Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/235222456
CITATIONS READS
0 1,689
3 authors:
Elpidio Romano
Università Telematica Internazionale UNINE…
72 PUBLICATIONS 170 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Elpidio Romano
letting you access and read them immediately. Retrieved on: 26 September 2016
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
the surrounding area so, people who lives in that area are Authority, provides sufficient assurance for th general public
unconsciously exposed to aircraft accidents risk. Actually, to use the services without having to worry about doing their
local risk levels are higher than might be expected. In fact, own risk assessments of the transport infrastructure before
even if it is true that the accident per flight index is very low deciding how to travel [31].
(typically 1 per 106), statistics demonstrate that accidents Currently, were moving to a wide adoption of Safety
mostly happen during take-off and landing phases and hence, Management Systems (SMS) which need to undertake risk
close to the airport (Fig. 2). assessment activities, either qualitative or quantitative.
As an example in the UK, the Civil Aviation Authority
Taxi, Climb
load,
Take- Initial
(flaps Cruise Desc.
Initial Final
Land.
(CAA) describes the means of implementation of SMS by an
off climb appr. app.
parked up) aircraft operator [8]. Risk assessment is an essential part of
Accidents 5% 12% 5% 8% 6% 3% 7% 6% 45%
such a system, and CAP712 therefore includes a risk
Fatalities 0% 8% 14% 25% 12% 8% 13% 16% 2% tolerability matrix for use when quantifying risk.
Exposure = % of flight
time based on flight TABLE 1 - WORLD WIDE AIR TRAFFIC
duration of 1,5 hours
(BLN. PASSENGERS/KM, NO CHARTER FLIGHTS), ICAO, 2006.
Effective Expected
1% 1% 14% 57% 11% 12% 3% 1% Air Company
and Mean Annual
1995 2005 2006 2007 2008
Geographic Area Variation (%)
Fig. 2 - Statistical Summary of Commercial Jet Airplane Accidents, Boeing, Africa 49,9 84,8 5,4 90,7 96,4 101,9
2005.
Asia/Pacific 556,5 967,4 5,7 1036,1 1105,5 1176,3
Risk level around large airports are, in effect, of the same WORLD 2248,2 3719, 5,2 3947,8 4177,0 4411,2
order as those associated with participation in road traffic.
An increase in airport capacity usually involves changes to The discipline of risk assessment has been applied in the
runways layout, route structures and traffic distribution, which aircraft systems as required for aircraft certification under
in turn effect the risk level around the airport. For these FAR23, FAR25 in U.S.A and under EASA Certification
reasons third party risk becomes an important issue in airports Specification (CS)-23/25 in Europe. Techniques to
development. accomplish safety assessment are quoted by the SAE in their
In the late 90’s the world’s airline fleet consisted of more Aerospace Recommended Practice (ARP)-4761.
than 15.000 aircraft, flying a network of approximately 15 Flight Operations Risk Assessment System, known as
million km and serving nearly 10.000 airports. The sector FORAS [16], is a risk management tool to “encode” human
directly employed more than 3.3 million people, with over 1.4 knowledge about a type of risk. The FORAS methodology
million in USA (Table 1) [3]. About 12 billion people and 23 employs a fuzzy expert system to identify the factors which
million tonnes of freight are being moved annually. The have the greatest impact on overall risk.
freight part represents approximately one third of value of the A different approach has been adopted by Kawasaki [25]
world’s manufactured exports. who developed the Aviation Safety Risk Model (ASRM). This
A variety of international institutions, organisations and system is based upon the Human Factors Analysis &
agencies deal with forecasting future trends, including Classification System (HFACS) proposed in [35]. HFACS is a
International Civil Aviation Organization (ICAO) and classification scheme which has been developed to capture
International Air Transport Association (IATA). The airspace and analyze the different types of human error that may occur.
manufacturers such as Airbus Industry, Boeing and Rolls The framework draws on [27], in which was developed the so-
Royce also make projections. called “Swiss-cheese” model of accident causation. ASRM
Historically, when there has been relatively rapid growth in was originally developed for military use (US Naval
air transport, it has often been followed by a series of Aviation), but it has been used more widely within the
accidents. The occurrence of such events has stimulated the aviation industry. The ASRM uses Bayesian Belief Networks
introduction of technical and operational measures. As a to model the uncertainty within the model, using either data or
result, overall safety has improved over time. However the opinion of “experts”.
