SAFTEFAST Evidencebased Aviation Fatigue Risk Management Steven R

Major Fatigue Factors ● Time of Day: between midnight and 0600 hrs. ● Recent Sleep: less than eight hours in last 24 hrs. ● Continuous Hours Awake: more than 17 hours since last major sleep period. ● Cumulative Sleep Debt: more than eight hours accumulation since last full night of sleep (includes disrupted sleep). ● Time on Task/Work Load: continuous work time without a break or intensity of work Copyright 2008, IBR

An Objective Fatigue Metric ● No Blood Test for fatigue, yet ● The conditions that lead to fatigue are well known. ● A fatigue model simulates the specific conditions and determines if fatigue could be present. ● The model can estimate the level of degradation in performance and provide an estimate of schedule induced fatigue risk. Copyright 2008, IBR

SAFTE ● The Sleep, Activity, Fatigue, and Task Effectiveness (SAFTE) Model is based on 12 years of fatigue modeling experience. ● Validated against laboratory and simulator measures of fatigue. ● Validated and calibrated to predict accident risk by the Department of Transportation. ● Peer reviewed and found to have the least error of any available fatigue model. ● Accepted by the US DOD (Air Force, Army, Navy, Marines) as the common warfighter fatigue model. Copyright 2008, IBR

SAFTE Model Components Copyright 2008, IBR

Least Error for Conditions of Sleep Restriction 2002 Seattle Fatigue & Performance Modeling Workshop, of all models tested against laboratory measurements: ● SAFTE had least error predicting objective vigilance performance. ● SAFTE had least error predicting subjective ratings of fatigue. ● Advances since 2002 have further validated the sleep and performance assumptions and prediction of accident risk and severity. Von Dongen, Aviation, Space, and Environmental Medicine, March, 2004, vol. 75, no. 3, 6 section II Copyright 2008, IBR

Accuracy of Predicting Sleep Pattern and Duration Measure Signalman Maintenance of Way Dispatchers (less night workers) Train and Engine Mean Agreement 92% 90% 88% Daily Sleep (Est. - Log) -24 min -21 min -3 min - 10. 8 min 7 Copyright 2008, IBR

Economic Risk and Effectiveness Railroad Accident Relative Risk No Fatigue High Fatigue Copyright 2008,

Validated Fatigue Modeling Tools SAFTE/FAST Fatigue Science has the exclusive license from the US

Practical Software for Implementation ● Fatigue Avoidance Scheduling Tool (FAST) Fatigue assessment tool using the SAFTE model Developed for the US Air Force and the US Army. Features ● FAST DOT / FRA sponsored work has lead to Sleep estimation algorithm enhancements for transportation applications Graphical analysis tools Dashboard of fatigue factors Data based of all effectiveness scores Copyright 2008, IBR

FAST Aviation Sleep Estimation ● Accurate estimation of sleep is critical: Measure: actigraphy or log books Estimate: algorithm to simulate sleep behavior ● Aviation specific estimates that can be refined with actrigraphic measurement. ● Considers time zone changes and is valid for any city pair. Copyright 2008, IBR

Sleep Estimator Tailored to Aviation Environment: Copyright 2008, IBR

FAST Aviation Specific Auto. Sleep ● ● ● ● Mimics typical sleep patterns Tailored to workgroup and schedule demands Considers total duty period and commuting Naps prior to anticipated late starts Considers time zones Slits sleep when appropriate Automatically inserts in-flight sleep for augmented crews. User defined parameters: amount of augmentation and quality of sleep environment ● Adjustable settings can be saved to file 13 Copyright 2008, IBR

Fatigue Risk. Improvement Management. Process System Continuous Measure Define the situation Schedule evaluation Actigraph recordings Monitor Assess operational indicators Individual self-evaluation Feedback to process Model & Analyze FRM Steering Committee Involves all stakeholders at each stage: management, labor, aided by science Model the fatigue problem Analyze sources and Fatigue factors SAFTE/FAST Modify/Mitigate Shared Responsibility • Operating practices • Labor agreements • Individual “life style” Manage Collaborate for solutions Obtain commitment to solve problem Enablers Employee training Medical screening Economic analysis Technology aids Copyright 2008, IBR

