You are on page 1of 8

16-l

Validation of the Siukamow Calibration of
Aircraft Position Error and Airflow Angles
Using a Differential GPS Technique on a Helicopter
Kenneth Hui Stewart Baillie
Flight Research Labotatoty, Institute for Aerospace Research. National Research Council
Ottawa, Ontario. Canada KIA OR6
Tel: (613)~99%96t32 Fax: (613)-952-1704 E-mail: Ken.Hui@NRC.CA
1.0 SUMMARY
This paper describes the validation of a technique for the
simultaneous determination of pit&static position ermr and
the calibration curves for angle of attack and sideslip sensots.
7he SCADS (simultaneous calibration of airdata system)
technique involves flying the aircraft in a “wind box” pattern
while recording a suite of standard flight test parameters and
Differential Global Positioning Sysvm (DGPS) measurement%
This simultaneous calibration technique combines tbc
calibration pmcedtue for both position error and airilow angle
calibration, and eliminates the need for flying close to the
ground during the tests. During the development of this
technique using the NRC Falcon 20 aircraft, the results
demonsrated that accurate calibrations could be obtained with
reduced flight time and cost over conventional calibration
techniques. The present paper describes the application of the
SCADS technique to the NRC Bell 206B helicopter. Tbe
calibration results are presented and arc compared with data
tium other standard calibration methods and verified with
tnanoeuwes not used in the model development Tbe results
from using the SCADS technique have demonsmated better
efficiency and accuracy.
2.0 INTRODUCTION
CAE Elcctmnics. Montreal, and the Flight Reseaxh
Laboratoty (FRL) of the National Research Council Canada
(NRC) have condu$tcd several joint programs to develop Level
D flight simulators These pmgmms have involved the use d
a leased airwaft to gather the flight test data for the simulator
mathematical model development To minimize the cost d
this and other projects. the FRL has concentrated on increasing
the efficiency of the flight tests. One area of emphasis has
Lwen the development of a more efficient method of calibrating
airdata systems.
The current FRL method of calibrating airdata systems is a
two-step process. The tirst step is to obtain the position error
correction (PEC) and flight data by:
Q-stopping at a reference position on a runway to record
baseline position and pressure data;
b)-performing a single, low altitude fly-by down the runway to
provide a single calibrated point of the airdata system; and
c)-performing a ties of fly-bys over a relatively flat surface,
such as a lake, to gather additional “pseudo-tower-fly-by”
data.
T5e PEC coefficients are then determined using the flight data,
as collected above. and the standard atmosphere equations.
Ihe second step is to obtain the calibration curves for the
angles of attack and sideslipi and the FRL flight path
reconsmtction (FPR) technique IS used for this pupose with a
variety of flight test manawns such as bim points and beta
sweeps. This technique has been used successfully on flight
test programs for several types of aircrafr’. and requires a fully
calibrated airspeed system. Even though these methods of
calibration have been proven to be accurate. they are tedious
and complex. The two processes cannot be flom
simultaneously, and the requirement to fly the test points in
series results in additional flight test time and cost l%e total
flight test time teqttired for this calibration is about five hours.
Additional drawbacks of this technique are that the FPR
process requites high accuracy inertial measuremena and
assumes the angle of sideslip to be zem during trim points.
These drawbacks have led to a search for improved methods of
calibrating airspeed and aittlow angles.
Recent advances in the Global Positioning System (GPS), in
terms of accuracy and price, have made it possible to use this
system in the calibration of aircraft airdata systems. Giber
agencies have used GPS for airspeed calibration. Kimberlin
showed in 1992 that the use of GPS for the pita-static system
calibration gave results similar to the two convmtional
methods, namely the “Speed Course” and ‘Tower-Fly-By”
methods. This pm&et was conducted cm the Prithwon
University in-flight simulator (a Ryan Navion aircraft).
FRL has developed a pmcess based on GPS, error modelling
and minimisation techniques to calibmte airdata systems in a
single step. This technique has been named SCADS
(Simultaneous Calibmtiott of Airdata System). Hui.
Srinivasan and Baillie’ showed in 1996 that the SCADS
technique provided accurate results on the NRC Falcon 20
aircraft The &data measurements were obmbwd from B
pmjectdcdicated pita-static system. while nosecone
differential pressure ports provided angle of attack and
sideslip measuntnents.
