Method for the Impr mprove veme ment of Aircraft Take ke-off - - PowerPoint PPT Presentation

Method for the Impr mprove veme ment of Aircraft Take ke-off Trajectory Simu mulation using Variability Analysis

Speaker: George Koudis ICRAT 2014 28th May 2014 1/24

Speaker background

• George Koudis
• 2nd year PhD student at Imperial College London
• Member of the Centre for Transport Studies
• Supervised by Dr. R. North, Dr. A. Majumdar,
• Dr. W. Schuster and Prof. W. Ochieng
• Researcher in the environmental impact of airport operations
• Working on the Sensor Network for Air Quality (SNAQ) at

LHR project with Prof. J. Polak, Dr. R. North and Dr. S. Hu

• Sponsored by the Lloyds Register Foundation

2/24

Sensor Network for Air Quality (SNAQ): Heathrow Project

• Consortium of 4 UK based universities and industry partners.
• Main project aim is to validate the use of sensor network for monitoring

local air quality (LAQ).

• Installed a network of air quality sensor nodes around London

Heathrow, which is being used as a case study.

3/24

Presentation outline

1. Introduction

• Identification of research area
• Outline aims and objectives

2. Review of state-of-the-art

• Current modelling approaches
• Factors affecting aircraft operations

3. Proposed methodology 4. Analysis

• Identify and quantify relationships between operations and factors

5. Application of analysis

• Modification of basic trajectory models informed by research

6. Summary and conclusions

• Potential modelling improvements and implications

4/24

Introduction

5/24

Introduction

5/24

Introduction

• Negative impacts of aircraft activity on local air quality (LAQ) around

airports increasingly constrains planning for airport growth.

• Aircraft movements in the landing and take-off (LTO) cycle are among the

most polluting.

• Aircraft emit substantial levels of NOX, CO, HC and PM10 (among others)

and therefore require regulation and management.

• To facilitate a reduction in impact on LAQ, emissions arising from

aircraft activity must be accurately and realistically represented.

5/24

Research issue to be addressed

• Aircraft emissions can have a considerable impact on LAQ.
• Aircraft LTO activities and consequent emissions require modelling at the
• perational level when analysing operational impacts.
• Current techniques operate at different levels of sophistication but do not

capture variability in activity.

• Variability may cause significant differences in the LTO trajectory and

consequent emissions.

• There is currently no method for modelling aircraft activity variability

with low computational expense and limited data availability.

6/24

Aims and objectives

• Can recorded, high-resolution data be used to enhance simple LTO

trajectory models and reflect actual aircraft operations?

• 1. Do the currently available emissions models of aircraft LTO activity

sufficiently represent actual activity?

• 2. Can the relationships between aircraft activity and the factors that

influence aircraft operations be identified and quantified?

• 3. Can knowledge regarding these relationships be used to modify basic

LTO trajectory models where limited data is available?

• This will result in an increased level of realism reflected by models

without high levels of data required and high computational expense.

7/24

ICAO modelling approaches

Approach Simple Advanced Sophisticated Aircraft and engine fit Aircraft group Aircraft type Aircraft specific Operational profile Standard LTO time in mode and thrust LTO time in mode, thrust per aircraft type LTO time in mode, thrust per specific aircraft Emission indices ICAO per group ICAO per movement ICAO per movement Spatial distribution Airport aggregate Airport aggregate Aircraft movement 8/24

Airport'opera*ons'

Time,'aircra1,'tail'number,''stand' ID,'runway'ID,'origin,'des*na*on'

Trajectories'

Fuel'flow,'x,'y,'z,'t'

Emissions'func*ons'

Thrust,'emission'inventory'

Aggrega*on'

Spa*al','temporal'aggrega*on'

General'procedure' Sophis*cated'approach' Simple'approach' Reference'cycle'and' spa*al'elements'

Recorded' aircra1'data' (QAR)' Simulated'e.g.' HIPERMTP' ICAO'EEDB' BFFM2' Interpola*on' Engine' assignment' database' ICAO'EEDB' modeMspecific' EI' Engine' assignment' database'

