, will be presented.

Acknowledgments

This work was supported by the UK Engineering and Physical Sciences Research Council [grant number EP/M017915/1].

References

[1] S.J. Daniels, A.A.M. Rahat, R.M. Everson, G.R. Tabor, and J.E. Fieldsend. A suite of computationally expensive shape optimisation problems using computational fluid dynamics. In International Conference on Parallel Problem Solving from Nature-PPSN XV, Part II, pages 296–307, 2018.

[2] S.J. Daniels, A.A.M. Rahat, G.R. Tabor, J.E. Fieldsend, and R.M. Everson. Redesign of industrial apparatus using multi-objective bayesian optimisation. In Tenth International Conference on Computational Fluid Dynamics (ICCFD10), Barcelona, Spain, July 9–13, 2018.

[3] A.A.M. Rahat, R.M. Everson, and J.E. Fieldsend. Alternative infill strategies for expensive multi-objective optimisation. In Proceedings of the Genetic and Evolutionary Computation Conference, GECCO ’17, pages 873–880, New York, NY, USA, 2017. ACM

In the field of heterogeneous catalysis, a detailed understanding of the macroscopic mass transport and the hydrodynamic dispersion is highly important for process intensification. In the present work, we propose a new OpenFOAM® solver named disTrackFoam that utilizes the Lagrangian Random Walk Particle Tracking Method [1] to simulate the convective and diffusive mass transport of ideal, massless tracer particles as concentration equivalents to calculate important reaction engineering characteristics such as dispersion coefficients, residence time distribution and tortuosity.

With this approach, numerical investigations in the industrial and technical relevant region of large ratios of convective to diffusive mass transport (high Péclet numbers) are feasible without the occurrence of common problems regarding numerical dispersion as well as oscillation. On the basis of these findings it is for example possible to optimize the geometrical parameters (cell size, unit cell type, strut diameter) of additively manufactured porous open cellular catalyst support structures.

Besides the implementation of the particle tracking equation, boundary conditions for the handling of particle contacts with solid surfaces specially suitable for the physically correct treatment of massless and diffusive tracer particles were introduced.

Additionally, based on the cyclicAMI boundary conditions, a procedure to calculate and track the particles actual position in periodic simulation domains was implemented, reducing the computational costs in these cases significantly.

Furthermore, we focused on user friendly and individually adjustable pre- and post-processing features such as the disTrackPositionCreator, which is a specially developed OpenFOAM® utility to generate the kinematicCloudPosition file for the start positions of particles in complex porous materials based on the domain's polyMesh information.

For validation,the numerical results of a pure one-dimensional diffusive mass transport experiment (Dirac delta distribution) as well as the three-dimensional diffusion superimposed by a convective flow (Taylor-Aris dispersion) are compared to their analytical solutions, showing excellent accordance.

Afterwards, using the additively manufactured so-called Periodic Open Cellular Structures (POCS, [2, 3]) as a demonstration case, the disTrackFoam solver is able to prove its potential to investigate the macroscopic mass transport especially in the field of chemical reaction engineering.

[1] Freund et al. 2005.

[2] Inayat et al. 2011.

[3] Klumpp et al. 2014.

The lean combustion is one of solutions for realizing the highly-efficient internal combustion engine. In the development of the lean combustion technique, the combustion stability is one of the huge problems. In this context, the ignition system is considered as one of important components to improve the combustion stability. To develop the ignition system for the lean combustion, we need to investigate the detail of physics in the ignition, which may involve many complex physical phenomena such as breakdown, discharge, turbulence and chemical reaction. In this work the spark model is introduced in OpenFOAM. The spark model is based on the AKTIM (arc and kernel tracking ignition model) by Duclos(2001), it composes of following four submodels: (i) electric circuit model, (ii) discharge channel model, (iii) restrike model and (iv) flame kernel model. The current ignition model also employs Masuda(2017)’s approach, in which both the discharge channel and the flame kernel are treated by lagrangian parcels; the discharge channel parcel move according to the flow, and flame kernel parcels grow up depending on the discharge energy and the turbulent flame speed. The premixed flame propagation model is also implemented by the author, based on the hyperbolic tangent approximation of the combustion progress ratio expressed by the mass fraction of species. The lagrangian ignition model is coupled with the combustion model; when the initial flame kernel grows up to a certain size, the reaction state is passed from lagrangian parcels to the computational mesh. The
current research is the validation study of the recent ignition model; the flame kernel growth were validated using the experimental results obtained by Herweg(1992) and shows a good agreement, while the coupling procedure of the current ignition model and combustion model were validated using experimental data of combustion engine with different air fuel ratio (A/F) conditions, and the in-cylinder pressure time history obtained by the simulation shows a good agreement with the experimental results.

A numerical study for simulating heat transfer between the reacting gas in a porous catalyst bed, reactor tube, and coolant in a Shell and Tube reactor for an exothermic reaction. The open source toolbox OpenFOAM is modified to solve such a problem by taking chemical reaction and multi-phase heat transfer into account. The Shell and Tube reactor simulated here is a 2-meter long, 28-millimeter diameter reactor, while the catalyst packed region is 1.5 meter in the center. The catalyst bed is assumed to be a continuous porous medium. The results show that the solver captures the trend of temperature profiles along both axial and radial directions. The maximum catalyst temperature predicted by this solver agrees with the measured temperature.

