2. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Results and Discussion
Potential for Real-time Applications
Real-
Conclusions
Backup Slides
2
3. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Results and Discussion
Potential for Real-time Applications
Real-
Conclusions
Backup Slides
3
4. Demanding Operational Conditions/ Factors
– Main Groups
INTELONICS LTD.
1 2 3 FC 4
to, p, ρ
in-
in-flight icing of lift/ control heavy rain, non-
non-standard water-/ice-/snow-
water-/ice-/snow-covered
surfaces tropical shower atmospheric conditions (slippery) runway
H
5 6 7, 8
Wxg, yg,
Wxg, Wyg, Wzg
terrain/ traffic/other human pilot error cross-/tail-
cross-/tail-wind, wind-shear, ‘microburst’,
wind-
external threat or inattention atmospheric/wake turbulence
T1
9 10, 11 12 E1 t =120s E90
T5
start...
P1
IAS=250 E45
E46
? P6
nz < 0 .7
P3 …
flight control automation engine malfunction; flight deviation from standard flight
logic/data error or imperfection control mechanical failure scenario
Normally, single operational factors are not critically dangerous. However, any [physically or
logically] meaningful n-factor combination (Φ i(1)∧Φ i(2) ∧ … ∧Φ i(j)∧ … ∧Φ i(n)) can result in a complex
(multifactor) accident-prone situation, i(j)∈{1, 2, … 12, …}, n∈{2, 3, 4, …}.
multifactor) accident- situation, (j)∈ 12, 4
5. Multifactor Flight Situation Build-up Chain
Build-
(Takeoff Example)
INTELONICS LTD.
catastrophic state
Legend:
Legend:
Ф4, Ф6, Ф7, Ф9, Ф10 – operational/design factors
(or risk/ ’what-if’/ flight path branching factors)
’what-
– colors for coding flight safety levels
– ‘bud’-type event for ‘implanting’ additional operational
‘bud’-
factors into a [less complex] flight situation scenario
safe
state S0, S1, …, S5 – notional flight situation scenarios in the order of increasing aircraft
motion dynamics & control complexity and operational risk
Ω(Ф|Sk) – subset of operational factors affecting scenario Sk, k = 0, 1, …, 5: Ω(S0) = ∅, Ω(Ф|S1) = {Ф4},
Ω(Ф|S2) = {Ф4, Ф10}, Ω(Ф|S3) = {Ф4, Ф7, Ф10}, Ω(Ф|S4) = {Ф4, Ф6, Ф7, Ф10}, Ω(Ф|S5) = {Ф4, Ф6, Ф7, Ф9, Ф10}.
5
6. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Results and Discussion
Potential for Real-time Applications
Real-
Conclusions
Backup Slides
6
7. Problem Formulation
INTELONICS LTD.
How to analyze/assess, predict and protect aircraft safety
performance in multifactor (complex, abnormal, anomalous,
uncertain) flight situations ?
What is the cause-and-effect mechanism of irreversible (‘chain-
cause-and- (‘chain-
reaction’ type) developments of the aircraft flight path under
multifactor conditions: accident precursors, key contributing factors,
conditions:
‘last-change-for-
‘last-change-for-recovery’ point, good and bad control rules
(“do’s” and “don’ts”) for human pilot or/and control automaton,
automaton,
etc. ?
Are there reliable and affordable techniques available for
implementation – during the aircraft’s life cycle (phases: design, test
& certification, training, operation, accident investigation) – in order
to help identify and avoid (or recovery from) a potentially
dangerous multifactor situation ?
7
8. Solution Approach
Main Principle INTELONICS LTD.
‘Knowledge is Power’: the ‘pilot – /automaton – aircraft – operational
environment’ system model is employed as ‘knowledge generator’.
Research Goal
Develop and demonstrate a technique for prediction and mapping of
aircraft/ project safety performance in multi-factor flight situations in
multi-
advance, before the vehicle is built/flown.
Techniques Employed
Applied aerodynamics, flight dynamics, situational control, multifactor flight
aerodynamics, dynamics, control,
domain theory, mathematical modeling, numeric techniques, computer
techniques,
simulation experiment, artificial intelligence (AI), graph theory, dynamic data
experiment, (AI),
structures, computer graphics, VATES (Virtual Autonomous Test & Evaluation
graphics,
Simulator, v. 5/7) software tool, Pentium-IV PC, MS Office, MAGE, etc.
Pentium-
Classic techniques + Modern techniques = New analytical potential.
potential.
8
9. Main Tasks. Customer Groups
Main Tasks INTELONICS LTD.
1. Automated design and fast-time examination of a broad set of realistic
fast-
scenarios and multifactor operational hypotheses in order to explore
potentially unsafe/anomalous flight situations – using the system model,
the vehicle’s ‘parametric definition’, and autonomous flight simulation
techniques.
2. Automated ‘mining’ from flight M&S output data, ’granulation’ (by L.
data,
Zadeh)
Zadeh) and generalization of new safety-related knowledge using a set
safety-
of anthropomorphic ‘knowledge maps’.
Target Customer Groups
{ Designer, Test Pilot/Engineer, Regulator, Educator/Instructor, Line Pilot,
Investigator, … }
In autonomous M&S setup, a research pilot is not needed in the simulation loop. During a
simulation experiment, realistic operational hypotheses (meaningful combinations of several
operational factors) are automatically generated and added to a baseline scenario for fast-time
fast-
virtual testing on a 6DOF flight model.
model.
9
10. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Potential for Real-time Applications
Real-
Results and Discussion
Conclusions
Backup Slides
10
11. ‘Pilot /Automaton – Aircraft – Operational
Environment’ System Model as Knowledge
Generator Operational system
INTELONICS LTD.
