Professional Documents
Culture Documents
Analytical Assessment of The Spin Susceptibility of The NASA Generic Transport Model
Analytical Assessment of The Spin Susceptibility of The NASA Generic Transport Model
Aditya A. Paranjape∗
University of Illinois at Urbana-Champaign, Urbana, IL 61801, United States
Stephen J. Gill†
University of Bristol, Bristol, England, BS8 1TR, United Kingdom
Narayan Ananthkrishnan‡
B.257, IIT Bombay, Powai, Mumbai 400 076, India
§
Mark H. Lowenberg
University of Bristol, Bristol, England, BS8 1TR, United Kingdom
A recently derived analytical criterion for predicting the spin susceptibility of aircraft
is applied to the Generic Transport Model (GTM). The criterion is unique among spin
prediction criteria in that it involves computing the zeros of a single nonlinear expression
which is a function of the aerodynamic moments acting on the aircraft and their derivatives.
The results obtained from the criterion are compared with traditional spin-susceptibility
criteria and with results obtained from bifurcation analysis, and a good correlation is found
for spin susceptibility in two angle-of-attack regions.
I. Introduction
Aircraft spin is one of the most dangerous phenomena encountered in flight. Although spins have been
recorded for nearly a century (since 1916 to be precise), it is only in the last thirty years or so that it has
found a rigorous explanation within the ambit of our formal knowledge of flight mechanics.
The first analytical criterion to predict spin susceptibility of aircraft came from Weissman in 1975,1
a probabilistic criterion based on Cnβ ,dyn and the so-called lateral control departure parameter (LCDP).
Bihrle and Barnhart2 correctly identified that equilibrium spin could be modelled accurately using only
Euler’s equations for rotational dynamics. Although their model did not yield an analytically tractable
criterion and was restricted to equilibrium spins, they correctly observed that, for a given control surface
deflection, spin equilibria (if any) coexist with benign trim conditions.
The traditional view of spin as an instability (along the lines of Dutch roll, wing rock, etc.) was refuted
by analyses of complete nonlinear flight dynamic models which became possible after the introduction of
bifurcation and continuation methods. It became clear that spins were primarily high-α steady states with
large angular rates. The spin solutions could themselves be stable or unstable. Entry into spin was typically
found to occur by way of a jump phenomenon due to onset of an instability on a coexisting low-α branch
of steady states.4 The nature of the spin (steady or oscillatory, erect or inverted, flat or steep, left- or
right-hand) could be predicted from a bifurcation analysis and strategies to recover from spin could be
deduced.6
Thus, traditional spin onset criteria may not always provide a good indicator to the spin susceptibility
of a given configuration. Instead, it makes sense to evaluate the spin susceptibility of an aircraft by looking
for the existence of high-α, high-angular-rate steady states regardless of their stability. The presence of
unstable spin equilibria merely implies that the spin dynamics is more complicated, perhaps a limit cycle or
∗ Post-DoctoralResearch Associate, Department of Aerospace Engineering and AIAA Member
† Ph.D. Student, Department of Aerospace Engineering/Department of Engineering Mathematics
‡ Independent Consultant and AIAA Associate Fellow
§ Reader, Department of Aerospace Engineering and AIAA Senior Member
1 of 9
2 of 9
Note that p = r cot α is obtained as a result of assuming that the flight path angle γ ≈ −90 deg.
Equation (5) locates all SNBs in the α − r parameter space. Of all SNBs in the α − r parameter space,
we wish to isolate the ones with the special property of being an incipient spin solution. An elegant way to
do this is to obtain the accessible region plot, which gives all possible equilibrium states at which the aircraft
may be trimmed (balance of moments about all three axes) subject to availability of control. As suggested
in Ref. [4], the accessible region may be computed by solving the trim moment equations in the form below:
qSc Iz − Ix 2
Cl = 0; Cm = − r cot α; Cn = 0 (6)
Iy Iy
A. Numerical Computations
The accessible region (AR) is composed of all those trims for which the control deflections are within
permissible limits. The AR is computed by calculating the control deflections required to trim across a grid
of α and p values, with the aforementioned assumption that r = p tan α. The aerodynamic data for GTM
is provided in a tabular forma , and linear interpolation is used to compute the coefficients at intermediate
values of α, p and r for the purposes of computing the accessible region.
