You are on page 1of 33

ICAO/ASPA Regional Seminar

Mexico City, May 18-19

Safety implications of Fatigue Risk


Management Systems
A Case
C S
Study

Emma Romig, Boeing Commercial Airplanes


Tomas Klemets, Jeppesen Systems AB
Hans Eriksson,
Eriksson Jeppesen Systems AB
David Hellerstrom, Jeppesen Systems AB
Part I
The Evolution of Flight Time
Limitations
A bit of aviation history

787
S i l
Serial
Number
001

1928
Boeing
Model 40
Flight Duration and Range Over Time

25

range in 1000km
i 1000k B777 - 2005 –
23 hours
range in hours
20

15

10
Wright
Brothers -
1909
5 –
23 minutes

0
1900 1920 1940 1960 1980 2000 2020
Flight Duration and Range Over Time

25
The majority of
range in 1000km
i 1000k US Flight Time
range in hours Limits
20 developed:
1931 - 1942

15

10

0
1900 1920 1940 1960 1980 2000 2020
Background

„ Study funded by Boeing’s China Safety, Efficency,


and Capacity group for the Civil Aviation
Administration of China (CAAC)
„ Purpose
„ Compare FTL’s between regions in regard to
crew p
productivityy / efficiencyy and ability
y to
protect from fatigue
„ Develop and demonstrate methodology for
improving FTL’s
„ Assumptions
„ Using BAM (The Boeing Alertness Model), based
on the ”Three Process Model of Alertness”
− BAM iis currently
tl bbeing
i validated
lid t d
operationally
„ Only shorthaul production on three airline
networks
− (long haul and/or other networks may have
different results)
6
Part II
Applying science to crew scheduling
Building realistic schedules using crew
scheduling software

8
Alertness Models
Alertness Models represent the most S + C + U + other
practical method of applying current effects are summed to
predict alertness as a
fatigue/sleep science to crew scheduling:
function of prior work
and sleep history

S represent the
homeostatic effect of time
awake

S’ represents the recovery


effect associated with sleep

+U
C represents the effect of
the ~24hr circadian
U rhythms

WOCL Wi d
Window off Ci
Circadian
di
Low

9
The Boeing Alertness Model in Scheduling

Alertness is predicted on a 0-10,000 scale, with


0 being the least alert, and 10,000 being most
alert
10
BAM-alertness predictions for a single crew member

Layover
Sleep opportunity at
Flight home base

Predicted
Alertness
(Center of distr.)
distr ) Predicted
Scale 0-10000 Sleep

11
“Context” affects fatigue

5503 4433

5863 5165

Removed

Removed
6490
12
A fleet-wide assesment of alertness

L
Lowest
t Predicted
P di t d Alertness
Al t in
i Flight
Fli ht – Fleet
Fl t View
Vi –1M
Month
th Objective 3: Maintain
Productivity
Objective 2: Increase (Productivity)
Overall Alertness
(All Flights)

Objective 1: Increase
Alertness on Worst
Flights
(Worst Flights)

13
Part III
How well do today’s prescriptive rules
protect alertness?
p
The Scheduling System Can Be Used Analytically
3 FTL Conditions

3 Airline Networks

Analyze Productivity
Statistics of Solution
15
A Comparison of 3 FTLs in terms of productivity
9:36

BAM
7:55

FAR BAM
FAR BAM
7:12 6:43 JAR FAR 6:51
6:37
CCAR 6:28 CCAR 6:20 6:23 CCAR
5:59 6:01 6:07
JAR JAR
ock Hrs/Day
y

5:31

Desired T
5:24

4:48
Blo

Trend
2:24

0:00
CN US EU

Network
16
Differences in FTLs the effect productivity &
alertness
FAR
6:37 Some Differentiating Factors:
CCAR
5:59
JAR
• Minimum Rest 8 hrs
5:24

FARs
• Block Time Limits
• No effects from # of
flights or time of day

• Minimum Rest 10 hrs

Rs
“BAM”: • Block & Dutyy Time

CCAR
• No Minimum Rest Limits
Rule • Small effects from # of
• No Block/Duty flights or time of day
Limits
• Alertness > 1500 • Minimum Rest 10 hrs
JARs
• Incentive to • Block & Duty Time Limits
create alertness • Reduced duty limits
b
basedd on # off flights
fli ht and
d
CN WOCL

