Aeroelastic Flutter

Embed Size (px)

DESCRIPTION

Coding to solve flutter

Citation preview

EAS 3406AEROELASTICBYDR. MOHAMMAD YAZDI HARMIN

ASSIGNMENT 2:FLUTTER

NAME: ONG THIAM CHUNNO. MATRIC: 158347

DISCUSSIONa) This aeroelastic calculation assumes that the unsteady aerodynamic damping is constant. In order to take account of structural damping in flutter analysis, one needs to obtain Rayleigh coefficients. Due to complexity in finding Rayleigh coefficient, the structural damping is also neglected. Based on Introduction to Aircraft Aeroelasticity and Loads by Jan R. Wright, the inclusion of structural damping actually has small effect on delaying flutter speed.

Figure 1: Vf and Vg plots against Air Speed (m/s)Critical flutter speed takes place where one of the modes has negative damping ratio. As the airspeed increases, the frequencies begin to converge. One of the modes has increasing damping ratio, as a contrast to this phenomena, another mode shows decreasing damping ratio which eventually become negative at around 111.6m/s. The frequency does not coalesce when flutter happens. This damping ratio decrement happens at a shallow gradient, so we called the flutter as a soft flutter.

b)Figure 2: Effect of changes in flexural axis on flutter speedThe mass axis is fixed while flexural axis is varying. Notice that even the flexural axis and mass axis are coincide at mid-chord, flutter is still taking place. When flexural axis is located at mid-chord, there is no inertial coupling. However, the two mode shapes are coupled when flexural axis is no longer located at mid-chord. From Figure 2, the flutter speed decreases as the flexural axis is moving further away from leading edge. An increment of flexural axis also indicates the increases in eccentricity between flexural axis and aerodynamic centre. This phenomenon can be explained based on moment of the wing. As the eccentricity increases, lift forces at aerodynamic centre will create a greater moment effect as the moment arm (eccentricity) of lift force to flexural axis is longer. This moment may push the leading edge upwards, lead to higher effective angle of attack, and this is where the degradation on flutter speeds performance started with the increasing distance of flexural axis and leading charge.

c)

Figure 3: Effect of flap stiffness on flutter speed

Figure 4: Effect of pitch stiffness on flutter speed

Generally, Figure 3 shows that flutter speed decreases along with the increases in flap stiffness and have a sudden increase again at 4.6 x 107 Nm/rad of flap stiffness. On the other hand, the increment in pitch stiffness improves its flutter speed except in the lower pitch stiffness (rapid increase with a sharp drop), which is opposite to that in increment of flap stiffness. The result may indicate that the changes in flap and pitch stiffness does not necessary improve dynamic aeroelasticity performance in the same trend (example: increases in pitch stiffness or decreases in flap stiffness improve dynamic aeroelasticity performance), there is uncertainty shown in Figure 3 and Figure 4. One may conclude that this theoretical analysis serves as reference for suitable flap and pitch stiffness to improve dynamic aeroelasticity performance in this particular wing configuration and altitude. The limits of increment in pitch stiffness and decrement in flap stiffness should be noticed to avoid degradation in performance. More studies should be done to investigate the factors for such inconsistency in the result shown in Figure 3 and Figure 4.

d)Figure 5: Vf plotFigure 6: Vg plot

Figure 7: Output of the calculationIn frequency matching k-method, the stable region is where the damping coefficient, g are negative and the system becomes unstable when g become positive. With the same configuration in a), the k-method analysis yields a slightly higher flutter speed than that in a). This is due to the inclusion of unsteady reduced frequency effects. The unsteady aerodynamic damping term, is assumed as constant in the analysis in part a) at different air speed. However, the aerodynamic stiffness and damping matrices are reduced frequency-dependent. The frequency dependency of is given by the equation below:

Frequency Matching k-method seems to show a more realistic result than the method using in part a) because it includes the unsteady aerodynamic condition and reduced frequency. If an aircraft is analysed based on steady condition or neglecting some unsteady terms, the aircraft may perform well below the actual capable performance. Steady state analysis is much simpler and quicker within safety margin of aircraft performance but come with a cost of unrealistic and degrading the actual performance of aircraft. K-method or p-k method is the better analysis method for such dynamic aeroelastic problem which in real life is unsteady and frequency dependent. REFERENCEBabister, A. W. (June 1950). Flutter and Divergence of Sweptback and Sweptforward Wings. Retrieved from https://dspace.lib.cranfield.ac.uk/bitstream/1826/7209/3/COA_Report_No_39_JUN_1950.pdf on 21 November 2013.Wright, J. R. (2007). Introduction to Aircraft Aeroelasticity and Loads. Dynamic Aeroelasticity Flutter (pp167-pp199). West Sussex, England: John Wiley & Sons, Ltd.Mohammad Sadraey. Wing Design. School of Engineering and Computer Sciences Daniel Webster College, 2013. Mohammad Yazdi Harmin. (2013). Lecture on Dynamic Aeroelasticity Flutter. Personal Collection of Mohammad Yazdi Harmin, University Putra Malaysia, Selangor.