technical and technological limits are near to be reached and An additional technique has been adopted by Bazargan and
this is demonstrated by the slowdown of aviation safety Ross [5]. They used the proportional occurrence of causal
improvements in the last few decades [35]. A formal risk factors obtained from accidents reports, where fatalities or
assessment carried out by the service providers, with support serious injuries were reported. These information are then
from equipment suppliers, and accepted by a Regulatory combined with experts judgments on the relative importance
53
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
of the flight attributes using the Analytical Hierarchy Process the fact that there is risk if there exists a potential source of
(AHP) by Thomas Saaty. damage, or hazard. When an hazard exists (e.g. a system
The purpose of this paper is to describe and analyze safety which in certain conditions may cause undesired
issues in airports paying attention to the following aspects: consequences), safeguards are typically devised to prevent the
occurrence of such hazardous conditions and its associated
• a strategic approach to improve airport safety, which undesired consequences.
includes the use of failure and hazard analysis
techniques and fast time simulation modeling; TABLE 2 - LIKELIHOOD ESTIMATE GUIDE
LIKELIHOOD
• safety of land side operations; RATING
DEFINITION
• certification aspects. Is expected to occur in most circumstances.
A Almost Certain
(100/year)
To avoid subjective interpretations and to increase model Will probably occur in most circumstances.
B Likely
(10/year)
accuracy, however, risk information were colleted turning to
Might occur at some time.
data records and statistically analyzed. Results were used to C Possible
(1/year)
draw risk contour maps so to provide a guide to local and Could occur at some time.
D Unlikely
national authorities, to population who live around the airport (1/10 years)
and to airports operators. May occur only in exceptional circumstances.
E Rare
(1/100 years)
The paper is organized as follows: in Section II some
definitions will be provided about risk management and risk TABLE 3 - CONSEQUENCE ESTIMATE GUIDE
assessment specifying some terms used in the civil aviation CONSEQUENCE
DEFINITION
field; in Section III gathered data will be analyzed using some RATING
safety data records and arranged in some useful tables; in Risk impact would be negligible or no risk
1 Insignificant impact can be identified to community or
Section IV the probabilistic model will be presented and
business.
remarked. Risk impact would result in few consequences,
2 Minor such as minor disruption to community and/or
II. DEFINITIONS business, but of limited overall consequence.
Risk impact would result in some
A risk is “the combination of the probability, or frequency, 3 Moderate consequences, such as short-term disruption to
of occurrence of a defined hazard and the magnitude of the community and/or business.
consequences of the occurrence” [8]. Risk impact would result in serious
The combination of these parameters determinates a two 4 Major consequences, such as medium-term disruption
dimensional quantity. So, if the risk is to be reduced, it can be to community and/or business.
Risk impact would result in disastrous
either be done on the severity axis, on the likelihood axis, or
5 Catastrophic consequences, such as long-term disruption to
considering both of these dimensions. The last approach may community and/or business.
be considered the best one to risk reduction.
For natural hazards such as an earthquake, typically we However, the presence of an hazard does not suffice itself
cannot do anything to reduce the likelihood, but there is much to define a condition of risk. Indeed, there is the possibility
that can be done to reduce the consequences: special building that the hazard evolves from potential to actual damage. Thus,
regulations can be put in place and “earthquake kits” can be the notion of risk involves some kind of loss or damage that
pre-distributed to inhabitants. Alternatively, there is much that might be received and the uncertainty of its transformation in
can be done to reduce the chances of happening of a mid–air an actual loss or damage so:
collision of two aircraft: the air traffic control system and on-
board radars deal with monitoring and maintaining both Risk = Damage + Uncertainty
vertical and horizontal separation.