Fubxe FRL plans include the nquitement for numetws
helicopter flight tests. With the tccognition that the nasecone
diffenntial pressure technique is not applicable to helicopters,
and that airdata tneasutrtncnt errors are mom difftcult to
determine for helicopter than for tixed-wing aircmfs a
validation of the SCADS technique on a helicopter with a
noseboom installation was squired.
This paper reviews the development of the airdata
aemdynamic model algorithms. and the flight test manoeuvres
for calibration using the SCADS technique. and fcatses on the
validation of the SCADS technique for helicopter applications.
In particular, the wr will describe:
1) position error characteristics and calibration methods:
2) flow angle modelling for a nos&oom installation;
3) effects of wind on airdata calibrations;
4) the SCADS technique and the specific details for the
helicopter application;
5) the NRC Bell 206B and its instrumentation and
6) the results of a SCADS flight test on the NRC Bell 206B.
Paper presented at rhe AGARD FVP Symposium on “Advances in Flight Testzng”,
held in Lisbon, Portugal, 23-26 September 1996, and published in CP-593.
3.0 POSITION ERROR CHARACTERISTICS AND
CALIBRATION METHODS
Brown’ has described that the flow field around at aircraft in
flight is distorted and. in turn. the local static pressure usually
differs from that at infinity. The magnitude of this etmr
generally increases with speed in proportion to acmdynsmic
forces and compressibility effects, and is also strongly affected
by the location of the pressurr sensor so”lfc on the aircraft
Gracey’ showed that it is generally possible to design a pitot
tube installation to avoid measurable total pressure etmr for
typical flight conditions. As a result, to produce BCCUIIL~~
ainpeed and altitude data on most airctnft, a calibration is
only needed for the static pressure source position etmr.
References 6 through IO discuss pitot-static calibration
techniques, specific pmccduns and expected accuracies.
Some references also include the analysis of airflow snglc
calibration methods. These methods include the ground speed
course, trailing bomb. pacer method, tower-fly-by method and
Radar techniques. These methods fall into three general
categories, namely (I) the dinct spzed reference methods
which depend on favourable wind conditions, (2) methods
which do not require any special abnosphetic measutemmts,
and (3) radar methods using amnsphetic meastuemcnD or
meteorological analysis for the mference pressum altimetry.
4.0 FLOW ANGLE MODELLING
The vanes on the noseboom measure the angles between the
local velocity vector and the nosebcam axes. These two angles
arc defined in Etkin” as angle of attack and flank angle of
attack. In this paper. the subscript of the noseboom was
dropped for the angles of attack and sideslip. flank angle of
attack and the tnre airspeed vector and i0 components.
Angle of attack: a - tan’l (w/u)
Flank angle of attack pF -tan” (v / ” )
when u. v, w a the three components of the mte airspeed
vector while U is the magnitude of the tme airspeed vector and
Angle of sideslip: B -sin-’ (v / U )
Traditionally. flank angle of attack. BF, is equated to angle d
sideslip. 6”: however, these two quantities ate not exactly the
same. Flank angle of attack is the rotation of the freesttam
velocity vector about the body Z axis. Angle of sideslip is the
rotation of the freestream velocity vector about the stability Z
axis. Comftions for the following mu*t be applied to the
measured angles a,n, BF,~. tn obtain the mte angles of attack
and sideslip: I) nosebcom misalignment, 2) noseboom
bending. 3) upwash and sidewash. 4) transfotmation of ftank
angle of attack to angle of sideslip and 5) aircraft angular rate.
4.1 Noseboom MkaI@ment CorrectIons
The Bell 2068 noseboom was installed with misalignments m
the order of 0. I - 0.3 degnxs horn the aimraf? axis system.
Fortunately. for misalignments of tbis magnihldc. the
significant tetms in the axis tnmsfmmation occur as minor
biases and scale factor errors which may be implicitly in&ded
in the upwasNsidewash comctions to be discussed in an
upcoming section.
4.2 Nasebmm BendIng Corrcctb~ns
During elevated-g matmewing, the tmsetcom will deflect
with the increasing load. For the wind box manoeuvre and
mast flight test manauvres of interest. the g-loading is either
small or short-lived; therefore, the noseboom bending effect
was ignored in this application.
