Upload
sumit-malik
View
222
Download
0
Embed Size (px)
Citation preview
8/21/2019 Application of Panel Methods
1/52
PEMP
ACD2506
Session 02
Application of Panel Methods
Session delivered by:Session delivered by:
. . .. . .
102 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
2/52
PEMP
ACD2506
Session Objectives
-- At the end of this session the dele ate would have
understood
How to test a anel code for accurac
How to validate and interpret results from a panel code
How to interpret airfoil results
The meaning of a higher order panel method
How to extend the method to 3-D cases
202 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
3/52
PEMP
ACD2506
ess on op cs
1. Revision of the Panel Methods2. Verification of the quality of Panel
e o esu s
3. Validation of the Panel Method Results
4. Airfoil Characteristics- Lift PitchinMoment, Drag
5. Special Airfoils, Inverse Design
. -
302 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
4/52
PEMP
ACD2506
INTRODUCTION
n ncompress e ow, e e ec s o v scos y can e
neglected, satisfies the Laplace equation. The continuity equation and conditions of irrotationality lead
to the Laplace equation.
The Laplace equation is a linear PDE and hence a large body
found accurately and efficiently. We are solving only one PDE.
Flows with Mach numberM< 0.3 can be approximated as
ncompress e ows. ence, e v scous e ec s arenegligible, a lowMflow can be approximated by the Laplace
equation and solved conveniently.
402 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
5/52
PEMP
ACD2506
A summary of the Panel method is given below by listing the steps
a en n e r o xamp e.
1. The geometry is to be represented by N panels. It is a closed
surface.
2. Boundary condition
representing a closed surface at infinity
3. Geometrical quantities are calculated. They include: areas,
slopes i , unit vectors normal and tangential to the panels
n i, i , no a or co oca on po n s a ong w e rcoordinates.
4. On each of theNpanels assume uniform source strength q i and
502 M.S. Ramaiah School of Advanced Studies, Bengaluru
the vortex strength , assumed to be same on each panel.
8/21/2019 Application of Panel Methods
6/52
PEMP
ACD2506
5. Now at any point in the field can be expressed in terms of (N
va ues o s ngu ar es an n n y va ue
:
s s v v
relate velocities to singularities. These influence coefficients are
geometric relations and can be evaluated.
602 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
7/52
PEMP
ACD2506
7. The velocities are related to the singularities by the relations:
8. Apply the tangency boundary condition for velocities on theN
pane s. ese are res equa ons or s ngu ar es q i an :
702 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
8/52
PEMP
ACD2506
9. Apply the Kutta condition at the trailing edge to get one more
equa on. s equa on s
10. Thus we have generated (N+1) equations to close the system.
This linear system of equations can be solved for (qi
and ).
can be evaluated at each panel (Bernoullis theorem) and then
lift, drag & moment coefficients can be calculated.
802 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
9/52
PEMP
ACD2506
Variation of pressure coefficient on the top and bottom surfaces is
s own or a r o a ong w e exac resu s o a ne
by conformal mapping.
= 6 1. Notice max & min values.
2. How to get CL, CD, CM?.
4. What do you mean by
exact results here?
. omment on agreement.6. The method can be
extended to 3-D.
902 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
10/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (1)
1. The discussion here is the same as grid independence study in
CFD.
2. T e resu ts we get s ou not epen on t e num er o pane s.
3. The number of panels is not a part of the original flow problem. It
is a numerical artefact we have introduced to solve the roblem.
Hence if the results depend on the number of panels, they are not
the solution of the original problem, i.e., they are not the results.. ,
better.
5. Also, an increase inN, results in better representation of the
s ngu ar ty str ut ons. emem er q i s un orm on a pane .6. The integral representations we use are exact. But they are
restricted to irrotational flows onl .
1002 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
11/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (2)
1. We increase the number of panelsN, and observe if the results
change. If they change we have not reached the desired stage.
2. Q: I t ey o not c ange w t respect to an ncrease n , w at
can we conclude?
3. When we increase , it is uite im ortant to have a strate . Also,
keep in mind that coarseness ofNmeans part of the flow geometry
is not represented properly. The strategy should be such that asN .
4. The effect of an increase inN, should be concluded only if this
increase is sufficiently large. (Not changingNfrom 200 to 210
an t en conc u e t at t e resu ts are nsens t ve to t s c ange5. What is the level of sensitivity of the results to a change inNthat
is acce table? It de ends on what uantities we are ins ectin and
1102 M.S. Ramaiah School of Advanced Studies, Bengaluru
the purpose of our study and also the resources available.
