CHAPTER 2
CHAPTER 2 Existing Software Used in Development of PREDAVOR The PREDAVOR methodology uses as its foundation the capabilities of
several existing ~ools . The input and outpu~ 0: these ~ools are then
linked ~oge~her with new code to produce an overall methodology.
The advantages and disadvantages of using existing code, rather than developing completely new code , were examined carefully when planning the PREDAVOR framework . Using existing tools eliminated the need to duplicate effort. It makes little sense to write code to perform a task when such a code already exists. In addition, it can be assumed that an existing code is further along in its validation process, and thus more robust. The chief disadvantage to using several different codes is linking the codes together in a cohesive manner. Different codes imply different input and output format, different programming languages , and potentially different operating environments .
In this particular case, two primary codes were heavily in use prior to ~he project development . The decision was made to use these codes as the foundation for PREDAVOR, and to link the software packages together using new code.
ACSYNT The ACSYNT aircraft design code is used to generate the wireframe model used in the PREDAVOR analysis . The workstation-based ACSYNT
.J
(AirCraft SYNThesis) is modular in design. Each discipline, such as
I
) aerodynamics, weights, or economics, is contained in an individual module, and these modules are linked together through an analysis package. The code is capable of analyzing a wide variety of aircraft including civil and military aircraft, fighters , bombers , and transports. The modular components of ACSYNT allow analysis of a single discipline, or the modules can be combined in ord er to evaluate the integrated results{l). Currently,
J
.J
.. -- '- - _.- ._--
r~ --~' - -- -- -- ... - ----~ I I ACS~t,,! ::.s adm ' is"Cered by V irgin':'a ?oly"Cechnic lJniversity i ~ ) . Original:y, nowever, ACSYNT was developed by NASA hffies Research Center for concep"Cual deslgn s"Cudies of advanced aircra="C and is still heavily in use today.
The real power of ACSYNT lies ~n i s non-linear optimization code.
This methodology allows the vehicle to be optimized for a particular oo:ec"Cive :unC"ClOr. or =unc"Cions ( su~h as gross "Cakeoff weight ) , given var~ous restralnts . In order for ~he non-linear optimization code "[0 be as realistic and feasib l e as possible, it is important for all of the components of the syn"[hesis process to be modeled correctly. For "[his reason, the modules in ACSYNT are parameter driven with equations derived from theory as opposed to table look -up methods (3; .
It is future goal of this project to use this optimization package to automate the handling qualities optimization scheme.
The version of ACSYNT currently being used in the PREDAVOR project includes a CAD interface written entirely in the three-dimensional graphics standard PHIGS (Programm er's Hierarchical Interactive Graphics System) [ ~ J .
This CAD package allows a model of the aircraft to be rapidly constructed using component templates ( see Figure 2 . 1.) Once the model is completed, it may be easily transferred into a file format that can be used by the other codes in the PREDAVOR methodology .
)
VORLAX
<:1
PREDAVOR uses a vortex lattice method called VORLAX to generate the I
J
forces and moments on the model that are used to calculate the stability derivatives . Variations of t he basic vortex lattice method are currently being used to analyze both planar and non-planar aircraft configurations.
I The beauty of the vortex lattice metho d lies in the simplicity of its
J
numerical technique as well as its high degree of accuracy (within the limi ts of the basic theory) [ 5 ! .
J
J
-------~-.-------.-------- .
r--- .- _. -- ----l
ENI'En COLOR NJMSER , £NI[R COl.CR ., OR SEu:cT NDol.J IfD,jI : igu re 2. 1 .
ACSYNT Screen and Model with Wing Template Basic Vor :ex Lattice Theory The basic vortex lattice me hod involves superimposing a finite
number of horseshoe vortices of different strengths rn on to the surface of
the model. Consider , for examp e , pa r t of a finite wing shown in Figure
2 . 2. A horseshoe vortex (abcd) of strength rn is placed upon a
.J
representative trapezoidal panel. The velocity induced at an arbitrary point P(x,y) by this single horseshoe vortex can be calculated using the J . ..1 SioL-Savart: Law : Vortex Rlament '\
of strength r
, / r dI x r
J dV
41r
Irr
p
J
dV
J
J
Ir. order to analyze an aircraft ( or any other shape ) , the entire surface is replaced with a series of representative trapezoids (Figure 2 .3 ) . A horseshoe vortex is then placed on each trapezoid. The total induced velocity at point P(x , y) may again be found using Biot - Savart. By applying the flow tangency condition to all control points, a system of slmultaneous equation may be obtained and solved for the unknown circulation s ( r r. ' s ) i 6 ] • These , in turn, directly correspond to the forces and moments acting upon the model.
. F'(x.y) ."' e d Figure 2.2 Schematic of a Single Horseshoe Vortex Source: Anderson , J. D. Fundamentals of Aerodynamics. New York : McGraw- Hill, 1991 .
Figure 2 . 3
J
Trapezoidal Half - Model of Aircraft
I
~~~houg~ vortex ~a~~ice methods are cu ~ren tly being used and have p~oveo ~o be prac icai and ver sat ile tools, most analysis has been largely st.:bsonlc. The applicabi_ity of tne basic techniques of vortex lattice i7 theory to supersonic flow has been _argely ignored ; . VORLAX, developed by Lockheed in 1977 , is applicable to both subsonic and supersonic flight CGr.ai::lons. ~he supersonic capabili~y is justified as follows. Assume tha~ t~e discrete vc~tex lattice app~oximates the vorticity on the surface The mathematical representation of this includes an integ~al tha~ has a residual term of the velocity field. Using this residual term correctly by including it in the resulting velocity field generated by the vortex lines , allows the calculation , and thus applicability, for supersonic flow lSl .