(DO∪DT+M)⊂DV TV >>(TO + TT+M) SV < ST+M
>>( (‘pilot/ automaton – aircraft –
operational environment’)
DV – flight domain explored in situational DO – flight domain Pilot
M&S [virtual flight] experiments explored in operations (automaton)
automaton)
Operational
Aircraft
environment
Flight test-bed + M&S stand
test-
Research pilot
(automaton)
automaton)
Simulated
Research flight
operational
simulator
environment
DT+M – flight domain Assigned
explored in flight tests operational Virtual flight test-bed
test-
constraints (situational system model)
and manned simulations
Mathematical
Legend:
Legend: model of pilot
TT+M - test flight and M&S time (flight test and manned M&S experience) (automaton)
automaton)
TO - operational flight time (operational flight experience) Mathematical Math model
model of flight of operational
ТV - virtual test flight time (‘virtual’ test experience) dynamics environment
S - R&D cost (flight safety analysis, prediction and protection)
11
12. Two-
Two-Level Knowledge Structure of Complex
Flight Situations Domain
INTELONICS LTD.
Micro-
Micro-structure Macro-
Macro-structure
Elementary Flight situation Situational
situation scenario tree C1
Event ... ...
E Ek ... П14
П2 E6
Пj П9 C2 B0
П B1
Ei E2
Process П8
П10
E4 П5
П1 П3 B2
Legend: Ei - flight event; Пj – flight E8
П4 ...
process; Cm – fuzzy constraint; - E1
system reference state; - system
E5
...
branching state (‘bud’); - system П7 П11
target state (‘leaf’); - system source П15 П6
state (‘root’); B-1 – parent branch; B0 – E7
main branch or ‘trunk’ (baseline E4
scenario); Bn – nth-order derivative П13 C4
П12
C3
branch (scenario with n operational ...
factors , n = 1, 2, …). ... B-1
Micro-
Micro- and macro-structure are two interrelated components of a generalized knowledge model
macro-
of a complex flight domain. Each flight path (a branch in the situational tree) is modeled according
(a tree)
to an event-process type scenario with a combination of n operational factors, n = 0, 1, 2, ….
event- ….
12
13. M&S Based Flight Safety Virtual Testing Cycle 13
1 Wind tunnel (*) 2 Experimental data measurement 3 Output test data files 4 Computational aerodynamics (‘virtual wind
and processing system (‘3 forces and 3 moments’) tunnel’), aircraft parametric definition database
INTELONICS LTD.
formation tools
test rig
aircraft model
8 Autonomous situational model of the ‘operator (pilot,
automaton) – aircraft – operational environment’ system
behavior (VATES)
7 Library of flight situation 5
Aircraft flight model input
scenarios for virtual testing and А Flight Aircraft model B database (aerodynamics, thrust,
dx
certification situation = f (x,u,w,t) ‘parametric inertias, geometry, etc.)
scenario dt
definition’
6 Flight situation content requirements: 9 Customer (aerodynamicist, 10
dynamicist, pilot, …) Computer
АП/FAR/JAR, compliance testing
methods, or flight test programs, or 13
Pilot’s Manual, or flight test/ accident
‘Flight’
records
15 Maps of aircraft’s safety performance in complex
situations (system-level knowledge of multifactor
effects of operational conditions on flight safety)
12 14 Systems model’s output database (‘flights’,
Virtual hypotheses, statistics, etc.)
‘test-bed’
&size [n_columns] [n_rows]
&name time [var01] [var02]
&unit s [unit01] [unit02]
11 Operational (‘what …, if …?’) &format (f6.2, 20f10.4)
[time] 499.9999 236.1820 3.8520
hypothesis for virtual testing – [time] 499.9782 236.2703 3.8821
[time] 499.8870 236.3342 3.9107
situational tree ‘genotype’ ...
Legend: - direction of information flow processing; 1, …, 15 - flight safety T&C process components; (*) – courtesy of Dr.
N.Sokhi; - feedback link; A and B – system model’s two main input data sets. 13
14. Safety Palette. Fuzzy Constraint
INTELONICS LTD.
Safety Palette green (‘norm’), ξG
yellow/ amber (‘attention’), ξY Color is natural and, perhaps,
the most effective and economic
red (‘danger’), ξR medium for communicating safety-
safety-
black (‘catastrophe’), ξB related information to / from an
operator (a pilot or automaton).
grey/white (‘uncertainty’), ξW
‘uncertainty’)
Fuzzy Constraint µC(VFL.D.) C: ‘permitted flaps-down flying IAS’
flaps- IAS’
1
с d
Legend:
Legend: c, d – characteristic …
points of the carrier of fuzzy 0 …
… 390 410 470 VFL.D.. [km/h]
FL.D
set-constraint C, µC(x) – L.A.
set-
Zadeh’s fuzzy set membership …
function ‘green’ ‘yellow’ ‘red’ ‘black’
Operational constraints under multi-factor flight conditions are not known precisely. They are
multi-
inherently ‘fuzzy’. The notions of fuzzy constraint (by L.A. Zadeh) and safety palette are employed
‘fuzzy’. Zadeh)
for approximate measurement of the compatibility of current (i.e. measured at time instants t)
system state with operational constraints for key (monitored) system variables.
14
15. Partial Safety Spectra. Integral Safety
Spectrum
INTELONICS LTD.
