1. http://kucg.korea.ac.kr R EAL TIME SIMULATION OF A TORNADO Shiguang Liu, Zhangye Wang, Zheng Gong, Lei Huang, and Qunsheng Peng

2. http://kucg.korea.ac.kr A BSTRACT simulating a tornado scene Based on Reynold-average Navier-Stokes equations. The dust particle flow is modeled by non-viscosity Navier–Stokes equations Multi-Fluid Solver is designed and implemented on GPU. Efficient method is proposed to simulate the tornado’s interaction with surrounding large objects.

3. http://kucg.korea.ac.kr I NTRODUCTION RATFM is proposed to simulate the chaos appearance of tornados more realistically than previous methods A novel two-fluid system solver is designed to achieve real time simulation To our knowledge, it is the first attempt to simulate damage from a tornado on surrounding objects Our system is easy to implement. By inputting different initial parameters, different tornado scenes can be produced automatically

4. http://kucg.korea.ac.kr R ELATED W ORK simulating natural phenomena smoke,water & fire semi-Lagrangian method Mizuno volcanic clouds consist of two fluids Müller et al. smoothed particle hydrodynamics Losasso et al. particle level set method simulate the interactions among multiple liquids Zhu et al. two-fluid lattice Boltzmann model

5. http://kucg.korea.ac.kr R ELATED W ORK Modeling the motion of dust particles contact force, normal force, and shear force Ding et al. propose an approach for tornado simulation To use many particles (not real time)  Our  TFM method  Real-time  interaction with large objects

6. http://kucg.korea.ac.kr R EYNOLD - AVERAGE N AVIER –S TOKES EQUATIONS

7. http://kucg.korea.ac.kr Interaction force Between air flow and dust particle flow plays an important part in modeling a tornado

8. http://kucg.korea.ac.kr Reynolds shear stress

9. http://kucg.korea.ac.kr R EYNOLD - AVERAGE TWO - FLUID MODEL dust particle flow model non-viscosity, incompressible fluid

10. http://kucg.korea.ac.kr B OUNDARY CONDITIONS

11. http://kucg.korea.ac.kr TORNADO ’ S CONDITIONS horizontal velocity field : rotating vertical velocity field : uplifting

12. http://kucg.korea.ac.kr M ODELING CONTRAST BETWEEN RATFM AND TFM ( RESULTS ) RATFM TFM

13. http://kucg.korea.ac.kr T HE TORNADO ’ S INTERACTION WITH LARGE OBJECTS Tonado 에 의해 부서지는 object 를 시뮬레이션 Object 는 voxel 에 연결 되어 있음 Voxcel 이 큰 압력을 받으면 연결된 object 부분을 부 숨

14. http://kucg.korea.ac.kr T HE TORNADO ’ S INTERACTION WITH LARGE OBJECTS Object 가 받는 힘 Torque

15. http://kucg.korea.ac.kr T HE TORNADO ’ S INTERACTION WITH LARGE OBJECTS

16. http://kucg.korea.ac.kr T HE TORNADO ’ S INTERACTION WITH LARGE OBJECTS ( RESULTS )

17. http://kucg.korea.ac.kr T HE TORNADO ’ S INTERACTION WITH LARGE OBJECTS ( RESULTS )

18. http://kucg.korea.ac.kr T HE TORNADO ’ S INTERACTION WITH LARGE OBJECTS ( RESULTS )

19. http://kucg.korea.ac.kr M ULTI -F LUID S OLVER ON GPU Our model describes a multiple fluid system Air flow particle flows. We solve the multiple Navier-Stokes equations in parallel in one rendering pass by combining multiple field data texture into one texture. It reduces the calculating time Flat 3D texture technique It’s easy to read and store velocity data

20. http://kucg.korea.ac.kr M ULTI -F LUID S OLVER ON GPU Flow chart of Multi-Fluid Solver With this, we can solve multiple NS in parallel in one rendering pass.

21. http://kucg.korea.ac.kr M ULTI -F LUID S OLVER ON GPU ( RESULTS )

22. http://kucg.korea.ac.kr R ESULTS AND D ISCUSSION Successfully generated dynamic tornado scenes Calculating the Poisson equation use the Jacobi iterative method 25 frames per second More iterations, lower frame rate

23. http://kucg.korea.ac.kr C ONCLUSION AND F UTURE W ORKS Simulating realistic tornado scenes To use RATFM The tornado’s interaction with surrounding large objects was simulated Future Works mixtures with three or more fluid components Water & Oil Other phenomena