You are on page 1of 9

INTER-NOISE 2006

3-6 DECEMBER 2006


HONOLULU, HAWAII, USA

Time stamped aircraft noise prediction - replacing the ‘Average


Day’ with the ‘Composite Year’

David G Southgatea Stephen J Sedgwickb


Australian Government Department of Australian Government Department of
Transport and Regional Services Transport and Regional Services
GPO Box 594 GPO Box 594
Canberra, ACT, 2601 Canberra, ACT, 2607
Australia Australia

ABSTRACT

Aircraft noise predictions based on the annual average day are often treated with little respect
by members of the public. Averaging of information is commonly seen as a deliberate attempt
to hide the extremes of noise exposure and to avoid revealing what is happening at sensitive
times. There is a clear need to progress beyond the use of averaged aircraft noise information if
trust is to be built between airports and communities. The Australian Government Department
of Transport has put considerable efforts into developing concepts which use time stamped
data, rather than the average day, as the basis of aircraft noise description and assessment. This
work has led to the development of a tool called TNIP Runway Allocator which is used to
build time stamped data sets for use in aircraft noise prediction. These data sets contain
information, including runway used, for each individual movement at an airport in a given
future time period. Testing of this tool has shown that robust future ‘movement by movement’
data sets can be constructed for even complex airports. When required, time stamped data sets
can be reduced to an average day to produce statutory lines such as land use planning contours.

1 INTRODUCTION
Public acceptance of the noise assessments in an environmental impact statement (EIS)
depends very heavily on a good match being achieved between prediction and actual outcome.
Once a project is in operation, if there is a public perception that the noise outcome is
significantly different to that predicted in the EIS both the credibility of the environmental
assessment process and the technical competence of the noise expert are likely to be seriously
questioned.

When a new runway opened at Sydney Airport in late 1994 there was an immediate strong
adverse public reaction with widespread claims that the noise patterns arising from the operation
of the new runway were very different to those predicted in the EIS. This adverse reaction led the
Australian Federal Parliament to set up an Inquiry in mid 1995. The report of this Inquiry was
very critical of the aircraft noise assessment contained in the EIS for new runway project with
many aspects of the EIS noise analysis being criticized. One particular area that the public
claimed was misleading was the use of the ‘annual average day’ to describe aircraft noise [1]. In
essence, the perception was that a noise which fluctuates very widely, and unpredictably, from
hour to hour, day to day, season to season and year to year cannot be accurately described by
using annualized averages.

a
Email address: david.southgate@dotars.gov.au
b
Email address: stephen.sedgwick@dotars.gov.au
Over the latter half of the 1990s through working with the Sydney community new concepts
for describing aircraft noise emerged which avoid the use of the average day. These concepts use
aircraft movement data sets in which each movement is time stamped. The techniques used
facilitate the reporting of the temporal distribution of aircraft noise and enable the user to rapidly
ascertain, for example, annual, seasonal and daily variations in aircraft noise exposure and to
easily ascertain noise exposure at sensitive times such as evenings and weekends [2]. Until
recently, time stamped data sets had only been used in Australia to report past aircraft noise
exposure patterns. However, since their inception there have been requests for these data sets to
be used in the carrying out of aircraft noise assessments in EISs.

Given the imperative for the public to have confidence in aircraft noise predictions, and the
evident public distrust of aircraft noise assessments based on the average day, the Department
has put a significant effort into the development of aircraft noise description and assessment
techniques based on time stamped, rather than average day, data. In the early part of this decade
the Department developed a prototype software application which constructed time stamped
future aircraft movement data sets for use in aircraft noise assessments. These data sets were
constructed through cross matching future aircraft movement data sets with fine resolution wind
data sets for past periods. The Department reported on this initial work in 2004 [3]. Since that
time the ‘proof of concept’ software has been developed into a robust general model that can be
used to construct future time stamped aircraft movement data sets for airports with almost any
runway combination and/or traffic demand patterns. This paper reports on that work.