δ
IAS (δF = 0, airborne) Σ1 Legend: Σ k – partial safety spectrum
δ
IAS (δF > 0, airborne)
Sideslip Σ2 for variable xk, k = 1, …, p; p – total
Monitored variables/ constraints
number of monitored constraints/
Partial flight safety spectra
Load_factor
variables, p = 20. Σ – integral safety
East_rate (groundroll)
East (groundroll) ... spectrum; t – flight time; ξi – color from
North (groundroll) safety palette, i ∈ {B (black), R (red), Y
Bank (airborne) (yellow), G (green),…}; < – ‘colder than’
Bank (groundroll) operation for comparing two safety
Pitch (airborne)
Pitch (groundroll)
Σk colors; max – operation of selecting the
‘hottest’ color at time instant t; || -
Vert_rate (airborne) operation of safety colors
δ
AoA (δF = 0) concatenation in Σ ; [t*; t*] – examined
δ
AoA (δF > 0) flight time interval; ∆ – spectrum
Wheels (airborne) ... construction time step.
Wheels (groundroll)
Elevator (airborne) - green (‘norm’), ξG
Elevator (groundroll) - yellow (‘attention’),
Aileron
Rudder Σ20 ξY (‘danger’), ξR
- red
Integral spectrum Σ - black (‘catastrophe’), ξB
- gray/white (‘uncertainty’), ξW
time, s
Integral Safety Spectrum Calculation Algorithm:
(∀t) (t∈[t*;t*]) (∃ξ(xk(t)) (ξ(xk(t))∈{ξW, ξG, ξY, ξR, ξB, …} ∧ (ξW < ξG < ξY < ξR < ξB))
(ξ(t) = max ξ(xk(t)), k = 1, …, p) ⇒ (ξ(t)∈Σ ∧ Σ = ξ(t*) || ξ(t*+∆) || ξ(t*+2∆) || … || ξ(t*))
Σ
For each simulated situation, its safety level is measured for selected key variables xk at recorded
time instants. As a result, a family of Partial Safety Spectra Σk, k = 1, …, p, and an Integral Safety
Spectrum Σ can be calculated for this situation. The integral safety spectrum is a color-coded time-
history of all violations and restorations of the monitored fuzzy constrains during the situation.
15
16. Flight Safety Classification Categories
INTELONICS LTD.
In order to measure safety performance for a flight situation in overall, a special ‘safety
ruler’ consisting of five classification categories I, …, V is employed. Why five? – because
experts cannot reliably recognize and use more than 5-10 gradations of a complex, difficult-
to-formalize system-level property (e.g.: Cooper-Harper scale). ‘Light green’, RGB (192; 255;
0), and ‘orange’, RGB (255; 192; 0), are interim colors used to denote Categories II-a and III.
16
17. Situational Tree of Flight
INTELONICS LTD.
T130: “Takeoff and initial
climb, ‘very strong’ wind-
shear, variations/ errors
of commanded flight
path (ΘG) and bank (γG)
angles”
‘Virtual flight test
experience’ accumulated
in tree T, hrs:
Legend: ΘG∈{2о, …, 20о} – commanded flight path errors, γG∈{-45о, …, +45о} – commanded bank angle errors,
T130|Г(Ф1×Ф2×Ф3)={F2682, …, F2811} – situational tree, Г(Ф1×Ф2×Ф3) – tree’s genotype (operational hypothesis), Фk –
operational factor, Ф1≡ΘG, Ф2≡γG, Ф3≡(Wxg,Wzg=f(t)) – ‘very strong’ wind-shear; N(T130)=130 – number of branches in
T130, ∆t(Bi)=60s – branch ‘length’, i=1, 2, …, 130; - safety palette
A composition of a baseline situation scenario and an operational multifactor combination in
M&S experiment results in a situational tree. The tree’s branches (flight paths) stand for ‘what-if’
derivative (non-standard) situations. All branches are color-coded using ‘integral safety spectra’.
17
18. Safety Window. Safety Chances Distribution
Example of mapping a situational tree S1⋅Г11: Takeoff. Errors of Selecting INTELONICS LTD.
Commanded Flight path and Bank Angles in Climb
Safety Window A
Safety Chances
Distribution Pie Chart
C
B
3
Let us map safety levels (categories) obtained for all situations from a Category ξj nj χj , %
tree onto a two-factor plane. This results in a Flight Safety Window (FSW). In
two- plane.
I 37 28
FSW above, cell C is located at ‘column A - row B’ crossing. This cell depicts
safety status of one flight path-branch from the tree. This is a non-standard
path- tree. non- II-a 8 6
situation with values of 14 o and 30o of factors Φ and Φ II-b 29 22
7 11 in S1. The cell is
colored using the safety category color ‘orange’. Note that the FSW has a
orange’. the has III 1 1
dangerous ‘corner’ (upper-left). Rapid transition (3) from safe (‘salad
(upper-left). (3
IV 55 43
green’) to dangerous (‘red’) zone is possible (Cat. II-a → IV), bypassing
IV)
interim zones (II-b, III). Flight control at such ‘corners’ obviously requires
(II- III) V 0 0
enhanced pilot attention. Σnj, Σχj | S1⋅Γ11
Γ 130 100
18
19. Flight Safety ‘Topology’
Operational/ design factor Ф1
3 6 2 INTELONICS LTD.
5
I
2 1 II-a
II-b
3 III
3 IV
V
4
Transitions 6 must be Transitions 3 must be
Flight Safety
Categories
known and prevented! known and controlled!
6 3
1 4 2
Operational/ design factor Ф2
1 ‘Abyss’ (catastrophe) 4 ‘Valley’ (standard safety, norm)
1, 2,…, 6 - main
2 ‘Hill’ (danger) 5 ‘Lake’ (maximum safety, optimum)
object types of flight 3 ‘Slope’ (reversible state 6 ‘Precipice’ (abrupt, irreversible
safety ‘topology’: transitions) state transitions, or ‘chain reaction’)
19
20. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Results and Discussion
Potential for Real-time Applications
Real-
Conclusions
Backup Slides
20
21. Baseline Flight Scenarios 21
INTELONICS LTD.