17
The FRM-based approach outperforms FTLs in
virtually all cases
9:36

BAM
7:55

FAR BAM
FAR BAM
7:12 6:43 JAR FAR 6:51
6:37
CCAR 6:28 CCAR 6:20 6:23 CCAR
5:59 6:01 6:07
JAR JAR
ock Hrs/Day
y

5:31

Desired T
5:24

4:48
Blo

Trend
2:24

0:00
CN US EU

Network
18
The Scheduling System Can Be Used Analytically
3 FTL Conditions

3 Airline Networks

Analyze Alertness
Statistics of Solution
19
Visualizing Fleet-wide Alertness

20
Average Predicted Alertness of Worst 5% of
Flights
3500
hts

BAM
st 5% of Fligh

2974
3000
BAM
2617
BAM
2500 2418
ess of Lowes

JAR
JAR
2100
CCAR 2043

Desired Trend
2000 1933 CCAR
1772
JAR
FAR 1566
FAR
Avg Prediicted Alertne

1475
1500 CCAR
1371
FAR 1309
1180

1000

500

0
CN US EU
Network
21
The FRM-based approach protects alertness
better than any of the FTL formulations
3500 Fixed Limit in BAM
@ 1500 – no flights
hts

BAM

below
st 5% of Fligh

2974
3000
BAM
2617
BAM
2500 2418
ess of Lowes

JAR
JAR
2100
CCAR 2043

Desired Trend
2000 1933 CCAR
1772
JAR
FAR 1566
FAR
Avg Prediicted Alertne

1475
1500 CCAR
1371
FAR 1309
1180

1000

500

0
CN US EU
Network
22
Visualizing Alertness and Productivity Together

Alertness
Desired Trend

Opportunity
for
Improvement

The “gap” between


represents
p optimal
p
solution for the network

Opportunity
Desired T

for
Improvement
Trend

Productivity
23
Part IV
How Flight Time Limits can be improved
Looking for improvements for the CCARs

Alertness
Desired Trend

Opportunity
for
Improvement

The “gap” between


represents
p optimal
p
solution for the network

Opportunity
Desired T

for
Improvement
Trend

Productivity
25
Looking for improvements for the CCARs

Alertness

CCAR
Desired Trend

R
Opportunity

BAM
for
Improvement

The “gap” between


represents
p optimal
p
solution for the network

Opportunity

BAM
Desired Trend

for
Improvement
CC
CAR
T

Productivity
26
Looking for improvements for the CCARs

Alertness

CCAR
Desired Trend

New
New

+P
Alertness eRule

CC
+A
New

+
+A
R
AlertnessNRRule
le

CAR +
Opportunity (A) New
Alertness Rule
(A)
Alertness Rule
for (A)
Improvement (A)

The “gap” between


represents
p optimal
p
solution for the network

Opportunity New
New
Desired Trend

for Productivity

CCAR +
New
Productivity
Improvement Rule
Productivity
Rule
CC

+P
((P)Rule
)

+A
CAR

P
T

(P)

A
+
+A
(P)
Productivity
27
An Example: Analysis of the CCAR+ Rules

Alertness
Desired Trend

CCAR CCAR+ BAM


Opportunity 1933 2185 2617
for
Improvement

The “gap” between


represents
p optimal
p
solution for the network

Opportunity
Desired T

for CCAR CCAR+ BAM


Improvement 5:59 6:21 6:28
Trend

Productivity
Productivity
28
Part V
The Path Forward: From FTLs to FRM
Some of the remaining challenges to FRMS

- Role of Modeling Æ Application


without abuse/misuse/etc.
- Setting Limits Æ Defining cut-offs
- Individual Differences Æ Coping
with differences between people,
lifestyle, culture, choices, etc.
-Complexity Æ Tracking alertness vs.
tracking duty/flight hours
-Labor Issues Æ How to separate
fatigue (and safety) issues from
quality-of-life

30
Moving forward

Refine
Collect Data M d l/FRMS
Model/FRMS
Validation / System
Maturity Achieved

FAA FTL NPRM

Operate FRMS-based
Flights Operations

Improve FTLs 31
Thank You

32
Thank you

Please feel email if interested in o


ourr white
hite paper
with full details of the analysis and improvement
methodology

Emma Romig – emma.romig@boeing.com


emma romig@boeing com
Tomas Klemets – tomas.klemets@jeppesen.com
David Hellerström - david.hellerstrom@jeppesen.com

33

You might also like