4.3 Effect8 of AhwaR Induced Upwash and Sidewash
This is the aerodynamic effect that the helicopter and/or the
noseboom induces on the local velocity vector. llte vanes on
a noseboom tneasure the effective angle of attack and flank
angle of attack (%. bum), not the free-stream angles of attack
and sideslip (a, b). ‘Ihe difference between these angles is
defined as the upwash or sidewash. To adjust the flow angle
model for this difference let:
a, - a + Aa
BFrn -BF + Al%
For both (1 and B ranging between +/- 20 degrees, Moes and
Ultitmorc” showed that the up- and side- wash comction
terms for a wingtip boom mounted on a fixed-wing aircraft ate:
Act - mar”+ aby.
Al% * ” BF. + IL
where m (upwash factor) and n (sidewash factor) atr
approximately constant at low speeds and less than 1.0.
With this as a basis, the vane calibration equations can bc
tarranged into the form:
a - (1-m) h + abi.
BF - (1-n) BF. + tk.ul
This form of correction will k assumed for the helicopter
noseboom case presented he.
4A Transformation of Flank An@ of Attadc to Angle of
Sideslip
The tmnsfomution of the flank angle of attack to angle of
sideslip is governed by the following equation:
For small angles of attack, less than +/- 10 degrees, the
difference ktwcen b and BP is minimal and, therefore, can be
neglected. For larger angle of attack ranges. however, this
nonlinear term must be included in the corrections.
45 Aircraft An@w Rate Corrections
Since the maeboom is not located at the aircraft cenflc of
gravity, any aircraft angular rate will induce an additional
airspeed at the vane location. The noseboom components of
velocity must be corrected to the centre of gravity (03) of tbc
&craft. These comxtions are presented in detail in Reference
5. As an example, for a centreline-mounted a-vane, this
cotrecfion is given by the equation:
c4-t&t (utma+Qx,)/(u-Qz,) I
where u-Ucosacosp;
cq is the true angle of attack at the CC of tJte aircraft:
xi, li are the distances from the centre of gravity of
the aircraft to the noseboom and
Q is the pitch rate of the a&aft.
5.0 EFFECTS OF WIND ON AIRDATA CALIBRATION
During a flight tescarrh program, phemberger” acquired and
analyzed a set of wind variation data. ‘Ihese data indicated
(Pd ) and hue static pressure (p, ) of the Bell 206B helicopter
are approximated by:
AP-cm + CP,*p* + cpl*p*2
Pd-PL +AP
P,-P, -AP
where
Pa , P,, are the indicated dynamic and static pressures
The second order etmr term is uniquely used in this
application on a Bell 2O6B. The helicopter up- and side-wash
effects, as previously discussed in Section 4.0, ‘Flow Angle
Modelling’, may be implicitly expressed in the vane
calibration equations as:
cl - cm + CN *am
I3 - Cm + CSI *BF.
that wind variations are often larger than airdata calibration
accuracy standards (typically 0.003 Mach). Ehemberger’s
data showed typical variations in wind of 0.075 Knotslmile
and typical RMS values of 6 Knots over a 3.5 hour time
p&cd. His results agreed reasonably well with wind
variability determined from larger data sets produced by the
National Meteorological Center and the USAF Global Weather
CClltfX.
Chat~‘~ did a similar study of a vertical wind. For turbulent
flight. it was possible b have large excursions in vertical wind
data (+/- 6 knots peak-to-peak in two minutes) during a 2O-
min session, with the vertical wind returning to a meart value
(2 knots) after the perturbation. chm also concluded that, for
most of the flight, the atmosphere was smooth. and the vertical
winds generally averaged zero over a long period of time. In
consideration of the above data. the SCADS technique
employs a wind model which can vary linearly with time.
6.0 THE SCADS TECHNIQUE
‘ll~e purpose of the flight test manceuwe in the SCADS
technique is to create an aim-aft time histoty for which the
emxs in the airdata system are independent of wind speed and
direction. The standard mnhod of eliminating the emrelation
between wind speed and pitot-static errors is to fly the aircraft
on nxipmcal headings. This approach does not, however,
adequately discern enum in the tneasurement of sidmlip angle.
