Ecological Interface Design in the Nuclear Domain

Ecological Interface Design in the Nuclear Domain

(Parte 4 de 5)

The balances and temperature-pressure plot, which communicate the information identified in the abstract and generalized functions, respectively, share some means-ends relationships. These relationships are depicted by the projection lines from data points on the plot (i.e., circle and blue diamond) to the energy balance that can support operators in monitoring tasks (Fig. 15). These projection lines visually translate deviation of heat exchange efficiency in terms of temperature into energy. As mentioned, energy is a useful indicator because it accounts for both mass and temperature. Large energy deviations can be alarming as stable shifts between energy states are usually slow. The projection lines also facilitate pattern recognition or rule-based behaviors. As the mass, energy, temperature, and pressure deviate from steady-state conditions, the projection lines form a distinctly different visual pattern, which is in the form of either a signal or sign to facilitate information processing of potential process disturbance (Fig. 16).

Authorized licensed use limited to: UNIV ESTADUAL PAULISTA JULIO DE MESQUITA FILHO. Downloaded on July 13, 2009 at 17:52 from IEEE Xplore. Restrictions apply.

LAU et al.: ECOLOGICAL INTERFACE DESIGN IN THE NUCLEAR DOMAIN 3591

Fig. 1. The second ecological display for the feedwater division. (a) Feedwater pump monitoring units and mass balance. (b) Valve monitoring unit. (c) Mimic diagram depicting process flow from the feedwater tank to the reactor. (c) Reactor tank embedded with equipments and a level trend graph. (d) Temperature profile depicting heat exchanges in the mimic diagram. 179 143 m (300 300 DPI).

Condensate Supply Monitoring: Fig. 10(a) presents the graphical form for monitoring the condensate pump operations. The information requirements are defined based on the AH of the condensate subsystem (Level 3 part-whole decomposition) in the feedwater division. The functional purposes are to satisfy condensate demand of the feedwater tank and to maintain target feedwater tank pressure, both of which are achieved by means of mass transport according to conservation of mass at the abstract function level.

Mass transport is achieved by means of pumping at the generalized function level. One constraint identified in the WDA is that mass transports can only be established by pumping the condensate at a discharge pressure higher than the pressure of the receiving tank. One or two of the three condensate pumps and a common recycle valve (which appeared at the physical function level) regulate the pumping process. The condensate pumps are fixed-speed centrifugal pumps, which have an operating capacity constrained by a set of pump curves that specify the range of pressures and flow rates. If the condensate pressure is higher than the pre-defined limit, the recycle valve returns the fluid to the condenser, and in effect, reduces the efficiency of the condensate subsystem.

A mass balance and a pressure-flow rate plot (Fig. 10(a)) depict the identified work domain information. The mass balance (Fig. 17 or the left part of Fig. 10(a) illustrates the functional purpose and abstract function in terms of satisfying condensate demand of the feedwater tank and adhering to the law of conservation of mass. The left and right vertical bar graphs are the total mass leaving and entering the control volume (which is the area colored in the lightest grey in Fig. 10(c), respectively. These are also summation bar graphs that show distribution of input from different sources and output to different sinks. When the two bar graphs are the same, the line connecting them should be horizontal and the center of the bubble/oval should be aligned to the center of the connecting line. When the mass balance constraint is violated across the control volume, the line rotates and the bubble slides along the line towards the bar graph of greater value, similar to a bubble in a spirit level. The displacement of the bubble is scaled to be sensitive to small deviations between the bar graphs. The structural means-ends relationship between

Authorized licensed use limited to: UNIV ESTADUAL PAULISTA JULIO DE MESQUITA FILHO. Downloaded on July 13, 2009 at 17:52 from IEEE Xplore. Restrictions apply.

3592 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 5, NO. 6, DECEMBER 2008

Fig. 12. Simplified, distilled level 3 AH for condenser subsystem in the condenser division. 87 93 m (600 600 DPI).

Fig. 13. Mass and energy balances depicting the FP and AFn of the condenser. 8 71 m (600 600 DPI).

first principles and the ensuing process is illustrated by a line connecting the bar graph at right to the pressure and mass flow operating point (i.e., white circle) on the pressure-flow rate plot (Fig. 18, or the right part of Fig. 10(a).

