Air Traffic Volume and Air Traffic Control Human Errors

Navigable airspaces are becoming more crowded with increasing air traffic, and the number of accidents caused by human errors is increasing. The main objective of this paper is to evaluate the relationship between air traffic volume and human error in air traffic control (ATC). First, the paper identifies categories and elements of ATC human error through a review of existing literature, and a study through interviews and surveys of ATC safety experts. And then the paper presents the results of an experiment conducted on 52 air traffic controllers sampled from the Korean ATC organization to find out if there is any relationship between traffic volume and air traffic controller human errors. An analysis of the experiment clearly showed that several types of ATC human error are influenced by traffic volume. We hope that the paper will make its contribution to aviation safety by providing a realistic basis for securing proper manpower and facility in accordance with the level of air traffic volume.


Introduction
The objectives of the air traffic services shall be to prevent collisions between aircraft, prevent collisions between aircraft on the manoeuvring area and obstructions on that area, expedite and maintain an orderly flow of air traffic, provide advice and information useful for safe and efficient conduct of flights, notify appropriate organizations regarding aircraft in need of search and rescue aid, and assist such organizations as required [1].In order to achieve this purpose, air traffic controllers should be especially apt to deal with the interaction between humans and mechanical devices while they provide directions or advices to pilots to maintain vertical and horizontal separations between aircrafts and avoid aircraft collision.Accordingly, the air traffic controllers need to conduct multiple functions at the same time, such as thinking, listening and speaking.Considering the complicated nature of this task, one would wonder whether there is a higher probability for an air traffic controller to make mistakes when traffic is heavier.The objective of this paper is to investigate if air traffic control (ATC) human error is influenced by the size of traffic volume.
First, categories and elements of ATC human error was studied through existing literature, and interviews and surveys of ATC safety experts.And then the paper presents the results of an experiment conducted on 52 air traffic controllers sampled from the Korean ATC organization, to find out if there is a relationship between traffic volume and air traffic controller human error.It should also be noted that we used an ATC simulator invented by the Korean Government for this experiment.

ATC Human Error and Air Traffic Volume
It is known that a lot of aircraft accidents are caused both directly and indirectly by human factors.Considering that human factor involves all aeronautical personnel who are related to aircraft operations, it is critical to do an in-depth research on the human factors of pilots and air traffic controllers who take the most crucial roles in aircraft operations.As for pilots, there are various studies and solutions that deal with issues such as the Cockpit Resource Management (CRM) and incorporation of a human resource management system into the Line Oriented Flight Training (LOFT).However, researches on the ATC human factor have been relatively inactive.
According to the Boeing Company, it turned out that of the commercial aircraft accidents for the past 10 years, 55% were caused by pilot error, 17% by aircraft defect, 13% by weather condition, 5% by airport and ATC, 3% by maintenance and 7% by miscellaneous matters [2].
Although ATC accounted for only 5% of commercial aircraft accidents, which is comparatively lower than other factors, it should not be overlooked that the 55% portion for which pilot error accounts, either directly or indirectly involves ATC because the cooperation between a pilot and an air traffic controller composes a significant part of aircraft operation.
Inspired by the current CRM program originally designed for the airline cockpit crew, EUROCONTROL has developed Team Resources Management (TRM) in order to research human factors in air traffic controllers [3].FAA has also created a new area called "ATC-CRM" for the study of controllers' errors [4].
The definition of air traffic volume used for ATC purposes is the maximum number of aircraft entering a sector in a given length of time.It is generally accepted that heavy traffic volume may present an excessively heavy workload to ATC personnel [5] and may thus result in a higher probability of error [6].So, it is also necessary to examine workload increase according to traffic volume increase.US Federal Aviation Administration (FAA) has reported that supplementary manpower in ATC is not provided in a timely manner, which conesquently causes a heavy workload and finally leads to more accidents [7].The ATC workload standard of EUROCONTROL is as described in the following table [8].
As ATC control sectors become more complicating because of the increase in air traffic volume worldwide, there have been efforts to rearrange sector structures and introduce more enhanced and automated ATC systems all around the world.However, due to the increase of information as well as air traffic volume, air traffic controllers are exposed to problems and situations that they have never experienced before, consequently increasing workload which is the cause of human errors [9].The following two accidents are example cases of air colli-

