Comparison Of Different CFD Techniques For Transient .

15th International Conference on Environmental Science and TechnologyRhodes, Greece, 31 August to 2 September 2017Comparison of different CFD techniques for transientmodelling of virtual breathing thermal manikinsMijorski S.1, Ivanov M.21PhD, SoftSim Consult Ltd., Consultant at Technical University of Sofia, FPEPM, Department: ”Hydroaerodynamics andHydraulic Machines”, Sofia 1000, Bulgaria2Senior Assist. Professor, PhD, Technical University of Sofia, FPEPM, Department: ”Hydroaerodynamics and HydraulicMachines”, Sofia 1000, Bulgaria*corresponding author: Martin Ivanove-mail: m ivanov @tu-sofia.bgAbstractThe presented paper reveals comparative analyses of threedifferent CFD based transient modelling techniques(URANS, DES and LES) for flow simulations with virtualthermal manikins. The interaction between the simulatedbreathing flow and the free convection flow from theheated virtual manikin’s surface for two breathing phasesis performed under controlled room conditions. Recentstudies of the authors’ show that simulations under steadystate conditions can lead to overprediction of the resultantfields, so the implementation of transient simulationmethods is recommended in that case. Qualitative analysesbetween the different techniques are made, in terms oftemperature and velocity fields’ comparison. Consideringthat these virtual thermal manikins are modern complextools for virtual design and assessment of the occupants’thermal comfort, as well as for virtual analyses of indoorair quality, the results achieved in the present study willprovide new and valuable approach for the integration ofvarious modelling techniques in the presented area. TheCFD results has demonstrated a flow pattern similarities inboth DES and LES solution methods, while in URANSsimulations it was observed a deflection of the thermalplume with almost 0.4 m at 2.5 m height from the floor,under free convection conditions. Nevertheless, for theexhale phase of the breathing cycle, there was a goodcorrelation between the different techniques, in thebreathing zones of the manikin.Keywords: CFD, URANS, DES, LES, Virtual ThermalManikin, Breathing Flow Modelling1.IntroductionThe breathing thermal manikins are considered as veryimportant tools in the environmental engineering practice.They represent accurate models of the human body andallow simulation of different levels of physical action, aswell as some human activities such as breathing, sweating,sneezing, coughing and others. They are very complexinstruments and are used also to study the convective flowsaround human bodies in different conditions, withoutexcessive risk of exposure to the people themselves[Madsen T. (1999)]. However, experimental studies withreal thermal manikins are expensive, time consuming,require highly skilled labour and are relatively difficult toconduct. Therefore, the use of Virtual Thermal Manikins(VTMs), particularly at the design and prediction stage ofthe indoor environment, seems to be appropriatealternative to the actual thermal manikins’ experiments.Implementation of diverse Computational Fluid Dynamics(CFD) based simulation techniques is required to assessthe performance level and accuracy of the creatednumerical models. Recent findings suggest that simulationsunder steady state conditions may over-predict theresulting parameter’s impacts in the “breathing flowmodelling” area [Ivanov and Mijorski (2017)]. It issuggested also that implementing transient conditions inthese simulations will further improve the developedmodels of breathing VTMs. But implementing transientconditions in breathing flow and free convection flowmodelling is not an easy task, and may require significantmodelling skills and computational resources. The study ofVillafruela et.al. (2016) summarizes that in the recentyears, CFD simulation tools have been used to study thediffusion of inhaled and exhaled contaminants fromdifferent ambient environments. Also, complex airborneinfection routes have been studied with CFD, including thesneezing and coughing mechanisms [Villafruela et.al.(2016)]. Steady-state breathing flow simulation study hasbeen previously performed by the authors, to analyse theinteraction between the inhaled/exhaled flow and thethermal plum characteristics, around the body of a thermalmanikin [Ivanov and Mijorski (2017)]. Considering itsnature, the breathing process in humans is transient, andthe breathing flow parameters are changing completely andrapidly in relatively short period of time as demonstratedby the authors in [Ivanov and Mijorski (2017)]. This makesdifficulties in the definition of boundary and initialconditions. But for accurate modelling, the breathingprocess should not be considered as steady-state andconstant inhalation/exhalation velocity should not beassumed. Moreover, the convective flow around thethermal manikins is buoyancy driven, associated with highnonlinearities and fluctuations. In order to capture theseeffects in the presented study, three different transienttechniques were implemented during the analysis: Unsteady Reynolds Averaged Navier Stokes(URANS): represents time-averaged equations of fluidCEST2017 00415