In addition, the VORLAX method ~ncludes special techniques for simulatlng the thickness of lifting surfaces using a double (bi-planar) vortex lattice layer. VORLAX is also capable of analyzing fusiform bodies by arranging a vortex grid on a series of concentric cylindrical surfaces .
These concepts are all illustrated in Figure 2.4 which shows a generalized vortex lattice model of a wing - body configuration.
I Ho rresh oe Free Legs .J
\\
I '. I / '; '/ J / Figur e 2 . 4 Generalized Vortex Lattice Model of Wing-Body Configuration
J
Source: Recreated from Miranda , L . R ., a nd R.D. Elliot and W . M . Baker.
"NASA CR-2865 A Generalized Vortex Lattice Method for Subsonic and Supersonic Flow Applications, " NAS11 - 12972 . Dec . 1977.
J
---- - ,. '-' -- _ .. -- c-- - -- -
l
:r. ~ rder LO fac~liLaLe Lhe ~~~~e~ complex inpu~ LO VORLAX , NASA Ames has developed a grapr.ical pre-pro~ess~r to Lhe code, called VORVIEW [9, . The lnp~L LO VORLAX had consisted of lengLhy files that numerically defined ~he coordinaLes of each L~apezold, as we~~ as oLher required informaLion for ana~ysis , There was no visual feedjack of ~he model being analyzed , and cha~ges LO the model were manual a~c Ledious, VORVIEW , on the other hand , uses as its input the wireframe geometry generated by ACSYNT (Figu re 5) , This file, together with a data file containing flight conditions, is used to launch VORVIEW, The wireframe model may then be "sliced" from wi~g Lip to wing Lip , and subdivided into trapezoids, Instead of defining each trapezoid numerically , as the input LO VORLAX requires, VORVIEW allows the trapezoids to be created graphically and the manual input file to VORLP~ created automatically.
,, -1 . , Figure 2.5
J
Wire frame Model Generated by ACSYNT and used by VORVIEW
I
I
I
J I
~
Wh~le VORVIEW does a nice job slicing the model in the planform view , it does ~ot currently posses the capability to create ve rtical surfaces automat ic ally . These panels can , however, be created by hand.
Figure 2.6 shows a sliced and subdivided VORVIEW model .
A:~er the model has been sliced and subdivided, VORVIEW transforms ~he ja:~ ana runs VORLAX .
The OUtput ~s shown both graphically as a Cp distric~~ion ( Figure 2 . 7 ) and numerica:ly as forces and moments in an output :::':'le.
Figure 2 . 6 Sliced and Subdivided VORVIEW Model <1 .22 0 4 . 643 I I
.J
-0 . 00 2
J Figure 2 . 7
VORVIEW Output Showing Cp Distribu t ion
·. 1
J
CHAPTER 3
CHAPTER 3 The PREDAVOR Code Methodology Bef o re PREDAVOR Al~hough rat her tedious, the tools of the previous sect ion cou l d be used in succession to generate the stability derivatives of a given mode~. The meLhodology would be as follows: Crea Le a model using ACSYNT .
2 . EdiL the input file for iniLial flight conditions.
3 . Run VORVIEW/VORLAX .
s. Manu a lly parse out the resu l - ing forces and moments from the
output file.
5. Edit the input file to contain a perturbation of the flight condiLions . ( F or example , change a = 8 deg to a = 2 deg. ) 6. Re - run VORVI EW /VORLAX .
Manually parse out the new results.
5 . Manu all y calcula te the stab ility derivative from the results of both runs.
9. Edit the input file to undo he perturbation.
10 . RepeaL steps 1 - 9 for each deri v aLive.
In o rder to generate a complete set of deri vative s, the VORVI EW /VORLAX combination would need to be run once at unperturbed . j conditions , and once for each perturbation needed (al ph3, beta, pitch rate, yaw rate, roll rate , control surface deflections, and change in forward velocity.
The results of each of these runs must be parsed, and each deri v aLive calculated by hand . Thus , to generate a standard set of derivatives, many runs of VORVIEW/VORLAX must be made and many sets of manual calculations performed. The entire process must be repeated if J
.J
analysis is needed at a different flighL condition.
While iL is certainly possible to generate sets of stability
J
J
derivatives in Lhe above manner, it is not practical. PREDAVOR was designed LO aULomaLe Lhis process. This has several advantages. The :irst is Lhe eliminaLion of tedious hand calculations involving multiple
ru~s 0: Lhe code , rna ual manipulatlon of ~he input files, parsing large
OULPUt f~les, manual axes transformatlo~s , and the calculations of Lhe stability derivatives hemselves . Secondly, accuracy may be improved through the elimination of many sources of human error. Thirdly, time is saved through multiple autonomous runs of the VORLAX code. And finally, by automating this process, it is possible to one day incorporate the PREDAVOR methodology into a mathematical optimizaLion scheme, such as COPES/CONMIN associated with the ACSYNT package ,l . .
PREDAVOR Architecture fig 3.1 illustrates the overall PREDA VOR architecture . The first step is the creation of the three - dimensional wireframe model using the CAD package in ACSYNT. Next , generic flight conditions and a few basic geometric parameters are added to the VORVIEW input file .