Baseline scenario Si is a plan of some ‘central’/reference (any standard or non-standard)
flight situation, which variations (derivative cases) are virtually tested in autonomous M&S
experiments. The goal is to evaluate combined effects of selected operational/design factors
on flight safety in these scenarios. The sources of data for baseline scenarios are: airworthiness
requirements, flight test data/programs, ACs, Pilot’s Manuals, real flight data records, flight
accidents/ incidents statistics.
21
22. Joint Graph of Baseline Scenarios
W1: crosswind 10 m/s
(left-to-right)
F1: left-hand engine S3 S5 INTELONICS LTD.
E44: engine failure …
44 out speed
S2 E88: altitude 200 m 88
6 P3: wheels - up
S1 E6: altitude 10.7 m …
T2: maintain T5: maintain
commanded bank γG & commanded bank γG
P1: set engines #1,2 levers … 7 P4: flaps - up
… heading ΨG angles & sideslip β G angles
to takeoff rating E7: altitude 120 m
T1: maintain path in
1 groundroll along runway’s 55 190
centerline
E1: situation start E55: in airborne E190: situation end
T2: maintain commanded
bank γG and heading
3
P2: elevator – ΨG angles
up for rotation T4: maintain
E3: VR achieved …
P5: maintain given commanded flight
T3: maintain commanded flight indicator airspeed path angle θG2 (2nd
path angle ΘG1 (initial phase of phase of climb)
5 climb)
E5: pitch 8о E12: flaps retracted
12
S4
Legend: W2: ‘strong’ wind
W1: crosswind -10
m/s (right-to-left) shear
44 Scenario consists of events and processes. It can be depicted as a
E44: engine out speed- event
directed graph. The scenario defines logic and content of a flight
F1: left-hand engine failure situation. It is also clear to the pilot. Scenarios S1, …, S5 are structurally
- process
close. They can be easily modified.
22
22
23. Operational Factors Selected for Testing
INTELONICS LTD.
Operational /design factors are modified or new events and/or processes in a flight scenario,
which can improve (or worsen) the aircraft safety performance. There are three groups of
operational factors: ‘operator’, ‘aircraft’ and ‘external environment’. The sources of information
on operational factors are airworthiness requirements, FMEA, statistics on flight operations, and
accidents/incidents data.
23
24. Design Field of Operational Hypotheses
Elevator deflection for Wheels - runway surface
rotation Ф3 Ф4 adhesion factor
∆δe µ Crosswind
INTELONICS LTD.
Rotation Ф5 velocity
Г2
airspeed Ф2 Wyg
VR
Г1 Г10 Г4 Flaps-up start
Ф6
HFL altitude
Г3
Longitudinal
C.G. Ф1 xCG Г5 Г6
Commanded flight
Г7 θG2 Ф8 path angle during
2nd phase of climb
Left-hand
engine Ф13
ζLHE Г8 Г13
failure at VEF Г9 θG1 Ф7
Commanded flight
Г11 path angle during
VEF Г12 initial phase of climb
‘Engine out’ Ф12
indicator airspeed kW Ф9 Intensity of wind-
kP γG shear
Engines power Ф10
Ф11
Legend: rating at takeoff Many operational factors from this list
Commanded bank are not critically dangerous alone. Much
Г13 - operational hypothesis angle more important to learn in advance
independent - link between
effects of unfavorable combinations of
Cross wind - operational factor
Wyg Ф5
velocity dependent these factors on flight safety.
factors in Г
24
25. Plan and Statistics of M&S Experiments
INTELONICS LTD.
Legend: i – code of baseline scenario Si, i=1, …, 5; k – code of operational hypothesis Гk, k=1, …, 13; N(Ф) – number
of operational factors in Гk; n – size of ‘flight’ series Ωk(F), Ωk(F)={Fi1, …, Fj, …, Fin}, n=in-i1+1, j – ‘flight’ code; ∆t –
planned duration of ‘flight’ Fj, Fj∈Ωk(F); ℑ|Si⋅Гk – ‘virtual flight test experience’ accumulated in tree Si⋅Гk; notation of
coordinate axes corresponds to ISO 1151.
Composition of baseline scenario Si and operational hypothesis Гk results in a family of
derivative (‘neighboring’) situations – a ‘situational tree’ Si⋅Гk. Construction of a ‘forest’ of such
trees - based on FMEA, flight test/operation/ incidents/accidents data - and studying their safety
‘topology’ in autonomous M&S experiments is the goal of virtual flight T&C.
25
26. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Results and Discussion
Potential for Real-time Applications
Real-
Conclusions
Backup Slides
26
27. Composition S1⋅Г1 Normal Takeoff. Variations of
Takeoff.
Flight situation code
C.G. and VR Speed (with
Correction of Elevator
Integral Safety Spectra # mP mF VR ∆δe
Deflection in Rotation) INTELONICS LTD.
Tested operational
factors
Safety Chances Distribution
0, 0%
Category ξj χj , %
I 100
II-a 0
II-b 0
III 0
IV 0
V 0
66, 100%
‘Flights’ in total - 66 100
Legend: in nj, χj% nj – number of ‘flights’ belonging to
Cat. ξj, χj% - percentage of ‘flights’ of Cat. ξj, j=I, …, V.
