Home Technology Inverse problems using Artificial Neural Networks in long range atmospheric dispersion
Article
Licensed
Unlicensed Requires Authentication

Inverse problems using Artificial Neural Networks in long range atmospheric dispersion

  • P. K. Sharma , B. Gera and A. K. Ghosh
Published/Copyright: April 5, 2013
Become an author with De Gruyter Brill
Submit Manuscript Author Information
Kerntechnik
From the journal Kerntechnik Volume 76 Issue 2

Abstract

Scalar dispersion in the atmosphere is an important area wherein different approaches are followed in development of good analytical models. The analyses 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. Both forward and backward calculation methods are being developed for atmospheric dispersion around NPPs at BARC. Forward modeling methods, which describe the atmospheric transport from sources to receptors, use forward-running transport and dispersion models or computational fluid dynamics models which are run many times, and the resulting dispersion field is compared to observations from multiple sensors. Backward or inverse modeling methods use only one model run in the reverse direction from the receptors to estimate the upwind sources. Inverse modeling methods include adjoint and tangent linear models, Kalman filters, and variational data assimilation, and neural network. The present paper is aimed at developing a new approach where the identified specific signatures at receptor points form the basis for source estimation or inversions. This approach is expected to reduce the large transient data sets to reduced and meaningful data sets. In fact this reduces the inherently transient data set into a time independent mean data set. Forward computations were carried out with CFD code for various cases to generate a large set of data to train the Artificial Neural Network (ANN). Specific signature analysis was carried out to find the parameters of interest for ANN training like peak concentration, time to reach peak concentration and time to fall. The ANN was trained with data and source strength and locations were predicted from ANN. The inverse problem was performed using the ANN approach in long range atmospheric dispersion. An illustration of application of CFD code for atmospheric dispersion studies for a hypothetical case is also included in the paper.

Kurzfassung

Skalare Dispersion in der Atmosphäre ist ein wichtiges Gebiet, in dem verschiedene Ansätze bei der Entwicklung guter analytischer Modelle verfolgt werden. Die Analysen auf der Basis der numerischen Strömungsmechanik führen zur Modellentwicklung auf der Basis grundlegender physikalischer Prinzipien und haben deshalb einen Vorteil gegenüber bereits existierenden Modellen. Verschiedene Methoden für die Berechnung der atmosphärische Dispersion um die Kernkraftwerke bei BARC wurden entwickelt. Vorwärts modellierende Verfahren, die den Transport in der Atmosphäre von der Quelle zum Empfänger beschreiben, verwenden vorwärts laufende Transport- und Dispersionsmodelle oder Modelle auf der Basis numerischer Strömungsmechanik mit vielen Durchläufen und das daraus resultierende Dispersionsfeld wird verglichen mit Beobachtungen multipler Sensoren. Rückwärts oder inverse modellierende Methoden verwenden nur einen Modelldurchlauf in inverser Richtung vom Empfänger um die gegen den Wind gerichteten Quellen zu bestimmen. Inverse Modellierungsverfahren umfassen adjungierte und tangentiale lineare Modelle, Kalman Filter, abweichende Datenassimilation und neuronale Netze. Ziel der vorliegenden Arbeit ist die Entwicklung eines neuen Ansatzes bei dem die identifizierten speziellen Signaturen an den Empfängerpunkten die Basis bildet für die Bestimmung der Quelle oder der Inversionen. Es wird erwartet, dass dieser Ansatz die langen Transienten-Datensätze auf aussagekräftige Datensätze reduziert. Tatsächlich reduzieren sich die transienten Datensätze zu zeitabhängigen mittleren Datensätzen. Vorwärtsrechnungen wurden durchgeführt mit Hilfe eines CFD Rechencodes für verschiedene Fälle, um einen großen Datensatz zum Trainieren der künstlichen neuronalen Netze (KNN) zu erzeugen. Die Analyse spezifischer Signaturen wurden durchgeführt um die für das KNN Training interessanten Parameter zu finden, wie z.B. Spitzenwerte der Konzentration, Zeit zum Erreichen dieser Spitzenwerte und Zeit bis zum Abfall der Konzentration. Das KNN wurde trainiert mit Daten und Quellstärke und Positionen wurden damit bestimmt. Inverse Probleme wurden mit Hilfe der KNN gelöst für weitreichender atmosphärische Dispersion. Diese Arbeit beinhaltet auch eine Erläuterung der Anwendung des CFD Codes für Studien der atmosphärischen Dispersion eines hypothetischen Falls.

