COMPARATIVE ANALYSIS OF THE PARALLEL COMPUTING EFFICIENCY FOR EXPLICIT AND IMPLICIT DIFFERENCE SCHEMES FOR COMPUTATIONAL AERODYNAMICS PROBLEMS
Abstract
One of the main areas of parallel computing application is solving of computational aerodynamics
problems. The paper considers parallel modeling of gas dynamics with quasi-onedimensional
system of equations. The system describes the gas flow through a channel with variable
cross section using implicit and explicit difference schemes. The purpose of this work is to
study the methods efficiency for organizing parallel computations with implicit and explicit difference
schemes for solving internal aerodynamics problems. The article presents a comparative
analysis of the proposed parallel models for quasi-one-dimensional equations system of gas dynamics.
The system describes flows in a channel of variable cross section. Various parallel algorithms
are used to solve systems of such type numerically. The method of splitting by physical processes
is used for an implicit difference scheme. To suppress the solution oscillations a smoothing
operator at the correction stage is introduced in calculations according to the scheme of the predictor-
corrector type. In scheme fractional steps the parallel scalar sweep algorithm is used to
solve tridiagonal systems with the parametric unknown choice. Besides to compare scheme mentioned
above a parallel algorithm is constructed for McCormack's explicit scheme. Such algorithm
is widely used in computational aerodynamics. Parallel computations are held by computing
structures with distributed memory and by another one – with linear switching dependence between
computing devices of the working field. The paper presents time estimations for each computing
stage (both by implicit and explicit difference schemes). It helped to calculate the developed
parallel algorithms efficiency. It is concluded that the acceleration factor in explicit scheme depends
linearly on the computing devices number.
References
aerogidrodinamiki [Splitting algorithms for solving multidimensional problems of
aerohydrodynamics]. Novosibirsk: Izd-vo SO RAN. 2014, 280 p.
2. Kovenya V.M. Splitting method in the problems of CFD, Computational Fluid Dynamics
Journal, 1999, Vol. 8, No. 2, pp.186-194.
3. Voevodin V.V. Mathematical foundations of parallel computing, Series in computer science.
World Scientific Publishing Co, 1992, Vol. 33, 343 p.
4. Hoffman J.D. Numerical methods for engineers and scientists. 2nd ed. revised and expanded.
New York: Marcel Dekker, Inc. 2001, pp. 651-701.
5. Borisov A.V., KovenyaV.M., Yanenko N.N. Ob odnoy konservativnoy raznostnoy skheme dlya
resheniya statsionarnykh uravneniy gazovoy dinamiki [On one conservative difference scheme
for solving stationary equations of gas dynamics], Matematicheskiye modeli techeniy
zhidkosti: Tr.6-go Vsesoyuznogo seminara po chislennym metodam mekhaniki vyazkoy
zhidkosti [Mathematical Models of Fluid Flows: Proceedings of the 6th All-Union Seminar on
Numerical Methods of Viscous Fluid Mechanics, Kungura autumn 1976]. Novosibirsk:
ITiPM, 1978, Vol. 10, No 6, pp. 15-22.
6. Gamolina I.E., Semenistyy V.V. Parallel'naya organizatsiya vychisleniy pri raschete zadach
aerogidrodinamiki pryamymi metodami. Mezhdunarodnoe nauchnoe sotrudnichestvo,
obrazovanie i kul'tura [Parallel organization of calculations in the calculation of problems of
aerohydrodynamics by direct methods. International scientific cooperation, education and culture].
Rostov-on-Don: Summa-Rerum. 2014, No. 3 (4), pp. 23-28.
7. Nour-Omid B., Raefsky A., Lyzenga G. Solving finite element equations on concurrent computers.
A.K. Noor, ed.. American Soc. Mech. 1986, pp. 291-307.
8. Yanenko N.N., Konovalov A.N., Bugrov A.N., Shustov G.V. Ob organizatsii parallel'nykh
vychisleniy i rasparallelivanie progonki [On the organization of parallel calculations and parallelization
of the run], Chislennye metody mekhaniki sploshnykh sred [Numerical methods of
continuum mechanics], 1978, No. 7, pp. 136-139.