All situations from Composition S1⋅Г1 are safe, i.e.
they belong to Category I cluster. Note how location
of events E3 and E7 on integral safety spectra is
changed due to situation (operational factors).
time, s 27
28. Composition S1⋅Г1 Normal Takeoff. Variations of C.G.
Takeoff.
and VR Speed (with Correction of
Elevator Deflection in Rotation)
S1:Normal takeoff, steering INTELONICS LTD.
commanded flight path and ⇒ In FSW below, cell 1 located at ‘column 2 - row 3’
bank angles during initial climb crossing is a color code of flight safety Category of
one situation from Composition S1⋅Г1. This situation is
obtained by combining values 4 and 5 of
operational factors 6 and 7 in scenario S1.
Flight Safety Window 2
6
4
7
3 5 1
This Flight Safety Window constructed for Composition S1⋅Г1 situations has ‘trivial topology’ –
one continuous green ‘valley’. That is, for a given aircraft/project all examined combinations of
longitudinal C.G. location and VR speed variations are acceptable safety-wise (NB: provided
that all other conditions of scenario S1 are fulfilled).
28
29. Composition S2⋅Г2 Normal Takeoff. Variations of
Crosswind Velocity and ‘Wheels -
Integral Safety Spectra # µ k⋅Wy
g
Runway Surface’ Adhesion Factor INTELONICS LTD.
k=10-1
Safety Chances Distribution
21; 33% 22; 35%
12; 19% 6; 10%
2; 3%
Variants with strong crosswind of |15|…|20|
m/s exhibit danger during groundroll up to event
E3 (VR) - ref. next slide for FSW. These variants
constitute 45% of all tested flight situations from
composition S2⋅Г2. Remaining situations (55%) are
safe - they belong to Categories I and II. Note
how the location of events E3 and E7 in IFSS is
changed due to the effect of (µ, Wyg)
combinations.
29
30. Composition S2⋅Г2 Normal Takeoff. Variations of
Crosswind Velocity and ‘Wheels –
Runway Surface’ Adhesion Factor
INTELONICS LTD.
S2: Normal takeoff under cross-wind and varying conditions of runway surface,
steering commanded flight path and bank angles during initial climb
Flight Safety Window
1 1
2 2
Shown above is Flight Safety Window constructed for situational tree S2⋅Г2. It contains one
central green ‘valley’, two side red ‘hills’ and two connecting ‘slopes’: (1) a steep ‘slope’ – for
dry and semi-wet runway, and (2) not steep ‘slope’ - for wet and water-covered runway. As the
absolute value of cross-wind velocity increases, transitions from safe to dangerous states occur
(1) sharply and (2) gradually, respectively. The shape and position of ‘crosswind velocity –
adhesion factor’ constraints can be seen as well.
30
31. S1⋅Г3 Normal Takeoff. Forward C.G. Location.
Variations/Errors of Selection of Commanded
Flight Path Angles (Initial and 2nd Phases of INTELONICS LTD.
Climb) and Flaps-up Start Altitude
Flaps-
Integral Safety Spectra # θG1 θG2 HFL
Safety Chances Distribution
0, 0%
5, 14%
0, 0%
0, 0%
5, 14%
25, 72%
14% of variants from situational tree S1⋅Г3,
14%
which have commanded flight path angle
(during initial phase of climb) more than
12o, exhibit danger. Note also how, for
danger.
example, event E7: ‘altitude 120 m’
changes its location in IFSS due to θG1.
31
32. S1⋅Г3 Normal Takeoff. Forward C.G. Location.
Variations/
Variations/ Errors of Selection of Commanded
Flight Path Angles (Initial and 2nd Phases of Climb) INTELONICS LTD.
and Flaps-up Start Altitude
Flaps-
S1: Normal takeoff, steering commanded flight path and bank angles during initial
climb
Flight Safety Window
For composition S1⋅Г3, sharp transitions (1) from safe situations to unsafe ones are observed at
commanded flight path angles θG1/θG2>12/10o for all values of HFL. Owing to high thrust-to-weight
ratio, errors in selection of flaps-up start altitude do not worsen the aircraft’s flight safety
performance, provided (NB) that other conditions of scenario S1 are preserved.
32
33. S3⋅Г5 Continued Takeoff. Left-hand Engine Out At VEF=150
Takeoff. Left-
km/h.
km/h. Variations/ Errors of Selection of Commanded
Flight Path Angles During Initial and 2nd Phases INTELONICS LTD.
Integral Safety Spectra # θG1 θG2 Safety Chances Distribution
7, 17% 18, 43%
8, 19%
0, 0%
1, 2%
8, 19%
If left-hand engine fails during ground-roll (at
left- ground-
VEF=150 km/h) takeoff safety cannot be
km/h)
secured at commanded flight path angle
θG1≥5o (during initial phase of climb). For
climb).
examined domain of operational factors, share
of safe situations is 36%.
36%
33
34. S3⋅Г5 Continued Takeoff. Left-hand Engine Out at VEF=150
Takeoff. Left-
km/h.
km/h. Variations/ Errors of Selection of Commanded
Flight Path Angles During Initial and 2nd Phases INTELONICS LTD.
S3: Continued takeoff (left-hand engine out at given VEF), steering commanded
flight path and bank angles during initial climb
Flight Safety Window
Left-hand engine failure during ground-roll decreases the limit of flight path angle admissible
in initial climb to 2o…4o compared to θG1=10o …12o in composition S1⋅Г3. ‘Precipice’ type
transitions (1) are observed at θG2=0o. ‘Abyss’ type states are likely to occur at flight path angles
θG1>4o (initial climb) for any θG2 (2nd phase of climb).