2 DEFINITION OF TERMS

2.1 ‘Average day’ data sets


An extract from a typical average day data set is
shown in Figure 1. This data set contains information on
the predicted traffic distribution between runways at the
study airport on a future day (commonly an ‘annual
average day’ when using metrics such as DNL, Lden,
ANEF, etc). Average day data sets of this form usually
underpin most conventional aircraft noise assessments.
These are typically constructed through a two stage
process. Firstly, trends in traffic growth are examined to
build a data set of predicted traffic movements at the Figure 1: Extract from average day data set
airport on the average day for a selected future year.
These movements are then allocated to runways by, in effect, computing the wind regime on the
average day using some form of wind rose technique - commonly averaging the wind at the
airport for the last 10 years. In this approach temporal information on both movement and wind
patterns is lost.
2.2 ‘Composite year’ data sets
The traffic predictions in ‘composite year’ data sets are built up in a similar manner to that in
average day data sets. However, rather than constructing a picture of the movements on a future
average day, an entire year of future movements is used with no averaging of the data. This
approach requires the storage of data on each individual predicted movement for every day at the
airport in the selected future year. Each of the aircraft movements in this data set is allocated to a
runway by relating each separate movement to time specific wind information extracted from
historic fine resolution wind data sets for the airport.
The composite year data set (example shown in Figure 2) is constructed from a single file
containing time stamped information on the predicted aircraft movements in the selected future
year. When the movement by movement allocations to runways have been made this file builds
up into a series of sub data sets containing runway allocated data for the past 10 years. Each of
the ten data sets will typically have between 200,000 to 500,000 lines for a medium to large
sized airport (the composite file size will be up to about 200 MB). Note that while each of the 10
separate files will have the same traffic distribution, operations with the same time stamp in each
of the files will not necessarily use the same runway due to the differences in wind patterns
between the years.

Figure 2: Composite year data set

3 TNIP RUNWAY ALLOCATOR – A TOOL TO GENERATE COMPOSITE YEAR DATA SETS

3.1 Background
Construction of a composite year data set would have been a very difficult task until recent
times due to the size of files involved. However, the significant advances in computing power in
the past few years have made this task easily manageable on a standard personal computer. Very
encouraging results from the Department’s 2004 proof of concept version of software designed
to construct composite year data sets led the authors to take the concept further and produce a
model which has general applicability at any airport. The
first version of this refined model – TNIP Runway Allocator
(TNIP RA) v1.0 – was released in early 2006 and is now
being progressively upgraded [4]. The program’s main menu
is shown in Figure 3.
Time stamped data sets can be built up without the
development of a generic model simply through carrying out
one-off analyses. This approach is currently being used by
the consultants carrying out the EIS noise assessments for a
new runway project at Brisbane Airport [5]. However, the
Department has an interest in developing tools that make
new aircraft noise analysis concepts accessible to as wide an
audience as possible and hence the focus has been on
producing a tool with general applicability. Figure 3: TNIP RA main menu
3.2 TNIP RA interface overview
TNIP RA lets a user interactively set up the operating parameters for an airport, load the
appropriate data, allocate aircraft movements to runways, examine and report on the allocation
results, carry out what-ifs and export files to other models for direct use in assessment of aircraft
noise exposure patterns. These options are displayed on the program’s main menu (Figure 3). All
of these functions are accessed through a series of user friendly input screens.
3.3 Setting up the airport
The set up area of the program enables the user to input a range of parameters which
establish the criteria which the program uses to make its runway allocation decisions. These
include specifying the wind limits for selecting runways, the operating capacity of individual
combinations of runways (termed runway modes) and the runway mode noise hierarchy for any
user defined period in the day.
After setting up the airport operating parameters, the files containing wind & aircraft
movement data and aircraft type and airport codes are loaded. These are simple .csv files and are
relatively small (eg about 600 kB for a 30 min resolution wind file for one year).
3.4 Allocating movements
Prior to allocation the user is required to enter certain allocation rules, in particular criteria
relating to the way Air Traffic Control (ATC) makes its runway use decisions. The allocation
process, during which the program progressively allocates each movement in the loaded aircraft
movement file to a runway, is relatively quick (eg for one major airport tested it takes
approximately 20 minutes to allocate more than 400,000 movements). The program can be set up
to automatically produce a number of allocation reports, such as that shown in Figure 5 (in the
next section), once the allocation process has been completed.
3.5 Examining allocations
The program contains comprehensive features for analysing the outcomes of the allocation
process. In particular the ‘Analyser’ (screen shot shown in Figure 4) allows the user to examine
the reason for each runway allocation decision. This is very useful when an airport is being
‘calibrated’ (ie when the airport’s existing operating parameters are being tested). The actual and
computed data sets can be searched and filtered using a comprehensive search tool. Allocation
reports, shown in Figure 5, can be rapidly produced for any filtered data set.