Emissions'inventory'

• n'spa*al'elements'with'

temporal'data'

Emissions'database'

spa*al'and'temporally' distributed'

General modelling approach

9/24

Proposed modelling approach

10/24

Airport'opera*ons'

Time,'aircra1,'tail'number,''stand' ID,'runway'ID,'origin,'des*na*on'

Trajectories'

Fuel'flow,'x,'y,'z,'t'

Emissions'func*ons'

Thrust,'emission'inventory'

Aggrega*on'

Spa*al','temporal'aggrega*on'

Recorded' aircra1'data' (QAR)' Simulated'e.g.' HIPERLTP' ICAO'EEDB' BFFM2' Interpola*on' Engine' assignment' database'

±'Variability'

=f(Aircra1'type,'TOW,' Meteorology,'etc.)''

Factors with impact on aircraft trajectory

• There are many factors that introduce variability in the LTO

trajectory: including operational, meteorological and human.

• Improving the level of understanding of the impact of these factors on

trajectory variability relies on high-resolution data availability.

• This can be either actual recorded or simulated data.
• Is it possible to validate generic data for use with multiple airports?

Operational

• Aircraft factors
• Airline operations
• Airport procedures
• Ageing
• Maintenance

Meteorological

• Wind direction
• Wind speed
• Temperature
• Humidity

Human

• Pilot preferences
• ATC decision

choices 11/24

Factors with impact on aircraft trajectory

Operational

• Aircraft factors
• Airline operations
• Airport procedures
• Ageing
• Maintenance

Meteorological

• Wind direction
• Wind speed
• Temperature
• Humidity

Human

• Pilot preferences
• ATC decision

choices 12/24

• There are many factors that introduce variability in the LTO

trajectory: including operational, meteorological and human.

• Improving the level of understanding of the impact of these factors on

trajectory variability relies on high-resolution data availability.

• This can be either actual recorded or simulated data.
• Is it possible to validate generic data for use with multiple airports?

Data availability

13/24

• Four datasets used:
• BOSS – documents all activity and provides general information

(8,331 flights)

• QAR – high-resolution (1-4Hz) aircraft data record containing

detailed activity data (17.9 mil. aircraft data rows)

• ICAO EEDB – Fuel flow and emission indices data for specific

engine types

• BUCHAir – Third party engine assignment information

Data availability (cont.)

Group Item Temporal Actual date/time Scheduled date/time Stand on/off time Spatial Type of activity (arrival or departure) Origin or destination Stand number Runway number Aircraft Information Airline operator Aircraft (type specific) Engine type Aircraft tail number

• Max. number of passengers
• Max. take-off weight (MTOW)

Group Item Units Temporal Date dd/mm/yy Time hh/mm/ss Time from engine start s Ground speed kts Spatial Latitude

• , ‘, “

Longitude

• , ‘, “

Pressure altitude ft Radio altitude ft Ambient Outside air temperature

• C

Total pressure hPa Engine information Fuel flow* kg/s Engine pressure ratio*

• Turbine gas temperature*
• Thrust*

% (of max) Aircraft information Flight phase

• APU usage
• BOSS data

QAR data

Data use

BOSS$

Trajectories$

Fuel$flow,$x,$y,$z,$t$

Emissions$func<ons$

Thrust,$emission$inventory$

Aggrega<on$

Spa<al$,$temporal$aggrega<on$

QAR$

ICAO$ EEDB$ BUCHAir$

±$Variability$

=f(AircraN$type,$TOW,$ Meteorology,$etc.)$$

14/24

Quality control

• Sampling frequency
• Erroneous values:
• Latitudes and longitudes outside boundaries of managed airspace
• Ground speed
• Elevation

Aircraft type Aircraft jet size Total records (pre-QC) Total records (post-QC) Percentage loss

A319

Small 1412 1398 1.0

A320

Small 1145 1122 2.0

A321

Small 425 416 2.1

A767

N/A 287 100.0

B77A

Medium 387 387 0.0

B772

Medium 67 67 0.0

B77W

Medium 55 54 1.8

B747

Large 470 453 3.6

Total 4248 3898 8.2

N.B. Aircraft jet size is for analysis purposes 15/24

Take-off roll trajectory

• Analysis of variability of the start, end and duration of take-off roll

trajectory is conducted in this research.