In order to compute the derivatives of the aerodynamic coefficients, we need to compute partial derivatives
of the aerodynamic coefficients. In order to compute the partial derivatives of aerodynamic coefficients, we
use the following two-step approach. In the first step, we prepare an α - p grid which is refined to the desired
accuracy (along with the condition r = p tan α) and obtain the aerodynamic coefficients at the grid points
from the tabular data using linear interpolation. Next, we use finite difference over α and p, together with
a smooth fit over r, to obtain the partial derivatives.
For example, consider Cl and suppose that we wish to compute Clp . Let the α − p grid have δa and
δp as the spacing between the grid points along α and p axes, respectively. We seek to compute its partial
derivative, Clp at a point (α, p, r). Note that the projection of this point on the α − p space coincides
with the grid point (α, p, 0). Let (α, p, r1 ) and (α, p, r2 ) denote the nearest grid points, so that r1 < r < r2
(without loss of generality). Then, we write
where
Cl (α, p + δp , ri ) − Cl (α, p − δp , ri ) r − r1
Clp ,i = and f =
2δp r2 − r1
A similar approach is used to compute other first and second order partial derivatives.
a GTM data to which the public can request access is designated as Design Sim v0912
3 of 9
4 of 9
30
25
20
-3 -2 -1 0 1 2 3
Roll Rate (rad/s)
(a) Finite difference and accessible set with δa , δr ≈ 0 and throttle 0.22
40
35
[deg]
30
25
20
-4 -3 -2 -1 0 1 2 3 4
Roll rate [deg]
p [rad/s]
(b) Polynomial approximation and the complete accessible set for throt-
tle held at 0.22
Figure 1. Plot showing the accessible region (blue) and the saddle node bifurcations (red) in the α - p parameter
space. Black ellipses are used to identify accessible incipient spin solutions (SNBs).
and 4(d) which show the magnified regions around the saddle nodes. The first saddle node marks the location
where the primary equilibrium branch turns first turns around, before leading to the second SNB where it
turns around again to create an (unstable) equilibrium spin branch. The SNB marked in Figure 4(c) is seen
in left as well as right “turning” flight. On the other hand, the second pair of SNBs corresponding to left
and right spins, marked in Fig. 4(d), are not coincident in the α − δe projection. The sense of the turn is
clear from the roll rate in Fig. 4(b), as the roll rate has the same sign as the turn rate.
The SNBs from Fig. 1 correlate well with two SNBs in Fig. 4: the first around α = 27 deg in Fig. 4(c)
5 of 9
and α = 36 deg in Fig. 4(d). The reader is reminded that the time histories in Fig. 3 and the spin solutions
in Fig. 4 represent fully developed spins and therefore, it is natural to expect a mismatch between the
trim angular speeds computed as part of the criterion and those obtained through actual simulation of a
6-dof model. The important point, though, is that incipient spin solutions are identified accurately by the
criterion - this enables the criterion to identify spin boundaries correctly although spin behaviour away from
6 of 9
200
35 Oscillatory spin
150
30 Oscillatory spins
100
25 Steep spiral
50
α (deg)
p (deg/s)
20 0
Steep spirals
−50
15 Steep spiral
−100
10
−150
5
−200
Oscillatory spin
0 −250
−80 −60 −40 −20 0 −30 −25 −20 −15 −10 −5 0 5 10 15
δe (deg) δe (deg)
(a) α vs δe (b) p vs δe
40 37
36
36.5
32
α (deg)
α (deg)
36
28
35.5
24
35
−54 −52 −50 −48 −46 −44 −42 0 1 2 3 4 5 6 7 8 9
δe (deg) δe (deg)
Figure 4. Bifurcation diagrams showing the angle of attack α and the roll rate p as functions of the elevator
deflection. Solid blue lines denote stable equilibria, while red dotted lines denote unstable equilibria. Green
and yellow lines denote the maximum and minimum values of the corresponding state on stable and unstable
limit cycles, respectively. Stars denote Hopf bifurcations, while empty circles (◦) denote torus bifurcations.
Saddle nodes marked with arrows are correlated to SNBs in Fig. 1.
the boundaries need not be predicted precisely. But then, spin behaviour away from the spin boundary is
not germane to predicting spin susceptibility.
Figures 4(a) and 4(b) show an isolated limit cycle branch between δe = −15 deg and δe = −30 deg.