* E-mail:

References

1 Edwards, L. L.; Freis, R. P.; Peters, L. G.: The use of nonlinear regression analysis for integrating pollutant concentration measurements with atmospheric dispersion modeling for source term estimation. Nuclear Technology101 (1993) 168–180Search in Google Scholar

2 Raza, S. S.; Avila, R.; Cervantes, J.: A 3-D Lagrangian stochastic model for the meso-scale atmospheric dispersion applications. Nuclear Engineering & Design208 (2001) 15–28Search in Google Scholar

3 Venkatesan, R.; Mathiyarasu, R.; Somayaji, K. M.: A study of atmospheric dispersion of radionuclides at a coastal site using a modified Gaussian model and a mesoscale sea breeze model. Atmospheric Environment36 (2002) 2933–2942Search in Google Scholar

4 JeongHyo-Joon; Eun-HanKim; Kyung-SukSuh; HwangWon-Tae; Moon-HeeHan; Hong-KeunLee: Determination of the Source Rate Released into the Environment from a Nuclear Power Plant. Radiation Protection Dosimetry (2005), 1–6Search in Google Scholar

5 Nazaroff, W. W.; Alvarez-Cohen, L.: Environmental Engineering Science. New York: John Wiley & Sons Inc., 2001Search in Google Scholar

6 Rege, M. A.; Tock, R. W.: Estimation of point source emissions of hydrogen sulfide and ammonia using a modified Pasquill–Gifford approach. Atmospheric Environment30 (1996) 3181–3195Search in Google Scholar

7 Sohier, A.: A European manual for off-site emergency planning and response to nuclear accidents. SCKCEN, Mol, Report R-3594, 2002Search in Google Scholar

8 de Sampaio, P. A. B.; Junior, M. A. G.; Lapa, C. M. F.: A CFD approach to the atmospheric dispersion of radionuclides in the vicinity of NPPs. Nuclear Engineering and Design238 (2008) 250–273Search in Google Scholar

9 Rojas-Palma, C.; et al.: Data assimilation in the decision support system RODOS. Radiation Protection Dosimetry104 (2003) 31–4010.1093/oxfordjournals.rpd.a006160Search in Google Scholar

10 Rege, M. A.; Tock, R. W.: A simple neural network for estimating emission rates of hydrogen sulfide and ammonia from single point sources. J. Air Waste Management46 (1996) 953–962Search in Google Scholar

11 Rao Shankar, K.: Source estimation methods for atmospheric dispersion. Atmospheric Environment41 (2007) 6964–6973Search in Google Scholar

12 Korotky, S.: ABC++ of neural network. http://www.orc.ru/∼stasson/neural.htmlSearch in Google Scholar

13 Lee Wai Ming: A Preliminary Study of Application of Artificial Neural Network in Prediction of Fire Phenomena in Enclosure. Department of Building and Construction, City University of Hong Kong. http://www.glink.net.hk/∼ericlee/Search in Google Scholar

14 Lampinen, J.; Selonen, A.: Using Background Knowledge in Multilayer Perceptron Learning. http://www.lce.hut.fi/∼jlampine/scia97/scia97_web.htmlSearch in Google Scholar

15 Yu-LinHan; XiLiu; Shu-HoDai: Fatigue life calculations of flawed structure-based on artificial neural network with special learning set. International Journal on Pressure Vessels and Piping75 (1998) 263–26910.1016/S0308-0161(98)00040-4Search in Google Scholar

16 McAuley, D.: The Backpropagation Network: Learning by Example. http://www.cs.indiana.edu/∼port/brainwave.doc/BackProp.htmlSearch in Google Scholar

17 Sharma, P. K; Markandeya, S. G.; Ghosh, A. K.; Kushwaha, H. S.; Venkat Raj, V.: Application of Artificial Neural Network for Prediction of Sprinkler Actuation Time in Fire. International Conference on Mechanical Engineering, Dhaka, 2001Search in Google Scholar

18 Verma, V.; Ghosh, A. K.; Sharma, P. K.; Kushwaha, H. S.: Thermal Analysis of a Quarantined Pressure Tube of a PHWR. ISHMT, Calcutta, 2002Search in Google Scholar

19 Sharma, P. K.; Markandeya, S. G.; Ghosh, A. K.; Kushwaha, H. S.: Computational Fluid Dynamics Simulation of Hydrogen Distribution in a Complex Multicompartment Geometry – A Parametric Study. NRT-1, First National Conference on Nuclear Reactor Technology with Focal Theme on Nuclear Safety, Nov. 25th-27th, Bhabha Atomic Research Centre, Mumbai, India, 2002Search in Google Scholar

20 Sharma, P. K.; Singh, R. K; Ghosh, A. K.; Kushwaha, H. S.: Application of Artificial Neural Network and CFD for Passive Dilution of Hydrogen in Vented Geometry. 33rd National – 3rd International Conference on Fluid Mechanics and Fluid Power, December 7–9, 2006, Indian Institute of Technology Bombay, 2006Search in Google Scholar

21 McGrattan, K.; McDermott, R.; Hostikka, S.; Floyd, J.: Fire Dynamics Simulator (Version 5), User's Guide NIST Special Publication 1019-5, In cooperation with: VTT Technical Research Centre of Finland, National Institute of Standards and Technology, Gaithersburg, Maryland, June 23, 2010Search in Google Scholar