The graphical pre - processor VORVIEW is then used.
VORVIEW's current capabilities allow the user to slice the planform view of the aircraft from wing tip to wing tip. In order to calculate the latera l derivatives, however, a model of the vertical i surfaces needs to be included. These vertical panels may be created ~ manually by editing a supplementary file that includes the geometric
j
slice data. The user simply adds the X, y, and Z locations of each of the four points of the trapezoid to be created to the file. VORVIEW allows the newly created trapezoid to be vi ewed graphically. Figure 3.2 shows a three-view of a model created in this manner.
Once a satisfactory slice model is created, VORVIEW is run once to create the appropriate input file to VORLAX . Once this file is created,
J
PREDAVOR edits it automatically, allowing multiple runs of VORLAX to be performed independently of VORVIEW.
j
1 - =L.~ , B§ l - 1
----.- - .. - -- GrapbiW Interface Figure 3.1 PREDAVOR Code Architecture Figure 3 . 2
J
VORLAX Model with Manually Created Vertical Panels
J
At :~~s ?oint the user may add control surfaces to the aircraft.
Thls ~s aone via a control s urface menu in VORVIE~. A planform of the sl~ced a~rcra:t is shown, and con::rol surfaces added by clicking on ::he appropr~a::e ?anel . Control surface type , per cen:: chord length , and deflection angle are all inputs. The control surface part of VORVIEW was modif~ed to allow separate input files to be created for each control surface (Figure 3 . 3) . A toggle button allows the user to choose between elevc.::or , aileron, and " other ". The control surface is created using a po~n:: and click technique, and the user presses the "SE T INPUT" button to create the new control surface input file. Because the control surface process in VORVIEW works only from a pl anform view , rudders may not be creat ed explicitly in this manner . The " other " option was created to anticipate VORVIEW ' s future ability to create vertical panels automatically. Until then, the user simply creates a deflected rudder manually, using the method described earlier to create vertical panels by hand . The derivative may be calculated using the , steps outlined at the beginning of this chapter.
The next step for the user is to edit the PREDAVOR input file to include the prop er flight conditions, baseline flight variable v alues, and perturbed conditions . This flight conditions file , together with the input file(s) created by VORVIEW , are used to run the PREDAVOR code.
PREDAVOR makes multiple runs through VORLAX, changing its input I
J
file automatically to reflect the necessary perturbations. PREDAVOR sifts through the rather large output data files and parses out the ~
I
, necessary data.
The stability derivatives are calculated, along with the dimensional derivatives, and the handling qualities parameter CAP.
Options exist to calculate the downwash due to the horizontal tail, and to perform the transformation from wind axes to body axes.
J
PREDAVOR may be used in a manual handling qualities optimization scheme . Geometric changes to the model may be made, and the process to ~h~s po~~: repeat ed. A flag in :he ?REDAVOR input files allows al~ ha~cl~n~ qualiti es data to be concatenated to a slngle =lle , until :he f~ag ~s changed . The CAP graphical i~terface then uses :his informatlon tc crea:e a CAP plot and presents it :0 the user , allowi.g them to ldenti=y handling qualities trends and optimize their aircra=t 0 :he rest.;~::s.
~
I 1
J t ,
=-
" 1 <
J
Figure 3.3
I
VORVI EW Screen Shot Showing Addition of Control Surfaces I;:ou : FLes The goal of project PREDAVOR is to rapidly estimate stability
J
derivatives using g i ven existing tools . Automating as much of the
J
process as possible aids in obtaining this goal. An intuitive, comprehensive input file is therefore logical. The PREDAVOR input file is called aircraft. edit. The format is line-delineated, with one data per line. The value of the variable is the first entry on the line, followed by a brief name and explanation for the variable (the name and explanation serves only to aid the user in the creation of the file).
The unperturbed variables are identified by a 0, as in alphaO and betaO.
Similarly, the perturbed values are followed by l's (alphal and betal ) .
Flight condition data is included, as well as moments of inertia.
Output Files There are two primary output files for PREDAVOR. The first is called STABDATA. The file was designed to be an intuitive snapshot presentation of both input and output data, presented in an easy to understand format. Stability derivatives are presented in matrix form rather than listed.
The second primary output file is a repeat of the data presented in what is called a SAD format. As discussed in Chapter 4, the PREDAVOR project is part of a larger project at Cal Poly called PANGLOSS. This project is comprised of several interactive analysis tools. An attempt by the PANGLOSS team is being made to define a standard aircraft data file, referred to as a SAD file. Each SAD file is comprised of data corresponding to a unique flight condition. The multiple SAD files created by varying flight conditions, for a single aircraft, is called a SAD book. A SAD book, containing an entire envelope of data for a single aircraft, thus lends itself well to table lookup schemes inherent in such tools as flight simulators. The STABDATA file is output automatically, and the SAD file format will be added as soon as a format decision is reached by the PANGLOSS team.
2.9 Ooere:::':,"},:;} ::' .. vironment doth ACSYNT and VORVIEW were designed to operate on Silicon Graphics (SGI ) workstations, optimally running IRIX version 4.0.2 .
. l\CS~t~: ~s wri::ten mostly i;; FORTRAN , while VOR lEW is primarl.ly written 1n ANSI C. Both, however, have graphical interfaces that are compatible witn the SGI's. PREDAVOR , in order to ensure compatibility, was written in ANSI C and runs on the SGI workstations .