Figure 4(b) reveals these isolated sets to be left spins. These two sets appear in the form of closed loops in
the bifurcation diagrams, and their origin is not altogether apparent. One may conjecture that these limit
cycles arise from the bifurcation equilibrium branches which occur for a different combination of control
inputs than that for which the bifurcation diagrams are plotted, in which case it may be possible to correlate
them with SNBs from Fig. 1.
It is of interest, for the sake of completeness, to compare the aforementioned results with traditional spin
prediction criteria. Two traditional spin criteria are shown in Fig. 5: the revised Weissman criterion and the
Kalviste criterion, and the legend is been explained in Table 1.
The results in Fig. 5 can be summarized as follows. The aircraft has a moderate to strong tendency
for departure and spin in the range α = 18 deg to 31 deg. In the range from α = 33 − 41 deg, it has
7 of 9
(deg)
F =40 deg
LCDP
0 15
−1 10
E
−2
D = 46 deg 5
−3
C 0
−4 B
−5
−6 −4 −2 0 2 4 6 8 10 12 −40 −30 −20 −10 0 10 20 30 40
Cn Dynamic −3 (deg)
x 10
Figure 5. Revised Weissman criterion and Kalviste’s criterion applied to the GTM.
a weak tendency to spin. In Kalviste’s criterion, these regions are marked by combination of lateral and
longitudinal divergence. These results can be explained in terms of Fig. 1 as follows. The first set of SNBs
are encountered in Fig. 1 in the region α = 20 deg to 30 deg, while the second set of SNBs are encountered
in the high-α range of α = 35 − 38 deg. The first set of SNBs typically correspond to a departure from
“nominal” equilibrium flight. The mechanisms are typically lateral-directional, and primarily yaw departure.
The second set of SNBs give rise to spin equilibria which could be stable or unstable, yielding equilibrium
spin and oscillatory or more complex spins, respectively. Notice that the second set of SNBs in Fig. 1 coincide
with the region demarcated by yellow lines (i.e., corresponding to longitudinal divergence) in the Kalviste
criterion plot shown in Fig. 5.
One interesting difference between the SNB-based criterion presented in this paper and the traditional
criteria is the dependence of the latter on β and its aerodynamic derivatives. Despite this fundamental
difference, the spin susceptibility assessments match. The reason for the concurrent assessments remains an
open problem.
8 of 9
Acknowledgements
The authors gratefully acknowledge the support of Dr. Austin Murch, Mr. Bruce Owens and others
members of the NASA’s Langley Flight Dynamics Branch and Dynamics Systems and Control Branch for
provision of the GTM model and advice on its use and flight characteristics. Dr. Murch kindly allowed us
to liberally use material from his Masters thesis which presented a series of spin test results for the GTM.
The first author would like to acknowledge his supervisor at UIUC, Prof. Soon-Jo Chung, for his support
and mentorship. The second author is supported by an Engineering and Physical Sciences Research Council
(EPSRC) Award grant in collaboration with Airbus.
References
1 Weissman, R., “Status of Design Criteria for Predicting Departure Characteristics and Spin Susceptibility,” Journal of
Computation of Attainable Equilibrium Sets,” Journal of Guidance, Control and Dynamics, Vol. 31, No. 2, 2008, pp. 329-339.
5 Fremaux, C.M., “Spin Tunnel Investigation of a 1/28-Scale Model of the NASA F-18 High Alpha Research Vehicle
(HARV) With and Without Vertical Tails,” NASA CR-201687, Apr. 1997.
6 Raghavendra, P. K., Sahai, T., Kumar, P. A., Chauhan, M., and Ananthkrishnan, N., “Aircraft Spin Recovery, with and
without Thrust Vectoring, using Nonlinear Dynamic Inversion,” Journal of Aircraft, Vol. 42, No. 6, 2005, pp. 1492-1503.
7 Foster, J. V. et al, “Dynamics Modeling and Simulation of Large Transport Airplanes in Upset Conditions,” AIAA
Guidance, Navigation and Control Conference and Exhibit, 2005, AIAA 2005-5933.
8 Murch, A. M., and Foster, J. V., “Recent NASA Research on Aerodynamic Modeling of Post Stall and Spin Dynamics
of Large Transport Airplanes,” 45th Aerospace Sciences Meeting and Exhibit, 2007, AIAA 2007-463.
9 Murch, A. M., Aerodynamic Modeling of Post-Stall and Spin Dynamics of Large Transport Airplanes, M.S Thesis,
9 of 9