Figure 4: TNIP RA Analyser


When the output from an allocation run has been examined and reported/stored the user can
make rapid what-if assessments by changing one or a number of the allocation criteria (eg cross
wind or down wind limits) and re-running the allocation process.
3.6 Exporting runway allocated data
When the software is being used for aircraft noise assessments its key function is to produce
data sets which can be exported to either the Department’s noise information/analysis program
TNIP (Transparent Noise Information Package) [6] or the US Federal Aviation Administration’s
noise contouring program INM (Integrated Noise Model) [7]. The use of these data sets in noise
assessment is discussed in Section 5.

4 MODEL VALIDATION
Testing of the prototype of the general applicability version of TNIP RA was carried out
using data from the current operating configuration at Sydney Airport. The high operational
complexity combined with the ready availability of historic runway use and fine resolution wind
data made this airport an excellent case to use for model validation.
The model has been designed in a way that facilitates rapid and transparent validation testing.
The user is able to take an aircraft movement data set for a past time period, strip off the runways
that were actually used during that time period and then compute runway allocations using fine
resolution wind data sets for the period. When the program is in this mode a report showing both
the ‘actual’ and ‘computed’ runway allocations is produced. An example of such a cross
comparison is shown in Figure 5.

Figure 5: Allocation validation check. Data for Jan – Jun 2005 Sydney Airport
It can be seen that the validation test for this particular period gave a very close agreement
between actual and computed runway use. Similar results to this were achieved for Sydney
Airport data sets when tested across a range of time periods with widely differing wind regimes.
This high level of agreement had not been expected given the complexity of the task and the
demonstrated history of difficulty in predicting runway use at Sydney Airport [8].
As the next step in the testing/validating process the software was set up for Brisbane
Airport. This is operationally a much simpler airport and the authors were therefore very
surprised when the initial testing resulted in little correlation between the actual and computed
runway allocations. However, it soon became apparent that the discrepancy in the model’s
performance between the two airports was due to differences in the way in which air traffic
control (ATC) makes runway use decisions at the two airports.
At Sydney, due to special circumstances surrounding the way the airport is required to
operate in order to implement a noise sharing regime, ATC makes its runway use decisions in
what is termed an ‘active’ manner. That is, ATC will change the runways in use at any time if
the operating conditions (usually determined by wind and traffic conditions) allow different
runways with a higher noise preference to be used. This behaviour differs at other airports,
including Brisbane, where runway use decisions are made ‘passively’. In this case ATC only
changes runways when the current runway combination can no longer be safely used (eg due to
wind changes or increases in traffic). When the active/passive ATC behaviour was factored into
the model’s decision making process the runway allocation performance achieved for Brisbane
was similar to that achieved for Sydney.
To date the model has been tested for three overseas and four Australian airports ranging in
size from major international airports to relatively small regional airports. An agreement between
actual and computed runway use of about ± 2% has been achieved at all these airports. These
results indicate that the model can generate robust predictions of how an airport will operate
under given wind and traffic conditions and give confidence that composite years developed with
TNIP RA can be used to positive effect in aircraft noise assessments.

5 USING COMPOSITE YEAR DATA SETS IN AIRCRAFT NOISE ASSESSMENT


One of the key goals in aircraft noise assessment is to provide decision makers and the public
with a picture of future noise which they can understand and which ultimately provides a good
match between ‘expectations’ and ‘outcome’ when a new project comes on stream. The
opportunity provided by composite year data sets to go beyond the average day is a major
advance toward achieving this end.
5.1 Flight path based descriptors
Experience in Australia has shown that flight path based aircraft noise descriptors, derived
from time stamped data sets, give the interested parties a good appreciation of aircraft noise
exposure patterns. Aircraft noise assessments based on time stamped data sets provide a great
deal of information that is automatically lost when assessments are based on the conventional
average day data sets. For example, data on annual variations in aircraft noise is lost during the
averaging process to build up the conventional average day data set. In a similar vein, differences
in noise exposure patterns between seasons, months, days of the week and even times of day can
be readily revealed when composite year techniques are adopted. The Department’s software
package, TNIP, has been designed to facilitate rapid interrogation and comprehensible reporting
from composite year data sets.
Composite year techniques were recently applied by a consultant engaged to assess the future
noise exposure patterns around a regional airport in Australia [9]. This application, involving the
use of TNIP RA, integrated with other elements of TNIP, provided a package that enables a
person to select the wind regime for a past year and interactively see what the movements on the
flight paths around the airport would be in the future year under the wind regime that occurred in
the selected past year. This is illustrated in Figure 6. Informal feedback from the manager of the
airport indicated that this application was very positively received by the local community [10].
Figure 6: Example TNIP RA noise assessment application

