17 < > Aerodynamc Desgn Optmzaton of Aeroengne Compressor Rotor, NASA JAXA, Brook Park, Oho, U.S.A., oyama@flab.eng.sas.jaxa.jp Meng-Sng LIOU, NASA Glenn Research Center, Brook Park, Oho, U.S.A. meng-sng.lou-1@nasa.gov,, 98-8577 2-1-1, E-mal: obayash@eee.org Akra OYAMA, NASA Glenn Research Center (currently: JAXA), Brook Park, Oho, U.S.A. Meng-Sng LIOU, NASA Glenn Research Center, Brook Park, Oho, U.S.A. Shgeru OBAYASHI, Insttute for Flud Scences, 2-1-1 Katahra Senda, Myag, Japan A relable and effcent aerodynamc desgn optmzaton tool usng evolutonary algorthm has been developed for transonc compressor blades. The real-coded adaptve-range genetc algorthm s used to mprove effcency and robustness n desgn optmzaton. To represent flow felds accurately and produce relable desgns, three-dmensonal Naver-Stokes computaton s used for aerodynamc analyss. Aerodynamc redesgn of NASA rotor67 s demonstrated to ensure feasblty of the present method. Entropy producton s consdered as the objectve functon to be mnmzed. The computaton s parallelzed on the SGI ORIGIN2 cluster at Insttute of Flud Scence, Tohoku Unversty, by dstrbutng flow analyses of desgn canddates to 64 processng elements. The present method successfully obtaned a desgn that reduced entropy producton by more than 19% compared wth the rotor67 whle satsfyng constrants on the mass flow rate and the pressure rato. 149 53 57 NASA CFD TRAF3D (4),(5) TRAF3D Jameson (6) Baldwn-Lomax CFD (7) (8) 2 TRAF3D TRAF3D (4),(5) CFD CFD (1),(2) (ARGA) [ p,mn, p,max ] p r SGI ORIGIN 2 r = p (1) p, mn r p, max r C r = ( p p, mn )/( p,max p,mn ) (2) 1 (3) r Wake 1 ARGA 1(d) 21 p 1
17 < > pn r (,1) r 3 CFD CFD pn r = N(,1)( z) dz (3) 1% p µ pn = (4) σ µ σ (17) NASA rotor67 197 streamlne-analyss 22 1.56 µ σ 3.11 1.29 1.63 33.25 kg/s 16,43 rpm ARGA 1.38 1.797x1 6 M ARGA ω σ m m / m. (6) σ = σ + ωσ σ σ ) (5) updated present ( sampled present M 4-2,.1-.5 4 ARGA ARGA %( ), 31%, 62%, 1%( ) B 5 14 56 (9) (1) (11) ARGA 1% (12) 67% 92% (13) SUS (14) 78% SGI ORIGIN 2 blended crossover (15) (16) 64 78% CFD 6 19% 1.8% SGI ORIGIN 2 64 3% 9% 2 ( ) 5 desgn rotor 67 ) rotor 67 ( P ) P. 1 P (7) rato, desgn rato, rotor67 / rato, rotor67
17 < > 8 33% Grds, Journal of Propulson and Power, Vol. 8, No. 2, 1992, pp.41-417. 9 Arnone, A., Lou, M.-S., and Povnell, L. A., Multgrd Calculaton of Three-Dmensonal Vscous Cascade Flows, 3% NASA-TM-15257, 1991. Arnone, A, Vscous Analyss of Three-Dmensonal Rotor Flow 1 9% Usng a Multgrd Method, ASME Journal of Turbomachnery, Vol. 116, No. 3, 1994, pp. 435-445. Jameson, A., Schmdt, W., and Turkel, E., Numercal Solutons of 11 the Euler Equatons by Fnte Volume Methods Usng Runge-Kutta Tme-Steppng Schemes, AIAA Paper 1981-1259, 1981. Jameson, A., The Evoluton of Computatonal Methods n Aerodynamcs, Journal of Appled Mechancs, Vol. 5, No. 4b, 1983, pp.152-169. 12 Jameson, A., Transonc Flow Calculatons, MAE Report 1651, MAE Department, Prnceton Unversty, Prnceton, NJ, 1983. 2% Oyama, A., Obayash, S., and Nakahash, K, Real-Coded Adaptve 6% 9% Range Genetc Algorthm and Its Applcaton to Aerodynamc Desgn, JSME Internatonal Journal Seres A, Vol. 43, No. 2, 2, pp.124-129. 13 Oyama, A., Obayash, S., and Nakamura, T., Real-Coded Adaptve Range Genetc Algorthm Appled to Transonc Wng Optmzaton, Appled Soft Computng, Vol. 1, No. 3, 21, pp.179-187, http://www.elsever.com/locate/asoc. Sasak, D, and Obayash, S., Low-Boom Desgn Optmzaton for SST Canard-Wng-Fuselage Confguraton, AIAA Paper 223-3432. CFD Mchalewcz, Z, Genetc Algorthms + Data Structures = Evoluton Programs, thrd revsed edton, Sprnger-Verlag, Berln, Germany, 1996. Deb, K., Pratap, A. and Motra, S., Mechancal Component Desgn for Multple Objectves Usng Eltst Non-Domnated Sortng GA, SGI ORIGIN Lecture Notes n Computer Scence 1917 Parallel Problem Solvng 2 from Nature PPSN VI, Sprnger, Berln, Germany, 2, pp.859-868. 