Generally speaking, risk assessment procedure aims [30]: This qualitative analysis is reflected in various dictionary
definitions of risk, such as: “possibility of loss or injury and
• to derive the likelihood value and the severity of the degree of probability of such loss”.
consequences value for each hazard; However, let x and p respectively be a given damage and
• to use gathered information as a means of prioritizing the probability of receiving such damage. From a quantitative
actions, for example i.e. which hazard is more time point of view, a measure of the associated risk R is:
consuming and should be tackled as first one?;
• to specify mitigating features as appropriate to each R = x⋅ p (1)
hazard;
• to predict the effectiveness of those features in
In practice, the perception of risk is such that the relevance
reducing the risk.
given to the damage x is far greater than that one given to its
probability of occurrence p so that (1) is slightly modified to:
A first, intuitive definition of the term “risk”, comes from
54
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
R = {( St , Pt , X t )} (3)
wheels, tires, and flaps. It also excludes bent III. SAFETY DATA RECORDS
aerodynamic fairings, dents in the aircraft skin, small The airport risk assessment includes a series of connected
punctures in the aircraft skin, ground damage to activity:
propeller blades, or damage to only a single engine;
• Fatal accident: An accident that results in fatal • events historical analysis;
injury; • accident frequencies determination;
• Fatal injury: An injury that results in death within 30 • magnitude and the risk evaluation.
days as a result of an accident;
• Serious injury: An injury sustained in an accident Information were acquired investigating:
that:
• aircraft accidents causes;
9 Requires hospitalization for more than 48 hours • accident location;
that begins within 7 days of the date of injury;
• accident consequences.
9 Results in a fracture of any bone (except simple
fractures of fingers, toes, or nose);
In terms of data used in this phase, one the most regular
9 Produces lacerations that result in severe
sources of accident data is the Statistical Summary of
hemorrhages or nerve, muscle, or tendon
Commercial Jet Accidents [6]. This data represents
damage;
information drawn from worldwide aircraft operations from
9 Involves injury to any internal organ;
1959 to 2003. It presents data on the types of accidents and
9 Involves second or third degree burns over 5%
the phase of flight in which they occurred. Similar reports are
or more of the body;
available from the UK CAA Safety Regulation Group, which
9 Involves verified exposure to infectious
has produced reports containing data drawn from accident and
substance or injurious radiation.
incident sources world-wide, in the Global Fatal Accident
Review CAP681 [7]. This report is significant in that it
Moreover, when building a large system from a number of
contains judgments of the causal factors of each fatal accident
smaller ones we find that many of the hazards arise from the
to aircraft >5.7 tonnes. It also contains circumstantial factors
intra-system interfaces [30].
which could have had a bearing on each accident.
When performing a risk assessment, then, we can start
So, accident data are obtained, when available, from
identifying those interfaces and the hazards arising from them.
government accident reports. Otherwise, information is
Where a system is made up of subsystems from different
solicited from operators, manufacturers, various government
suppliers their domains of influence also need to be
and private information services (Table 4).
considered. An airport has a lot of interfaces with outside
Such information is inferred by a historical analysis of the
world: air traffic control has radio and telephones, there are
events, making reference to:
navigational aids that communicate with aircraft (instrumental
landing systems), there are road/rail links, etc. We will
1. local files (ANSV);
consider only one airside interface, the runway (Fig. 4): which
2. world files (AAIB, AAIU; ATSB; NTSB; TSB, etc.).
is the interface between the air navigation system and the
ground handling area.
In order to structure a tool which could allow a brief, but
exhaustive, description of the analyzed data and could be a
useful support to record the first news of our investigations, a
report has been compiled (Fig. 5).
In this report the ID_NUMBER is the code of the analyzed
report whereas the field DATE AND HOUR indicates the date
and the time in which the accident has happened (in
conformity with the prescriptions of the ICAO Annex 13, it is
express in local or coordinated universal schedule UTC,
Universal Time Coordinated). LOCATION is the place in
which the accident is occurred and AIRCRAFT_ID is the
typology of aircraft interested by the accident (in our case
commercial airplanes such as: Boeing, Airbus, Mac Douglas).
The field CLASS indicates the class of the aircraft defined
in relationship to its maximum take-off weight (MTOW),
Fig. 4 – Perth airport runway identified by the following letters:
56
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
engine;
• B: aircrafts with MTOW < 6.750 Kg and two • run off: it is frequent in the case of long landing or
engines; aborted take-off;
• C: aircrafts with 6.750 kg < MTOW < 136.000 kg; • veer off: it is relative to an aircraft side off and can
• D: aircrafts with MTOW > 136000 Kg and more than happen both in take-off phase and landing. It can be
two engines. due to a high value of the transverse component of
the wind, to a mechanical breakdown, etc;
TABLE 4 – THE 15 WORST AVIATION DISASTERS (2007-1995), BOEING • short landing: it happens when an airplane touch the
STATISTICS
ground before the runway threshold. It is mainly due
Fatal Date Location Carrier Type to bad meteorological conditions;
349 11/12/1996 New Delhi, India Saudi / Kazastan B747 / Il76 • run incursion: it occurs both in take-off phase and
Islamic Revolution's
landing phase. It can concern both aircrafts or other
275 02/19/2003 Shahdad, Iran Il-76MD
Guards Co. vehicles.