34
35. S4⋅Г6 Normal Takeoff. Variations of Wind-shear
Takeoff. Wind-
Intensity and Errors of Selection of Flaps-up Start
Flaps-
Altitude INTELONICS LTD.
S4: Normal takeoff under windshear conditions, steering commanded flight path and
bank angles during initial climb
Flight Safety Window
2
In scenario S4 we have θG1/θG2=8o/8o. If ‘strong’ or worse windshear is expected (kW≥1) takeoff
is prohibited. In order to evaluate possibility of safe outcomes at kW<1 it is expedient to expand
Flight Safety Window downward. If windshear intensity increases from ‘very strong’ (kW>1.4) to
‘hurricane’ (kW=2), ‘precipice’ type transitions (1) are most likely to occur at flaps-up start altitude
HFL∈[60; 70] м. If aircraft unintentionally enters a zone of ‘very strong’ windshear (kW=1.2 …1.6)
flaps must be retracted as late as possible to stay within ‘orange’ zone (2).
35
36. S4⋅Г7 Normal Takeoff. Forward C.G. Location. Variations
Location.
of Wind-shear Intensity and Commanded Flight Path
Wind-
Angles (During Initial and 2nd Phases) INTELONICS LTD.
S4: Normal takeoff under windshear conditions, steering commanded flight path and
conditions,
bank angles during initial climb
Flight Safety Window
For composition S4⋅Г7 main objects of safety ‘topology’ are: small green ‘valley’ (at left lower
are:
corner), orange ‘slope’, extensive red ‘hill’ adjacent to black ‘abyss’ (at right upper corner). At
corner).
takeoff under ‘strong’ and ‘very strong’ windshear conditions (1<kW≤1.6): maximum safety is
achieved at θG1/θG2=5o/3o; it is prohibited to climb at θG1/θG2>7o/5o; irreversible transitions are
likely at θG1≥12o.
36
37. S5⋅Г10 Continued Takeoff. Left-hand Engine Out at VEF.
Variations of Left-hand Engine Out Speed and
Cross-wind Velocity
INTELONICS LTD.
S5: Continued takeoff (left-hand engine out at VEF), under cross-wind conditions, steering
commanded flight path and bank angles during initial climb
Г10 = Ф13×Ф12×Ф4 ≡ ζLHE×VEF×Wyg
Flight Safety Window
This Flight Safety Window has central green ‘valley’ and two side red ‘hills’. Adjacent to left
‘hill’ is a potentially catastrophic ‘abyss’ located at lower left corner. It is created at small and
medium values of VEF and is linked to ‘valley’ by ‘precipice’ type transitions. Small ‘abyss’ is also
revealed at crosswind velocity of ~18 m/s and VEF∈[175; 190] km/h.
37
38. S1⋅Г11 Normal Takeoff. Variations/ Errors in Selection of
Commanded Flight Path and Bank Angles
(During Initial Phase of Climb) INTELONICS LTD.
S1: Normal takeoff, steering commanded flight path and bank angles
takeoff,
during initial climb
Г11 = Ф7×Ф11 ≡ θG1×γG
Flight Safety Window
This Flight Safety Window has a potentially dangerous ‘corner’ corresponding to (θG1, γG) ≅
(12o…14o, -30o…-37.5o). Sharp transition (1) of states from safe (‘green’) to dangerous (‘red’) zone
37.
is possible (Cat. I→IV), bypassing interim zones (Cat. II, III). Flight at such ‘corners’ requires
Cat. IV), (Cat. III)
enhanced attention and accurate piloting from pilot. pilot.
38
39. S4⋅Г13 Normal Takeoff. ‘Very’ Strong Wind-shear.
Wind-shear.
Variations /Errors of Selection of Commanded Flight
Path and Bank Angles in Climb INTELONICS LTD.
Integral Safety Spectra # θG1 γG Integral Safety Spectra # θG1 γG
Safety Chances Distribution
7, 5%
26, 20%
20, 15%
10, 8%
10, 8%
57, 44%
‘Very strong’ wind-
wind-
shear may worsen flight
safety ‘topology’ of
takeoff catastrophically at
small values of
commanded flight path
angle θG1≤4o.
39
40. S4⋅Г13 Normal Takeoff. ‘Very’ Strong Wind-shear.
Wind-shear.
Variations /Errors of Selection of Commanded
Flight Path and Bank Angles in Climb INTELONICS LTD.
S4: Normal takeoff under windshear conditions, steering commanded flight path and
conditions,
bank angles during initial climb
Г13 = Ф9×Ф7×Ф11 ≡ kW×θG1×γG (kW=1.5)
Flight Safety Window
Flight safety ‘topology’ obtained for ‘very strong’ wind-shear conditions at small θG1 and any
wind-
γG contains a stable catastrophic ‘abyss’ (black strip in the bottom) and ‘‘precipice’ type
transitions (1). That is, an attempt of initial climb at small values of commanded flight path angle
(2o…4o) inevitably leads the vehicle to a fatal outcome.
40
41. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Results and Discussion
Potential for Real-time Applications
Real-
Conclusions
Backup Slides
41
42. Dynamic Safety Window Sequence
INTELONICS LTD.
t = t0
(‘benign Normal Takeoff. Variations
weather’)
of Wind-shear Intensity,
Wind-
Errors/ Variations of
Optimal modes -
maximum safety
Selection of Commanded
Flight Path and Bank Angles
in Initial Climb – ‘forest’ of
t = t1 situational trees
(‘strong’
wind-shear)
The developed safety
‘topology’ maps, including Flight
Safety Window, Safety Chances
Pie Chart and other formats, can
be potentially useful for flight
operations.
t = t2
(‘very strong’ The goal is to monitor operational
wind-shear) constraints and dynamically
adapt piloting tactics under
multifactor conditions in real time,
provided that there exist onboard
technical means to measure
operational factors in real time.