22 McGrattan, K.; Hostikka, S.; Floyd, J.; Baum, H.; Rehm, R.: Fire Dynamics Simulator (Version 5) Volume 1 Mathematical Model. Technical Reference Guide NIST Special Publication 1018-5, In cooperation with: VTT Technical Research Centre of Finland, National Institute of Standards and Technology, Gaithersburg, Maryland, October 1, 2007Search in Google Scholar

23 McGrattan, K.; Hostikka, S.; Floyd, J.; Baum, H.; Rehm, R.: Fire Dynamics Simulator (Version 5) Volume 2, Validation. Technical Reference Guide NIST Special Publication 1018-5, In cooperation with: VTT Technical Research Centre of Finland, National Institute of Standards and Technology, Gaithersburg, Maryland, October 1, 2007Search in Google Scholar

24 McGrattan, K.; Hostikka, S.; Floyd, J.; Baum, H.; Rehm, R.: Fire Dynamics Simulator (Version 5) Volume 3, Verification. Technical Reference Guide NIST Special Publication 1018-5, In cooperation with: VTT Technical Research Centre of Finland, National Institute of Standards and Technology, Gaithersburg, Maryland, October 1, 2007Search in Google Scholar

25 Sharma, P. K.; Markandeya, S. G.; Ghosh, A. K.; Kushwaha, H. S.: Approaches for Modelling of Dispersion of Pollutants in Atmosphere with Emphasis on Computational Fluid Dynamics. NEHU, Shillong, 2004Search in Google Scholar

26 Sharma, P. K.; Ghosh, B.; Ghosh, A. K.; Kushwaha, H. S.: 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, Maulana Azad National Institute of Technology, Bhopal, April 6–8, 2007Search in Google Scholar

27 Gera, B.; Sharma, P. K.; Singh, R. K.; Ghosh, A. K; Kushwaha, H. S.: 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, NPCIL, Mumbai, January 5–7, 2008Search in Google Scholar

28 Sharma, P. K.; Gera, B.; Ghosh, A. K.; Kushwaha, H. S.: 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, College of Engineering, Pune, December 17–19, 2009Search in Google Scholar

Received: 2010-8-20
Published Online: 2013-04-05
Published in Print: 2011-05-01

© 2011, Carl Hanser Verlag, München

Articles in the same Issue

  1. Contents/Inhalt
  2. Contents
  3. Summaries/Kurzfassungen
  4. Summaries
  5. Technical Contributions/Fachbeiträge
  6. Comparison between CAREB code calculations and LOCA test results in the FUMEX III project
  7. Calculation of moderator circulation in IPHWR using a porosity approach
  8. Simulation of natural circulation in a rectangular loop using CFD code PHOENICS
  9. CFD analysis of passive autocatalytic recombiner interaction with atmosphere
  10. Review and investigations of oscillatory flow behaviour of a horizontal ceiling opening for nuclear containment and fire safety analysis
  11. CFD simulation of thermal discharge behaviour in the Kadra reservoir at the Kaiga atomic power station
  12. Inverse problems using Artificial Neural Networks in long range atmospheric dispersion
  13. Sipping tests for the irradiated fuel elements of the TR-2 research reactor
  14. Neutron multiplication in source driven subcritical nuclear systems
  15. Cyclotron production of 101Pd/101mRh radionuclide generator for radioimmunotherapy
  16. Investigation of cross sections of reactions used in neutron activation analysis
  17. Modified UN method for the reflected critical slab problem with forward and backward scattering
https://doi.org/10.3139/124.110116

Articles in the same Issue

  1. Contents/Inhalt
  2. Contents
  3. Summaries/Kurzfassungen
  4. Summaries
  5. Technical Contributions/Fachbeiträge
  6. Comparison between CAREB code calculations and LOCA test results in the FUMEX III project
  7. Calculation of moderator circulation in IPHWR using a porosity approach
  8. Simulation of natural circulation in a rectangular loop using CFD code PHOENICS
  9. CFD analysis of passive autocatalytic recombiner interaction with atmosphere
  10. Review and investigations of oscillatory flow behaviour of a horizontal ceiling opening for nuclear containment and fire safety analysis
  11. CFD simulation of thermal discharge behaviour in the Kadra reservoir at the Kaiga atomic power station
  12. Inverse problems using Artificial Neural Networks in long range atmospheric dispersion
  13. Sipping tests for the irradiated fuel elements of the TR-2 research reactor
  14. Neutron multiplication in source driven subcritical nuclear systems
  15. Cyclotron production of 101Pd/101mRh radionuclide generator for radioimmunotherapy
  16. Investigation of cross sections of reactions used in neutron activation analysis
  17. Modified UN method for the reflected critical slab problem with forward and backward scattering
Sign up now to receive a 20% welcome discount
Institutional Access
How does access work?
Have an idea on how to improve our website?
Please write us.
© 2025 De Gruyter Brill
Downloaded on 11.12.2025 from https://www.degruyterbrill.com/document/doi/10.3139/124.110116/pdf
Scroll to top button