The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.1 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
CFD for atmospheric dispersion prediction in close range B. Farges, P. Brocheny, C. Thelier, N. Goreaud AREVA NP RESTRICTED AREVA The information in this document is AREVA property and is intended solely for the addressees. Reproduction and distribution are prohibited. Thank you Engineering & Projects Organization
Outline of the presentation 1. Context 2. Methodology setup and validation 2a. Hydraulics 2b. Dispersion 3. Industrial application 4. Conclusion and outlook The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.3 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
1. Context (1/2) Some accidental or incidental scenarios can lead an industrial site to release gases or particules in the atmosphere Depending on the concentration, of the weather conditions and of the quantity, it will form a plume of a given size / concentration Utilities must demonstrate that these scenarios are properly handled and that the consequences are acceptable � For the public outside the facility fences => long range � More and more : for the workers on site => close range Some tools exist on the market dedicated to these issues, but � No or very limited capacity for detailed representation of a given industrial site (buildings, complex obstacles…) The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.4 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
1. Context (2/2) AREVA has a long history in using CFD � Variety of problems � Trained engineers � Computation power available � Validation file on numerous applications � CFD used as a support to safety analyses For this reason : choice to use CFD for the modelling of near- field atmospheric dispersion � Detailed representation of site geometry (buildings…) � Recirculations, turbulence, buoyancy,… taken into account The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.5 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2. Methodology setup Approach Step by step : growing complexity a. Hydraulics only : • On an empty domain • With a single building • Benchmark against litterature b. Addition of dispersion : • On an empty domain first • Validation against the Prairie Grass experiment c. Addition of reacting flows (not part of presentation today) : • Possibilities to account for chemical reactions / power decay during dispersion Illustration on the sizing of protection on an AREVA facility The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.6 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2.a. Methodology setup Hydraulics on empty domain Assumptions � Surface boundary layer � Perfect gas � Stationnary wind profile : Assumtion that meteo variation are slow compared to dispersion transient Computation assumptions � In accordance with best practices Identified through a litterature survey � Large domain dimensions in order to prevent side effects . − ε k � Turbulence model : standard (was found to be the best compromise) � Dispersion in neutral conditions (classical choice : Richards & Hoxey, Hargreaves, Vendel) B.Blocken et al, 2007 Equations of velocity , temperature and turbulent quantities profile relative to a neutral atmosphere prescribe at inlet. The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.7 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2.a. Methodology setup Hydraulics on empty domain Profiles prescribed at domain inlet ( from similitude theory from Monin- Obukhov) Atmosphéric profiles ( ) = − ⋅ T z T z Temperature 0 , 0098 0 + u z z * = κ u Profiles 0 ln( ) Velocity z maintained 0 u 3 * Turbulent ε = κ z + z dissipation ( ) 0 Local anomaly u * ² Turbulent typical for = k standard CFD C kinetic energy µ codes Acceptable situation on empty domain The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.8 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2.a. Methodology setup Hydraulics with a single building Behaviour of flow around an obstacle (square building) � Presence of turbulent structures � Classical k-epsilon limitations (consistent with litterature) : Slight Slight underestimation of surestimation of recirculation zone turbulent kinetic energy upstream Slight 10 m surestimation of the recirculation zone downstream Results are consistent with the state of the art and judged acceptable The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.9 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2.b. Methodology setup Dispersion Reference experiment : Projet Prairie Grass (1956) � SO 2 release in an open plain � 68 tests about 10 of them in neutral conditions • Three configurations chosen to validate the methodology Vervecken et al. 2013 Two means to calibrate the results : � Calibration of a parameter : S ct => fast, provides order of magnitudes at the plume centerline � Statistical correction => more complex, more reliable The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.10 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2.b. Methodology setup Dispersion Simulation of the Prairie Grass experiment � Calibrating Sct enables to get the right concentration in the axis of the plume But the shape of the plume is incorrect : overestimation in the axis of the plume and underestimation of its width . (Known phenomenon : Vervecken, Riddle ) Acceptable for an order of magnitude determination. The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.11 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2.b. Mise en place de la modélisation Dispersion Simulation of the Prairie Grass experiment � Accounting of wind oscillations (litterature : Vervecken) • Statistical fluctuations around the average wind direction : a statistical fluctuation with standard deviation σ α has been measured : Fluctuations de la direction du vent (essai n°57) 0,06 Densité de probabilité 0,04 0,02 0 58 90 122 Angle en degré • Turbulence modelling already accounts for some deviation σ m � definition of an additive deviation σ e to reach σ α • Performance of several computations for several angles • Gaussian ponderation of the results The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.12 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
2.b. Mise en place de la modélisation Dispersion Simulation of Prairie Grass experiment � Results with second method • Shape of the plume and concentration of the plume are better predicted • Drawback : several computations are necessary Correction statistique sur les capteurs à 50 m 600 Mesures test 57 concentration de SO2 en mg/m3 500 Simulation 57 - 400 Sct0,7 Simulation 57 - 300 Sct 0,7 - correction statistique 200 100 0 60 70 80 90 100 110 120 Angle en degré The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.13 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
3. Industrial application Milling facility Industrial problem � Ammonia stored on site under pressure � If a leakage happens, need to ensure acceptable doses are not overcome First step � CAD repair inside STAR-CCM+ from site data � Large domain generation � Meshing • Less than 5 million cells for the total volume • Enabling a fast and smooth transient run The information in this document is AREVA property and is intended solely for the addressees. STAR Global conference AREVA - dispersion – March 17-19th 2014 - p.14 Reproduction and distribution are prohibited. Thank you RESTRICTED AREVA
Recommend
More recommend