265 11/12/2001 Queens, New York American Airlines A300
Buah Nabar, Garuda Indonesia
234 09/26/1997 A300
Indonesia Airlines
East Moriches, New
230 07/17/1996 Trans World Airlines B747
York
Off Nova Scotia,
229 09/02/1998 Swissair MD11
Canada
228 08/06/1997 Agana, Guam Korean Airlines B747
Essentially we refer to human factors, mechanical factors or allows a precise approach without any decision
environmental factors. Even if these factors are not height and a RVR between 200 and 50 meters;
interdependent, they can interact. The reduction in accidents • the pavement conditions during the accident. To
attributable to human errors is not as much as function of define aforesaid conditions we referred to the ICAO
interventions aimed at aircrew, as it is improvements made to terminology. The followings terms have been used:
the aircraft. After all, it is known that the opportunities for
human errors considerably grow up when a mechanical failure 9 damp, to point out that the surface showed
occurs. So, it would appear that many of the interventions changes of color because of the damp;
aimed at reducing the occurrence or consequence of human 9 wet, to point out that the surface is full water,
errors have not been as effective as those directed at but there is no puddles;
mechanical failures [35] 9 water patches, to point out that on the surface
However, mechanical and environmental factors are they were visible puddles;
obviously unchangeable in the brief period: there’s only the 9 flooded, to point out that on the surface they
possibility to act on human errors applying preventive were visible ample zones covered with water.
measures that aim to reduce the accidents.
In case of some ice on the runway the terms used are: rime
TABLE 5 - CAUSES OF EVENTS (%) or frost, dry snow, wet snow, slush, ice, compacted or ruled
Cause 1950s 1960s 1970s 1980s 1990s 2000s Total snow, frozen ruts or ridges.
Pilot Error 41 37 29 30 31 30 33
The EVENT_SCHETCH is a graphical representation of the
Pilot Error
11 17 15 16 19 19 16 accident in which the points discussed in the synthetic
(weather based )
Pilot Error description of the accident and some photographic
7 3 4 4 6 3 4
(mechanical based) documentation are reported.
Total Pilot Error 59 57 48 50 56 52 53
Other Human Errors 4 7 10 6 7 9 7
Weather 14 11 10 12 9 8 11 A. Experimental analysis: acquisition and elaboration data
Mechanical Failure 20 19 21 21 21 25 21 For each event the proposed report has been compiled.
Sabotage 3 4 9 10 7 6 7 Acquired data allowed to establish that the 46,4% of the 1.174
Other 0 2 2 1 1 0 1 commercial airplanes accidents concerned the airport (58,5%
concerning the RWY, 33,6% concerning the apron and 7,9%
As concern failures typology, the following classification concerning the TWY) and the 18,1% the approach paths.
can be made: As concern risk events typology: 40,1% are accident,
54,7% are incident and 5,2% are serious incident.
• active failures (errors or active drawbacks): errors or From these results emerges that during the taxing
drawbacks that have an immediate negative effect; maneuvers, from and for the runway, and those of standstill in
• latent failures: failures existing before the event. the terminal area, there aren’t human damages if we except for
fear or light injuries. On the other hand, the accidents during
A description of the features interested by the accident is the take-off or landing phases are characterized by an high
reported in the AIRPORT FEATURES INTERESTED field: percentage of injuries and deaths. Indeed, in this paper “apron
RWY, TWY, Apron or also the zone where the accident maneuvers” have not been considered.
occurred, as well as the state in which was found during the Investigating causes of those fatal aircraft accidents is
accident. Some interesting airport features about runways are: difficult because they generally stem from a complex system
of mutually dependent, sequential factors. These factors can
• a synthetic description of the geometric be classified in several ways. At first, according to the current
characteristics: length, width, longitudinal and state-of-knowledge, they can be categorized into:
transversal inclination, presence of stop way and his
dimensions, TORA (Take Off Run Available), TODA • known and avoidable;
(Take Off Distance Available), ASDA (Accelerate • unknown and unavoidable causes.