In the SCADS technique. therefore, the a&aft is flown in a
wind box pattern (see Figure I). To further produce variation
in angles of attack and sideslip and, therefore, to improve the
information content of the time history. variations in airspeed
and beta sweep manocuvns have ban incorporated in the
various “legs” of the wind box.
4-J
beta sweep
beta swee deceleration
Fire 1 Typical Wind Ilox Pattern
According to the SCADS results from previous flight tests ~1
the NRC Falcon 20’. a wind box with an
acceleration!decelcmtiation manozuwe in the first two legs of the
wind box was ktter than a wind box flow at a single cOnstaX
speed. In the present work, the requinment for a variation in
airspeed over the SCADS manoeuvre was met by accelerating
or decelerating the aircraft either cm straight segmettts 01
during turns. For ail wind boxes, a gradual sideslip change
with constant tmek was executed in both directions (known a8
a beta sweep) on the third and last legs of the wind box. Each
wind box covmd the entire airspeed envelope of UIC
helicopter. All wind boxes were pxfotmed within 25 km d
the GPS nfennce station to maintain the optimum GPS
rneas”rement accuracy.
The general equations for SCADS were ptrsented in
Refetrnce 5. Some of the equations specific to the SCADS
application in helicopters will be listed below. Ihe equations
for the position error correction (APL true dynamic pressure
where the terms C*a and Cm reflect the sum of biases firm the
misalignment of the noseboom plus the up- or side- wash
effects CAI and Cal are sensitivity factors. and %, BFrn are
the geometric or static angle of attack and flank angle of attack
tneasures.
To summarize the SCADS technique. the following parameters
were meawed during the wind box manoeuvre:
-dynamic pnssurc. static pressure. 3 attitudes (6, +, 8).
-total temperahue, 2 noseboom vane deflections.
- 3 angular rates (P, Q. R). 3 GPS Snmd speeds ( VON,
V,.V,o). and GPS altitude (Zo ).
The aircraft ground speed vectm for the manoeuvre is
calculated from the hue airspeed vector and an assumed wind
model. ll~e Direct Search Complex Algorithm n~ethod’~”
varies the cafficiene of I) the position error comction
equations, 2) the airflow angle model and 3) the wind model
in order to minimile the weighted sum of the exmta between
the GPS-measured ground speed vector and the calculated
ground speed vector. The minimisation also considers Uw
diffewtce between calculated pressure altitude and GPS-
derived altitude.
7.0 AIRCRAFf DESCRIFIION
The validation of the SCADS technique was performed using
the NRC Bell 2068 JetRanger (see Figure 2). This aircraft is
a single engine utility-type helicopter. The main mtor is a
two-bladcd. semi-rigid see-saw type. ?he main rotor blades
are of all-metal consm~ction of ahtminum alloy monowque
type. The diameter of the main rotor is 33 ft 4.0 in and the
length of the chord is I3 inches.
Figure 2 The NRC Bell 206 HeIiipter
8.0 INSTRUMENTATION SYSTEM 83 Global Positic~ning System
For the SCADS validation pmjecS a Litton LTN-90-IOU
Inertial Refenncc System (IRS) and the FRL high-accuracy
portable insrmmentation system and a data acquisition systi
(Micmpak) were installed in the Bell 2068. This system
included:
I ) sensor for measuring the aircraft’s inertial
pWd”Eters,
2) a noseboom airdata system, with putentiometer
mcasumnents for the airflow angle vanes. and
3) a PC-mounted GPS receiver.
A NovAtel3151 GPS PC card receiver was installed in the test
aircraft with a model 51 I antenna mounted atop the fuselage
behind the main rotor of the helicopter. The unit was a single
fiequmcy (Ll at 1575.42 MHz) receiver card mounted in the
backplane of a 486amputer system. The Differential mode
of the GPS was used to obtain accurate tt~easutes of aircraft
position and gmtmd sped. The GPS reference station
included a NovAtel 501 antenna with a choke-ring mounted
atop the FRL hangar and a 3151 GPS PC card receiver. GPS
data and status information could be displayed and conboIled
on either PC system by a NovAtel program called Winsat
The time cc&d parameters were worded onboard the
helicopter at 64 Hz.