42
43. Potential Contribution to Integrated
Intelligent Flight Deck Initiatives (1)
INTELONICS LTD.
Safety Window
(commanded
‘flight path angle
– bank angle’ )
Safety Chances
Distribution
Wind-
Wind-shear ‘benign weather’
weather’ ‘strong’
strong’ ‘very strong’
strong’
forecast
- Optimum (safety-wise)
(safety- wind-
wind-shear impact
piloting modes real-
real-time analysis
The concept of dynamic safety window can be potentially useful to help pilot/automaton
predict aircraft safety performance in various ‘what-if’ scenarios and find optimum control
‘what-
tactics under demanding conditions (in this specific takeoff and initial climb case – commanded
‘flight path angle – bank angle’ pairs).
43
44. Potential Contribution to Integrated
Intelligent Flight Deck Initiatives (2)
INTELONICS LTD.
Dynamics Safety Window Tree. Safety Chances Distribution Time-history
Time-
19 13
L
18 12 G
K
17 11
J 16 S↑ S↓ 10 Dynamics Safety
F Window Tree and
15 9
I Safety Chances
14 8 Distribution Time-
Time-
H E history maps
7 7
D D are expedient to
6 6 study as analytical
C C tools for supporting
5 5
automatic or
4 4 manual recovery
S0 S0 decision-making in
3 3
B B emergency
2 2 situations
1 1 under uncertainty.
A 0 0 A
-1 -1
100 80 60 40 20 0 0 20 40 60 80 100
i i
χj, % χj, %
44
45. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Potential for Real-Real-time Applications
Real-Real-
Results and Discussion
Conclusions
Backup Slides
45
46. Conclusions
The developed two-level ‘pilot / automaton - aircraft –
two- INTELONICS LTD.
operational environment’ system model :
is powerful, affordable and easy-to-use system-level
easy-to- system-
safety mapping, analysis and prediction tool
focuses on complex (multifactor, uncertain,
(multifactor,
anomalous) flight situation domains
enables systematic aircraft safety research beginning from
early design phases
incorporates advanced safety ‘knowledge-mapping’ techniques
knowledge-
including ones for potential real-time applications
real-
provides 102-103 times increase in M&S based structured
(‘granulated’) information on flight safety in advance
helps enhance aircraft flight safety performance
apriori,
apriori, i.e. not necessarily based on accident statistics
complements flight testing & manned simulations,
especially when studying multi-factor cases
multi-
does require, however, a complete ‘parametric definition’ of
‘parametric
the vehicle/project for the flight domain of interest.
interest.
46
48. Presentation Outline
INTELONICS LTD.
Multifactor Flight Situation Domain
Problem and Solution Approach
Research Methodology
Modeling & Simulation Experiment Setup
Results and Discussion
Conclusions
Backup Slides
48
49. System Model Application Experience
(Simulated Aircraft Types/Projects: 1978-2010)
1978-
INTELONICS LTD.
A400M Prototype Military Transport (FLA F-93A)
F- Ilyushin- Medium-
Ilyushin-86 Medium-Range Airliner
Project (Cranfield University, UK)
(Cranfield Ilyushin-96-
Ilyushin-96-300 Long-Range Airliner
Long-
Advanced Hypersonic Maneuvering Aerospace Kamov- Multi-
Kamov-32 Multi-Purpose Helicopter
Plane Project *** Mil-26 Heavy-Lift Helicopter
Mil- Heavy-
Advanced Notional 4++ Generation Highly-Highly- Mil-8 Medium Multi-Purpose Helicopter
Mil- Multi-
Maneuverable Fighter (TVC) Project *** Sukhoi-
Sukhoi-49 Primary Pilot Training Airplane ***
Airbus A300-600 Long-Range Airliner
A300- Long- Sukhoi-80GP Multi-
Sukhoi-80GP Multi-Purpose Commuter
Amphibious Wing-In-Ground GA Plane Project
Wing-In- Airplane ***
Antonov-
Antonov-28 Commuter Airplane Supersonic Business Jet (SSBJ) Project (GIT)
(SSBJ)
Beriev-
Beriev-103 Amphibious GA Airplane *** Tupolev-134A Short-
Tupolev-134A/B Short-Range Airliner
Boeing-737-
Boeing-737-300 Medium-Range Airliner (GIT)
Medium- Tupolev-
Tupolev-136 Regional Cargo-Transport
Cargo-
Buran Hypersonic Aerospace Vehicle Project (cryogenic LNG-fuel) ***
LNG-
Cessna Citation X Business Jet (UTA) *** *** Tupolev-
Tupolev-154, -154M Medium-Range Airliner
Medium-
Concord Supersonic Passenger Airplane Tupolev-
Tupolev-204 Long-Range Airliner
Long-
High-
High-Speed Civil Transport (HSCT) Project (GIT)
(HSCT) Tupolev-334-
Tupolev-334-100 Short-/Medium Range
Short-
Hybrid (Aerostatic + Aerodynamic) Multi-Purpose
Multi- Airliner ***
Transport Aircraft Project (GTLA) *** ***
(GTLA) UAV and UUV Projects *** ***
Ilyushin-
Ilyushin-114 Regional Transport/Cargo Airplane XV-15 Bell Helicopter Textron Tilt-Rotor (GIT)
XV- Tilt-
Ilyushin-62M Long-
Ilyushin-62M Long-Range Airliner Yakovlev- Medium-
Yakovlev-42 Medium-Range Airliner
Legend:
Legend: 31 aircraft and projects in total, including: Hypersonic (2), Supersonic (4), Subsonic
(25, including 21 fixed-wing and 4 rotary-wing vehicles). GIT – Georgia Institute of Technology (USA).
