Transcript Slide 1
SNAME H-8 Panel Meeting No. 124 Oct. 18, 2004 NSWC-CD Research Update from UT Austin Ocean Engineering Group Department of Civil Engineering The University of Texas at Austin Prof. Spyros A. Kinnas Dr. Hanseong Lee, Research Associate Mr. Hua Gu, Doctoral Graduate Student Ms. Hong Sun, Doctoral Graduate Student Mr. Yumin Deng, Graduate student Topics ►MPUF/HULLFPP .vs. PROPCAV/HULLFPP ►Effective wake evaluation at blade control points ►Modeling of cavitating ducted propeller ►Blade design using optimization method 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 2 MPUF3A- and PROPCAV- /HULLFPP (Steady wetted case: H03861 Propeller) ►Propeller and hull geometries * 4 blades * User input thickness * User input camber 10/18/2004 * uniform wake * Froude number Fr=9999.0 * Advance Ratio Js =0.976 * IHUB = OFF SNAME panel H-8 mtg. No. 124, NSWC-CD 3 MPUF3A- and PROPCAV- /HULLFPP (Pressure distribution on the hull) ► From MPUF3A/HULLFPP 10/18/2004 ► From PROPCAV/HULLFPP SNAME panel H-8 mtg. No. 124, NSWC-CD 4 MPUF3A- and PROPCAV- /HULLFPP ►Circulations from MPUF3A and PROPCAV * Not considering induced velocity effect * Match the transition wake geometry from PROPCAV with that from MPUF3A 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 5 MPUF3A- and PROPCAV- /HULLFPP ►Field Point Potential from MPUF3A and PROPCAV 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 6 MPUF3A/HULLFPP (Effects of the ultimate wake singularities) ►Previously, it was assumed that only the steady part of the circulation at the blade TE shed into the ultimate wake, and a decay function was applied to the transition wake ►In the improved approximation the unsteady vorticity is shed into the ultimate wake ►This improvement was verified by several cases using uniform inflow 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 7 MPUF-3A/HULLFPP ► General wake geometry 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 8 MPUF-3A/HULLFPP (Steady cavitating case) ►Hull geometry and run conditions * Uniform wake * IHUB =ON * TLC = ON 10/18/2004 * Cavitation number n 1.731 * Froude number Fr = 3.0789 * Advance Ratio Js =1.177 SNAME panel H-8 mtg. No. 124, NSWC-CD 9 MPUF-3A/HULLFPP (Pressure distribution on the hull) ► Improved Approximation 10/18/2004 ►Using decay function SNAME panel H-8 mtg. No. 124, NSWC-CD 10 MPUF-3A/HULLFPP (Unsteady cavitating case) ►Cavitating run conditions ►Cavity patterns 20x18 * Effective wake * Cavitation number n * Froude number Fr=4.0 * Advance Ratio Js =1.0 * IHUB = OFF * TLC = ON 10/18/2004 2 .7 SNAME panel H-8 mtg. No. 124, NSWC-CD 11 MPUF-3A/HULLFPP (Pressure distribution on the hull) ► Improved approximation 10/18/2004 ►Using decay function SNAME panel H-8 mtg. No. 124, NSWC-CD 12 NEW EFFECTIVE WAKE CALCULATION 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 13 Effective wake evaluation at blade control points Previous method: evaluates effective wake at a plane ahead (by one cell) of the blade. New method: Evaluates the effective wake at the blade control points. Effective wake evaluation at blade control points Interpolation of total axial velocity on control points Interpolation of total tangential velocity on control points Effective wake evaluation at blade control points At the MPUF-3A control points, the induced velocity may be in error due to the local effect of blade singularities. The bad points need to be removed before the induced velocity is (time) averaged. The figure shows the induced velocity at a control point at chord index 9 and span index 8. Effective wake evaluation at blade control points At each control point, Ue = Ua -Uin is applied, the expected effective wake should be 1.00 at all points, there is still a maximum of 4% error in this case. Effective wake evaluation