The pressure-flow rate plot depicts the operation of the condensate pumps in the mimic diagram (Fig. 10(c)). The mass balance and pressure-flow rate plot share a common vertical axis of mass flow rate, enabling this line to represent the mass input by the condensate pumps in connection to their operating parameters (i.e., flow rate and pressure). Two work domain constraints are embedded in this pressure-flow rate plot. First, a vertical line marks the feedwater tank pressure—the minimum that the condensate pumps must overcome to establish mass transport. Second, the operating points of the condensate pumps are

Fig. 14. Temperature-pressure graph of the condenser. 8 6 m (600 600 DPI).

Fig. 15. Integrated graphical forms depicting efficiency gaps of the condenser. 8 65 m (600 600 DPI).

restricted by the two pump curves. The upper and lower pump curves describe two-pump and one-pump operating modes, respectively. Under normal operating conditions, the operating point may only move along one of these curves, with the active one being highlighted.

The detailed description verifies that the graphical form embeds information specified by the work domain analysis to be necessary for monitoring under all situations. Furthermore, this graphical form is verified to convey relevant information in such a way that promotes skill- and rule-based behavior. It is also shown to support knowledge-base behavior by mapping relations between operating parameters of the components, purposes of the subsystem, and domain invariants.

The establishment of mass transport to achieve the primary objective of supplying condensate is demonstrated with the mass balance. The summation bar graphs are designed to communicate inefficiencies. In the case of substantial recycling of condensate, the portion of the bar graph at left, designated to recycle condensate mass output to the condenser, would become increasingly visible or even dominating suggesting an inefficiency typically caused by low total flow rates. On the other hand, faults violating mass conservation, such as leakage

Authorized licensed use limited to: UNIV ESTADUAL PAULISTA JULIO DE MESQUITA FILHO. Downloaded on July 13, 2009 at 17:52 from IEEE Xplore. Restrictions apply.

LAU et al.: ECOLOGICAL INTERFACE DESIGN IN THE NUCLEAR DOMAIN 3593

Fig. 16. Integrated graphical forms facilitating pattern recognition for monitoring. (a) Normal conditions. (b) Abnormal conditions. 179 81 m (600 600 DPI).

Fig. 17. Mass balance of the graphical form for monitoring condensate subsystem. 8 85 m (600 600 DPI).

within the control volumes, are communicated through the line rotation and bubble location of the mass balance (Fig. 17). The pressure-flow rate plot illustrates operating constraints. The constraint of minimum discharge pressure for the pumps to deliver condensate can be detected by locating the operating point (white circle) with respect to the vertical line that depicts the feedwater tank pressure. In other words, an operating point located to the left of this vertical line (darker area of the plot) means that condensate is not being transported to the feedwater tank. Operating points deviating from the pump curves indicate equipment malfunctions. The intersection between each pump curve and the vertical line of the feedwater tank pressure defines the maximum flow capacity (an operating limit) for each operating mode. This graphical form captures

Fig. 18. Pressure-flow rate plot for monitoring condensate subsystem. 83 85 m (600 600 DPI).

information content and structure with minimal use of symbols (i.e., numbers) allowing the operators to rely on lower levels of cognitive control for monitoring and diagnosis.

Valve Panel: The graphical form for monitoring valve operations was first designed for the feedwater subsystem in the feedwater division and later adopted in the other ecological displays (e.g., Fig. 1(b)). In contrast to other graphical forms of the ecological displays, the valve monitoring unit design is largely based on information contained in the Level 4 part-whole model, although the design was also guided by AH models of feedwater subsystem Level 3 part-whole decomposition. As mentioned, the analysts found that WDA at Level 3 decomposition (stopping at units of components) was useful for understanding the purpose, first principles, and

Authorized licensed use limited to: UNIV ESTADUAL PAULISTA JULIO DE MESQUITA FILHO. Downloaded on July 13, 2009 at 17:52 from IEEE Xplore. Restrictions apply.

3594 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 5, NO. 6, DECEMBER 2008 processes of the feedwater subsystems, but not sufficiently detailed to capture the inter-component relationships necessary for operators to optimize control of the bleed steams7. Thus, a Level 4 part-whole analysis was conducted.