25, fixed- rotary-
UTA – University of Texas at Arlington (USA). *** – VATES v.7 based macro-structural M&S (other – VATES
macro-
v.5 based micro-structural M&S). TVC – thrust vectoring control. *** – ongoing M&S research.
micro-
49
50. Overview INTELONICS Ltd. – Dr. Ivan Burdun
INTELONICS LTD.
EDUCATION CONTRACTS/COLLABORATION
1997 Special Non-Degree Research Course Boeing Company, USA
Georgia Institute of Technology (GIT), USA Central Aero-Hydrodynamic Institute (TsAGI), Russia
1993-1996 Doctorate Degree Research Course Cranfield Central R&D Inst. of Aerospace Systems Ltd. (TsNIIARKS Ltd.),
University, UK (thesis writing up not finished) Russia
1982 PhD Award RCAEI, USSR Chinese Aeronautical Establishment, P.R. China
City of Moscow Government, Department of Science &
1977-1980 Doctorate Degree Research Course Riga Civil Technology Policy (DNPP), Russia
Aviation Engineering Institute (RCAEI), USSR Cranfield University, UK
1971-1977 MSc Course in Aviation Mechanical Flight Safety Service of MoD Aviation, Russia
Engineering Riga Civil Aviation Engineering Georgia Institute of Technology (GIT), USA
Institute, USSR Ilyushin Design Bureaux, USSR
Kiev Civil Aviation Engineering Institute (KII GA), USSR
PROFESSIONAL BACKGROUND Ministry of Civil Aviation (MGA), USSR
MoD Flight Test Center, Russia
Molniya Science & Production Holding (NPO Molniya), USSR
Since 2007 Chief Scientist & Director INTELONICS Ltd., Moscow Civil Aviation Engineering Institute (MII GA), USSR
Russia NASA Langley Research Center, Aeronautics Systems Analysis
Since 2006 Associate Prof. Novosibirsk State Technical Branch (ASAB), USA
University, Russia Riga Civil Aviation Engineering Institute (RCAEI), USSR
2000-2007 Chief Research Officer Siberian Aeronautical Riga Technical University Aviation Institute (RTU), Latvia
Research Institute, Russia Siberian Aeronautical Research Institute (SibNIA), Russia
1997-2000 Research Engineer II GIT, USA Siberian State University of Telecommunication and Informatics
1983-1993 Senior Research Officer Riga Branch of Civil (SibGUTI), Russia
Aviation State Research Institute, USSR/Latvia State Research Inst. of Civil Aviation (GosNIIGA), USSR/Russia
Sukhoi Design Bureaux, Russia
1980-1983 Lecturer Assistant Riga Civil Aviation The University of Oklahoma, USA
Engineering Institute, USSR Tupolev Design Bureaux, Russia
1977-1983 Graduate Research Assistant Riga Civil Ulyanovsk Civil Aviation Training School (UVAUGA), Russia
Aviation Engineering Institute, USSR University of Texas at Arlington (UTA), USA
50
51. Acknowledgements
INTELONICS LTD.
The author wishes to thank:
Prof. David Allerton (University of Sheffield, UK),
Dr. Jean-Pierre Cachelet (Airbus, France),
Dr. Bernd Chudoba (University of Texas at Arlington, USA),
Dr. Dimitri Mavris (Georgia Institute of Technology, USA), and
Dr. Andrew Moroz (Lommeta JSC, Russia)
- for their multi-aspect support of this research, valuable advice and
cooperation.
51
52. Selected References
INTELONICS LTD.
1. Burdun I.Y. UAV ‘Built-in’ Safety Protection: A Knowledge-Centered Approach. Proc. of AUVSI
Unmanned Systems Europe 2007 Conference & Exhibition, 8-9 May 2007, Köln, Germany, 49 pp,
2007.
2. Бурдун Е.И. Прогнозирование безопасности полёта самолета гражданской авиации в
сложных условиях. Автореферат диссертации на соискание ученой степени доктора
инженерных наук, РТУ, Рига, 36 с, 2008 [in Russian].
3. Burdun I.Y. A Technique for Aircraft «Built-In» Safety Protection in Complex (Multifactor)
Conditions Based on Situational Modeling and Simulation. Proc. Of VIII International Conference
‘System Identification and Control Problems’, V.A. Trapeznikov Institute of Control Sciences,
Russian Academy of Sciences, January 26-30, 2009, Moscow, Russia, 44 pp, 2009. [in Russian].
4. Burdun I.Y. The Intelligent Situational Awareness And Forecasting Environment (The S.A.F.E.
Concept): A Case Study. Proc. of 1998 SAE Advances in Flight Safety Conference and
Exhibition, April 6-8, 1998, Daytona Beach, FL, USA, Paper 981223, pp 131-144, 1998.
5. Программно-моделирующий комплекс (ПМК) для исследований безопасности поведения
системы «оператор (лётчик, автомат) – летательный аппарат (ЛА) – эксплуатационная
среда» в сложных (многофакторных) полётных ситуациях (ПМК VATES). Свидетельство об
официальной регистрации программы для ЭВМ № 2007613256, выданное Федеральной
службой по интеллектуальной собственности, патентам и товарным знакам РФ.
Правообладатель: ООО «ИНТЕЛОНИКА». Автор: Бурдун И.Е. Зарегистрировано в Реестре
программ для ЭВМ 02.08.2007, Москва, 1 с, 2007 [VATES Patent, in Russian].
52