8/21/2019 Application of Panel Methods
12/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (3)
1. Sensitivity of the drag coefficient to the number of panelsN .
2. What should be the asymptotic value asN becomes large?
1202 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
13/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (4)
1. Sensitivity of the lift coefficient to the number of panelsN .
2. Keep in mind the expanded scale used.
3. Comment on t e non-monoton c e av our o t e t coe c ent
to the number of panelsN .
1302 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
14/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (5)
Based on the sensitivity of the results toN as seen in the previous
two figures consider:
1. Comment on the choice of N = 20. See that the lift values are
the same for N = 20 & 100 !
2. Is the value N = 80 or 100 a satisfactory choice?
3. Suppose we decidedN = 100 to be a satisfactory choice (last,
choice for = 2 ? Is it likely to be a satisfactory choice for = 12 ?
4. If we increase the number of panels sufficiently, can we resolvethe boundary layer and make the results more realistic?
1402 M.S. Ramaiah School of Advanced Studies, Bengaluru
.
viscous flow?
8/21/2019 Application of Panel Methods
15/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (6)
1. The same results are plotted now as a function of ( 1/n ) which is
proportional to (or a measure of average) panel size.
2. Sens t v ty o t e rag coe c ent to ( 1 n ).
1502 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
16/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (7)
1. The same results are plotted now as a function of ( 1/n ) which is
proportional to (or a measure of average) panel size.
2. Sens t v ty o t e t coe c ent to ( 1 n ).
1602 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
17/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (8)
1. The same results are plotted now as a function of ( 1/n ) which is
proportional to (or a measure of average) panel size.
2. Sens t v ty o t e moment coe c ent to ( 1 n ).
1702 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
18/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (9)
Based on the sensitivity of the results to ( 1 / n ) as seen in the
previous three figures consider:
1. As the number of panels becomes very large ( 1 / n ) 0 and
we should et exact results.
2. If we plot these graphs on log scale it may be possible to
evaluate the accuracy of the numerical scheme.. D
4. Is it possible to extrapolate CL curve to ( 1 / n ) 0 ?
5. Is it possible to extrapolate Cm curve to ( 1 / n ) 0 ?
. e cons er ng t e top t ree quest ons, ta e a oo at t eprevious two figures also.
1802 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
19/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (10)
The quantities we used ( CD , CL & Cm ) for comparison are integral
values. We will compare now CP distribution. This is a more
eta e an sens t ve test. Argue w y we nee more pane s near
the leading edge.
1902 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
20/52
PEMP
ACD2506
Sensitivity of the Results to the Number of Panels (11)
1. In the previous figure we see that with 20 panels we do not
recoverCP = 1 at the leading edge. But results of 60 & 100
pane s are n st ngu s a egrap ica y.
2002 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
21/52
PEMP
ACD2506
Validation of the Results (1)
1. Step 1: In the previous comparisons we have assumed that the
inviscid model is valid and verifiedthe computational procedure
to ma e sure t at our approx mate so ut on s n ee c ose to t e
exact solution of the assumed model.
2. Ste 2: Once we are convinced of this, it is ossible to check
(validate) how close is the assumed theoretical model to physical
reality.
. .
4. In the validation step 2 here we usually make comparison for a
wider range of parameters of practical relevance.
. t s o ten cu t to ver y our computat ona proce ure orevery possible range of parameters. We, of course, select the
ran e of arameters of ractical interest and then make
2102 M.S. Ramaiah School of Advanced Studies, Bengaluru
verification tests to more stringent cases.
8/21/2019 Application of Panel Methods
22/52
PEMP
ACD2506
Validation of the Results (2)
made here withthe experimental
data.
See next slide
for comments.
2202 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
23/52
PEMP
ACD2506
Validation of the Results (3)
NACA 0012 and NANA 4412. The fist airfoil is symmetrical
and both of them have 12 % thickness.
. e pane met o g ves t e correct s ope o , pre cte y
the thin airfoil theory (Verify).3. For NACA 0012 the curve passes through the origin but for
NACA 4412 the zero lift angle is about Z L = - 4 .
4. Disagreement with the experimental data is gradual as ncreases, ut t s g er or t e a r o .
5. Q: Is it possible to argue that by shifting one set of curves that
the disagreement is roughly the same in both the sets? Keep
thin airfoil theory in mind.6. For the cambered airfoil the flow separates first at the trailing
2302 M.S. Ramaiah School of Advanced Studies, Bengaluru
.
separates first at the leading edge and stall is abrupt.
8/21/2019 Application of Panel Methods
24/52
PEMP
ACD2506
Predicting the Pitching Moment (1)
1. Pitching moment plotted is about the quarter chord point.
2402 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
25/52
PEMP
ACD2506
Predicting the Pitching Moment (2)
1. For the NACA 0012 airfoil experimental data indicate that the
quarter chord point is the aerodynamic centre (recall the
.
indicating that the quarter chord point is not the aerodynamiccentre.
