K Goebel (NASA), I Schagaev (Londonmet)
FLIGHT SAFETY SYMPOSIUM 2015
Active System Control for Safety, Maintenance, Efficiency
London Heathrow, UK 15th-16th Sept
1
active system control: why? Aviation safety…? OR? WHY? - is safety not enough? WHAT else? WHERE is maintenance, efficiency? Can we combine it?
NEXT: ABBAT, ASTI… tbc
HOW to achieve: new theory, technologies, infrastructures, policies
WHERE: aircrafts manufacturers, regulatory bodies, avia-, insurancecompanies,maintenance companies
2
safety accident aspect - I
https://www.academia.edu/8637255/Active_System_Control_-_Ch1-2 3
safety accident aspect - Ia Flight data recording (on ground) Leak visual manifestation Accident
Leak
FLIGHT (blast off) Flight able to continue in a safe state after recovery operation
Time for possible emergency reconfiguration and recovery of Challenger to safe state
10 seconds
72 seconds
4
safety accident aspect - II x9
Total pressure probe 1 x1 x2
Full pressure air-line x3
Airspeed Indicator
Static Head 1
x6
x4
Altimeter
x8
Total pressure probe 2
Static pressure air-line Air Data System
x5
x7
Variometer
Static Head 2 x10
In case of icing the pressure sensors TPP1 and TPP2 may be blocked. Then required pressure in the full pressure pipe is not available, devices AI and ADS show wrong values. “Correctness” of variables x1-x10 becomes questionable and Air pressure system goes to uncertain state. Resolving of this situation is possible: https://www.academia.edu/7124307/Analytical_Synthesis_of_Aircraft_Control_Laws Ignoring leads to: http://www.bea.aero/fr/enquetes/vol.af.447/note29juillet2011.en.pdf 5
Safety, what is it ? There is a difference in approach to aviation safety between the various parties involved: § Insurance companies § Airline operators § Airport operators § Operational regulators § Standards regulators § Aircraft and equipment manufacturers § Flight crew and ground maintenance staff § Passengers Importance of safety for them is financial, political or vital. 6
erning’ laws is shown in Table 1 below:
Safety and governing laws
Table 1 Features of system design using various laws and regulations Features
Laws physical
biological
social
aviation safety
Duration
forever
20-30 years
50-100 years
25 years
Avoidability
impossible
rare, but possible
rare but possible
possible
Application Rate
always
high
medium
medium
Ask yourself : Physical effect this is continuous and every system and material object is Whatlaws amareI unavoidable, doing totheir make shift?: governed by them absolutely. However, the effect of such physical laws often depends on context, e.g. less gravitational effect at very high speed and high altitude, greater heat accelerating chemical reactions. Source: https://www.academia.edu/8637255/Active_System_Control_-_Ch1-2
7
Systems which rely or similar in behaviour to biological laws in their design and operation have a
Classification of aircrafts? For what? …classification aims a formation of technical portrait of a typical aircraft including design, technological and management features… they altogether have substantial impact on aircraft reliability, maintainability, and, therefore, safety: https://www.academia.edu/8637255/Active_System_Control_-_Ch1-2
...existing schemes of aircraft (not air traffic) safety management are mostly conservative, oriented on after flight (and accident) analysis (CA, military) or exist at rudimental level (GA). …all these schemes are easily avoidable by aircraft owners and users, as they depend upon ‘human’ factor (the weakest link in the chain can’t be relied on to fix the chain)… https://www.academia.edu/7119860/The_Concept_of_Dynamic_Safety 8
Existing scheme of safety management
monitoring of hardware
data recording and archiving
control registration checking of pilot analysis control decision
transporting of hardware, instruments, and data
data processing and analysis
9
ASC definition Active System Control is an approach to: continuously evaluate and process the state of an aircraft in real time of flight; define when necessary and implement best possible RT recovery action or the scheme of graceful degradation Implementation of ASC for aviation assumes: - Process-oriented Information model of aircraft; - Flight data object and predicates; - Modeling of main elements; - Creation of dependency and recovery matrices Dependency matrix cortege includes matrixes of: input data; state and element dependency (more next slides) https://www.academia.edu/7110728/ PRINCIPLE_OF_ACTIVE_SYSTEM_SAFETY_FOR_AVIATION_CHALLENGES_SUPPORTIVE_THEORY_IMPLEMENTATION_APPLICATION_AND_FUTURE 10
ACS, arguments for: Practice speaks:
Symptoms of deviations from expected scheme, behaviour or rules should ignite search of “guilty” element(s) Known solutions for the localisation of problem are based on assumptions about analysed system and fault models, including fault tree analysis Regretfully, multiple faults are real, and fault tree analysis is… Also: probabilistic models of location of faulty element are not always adequate Real time detection and recovery and/or graceful degradation needs monitoring, thus…
New model of active system control is required 11
Active System Control vs. FTA FAULT TREE ANALYSIS
COMMENTS/REQUIREMENTS
The class of all faults is known before operation of the application
The class of faults changes during the life cycle of a system due to ageing, maintenance errors, incompleteness or poor quality or external reasons.