at blade control points The error brings lower circulation for this case, which still needs improvement. CAVITATING DUCTED PROPELLER 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 19 Modeling of cavitating ducted propeller (duct: panel method, propeller: PROPCAV) ►NACA0015 Duct Straight Panel Paneled with pitch angle (45o) Modeling of ducted propeller ►NACA0015 Duct Modeling of ducted propeller ►NACA0010 Duct Straight Panel Paneled with pitch angle (45o) Modeling of ducted propeller ►NACA0010 Duct Modeling of ducted propeller ►NACA0015 Duct + N3745 Propeller * Uniform wake * Advance ratio Js =0.6 Circulation Distribution BLADE DESIGN VIA OPTIMIZATION 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 25 CAVOPT-3D (CAVitating Propeller Blade OPTimization method) Mishima (PhD, MIT, ’96), Mishima & Kinnas (JSR ’97), Griffin & Kinnas (JFE’98) CAVOPT-3D ►Allows for design of propeller in non-axisymmetric inflow and includes the effects of sheet cavitation DURING the design process ►MPUF-3A is running inside the optimization scheme until all requirements and constraints are satisfied ►Takes about 600-1000 MPUF-3A runs to produce the final design (3-6 hrs) ►New versions of MPUF-3A (that include duct, pod, etc) can be incorporated ►Not practical as a web based instructional tool 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 27 New Optimization Method ►Start with a base propeller geometry. ►Given conditions are: Js, inflow (can be nonaxisymmetric), cavitation number, Froude number, and thrust coefficient. ►The optimum design is searched for within a family of propeller geometries such that: P / D X 1 * ( P / D)base c / D X 2 * (c / D)base f / c X 3 * ( f / c)base X1, X2, X3 are factors (constant initially, to be varied later) 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 28 ► Hydrodynamic coefficients and cavity planform area are expressed in terms of polynomial functions of X1, X2 and X3. KT KQ CA T n D 2 4 Q n D 2 5 f1 ( X 1 , X 2 , X 3 ) f2 ( X1, X 2 , X 3 ) cavity area blade area f3 ( X 1 , X 2 , X 3 ) f i ai ,1 X 1 ai ,2 X 2 ai ,3 X 3 ai ,4 X 1 ai ,5 X 2 ai ,6 X 3 2 2 2 a i ,7 X 1 X 2 ai ,8 X 1 X 3 ai ,9 X 2 X 3 ai ,10 While: min X1 X1 min X2 X2 min X3 X3 X1 X2 X3 max max max The function coefficients are determined by Least Square Method (LSM), using the predictions of a large array (e.g. 10x10x10) of MPUF-3A runs 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 29 The Optimization Scheme (based on CAVOPT-2D, optimization method for cavitating 2-D hydrofoils) ► The optimization problem of the propeller design is : Minimize : f ( x) Subject to : gi ( x) 0 hi ( x) 0 Where f ( x) is the objective function to be minimized. x is the solution vector of n components. gi ( x) 0 ( i=1…m ) are inequality constrains and hi ( x) 0 ( i=1…l ) are equality constrains. The constrained optimization problem is changed to an unconstrained optimization problem by using Lagrange multipliers and penalty functions. For more information, please refer to the JSR paper by Mishima & Kinnas, 1996. 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 30 In current case, the problem reduces to: Augmented Lagrange function: ~ ~ 2 ~ F ( x, 1 , 1 , c1 , c1 ) K Q ( x) 1h1 ( x) 1[ g1 ( x) s1 ] 2 2 ~ c h ( x) c1[ g1 x s1 ] 2 1 1 With: h1 ( x) KT ( x) KT 0 0 g1 ( x) CA( x) CAMAX 0 KT 0 and CAMAX are user defined. The function to be minimized is KQ ( x) , while x is the vector ~ (X1, X2, X3), 1 and 1 are Lagrange multipliers, c1 and c1 are penalty function coefficients. 