It must be noted, however, that the only part of the PREDAVOR process that requires graphical, workstation abilities is the creation Once these steps are of the model and the initial run of VORVI EW .
completed, a user may download the necessary files to any system that is The rest of the process and the ca~able of running compiled C code.
analysis may then be completed on the new system.
PREDAVOR Calculations In addition to editing input files , performing multiple VORLAX runs, and parsing output data, PREDAVOR performs internal calculations to generate the stability derivatives , the dimensional derivatives, and axes transformations.
Stability Derivative Calculation The output of VORLAX contains the total forces and moments upon the analyzed model. These forces and moments are in turn used to calculate the non -d imension al stabil ity deri v at ives of the model at that Usually , st abilit y derivative data, such as flight flight condition .
test data, wind tunnel resul ts , and theoretical computations, are given in non-dimensional stabil it y derivatives . This fac ilitates comparison of aerodynamic characteristics of different aircraft as well as those of the same aircraft at d iff erent flight conditions 121. The stability derivatives generated by PREDAVOR are thus of the non-dimensional form .
An example of a stability derivative calculation is as follows.
Each derivative is non -dimens ionalized as appropriate .
ax
=
ou
Dimensional Derivative Calculation In order to calculate the handling qualities parameter CAP, some dimensional derivatives are needed. Dimensional stability derivatives are used when determining the analytic transfer function of the model.
They directly correspond to the coefficients of the differential equations that describe the dynamics of the model [3J . It is these dynamics that the CAP parameter interprets into a useful metric of aircraft performance.
The dimensional derivatives are calculated according to the definitions shown in Table 3.1. It is assumed Cd = O. The moments of o inertia were provided in the input file aircraft. edit.
Drag Considerations It is important to note that, due to the limitations of the vortex lattice method, only aerodynamic (induced) drag can be estimated.
Therefore, a good estimation of Cd must be obtained using other o methods.
Axes System In order to ensure appropriate comparisons between sets of stability derivatives, it is necessary to look at them in a common axes system. Choice of axes system ' often depends on the method and location of data generation. VORLAX uses a non-conventional axes system, shown in Figure 3.4. Anticipating the average user to be familiar with the more conventional axes system used in aircraft analysis, PREDAVOR internally corrects for the change in systems. Thus both the input and the output of the code are in traditional coordinates. These coordinates correspond to the body axes, and PREDAVOR has been designed to give its output in the body axes. The transformation from wind axes to body axes is given below: Wind Body costa) sin(a) [ cos( a) cos.8) -Sin(a)rCD. ]
[ -m]
+CY. - sin(.8) cos(.8) o +CYw -CLo. sin(a) cos(.8) sine a) sin(.8) costa) -CL- cos(a)sin(a) [cos(a) cos.8) -sin(al C,w] Cm. - sin(JJ) cos(.8) o Cmw
[ Ct']
Cn. sine a) cos(.8) sine a) sin(.8) costa) Cnw Table 3.1 Dimensional Derivative Definitions Longitudinal Dimensional Derivatives - (CD. + 2C )QS DO Xu mu O -(CLu + 2C )QS LO Zu mu O -(C + 2C )QS C La oo
Z =
Z& = - C &-QS I (uom) w w Za 2u mu O o
Z =
UoZ Z& = UOZ& a w a w c -C -QSlm Zq Zq 2u C (QSC) Mu mu I U o y C (QSC) c QSc Mw = C & ---- Mw mu of rna 2u uOl y U Y o
Ma = uoMw Ma = uOM&
w Table 3 . 1, continued Lateral Dimensional Derivatives np QSb C (5. 1) I , Sb 2C QSbC 1) Q -nr (5.
__ yr (ft / s) N Y, r 2I,uo 2muo L , QSb C'r (5·') 2Ixuo ylia Y = QSC (ft I 5 2) lia m N = QSbCna- a- I , L = QSbC'lia L = QSbC'a- a- liasl I I , x Source: Nelson , Robert C. Flight Stabilit y and Automatic Control. New York: McGraw-Hill, 1989. 127, 166.
y Traditional Axes System used in PREDAVOR Non-traditional Axes System used in VORLAX Figure 3 . 4 Axes Systems U sed i he PREDAVOR Project Handling Qualities In order to validate the handling qualities optimization scheme, it was determined that a single baseline metric was necessary. Several metrics were considered, including classic Neal Smith analysis, bandwidth and time delay, Nichols chart analysis, and control anticipation parameter ( CAP ) . The CAP metric was chosen for several reasons. First, a longitudinal metric was thought to be most appr opri ate . Pitch control, both as a primary control axis and as an indirect way of controlling flight path, has long been identified as a vital component of flying qualities [4). The CAP parameter is readily calculated using available information. The only additional information that needs to be supplied that is not inherent to the VORLAX analysis is the moments of inertia of the model. In addition , the CAP parameter can be calculated internally. Several of the other metrics have calculation schemes that require intensive calculations. These calculations could be performed quite readily using existing packages such as Matlab or Program CC. This would, however, necessitate yet another interface between pieces of code. Using an easily calculated metric such as CAP reduces both complexity and computational time. Finally, the CAP metric is very intuitive in nature. An easy to understand metric aids in using PREDAVOR as an educational tool. The primary disadvantage to using the CAP parameter is its non-applicability to unconventional aircraft.