Because the Litton IRS was designed for the navigation mlc on
commercial aircraft, its parameters. especially normal
accelerations, are heavily low pass filtered. For this t’easo”.
the measures of aircraft acceleration and angular rate were
taken from the thetmally-modelled accelerometers and rate
sensot~ of the Micropak while attitude measures were
obtained fmm the IRS.
a.2 Airdatasystem
Prior to this experiment, a swivelling pitot-static noseboom
system, * Space Age Conuol, self-aligning airdata pmbe, was
installed on due NRC Bell 206B helicopter for project purposes
(Figure 3). The swivelling probe was separate from the
pilot&-pilot systems and consisted of a combined pitot-static
tube with four tins attached to dx end of (he tube to allow for
aerodynamic alignment of the pmbe with the local flow. There
were six static orifices placed at dOdegree intervals around
the tube. The swivelling prOa was designed to align itself
with the local flow to effectively eliminate total snd static
pressure losses as a result of local angles of attack and sideslip
effects. The nose pitot section was mounted to permit a 21
degtw swivel in any direction. Manufactuter’s data suggests
that the vines can maintain their accuracy up to +/- 40 degrees
of angles of attack snd sideslip. The static and diffenntial
pressure transducers wcn PamScientific “intelligent”
!nmduccn which are thermally modelled, and accumte to
within 0.01 percent A Rosemount total tetnpemture pt’cbe
was mounted vertically behind the sideslip vane and below the
boom.
The GPS data post-pnxessing software used ambiguity
resolution and cycle slip handling techniques”~‘p” developed
by GmNav System for single and dual fiequeney systems.
For this experiment, with single-fxquency observations. this
GPS post-processing software. called P-RTK. was expxted to
generate its kst accuracy for baselines (distance betw&n the
airraft and the gmtmd station) of 25 Irm or less. The
combination of P-RTK and the NovAtel 3151 compatible
receiver was expxted to pmduce position accuracies of 2 cm +
2ppm (i.e.. 2 mm error for way k”t) for fixed-wing aint&
when at least four satellites wee observed. llte dme-ccded
GPS data. pmcessed with the P-RTK sofhvare. provided
measures of longitude. latitude, height and the dmx
compncne of velocity with tespect t” the ground. The WGS-
84 convention was used to convert the longitude. latitude and
height into local Cattesian coordinates.
The use of GPS in the helicopter envimttme”t is not as
advanced as in fixed-wing applications. Gne pmblem for
helicopten is that the rotm blades mommmtiiy blank oiT the
satellites. This blocking effect lowers the signal-to-noise level
of the satellite measutements. With low s&al-toa0i.w level.
the ocfumllfe of what are knovm as cycle-slips becomes man
prevalent These complications made the professing of
helicopter GPS measurements tno~c difficult.
At the begirming of the SCADS validation flights, the NovAt.
GPS receiver was configutrd to track ephemeris and the
satellite pseudoranges with options that wete validated in
previous fixed-wing studies. This setup tesulted in GPS data
which was unusable. Newer options for the NovAtel
contigumion were bied and much improved data resulted.
Attention was also paid fo the manner in which the helicopter
was flown. High quality GPS dam was fdly achieved by
using the newer contigtmttion options and by flyins the
helicopter with the following limimdwts@madutes:
. Position the helicopter at a locatio” where no obstacles
block the signal from the satellites
.
Set ICKY% Nr (rotor RPM)
. Start Winsat snd wait until at least 4 satellites are tracked
before starting the data log
. Collect 5 minutes of GPS data prior to moving dw
helicopter
. Climb to the test altitude gradually
. Limit turns to less than 40 deg bank and limit roll/yaw
angular rates to less dta” 30 deg/scc
. Return to the starting lax&ion and. with Nr -100%.
tecord 3 tnon minutes of GPS data
All GPS data used in this paper were collected in the above
FIgureJ TheBell2o6Nosehoom
manner and resulted in a” i%ger ambiguity solution with M
special user interaction.
9.0 SCADS FLIGHT TEST ON THE BELL 2068
The NRC Bell 206B SCADS flight test comprised 3.7 flight
hours in 3 flights over a 3 day period. The objectives of this
flight test were to optimise the SCADS manoeuvre for the
helicopter application. to calibrate the Bell 2068 airdata
system using the SCADS technique and to assess the accuracy
of that calibration by collecting and analyzing time histories of
other ‘ypes of aircraft manoeuvres. The following sections
will summaize the results and discuss them in relation to
these objectives. Wind conditions for the three flights are
described in Table I.