2. Once the flow separates disagreement is large.
3. For the cambered NACA 4412 airfoil experimental data indicate
- .
4. Recall that for this airfoil flow separates at the trailing edge and
disagreement is gradual as is increased. Pitching moment is
large and there is a qualitative difference between the two setsof results.
5. Com arison made here is ver exactin and also of ractical
2502 M.S. Ramaiah School of Advanced Studies, Bengaluru
importance for stability. But failure is not of computations.
8/21/2019 Application of Panel Methods
26/52
PEMP
ACD2506
Discussion of the Airfoil Results by Panel Methods (1)
1. The panel methods solve for an irrotational flow. Hence the results
may be OK for thin airfoils at low Mach number (M< 0.3) and small
.
2. Results may be acceptable for pressure distribution and lift but notfor drag. See the next figure.
3. The 2-D drag value predicted by the panel methods should be strictly
zero but there may be a small value indicating numerical error.
4. The skin friction dra of an airfoil cannot be redicted b the anel
methods. Because of viscous effects the rear stagnation pressure will
not be fully recovered and will contribute for a small value of
. .
be predicted by an inviscid model.
5. Since the panel methods are accurate and fast, they can be used to
2602 M.S. Ramaiah School of Advanced Studies, Bengaluru
study the effect of geometry keeping in mind viscous effects.
8/21/2019 Application of Panel Methods
27/52
PEMP
ACD2506
Discussion of the Airfoil Results by Panel Methods (2)
1. Key areas of interest in airfoil pressure distribution.
2702 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
28/52
PEMP
ACD2506
Airfoil Characteristics (1)
Effect of Angle of Attack :NACA 0012 Airfoil
CP minimum
value decreases
with increase in
A larger
deceleration with
2802 M.S. Ramaiah School of Advanced Studies, Bengaluru
8/21/2019 Application of Panel Methods
29/52
PEMP
ACD2506
Airfoil Characteristics (2)
ec o r o c ness:
= 0 NACA 0012 Airfoil
CP minimum
value decreases
with increase in
thickness
A larger
increase in
thickness
2902 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
30/52
PEMP
ACD2506
Airfoil Characteristics (3)
ec o r o c ness:
= 4 NACA 0012 Airfoil
The thinnest airfoil
shows a large expansion /
recom ression due to the
stagnation point being
below the L.E.
The thicker airfoil
results in milder expansion
and subsequentrecompression.
3002 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
31/52
PEMP
ACD2506
3102 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
32/52
PEMP
ACD2506
Airfoil Characteristics (4)
Effect of on Cambered Airfoil CP:
= 0 & 4 NACA 4412 Airfoil
Due to camber lift is
generated without a large
. .
subsequent recompression.
T s re uces t e
possibility of L.E.
se aration. See next
figure for comparison.
3202 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
33/52
PEMP
ACD2506
Airfoil Characteristics (5)
am er ec s on r o P:
CL = 0.48 for both the airfoils Due to camber lift is
generated without a
lar e L.E. ex ansion and
subsequent
recompression.
Next two figures
indicate it for CL = 0.96
& for CL = 1.43.
3302 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
34/52
ACD2506
Airfoil Characteristics (6)
am er ec s on r o P:
CL = 0.96 for both the airfoils
3402 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
35/52
ACD2506
Airfoil Characteristics (7)
am er ec s on r o P:
CL = 1.43 for both the airfoils
3502 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
36/52
ACD2506
Even though the panel methods do not give drag values, it is
nstruct ve to oo at t e rag va ues or t ese two a r o s anappreciate the effectiveness of the camber.
3602 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
37/52
ACD2506
Airfoil Characteristics (8-A)
ec s o am er on r o P:NACA 6712 Airfoil
E ect o extreme a t cam er was use y R c ar W tcom n
the development of the NASA supercritical airfoils. It o ens u the ressure distribution near the leadin ed e. See
next figure.
3702 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
38/52
ACD2506
Airfoil Characteristics (8-B)
ec s o am er on r o P:NACA 6712 Airfoil Aft camber opens
up the pressure
distribution near the
leading edge.
But it has led to a
large zero lift pitching
.
It has also led to
delayed and thenrapid pressure
recover and ossible
3802 M.S. Ramaiah School of Advanced Studies, Bengaluru
early B.L. separation.