and Stop Distance Available), LDA (Landing
Distance Available), runway instrumentations, ILS The former should be considered conditionally in the sense
system for landing. Particularly, in relationship to the that immediately after an accident the real causes are seldom
runway visual range and to the decision height, the fully known but as the investigation progresses they become
ILS is divided in ILS of CAT I, it allows a precise known and avoidable. Then, with respect to accident type, the
approach till a height of decision of 60 meters and a main causes can conditionally be classified into:
RVR of the 800m, ILS of 2CAT II, it actually allows
a precise approach till a decision height of 30 meters • human errors;
and a RVR of the of 400 meters, ILS of CAT III, it • mechanical failures;
58
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
ground or aborted the take-off in point A (the center of the obtained and used with the probability scheme. For example
generic patch of the grid). on 11/07/2001 in Fiumicino Airport (Rome) an accident
interested a MD-11 airplane. Runway 16C dimensions were
9.850 ft (length) and 150 ft (width). Dividing these dimensions
by the "standard runway" dimensions it were obtained the
following values: Cx = 1,015, Cy= 1,000.
A
1) The proportional accident model
B
This model involves the statistical modeling of the
occurrence of air accidents over time. A Poisson process is
often deployed. Such a process is based on the following
assumptions:
Fig. 9 - Discretization of the interested area. Capodichino Airport, Naples • an event can occur randomly, at any time and in any
point of the space. Past aircraft accidents had these
Then we can assume: features: they occurred in a random way in different
countries of the world;
ni • the occurrence of an event in a given time or space
PrIi {B} = ⋅ Pr i {I }⋅ Pr i {A}⋅ Pr i {B / A} (4) interval, is independent on what happened in any
∑ ni other non-overlapping interval. Air accidents, except
very rare mid-air collisions, have occurred as the
where: series of independent events in time and space;
• the probability of an event occurring in a small
ni interval Δt is proportional to Δt and can be calculated
• is the percentage of the airplanes belonging by λ · Δt where λ is the mean rate of occurrence of the
∑ ni event. It is assumed constant and equal to 1/Ta, where
to the ith weight class related that land or take-off Ta is the average time interval between two
from the considered airport; consecutive events. The probability of two or more
ni occurrences in Δt is negligible.
• ⋅ Pr i {I } represents the proportion of aircraft
∑ ni In Poisson processes the time intervals between two
that crashed within the airport surrounding, that run- consecutive events is exponentially distributed, indicating
off the runway, etc…So, this term represents the non-memory properties: this means that future events do not
particular type of accident the aircraft may have; depend on the number of previous events or on the time in
• Pr i {A} is the probability that the airplane belonging which previous events happened. This would logically seem
to be the case with air accidents. Mathematically, let T be the
to the ith weight class touches the ground in the
random variable representing the time interval between any
landing phase or aborts the take-off in a point of a
two consecutive events. This variable is exponentially
specific patch of the grid represented by the point A;
distributed. The probability that no accident will occur in a
• Pr i {B / A} is the probability that the airplane time interval t is:
belonging to the ith weight class, departing from the
point A stops (or major debris stoop) in a point of a P (T > t ) ≅ P ( X t = 0 ) = e − λ ⋅t (5)
specific patch of the grid represented by the point B.
where, Xt is the number of air accidents in time interval t and λ
Adding, for each point B, representative of a particular
is the average accident rate. Similarly, the probability of the
patch of the grid, the probabilities determined with (4) and
occurrence of at least one event in time t is:
considering each weight class category, cumulative
probabilities are calculated. Finally, the above mentioned
isofrequency lines were obtained enveloping the points P (T ≤ t ) = 1 − P (T > t ) = P ( X t ≠ 0 ) = 1 − e − λ ⋅t (6)
characterized by the same cumulative probability.