FAST Aviation Fatigue Assessment Process Airline Specific Schedule Database XML format City-pairs/Trips or 30 -day Bids Aviation Auto. Sleep SAFTE Model Output results to folder Links to Manager Standard FAST schedule is created by FAST Aviation Modeler FAST Aviation Manager FAST Aviation Modeler • • Translation Tools available for any scheduling system • • Sorts by Criterion Displays results Links to Analyzer Fleet level reports FAST Analyzer Individual Schedules • • • Examine schedules Effectiveness Graph Fatigue Factors “What-If” Drills Individual reports Modular Process for Speed and Flexibility Copyright 2008, IBR

FAST Aviation Modeler Model the schedules: 1. Set Auto. Sleep parameters if necessary 2. Name the Output folder a unique name 3. Choose either a City Pairs file or a Bid Schedules file for modeling Airline Schedule Database Processed Schedules Modeling Results Copyright 2008, IBR

FAST Aviation Manager Sort by: v Flight Time Below Criterion Level (FTBCL) v Critical TBCL (30 min associated with takeoff/landing) v Average Effectiveness v Minimum Effectiveness overall v Minimum Effectiveness during critical periods v Maximum Workload (high workload score in schedule) v Median Workload (central score across schedule) Save table to text file Click on any line in Aviation Manager and schedule opens in FAST for detailed analysis. Copyright 2008, IBR

If fatigue is present, what do you do about it? ● Modeling tools must do more than give you a fatigue score: It must estimate fatigue risk It must show detail of each schedule It must calculate fatigue factors It must suggest conditions that lead to fatigue so mitigations can be implemented by an FRMS 18 Copyright 2008, IBR

Detailed Analysis Results ● Creates detailed database that shows: All duty periods and estimated sleep intervals Effectiveness in each half hour of each duty period Effectiveness at each half hour of the clock Distribution of duty time in effectiveness categories Allows results be sorted based on user defined categories Individual ID reports with effectiveness at the 1 19 min resolution Copyright 2008, IBR

FAST Aviation Analyzer San Francisco to Sydney Pairing 14. 5 hr flight Pre-flight nap Sleep Timing based on both physiological and social cues In-flight sleep Dashboa Fatigue Pre-flight nap Schedules in Aviation Manager link to FAST for detailed analysis. Copyright 2008, IBR

Dashboard Information Content based on fatigue analysis workshop hosted by NTSB and conducted by Drs. Mark Rosekind & David Dinges, funded by FRA Office of Safety. l Sleep (last 24 l l Flags are fatigue indicators l l l hrs) Chronic Sleep Debt Hours Awake Time of Day Out of Phase Performance Values l l l Value at point in schedule Criteria l l Effectiveness (vigilance) Mean Cognitive Lapse Index Reaction Time Reservoir Copyright 2008, IBR

Schedule Files for Evaluation A: 90 Short haul pairings, 1094 active flights B: 56 Short haul monthly rosters, 3963 active flights C: 47 Long haul pairings, 188 active flights D: 64 Long haul monthly rosters, 1006 active flights ● Translated the spread sheets using Access database into the required XML file structure. ● Batch processed through FAST Aviation ● Used FAST Manager to rank order Pairings and Rosters ● Used output spread sheet to rank order segments 22 Copyright 2008, IBR

Fatigue Metrics ● Typically, FAST is used to assess the “tail of the distribution” – how much critical duty time is spent at low effectiveness. ● For this exercise, we were asked to rank order all segments and schedules, not just the extreme cases. ● We rank ordered segments by minimum effectiveness at critical times of flight – take-offs and landings. ● We rank ordered pairings and rosters by minimum effectiveness and “critical time below criterion” 23 which is more useful for entire schedules. Copyright 2008, IBR