During a severe nuclear reactor accident, large amounts of hydrogen, steam and other non-condensable gases are released into the reactor containment. Hydrogen gas can accumulate in significant proportions in some parts of the containment, and subsequent hydrogen combustion (like in Fukushima Daiichi NPP) may result in structural failures, which may allow the release of harmful radioactive material into the environment. Hence, it becomes essential to predict hydrogen and steam distribution in containment for the design and assessment of effective hydrogen mitigation methods.

Ongoing work at Forschungszentrum Jülich aims at developing a dedicated solver ‘containmentFOAM’ which has the capabilities to handle multi-species turbulent gas transport, conjugate heat transfer, aerosol/particle transport, and radiative heat transfer. Besides the solver, a comprehensive model library is being built. The ‘reactingParcelFoam’ solver available in OpenFOAM-6 is chosen as the base solver for ‘containmentFOAM’ since it has the capabilities to handle multi-species gas transport, aerosol modeling and radiative heat transport.

In the application mentioned above, gas mixing phenomenon occurs among gases with significant difference in specific heat capacities (C_P of H₂ = 14.32 kJ/kg.K, C_P of Air = 1.01 kJ/kg.K, C_P of H₂O = 1.864 kJ/kg.K). The base solver, ‘reactingParcelFoam’ has deficiencies in modeling such multi-species gas transport phenomenon and produces incorrect temperature field even on isothermal cases. Multi-species gas mixing modeling involves a set of tightly coupled system of equations (continuity, momentum, the mass fraction of individual species and enthalpy). Incorrect enthalpy transport equation results in a wrong temperature field and hence an incorrect density field, eventually leading to solver instability and poor results.

The species transport equation was extended to account for molecular and turbulent diffusion. The energy equation was modified to account for (i) enthalpy transport due to inter-species diffusion (ii) varying specific heat values of different species. The overall solver stability is improved by introducing an additional density corrector step in PIMPLE algorithm. The paper discusses the successive modifications and highlights their effects systematically based on simple 1D cases. Finally, the applicability of the solver is demonstrated on a 2D mixing layer case.

In this paper, a new method for applying boundary conditions in OpenFOAM, is presented. In this new framework, standard boundary conditions (fixedValue, zeroGradient etc.) can be applied on new immersed boundaries which are constructed from existing faceZones and produce identical results to the standard boundaries. This framework is powerful in applications with moving parts such as rotating gears, FSI, multiplase applications etc. Currently, in these applications, the mesh motion algorithm moves the boundaries until re-meshing is required. This method, in most industrial applications, is inefficient or even impossible. Using GIB, the point coordinates of the faces near the interface are snapped on the interface. After the snapping, a faceZone which contains the faces which are located exactly on the interface is constructed. A new boundary is created based on the faceZone and boundary conditions are applied. The matrix contributions of each OpenFOAM operator (fvm::, fvc::) using GIB and body fitted meshes are the same which guarantees that the results are identical. The implementation is generic and no additional numerical schemes or executables are required. In this paper we focus on the GIB coupling with the multiphase solvers (interFoam family of solvers). Special care is taken in the non-standard FV operators such as MULES. Finally, results are shown in multiple applications.

A 2D-3D coupling solver for free-surface fluid numerical simulations has been implemented by Mintgen and Manhart (2017) based on OpenFOAM framework. The coupling solver shallowInterFoam combines the 2D Shallow Water solver shallowFoam and the 3D Navier-Stokes solver interFoam providing good assessments with reasonable computational costs. All of these solvers are free to use in the Finite Volume Open Source code OpenFOAM; shallowFoam is published on the gitHub platform.

The Dirichlet-Neumann decomposition method is employed in the coupling algorithm where the information from the non-overlapping sub-domains (i.e. 2D regions and 3D regions) exchange at the common boundary, the 2D-3D interfaces. As a result, the 2D-3D coupling requires strictly a perfect match of the mesh in horizontal directions (i.e.x and y directions) at the 2D-3D interface.

Currently, only cases with regular bottom have been tested; where the coupled meshes are created by using blockMesh, a mesh generation utility supplied with OpenFOAM. For these simple geometry cases, the coupling solver produces accurate results with less computational efforts comparing to the full 3D solver (Mintgen and Manhart, 2017).

When creating complex geometries, another mesh generation tool snappyHexMesh can be used. It generates three-dimensional body-fitted meshes by morphing the refined split-hex background meshes to the object surfaces. However, the process of mesh generating using snappyHexMesh leads to unmatched meshes at the interface. Some cell faces are ill-grouped to the wrong boundary patch as a result of the surface feature extract in the meshing process.

To apply the coupling solver in wider application scenarios textit{e.g.} flood event in urban areas, mesh generation for complex geometries and irregular bottoms are demanded. The present study introduces a mesh re-regulation method that the coupling solver uses. The proposed method reports all ill-grouped faces in the first place. Furthermore, these ill-grouped faces will be set to the appointed boundary. In this manner, the coupling solver is able to deal with more complicated geometries.