9.1 Summary of Results
The domwash fivm the main IO!M clears the nos&com-vanes
when the helicopter exceeds a forward spetd of 20 Knots.
Therefore. most of the SCADS wind box manoeuvres were
performed within an IAS range of 40 to I IO Knots. The wind
boxes were performed at altitudes of either loo0 ft or 5000 ft
and in one of three wind conditions; calm. strong and steady,
or moderate and mrbulent. The SCADSdetived airdata
calibration coefficients .w presented in Table I. The File
designator indicates the wind conditions (i.e., S - Steady
strong wind of IS Knots, U - Moderate but turbulent wind of 7
Knots and C - Calm wind of I.5 Knots). The Type designatot
indicates how the variation in airspeed was accomplished: L -
Acceleration/deceleration within the first two legs of the wind
box. T-the acceleration/deceleration within the turns and T2 -
similar ta T except the last leg has a different constant speed.
All airdata calibration coefftcients atr telatively consistent in
value. The values of the airdata calibration cccff~cientr have
the best agnement for calm wind cases. ‘llte predicted wind
magnitudes for all ca?es agree ta within l-2 knots. lbe
calculated wind direction is also relatively consistent. These
wind magnitudes and directions also agree well with the actual
and predicted weather measures over the Ottawa/McDonald
Cattier International airport.
9.2 Comparison of Results tim Dtfferent SCADS
hfslnoeuvres
Inspecting the values presented in Table I, with consideration
of the acceleration/deceleration type. leads to the conclusion
that the T2 manceuvre type is slightly superior. Unfortunately.
T2 manoeuvtes were only flown in calm conditions; themfore.
the comparison may be misleading. Fmm a qualitative
considetation of the more uniform distribution of angle of
attack achieved during TZ manoeuwes it is felt however, that
the T2 manocuvrc is slightly superior u, others.
9.3 Bell Xl6 Airdata Calibration
The air&ta calibration coefticieno for the Bell 206B are
summarized in Table 2. ll~ese coefficients were sorted by
meteorological wind conditions. The overall values reflect an
average of airdata calibration coefficients which wcn derived
hum all the available casts. The range of deviations from the
average value for the calm air condition group were
significantly lower than for the other two groups. Under these
conditions, the wind is more repeatable and the airdata
coefficients are more robust. Based upon the consistency of
the calm air coeffXenb and the wind data, on a case by case
basis. the averaged calm air caff~ciena were chosen as the
final NRC Bell 206B calibration. Note that this choice
represents only 0.6 hours of flight data, including aircraft
transit time.
Designator Position Error Cortrction a
B
Resulting Wind Model
I
File LAS
Type CW CPI CPl CA0 CA, CEO CBI % we0 W&IO
Kt PSI /psi fr/sec f&c f&c
so1 40-110 L&x/da 0.011 -0.067 0.637 0.403 0.887 -.559 0.964 24.93 -.79 I.41
Table 1 Results from the Eleven Wind Box Manoeuvres
16-6
Table 2 Summary of Airdata Calibration Coeffkients
9d Typical SCADS Identitication Results
Figure 4 shows the typical error behveen the GPS-measured
gmmd speed/altitude time histories and those calculated from
a SCADS estimation. in this case run COZ. The significant
features of this plot are:
I) groundspeed and altihlde errors atr small during
constant heading sections;
2) these mm are large during turns (turn data is not
used in the SCADS minimisation routine).
9.5 Verfflcation of Bell 206 PRC Calibration with
Standard Calibration Techniques
The Bell 206B was flown up and down the runway in calm air
conditions at diffetent speeds using a ground speed cour8.c
@SC) technique. The runway was %5l ft long. The PEC
coefficients derived tkm this method and those fmm the
SCADS estimation arc listed in Table 3. The large difference
between the estimated PECs at the lowest airspeed is
attributable to a decease in accuracy in the GSC method at
lower speeds. In such cases, a small variation in wind will
cause a large percentage variation in the PEC estimate.
