Giulio Di Gravio 1 , Maurizio Mancini 2 , iarca 1 , Francesco Costantino 1 Riccardo Patr tria 1 Department of Mechanical and Aerospace Engineering 2 ENAV s.p.a. Operations Directorate Fourth Fou h SESAR R Inno nnovatio tion Days th November 27 th 25 25-27 r 2014 2014 ATM sa safety management: rea eactive an and pr proactive ind indicators Forecasting and monitoring ATM overall safety performance
WHE HERE DO DO WE E STARTED (i (i kno know you you kno know it it) ESARRs EUROCONTROL is developing a ha harm rmonised framework rk for or the he sa safety re regu gula latio ion of of ATM, for implementation by States. Its core is represented by ESARRs Safety Oversight in ATM Reporting and Assessment of Safety Occurrences in ATM. In Italy ENAV s.p.a. follows ESARRs. Use of Safety Management System by ATM Service Providers. We analysed the safety events’ reported data collected from 2008. Risk Assessment and Mitigation in ATM. ATM Services’ Personnel Software in ATM Systems Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 2
WHE HERE DO DO WE E STARTED (al also thes hese thi hings …) ICEB EBERG OF OF SA SAFETY REASON SWISS CHEE REA EESE MOD MODEL THESE MODELS are • Clear • Evocative • Easy to o un understand BUT … HOW TO O APPLY THEM FOR EVALUATING THE GLOBAL SA SAFETY LEVELS OF ATM? Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 3
WHE HERE WE E WOUL ULD LI LIKE TO GO GO (no now you you kno know al also it! it!) ESARRs’ reporting structure + ENAV s.p.a. reporting database permit the construction of SMARTER and HOLISTIC INDICATORS ATM related safety occurrences NUMBER OF STANDARD APF INDICATORS REPORTING DATABASE OCCURRENCES INDICATORS ATM - specific Accidents Incidents occurrences Mid Air Potential for Inability to Collision Near Failure of collision or provide CFIT Collision ATM service near collision ATMS Collision Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 4
RE REACT CTIV IVE INDIC ICATORS (stepwise saf afety 1/ 1/2) AER EROSPACE PE PERFORMANCE FACTOR • Determine the organizational factors that influence performance (ESA SARR 4) • Determine the information available on those factors (Litera rature re analy nalysis is) • Organize the influencing factors (Accid cidents/ s/In Inci cidents/Is Issues) • Determine the relative importance of the factors (AHP AHP) • Display information for decision makers (Sa Safe fety Ind ndex) ANALYTIC IC HIE HIERACHY PR PROCESS • Structure the problem (Min Mind Ma Map) • Construct a set of pairwise comparison matrices ( Subject matter experts’ evaluations ) • Use the comparison to obtain the weights (FAA: Rob Robust Mod Model) 𝐹𝑤𝑓𝑜𝑢 𝑗 𝐵𝑄𝐺 𝑇𝑏𝑔𝑓𝑢𝑧 𝐽𝑜𝑒𝑓𝑦 = 𝐹𝑤𝑓𝑜𝑢 𝑗 𝑏𝑜𝑜𝑣𝑏𝑚 𝑑𝑝𝑣𝑜𝑢 𝑈𝑃𝑈𝐵𝑀 𝑢𝑠𝑏𝑔𝑔𝑗𝑑 𝑑𝑝𝑣𝑜𝑢 𝐹𝑤𝑓𝑜𝑢 𝑗 𝐵𝐼𝑄 𝑥𝑓𝑗ℎ𝑢 𝒐 𝑩𝑸𝑮 𝑻𝒃𝒈𝒇𝒖𝒛 𝑱𝒐𝒆𝒇𝒚 𝒌 = 𝑭𝒘𝒇𝒐𝒖 𝒋 𝑩𝑸𝑮 𝑻𝒃𝒈𝒇𝒖𝒛 𝑱𝒐𝒆𝒇𝒚 𝒋 Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 5
REACT RE CTIV IVE INDIC ICATORS (stepwise saf afety 2/ 2/2) We have buil ilt 4 Indexes: EN EN-ROUTE AIRP RPORT SAFETY INDEX 1 APT SAFETY INDEX 1 ENR SAFETY INDEX 2 (ATM) APT SAFETY INDEX 2 (ATM) ENR SAFETY INDEX 1 APT Accidents ( 0,5 ) Events ( 0,35 ) Issues ( 0,15 ) Near collision (0,752) Procedural (0,375) • INSA + INSN Accidents (1,000) • ATO • NCFIT • MA • RINA + RINN (AAY) • EME • SMI • PRI • TWI (AAY) External (0,176) Potential for collision or • BS near collision (0,107) • LASER • TRA • DATC / DATS • WS • LBS • OTH • PSMI • REX • RINA+RINN (AAN) Communication (0,448) • TWI (AAN) • UPA • SCS • CSC • PLCC System failure (0,142) • AIS • ASP • MET Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 6
RE REACT CTIV IVE vs PR PRO-ACT CTIVE INDIC ICATORS (no (no cla clash, , bo both ar are winn nner ers) We obtained RE REACT CTIV IVE safety ana nalysis based on: • ENAV s. s.p.a. p.a. sa safe fety da database • APF Sa Safe fety Ind ndexes We aim to obtain PR PRO-ACT CTIV IVE safety analysis Safety REACTIVE PRO-ACTIVE Index SAFETY SAFETY ? 2011 2012 2013 2014 Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 7