SAFTE/FAST - Segment Analysis A SH Pairings Rank 1 2 3 4 5 A 10040 A 10002 A 10045 A 10002 Start LAX LPB ATL VVI IQQ End Min Critical Time Minimum E Rank ATL IQQ SEA LPB SCL 5/15/2010 5: 55 2/4/2011 8: 15 5/14/2010 1: 30 2/4/2011 6: 30 2/4/2011 10: 10 5/15/2010 10: 08 2/4/2011 9: 19 5/14/2010 6: 57 2/4/2011 7: 44 2/4/2011 12: 24 70. 84 74. 64 74. 8 74. 99 75. 65 3 7 8 9 10 C LH Pairings 1 2 3 4 5 C 10027 ANC C 10014 PDX C 10033 DFW C 10025 BRU C 10030 LAX B SH Rosters LAX NRT BRU SHJ AMS 9/16/2010 15: 05 5/17/2010 21: 10 9/23/2010 18: 15 9/27/2010 16: 40 9/16/2010 23: 10 9/16/2010 20: 09 5/18/2010 7: 59 9/24/2010 3: 34 9/27/2010 23: 09 9/17/2010 9: 49 70. 32 70. 45 72. 06 73. 62 73. 81 1 2 4 5 6 1 2 3 4 5 B 10049 B 10046 B 10052 B 10049 B 10044 SCL SCL SCL 12/22/2010 5: 45 1/10/2011 5: 30 1/8/2011 3: 40 1/16/2011 2: 55 1/10/2011 4: 25 12/22/2010 8: 09 1/10/2011 7: 44 1/8/2011 5: 29 1/16/2011 5: 19 1/10/2011 6: 39 72. 59 72. 84 73. 66 74. 17 74. 39 6 7 8 9 10 1 2 3 4 5 D LH Rosters D 10029 GUM D 10032 LIM D 10024 SLC D 10052 HKG D 10024 SLC NRT 5/26/2010 5: 00 5/26/2010 8: 54 SCL 1/4/2011 6: 15 1/4/2011 9: 34 ANC 5/18/2010 3: 35 5/18/2010 8: 29 HEL 11/17/2010 17: 15 11/18/2010 4: 19 ANC 6/2/2010 3: 35 6/2/2010 8: 29 66. 82 67. 76 67. 78 68. 59 68. 88 1 2 3 4 5 ARI IQQ ANF ARI IQQ 24 Copyright 2008, IBR

SAFTE/FAST - Roster Analysis Rank A SH 1 2 3 4 5 Critical TBCL (77) TBCL Overall Rank 149 2 88 4 61 6 53 9 41 10 Pairings A 10002 A 10001 A 10008 A 10040 A 10000 C LH 1 2 3 4 5 Pairings C 10014 C 10033 C 10019 C 10038 C 10027 182 143 77 61 61 1 3 5 6 6 3 B SH 1 2 3 4 5 Rosters B 10049 B 10050 B 10046 B 10045 B 10047 201 194 126 124 123 1 2 4 5 6 1 D LH 1 2 3 4 5 Rosters D 10024 D 10043 D 10044 D 10022 D 10029 180 111 91 85 79 3 7 8 9 10 3 Minimum E Rank 2 4 5 1 8 4 7 1 2 6 2 8 5 9 6 16 25 1 Copyright 2008, IBR

Minimum Critical Time Effectiveness SEGMENTS 26 Copyright 2008, IBR

Short Haul Pairing A 10040 LAX to ATL Segment Early Start Daytime Rest Night

LH Pairing - C 10027 Anchorage-LAX Segment 86% reservoir 5 hrs 28 Copyright 2008,

LH Pairing - C 10014 – Narita Segment Narita 29 Copyright 2008, IBR

Short Haul Pairing - A 10002 Santa Cruz, Bolivia – Santiago, Chile 1. 5

A 10002, continued Santa Cruz, Bolivia – Santiago, Chile Possible Mitigation 3 hr Nap

Long Haul Pairing - C 10014 Honolulu to Salt Lake 93% Res 3: 15

C 10014 Honolulu to Salt Lake Altered Sleep Pattern 33 Copyright 2008, IBR

Long Haul Pairing C 10033 Kuala Lumpur Based Pilot 34 Copyright 2008, IBR

Greatest Time Below Criterion ROSTERS 35 Copyright 2008, IBR

Long Haul Roster - D 10029 11 Day - Closer Examination 36 Copyright 2008,

D 10029 Detroit– Narita- Guam Segment 37 Copyright 2008, IBR

Long Haul Roster D 10029 Guam – Narita Segment 38 Copyright 2008, IBR

Short Haul Roster - B 10049 92 Segments in 57 Days 25 th Ranking

B 10049 Early Start on 12/19 40 Copyright 2008, IBR

B 10049 Multiple Segments at Night on 12/21 starting 1730 to 0510 41 Copyright

B 10049 Multiple Segments at Night 42 Copyright 2008, IBR

Long Haul Roster - D 10024 Salt Lake-Anchorage-Minn 43 Copyright 2008, IBR

D 10024 Six Day Series Salt Lake-Anchorage-Minn 44 Copyright 2008, IBR

D 10024 Salt Lake-Anchorage-Minn Explanation Two Early Starts (0700 and 0500) Two Consecutive Night