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 31 Optimization Samples: Sample 1: Fully wetted run based on N4148 propeller (with prescribed skew distribution) -- Design conditions: • K T 0 0.15 , K Q to be minimized • J 1.0 , n 999 , Fn 999 S • uniform inflow • 20x9 grid size -- Range of variables: 0 .8 x1 2 .2 0 .8 x 2 2 .0 0 .8 x 3 2 .0 -- Database and Interpolation : 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 33 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 34 How good is the interpolation method? 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 35 -- Optimum solution and comparisons with CAVOPT-3D 3rd order functions are used to approximate both KT and KQ code KT 10KQ Efficiency OPT 0.1490 0.2994 0.7921 CAVOPT-3D 0.1504 0.2996 0.7991 The solution of OPT are : X1 = 1.28865 X2 = 0.80000 X3 = 2.00000 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 36 Propeller geometry comparison: OPT vs. CAVOPT-3D 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 37 Circulation comparison: OPT vs. CAVOPT-3D 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 38 Pressure distribution comparison: OPT vs. CAVOPT-3D 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 39 Blade geometry comparison: OPT vs. CAVOPT-3D 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 40 Sample 2: Cavitating run based on N4148 propeller and presribed skew distribution -- Design conditions: • K T 0 0.25 , K Q to be minimized • J 1.2 , n 2.5 , Fn 5.0 S • C AM AX 40% • effective wake file • 10x9 and 20x9 grid size -- Range of variables: 1 .0 x1 2 .0 0 .8 x 2 2 .0 0 .8 x 3 2 .0 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 41 -- Wake file used: 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 42 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 43 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 44 -- Database and Interpolation : 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 45 -- Optimization solution from OPT (MPUF-3A: 10X9) and comparisons with CAVOPT-3D (MPUF-3A: 10X9) 4th order functions are used for KT, KQ and CAMAX Initial guess: ( 0.8, 1.0, 1.0 ) code KT 10KQ CA Efficiency OPT (10x9) 0.2505778 0.5528295 18.51% 86.6% CAVOPT-3D 0.2267294 0.5193282 18.97% 83.4% The solution of OPT are : X1 =1.43586 X2 =2.00000 X3 =1.89869 Several initial guesses were tested, they led to the almost same optimization results. 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 46 Propeller geometry comparison: OPT (10x9) vs. CAVOPT-3D (10x9) 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 47 Circulation comparison: OPT (10x9) vs. CAVOPT-3D (10x9) 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 48 Blade geometry comparison: OPT (10x9) vs. CAVOPT-3D (10x9) 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 49 Cavitations comparison: OPT (10x9) vs. CAVOPT-3D (10x9) 18.51 % 10/18/2004 18.97 % SNAME panel H-8 mtg. No. 124, NSWC-CD 50 -- Optimization solution of OPT (20x9) 4th order functions are used for KT, KQ and CAMAX Initial guess: ( 0.8, 1.0, 1.0 ) code Grid KT 10KQ CA Efficiency OPT 20x9 0.2497656 0.5558742 20.85% 85.8% CAVOPT-3D 10x9 0.2267294 0.5193282 18.97% 83.4% The solution of OPT are : X1 =1.44072 X2 =1.94383 X3 =2.00000 Several initial guesses were tested, they led to almost the same optimization results. 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 51 Propeller geometry of OPT (20x9): 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 52 Circulation of OPT 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 53 Blade geometry and cavity of OPT (20x9) 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 54 Conclusions and Future work (on optimization) ►The interpolation scheme can approximate the database very well using higher order functions. ►The optimization scheme works well for the fully wetted run. For cavitating runs, both CAVOPT-3D and OPT should be improved. ►Include more parameters in current optimization scheme. ►Improve the approximation of cavity area. 10/18/2004 SNAME panel H-8 mtg. No. 124, NSWC-CD 55