INTERNATIONAL JOURNAL OF ENVIRONMENTAL SCIENCES Volume 1, No 6, Copyright 2010 All rights reserved Integrated Publishing Association

Transcription

1 INTERNATIONAL JOURNAL OF ENVIRONMENTAL SCIENCES Volume 1, No 6, 2011 Copyright 2010 All rights reserved Integrated Publishing Association Research article ISSN Comparison of Atmospheric Dispersion in a Large Terrain of Kaiga Atomic Power Plant with Simple Terrain Assumption against actual Complex Terrain Modelling Reactor Safety Division, Bhabha Atomic Research Centre, Trombay, Mumbai, India pa1.sharma@gmail.com ABSTRACT Pollutant dispersion in the atmosphere is an important area wherein different approaches are followed in development of good analytical model. The analysis based on Computational Fluid Dynamics (CFD) codes offer an opportunity of model development based on first principles of physics and hence such models have an edge over the existing models. The present paper is aimed at bringing out some of the distinct merits and demerits of the CFD based models. A brief account of the applications of such CFD codes reported in literature is also presented in the paper. The choice of codes and the features to be employed from within the code for a specific problem needs expertise in thermal hydraulics and numerical techniques. Brief guidelines towards this objective are also covered in the paper. An illustration of use of CFD code for pollutant dispersion studies is also included in the paper which clearly brings out the success and advantage of CFD based approach for modelling complex terrain. Keywords: Atmospheric dispersion, CFD, complex terrain 1. Introduction The rapid growth in technological advancements has led to steep rise in various industries the world over, all aimed towards improving the quality of life. However, the consequent deterioration in the quality of environment also became unavoidable if not altogether uncontrollable. In India, the rising levels of air pollution have now been identified as one of the potential threats to the health of urban population. Besides the pollution threat from the normal industrial operations and other regular polluting sources, the accounts of damages caused by accidental release of harmful chemicals are well known; the Bhopal tragedy in 1986 is only one of such examples. No wonder, the atmospheric dispersion of pollutants in the form of solid particulates, chemical vapours of condensing nature or in gaseous form has been one of the key issues which attracted the attention of researchers even as early as first half of the last century. The modeling approaches for atmospheric dispersions have undergone evolutionary changes over the last 50 years or so and these are seen to be reviewed in open literature from time to time (Turner 1973, Turner 1985 and Wilson 1993). In the last decade, use of the computational fluid dynamics (CFD) based models are also seen to be rising. The high computational powers provided by the latest high end computational machines together with applications of advanced numerical techniques have made it possible to analyse the complex pollutant dispersion studies to a significant details never thought before. However, the use of CFD based techniques could be wonderful tools if employed with proper understanding of associated fluid dynamics and various dispersion mechanisms in the context of problem to be analyzed. The present paper is aimed at bringing out some of the salient features of CFD based modeling of pollutant dispersion studies and its advantages Received on December, 2010 Published on April

2 over the conventional methods adopted earlier. Application of CFD tools for scalar dispersion over a complex terrain of Kaiga with and without taking the complex topography are also discussed in support of the present theme of this paper. 1.1 Existing Models A Brief Account Historically, diffusion theory based calculation used in gas warfare during World War I remained employed for quite some time for the pollutant dispersion studies. A practical approach suggested by F. Pasquill in 1961 was to use Gaussian framework for dispersion, with assigned values of dispersion coefficients derived empirically from curve fits to experimental data and a designation of mixing conditions based on stability scale. Today there are many variations on this theme adapted for rural/urban/coastal/valley perturbations using expressions regressed against additional field or laboratory data. These expressions have bee integrated into large numerical air pollution programs which consider details of local climatology, release conditions, atmospheric chemistry, etc. (Hanna 1982; Turner 1994; Venkatram & Wyngaard 1988; Zanetti 1990). Wind tunnel based modeling has been used for stack modeling. In summary, some of the widely used methods for modeling of dispersion of pollutants in atmosphere can be listed as ; a) Empirical rules, b) Gaussian methods, c) Modified Gaussian Methods, d) Isolated geometry models, e) Street canyon Models, f) Physical Methods and so on. A brief account of some of the prominent model, in no case an exhaustive study, is presented below. Some of the reported models, to name a few, are; Industrial Short Complex Short Term (ISCST3) (USEPA, 1995a), Complex Terrain Dispersion Model plus (CTDMPLUS) algorithms for unstable situations (Perry, 1989), Offshore and Coastal Dispersion (OCD) model (DiCristogara & Hanna, 1989), CALPUFF model (USEPA, 1995b). The ISCST3 model is a popular Gaussian model designed to predict short duration (hours), short range (1 10 km) concentrations of pollutants from industrial sources. The model CTDMPLUS has capability of accounting more complex terrains. The OCD model was developed to assess the impact of offshore emissions on the air quality of coastal region and had special algorithms to account for atmospheric conditions unique to the coastal environment. The CALPUFF model is not a Gaussian model; rather it tracks puffs of pollutants through a temporally and spatially changing atmosphere. The CALPUFF model still uses empirical plume rise formulae and simplified rules to track the pollutants over terrain features such as hills and mountains. Integrated plume and puff diffusion with application of CFD has also been tried (Akira Mori, 2000).The potential shortcomings of these types of models are their empirical nature rather than exploitation of the first principles. It could be erroneous in many complex situations to predict the plume rise based on empirical correlations. Such calculations would further lead to substantial errors in the predicted downwind concentration values. In the case of stagnant environment with isothermal emissions, the Gaussian models can be expected to give a reasonable answer. However, if the plume comes with little initial velocity or the plume is thermal and entering into a real environment the dynamics of the plume induced flow field must be included in the simulation. Simple empirical expressions often include entrainment parameters calibrated for different source characteristics, but these usually do not encompass the regime of large, buoyancy dominated plumes. 2. Computational Fluid Dynamics (CFD) Based Modelling As mentioned above the analysis based on the first principles employing conservation laws of mass, momentum and species should offer substantial improvements in the pollutant dispersion calculations over the so far adopted empirical framework. Such governing set of equations, although highly complex, can be best tools to include even many of the complex 1284