Since examined data interested different runways (with The probabilistic assessment of accidents uses a sample of
different length and/or width), it was necessary to 101 accidents over the period 1995 - 2003. The distribution of
adimensionalize distances information. To such purpose we the time intervals between these events is shown in Fig. 10.
introduced the "standard runway" with a length of 10.000 ft A simple calculation provides an estimation of the average
and a width of 150 ft. Comparing these dimensions with the accident rate λ = 7.851 accidents per year.
real ones, the scale coefficients (Cx and Cy) have been
60
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
0,55
0,5
0,45
38 38 Empirical data
0,4
Theoretical data
0,35
0,3
0,25
0,2
14
0,15
0,1 6
4
0,05 0 1 0 0 0 0
0
15 45 75 105 135 165 195 225 255 285 315
The time between the air accidents (days)
61
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
⎧ Δx Δx ⎫
Pr i ( A) = Pr ⎨ x A − ≤ x ≤ xA + ⎬ =
⎩ 2 2⎭
Δx
xA + 1 ⎛ x − μ ⎞2 (10)
2 − ⎜
1 σ ⎟⎠
= ∫ Δx σ 2π ⋅ e ⎝
2
Fig. 15 – Stop point distribution for C weight class airplanes on x axis
xA −
2 For D weight class aircrafts accidents with a touchdown
point distance far than 2.000 ft, preliminary data distribution
3) Model formulation: Pri {B/A}, landing phase analysis suggest a normal approximation with μ = 8,05 and σ
Also in this case, to define a statistical model that could = 1,46.
allow determining the probability that an airplane of the ith Therefore the probability that an aircraft that touched the
weight class stops in point B, after it has touched the runway runway in point A stops in the point (xB; 0) is given by:
in point A, for each weight class of aircrafts and for each
touchdown zone, the stop distances were computed along the Δx
xB +
x and y axes in homogeneous intervals: e − x x α −1
2
Pr i {x B / A} = ∫Δx Γ(α ) dx (14)
xmax − xmin xB −
Δx = 2
k (11)
y − ymin for C weight class aircrafts and for D weight class aircrafts
Δy = max with stop distance lesser or equal to 2.000 ft. For D weight
k class aircrafts with stop distance higher than 2.000 ft, the same
probability is given as:
with:
Δx
xB + 1 ⎛ x−μ ⎞ 2
k = 1+ 3,3 log10 nij (12) 2 − ⎜ ⎟
Pr {x B / A} =
1
∫
i 2⎝ σ ⎠
e dx (15)
xB −
Δx σ 2π
where nij is the sample numerousness related to the ith weight 2
class and to the touchdown interval j.
So, as previously done, global distributions on x axis and y Likewise along the y axis, sample data are distributed as a
axis were divided in two sub-sets corresponding to C and D normal function with:
weight classes. For example Fig. 15 reports the stop point
distribution (airplane or major debris) for C weight class • μ = 0,856 and σ = 1,439 for C weight class aircrafts
airplanes. and touchdown point distance lesser or equal to 800
Statistical tests demonstrates that this distribution may be ft;
approximated by a standard Gamma function: • μ = 0,380 and σ = 1,434 for C weight class aircrafts
and touchdown point distance between 800 and
2.000 ft;
• μ = - 0,309 and σ = 2,408 for C weight class aircrafts
62
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
and touchdown point distance far than 2.000 ft; and touchdown point distance far than 4.500;
• μ = - 0,248 and σ = 2,321 for D weight class aircrafts • μ = 1,563 and σ = 3,898 for D weight class aircrafts
and touchdown point distance lesser or equal to and touchdown point distance lesser or equal to
2.000 ft; 4.500 ft;
• μ = - 0,44 and σ = 0,983 for D weight class aircrafts • μ = 0,167 and σ = 0,518 for D weight class aircrafts
and touchdown point distance far than 2.000 ft. and touchdown point distance far than 4.500 ft.
Therefore the probability that an aircraft that touched the and the probability function is as in (18).
runway in point A stops in the point (0; yB) is given by:
Δy 2
yB + 1 ⎛ y −μ ⎞
2 − ⎜ ⎟
Pr i {y B / A} =
1
∫
2⎝ σ ⎠ (18)
e dy
Δy σ 2π
yB −
2
Δx Δy 2
xB + yB + 1⎛ y−μ ⎞
2 − x α −1 2 − ⎜ ⎟
e x 1
Pri {B / A} = ∫Δx dx ⋅ ∫Δy e 2⎝ σ ⎠
dy (19)
Γ(α ) σ 2π
xB − yB − ∞
2 2
V. CONCLUSION
Risk reduction is one of the key objectives pursued by
transport safety policies. Risk assessment is an essential
process in defining policy for risk management. By
identifying the nature and the potential impacts on consumers
or employees, risk assessment can assist regulatory authorities
Fig. 18 - Isofrequency lines around the airport applying the model
and business organizations to determine what kind of actions
are needed
In fact in the aircraft accident probability model it is When building a large system (the airport) from a number
hypothesized a relation between the probability of finding the of smaller ones, we find that many of the hazards arise from
wreckage in the point B (P(B)) and the mix index traffic the intra-system interfaces.