APPENDIXAeroelastic Equation

where=

Frequency Matching k method

MATLAB codingclear all; ; clc%Parametersm = 100; % unit mass / area of wingc = 1.7; % chord in ms = 7; % semi span in mxcm = 0.5*c; % position of centre of mass from nosexf = 0.5*c; % position of flexural axis from nosee = xf/c-0.25;% eccentricity between flexural axis and aero centre (1/4chord)K_k = 1.2277e7; % Flap Stiffness in Nm/radK_t = 7.2758e5; % Pitch Stiffness in Nm/radMthetadot = -1.2; % unsteady aero damping termroll = 1.225; % air density in kg/m^3aw = 2*pi; % 2D lift curve slopeM = (m*c^2/(2*xcm))-m*c; % leading edge mass termb=c/2; %Inertial, Damping and Stiffness Matrixa11=(m*s^3*c)/3 + M*s^3/3; % I kappa based on pg 179a12 = ((m*s^2)/2)*(c^2/2 - c*xf) - M*s^2*xf/2; %I kappa thetaa21 = a12;a22= m*s*(c^3/3 - c^2*xf + xf^2*c) + M*(s*xf^2); % I thetaA=[a11,a12;a21,a22]; b11=c*s^3*aw/6;b12=0;b21=-e*c^2*s^2*aw/4;b22=-c^3*s/8*Mthetadot;B=[b11 b12; b21 b22] c11=0;c12=c*s^2*aw/4;c21=c11;c22=-e*c^2*s*aw/2;C=[c11 c12; c21 c22] D=zeros(2) e11=K_k;e12=0;e21=e12;e22=K_t;E=[e11 e12; e21 e22] % %Find Eigenvalue of Lambda, Frequency and Damping Ratioii=1;for V=1:0.1:180 Q=[zeros(2) eye(2); -A\(roll*V^2*C+E) -A\(roll*V*B+D)]; lambda=eig(Q);for j = 1:4im(j) = imag(lambda(j));re(j) = real(lambda(j));freq(j,ii) = sqrt(re(j)^2+im(j)^2);damp(j,ii) = -100*re(j)/freq(j,ii);freq(j,ii) = freq(j,ii)/(2*pi); endVel(ii) = V;ii=ii+1;end %Find Flutter speedfor dd=1:1:4for l=1:1:numel(damp(1,:)) if damp(dd,l)1 vflut_high(dd)=Vel(l); vflut_low(dd)=Vel(l-1); dd_high(dd)=damp(dd,l); dd_low(dd)=damp(dd,l-1); coef(dd)=dd_high(dd)/-dd_low(dd); vflut(dd)=(vflut_high(dd)+coef(dd)*vflut_low(dd))/(coef(dd)+1); break else vflut(dd)=1000; end endendvflutter=min(vflut); %Display resultfigure(1)subplot(2,1,1); plot(Vel,freq,'k');title('Vf plot','fontsize',15);xlabel ('Air Speed (m/s) '); ylabel ('Frequency (Hz)'); grid subplot(2,1,2);plot(Vel,damp,'k'); hold ontitle('Vg plot','fontsize',15);xlabel ('Air Speed (m/s) '); ylabel ('Damping Ratio (%)'); grid on; hold onplot(vflutter,0,'ro','markerfacecolor','r');text(vflutter,0,['\leftarrowFlutter Speed:',... num2str(roundn(vflutter,-2)),'m/s'],'fontsize',12,'fontweight','bold') fprintf('The flutter speed is %5.2f. m/s.',vflutter) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%b) effect of flexural axis on flutter speed%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% iii=1;for xf=0.1*c:0.1*c:0.8*ce=xf/c-0.25; %Inertial, Damping and Stiffness Matrixa111=(m*s^3*c)/3 + M*s^3/3; a122= m*s*(c^3/3 - c^2*xf + xf^2*c) + M*(s*xf^2);a112 = ((m*s^2)/2)*(c^2/2 - c*xf) - M*s^2*xf/2; a121 = a112;A1=[a111,a112;a121,a122]; b111=c*s^3*aw/6;b112=0;b121=-e*c^2*s^2*aw/4;b122=-c^3*s/8*Mthetadot;B1=[b111 b112; b121 b122]; c111=0;c112=c*s^2*aw/4;c121=c111;c122=-e*c^2*s*aw/2;C1=[c111 c112; c121 c122]; %Find Eigenvalue of lambdakk=1;for V=1:1:200 Q1=[zeros(2) eye(2); -A1\(roll*V^2*C1+E) -A1\(roll*V*B1+D)]; lambda1=eig(Q1); Vel1(kk)=V; %Find frequency and damping ratiofor jj = 1:4im1(jj) = imag(lambda1(jj));re1(jj) = real(lambda1(jj));freq1(jj,kk) = sqrt(re1(jj)^2+im1(jj)^2);damp1(jj,kk) = -100*re1(jj)/freq1(jj,kk);freq1(jj,kk) = freq1(jj,kk)/(2*pi); end %Find Flutter Speedkk=kk+1;for ddd=1:1:4for ll=1:1:numel(damp1(1,:)) if damp1(ddd,ll)1 vflut_high1(ddd)=Vel1(ll); vflut_low1(ddd)=Vel1(ll-1); dd_high1(ddd)=damp1(ddd,ll); dd_low1(ddd)=damp1(ddd,ll-1); coef1(ddd)=dd_high1(ddd)/-dd_low1(ddd); vflut1(ddd)=(vflut_high1(ddd)+coef1(ddd)*vflut_low1(ddd))/(coef1(ddd)+1); break else vflut1(ddd)=1000; end endendendxff(iii)=xf/c;vflutter1(iii)=min(vflut1);iii=iii+1;end %Display resultfigure(2)plot(xff,vflutter1)title(['Flutter speed, V_f_l_u_t_t_e_r (m/s) vs Flexural axis,'... 