raalf / freewake Goto Github PK

View Code? Open in Web Editor NEW

10.0 10.0 9.0 7.37 MB

A higher-order potential flow code.

License: GNU General Public License v2.0

C++ 94.15% C 3.18% MATLAB 2.67%

freewake's People

Contributors

Stargazers

Watchers

Forkers

bill-b octwanna suozhenapple bramesfeld sjy353 budhyant wkemp-edu alexruiz95 unoah-dev

freewake's Issues

Outputs needs to be cleaned

Remove outputs that aren't used
Check format of outputs and clean up
Check values

Normal force is not symmetric when m>1 in non-circling flight

The A,B and C values for the trailing edge row of elements are symmetric, but the induced normal force is not symmetric. This is leading to a roll moment.

This condition doesn't work?

The value of k doesn't look right in some cases using the default input

FreeWake/Source/wing_geometry.cpp

Lines 168 to 175 in 8dea884

 while((panelPtr[k].left-1<=panel2[wing] && panelPtr[k].left-1 !=-1) \ 

 || (panelPtr[k].right-1<=panel2[wing] && panelPtr[k].right-1 !=-1)) 

 { 

 panel2[wing]=k; 

 panel1[wing+1]=k+1; 

 panel2[wing+1]=k+1; 

 k++; 

 }

check computation of total forces and moments about CG/reference point.

We know CFX is negative

Update viscous corrections to use unique velocities

Are these forces ever used?

These forces calculated in drag? are they ever used?

FreeWake/Source/drag_force.cpp

Lines 442 to 444 in f45eef5

 surfacePtr[i].Force[0] += R[0]; 

 surfacePtr[i].Force[1] += R[1]; 

 surfacePtr[i].Force[2] += R[2];

Are the CG and panels rotated correctly?

We move the CG to account for turning flight, and descending flight, but the CG is not rotated to account for bank angle or alpha (when horiz ==1).

We rotate panels (bank and alpha) about the global 0,0,0 origin, not the CG. Is this what we want?

side forces of planar wing without bank angle during turning flight

Need to minimize side forces without bank angle.

What does this mean?

induced velocity in FW18

add to induced_velocity.cpp line 146 'j++'
Otherwise, will not add up surface DVE velocities correctly.

if statement in equ_system.cpp causes errors with multiple m

FreeWake/Source/equ_system.cpp

Line 886 in d36cf24

if( (pLeft!=panel) && (\

Put most current version in header of program

Update that when program is executed, it also includes a date of the version that is being used.

Fix CDi which is negative during turning flight.

This probably requires a reference frame transformation.

make sure that camber and airfoil stuff still works, including their blending.

Rudder deflection doesn't always work

Model the following panel (vertical stab with rudder def. 45deg) and run with phi = 45.0, circling,horizontal flight, alpha 10.

Panel #: 1. Number of elements across span n = 2 and chord m = 4
Neighbouring panels (0 for none) left: 0 right: 0
xleft yleft zleft chord epsilon Bound.Cond. Airfoil Hinge Deflection
5.500 0.000 0.000 0.350 0.000 100 2 0.700000 45
xright yright zright chord epsilon Bound.Cond. Airfoil Hinge Deflection
5.500 0.000 2.500 0.350 0.000 100 2 0.700000 45

The rudder deflects in the plane of the vstab, not out of the plane.
Maybe an issue here with quadrants:

FreeWake/Source/wing_geometry.cpp

Lines 610 to 622 in fd3e186

 ceps = cos(surfacePtr[l].epsilon); //needed for tempA below 

 seps = sin(surfacePtr[l].epsilon); 

 //computing the mid-span location of the DVE LE 

 //temporarily saved in xleft 

 tempA[0] = x1[0]+xLE[0]*tempS; 

 tempA[1] = x1[1]+xLE[1]*tempS; 

 tempA[2] = x1[2]+xLE[2]*tempS; 

 //vector from LE mid-span point to reference point 

 delX[0] = surfacePtr[l].xsi * ceps; 

 delX[1] = surfacePtr[l].xsi * seps * sin(nu2); 

 delX[2] = -surfacePtr[l].xsi * seps * cos(nu2);

FW aileron trim won't work with equal deflections

When roll trim is ON, the check we do to make sure there is an aileron will fail if there are two ailerons with equal but opposite deflection. (Ex. -5 and +5)

FreeWake/Source/longtrim.cpp

Lines 122 to 130 in ba2bafc

 tempsum = 0; 

 for (i = info.panel1[0]; i <= info.panel2[0]; i++) { //for all panels on wing 1 

 tempsum = +panelPtr[i].deflect1; 

 } 

 if (tempsum == 0) { 

 printf("No aileron defined on wing 1.\n You must define a deflection on at least one panel on wing 1. \n"); 

 scanf("%c", &answer); 

 exit(1); 

 }

Also unrelated but we can maybe decrease the allowed error on Cl for roll trim.

mkdir doesn't work on windows

This doesn't work on windows.