f
0 ,oo
Figure 4 Typical SCADS IdentlIlcatIon Restdts
A&proximate IAS Avg Indicated Pd GSC _ PEC
psi PSI
40 0.02324 0.00545
60 0.05310 O.oQ895
80 0.09240 0.01428
100 0. I6670 1 0.02063
Table 3 PIE Coeftlcient Comparison SCADS Vs GSC
SCADS - PEC (SCADS-PEC) I P,
psi txmnt difference
0.00934 16.7
0.00962 1.26
0.01200 -2.47
0.02253 I.14
CaSe
Calm
Leg
Wind Magnitude Wind Direction North Wind East Wind Aircraft Heading
SCg”lC”t ft/s deg ftls ft/s deg
I I 1 2.41 I -34.3 1 1.16 I -0.79 1 341.3
3 2.92 22.1
4 3.82 30.6
Moderate I 10.65 271.9
2 10.78 289.0
3 12.63 293.4
4 13.32 290.5
Table 4 Two Cases of SCADS Results for PEC Verification
2.34 0.95 155.9
2.57 1.52 246.7
0.3 -10.5 337.0
3.5 -10.0 70.3
4.6 -11.5 160.7
4.8 -11.9 250.5
9.6 Verification with Ftight Data
Verification of the SCADS-produced Bell 2OfiB airdata
calibration was performed by using flight ncords which were
not used in the development of the PEC. Two cases were
chosen: a constant airspeed wind box flown in calm wind
conditions and a wind box with airspeed changes during the
turns flovm in moderate and turbulent wind conditions. For
each of the manoeuvns, the SCADS-derived calibrations wen
applied to the airdata measurements. and the wind WBS
determined from the difference between the
measured/calibraed airspeed vector and the GPS-meawed
groundspeed vector. The mean value of the horizontal wind
vector for each wind box leg gives an estimate of the aceumcy
of the airspeed. Table 4 shows these mean values.
Based on an inspection of Table 4 and a comparison of mean
wind magnitudes and directions for reciprocal heading legs,
the WOM case airspeed error (assuming the acbull wind was
consmnt) is 1.27 ftlscc (0.75 Knots), derived from the
moderate wind case. in the second and fourth legs. This gives
a conservative estimate of dx PEC accuracy.
A second verification can be performed by inspecting the
calculated wind for conelation with variations in angle of
attack and sideslip. Figure 5 shows the wind components
calculated during the Beta sweep of the calm air wind box.
Figure 5 shows a minimal conection in calculated wind with
@. Attributing all East wind variation (approximately
equivalent to body axis lateral wind) to ermm in the p
calibration rw.ults in a worst case p error of 0.21’ over the
range of +/- 13.5’. A similar look at vertical wind variation
correlated to angle of attack change produced a maximum
angle of attack error of 0.3 1’ over a range of +/- 3.54
10.0 CONCLUSIONS
A simultaneous calibration of the airdata system (SCADS)
technique was validated using the NRC Bell 206B helicopter.
Flight test results from the SCADS technique were compared
to other standard calibration methods and matched well. The
following specific conclusions can be reached:
. Calm weather conditions .we preferable for the calibration
of a helicopter airdata system.
. The SCADS technique provides better efficiency and
accuracy than pnvious FRI. methods.
. The SCADS method provided an airspeed calibration to
within 0.75 Knots. and abilow angle calibrations to
within 0.31’ for a and 0.21Ofor 8.
The aircraft insbumentation included a swivelling pitot-static
noseboom with airtlow vanes. a NovAtel DGPS and inertial
sensors. The aircraft was flown in a wind box pattern with
gradual beta sweeps and in an acceleration/deceleration
fashion. The initial airdata aerodynamic model was devised
from the empirical formulation based on the position error
characteristics. mx&amn airtlow modelling, and atmospheric
analysis for an airdata calibration. The SCADS technique
allowed a simple and accurate calibration of aircraft PEC and
&low angles. Ithis technique was validated by modelling the
Bell 2068 helicopter airdata system in the IAS range of 40 to
I10 Knots.
11.0 ACKNOWLEDGMENTS
We wish to acknowledge the contributions of Stephan
Carignan. pilot of the NRC Bell 206B helicopter. who
@onned and developed the flight test procedures which
ensured an integer ambiguity solution for tie GPS.
16-8
12.0 REFERENCES
‘Baillie, S. W.; Hui. K.: de Leeaw. J.: ‘lix Flight Test and
Data Analysis Program for the Development of a Boein@de
Havilland Dash 8 Simulator Model. AGARD CP 519, Paper
30, May 1992.