• The take-off roll is the phase of the LTO cycle which sees the highest

rate of NOX emissions produced as thrust ranges between 80-100%.

• Two cut-off points are defined for analysis purposes:

Cut-off point Criteria used for the data cut Start of take-off roll Thrust = 14% (ICAO idle thrust x2) & speed >100kts 60s later Wheels off Elevation > 10ft (above datum)

16/24

Take-off roll activity: aircraft type variable

Ordered by MTOW Ground speed at wheels off /kts 17/24

Take-off roll activity: aircraft type variable (cont.)

18/24

Variation in activity for nominally similar aircraft

• Blue boxplots refer to the ground speeds at wheels off of one A319

aircraft during the sample period (>40 events).

• Green boxplot shows the ground speeds at wheels off for all A319

activity during the sample period.

• Red boxplot shows the ground speeds at wheels off for all aircraft

activity during the sample period.

Ground speed at wheels off /kts

19/24

Length of roll - ground speed at wheels off variation

r = 0.853 Ground speed at wheels off /kts Length of take-off roll /m

Take-off roll activity: Start and end of roll location

• Box plot analysis of longitude at the start and end of take-off roll
• More variability in wheels off point that start of roll point

20/24 N.B. Not to scale

Impact of variation in aircraft take-off weight

• Fuel loading assumed to be the major contributor to variation in aircraft

take-off weight.

• Variability in distance between origin and destination airports used as a

proxy for variability in fuel loading.

• Fairly strong correlation found between both length of take-off roll and

ground speed at wheels off.

• Stronger for larger aircraft than smaller, this is assumed to be due to

reserve fuel loading forming a bigger proportion of shorter-haul (smaller aircraft) flights.

Aircraft jet size Length of take-

• ff roll

Ground speed at wheels off Small r = 0.273 r = 0.417 Medium r = 0.598 r = 0.565 Large r = 0.629 r = 0.664

21/24

Impact of variation in wind speed and direction

• No identifiable correlation between ground speed at wheels off and

wind speed/direction.

• Relationships may be obscured due to the low resolution of available

meteorological data.

22/24

Application to simple trajectory models

• The quantified values can be applied to simple trajectory models.
• The speed that a B747 wheels off occurs has a distribution with a

median of X knots, within the range of 0.86X knots and 1.29X knots.

• When considering the journey distance, the range of speeds at wheels
• ff is 0.86X – 1.04X if the distance is less than 5,000km, and 0.98X -

1.10X for journeys with a distance of greater than 10,000km.

• This has implications on the volume and spatial distribution of

emissions generated by the aircraft activity.

23/24

Application to simple trajectory models

• From BOSS data:
• Activity: Departure
• Aircraft: Boeing 747
• Stand/Runway: 501/27R
• Destination: ?

23/24

Application to simple trajectory models

23/24

• From BOSS data:
• Destination: Europe (<5,000km)

Application to simple trajectory models

23/24

• From BOSS data:
• Destination: USA (>10,000km)

Summary

• ICAO modelling approaches either lack sufficient realism or require an
• ften unattainable amount of input data to accurately reflect actual

aircraft activity.

• It is possible to identify relationships between aircraft activity and the

factors which impact it. This is demonstrated using different aircraft type and aircraft take-off weight. No relationship was identified between wind direction/speed and activity using the data available.

• The relationships can be quantified in a manner that allows the

modification of a simple trajectory model to incorporate a greater level

• f realism.
• Future work should be carried out to analyse the impact of additional

relationships, for different phases of the LTO cycle and to assess airport transferability.

24/24

Thank k you, , Any Any qu questions? estions?

Speaker: George Koudis Contact: gsk12@ic.ac.uk ICRAT 2014