3 issues listed earlier. The CFD based models, therefore have an edge over the existing non CFD based models in several respects. The CFD codes have been widely used to model various types of pollutant dispersion problems. At BARC; CFD codes are in use for quite some time [Sharma et al. 2004; Sharma et al. 2007; Gera et al. 2008; Sharma et al. 2009]. CFD forward computation coupled with signature analysis and neural networks have also been used by author in other paper [Sharma et al. 2010]. 2.1 Kaiga Atomic Power Plant Simulation The CFD tools has been utilised to carry out the dispersion of tracer gas (SF 6 ) at the Kaiga site. The computations were performed with the actual topographical data. The constant wind condition (Initial velocity: U0=2.75, V0=0.49 and velocity boundary conditions: Unormal= 2.71, Utengential= 0.67) have been applied. The Kaiga region topology is highly complex due to the presence of hills and river. The computational domain was 30000m (length) X 30000m (width) x 1500m (height). The source is released from 20000m, 20000m and 50m (height) from the left bottom corner. Total number of grids used in the simulation was 4.32 lakhs cells. The right side boundary was modeled as inlet where the available wind velocity was applied. The actual velocity was available at every 15 minutes. The SF 6 was released at a rate as shown in figure for a design basis accident of loss of coolant accident determined by another nuclear thermal hydraulic code RELAP and used in the present analysis as a time dependent boundary condition. Simulation started with 200 seconds delay to generate the wind field due to the modeled geometry. Simulation was performed for 6 hours. To understand the effect of actual topology Vis a Vis simple terrain computations (Gaussian plume equivalent computations) same computation has been revised by removing the topology obstruction data. The equivalent schematic is shown in figure. Figure 1: Numerical model for complex terrain 1285

4 g/s 1.80E E E E E E+01 g/s 6.00E E E E E Results and Discussion Figure 2: Scalar gas injection time history Figure 3: Mass fraction at 4088 sec (Simple) Figure 4: Mass fraction at 4088 sec Figure 5: Mass fraction at 7464 sec (Simple) Figure 6: Mass fraction at 7464 sec 1286

5 Figure 7: Mass fraction transient with simple terrain at two different locations Figure 8: Mass fraction transient with complex terrain at two different locations Figure 9: Mass fraction at 600 sec at 100 m height (Simple) Figure 10: Mass fraction at 3000 sec at 100 m height(simple) Figure 11:Mass fraction at 6000 sec at 100 m height (Simple) Figure 12: Mass fraction Figure 13: Mass fraction at Figure 14: Mass fraction at 1287

