Application control chart coApplication control chart concepts of designing a pre-alncepts of designing a pre-al

arm systemarm system

Sheue-Ling Hwang, Jhih-Tsong Lin, GSheue-Ling Hwang, Jhih-Tsong Lin, Guo-Feng Liang, Yi-Jan Yau,uo-Feng Liang, Yi-Jan Yau,

Tzu-Chung Yenn, Chong-Cheng HsuTzu-Chung Yenn, Chong-Cheng Hsu

Nuclear Engineering and Design 238 (2008) 3522–3527

IntroductionIntroduction

A well-trained operator in an NPP should possA well-trained operator in an NPP should possess the abilities not only to monitor the systeess the abilities not only to monitor the system information, but also to utilize this informatm information, but also to utilize this information to manipulate the system and predict otheion to manipulate the system and predict otherwise unforeseen situations.rwise unforeseen situations.

There are two approaches among the studies There are two approaches among the studies about advanced alarm designs in recent year: iabout advanced alarm designs in recent year: improving the alarm system and reducing the fmproving the alarm system and reducing the frequency of alarms.requency of alarms.

In this study, we implemented a quality controIn this study, we implemented a quality control method, which has no complex mathematical method, which has no complex mathematical model and which has been applied in several l model and which has been applied in several industries, to reduce the frequency of alarms iindustries, to reduce the frequency of alarms in a power plant control room.n a power plant control room.

In order to evaluate the efficiency of the pre-alIn order to evaluate the efficiency of the pre-alarm system, we conducted an experiment to siarm system, we conducted an experiment to simulate the reactor shutdown and the load-rejmulate the reactor shutdown and the load-rejection task in a control room to assess task peection task in a control room to assess task performance.rformance.

Quality control chart techniques Quality control chart techniques for fault detectionfor fault detection

Rule 1Rule 1: Any one point falls outside the upper c: Any one point falls outside the upper control limits (UCL)or lower control limits (LCL).ontrol limits (UCL)or lower control limits (LCL).

Rule 2Rule 2: Seven points in a row are continually in: Seven points in a row are continually increasing (or decreasing).creasing (or decreasing).

Rule 3Rule 3: Cyclical patterns of points occur.: Cyclical patterns of points occur. Rule 4Rule 4: Two out of three consecutive points fall : Two out of three consecutive points fall

beyond the twosigma limit.beyond the twosigma limit. Rule 5Rule 5: A run of five points falls beyond the on: A run of five points falls beyond the on

e-sigma limite-sigma limit

MethodMethod

Participants-Participants-Twenty-six participants. ThTwenty-six participants. The ages of participants were 36.5±9.8 yeae ages of participants were 36.5±9.8 years.rs.

Experimental tasks and procedure-Experimental tasks and procedure- The team members in the experiment weThe team members in the experiment we

re a reactor operator (RO), an assistant rre a reactor operator (RO), an assistant reactor operator (ARO) and a supervisor.eactor operator (ARO) and a supervisor.

In the normal state, the system was useIn the normal state, the system was used to simulate the shutdown of the reactd to simulate the shutdown of the reactor according to integrated operating proor according to integrated operating procedures.cedures.

In the abnormal state, the procedure of lIn the abnormal state, the procedure of load rejection was used. operators had toad rejection was used. operators had to find the solutionsfrom the procedure to find the solutionsfrom the procedure to deal with the events so that all events o deal with the events so that all events could be relieved in 10min.could be relieved in 10min.

During the shutdown and load-rejection During the shutdown and load-rejection procedures, the operators had to monitprocedures, the operators had to monitor six critical parameters of the system. Ior six critical parameters of the system. In order to evaluate the effects of the prn order to evaluate the effects of the pre-alarm pattern, we set some fault evente-alarm pattern, we set some fault events based on these parameters.s based on these parameters.

The secondary task was based on HwanThe secondary task was based on Hwang et al.’s (2007) task of comparing two ig et al.’s (2007) task of comparing two integers without using any computations.ntegers without using any computations.

The subjects used the mouse to choose The subjects used the mouse to choose “greater than”, “equal to” or “smal“greater than”, “equal to” or “smaller than” to respond to the number coler than” to respond to the number comparison.mparison.

Experimental variablesExperimental variables

There were two independent variables: There were two independent variables: alarm type and system state. alarm type and system state.

The alarm types had three levels – text, The alarm types had three levels – text, graphic pre-alarm and original graphic pre-alarm and original

The system state had two levels, the norThe system state had two levels, the normal and abnormal system states. mal and abnormal system states.

Each team operated six trials in the expeEach team operated six trials in the experiment.riment.

For the dependent variables, task perforFor the dependent variables, task performance, mental workload and team SA wmance, mental workload and team SA were measured.ere measured.

The performance of the primary task waThe performance of the primary task was measured by the operation time, the ss measured by the operation time, the searching time and the frequency of alarearching time and the frequency of alarm.m.

The secondary task performance was mThe secondary task performance was measured by the rate of correct responses.easured by the rate of correct responses.

Results-Results-Task performanceTask performance

The result for alarm frequency indicated The result for alarm frequency indicated that the pre-alarm type significantly affethat the pre-alarm type significantly affected the frequency of alarms (cted the frequency of alarms (FF2,72 = 142,72 = 141.7, 1.7, P P < 0.001). < 0.001).

Graphics and text were better than the oGraphics and text were better than the original alarm design, but there was no diriginal alarm design, but there was no difference between graphics and text.fference between graphics and text.

The results indicated that operation timThe results indicated that operation times (in seconds) were not significantly diffes (in seconds) were not significantly different when using different types of pre-aerent when using different types of pre-alarm systems.larm systems.

The results of ANOVA revealed that the The results of ANOVA revealed that the main and interaction effects on average main and interaction effects on average searching time were not significant.searching time were not significant.

Results-Results-Mental workloadMental workload

The results of ANOVA indicated that the pre-alThe results of ANOVA indicated that the pre-alarm type was a significant factor (arm type was a significant factor (FF2,144 = 7.972,144 = 7.973, 3, P P < 0.01) in workload.< 0.01) in workload.

The mental workload of either graphics or text The mental workload of either graphics or text was lower than the original, but there was no was lower than the original, but there was no difference between graphics and text.difference between graphics and text.

The NASA-TLX mean scores of the normal and The NASA-TLX mean scores of the normal and abnormal states were 62.24 and 70.47, respectabnormal states were 62.24 and 70.47, respectively.ively.

The ANOVA result indicated that neither The ANOVA result indicated that neither the main nor the interaction effect was sthe main nor the interaction effect was significant.ignificant.

Results-Results- Team SA Team SA

ConclusionConclusion

The results indicated that this system coThe results indicated that this system could detect the most common fault sympuld detect the most common fault symptom in the NPP and that implementing ttom in the NPP and that implementing the system could reduce operators’ mehe system could reduce operators’ mental workload.ntal workload.

Future studies could examine whether tFuture studies could examine whether the pre-alarm system could be expanded he pre-alarm system could be expanded to include the operator’s physiology into include the operator’s physiology indicesdices