ATM_Seminar_2021_Presentation_pdf_59.pdf

Apresentação da melhor conferência do mundo sobre ATM da aviação.

1. Presenter: Tejas Puranik
Authors: Wenxin Zhang, Carter Tegen, Tejas Puranik,
David Anvid, Rukmini Roy, and Dimitri N. Mavris
Fusion and analysis of data sources for
assessing aircraft braking performance
on non-dry runways

2. ❑Background
▪Motivation and Context
▪Research Objective
❑Data Sources and Processing
▪ASOS
▪FICON
▪Data Fusion
❑Data Analysis Results
❑Conclusions and Next Steps
Outline
2

3. 3
Background

4. ❑On December 8, 2005, Southwest Airlines Flight 1248 overran runway 13C at
Chicago’s Midway Airport (MDW) after landing on a runway contaminated with snow
and slush.The Boeing 737-700 aircraft exited the end of the runway and went
through an airport perimeter fence; striking an automobile and resulting in a fatality
❑Takeoff and Landing Performance Assessment (TALPA) Aviation Rulemaking
Committee (ARC) produced significant changes to the way aircraft braking is
evaluated and operationally addressed
❑Subsequent data collected over many years thus available for analysis
Motivation and Context
1
Federal Aviation Administration, “Runway Overrun Prevention,” Advisory Circular 91-79A, Nov. 2007 4
There exists an opportunity to leverage large volumes of routinely collected data in order to enhance the
understanding of aircraft performance on dry and non-dry runways to further improve safety
Source:
https://www.skybrary.aero/index.php/File:B737_Chicago_Midway_081205_.jpg
❑Runway overruns during the landing phase of flight account for approximately 10 incidents or accidents every year
with varying degrees of severity, with many accidents resulting in fatalities
1
❑Conditions identified were the dynamics of a tailwind approachand landing and wet or contaminated runways that
lead to substantially increased landing field lengths
❑Current braking models are based on testing and data collected using aircraft with braking systems which are no
longer the standard (e.g., early generation Anti-skid systems)

5. The overarching objective of the work presented today is to quantitatively explore how different factors may
work to cause or prevent poor braking performance by fusing and analyzing multiple sources of data
relating to runway conditions, characteristics, and prevailing weather conditions and comparing these with
the pilot reported braking action.
Research Context and Overall Objective
Dr. Tejas Puranik 5
Overall Project
Flight Data
Analysis
FICON Data
Analysis
Ongoing work Focus of today’s talk

6. Dr. Tejas Puranik 6
Data Sources and Processing
▪Automated Surface Observing System (ASOS)
▪Field Condition Reporting (FICON)
▪Runway and Airport Characteristics
▪Data Fusion

7. ❑Methods for assessing runway conditions (Type, depth, coverage)
❑Reporting of braking action by pilots (PIREPS)
❑Reporting of runway conditions through airport operators, the NOTAM system, and ATC
agencies (FICONs, ATIS, and ASOS)
❑Airplane performance data (i.e., Landing Field length plus margin)
❑Before landing performance assessments (Crew Procedures)
❑Terms used in runway condition reports and performance data (RwyCC and RCAM)
TALPA Recommended Standardizations
Dr. Tejas Puranik 7

8. ❑Automated Surface Observing System (ASOS) units are
automated sensor suites that are designed to serve
meteorological and aviation observing needs
❑More than 900 ASOS sites exist in the United States,
mostly located at airports
❑The National Oceanic and Atmospheric Administration
(NOAA) provides an archive of one-minute interval
data from ASOS sites in the United States each month
▪The one-minute interval data was used as the weather data
source
▪The raw data from NOAA is not suitable for data analytics
purposes
❑We developed scripts to automatically parse the raw
data into a suitable format for the research
Automated Surface Observing System (ASOS)
Dr. Tejas Puranik 8
erroneous data
Sensor malfunction
or human errors
missing data
Missing data due to
possible power
outages or other
reasons