6 at 600 sec at 100 m height 3000 sec at 100 m height 6000 sec at 100 m height Figure 15: Mass fraction at 5992 sec at 50m height (Simple) Figure 16: Mass fraction at 5992 sec at 200m height (Simple) Figure 17: Mass fraction at 5992 sec at 300m height (Simple) Figure 18: Mass fraction at 5992 sec at 50m height Figure 19: Mass fraction at 5992 sec at 200m height Figure 20: Mass fraction at 5992 sec at 300m height A typical study has been carried out to demonstrate the utility of the CFD code in the application for NPPs. Following salient conclusion have been brought out. Complex topography plays a dominant role in atmospheric dispersion of pollutant which is a function of space coordinates, time, wind velocity etc. Comparing Fig 3 with Fig 4 it is clear that the simple terrain assumption gives higher concentration at lower height at hill space where the pollutant actually should not exist in actual conditions and at high altitude it shows no scalars (where it should have significant values). Adjoining valley and hill regions also depicts entirely different atmospheric dispersion pattern (Fig 5 and Fig 6). Comparison of the graphs for two typical receptors shows some interesting observation (Fig 7 and Fig 8). The peak concentration at near to ground and at higher altitude can not be compared (in fact kind of inversion in trends exists). For the remaining mid range values the peak concentration observed with complex terrain are less as compared to simple terrain. Also there is a spread in the time to reach peak concentration values in complex terrain case due to enhanced mixing of scalar. Presence of complex terrain results in the increase in the fall time (from peak to insignificant values) values again due to enhanced mixing. Figs 9 to 14 show the comparison of pollutant concentration for simple and complex terrain at 100 m height at various times. Figs 15 to 20 show the same comparison at 5992 seconds at different heights. From these figures it is clear that dispersion is altogether different in complex terrain as compared to simple terrain Hence the dispersion modelling of complex terrain by adding obstructions of approximate size to a simple terrain will predict the dispersion behaviour that is not accurate. CFD based methods could be a versatile tool which can be used to handle most of the complex situations covering various scales. 1288

7 4. References Comparison of Atmospheric Dispersion in a Large Terrain of Kaiga Atomic Power Plant with Simple 1. Akira, M., 2000: Atmospheric environment, 34, pp DiCristogara, D.C., Hanna S.R., 1989: Volume 1: User s Guide, Sigma Research Corporation, Westford, MA, Report No. A Gera, B., Sharma, P.K., Singh, R.K., Ghosh, A. K. and Kushwaha, H.S. 2008: Approaches for Modelling of Atmospheric Dispersion with Emphasis on Computational Fluid Dynamics and Application of CFD for Atmospheric Dispersion in a Large Complex Terrain of Kaiga Atomic Power Plant, International Conference on Reliability, Safety & Quality Engineering, ICRSQE 2008, January 5 7,NPCIL, Mumbai. 4. Hanna, S., 1982: Review of Atmospheric Diffusion Models for Regulatory Purposes, World Meteorological Organization Tech. Rept. No. 581 Geneva, Switzerland pp Pasquill, F. & Smith F.B., 1983: Atmospheric Diffusion. 3 rd edition, New York: John Wiley pp Sharma, P. K., Gera, B., Ghosh, A. K. and Kushwaha, H.S., 2004: Modelling of atmospheric dispersion in a flat terrain of kakarapara atomic power plant (KAPP) including effect of building component, 36th National Conference on Fluid Mechanics and Fluid Power, December 17 19, College of Engineering, Pune. 7. Sharma P. K., Markandeya S. G., Ghosh A. K. and H. S. Kushwaha, 2004: Approaches for modelling of dispersion of pollutants in atmosphere with emphasis on computational fluid dynamics, NEHU, Shillong. 8. Sharma P. K., Ghosh B., Ghosh A. K. and H.S. Kushwaha,2007: Analytical and computational fluid dynamics simulation for atmospheric dispersion in a large simple terrain of Narora atomic power plant a predictive calculation, 12th Annual Conference of Gwalior Academy of Mathematical Sciences (GAMS) and All India Symposium on Computational Biology, April 6 8, Maulana Azad National Institute of Technology, Bhopal. 9. Sharma, P. K., Gera, B., Ghosh, A. K. and Kushwaha, H.S., 2010: Inverse problems using ANN in long range atmospheric dispersion with signature analysis picked scattered numerical sensors from CFD, Seventeenth National Symposium on Environment (NSE 17), May 13 15, Indian Institute of Technology, Kanpur. 10. Schlünzen, K.H., 1996: Validierung hochauflösender Regionalmodelle. Ber. aus dem Zentrum f. Meeres und Klimaforschung, Meteorologisches Institut, Universität Hamburg, A23, pp Turner, J.S.,1973: Buoyancy effects in fluids, Cambridge University Press, Cambridge, England. 12. Turner, J.S.,1985: Atmospheric environment, 19, pp

8 13. Turner, D.B., 1994: Workbook of atmospheric dispersion estimates: an introduction to dispersion modeling. 2nd Ed. Boca Raton: Lewis Publishers of CRC Press, pp US Environmental Protection Agency, 1995: User s guide for the industrial source complex (ISC3) dispersion models, Vol. II description and model algorithms, Publication EPA 454/B b. 15. US Environmental Protection Agency, 1995: A user s guide for the CALPUFF dispersion model, Office of air quality planning and standards, Research Triangle Park, North Carolina. Publication EPA 454/B Venkatram, A. and Wyngaard, J.C., 1988: Lectures on air pollution modeling, Boston, American Meteorological Society, pp Wilson, R.B.,1993: Atmospheric environment, 27B, pp Zannetti, P., 1990: Air pollution modeling: theories, computational methods and available Software. New York: Van Nostrand Reinhold, pp