FreeWake/Source/Main_PerfCode2020.cpp

Line 155 in de0d94b

mkdir(info.output,0777);//create output directory in output/

Instead, windows needs:
#include "direct.h"
_mkdir(info.output);
But I don't think that will work on mac. So we need a solution for both

first and last span indices are wrong

in the timestep output file (ex. timestep25.txt), the first and last span indices are incorrect with the new T tail geometry.

Adding CMoWing twice to residual moment

Is the zero lift moment supposed to be added twice? but only one is added if camber is off....

FreeWake/Source/longtrim.cpp

Lines 78 to 79 in d83d089

 //adding zero lift of wing only if camber is turned off 

 if(~info.flagCAMBER){CM_resid += info.CMoWing;}

FreeWake/Source/longtrim.cpp

Lines 104 to 105 in d83d089

 //adding zero lift of wing  

 CM_resid += info.CMoWing;

relaxed wakes are not connected to TE

Adding calculation of area

a few ways to do this. I can calculate the panel areas and add them up, and also calculate the projected panel areas and add them up. These values will not change with camber or flap deflection.

We can also add up the DVE element areas. Those values would change with camber, and the projected areas would change with flap deflection.

Also, what will these values be used for? for example, what if we want to non-dimensionalize to just the wing area and not include the tail? how will the user approach this?

deltae is sometimes -inf after first timestep and causes the main timestep loop to break

I've only seen the issue when n = 1 and trying the infinite wing case

FreeWake/Source/PitchMoment.cpp

Line 425 in 0d134a3

} while((deltae > info.deltae) && (timestep<info.maxtime));

Continue comparison with VoGen

What are you counting over? i or k?

FreeWake/Source/read_input.cpp

Lines 955 to 960 in 8dea884

 for (k = 0; i < Row; ++i){ 

 for (j = 0; j < Col; ++j){ 

 dataPtr[i][j][0] = 0; 

 dataPtr[i][j][1] = 0; 

 } 

 }

should uTE be nondimensionalized?

When moving the elements in the ind. drag computation, should uTE be nondimensionalized? U is nondimensionalized for the non circling case...

FreeWake/Source/drag_force.cpp

Lines 220 to 231 in 98a3009

 
 if(info.flagCIRC){ //Added by D.F.B. 03-2020 for circling flight 

 // NOTE: Move *induced point* in its own velocity direction 

 tempS=dot(delX,surfacePtr[index].uTE[k]); 

 scalar(surfacePtr[index].uTE[k],tempS,tempB); 

 } else{ // if not circling flight, move pts in freestream direction (U) 

 //delX projected into the freestream direction (magnitude) 

 tempS=dot(delX,surfacePtr[index].U); 

 //vector from TE of s(inducer) to TE of index (induced) projected into the freestream direction (with direction) to make tempB 

 scalar(surfacePtr[index].U,tempS,tempB); 

 }

eN and eL directions are calculated using control point velocities, not the LE velocities

eD direction is found with TE velocity

Number of DVEs on flap

For the current version, the LE of the DVEs has to match hinge line. For example, chord 10, flap 20%, m=5 works fine; m=6 does not work
Suggestion, adjust DVEs on flap that at least one DVE on flap, otherwise round m*percentage of flap to full number.
Issue: DVEs of one span wise location will have different chords.

rotating panels in this order causes the panel to have sweep

Rotating in Z-X-Y causes the panels to have sweep when using a bank angle. Switching to Z-Y-X fixes this problem, but I don't know if it causes another problem. Z-Y-X is a more common order.

FreeWake/Source/wing_geometry.cpp

Lines 280 to 295 in dee48ae

 //rotate X1 

 rotateZ(panelPtr[panel].x1,psi,tempA); 

 rotateX(tempA,-nu,tempAA); 

 rotateY(tempAA,eps,panelPtr[panel].x1); 

 //rotate X2 

 rotateZ(panelPtr[panel].x2,psi,tempA); 

 rotateX(tempA,-nu,tempAA); 

 rotateY(tempAA,eps,panelPtr[panel].x2); 

 }//END loop over panels 

 // II. updated CG location (RefPt) 

 //rotate reference point (CG) 

 rotateZ(info.RefPt,psi,tempA); 

 rotateX(tempA,-nu,tempAA); 

 rotateY(tempAA,eps,info.RefPt);

small errors in flap hinge position cause big problems

Depending on the compiler, this condition isn't always satisfied. Why.

The 1000* is just a sketchy fix for now.

FreeWake/Source/wing_geometry.cpp

Line 1467 in fd3e186

}else if(((tempA[0] / panelPtr[i].c1) -panelPtr[i].hinge1) < (1000.*DBL_EPS)){

FreeWake/Source/wing_geometry.cpp

Line 1507 in fd3e186

}else if(((tempAA[0] / panelPtr[i].c2 )-panelPtr[i].hinge2)<(1000.*DBL_EPS)){

Branch: FreeWake2020_dev - Different m values do not work with output files

Having multiple m values does work with the current timestep#.txt files. Unable to plot the results.

Does beta work?