The control anticipation parameter is defined as the ratio of the initial pitching acceleration to steady-state normal acceleration, and can be represented by ooSp 2/ (n/a), as shown in Figure 3 . 5 . Although there are several interpretations of CAP, the one used in PREDAVOR is the maneuvering stability margin interpretation. Because n/a is proportional to CLa, and OO~ 2 to Cma, OO~ 2 /(n/a) can be recognized as being related to static margin , Cma/CLa [5) . It therefore may be calculated as a function of stability derivatives, dimensional derivatives, and moments of inertia: ma
m 2 = -qSc C (C + SCp C )
sp I La C 4 mq y La m rna
msp = -c W (C + gcp C )
CAP (nl a) Iy C 4(W IS) mq La
The CAP parameter is plotted against the damping ratio, S, and the
point plotted on a graph with empirically defined regions for Levell, Level 2, and Level 3 handling qualities boundaries. PREDAVOR uses the CAP plot corresponding to the Category B Flight Phase.
The CAP metric was used as a baseline metric in order to test the handling qualities optimization scheme. Once the proof of concept of the scheme has been verified, other metrics may be considered for future incorporation. The PREDAVOR code was written · to establish the basic framework of the scheme, and additions and enhancements of the code should be encouraged.
M "n/a
M"n/a I
, co", . 5' + 21;co",S + co",' so::
u=
Odb co", co(log scale) -> , co",
CAP
M"n/a n/a Q) 2 ...
Figure 3 .5 Stability Margin Interpretation of CAP Parameter Source: Dept. of Defense. Military Standard Flying Qualities of Piloted Aircraft. MIL-STD-1797A. 30 Jan. 1990. 172-175.
CHAPTER 4
--- - - _._. - -- -- CHAPTER 4 The PANGLOSS Project The PREDAVOR methodology is part of a larger framework called the PANGLOSS Project . The goal of PANGLOSS is to provide students with accurate yet intuitive tools that would allow them to rapidly analyze and understand aircraft stability , control , and handling qualities .
PANGLOSS is an ongoing project at Cal Poly and team members consist mostly of graduate students designing analysis tools to be used at the undergraduate level.
One major branch of PANGLOSS is comprised of three projects that are designed to work together in a seamless methodology . PREDAVOR is an important part of this branch . The framework of this branch is shown in Figure 4 . 1 . In the upper left hand corner a burgeoning aerospace des~gn engineer conceives of an aircraft design . First they models their aircraft and obtain stability derivatives as well as a first cut handling qualities analysis from PREDAVOR .
Next , they can analy ze ' and manipulate this data using the intuitive graphical interface SAVIo Finally, they can input this new data , gained from PREDAVOR and SAVI, into a workstation-based simulator called RADIAN.
In this way, the designer can very rapidly conceive of a des ign , analyze it , and actually fly his design , all in a ma tte r of hours.
Q L _
~ MoOeI AIrcr aft
~ ( "'-- PREDAVOR
Cole S'obliTv Denllolrlles Handi ng QuoI~IElS Figure 4 . 1 PANGLOSS Project Overview I
.J
---- -- ----- .--.- - --~ .. -- - -. - ----
Project PREDAVOR is deal with ex~ens~vely i~ Chapter 3 of this documenL. The followlng secLio~s briefly summar~=e and highlighL Projects SAVI and RADIAN.
SAVI Proj eCL SAVI [lj wa s conceived as a wa y LO give designers direct access to simulator data in an easy to understand =ormat . Most simulators use Lable look - up me~hoas. These tables consist of thousands of daLa poinLs . If the aircraf ~ des · gner wishes ~o analyze or manipulate any of this daLo, .e must stop he sim~ ator , identify the daLa poinLs he wishes to change , an edit the files using a standard text editor.
The data must then be rel oa ded inLo the simulator and the slmulator started. SAVI al low s Lhe designer access ~o the table information in intuitive graphical interfaces . figure 4 . 2 shows the SAVI control window and Figures 4 . 3 and 4 . 4 show a 2- D and 3- D plot .
The 3 - D plot may be rot a t e d for b e tter viewing .
File Edit Plot Ed it 3D !:!.e lp View:
As a function .of J ALPHA1 C I
/ COO
Where : i DECAN 1& - 70
CLO CMO
And as a function of : I DECAN C I
~ DCDlEF where: IALPHA11= 90 DCLCAN DCLLEF Fixing the yalues of : DCLTEL iMACHI & 0 DCLTER DCMCAN Figure 4 . 2 S AV I Co ntrol Win do w _..J Onc e vi ewed , t he regio n s o f d a ta of interes t may t hen be i sola t ed using point a nd c lick me t h od s.
Dat a can be cha nged by c lic k ing a n d
.J
d ragging on a d a t a poin t or by en ter ing a ne w value . Al l data i s accesse~ ai~ect~y into s~mu~aLo~ memory so these changes can be mace while ~~e Slmu aLor is act ive . The aircraft in the slmu~aLor immedia:el y reacts with the new dynamics .
s.;::::::: contai.ns the :0 ~ovli.;)g features : - ~d:t:ng algo~ithms :o~ one or more points in 2 - ~ and 3 - D pars .