1.8% Baker, J. E., Reducng Bas and Ineffcency n the Selecton Algorthm, Proceedngs of the Second Internatonal Conference on Genetc Algorthms, Morgan Kaufmann Publshers, Inc., San Mateo, Calforna, 1987, pp 14-21. CFD Eshelman, L. J. and Schaffer, J. D., Real-Coded Genetc Algorthms and Interval Schemata, Foundatons of Genetc Algorthms.2, Morgan Kaufmann Publshers, Inc., San Mateo, Calforna, 1993, pp 187-22. De Jong, K. A., An Analyss of the Behavor of a Class of Genetc Adaptve Systems, Ph.D. thess, Unversty of Mchgan, Ann Arbor, Mchgan, 1975. Perzga, J. J., and Wood, J. R., Investgaton of the Mettnen, K., Makela, M. M., Nettaanmak, P. and Peraux, J. eds., Evolutonary Algorthms n Engneerng and Computer Scence, John Wlley & Sons Ltd, Chchester, U.K., 1999, Chaps.17-24. Dasgupta, D. and Mchalewcz, Z., eds., Evolutonary Algorthms n Engneerng Applcatons, Sprnger-Verlag, Berln, 1997, chapters 2 and 3. Arnone, A., Lou, M.-S., and Povnell, L. A., Naver-Stokes Soluton of Transonc Cascade Flow Usng Non-Perodc C-Type 3 Three-Dmensonal Flow Feld Wthn a Transonc Fan Rotor: Experment and Analyss, Journal of Engneerng for Gas Turbnes and Power, Vol.17, No.2, Aprl 1985, pp.436-449.
17 < > adabatc effcency.94.92.9.88 near stall.86 experments.84 computatons.82.9.92.94.96.98 1 mass flow rate/mass flow rate at choke Fg. 2 Comparson of computed and measured adabatc effcency. r Fg.3 pn Schematc vew of decodng of the real-coded ARGA. Evaluaton Intal populaton Optmum Every M generatons Selecton Samplng for range adaptaton Range adaptaton Reproducton by crossover+mutaton Reproducton by random dstrbuton Fg.1 Computatonal grd over the. Every other lne sshown. Fg.4 Flowchart of the real-coded ARGA. 4
z/c Fg.5 entropy producton Fg.6.2.1 Table 1. control ponts B-Splne curves t/c.6.4.2 control ponts B-Splne curves Control ponts and B-Splne curves for a mean camber lne and a thckness dstrbuton..1.9.8 optmzed desgns.7 2 4 6 8 1 generaton Optmzaton hstory n terms of entropy producton. Computed performance of rotor67 and the optmum desgn. Fg.8 y/c statc pressure 17 < >.12.1.8.6.4.2 1.2 1.8.6.4 optmzed desgn Comparson between the optmzed desgn and rotor 67 at 33% span. mass flow sentropc pressure entropy [kg/sec] effcency rato producton Rotor67 33.774.9189 1.6758.9467 Optmum 33.929.93528 1.6859.73263 entropy producton.1.8.6.4.2 optmzed desgn -.2 span Fg.7 Comparson of spanwse entropy producton dstrbutons. 5 Fg.9 Relatve Mach number contours of the optmzed desgn and rotor 67 at 33% span. Maxmum relatve Mach numbers are 1.39 for the and 1.19 for the optmzed desgn respectvely.
Fg.1 y/c statc pressure.4.3.2.1 -.1 1.8 1.6 1.4 1.2.8 optmzed desgn 1.6 Comparson between the optmzed desgn and rotor 67 at 9% span. 17 < > Fg.12 Ol flow patterns and statc pressure contours on sucton surfaces. adabatc effcency.94.92.9.88.86.84 optmzed desgn.82 31 31.5 32 32.5 33 33.5 34 34.5 mass flow rate/mass flow rate at choke Fg.13 Performance map comparson between the optmzed desgn and rotor 67. Fg.11 Relatve Mach number contours of the optmzed desgn and rotor 67 at 9% span. Maxmum relatve Mach numbers are 1.47 for the and 1.42 for the optmzed desgn respectvely. 6