Alarm Design

Introduction

• It is expected that every alarm system should appropriately alert the operator of an unexpected change in the state of the system, allow enough time for the operator to acknowledge and respond to the alarm and only present useful and relevant information. It is recognised that the design of any effective alarm system must take human capability and operator behaviour into consideration. The acoustic and visual properties of alarm design must be taken into consideration. These properties have been studied extensively as a result of several properties of human behaviour that have been acknowledged to affect the effectiveness of alarm through the design process.

Texaco Milford Refinery (1994)

• An example of a poor alarm design is the Texaco Milford Refinery explosion. In this incident, a severe electrical storm caused plant disturbances, which lead to the explosion injuring 26 people and causing £48bn in damages. In the last 11 minutes before the explosion, the two operators in the control room had to recognise, acknowledge and act upon 275 alarms all going off simultaneously. This was problematic as operators were unable to distinguish which area of the plant was in crises. Key factors within the investigation that have arisen as a result include the poor prioritisation of alarms, control room display unit format and the inadequate training the operators had for dealing with a stressful and sustained plant incident

Alarm prioritisation

• Edworthy (2011) suggests that to avoid alarm proliferation, the most hazardous situations should have unique alarms, less hazardous could be indicated by a signal alarm that indicates lower priority. This unfortunately is difficult in practice as it is not possible for even intelligent monitoring systems to detect and interpret criticality. Therefore it is required that operators should learn to sufficiently identify situations that are more critical than others and safely prioritise alarm correspondance.

Acoustic Properties

• Xiao (n.d.) writes how "acoustic properties are often poorly designed but more importantly, there seems to be a general lack of understanding of what kind of information warning signals provide" This could be dependent upon the sound the alarm is making and how easy it is for operators to be able to learn to associate a sound to to a particular meaning. In terms of alarm learnability, there are three types of sound that are often used. First there are abstract tones such as beeps, buzzes etc. Second, there are tonal/melodic alarms. Thirdly, there are auditory icon which are a large range of sounds, usually real-world sounds that bear a distinct relationship to the hazard the are indicating. Edworthy (2011) writes how "two research projects demonstrated that there are significant differences in how easy different classes of sound are to learn." It is believed that abstract and tonal alarms are difficult to learn. If there is a situation where an alarm sounds and it is not recognised, more attention will be placed on it than if it was a common occurrence. Also, there are several experimental databases showing how the urgency of an alarm can be reduced or increased by altering the acoustic properties such as loudness, pitch rhythm etc.

Inattentional deafness

• It has been found that a fundamental shortcoming of many of the alarm problems is with the interpretation of the significance of a state change in the monitored process. This can also be impacted by the state of the cognitive workload an operator is handling at time of alarm sounding. Inattentional deafness suggests that tasks involving high perceptual load can consume most of an individual's mental capacity, leaving little remaining for processing information that is not directly related to the focal task. Dehais et al (2012) conducted a study where pilots were fitted with an eye tracker and electrocardiogram reader and asked to perform landings in a flight simulator . During the critical landing, an auditory landing gear alarm was triggered while the participants also faced a windshear (rapid change in wind speed and direction). It was reported that 8 out of 14 pilots did not report hearing the critical alarm during debrief. Therefore, according to perceptual load theory, tasks involving high perceptual loads can consume most of attentional capacity, leaving little remaining for processing information that is not directly related to the focal task.