Hui. K.; Baillie S. W.; de Leeuw, J. H.: ‘Ihe Application af
Parameter Estimation to Identify the Aerodynamic Derivatives
of the Dash 8 Series 100. CASI FTS, Ottawa, Ontario, May
1992.
%inivasan, R.; Hui. K.; Baillie, S. W.: Airflow Angle
Measurement Using a Multi-hole Radome Differwttial
Pressure Technique. CASI Journal, Vol 41. No. 4, December
1995.
4Kimberlin. R. D.; Sims, J. P.: Airspeed Calibration Using
GPS. University of Tennessee, Space Institute, Tullahoma,
TN, AIAA 92-4090. August 2426.1992.
‘Hui. K.: Srinivasan, R.; Baillie. S. W.: Simultaneous
Calibration of Aircraft Position Error and Airflow Angles
Using Diffemntial GPS. CASI FJX, Gttawa. Ontario. Canada
April 1996 (submitted for CASI Journal publication.
lkcember 1996).
%-own, E. N.: Position Error Calibration of a Pressure Survey
Ailcraft Using a Trailing Cone. National Center for
Abnospheric Rexarch, Technical Note NCARAN-313+STR.
Boulder, CO., July. 1988.
‘Gmcey. W.: Measunment of Aircraft Speed and Altitude.
NASARP-1046.1980.
%eAnda. A. G.: m Standard Airspeed Calibration
Pmeduns, AFFTC-TIH-81-5, Edwards Air Face Base. CA,
June 1981.
%awfwd. I. A.; Nippress. K. R.: Calibration of Airdata
System and Flow Direction Sensor. AGARD-AG-300. Vol. 1.
Sept 1983.
‘%aming, E. A. Jr.: Airdata Calibration of a High-
Petfomxmce Aircraft for Measuring Atmospheric Wind
Profile. NASA TM-101714.1990.
“Etkin. B.: Dynamics of Flight Stability and Contml. Second
edition, John Wiley and Son. 1982.
‘Richardson. N. R.; Pearson. A. 0.: Wii Tunnel Calibration
of a Combined Pitot-Static Tuba. Vane-Tuk Flow-Dim&on
Transmitter. and Stagnation-Temperature Element at Mach
Numbers fmm 0.60 to 2.87. NASA TN D-122. 19.59.
“Mow, T. R, Whibnot’e. S. A.: Pmliminary L&t at
Techniques Used to Obtain Aitdata from Flight at High Angles
of Attack. NASATM-101729, December 19%.
‘%hmtberger. L. 1.; Haeting. E. A. Jr.; Lockhan M.G.; Teets,
E. H.: Atmospheric Analysis for Airdata Calibration on
Research Aircraft. AJAA 920293, NV., January 6-9.1992.
“Ghan. K. R.; Bowen, S. W.; Scott, S. G.; Bui. T. P.: Tbe
NASA ER-2 Mcaomlogical Measutrment System:
Instmmentation, Calibration and Intercomparison Results.
Appendix B.6 J. Geophys. Res.. 94.11573-l 1587.1989.
‘6Anon: Internal Mathematical Software Library (IMSL).
Vemion I .O, April 1987.
“Gill. P.E.: Murray, W.: Wrighf M.H.: Practical
Gptimiaation. Academic Press. New York, 1981.
“Chen. D.: FAST Ambiguity Search Filter (FASF): A Novel
Concept for GPS Ambiguities Resolution. Pmt. Gf GPS’93,
The Institute of Navigation, Alexandria, VA, pp. 781-787.
1993.
‘%en, D.; Lachapalle. G.: The Comparative Analysis of
FASF and the Least-Squaw Ambiguity Resolution Technique.
Pmt. Of KfS’94. Dept of Geomatics Engg.. Calgasy. AB..
1994.
‘OChen, D.: Petfomuncc Outline of P-RTK. Internal
documentation of GeoNav Systems. Calgary. Alberta. Canada.
1994.
“Croll. J.B.; Lochhead. H. M.; Liu, J.: Flight Validation of the
Position Accuracies of a NovAtel RT-20 Differential GPS
Installed in a Falcon 20 Aircraft Canadian National Research
Council, LTR-FR-130. May 1996.