- :~pu~ :~om data :ile O~ d~recL In erface wi.~h si.mula~ion me~ory 3ased on generic C a~d X- windows for portaoility - =~~lemented with Mot~f Ibraries for consistent look and feel ~ostscript output for hard copy of plots - ~~ML on - line user ' s manual - ~~rect interface to simulation memory Figure 4.3 SAVI Two - Dimensional Plot Window !
j J
.J
J
/
/
/ r Figure 4.4 SAVI Three - Dimensional Plot Window RADIAN i Project RADIAN "! consists of the development of a workstation- based flight simulator that will have the following features: - Six degrees of freedom simulator.
- Full non-linear equations of motion.
- Workstation-based , flight stick or mouse.
- Performance evaluation consisting of an "up and away" task and a landing task.
- Visual representation of model on screen .
The simulator will use data generated by PREDAVOR and SAVIo RADIAN contains two performance evaluation situations that allows the designer to qualitatively evaluate the aircraft dynamics . The up and away task, shown in Figure 4 . 5, consists of a floating cross with a light on one end. The light changes locations on the cross in a random
J
fa5;:2..0r .. :~e ?~:o: ~s :~ ?o~~: :ne aircra~t nose dlrec~ly The: ?oa: a"C :ne ::.gh"C. The ~~lo: gains 0 scs~e tr.o: correspo~ds :0 his success .
The a lgorl:hm :or :~~s :eature is s: ~ _ _ ~r. progress . The second "Cask ~s "(he landing task, S~8Wr. in Figu r e ~ . 6 . T ne pilot lands the air craf"( and gains a score nasec on , among otner ~arian':'es , the ra~e of descent a: "(ouchdown. T~e pi:2: ~s a~ded by a ver ::.sa : slope indicator in :he form K~en :he po~es are ':'e vel . :he alrcraft is on :ne of " "Ce:eph one: poles" .
flig ht pa:h .
Figure 4.5 RADIAN Up and Away Simulator Task
J
---- - ... . - -- ~--.- -~ , I , 3 0 Figure 4 . 6 RADIAN Landing Simulator Task Future Work At the time of this writing , Projects PREDAVOR and SAVI are completed and working in a stand - alone fashion. The RADIAN simulator is still under construction . Wh e n finished , the three independent codes need to be integrated into a seamless methodology and tested thoroughly , -.:,..: for robustness of method. Other PANGLOSS projects include Matlab-based packages for investigating handling qualities of aircraft , and a PC - based code to take aircraft geometry and determine state space matrices .
J
J
CE.L.PTER 5 Tes~ln~ and Results In order to val id ate the ?REDAVOR methodology , test cases were conducted . For the subsonic case , a Lear Jet Model 23 was used , and for the supersonlc case , the North l-_ erican XB - ~ O Valky rie was selected.
Both models were chosen because stability de::-i vati ve data as well as geometric data was readily avai~able . In aadi tion , a basic handling qualities analysis was conducted .
Sub sonic Case - Lear Jet Model 23 PREDAVOR was applied to a conventional subsonic aircraft , the Lear Jet Model 23 . This T- tail aircra:t features :uselage mounted engines as well as fuel tip tanks . The aircraft model , shown in Figure 5.1, was created using ACSYNT. The aircraft was analyzed at the flight conditions shown in Ta b le 5 . 1.
Figure 5.1 Wireframe Model o f Lear Jet Model 23 I .J The planform model of the aircraft was "sliced " automatically using the VORVI EW interface to create the ana ly sis panels. Vertical
J
panels were created by hand .
The slice mode l i s shown in Figure 5 . 2.
J
- --- -- .. - - Table 5 .1 - Flight Condition s for Lear Jet Model 23 Flight Condition Cru i se Max . Weight Altit:ude (ft) 40,000 Air Density (slugs/ . 000588 Speed ( fps ) 677 (M - O. 7 ) Init:ial At:t i t:ude ( deg ) 2 . 7 Geometry a n d Inertias ( ft- ) Wlng Area 231 . 7 7 Wing Span ( ft ) 34 .1 Wing Ge o . Chord ( ft ) 7 . 03 Weight: (lb s ) 13,000 ft- ) ( slug 28,000 Iy'xt' ft:- ) ( s l ug 18 , 800 I yy c ( slug ft" ) 47,000 I :z e ft-) I ( slug 1 , 300 xz b
I
J Figu r e 5 . 2 - Sliced Rep res ent a tion of Lear Jet Model 23 The model of th e Lear Jet was th e n a nalyze d u s ing 150 wing tip to I wing tip slices and 1500 subpolygons . The re s ulting stability -1 derivatives ar e shown in Table 5 . 2 . T he d e ri vatives were compared to those generated using empirical methods for the same aircraft at the
J
given flight conditions fli . Included in the table are relative
J
~ ---- -- -- ~- --~ - - - -_ .. _ --- . _ _ . _ - - . ~ - ---
lmpor~ance of the derivatives :: ' . The es~imated accuracy using ~he emplrlcal method is given in order to facilitate a comparison .
Table 5 . 2- S ability Derivati v es of Lear Jet Model 2 Longitudinal Stability Derivatives Est . P re d.
Der i vat~v es VO RLAX Emp. Da ta Importance
*
0 . 4100 C 0 . 2594 L 5 . 50 5.84 10 C:. a ±5 % 2 . 98 2 . 20 4 C" a O Ot ±40 % 9 . 93 4 . 70 3 C q L ± 20 % 8 . 37 0 . 40 5 ±20 % C :. " 0 . 0261 0.0335 CD - 0 . 3723 0 . 3000 5 C a c ±10 % - 0 . 0644 0 . 1040 6 C u r ±20 % - 0 . 0247 0 . 00 C~ - 0 . 5701 - 0 . 6400 10 C M • ±1 0 !i; - 4 . 9660 - 6.700 C a do t M ±4 0 % - 16 . 55 - 15.50 9 eM q ±2 0 % - 1 . 7991 0 . 050 8 M ±2 0 % C " Lateral Stability Derivatives VORLAX Es t . Pred .