PR PRO-ACT CTIVE SAFE FETY: 3 3 strategies (al all 3 3 roa oads lead ead to o Rome saf afety) SAFETY MONITORING MONTECARLO Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 8
HISTORIC HI IC FI FIT T (as as thi hings di did, , thi hings will do do) • Try to fit various probability distribution to data • Select, for each event type the BEST BEST distribution The BEST T one ne according to the AIC rank that considers the prin inciple of par arsim imony and has a good accu curac acy/ea ease e rati tio in case of many input data \ TRA gen-08 12 feb-08 8 mar-08 5 apr-08 4 mag-08 10 DIST STRIBUTION FI FIT giu-08 11 lug-08 1 Probability Relative Frequency ago-08 17 FOR EA EACH EVEN ENT set-08 13 ott-08 3 TYPE nov-08 5 dic-08 7 gen-09 5 feb-09 2 mar-09 12 apr-09 8 mag-09 20 giu-09 23 lug-09 3 ago-09 9 set-09 24 APF F PROCESS ott-09 7 nov-09 5 dic-09 9 gen-10 7 feb-10 5 mar-10 6 apr-10 6 TRA Number of Occurrences mag-10 13 giu-10 17 lug-10 12 ago-10 8 set-10 11 ott-10 10 nov-10 15 dic-10 10 Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 9
TI TIME SER SERIE IES ANALYSIS S (fore orecast with h mat maths) mean, μ volatility parameter, σ • Based on Yule’s and Box and Jenkins’ theories a linea ear fi filter has been built for each event type, according to its characteristics (AR, MA, ARMA, ARIMA, etc.) HISTORIC FIT (as things did, things will do) • Analyzing (transformation, trends, seasonality): TIME E SERIES TRA Safety Index 1 ENR Contribution ANAL NALYSI SIS S FOR EA EACH EVEN ENT TYPE EVENT TYPE HISTORIC SERIES APF F PROCESS Time [months] Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 10
CA CAUS USAL L FI FIT T (roots ar are e fr frui uits 1/ 1/2) HISTORIC HISTORIC DATA DISTRIBUTION Human YES: NO: Δ < ε Noise Equipment Procedure Factor STOP ITERATE MIXTURE MODEL Weighted CAUSAL Weighted Weighted Weighted Human DISTRIBUTION Procedure Noise Equipment Factor Human Equipment Procedure Noise Factor Weight Weight Weight Weight CAUSAL ANALYSIS ITERATION Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 11
CAUS CA USAL L FI FIT T (roots ar are e fr frui uits 2/ 2/2) Human Factor Procedures 5% 10% Distribution Distribution Causes Causal Factor mean effect Noise 0,05 12,9685 0,6484 Human Factor 0,0475 8,7573 0,4159 Equipment 0,8075 10,2097 8,2444 Procedures 0,095 10,2097 0,9699 Equipment 85% CAU CAUSAL AL ANAL NALYSI SIS S FOR EACH EVEN EAC ENT TYPE Probability Relative Frequency COMPAR ARED WITH HIST STORIC DATA EVENT TYPE HISTORIC SERIES APF F PROCESS TRA Safety Index 1 ENR Contribution Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 12
GL GLOBAL L RE RESULTS (t (the he bes best is is ye yet to come ome) 2010 2011 2008 2009 Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 13
GL GLOBAL L RE RESULTS (a (a vi view fr from af afar ar …) SAFE AFETY INDEXES LIK LIKE PAS PAST PE PERFORMANCE EVALUATORS Individuate cr criticalities by y X-R analysis based on Shew Sh ewart co control chart charts Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 14
GL GLOBAL L RE RESULTS (a (a bi bit clo closer …) CA CAUS USAL ANALY LYSIS IS vs vs RE REAL L DA DATA Probability Relative Frequency 3 3 meth ethodologies to FORECAST FO SAFE FETY INDEXES and thus, SAFE FETY LEV LEVELS LS TRA Safety Index 1 ENR Contribution CA CAUS USAL vs vs TI TIME SER SERIE IES vs vs RE REAL DA DATA 3.000 SAFETY INDEX 1 ENR SAFETY INDEX 1 ENR forecast 2.500 LOW limit Causal Fit 2.000 UP limit Causal Fit 1.500 1.000 Whi Which one one is is 0.500 0.000 the be the best st one one? Madrid, November 26 th 2014 ATM safety management: reactive and proactive indicators 15
Recommend
More recommend