Workload Factor ● According to the NTSB: “One factor that contributes to self-reported pilot fatigue, especially in short-haul flight operations, is the number of legs flown in a duty period. ”* ● The highest workload in a flight occurs at take-off and landing; increasing segments multiplies these high stress periods. ● FAST Aviation is the first fatigue assessment tool to provide an automated method to assess this source of fatigue. *NTSB Safety Recommendation, A-09 -61 through -66, August 7, 2009 Copyright 2008, IBR

Sample Workload Pattern Maximum te pa ssi di ad klo er v s o

B 10007 Top Ranking Workload 58 Segments in 12 Days 48 Copyright 2008, IBR

Advantages of Modeling Approach ● Validated model with history of outstanding performance under independent review. ● Explicit Sleep Estimator (Auto. Sleep) tailored to habits and policies of the airline. ● Aviation specific drivers of fatigue. Cognitive fatigue Workload related fatigue ● Analysis tools that lead to specific fatigue factors and mitigation approaches. ● Modular design can be tailored to customers needs. Copyright 2008, IBR

Fatigue Risk. Improvement Management. Process System Continuous Measure Define the situation Schedule evaluation Actigraph recordings Monitor Model & Analyze Assess operational indicators Model the fatigue problem Involves all Mitigations are Proportional to the Risk stakeholders Individual self-evaluation Analyze sources and at each stage: Feedback to process Evolutionary, management, Incremental Improvement. Fatigue factors labor, by science Responsive toaided Changing Circumstances Modify/Mitigate Shared Responsibility • Operating practices • Labor agreements • Individual “life style” Manage Collaborate for solutions Obtain commitment to solve problem Enablers Employee training Medical screening Economic analysis Technology aids Copyright 2008, IBR

D Fa iag tig no ue sis M ea M M od s ur el Fa ea es s in tig ur g & e u B e s& M ar M e Fa a rie od Ba s rs tig ur el rr e i i u ng er M e s& s e M Fa a B o tig sur de ar ue es lin rie g rs M & od Ba el rr in ie g rs Fatigue Risk Pyramid Ev en t Se qu e nc e Accidents & Incidents Fatigue Related Errors Subjective Awareness Job Performance Changes Level 3 Defense Employee sleep habits, traits, & conditions Level 2 Defense Work demands, schedules, and sleep opportunities Level 1 Defense Based on James Reason, “Managing the Risks of Organizational Accidents”, Figure 1. 6, Stages in the development and investigation of an organizational accident. Copyright 2008, IBR

If fatigue is present, what do you do about it? ● Modeling tools must do more than give you a fatigue score: It must estimate fatigue risk It must show detail of each schedule It must calculate fatigue factors It must provide context of conditions that lead to fatigue so mitigations can be implemented by an FRMS 53 Copyright 2008, IBR

Questions: Steven R. Hursh, Ph. D and Reid Blank Institutes for Behavior Resources 2104 Maryland Avenue Baltimore, MD 21218 (410) 752 -6080 shursh@ibrinc. org rblank@ibrinc. org Chris Hallman Baines Simmons America 17 Greenville St. , Suite 221 Newnan, GA 30263 (678) 343 -1635 Office (770) 251 -5654 Fax chris@bainessimmonsamericas. com www. safetyfromknowledge. com 54 Copyright 2008, IBR

FS Actigraph Data Processing Individual fatigue risk levels are amalgamated into a group report SAFTE evaluates the fatigue risk and effectiveness of each individual driver Personnel wear the actigraph that measures wrist movements Daily sleep/wake/work times are fed into the SAFTE risk evaluation model Wrist movements are recorded 24/7 and downloaded over the internet Downloaded data are converted to daily sleep/wake/work times Copyright 2008, IBR