5.2 Noise contour based descriptors


If desired, the data in a composite year data set can readily be reduced to an annual average
day. This may be necessary, for example, when developing statutory noise contours, based on
logarithmic annual average day metrics such as the DNL, to define areas where houses are
eligible for acoustic insulation.
While the form of the ‘average day’ data set derived from a composite year data set will be
the same as that generated by conventional wind rose techniques, our experience is that deriving
the average day data set from a time stamped data set, rather than through wind rose techniques,
is likely to generate much more robust runway allocations. The time stamped approach also
facilitates the rapid generation of ‘what-if’ average days.
Composite year data sets are particularly suited to the generation of partial noise contours
based on an arithmetic aircraft noise descriptor such as the N70. For example, average day
movement data sets can rapidly be generated for say summer and winter based over the 10 years
captured in the composite year data set. These can be exported to INM to allow the generation of
separate noise grids which can be displayed say as difference grids to illustrate the likely
seasonal variation in noise at any given point on the ground in the vicinity of the airport. This
concept is illustrated in the notional difference grid shown in Figure 7.
Figure 7: Notional summer/winter N70 comparison

6 CONCLUSIONS
Aircraft noise assessments based on average day data sets can only provide a limited picture
of future aircraft noise since the temporal information needed to give a good match between
‘expectation’ and ‘outcome’ is lost in the averaging process.
Recent advances in computing power now make it practical to generate, and manipulate,
disaggregated time stamped aircraft movement data sets. Work by the Department has indicated
that robust data sets of this form can be readily constructed through movement by movement
aircraft/wind cross matching techniques.
The use of composite year data sets opens up a range of options for describing and assessing
aircraft noise which provide a much greater level of detail than can be achieved using
conventional average day techniques. The temporal information provided by these approaches
has the potential to facilitate a greater level of public confidence in aircraft noise predictions than
is currently being achieved using average day techniques.
The use of composite year data sets does not preclude the generation of conventional noise
contours. All the information in an average day data set can be rapidly extracted from a
composite year data set.
7 REFERENCES
[1] “Falling on Deaf Ears?”, Report of the Senate Select Committee on Aircraft Noise in
Sydney, paragraphs 8.45 - 8.48, Commonwealth of Australia, Canberra, (1995)
[2] “Expanding Ways to Describe and Assess Aircraft Noise”, Department of Transport and
Regional Services, Commonwealth of Australia, Canberra, (2000)
[3] D. Southgate and J Firth, “Improving the accuracy of runway allocation in aircraft noise
prediction,” Proceedings of the Annual Conference of the Australian Acoustical Society,
(2004).
[4] TNIP Runway Allocator, Department of Transport and Regional Services, Commonwealth
of Australia, Canberra; http://www.dotars.gov.au/aviation/environmental/transparent_noise/tnip_ra.aspx
[5] Brisbane Airport New Parallel Runway Project, Brisbane Airport Limited, Brisbane, 2006
http://www.newparallelrunway.com.au/content/home.asp
[6] TNIP (Transparent Noise Information Package), Department of Transport and Regional
Services, Commonwealth of Australia, Canberra,
http://www.dotars.gov.au/aviation/environmental/transparent_noise/tnip.aspx
[7] INM (The Integrated Noise Model), United States Federal Aviation Administration,
Washington; http://www.aee.faa.gov/Noise/inm/
[8] Op cit [1], paragraphs 7.34 – 7.50
[9] “Scheduled Jet Flight Path Movements Charts – TNIP”, Sunshine Coast Airport, Airbiz
Aviation Strategies Pty Ltd, Melbourne, Australia (2006)
[10] Personal communication with Peter Pallot, Airport Manager Sunshine Coast Airport,
Queensland, Australia (2006)

You might also like