9. Semenistyy V.V., Gamolina I.E., Duryagina V.V. Konstruirovanie effektivnykh parallel'nykh
algoritmov dlya resheniya polnoy dvumernoy sistemy uravneniy Nav'e-Stoksa po yavnoy
skheme Mak-Kormaka [Designing effective parallel algorithms for solving a complete
twodimensional system of Navier-Stokes equations according to the explicit McCormack
scheme], Sb. materialov II mezhdunarodnoy nauchno-prakticheskoy konferentsii
«Issledovaniya i razrabotki v perspektivnykh nauchnykh oblastyakh» [Collection of materials
of the II International Scientific and Practical Conference "Research and Development in
promising scientific fields"]. Novosibirsk. 2017, pp. 88-95.
10. Semenistyy I.I., Gamolina I.E., Duryagina V.V. Modelirovaniye sverkhzvokovogo techeniya
po kanalu peremennogo secheniya s ispolzovaniyem sredy ANSYS. [Supersonic flow simulation
through a channel with variable cross section using software ansys "KomTekh"]. Taganrog.
2019, pp. 138-145.
11. Semenistyy V.V., Gamolina I.E. Issledovaniye sposobov organizatsii parallelnogo resheniya
vneshnikh zadach aerodinamiki na osnove skhem rasshchepleniya [Study of parallel solution
organization for external aerodynamics problems based on splitting schemes], Izvestiya YuFU.
Tekhnicheskiye nauki. [Izvestiya SFedU. Engineering Sciences], 2020, No. 5, pp. 60-67.
12. Semenistyy V.V., Gamolina I.E. Organizatsiya struktur dannykh pri parallelnykh
vychisleniyakh zadach aerodinamiki [Data structure organization using parallel calculations
for problems of aerohydrodynamics], Mater. 4-toy Mezhdunarodnoy nauchnoy konferentsii
[Proceedings of the 4th International Scientific Conference, Maykop], 2021, pp. 262-265.
13. MacCormack R.W. The effect of viscosity in hypervelocity impact cratering, J. Spacecraft and
Rockets, 2003, Vol. 40, No. 5, pp. 757-763.
14. Terenkov V.I., Arsenii V.F., Evseev E.G., Lutskiy Ya.A., Semenistyy V.V. O korrektnosti i
ustoychivosti algoritma rasparallelivaniya progonki [On the correctness and stability of the
parallelization algorithm of the run], Tr. int-ta prikl. matemat. im. I.N. Vekua Tbilis. un-ta
[Proceedings of the I.N. Vekua Institute of Applied Mathematics]. Tbilisi, 1985, pp. 298-307.
15. Caughey Hafez. Contributions of Robert Mac-Cormack to Computational Fluid Dynamics,
Frontiers of Computational Fluid Dynamics, 2002, pp. 1-26.
16. Bram V. Upwind and High-Resolution Methods for Compressible Flow: From Donor Cell to
Residual-Distribution Schemes, Communications in computational physics, 2006, Vol. 1,
No. 2, pp. 192-206.
17. Barkalov K.A. Metody parallelnykh vychisleniy. [Parallel Computing Methods]. N. Novgorod:
Izd-voNizhegorodskogo gosuniversiteta im. N.I. Lobachevskogo, 2011, 124 p.
18. Kovenya V.M., Slyunyayev.A.Yu. Modifikatsiya algoritmov rasshchepleniya dlya zadach
gazovoy dinamiki i Navye-Stoksa [Modification of splitting algorithms for gas dynamics and
Navier-Stokes problems], Vychislitelnyye tekhnologii [Computational technologies], 2007,
Vol. 12, No. 3, pp. 71-86.
19. Kovenya, V.M., Eremin A.A. Predictor-Corrector Difference Scheme for Numerical Solution of
the Euler and Navier–Stokes Equations, Journal of Mathematical Sciences, 2016, Vol. 215,
No. 4, pp. 484-498.
20. Hoffman J.D. Numerical methods for engineers and scientists. Second edition revised and
expanded. – New York: Marcel Dekker, Inc.. 2001, pp. 651-701.
