Randall J. Mumaw

There Is More to Monitoring a Nuclear Power Plant than Meets the Eye

A fundamental challenge in studying cognitive systems in context is how to move from the specific work setting studied to a more general understanding of distributed cognitive work and how to support it. We present a series of cognitive field studies that illustrate one response to this challenge. Our focus was on how nuclear power plant (NPP) operators monitor plant state during normal operating conditions. We studied operators at two NPPs with different control room interfaces. We identified strong consistencies with respect to factors that made monitoring difficult and the strategies that operators have developed to facilitate monitoring. We found that what makes monitoring difficult is not the need to identify subtle abnormal indications against a quiescent background, but rather the need to identify and pursue relevant findings against a noisy background. Operators devised proactive strategies to make important information more salient or reduce meaningless change, create new information, and off-load some cognitive processing onto the interface. These findings emphasize the active problem-solving nature of monitoring, and highlight the use of strategies for knowledge-driven monitoring and the proactive adaptation of the interface to support monitoring. Potential applications of this research include control room design for process control and alarm systems and user interfaces for complex systems.

Operator monitoring in a complex dynamic work environment: a qualitative cognitive model based on field observations

Complex and dynamic work environments provide a challenging litmus-test with which to evaluate basic and applied theories of cognition. In this work, we were interested in obtaining a better understanding of dynamic decision making by studying how human operators monitored a nuclear power plant during normal operations. Interviews and observations were conducted in situ at three different power plants to enhance the generalizability of results across both individuals and plants. A total of 38 operators were observed for approximately 288 hours, providing an extensive database of qualitative data. Based on these empirical observations, a cognitive model of operator monitoring was developed. This qualitative model has important theoretical implications because it integrates findings from several theoretical perspectives. There is a strong human information processing component in that operators rely extensively on active knowledge-driven monitoring rather than passively reacting to changes after they occur, but there is also a strong distributed cognition component in that operators rely extensively on the external representations to offload cognitive demands. In some cases, they even go so far as to actively shape that environment to make it easier to exploit environmental regularities, almost playing the role of designers. Finally, expert operators use workload regulation strategies, allowing them to prioritize tasks so that they avoid situations that are likely to lead to monitoring errors. These meta-cognitive processes have not received much attention in the human information processing and distributed cognition perspectives, although they have been studied by European psychologists who have studied cognition in complex work environments. Collectively, these findings shed light on dynamic decision making but they also serve an important theoretical function by integrating findings from different theoretical perspectives into one common framework.

How do operators monitor a complex, dynamic work domain? The impact of control room technology

This article describes part of a research programme whose goal is to develop a better understanding of how operators monitor complex, dynamic systems under normal operations. In a previous phase, field observations were made at two older nuclear power plant control rooms (CRs) consisting primarily of analogue, hard-wired instrumentation. In this phase, additional field observations were conducted in a newer computer-based CR to determine the impact of CR technology on operator monitoring. Eleven different operators were observed in situ for a total of approximately 88 h. The findings indicate that there are many similarities in the monitoring strategies adopted by operators in the two types of CRs. However, in most cases, these same strategies are performed using different behaviours, thereby showing the shaping effect of the CR technology. A new way of conceptualizing the difference between traditional analogue CRs and modern computer-based CRs is proposed.

Using Cognitive Task Analysis to Define Human Interface Requirements for First-of-A-Kind Systems

Cognitive task analysis (CTA) methods have grown out of the need to explicitly consider cognitive processing requirements of complex tasks. A number of approaches to CTA have been developed that vary in goals, the tools they bring to bear, and their data requirements. We present a particular CTA technique that we are utilizing in the design of new person-machine interfaces for first-of-a-kind advanced process control plants. The methodology has its roots in the formal analytic goal-means decomposition method pioneered by Rasmussen (1986). It contrasts with other approaches in that it is intended: (1) for design of first-of-a-kind systems for which there are no close existing analogues, precluding the use of CTA techniques that rely on empirical analysis of expert performance; (2) to define person-machine interface requirements to support operator problem-solving and decision-making in unanticipated situations; and (3) to be a pragmatic, codified, tool that can be used reliably by person-machine interface designers.

Evaluation Issues for Computer-Based Control Rooms

The design of control centers is advancing toward totally computer-based man-machine interfaces. Computer based interfaces offer many potential advantages over traditional hardwired control panel interfaces including greater flexibility regarding the type of data displayed and its presentation. However, achieving this potential will require development of new interface concepts that will change the way operators interact with the plant. Extensive evaluation throughout the design process will be required to verify and validate the interface concepts. This paper describes a process for uncovering evaluation issues related to the computer-based control room concept and its relationship to cognitive activities of plant control. Important evaluation issues are presented.

Analysis of Complexity in Nuclear Power Severe Accident Management

A model of decision making has been developed for nuclear power plant operations and has been previously applied to the analysis of performance during emergency operations. The model was extended to identify the cognitive skills required, the types of complexity that can arise, and the potential for human error in severe accident management (SAM). Twelve SAM scenarios were developed to aid in this analysis. Potential sources of complexity and error are described and illustrated, and implications for training cognitive skills are discussed.

More About Operator Monitoring under Normal Operations: The Role of Workload Regulation and the Impact of Control Room Technology

This paper is part of a research program that we have been conducting to better understand how operators monitor a nuclear power plant under normal operations. A field study was conducted in a modern plant that has more automation and more computer-based displays than the two plants that we had observed in previous studies. Eleven operators were observed for a total of approximately 88 hours. The findings suggest that operators actively adopt strategies to regulate their workload so as not to exceed their resource constraints. In addition, the results suggest that computer-based control rooms require operators to have more knowledge about the interface (not the plant itself) than do the older analog, hard-wired control rooms. Moreover, computer-based designs require operators to reduce the degrees of freedom in the design to have context-sensitive information, whereas hard-wired designs require operators to expand the degrees of freedom.

Training Complex Tasks in a Functional Context

We have reviewed training programs for complex skills that have strong decision-making components, such as nuclear power plant operations and air traffic control. In each case, we found that an ISD approach is routinely applied to training-program design. The ISD framework can aid training designers in designing individual modules of instruction but seems to provide insufficient guidance on designing the larger training-program structure. We found two types of problems. First, because a good understanding of skill acquisition is not used to drive training-program design, training activities can be ineffective or inefficient. Second, because it is difficult to get insights on cognitive skills with traditional task analysis, the core decision-making task is not trained explicitly. Trainees are typically on their own to discover decision-making skills. We developed an alternative framework for training-program design called the Functional Context Approach. This approach attempts to restore efficiency to skill acquisition and improve training of critical decision-making skills.