9. FICON –Field Condition
A specialized field report that may be included in FAA-issued
Notices to Airmen (NOTAMs). FICONs may contain:
❑Report on the condition and braking action for a runway
❑Contains a description of the runway with precipitation type
and depth
❑Runway Condition Code (RwyCC): A number from 1-6 (6
being dry and ideal braking conditions) assessing the quality
of the runway surface; reported for each third of the runway.
❑Information on Non runway movement surface conditions
and taxiing hazards (e.g., snowbanks)
❑One of the most significant TALPA recommendations was the
introduction of consistent method for assessing runway
conditions, known as the Runway Condition Assessment
Matrix (RCAM)
Field Condition Reporting (FICON)
Dr. Tejas Puranik 9
Runway Condition Assessment Matrix
1
1
Federal Aviation Administration, “Airport Field Condition Assessments and
Winter Operations Safety”, Advisory Circular 150/5200-30D

10. Data Structure
Dr. Tejas Puranik 10
FICONCancel DateEnd Date
Pilot Reported
BA
Reference
ID
RWY
ID
Start DateAirportRwyCC 1RwyCC 2RwyCC 3 Description Status
'1/1/1
2/24/2017
11:22
2/25/2017
10:35
POOR 4537942336
2/24/2017
10:35
EAR 1 1 1
75 PRCT ICE 150FT WID OBSERVED AT
1702241035
Cancelled
'1/1/1
2/22/2017
18:04
2/23/2017
16:25
NIL 4536164829
2/22/2017
16:25
PNA 1 1 1
100 PRCT ICE OBSERVED AT
1702221625
Cancelled
'1/1/1
2/22/2017
18:04
2/23/2017
16:23
NIL 4536161611
2/22/2017
16:23
PNA 1 1 1
100 PRCT ICE OBSERVED AT
1702221623
Cancelled
'1/1/1
2/23/2017
16:12
POOR 4536151428
2/22/2017
16:12
DUB 1 1 1
100 PRCT ICE OBSERVED AT
1702221612
Expired
'1/1/1
2/23/2017
16:08
4536146810
2/22/2017
16:08
DUB 1 1 1
100 PRCT ICE OBSERVED AT
1702221608
Expired
'1/1/1
2/20/2017
17:13
2/21/2017
15:12
NIL 4534353222
2/20/2017
15:12
DIJ 1 1 1
100 PRCT ICE OBSERVED AT
1702201512
Cancelled
'1/1/1
2/15/2017
16:39
POOR 4529785613
2/14/2017
16:39
CQM 1 1 1
90 PRCT ICE OBSERVED AT
1702141639
Expired
… … … … … … … … … … … … …
… … … … … … … … … … … … …

11. ❑The FAA provides public access to a repository with airport and runway data that can be
used for this research
❑The data repository covers all FAR 139 certified airports in the United States
❑This source has been selected due to its high reliability and expansive coverage
Runway and Airport Data
Dr. Tejas Puranik 11
Airport
Facilities
Airport
Runways
Airport
Remarks
Airport
Schedules
Used within the context
of this research
https://www.faa.gov/airports/airport_safety/airportdata_5010/

12. Data Fusion Process
Dr. Tejas Puranik 12
Process Output

13. Dr. Tejas Puranik 13
Analysis
▪Data Statistics and Distribution
▪Analysis of Subsets of Interest

14. Data Statistics and Distribution
14Dr. Tejas Puranik
683150total records (rows) and ~25 relevant columns
▪568791(83.26%) contain Runway Condition Codes (RwyCCs)
▪11899(1.74%) contain Pilot Reported Braking Action
(PIREPs)
▪9906(1.45%) contain both RwyCCs and PIREPs
Geographic Distribution
▪More concentration of FICON reports from wet and cold
areas, dryer and warmer areas tend to have higher FICON
values
▪Ideal field conditions (6/6/6) not reported in the data set
Seasonal Variation
▪Majority of FICON reports from winter months
14

15. Homogeneity in Runway Condition Codes
Dr. Tejas Puranik 15
❑Majority of FICONs report homogeneous conditions for
the entire runway (~98%).
▪Among these 5/5/5 and 3/3/3 were the most
common reports
❑Very small fraction contain non-homogeneous reporting
across thirds of the runway
❑During analyses where a single value is desired, we
defaulted to the worst of the 3 non-homogeneous values.