Fault tree analysis is static, developed during design of the objects and subsystems Fault trees are known from design time; are applied after events/accidents to detect and understand the cause(s) using expert opinion Number of fault trees, and their interactions, is growing quickly with complexity of a system
Dependence between elements of the system is reflected in dynamic changes of structure and significance.
Markov Processes are used to analyze outcome of the flow of events
A realistic model of the system might not be Markovian: the rate of transitions is changing, as well as importance of consequences and the possibilities for recovery and or repair before harm is inflicted.
Structure of dependence is becoming variable and must be updated during life cycle using existing, historical and new flight information; otherwise the predictive power of prognosis is not sufficient. New system needs to identify, keep and manage of possible scenarios of system behaviour in changing environment before unpleasant sequence of event takes place.
Mutual exclusion of possibilities are appliedThe consequences of many common events ARE NOT MUTUALLY EXCLUSIVE. New, more realistic model should for event analysis address that
There is no possibility to use FTA in real time of system operation
A model for control of the system behaviour MUST be used in real-time to take account the operational state of the system 12
Active System Control: Model + Flight data This is what we model…
tn t0
t1
t2
This is what we record…
13
Flight data
Flight mode
Aircraft elements
Flight data selfdependencies
Flight data margins per flight mode
Flight data affiliation to aircraft elements
Flight mode selfdependencies
Dependencies aircraft elements upon flight mode
Aircraft elements
Flight mode
Flight data
Active System Control: dependency matrixes
Aircraft’s element dependencies
! http://it-acs.co.uk/files/Grant_for_a_patent.PDF
14
ASC vs.conditional and preventive maintenance
(Conditional maintenance)
(Preventive maintenance)
“profit” is here and here… active system control is doable… for more see:
https://www.academia.edu/7119690/Applied_Cyber-Physical_Systems
15
ASC; further steps?: Principle of active system control: § introduced in 1986; § prototype for military tested 91-94 § In 2004 -09 developed for GA; GA prototype model tested using BA & SwissAir expertise + Complexity of system analysis grows linearly, RT What is next? § Framework of dependency matrixes using prognostic of system state in real time; § Population of dependency matrixes (System health management); § Active Black Box for Aviation (ABBA) 16
System Health Management Raw Sensor Data Sensor Validation
Validated Data Faulted Sensors Flagged
Feature Extraction
Time-stamped Features, Event Messages &/or Parametric Data
Anomaly Detection/Id
Warnings & Alerts Coarse Granularity Id (subsystem level)
Diagnostic Analysis
Subsystem Failure Modes
Prognostic Analysis
Remaining Useful Life Estimation
Fault Accommodation
Operational Corrective Action Identification/Reconfiguration/ Impact Contingency Assessment Management
ACS next step in hardware: ABBA? This? Or that?
http://www.nasa.gov/centers/ames/multimedia/images/2005/ blackbox.html
18
Conclusions: § Active system control theory works and … in real time § Using NASA prognostics special system software for proposed framework active system control: - increases safety of aircrafts at order of magnitude; - reduces cost of maintenance substantially - 1.8-2.9; - increases and guarantee maintenance quality; - reduces insurance premium in general. If we want it done… - Widening of cooperation/collaboration between research EU, USA, China, Russia, integration of effort of leading centres, regulatory bodies (Eurocontrol, EASA,NTSB) industries EADS/ Airbus, ESA, Boeing, etc) is required. contact:
[email protected],,
[email protected],
[email protected]
19