Key elements discovered at this level were: 1) equally distributed bleed steam across heat exchangers results in higher efficiency, in accordance with the laws of thermodynamics; 2) valve positions provide a basis to infer relative flow rates; and 3) valve positions are set by the control system in the turbine division. The expected positions under steady-state conditions for automatically controlled valves are known for each power level, thus providing references for detecting deviations.

Fig. 1(b) portrays the information content discovered in the

Level 4 part-whole model. The position of each automatically controlled valve is depicted by a small circle along a vertical axis that indicates percentage of opening. Each blue diamond represents the expected valve position for a given power level. All circles should be ideally located inside their respective blue diamonds. Valves belonging to parallel flows are placed beside one another and connected with lines. These lines are perfectly horizontal under normal conditions (i.e., equally distributed flows and properly functioning valves). The lower part of the graphical form contains green bar graphs representing the difference between controller output and actual valve position. The bar graphs are invisible during fault-free operation. The description of these features verifies that this graphical form contains the information specified by the work domain analysis necessary to support knowledge-based tasks.

The graphical form for monitoring valve behaviors is again intended to support the operator at all three levels of cognitive control, adhering to the EID principle derivedfrom the SRK taxonomy. In terms of KBB, this form illustrates the relationships between valves, simplifying comparisons between positions of different valves. The horizontal lines connecting valves in parallel and the blue diamonds depicting expected values permit operators to rely on SBB or RBB for detecting anomalies. Similarly, the green bar graphs provide visually salient indications of discrepancies between set points and actual values, which usu- ally indicate fault conditions such as stuck valves. 3) Summary: Five ecological displays were designed to represent the information content and structure identified by the WDA. These displays contain graphical forms that permit operators to rely on lower level of cognitive control, an EID principlederivedfromtheSRKtaxonomytoimproveinterfacecompatibility with human information processing. To illustrate the viability of EID for improving verification, we described three graphical forms in detail to verify that information identified by the WDA (i.e., criteria for information content and structure) can be fully represented; and demonstrate the anticipated benefits for monitoring in terms of the SRK taxonomy (i.e., criteria for forms based on human perceptual capabilities) to verify that the displays are compatible with human information processing. This process indicates that EID is not only effective as a design

7The deign decisions here were also influenced by the lack of sensors for measurement or derivation of flow rates in the feedwater subsystem. This design could therefore only be based on valve position data. Sensor availability in the context of EID is included as a topic for future work.

tool but also as a verification tool that helps to ensure interface compatibility with operators.

We have illustrated that the EID framework can serve as both a design and verification tool to develop and evaluate displays for the secondary side of a BWR. To put this work in context with current research and practice, this section discusses the unique contributions of this study, its limitations, and prospects for future work.

A. Unique Contributions

This research offers several contributions. The EID products described here are unique to the open literature, providing the first detailed account of ecological displays for a simulator of an operating nuclear power plant. The detailed descriptions of selected graphical forms demonstrate anticipated benefits in supporting monitoring and diagnosis. In sum, this proof-of-concept expands on the foundation of prior research, confirming that the EID framework can be meaningfully scaled up to meet challenges faced by the nuclear industry (see [46]–[49], [52]).

The descriptions of the EID products serve as a detailed, representative example to assist NPP interface designers in putting the framework into practice. First, we have demonstrated that the WDA can specify information content and structure to support problem solving, especially during unanticipated events that are often precursors of major accidents (see [25]–[27]). The AH helps designers to conceptualize work domains in a psychologically relevant manner as opposed to relying on physical and process perspectives that do not explicitly support knowledgebased, ill-defined tasks. Second, based on the SRK taxonomy, we have demonstrated the anticipated benefits of the ecological displaysbyrepresentinginformationcompatiblewithhuman information processing. The taxonomy assists designers in predicting information processing performance of visual forms, thereby informing effectiveness of designs in capitalizing on innate human perceptual capabilities.