Beta has to be zero if using symmetry for force calcs:

FreeWake/Source/lift_force.cpp

Line 784 in d7c7a05

if (info.sym==1 && info.beta == 0)

Not sure if lift is calculated correctly with beta:

FreeWake/Source/lift_force.cpp

Lines 445 to 449 in d7c7a05

 //free stream lift 

 N_force[l][0] = N_free * sqrt(eN[0]*eN[0]+eN[2]*eN[2]); 

 if (eN[2]<0) N_force[l][0] *= -1; //neg. if resultend is downward 

 N_force[l][1] = dot(R,eL); //induced lift

We are not including the lift from the vortex sheets. Is this significant?

Calculation of 'r' vector

Since the gradient is used to calculate omega for only the XY components of velocity, we can define our gradient as Vk/ R, where Vk is Uinf*cos(alpha) during descending flight.

Since we want the 'r' vector in the xy plane, we should be calculating the 'r' vector as Uinf*cos(alpha) / gradient, which is equal to Vk/grad, and not Uinf/grad as is done now:

FreeWake/Source/PitchMoment.cpp

Line 88 in dee48ae

circCenter[1] = XCG[1]-info.Uinf/info.gradient;

And then we should not be multiplying the gradient by cos alpha here, because we will be calculating grad as Uinf cos(alpha) / R

FreeWake/Source/wing_geometry.cpp

Line 1269 in dee48ae

omega[2] = -info.gradient*cos(info.alpha);

fixed junctions with relaxed wake

More information?

	while((panelPtr[k].left-1<=panel2[wing] && panelPtr[k].left-1 !=-1) \
	\|\| (panelPtr[k].right-1<=panel2[wing] && panelPtr[k].right-1 !=-1))
	{
	panel2[wing]=k;
	panel1[wing+1]=k+1;
	panel2[wing+1]=k+1;
	k++;
	}

	surfacePtr[i].Force[0] += R[0];
	surfacePtr[i].Force[1] += R[1];
	surfacePtr[i].Force[2] += R[2];

	ceps = cos(surfacePtr[l].epsilon); //needed for tempA below
	seps = sin(surfacePtr[l].epsilon);

	//computing the mid-span location of the DVE LE
	//temporarily saved in xleft
	tempA[0] = x1[0]+xLE[0]*tempS;
	tempA[1] = x1[1]+xLE[1]*tempS;
	tempA[2] = x1[2]+xLE[2]*tempS;

	//vector from LE mid-span point to reference point
	delX[0] = surfacePtr[l].xsi * ceps;
	delX[1] = surfacePtr[l].xsi * seps * sin(nu2);
	delX[2] = -surfacePtr[l].xsi * seps * cos(nu2);

	tempsum = 0;
	for (i = info.panel1[0]; i <= info.panel2[0]; i++) { //for all panels on wing 1
	tempsum = +panelPtr[i].deflect1;
	}
	if (tempsum == 0) {
	printf("No aileron defined on wing 1.\n You must define a deflection on at least one panel on wing 1. \n");
	scanf("%c", &answer);
	exit(1);
	}

	//adding zero lift of wing only if camber is turned off
	if(~info.flagCAMBER){CM_resid += info.CMoWing;}

	for (k = 0; i < Row; ++i){
	for (j = 0; j < Col; ++j){
	dataPtr[i][j][0] = 0;
	dataPtr[i][j][1] = 0;
	}
	}


	if(info.flagCIRC){ //Added by D.F.B. 03-2020 for circling flight
	// NOTE: Move induced point in its own velocity direction
	tempS=dot(delX,surfacePtr[index].uTE[k]);
	scalar(surfacePtr[index].uTE[k],tempS,tempB);
	} else{ // if not circling flight, move pts in freestream direction (U)
	//delX projected into the freestream direction (magnitude)
	tempS=dot(delX,surfacePtr[index].U);

	//vector from TE of s(inducer) to TE of index (induced) projected into the freestream direction (with direction) to make tempB
	scalar(surfacePtr[index].U,tempS,tempB);
	}

	//rotate X1
	rotateZ(panelPtr[panel].x1,psi,tempA);
	rotateX(tempA,-nu,tempAA);
	rotateY(tempAA,eps,panelPtr[panel].x1);

	//rotate X2
	rotateZ(panelPtr[panel].x2,psi,tempA);
	rotateX(tempA,-nu,tempAA);
	rotateY(tempAA,eps,panelPtr[panel].x2);
	}//END loop over panels

	// II. updated CG location (RefPt)
	//rotate reference point (CG)
	rotateZ(info.RefPt,psi,tempA);
	rotateX(tempA,-nu,tempAA);
	rotateY(tempAA,eps,info.RefPt);

	//free stream lift
	N_force[l][0] = N_free * sqrt(eN[0]eN[0]+eN[2]eN[2]);
	if (eN[2]<0) N_force[l][0] *= -1; //neg. if resultend is downward

	N_force[l][1] = dot(R,eL); //induced lift

raalf / freewake Goto Github PK

freewake's People

Contributors

Stargazers

Watchers

Forkers

freewake's Issues

Recommend Projects

Recommend Topics

Recommend Org