flow motion, decomposing the flow into its time-averagedand fluctuating quantities;2. Detached Eddies Simulations (DES): represents ahybrid method that treats near-wall regions with aReynolds-Averaged Simulation (RAS) approach and thebulk flow with an LES approach;2.1. 3D Models Large Eddies Simulations (LES): represents amethod, where large turbulent structures in the flow areresolved by the governing equations, while all the smallereddies filtered by the Sub-Grid Scales (SGS) are modelled.The main aim of the presented paper is to performcomparison analyses of the described different CFDtechniques for transient modelling of virtual breathingthermal manikins. The obtained results will providesignificant understanding of the solution differencesbetween less computational expensive methods like RASand more advanced ones like LES and DES.3D model, spatial discretization and simulationsetupFor the purpose of the presented study, it is used apolygonal virtual manikin model with appropriate structureand design, which meets the basic requirements of theergonomic design area. The virtual manikin is designed byGeorgi Chervendinev (Engineering Design Lab at TUSofia), in order to match the overall 95th percentile of theanthropometric size of a standard person. It has anapproximate surface area of 2 m2 and height of 1.75 m.The nasal valve opening is constructed according to thestudy of Lin (2015) and is shown in Fig. 1. The openingarea of is 7.3 x 10-5 m2. The normal to the nasal openingwas specified to 45 degrees from the vertical body axis.Additionally, exhaust walls from the nasal valve to thenose end were inclined to 15 degrees according to thestudies of Nilsson (2006) and Lin (2015).Figure 1. 3D model and CFD boundary conditions2.2. Spatial discretizationFor the purpose of the comparison analyses between thedifferent CFD techniques, the computational domain wassignificantly reduced by introducing two symmetry planes,at the both sides of the manikin. This measure wasconsidered with clear understanding of the flow restrictionsthat will be introduced, but with an aim of simulation timereduction. As shown on Fig. 1, the model comprised just0.1 m section of the manikin, including the two nasal valveopenings, head and manikin body, excluding the hands andlegs. The 3D model of the polygonal thermal manikin isplaced in the context of rectangular shaped room. Thediscretization was done with snappyHexMesh utility, partof an ENGYS (www.engys.com) enhanced version of theCFD code OpenFoam (www.openfoam.com). As a result,the numerical grid was with total of 670 000 poly-meshcontrol volume elements. The base cell size was defined to2.5 10-2 m and to capture the nasal valve zones geometricalfeatures the maximum level of cells refinement reached to8 10-4 m. Additionally the manikin’s surfaces were refinedwith the introduction of prism layers. By first layer heightof 0.4 10-4 m, the y values over manikin surfaces werebelow 4 as recommended in the work of Spalart (2001).Thus, the models matched the basic requirements of LESmethod for resolving accurately the turbulent flow overobject surfaces.2.3. Solution MethodsThree unsteady/transient 3D simulations were performed,covering two different phases of the human breathingcycle. These included no breathing or the free convectionflow case for time duration of 20 seconds (allowing to getthe solutions to a fully developed convective flow aroundthe manikin and room space) and sequentially 2 seconds ofexhale phase. In order to keep maximum Courant numberbelow 1, the simulations were run with different time steps,as for former case it was set to Δt 1 10 -3 seconds, whilefor the lather it was reduced to Δt 25 10-5 seconds. Thesolver for the transient buoyant turbulent flow ofincompressible fluids buoyantBoussinesqPimpleFoam isused with combinations of PIMPLE algorithm forpressure-linking. The PIMPLE represents a merged SemiImplicit Method for Pressure-Linked Equations (SIMPLE)and Pressure Implicit with Splitting of Operator (PISO)algorithms, thus offering an improved solution and fasterconvergence for the transient solutions. The URANSsimulations were run with the Shear Stress Transport(SST) k-ω turbulence model initially proposed by Menter(1993). The model combines the k-ω approach in the innerparts of the boundary layer, but also switches to a k-εapproach in the free-stream regions of the computationaldomain. More details of the selected URANS turbulencemodel are given in the work of Menter, (2011). For theDES simulations, the Spalart-Allmaras turbulent modelwas implemented. Initially, the standard Spalart-Allmarasmodel was proposed by Spalart and Allmaras (1994), andthen its DES formulation was proposed by Shur et al.(1999). The model uses the distance to the closest wall asCEST2017 00415