PEMP
8/21/2019 Application of Panel Methods
39/52
ACD2506
Airfoil Characteristics (9-A)
ec s o am er on r o P:GA(W)-1 also known as NASA LS(1)-0417 Airfoil
T e a r o s s own ear er were eve ope n t e 1930s an t e
geometry was given by simple formulas. Modern airfoils developedin the 1970s are defined b tables of coordinates.
The figure below shows 17 % thick GA(W)-1 airfoil (General
Aviation) .
3902 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
40/52
ACD2506
Airfoil Characteristics (9-B)
ec s o am er on r o P:GA(W)-1 Airfoil, 17 % thick It has better
maximum lift and
stall characteristics.
The ressure
recovery rate is
constant and hence
.
If the camber is too
steep, flow mayseparate first at the
bottom surface .
4002 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
8/21/2019 Application of Panel Methods
41/52
ACD2506
Laminar Rooftop Airfoil by Liebeck for High Lift
Note the high lift curve in the next figure.
Design and tests from R.H. Liebeck,Re . MDC-J5667/01, Au ust 1972
From: R.T. Jones, Wing Theory.
4102 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506
8/21/2019 Application of Panel Methods
42/52
ACD2506
for High Lift
Note the pressure recovery in the
.
Design and tests from R.H.e ec ,
Rep. MDC-J5667/01, August
1972
From: R.T. Jones, Wing Theory.
4202 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506Drag value for Liebecks High Lift Airfoil
8/21/2019 Application of Panel Methods
43/52
ACD2506Drag value for Liebeck s High Lift Airfoil
Note the low drag value even at very high lift!
4302 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506
8/21/2019 Application of Panel Methods
44/52
ACD2506
Inverse Problem of Airfoil Design
1. The pressure distribution is specified. Of course, it has to be
something possible.
.
distribution.3. This is the inverse problem different from the analysis problem
where we obtain the pressure distribution corresponding to a
known geometry.
4. If we succeed in the inverse desi n it is ossible to enerate anideal airfoil section to meet an specific needs, provided it meets
other requirements, for example from the structural viewpoint.
.
does not exist. Hence the restriction in item 1 above.
6. Panel method being accurate and fast comes in a handy, specially
4402 M.S. Ramaiah School of Advanced Studies, Bengaluru
in 2-D problems. But we have to suitably modify them.
PEMP
ACD2506
A E l f I Ai f il D i
8/21/2019 Application of Panel Methods
45/52
ACD2506
An Example of Inverse Airfoil Design
4502 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506
8/21/2019 Application of Panel Methods
46/52
EXMP: Wing-body-tail configuration with wakes
4602 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506
8/21/2019 Application of Panel Methods
47/52
EXMP: Details of the wake model required.
4702 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506
8/21/2019 Application of Panel Methods
48/52
EXMP: The space shuttle mounted on a Boeing 747.
4802 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506
8/21/2019 Application of Panel Methods
49/52
It is tempting to apply the panel methods to 3-D flows.
A 3-D lifting body has induced drag even in a steady, irrotational
flow. Panel methods can be used to calculate the induced drag. But.
Application of the Kutta condition needs a careful consideration.
It applies to distinct edges. It can lead to difficulties.
ow o mo e e wa es a so nee s care u cons era on.
Since the 3-D bodies considered are usually complex, the wake
from one part may interfere with some other parts and hence the flow
may not be strictly irrotational. Usually quadrilateral panels are used to define a surface. It
4902 M.S. Ramaiah School of Advanced Studies, Bengaluru
.
PEMP
ACD2506
8/21/2019 Application of Panel Methods
50/52
Hi her Order Panel Methods
Panel methods are approximate methods.
,
we approximate the geometry, (2) how well we approximate theequations and (3) the round-off errors in the arithmetic. The last part
can be controlled and we will not bother about it here.
We need a larger number of panels for an accurate representation
of the eometr . Often non- lanar anels are used. In higher order panel methods singularity distributions are not
constant on the panel. This improves the formal accuracy of the
.
A smaller number of panels can be used in higher order methods.
However, in practice the need to resolve geometric details dictates
5002 M.S. Ramaiah School of Advanced Studies, Bengaluru
the number of panels and we cannot reduce the number of panels.
PEMP
ACD2506
8/21/2019 Application of Panel Methods
51/52
Summary
The following topics were dealt in this session
1. Revision of the Panel Methods
2. Verification of the quality of Panel Method Results
.
4. Airfoil Characteristics- Lift, Pitching Moment, Drag
5. Special Airfoils, Inverse Design
6. Higher order and 3-D Panel Methods
5102 M.S. Ramaiah School of Advanced Studies, Bengaluru
PEMP
ACD2506
8/21/2019 Application of Panel Methods
52/52
5202 M.S. Ramaiah School of Advanced Studies, Bengaluru