k Data confirms that in the airports case the runway is the
( ni / ∑ ni ). most critical intra-system to deal with.
i =1 In this paper we proposed a probabilistic method to analyze
accident probability deduced by real data analysis which can
be coupled with any method to assess accident magnitude.
C. Evaluation of accidents consequences The model helps to determine the probability that an
There are many different ways in order to assess the airplane of the ith weight class, that touched the ground in a
consequences of an airplane accident, either qualitative or point of a specific grid patch, which is represented by its
quantitative, even if it is rather a difficult issue. We may center (A), will be found in another grid patch, represented by
64
INTERNATIONAL JOURNAL OF SYSTEMS APPLICATIONS, ENGINEERING & DEVELOPMENT
Issue 2, Volume 2, 2008
its center (B). The same approach could be used for an [22] Kanafani A., “The analysis of hazards and the hazards of analysis:
reflections on air traffic safety management”, Institute of Transportation
airplane taking-off from the runway. Studies, University of California, Berkeley, Working paper UCB-ITS-
In this way some isofrequency lines (Fig. 18) to identify WP-84-1, 1984.
area with higher accident probabilities can be drawn. [23] Kuhlmann A., Introduction to Safety Science. New York: Springer,
1981.
This model represent a quantitative tool that can be used
[24] Lazarick R., “Systematic Assessment of Airport Risk”, presented at the
both in airport designing (to acquire specific information 16th Annual Security Technology Symposium, FAA Aviation Security
about safety distances between support/civil structures and the R&D, Williamsburg, Virginia, June 2000.
runway, to analyze the impact of future airport re-layouts on [25] Luxhøj J.T., “Probabilistic Causal Analysis for System Safety Risk
Assessments in Commercial Air Transport”, in Proc. of the Workshop on
safety) and in airport operations management (to take Investigating and Reporting of Incidents and Accidents, Williamsburg,
decisions about incoming and departing air traffic mix, to Sep 2003.
support risk managers in the assessment phase). [26] Owen D., Air Accident Investigation: How Science Is Making Flying
Safer. Patrick Stephens Limited, Yeovil, 1998.
[27] Reason J., A human error, New York: Cambridge University Press,
REFERENCES 1990.
[1] AA.VV., “Method for assessing Third Party Risk Around Airports”, [28] Rosenberg, B., “Air safety: the state of art”, Aviation Week and Space
National Aerospace Laboratory, Netherlands, 2002, unpublished Technology pp.51-66, 1987.
[2] AA.VV., “Assessing risk and setting targets in transport safety [29] Space Project Management, Risk Management, European Cooperation
programmes”, presented at the European Transport Safety Council, for Space Standardization Requirements & Standards Division ECSS-M-
Brussels, 2003. 00-03B, Noordwijk, The Netherlands, 16 August 2004.
[3] Air Transport Action Group, “The Economic Benefits of Air Transport”, [30] Spriggs J., “Airport Risk assessment: Examples, Models and
Geneva, 1996, unpublished. Mitigations”, presented at the 10th Safety-critical Systems Symposium,
[4] Barnett A., “Air safety: end of the golden age?”, Chance: New Southampton, England, 2002.
Directions for Statistics and Computing, vol. 3, no 2, pp. 8 – 12, 1990. [31] Trbojevic V., “Linking risk analysis to safety management”, presented at
[5] Bazargan M., Ross D. L., “A Comparative Risk Measure for General the 7th International Conference on Probabilistic Safety Assessment and
Aviation”, presented at the 17th International Conference on Multiple Management, Berlin, 2004.
Criteria Decision Analysis, Whistler, Canada, Aug 6-11, 2004. [32] Tocchetti. A., Infrastrutture ed impianti aeroportuali. Napoli: Guido
[6] Boeing, “Statistical Summary of Commercial Jet Airplane Accidents Angeli Editore, 1983.