Derivative s Emp Data Impor tan c e
*
- 0 . 3849 - 0 . 1100 10 el b ±20 % Cl - 0.4818 - 0 .4500 10 P ±15 %
·1
" Cl r 0 .2 2 5 2 0 . 1600 ±40 % Cn b 0 .5999 0.1270 10 ±15 % Cn p - 0 . 0797 - 0.0080 8 ±90 % Cn r - 0 .5475 - 0.2000 9 ±25 % Cy b -2 . 4666 - 0 . 7300 7 ±20 % Cy 0 .1759 0 . 0000 4 P ±50 % r 1 . 356 7 0 . 4000 4 Cy ± 30 % *Rel atlve Importance , 1 0= Ma]or , 5=M l nor , O=Negllglble, Roskam
J
J
The vortex lattice method did a good job predicting the longitudinal derivatives. C and CL (a dot ) were predicted satisfactorily, La as was C , , CM (a d ot) ' and CM C was significantly overpredicted. The Ma q. Lq fo rward perturbation derivatives were not predicted well. The derivatives based on drag are difficult to compare because the vortex lattice method, by nature, only predicts induced drag effects. A good estimation of Co o is needed to accurately predict the drag derivatives.
The lateral derivatives, in general, were predicted well, with beta derivatives consistently predicted high.
Supersonic Case- North American XB-70 In order to validate the code's capability to analyze supersonic configurations, a test case of the North American Valkyrie XB-70 was conducted. This aircraft was a canard delta wing aircraft designed as a strategic bomber in the 1960's. The wireframe model of the XB-70 is shown in Figure 5.3. The sliced model, generated in the same manner as the Lear Jet model, is shown in Figure 5.4, and the test case flight condition is given in Table 5.3.
Figure 5.3 Wire frame Model of the XB-70 , , Figure 5 . 4 Slice Model of the XB-7 0 Tabl e 5 . 3- F light Conditions for the XB - 70 Flight Condition Cruise Max. Weight Alti tude (f1: ) 60 , 000 Air Density ( slugs / . 0002237 (f ps ) Speed 2420 (M = 2.S) In itial Attitude ( deg ) 4 . 4 Geometry and Inertias ( ft- ) W i ng Area 6297 . 8 Wing Span ( ft ) 1 05 Wing Geo. Ch o rd (f t ) 78 . 53 Weight (l bs) 13 , 000 ft-) Ixxb (slug .1 8E7 ft-) (slug I yyb . 10 E8 I .J ft-) (s lug .22l E8 I zzo The stability derivatives for the XB-70 were calculated and are tabulated in Table 5.4. The derivatives for the most part agree with data from various sources , including flight test data (3) • Of the important deriva tiv es, Cl~ and Cn~ are again overpredicted, but still we~l wl~hin tolerable ~ange. This agreement illustrates the vortex In both :;'a~::::'ce c:ode ' s ability "to analyze supersonic configura"tions.
::ne so..:osonic and supersonlc case, i"t was found tha"t "this method is ex"t~e e~y sensi"tive ::0 "the placement of the center of g~avity. Handllng qua:~::es analysis s howed that the XB - 70 is a Level 1 aircraft a t both ::he subsonic and supersonic conditions tested . Optimization studies are In p.::-og.::-ess.
Table 5.4 - Stability Derivatives of the X8 - 70 Longitudinal Stability Derivatives VORLA){ Derivativ es Data** Importance
*
0 . 08 0 . 091 C:.
l. 13 l. 50 C:. a - 4 C:. ;, o ,,~ 0 . 7 0 9 3 C:. g - 1.88 5 L C " - C r - C::, ;, 0 . 0002 6 CD c - C M - 0 . 155 - 0 .14 10 C a M 0.017 0 . 0 7 C a dot M - 0 . 565 - 0 . 4 9 C q M 0.469 8 C U M Lateral Stability Derivatives I ::) Derivatives VORLA){ Data** Importance * I Cl b 0.005 0 . 013 10 I Cl - .065 - 0 . 07 10
J P
Cl r - 0 . 049 - 0 . 015 Cn b 0 . 097 0 . 05 10 Cn p - 0.048 - 0. 075 8 Cn r - 0.089 - 0 . 36 9 I Cy - 0 . 23 - 0 . 36 7 b . ~ Cy 0 . 11 4 P cy r 0 .2 0 4 *Rela~ive Impor ~ance, lO=Maj or , 5=Min or , O=Negligible , Roskam
J
**Source: Heffley , R. K, and W .F. Jewell. "NASA CR 2144 Aircraft Handling Qu al itie s," December 1972.
J
.
. \
! , , OptimizaLion of Wing and Horizontal Tail The opLimization s cheme was applied LO the geometry of the Lear Jet Model 23 by varying horizontal tail location and aspecL ratio .
Results are shown in Figure 5 . 5 . Fir s t , the longitudinal location of ~he horizonLal Lail was changed . Point 5 on the graph locates the aCLual posLion of Lhe horizonLal tail. The tail was then moved forward and aft in 3 foot increments . At its original location , the Lear Jet is a Levell aircraft to Category B tasks . As the tail is moved fore , the
aircraft moves away from the Level 1 space , with both CAP and S
increasing. As the tail approaches the moment center of the aircraft, the handling qualities stay solidly Le vel _.