'x_f at x_c_m = 0.5c'])grid onxlabel('Flexural axis, x_f per chord (m/m)')ylabel('Flutter speed, V_f_l_u_t_t_e_r (m/s)') %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% c)effect of flap and pitch stiffness on flutter speed%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %Reset valuexf=0.5*c;e = xf/c-0.25;%Find Eigenvalue of Lambda and Flutter Speedii=1;for K_k=0.5e7:0.1e7:5e7%Structural Stiffnesse211=K_k;e212=0;e221=e212;e222=K_t;E2=[e211 e212; e221 e222]; k=1;for V=1:1:200 Q2=[zeros(2) eye(2); -A\(roll*V^2*C+E2) -A\(roll*V*B+D)]; lambda2=eig(Q2); Vel2(k)=V; for j = 1:4im2(j) = imag(lambda2(j));re2(j) = real(lambda2(j));freq2(j,k) = sqrt(re2(j)^2+im2(j)^2);damp2(j,k) = -100*re2(j)/freq2(j,k);freq2(j,k) = freq2(j,k)/(2*pi); % convert frequency to hertzend k=k+1; for dd=1:1:4for l=1:1:numel(damp2(1,:)) if damp2(dd,l)1 vflut_high2(dd)=Vel2(l); vflut_low2(dd)=Vel2(l-1); dd_high2(dd)=damp2(dd,l); dd_low2(dd)=damp2(dd,l-1); coef2(dd)=dd_high2(dd)/-dd_low2(dd); vflut2(dd)=(vflut_high2(dd)+coef2(dd)*vflut_low2(dd))/(coef2(dd)+1); break else vflut2(dd)=1000; end endendendK_kappa(ii)=K_k;vflutter2(ii)=min(vflut2);ii=ii+1;end %Display Resultfigure(3)plot(K_kappa,vflutter2)title(...'Flutter Speed, V_f_l_u_t_t_e_r(m/s) vs Flap Stiffness, K_\kappa(Nm/rad)')grid onxlabel('Flap Stiffness, K_\kappa (Nm/rad)')ylabel('Flutter Speed, V_f_l_u_t_t_e_r (m/s)') %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%Reset ValueK_k=1.2277e7; %Find Eigenvalue of Lambda and Flutter Speedii=1;for K_t=1e5:0.1e5:10e5e311=K_k;e312=0;e321=e12;e322=K_t;E3=[e311 e312; e321 e322];k=1;for V=1:1:200 Q3=[zeros(2) eye(2); -A\(roll*V^2*C+E3) -A\(roll*V*B+D)]; lambda3=eig(Q3); Vel3(k)=V; for j = 1:4im3(j) = imag(lambda3(j));re3(j) = real(lambda3(j));freq3(j,k) = sqrt(re3(j)^2+im3(j)^2);damp3(j,k) = -100*re3(j)/freq3(j,k);freq3(j,k) = freq3(j,k)/(2*pi); % convert frequency to hertzendk=k+1;for dd=1:1:4for l=1:1:numel(damp3(1,:)) if damp3(dd,l)1 vflut_high3(dd)=Vel3(l); vflut_low3(dd)=Vel3(l-1); dd_high3(dd)=damp3(dd,l); dd_low3(dd)=damp3(dd,l-1); coef3(dd)=dd_high3(dd)/-dd_low3(dd); vflut3(dd)=(vflut_high3(dd)+coef3(dd)*vflut_low3(dd))/(coef3(dd)+1); break else vflut3(dd)=1000; end endendendK_theta(ii)=K_t;vflutter3(ii)=min(vflut3);ii=ii+1;end %Display Resultfigure(4)plot(K_theta,vflutter3)title(...'Flutter Speed, V_f_l_u_t_t_e_r(m/s) vs Pitch Stiffness, K_\theta(Nm/rad)')grid onxlabel('Pitch Stiffness, K_\theta (Nm/rad)')ylabel('Flutter Speed, V_f_l_u_t_t_e_r (m/s)')%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%d) K-method%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%Reset ValueK_theta=7.2758e5;j=1;for k=1:-0.01:0.1 kk(j)=k; Mthetadot=-5/(2+5*k); b11=c*s^3*aw/6; b12=0; b21=-e*c^2*s^2*aw/4; b22=-c^3*s*Mthetadot/8; B=[b11 b12; b21 b22]; F=A-i*roll*(b/k)*B-roll*((b/k)^2)*C; lambda4=eig(F*inv(E)); for l=1:1:2 F_array(l,j)=F(l); lambda_array(l,j)=lambda4(l); omega(l,j)=1/sqrt(real(lambda4(l))); g(l,j)=imag(lambda4(l))/real(lambda4(l)); V4(l,j)=omega(l,j)*c/(2*k); freq4(l,j)=omega(l,j)/(2*pi); end j=j+1;end for dd=1:1:2for l=1:1:numel(g(1,:)) if g(dd,l)>=0 && l>1 vflut_high4(dd)=V4(dd,l); vflut_low4(dd)=V4(dd,l-1); g_high(dd)=g(dd,l); g_low(dd)=g(dd,l-1); coef4(dd)=g_high(dd)/-g_low(dd); vflut4(dd)=(vflut_high4(dd)+coef4(dd)*vflut_low4(dd))/(coef4(dd)+1); red_freq(dd)=(kk(l)+coef4(dd)*kk(l-1))/(coef4(dd)+1); break else vflut4(dd)=1000; red_freq(dd)=1000; endendend vflutter4=min(vflut4)red_freq1=red_freq(vflut4==vflutter4)flutter_freq=(vflutter4*red_freq1/b)/(2*pi) Vel4(1,length(V4))=0;Vel4(2,:)=V4(1,:);Vel4(3,:)=V4(2,:);Freq(1,length(freq4))=0;Freq(2,:)=freq4(1,:);Freq(3,:)=freq4(2,:);