16. ❑Three descriptor categories
▪Runway Treatment: Modifications to the surface to
reduce standing water and hydroplaning potential
▪Runway Surface Type: Material used in the runway
construction. Asphalt, Asphalt/Concrete, and
Concrete
▪Runway Condition: Quality of the runway surface
and is an indication of proper maintenance by the
airport operator: Excellent, Good, Fair, Poor
❑Significant trend correlation between in the
type of runway treatment and history of
RwyCCs reported that is consistent with
expected correlation from literature
Subset with Runway Condition Codes (??????=������)
Dr. Tejas Puranik 16

17. Subset with Runway Condition Codes and Pilot Reports (??????=����)
❑Using current “time of arrival” best practices guidance,
one of the aircraft crew’s important sources of braking
action information is from the reported RwyCCs
❑Other static variables such as runway longitudinal slope,
polished/rutted wheel tracks, or runway lateral slope
might affect pilot reports
❑Expected trend of fewer GOOD / GOOD-MEDIUM /
MEDIUM PIREP reports for the lower RwyCCs is confirmed
❑1/1/1 Column suggests that RwyCC reports might be
conservative as 25% of PIREPs are MEDIUM or better
▪Could be examined or understood based on different variables
(contaminant or non-contaminant)
Dr. Tejas Puranik 17
??????=9580

18. Relationship with Contaminant Variables
Dr. Tejas Puranik 18
??????=1390 ??????=1194

19. Relationship with Non-Contaminant Variables
Dr. Tejas Puranik 19
More positive slopes (uphill) have a strong positive
relationship with “good” reported braking action
Like the previously seen plot for runway condition codes,
runway treatment also has an impact on the pilot reports

20. ❑A multivariate analysis using three of the
variables described so far was conducted
(treatment, slope, and runway condition code)
❑2-way and 3-way ANOVA tests were performed
to identify the significance of these variables
Multivariate Analysis
Dr. Tejas Puranik 20

21. ❑Existing data sets (FICON, NOTAMs, and ASOS) are manageable and
assessable for use in a longer timeframe analysis based on the methods
described in this work
❑Data cleaning process for obvious errors and outliers does not negate the
statistical importance of analysis on the remining data
❑General trends and correlations from the data analysis are consistent with
operator anecdotal experience and prior OEM physics-based models
Observations and Insights
21Dr. Tejas Puranik

22. Dr. Tejas Puranik 22
Conclusion

23. ❑Demonstrated a method for processing, cleaning, and fusing relevant data
sources for a quantitative braking assessment study
❑The relationship between runway surface conditions, airport and runway
characteristics, prevailing weather conditions, and pilot reported braking
action was studied over a large period using collected data
❑Provided data-driven substantiation of various trends that were based on
TALPA recommendations
Conclusion
Dr. Tejas Puranik 23

24. ❑The fused data sets lend themselves well to a supervised Machine Learning
study. Two different flavors are envisioned:
▪Given the weather and runway conditions, can we predict pilot reports and check whether they are
consistent with expectations
▪Conversely, from an airport perspective, if we use the pilot reports and prevailing conditions, can we
predict the runway condition code and is this more or less conservative than what has been assigned
❑An analysis beyond just the U.S. FICONs is desired although standardizing
the data collected might be a challenge
❑Next steps include analyzing the fused FICON data with flight data collected
from routine airline operations to gain insight into other aspects of
potentially degraded braking performance
Next Steps
Dr. Tejas Puranik 24

25. Acknowledgment
Dr. Tejas Puranik 25
The authors would like to express our gratitude to Somil Shah, Angela Campbell, Cliff
Johnson, Paul Giesman, Michel Hovan, and Raymond Zee for their feedback and support.
We would also like to acknowledge the Airport Technology Branch at the FAA Technical
Center for providing the FICON data and braking action reports used in this study. This
work was sponsored by the Federal Aviation Administration (FAA) via contract no. DTFACT-
14-D-00004 (task order 692M15-20-F-00641) to the Georgia Institute of Technology. The
views expressed in this paper are those of the authors and the information in this
research does not constitute FAA flight standards or FAA aircraft certification policy.

26. Dr. Tejas Puranik 26
Thank you. Questions?
Email: [email protected]