Structure of ATC Error Elements
Based on a literature review [10], and interviews and surveys of ATC safety experts, the study categorized the ATC error into three categories; communication error, procedure error, and instruction error.The definition of each error category is as follows: • Communication error refers to errors during radio communication.Communication error in ATC is divided into the two categories of errors that occur between a pilot and an air traffic controller, and the errors that occur between air traffic controllers.For instance, there are errors such as not challenging incorrect readback, using wrong call-signs, using non-standard phraseology, and missing and clipping the call sign.
• Procedure error involves incompliance with ATC procedures; for instance, failure to respond to an unanswered call, not responding to alarm, not identifying aircraft, failure to terminate radar services, not issuing approach clearance, not giving reasons for vectoring information, failure to deliver information to aircraft, etc.
• Instruction error occurs while conducting control procedures and communications.Specifically, there are errors such as delivery of incorrect information, issuing descent instruction late, issuing flight phase change instruction late, direction instruction error, clearance instruction error, etc.
This research also tried to define major error elements, which are components of each ATC human error cate gory, by the analysis of the data on ATC error items.These data are obtained from the interviews and surveys

Preparation for Experiment
This study conducted experiments on ATC duty performance and human error during duty with sampled air traffic controllers utilizing simulated approach control lab.The ATC task is generally divided into 4 major parts: area control, approach control, aerodrome control and ramp control.According to the Korean Government's data (2008), there are about 300 air traffic controllers who perform this duty.The sample group for this re- search was air traffic controllers who are currently or were previously in charge of ATC aerodrome control and approach control at international airports in Korea.The characteristics of this 52 sample group are as described in the following table.
The ATC simulation equipment utilized for this research is a kind of training device for air traffic controllers, developed by the Ministry of Land Transportation and Maritime (MLTM) of Korea in 2007.This device can simulate various flight situations that may occur during air traffic control duty.It has basic functions that give various control instructions related to flight maneuver such as climb, cruise, descent, and speed control of an airplane.It enables trainees to experience situations close to ones that occur during the actual duty of ATC by simulating situations and conditions such as the approach course diagram, initial take-off direction, instrument landing approach path, various weather conditions, etc.

Conducting the Experiment
In order to carry out the experiment, the above mentioned ATC simulator was installed and the training program was adjusted so that it could produce the data appropriate for the purpose of this research.There were 3 computers, 4 monitors, 5 radio communication devices and 2 speakers provided for the participant group consisting of 1 air traffic controller and 3 pilots.The seats for the pilots and controller is separated by more than 5 meters with a partition, so that verbal communication between the air traffic controller and pilot can be conducted in a situation similar to that of the actual radio communication between pilots and the controller.
The spatial background of the experimental scenario is the terminal management areas of two large international airports in Korea, Jeju International Airport and Gimhae International Airport.We conducted the experiment twice spending four months in total.The first period of experiment was from January 15, 2009 to March 15, 2009, and the second one was from July 1, 2009 to August 31, 2009.Each sampled air traffic controller was asked to perform air traffic control duty for 15 minutes with varied levels of air traffic volume.They were required to try out transfer of control, issuance of traffic information, and aircraft separation.In the first period there were three levels of traffic volume, designated as V1, V2 and V3.The second period experimented with twelve levels of traffic volume, designated as L1, L2, L3, L4, L5, L6, L7, L8, L9, L10, L11 and L12 (refer to Table 5).This volume spectrum is defined to accommodate the entire range of possible situations, from a very low level to an extremely high level of traffic volume.As one can see at Table 5, the level of traffic volume is defined by the number of aircraft controlled by each sampled controller during the 15 minute experimental session.The number of errors made by each sampled controller was counted utilizing the records of ATC duty during his/her experimental session.

Analyses
This study established four hypotheses, in order to discuss the relationship between air traffic volume and frequency of error occurrence: Hypothesis 1: Overall ATC Error will increase as air traffic volume increases.
Hypothesis 2: Communication error will increase as air traffic volume increases.
Hypothesis 3: Procedure error will increase as air traffic volume increases.
Hypothesis 4: Instruction error will increase as air traffic volume increases