%Display Resultfigure (5)title('Vf plot','fontsize',15)xlabel ('Air Speed (m/s) '); ylabel ('Frequency (Hz)'); grid on; hold onplot(V4(1,:),freq4(1,:),'b','linewidth',1.5); hold onplot(V4(2,:),freq4(2,:),'g','linewidth',1.5); hold onplot(Vel4,Freq,'--');hold onlegend('Mode 1','Mode 2','Reduced Frequency, k')text(max(V4(1,:)),freq4(1,j-1),['\leftarrownode for V_i at k_i'],... 'fontsize',12,'fontweight','bold') for l=1:1:length(freq4) for ll=1:1:2 plot(V4(ll,l),freq4(ll,l),'ko','markerfacecolor','k','markersize',5) hold on endend figure(6)xlabel ('Air Speed (m/s) '); ylabel ('Damping coefficient, g');grid on; hold on;title('Vg plot','fontsize',15)plot(V4(1,:),g(1,:),'linewidth',1.5);hold onplot(V4(2,:),g(2,:),'g','linewidth',1.5); hold onplot(vflutter4,0,'or','markerfacecolor','r');hold onlegend('Mode 1','Mode 2')text(vflutter4,0,['\leftarrowFlutter Speed:',...num2str(roundn(vflutter4,-2)),'m/s'],'fontsize',12,'fontweight','bold') fprintf('The flutter speed is %5.2f m/s.',vflutter4 )fprintf('\nReduced frequency, k= %5.4f',red_freq1)fprintf('\nFlutter frequency, f= %4.2fHz',flutter_freq)