LEVEL 2 -4 LEVEL 1 CAP 1 -1 -2 . S sec 3 _
J
.2 -2 .1 I
. i
.05 .J x Hor izontal 'rail inqaft IIIOV Aspect Ratio .02 winq, same of are a . 01 .1 .2 2 5 .5 1 D&mpl.ng Ratio, ~
CATEGORY B FLIGHT PHASES
J
Figu re 5. 5
J
CA P Graph fo r Le ar J et Model 23 ..
Nex::, the aspect ratlO was varied, keeping constant wlng area a:lO 2.~lowing the wingspan to change. The points on the CAP graph are :lumbered wlth the value of the aspect ratio. There is no clear ::-elaLionship between varying aspect ratio and the flying qualities of ::he aircra:::. Aspect raLio' s 2,3 , and 5 seem to form an increasing paLh, yeL aspect raLio of 4 is clearly an anomaly, as is aspect raLio 7 .
This type of analysis would be useful when aspect ratio is used as a :::onstraint on the preliminary design .
It would only be necessary to ensure thaL the aspect ratio given provides a Levell aircraft.
In both cases, the analysis was extremely sensitive to center of gravity location, more so than with the stability derivatives. Because of this sensitivity, this tool is recommended for use in identifying trends, rather than to force the optimization to a specific CAP value .
I .J
I
J
CHAPTER 6
--~ -- - .. - -- r--- !
I ' CHAPTER 6 Conclusior.s and RecommendaLloDs A com~rehensive worksLaLion-based tool to facilitate the opL~mizaLion of aircraft for handling qualities was designed and lmplemenLed. PREDAVOR rapidly calculates stabiliLY and dimensional der~vaLlves given a three d~menslonal model of ar. a~rcraft . It then eStimaLes tDe handling qua~it~es metrlC control antlcipaLion paramete~ ( CAP ) and plots iL via a graphical interface on a CAP ploL. In this way it allows Lhe user to rapidly assess aircraft geometry changes and identify trends as they pertain to handling qualities. The aircraft may then be optimized for these qualities.
In general, both the longitudinal and lateral derivatives were The inherent vorl ax lattice method has been shown to be extremely sensitive to center of gravity location , as is the CAP calculation. This sensitivity must be noted b y the user in order to use the tool effectively. The stability derivatives predicted are well within Lolerable ranges for such estimations .
Further research will include the possible implementation of this scheme into an existing optimization and aircraft design package, such as NASA's ACSYNT, in order to allow multidisciplinary optimization, I including handling qualities, of aircraft during the preliminary design
J
stage .
J
J
J
/---
r , ' '
I
I
No-ces
I
Chapter 1 R . Simms , "CE: Engineering a Change in the Design Process , " f Aerospace America , April 1993.
I
A . Mykelbust and p , Gelhausen, "Putting the ACSYNT on I Aircraft Design ," Aerospace America, Sept 1994 .
> McDonnell Douglas Company , The USAF Stability and Control Dlgital DATCOM , Vol 2 , "Implemen ation of DATCOM Methods", AFFDL - TR - 79 -3 032 ( S~. Louis Missouri : April 1979) ~ Dep-c . of Defense , Military Standards, Flying Qualities of Piloted ~ircraft , MIL - STD - 1797A, 30 Jan ., 1990 .
Chapter 2 A . Myklebust and P. Gelhausen.
NASA Ames Research Center and Virginia Polytechnic State University, ACSYNT V2 . 0 Overview and Installation Manual, Jan. 1993.
ACSYNT Manual ACSYNT Manual 5 L . R . Miranda , R . D. Elliott, and W. M. Baker, "NASA CR - 2865 A Generalized Vort ex Lattice Method for Subsonic and Supersonic Flow Applications ," NASll -1 2972 , Dec. 1977, .
o J . D. Anderson, Fundamentals of Aerodynamics, (New York: McGraw - Eill, 1991 ) J. D. Anderson L . R . Miranda et al.
9 VORVIEW , computer s oftware , J. R. Gloudemans and Dave Kinney, NASA Ames Research Center (Moffett Field, CA.)
i
Chapter 3 ·. -l ACSYNT Manual Duane McRuer, Irving Ashkenas, and Dunstan Graham, Aircraft Dynam ic s and Automatic Control ( Princeton, New Jersey: Princeton University Press, 1973) Duane McCruer J. D. Anderson I .1 -.J Dept. of Defense
J
J
------ .. ---- ----_. --- ..
Chapter 4
Notes, continu e d Chapter 4 Todd Whelan, "Development and Implementation of Software for Visualizing and Editing Simulation Input Data", (Master's Thesis, California Polytechnic State University, San Luis Obispo, 1996).
RADIAN Flight Simulator, computer software, Fritz Anderson and Todd Whelan (work in progress, California Polytechnic State University , 1996) Chapter 5 J. Roskam , Airplane Flight Dynamics and Automatic Flight Controls, PartI (University of Kansas: Roskam Aviation and Engineering Corpor atio n, 1979) R. K. Heffley and W. F. Jewell, "NASA CR 2144 Aircraft Handling Qualities," Dec. 1972 3 R. K. Heffley r' .
I ~ . ,