Worldwide Operations 1959-2003”, Seattle, May 2004. [33] Wagenmakers J.H., “A review of transport airplane performance
[7] CAA Safety Regulation Group, “Global Fatal Accident Review 1980- requirements might bene t safety”, Flight Safety Digest, vol.19, pp.1-14,
1996”. Civil Aviation Authority Rep. CAP 681, Gatwick, 1998. 2000.
[8] CAA Safety Regulation Group, “Safety Management Systems for [34] Wiegmann D.A., Shappell S.A.. “Applying Reason: The Human Factors
Commercial Air Transport Operations”, Civil Aviation Authority Rep. Analysis and Classification System (HFACS)”, Human Factors and
CAP 712, Gatwick, 2002. Aerospace Safety, vol.1, no.1, pp. 59-86, 2001.
[9] Cole J., “Overview of aviation safety issues”, presented at the 7th Annual [35] Wiegmann D. A., Shappell S. A., A human error approach to aviation
Aviation Forecast Conference, National Air Traffic Controllers accident analysis: The Human Factors Analysis and Classification
Association, Washington, 1997. System. London: Ashgate Publishing, 2003.
[10] Corrie S.J., “Potential growth in air travel demands renewed effort to [36] Zanelli S., Affidabilità e sicurezza nell’industria di processo. Pisa:
improve safety record”, International Civil Aviation Organization Zanichelli editore, 1998.
Journal, vol.7, no.9, pp. 7-9, December 1994. [37] Huan-Jyh Shyur (2008) A quantitative model for aviation safety risk
[11] Dose A., “Safety is best served by paying close attention to the key assessment, Computers and Industrial Engineering Volume 54, Issue 1
elements in its management”, International Civil Aviation Organization (February 2008).
Journal, vol.20, no.21, 1995. [38] Ginalber luiz de Oliveira Serra (2007), Proposal of Numerically Robust
[12] Edkins G.D., “The Indicate safety program: evaluation of a method to Algorithm for Stochastic Systems Identification, WSEAS
proactively improve airline safety performance”, Safety Science, vol. 30, TRANSACTIONS on ELECTRONICS.
no.3, pp. 275–295,1998. [39] Mirela Voicu, Gabriela Mircea (2005), Algorithms for exploiting
[13] Erto P., Probabilità e statistica per le scienza e l’ingegneria, Milan: multidimensional databases, WSEAS TRANSACTIONS on
McGraw-Hill, 1999. INFORMATION SCIENCE and APPLICATIONS Issue 12, Volume 2.
[14] Evans A.W., “Risk assessment by transport organizations”, Transport [40] Gao Shiqiao, Liu Haipeng, Jin Lei (2006), Fuzzy Dynamic
Reviews, vol.17, no.2, pp. 145-163, 1996. Characteristic of Concrete Material under Impact Loads WSEAS
[15] Floyd P., Nwaogu T.A., Salado R., George C., “Establishing a TRANSACTIONS on APPLIED and THEORETICAL MECHANICS
comparative inventory of approaches and methods used by enforcement Issue 1, Volume 1.
authorities for the assessment of the safety of consumer products
covered by the Directive 2001/95/EC on general product safety and
identi cation of best practices”, Final Report Prepared for DG/SANCO,
European Commission by Risk & Policy Analysts Limited, United
Kingdom, 2006.
[16] Hadjimichael M., McCarthy J., “Implementing the Flight Operations
Risk Assessment System”, presented at the 57th International Air Safety
Seminar Shanghai, China, Nov 2004.
[17] Gibbons J.H., “Airport System Development: OTA-STI-231”, presented
at Office of Technology Assessment Congress, Washington D. C.,
August, 1984. Library of Congress Catalog Card Number 84-601101.
[18] Guidelines to Safety Assessment, Group of Aerodrome Safety
Regulators, WP093, May, 2006.
[19] Guldenmund F.W., ‘The nature of safety culture: a review of theory and
Research”, Safety Science, vol. 34, no.1-3, pp 2215-2257, 2000.
[20] Hale A.R., Heming B.H., Carthey J., “Modelling of safety management
systems”, Safety Science, vol. 26, no. 1-2, pp. 121-140, 1997.
[21] Janic M., “An assessment of risk and safety in civil aviation”, Journal of
Air Transport Management, vol.6, no. 1, pp. 43-50, January 2000.
65