EID is also presented as a tool that could improve verification or analytical evaluation as proposed in [39]. Through the WDA, we specify verification requirements on information content and structure that explicitly aim to support problem solving, especially during unanticipated events. These requirements complement those generated by task analyses (or task support verification). As a potential framework for functional requirement analysis, the WDA information requirements ensure that the interface captures the necessary functional information (i.e., system objectives,system functions, and system-function relationships) for monitoring and diagnosis. The EID framework also specifies that the interface should not force operators to engage at levels of cognitive control higher than necessary. We evaluate the conformance to this specification or principle by verifying that information is primarily represented in the forms of signals and signs supplemented with symbols, thereby ensuring the maximum interface compatibility with human information processing.

The ecological displays have been implemented in a highfidelity simulator, have undergone functional testing by process

Authorized licensed use limited to: UNIV ESTADUAL PAULISTA JULIO DE MESQUITA FILHO. Downloaded on July 13, 2009 at 17:52 from IEEE Xplore. Restrictions apply.

LAU et al.: ECOLOGICAL INTERFACE DESIGN IN THE NUCLEAR DOMAIN 3595 experts, and have served as a test bed for empirical research (see [35]). This work therefore takes the first step toward a series of validation studies to determine the practical benefits that EID could bring to the nuclear domain.

B. Limitations

The highly representative nature of this design case study comes with several limitations. First, our scope was limited to the secondary side of the plant because we have greater access to turbine operators than reactor operators to serve as participants in upcoming empirical evaluation/validation studies. Our choice of analysis scope was also influenced by the fact that the secondary side was readily decomposed into three sections for allocation to the three design teams. Developing ecological displays for the primary side remains an outstanding research goal. The primary side introduces engineering complexities and threats to safety that exceed those encountered in secondary systems, providing a challenge for further verification research on EID in the nuclear domain.

Although implementation of this interface includes user interaction, we excluded the design of innovative control features to reduce training demands on the operators for the empirical study as described ina companion article [35]. Thus, the potential benefits of coupling control at the component level to higher-order properties such as energy have not yet been examined. The design of control features that satisfy the EID principles demands additional work.

C. Future Work

Future studies are necessary to address other viability issues associated with the EID framework that were not explored as part of this study. Technical requirements suchas additional sensors to satisfy information demands remain to be addressed (see [59], [60]). These requirements have direct implication on cost and, thus, viability of EID application. More generally, research on cost-benefit trade-offs related to interface development and implementation is also necessary to assess the viability of this framework. Finally, a more detailed account of the process of applying EID at an industrial scale would be an invaluable complement to the product description provided here.

The objective of this research program is to investigate the potential benefits that EID could bring to the nuclear industry. This article presents evidence that EID could improve interface design and verification practice in the nuclear industry. In applying the framework to the secondary side of an operating BWR plant, we have shown that EID is effective in addressing interface design challenges on a practical scale. In the descriptions of EID products, we have illustrated two concepts: 1) how a WDA reveals relevant domain characteristics that are theoretically necessary to support effective monitoring and diagnosis, and 2) how the SRK taxonomy assists the designer in predicting how information processing performance will be facilitated by visual representations of the domain information content and structure. We also illustrate that the EID framework could enhance current verification practice through the conduct of a WDA that specifies the functional information that should be contained in the interface to support problem solving. These findings contribute to the growing support in the academic literature for work domain-based interface design approaches. While demonstration or verification is illustrative of design features, interface effectiveness must be confirmed through validation. In a companion article [35], we present the first empirical evaluation of these ecological displays with professional operators, thereby, initiating the process of validating EID in the nuclear domain.

The authors would like to thank C. Nihlwing andH. Svengren of IFE for their technical support in providing detailed descriptions and explanations about the workings of the BWR simulator. They would also like to thank J. Kvalem, of IFE, for his efforts in making this collaboration possible, A. Teigen, of IFE, for his work in implementing our ecological interface, N. Dinadis for his insightful feedback on the interface design, and invaluable comments provided by G. Skraaning, Jr., and two anonymous reviewers.

[1] F. L. Toth and H. Rogner, “Oil and nuclear power: Past, present, and future,” Energy Economics, vol. 28, p. 1–12, 2006. [2] R. K. Lester, “New nukes,” Issues in Science and Technology, vol. 2, p. 39–46, Summer 2006. [3] World Nuclear Assoc., WNA Report: The New Economics of Nuclear

(Parte 4 de 5)

Comentários