“big data” in aviation- technology & adoption … data in aviation...“big data” in...

23
“Big Data” in Aviation- Technology & Adoption Evolution Richard Hayden, FLYHT Dr. Harrison Chin, MIT Presented at: THE POWER, PROMISE AND PERIL OF ANALYTICS & BIG DATA Fisher School of Business Ohio State University February 24, 2012

Upload: dangthu

Post on 08-May-2018

218 views

Category:

Documents


4 download

TRANSCRIPT

“Big Data” in Aviation-Technology & Adoption Evolution

Richard Hayden, FLYHTDr. Harrison Chin, MIT

Presented at:THE POWER, PROMISE AND PERIL OF ANALYTICS & BIG DATA

Fisher School of Business Ohio State University

February 24, 2012

Aviation

• We “lift things up and put them down”• Lifeblood of global economy-

• Civil Aviation: >8% of global GDP• 15 million jobs and $1.1 trillion GDP • 33 million jobs and $1.5 trillion GDP if aviation’s impact on tourism included

• Defense contribution significantly increases numbers and industry importance• Largest net export of US economy• If Aviation were a country, it would be in the top 10 GDP’s in the world

• Essential component of national defense• Manned and unmanned

• >1/4 million platforms (>20,000 airliners & 15,000 unscheduled jets)

Aviation Data

• Aircraft – passengers – cargo – fuel – ATC – security -consumables - environment – pilots – maintainers - costs

The data is “big” and we have a lot of it.The issues are to make it usable & timely

Let’s just talk about the aircraft

• Complex system with enormous flexibility,

but absolute limits

• Encounters wide range of environments

• Man-made—therefore, can be modeled• But not always perfectly or accurately

• Limits are specified with safety (uncertainty) margins that drive maintenance

• Human-operated and maintained• Can’t be modeled or always predicted

• Major influence on outcomes

Aircraft DataIn the beginning…

• Compass, fuel gauge, barometer, horizon, good eyes, calibrated stomach and derriere

Today…• Up to 3000 sensors per aircraft

• Digital autopilots, engine controls, engine health monitoring systems, maintenance computers, fuel systems, satcom, weather radar, collision warning systems, ground proximity systems, instrument landing systems, braking controls, flight data recorders, self-diagnostic avionics

• Each aircraft generates ~100MB of data per flight hour (or more). Flight hours per year range from ~500 to >4000. Do the math!

• Availability of data is not the problem in aviation, but too much data can be

Foundation of an aircraft data strategy

What problem(s) are we trying to solve?

Different answers for different clients using same data base

In aviation, fundamentally, the requirement is:“Fly at any time; be on time-all the time”

• Safety first

• Maximize asset productivity

All detailed lower level requirements must be derived from & tested against these top level requirements

Requirements & issues with aircraft data

• Accuracy & reliability: assure data quality / avoid false alarms

• Usability: Concise / easily interpreted / provides context

• Timeliness: aviation is a real-time business / data availability must match operational tempo

• Automation and end-to-end integration: humans need to focus on what the data means

• Unambiguous response procedures: must be defined in advance, documented, and trained (new procedures can be evolved from new insights, after validation)

Integrated solutions exist

• ACARS (Aircraft Communications and Reporting System)

• IVHMS (Helicopter Integrated Vehicle Health Management Systems) (a.k.a. “HUMS”)

• AFIRS (Automated Flight Information Reporting System)

• Many partial fragmented systems exist - hard to integrate into a total system picture

ACARS

• 40-year old technology - augmented “voice only” comm

• Limited message capability-mostly used for position reporting and flight operations

• “Dumb system” – and open loop

• Dependent upon terrestrial infrastructure that is not globally available (satcom now augments ground stations)

• All messages routed through one of 2-service providers

• Technology is “max’ed out” - not easily expandable to take advantage of proliferation of digital data

• AF447 lessons…..human “system” failure

AF447—how could this happen?*• Modern aircraft with best data sources, ACARS, and satcom

• Weather situation was known (or knowable)

• Autopilot disengaged requiring crew to “hand fly”

• “Data” on cockpit displays misinterpreted by flight crew (“..no valid indications”)

• SOPs not followed – confusing data

• ACARS sent several messages –in cryptic but familiar formats

• Messages were received at AF ops

• No one was watching (and no automated alerting existed)

• 4-6 hours before ATC and AF realized AF447 was missing **

• Locating aircraft, passengers, & black box took “forever”

* http://www.bea.aero/en/enquetes/flight.af.447/flight.af.447.php** http://www.bea.aero/en/enquetes/flight.af.447/triggered.transmission.of.flight.data.pdf

Health and Usage Management Systems (HUMS)

DoD Requirements:

“Affordable Readiness”

Condition-based maintenance

Weak Link

25 shafts, 27 gears, and 70 bearings24 high frequency accelerometers & 1 tachometer

Data is collected at 100 kHz for 4 seconds for each accelerometer channel and tach

Total size per acquisition is approx. 20 MB [100 MB/flight hr.(5 acquisitions/flight hr.)]