Test of Hypothesis 1
The error data obtained in the second period of the experiment was used to test hypothesis 1.The level of in-  fluence that traffic volume has on frequency of error occurrence is presented in Table 6.First, we performed a correlation analysis to see the level of correlation between the two variables, traffic volume and frequency of ATC human errors.The result showed that 0.984 (p < 0.01) was the correlation coefficient.So, it can be said that there is very strong relationship between the level of air traffic volume and the error frequency of the air traffic controller.And we also performed a simple regression analysis utilizing traffic volume as the independent variable "x", and frequency of error as the dependent variable, "y".The result of the regression analysis was "y = 0.0516 x + 0.1886" with an R 2 value of 0.9675, which also indicates high significance.
Although the overall equation shows a linear regression curve, there are some areas where one can detect unproportionally higher marginal increase in frequency of error.As you can see in Figure 1, the marginal increase in error frequency from the volume level 22.5 to 27.5 is higher than in other areas.
Table 7 shows the marginal increase in error frequency at each level of traffic volume.The value in the third column and sixth column of Table 7 is calculated utilizing the following equations; ∆у = у i+1 -у i у i+1 = frequency of error at x i+1 traffic volume у i = frequency of error x i traffic volume where, x i = traffic volume (number of aircraft controlled per unit time), y i = frequency of error occurrence Referring to Table 7, it can be said that the marginal     error frequency of each category was significantly affected by the level of traffic volume.It was confirmed that all three hypotheses were accepted with the significance level of 0.05.This means that the error frequency of each category of ATC human error is influenced by air traffic volume in a statistically significant manner.
increase in error frequency is highest at the level of traffic volume between 22.5 to 25.

Test of Hypotheses 2, 3 and 4
Hypotheses 2, 3, and 4 were tested with the data obtained in the first period of the experiment.Those hypotheses focus on error frequency variation of each error category, such as communication error, procedure error and instruction error, depending on the level of traffic volume.Table 8 and Figure 2 shows the distribution of average error frequency for each category of ATC human error.Each of the hypotheses was tested separately through ANOVA (Analysis of Variance) to see if the average error frequency of each category was significantly affected by the level of traffic volume.It was confirmed Table 9 is the distribution of error frequency for each element in the communication error category.Figure 3 presents this information as a diagram.Table 10 is the distribution of error frequency for each element in the procedure error category.Figure 4 shows this in the form of a diagram.Table 11 is the distribution of error frequency for each of the 6 elements in the instruction error category.Figure 5 shows this information in the form of a diagram.

Conclusions
Despite the utilization of high-tech equipment, ATC is still mostly dependent on individual decision-making [11], which is always subject to probability of occurrences of human error.The main purpose of this research was to test a few hypotheses that claimed that the frequency of ATC human error will be influenced by the level of air traffic volume.The required data were gathered through experiments that utilized a sample of air traffic controllers who are currently working for the Korean ATC organization, and the ATC simulator.
We found there are significant relationships between ATC human error and the level of traffic volume.As the air traffic volume increases, the frequencies of error occurrence for most ATC human error elements defined by this study, were verified to increase.Especially, the increase of procedure errors was remarkably high compared to other error categories.The marginal increase in frequency of error from traffic volume 22.5 to 25 (number of aircraft) was revealed to be the highest.Therefore, it may be efficient to limit traffic volume to less than 22 aircraft per 15 minutes.However, it is necessary to consider differences that originate from factors such as air traffic control system of each air traffic control facility, sector conditions and flight procedures when determining the appropriate traffic volume.
It would be effective to assess the determined traffic volume on a regular basis, and to take corrective measures such as supplementing manpower, increasing controlling seats and enhancing air traffic control systems in case determined traffic volume is exceeded.We hope that the paper will make its contribution to aviation safety by providing a realistic basis for securing the proper amount of manpower and facility in accordance with the level of air traffic volume.

Figure 1 .
Figure 1.Regression result; frequency of error and level air traffic volume.

Figure 3 .
Figure 3. Frequency of communication error by air traffic volume.

Figure 2 .
Figure 2. Air traffic volume and frequency of error.

Figure 4 .
Figure 4. Frequency of procedure error by air traffic volume.

Figure 5 .
Figure 5. Frequency of instruction error by air traffic volume.

Table 2 . Major airplane accidents related to ATC human factor.
sion in which human errors occurred directly or indirectly because of heavy workload due to high air traffic volume, and lack of ATC facility and manpower.

Table 3 . Structure of ATC human error elements.
of profoundly experienced ATC practitioners who participated to provide their opinions on ATC human error elements.Finally, we constructed a structure of error elements as shown in the following table.