Intermediate data = 3.5 MB per flight hour(Raw data spectra, signal averages, envelope spectra)

Condition Indicator (CI) data: 0.04 MB per flight hour (30 CI’s per mechanical component)

Health Indicator (HI) results:0.0015 MB per flight hour (1 HI per component)

On-board system data reduction (66,667:1)

Raw data / flight hour = 100 MB(“squiggly lines”- incomprehensible to users)

Processed data / flight hr.: 3.5 MB(Still needs expert interpretation)

29:1 reduction

88:1 reduction

27:1 reduction

CIs

Concise, unambiguous, actionable outputs

Measurable benefits-US Army experience

But, after 20 yrs. of HUMS technology V&V..

Major benefits are being “left on the table”

Despite-• Taxpayer-funded equipage of most (>4,000) aircraft• A thorough technology maturation program• Quantified payoffs on deployed aircraft• OEM buy-in and support

Key issues & lessons learned (Hayden’s opinion):• Continuity of policy and leadership• Fragmented responsibilities and commitment• Resistance to and subversion of change

AFIRS: Automated Information Reporting

“Blue Box”

Automated reporting

Voice & data - 2 way.

Connects to FDR and other data sources

Global satellite communications (Iridium) —no gaps or coverage limits

Relays secure information from Blue Box to server and back

Data transformed into timely messages and usable information- delivered to user IP address in seconds

N1XYZ DEPARTED: EGNR AT: 2011-05-16 16:36:26 FOB: 4693

G- XYZ Exceedance: N1 Overspeed Left EngineLIMIT: 99.5 OVER LIM SEC: 25 PEAK: 100.25

GMT: 16:36:05 Location: LAT: 53.184814 LONG: -2.9680176SAT: 14 PALT: 160 MACH: 0.24 IAS: 159

Real Time Event Notifications-concise & clear

G- XYZ Exceedance: N1 OverspeedLeft Engine

AFM LIMIT: 99.5OVER LIMIT:

25 sec

PEAK: 100.25

GMT: 16:36:05 Location:

LAT: 53.184814

LONG: -2.9680176

SAT: 14 PALT:160

MACH: 0.24

IAS: 159

Information Presentation Tools

Customizable for different subject matter experts

Quantified benefits of AFIRS

• Dispatch availability improvements

• In flight troubleshooting support

• Reduced unscheduled maintenance

• Extended time on wing for engines

• Accurate times and other data for operations

• Reduced fuel consumption and carbon emissions (3-5%)

• Tracking and communications in remote areas

• Savings pay for system and service in months (if data is used to change and improve SOPs)

What’s working?

TECHNOLOGY• Aircraft behavior is well understood

• Data sources are inherent in modern aircraft

• On-board data gatherers and processors

• Communications infrastructure

• Deployed technology for diagnostics, prognostics, reporting

PROCESSES TO UTILIZE DATA• Rules & SOPs exist (AFM, AMM, etc.)

• FOQA, FDM, SMS

BENEFITS / BUSINESS CASE PROVEN

What remains to be overcome? (IMHO)

PEOPLE and ORGANIZATIONAL INERTIA• Leadership – need visible change leadership before change

“management”. Must come from the very top

• Resistance to change

• Ignorance - of technology, benefits, and previous body of work

• Fragmented approach - this requires end-to-end integration

• Stovepipes and rice bowls – need incentives and consequences

• Role confusion - OEMs, operators, service providers

• Misguided competition – unbundling system fatal

• Policies - local, industry, national

Summary *

• Automated, reliable, verifiable, and accurate data collection, interpretation, and presentation is a reality in aviation today

• Infrastructure required is available, mature, and affordable

• Promise is proven but still unfulfilled

• Users must be prepared to change processes to take advantage of the information

• Early adopters are saving money, saving fuel, and saving lives

• An industry-wide policy that involves all stakeholders would help

* DISCLAIMER: Opinions expressed in this presentation are those of the lead author, not necessarily those of Dr. Chin, FLYHT, or MIT.

Your questions, please