Considerable interest has occurred of late over the use of hexagonal/radar/spider plot displays. While this type of display exhibits the high display proximity described by Dr. Christopher Wickens of the University of Illinois at UrbanaChampaign, it may not be the best way to show data for pattern matches or qualitative assessment. Dr. Kevin Bennett and Dr. John Flack of Wright State University in Dayton have argued that the key is not whether the data are presented in a single shape, but whether “emergent features” (patterns) result when the data change. In either case, the key is that unique patterns must correspond to a unique state of the system being measured.
Consider the use of hexagonal displays in a recent industry trade publication. The intent was to enable assessment of advanced control and whether the constraints were set near their upper or lower limit. In the example, 100% was good, while 0% was bad. The author used radar plots similar to those in Figure 1 to show a normal case and one in which the pressure control was turned “off”. The concept is that the change in the pattern of the hexagonal display would alert the user to the fact that (1) some of the variables could be altered and (2) a variable was “off”.
The use of this format for presenting the data fails on several counts. First, the patterns do not correspond to the state of the system, because they represent how the operator has configured the system. The only change they represent in system state is the change the operator causes by altering the limits. The shape implies some sort of interaction between the variables, when the actual task is to examine each variable separately and determine if it is less than 100%. Ironically, Wickens’ principle argues that displays with high proximity are harder for detailed inspection tasks of a single variable shown in the shape.
However, there are other ways to present this data. Figure 2 shows the same data presented in a simple bar chart, with the margin to optimal (value of 100%) shown with the grey vertical shading. More vertical shading is bad. The need to alter the constraints can still be easily seen, even though the data are not presented in a single shape. The preferred method would be to assess the performance of each type of display to determine which yielded superior performance in the task at hand.
Ideally, any display format chosen should be evaluated for its impact on operator performance versus an alternate display format. While standard in the aerospace industry, this is rarely done in the process control industry. Lacking such an evaluation, the key becomes matching intended use with the impact of different display characteristics. If your organization is interested in a more detailed discussion of different ways to present information, Beville offers half- and full-day seminars on human factors in display design.
Copyright © 2005 Beville Engineering, Inc., All Rights Reserved
RELATED EXTERNAL MEDIA
|Consortium Reports New Findings on Alarm Rates||Automation World|
|How Many Alarms Can An Operator Handle||Chemical Processing|
|Impact of Alarm Rates and Interface Design on Operator Performance||Automation World|
|Operator Interfaces: Moving from Comfortable to Most Effective||Automation World|
|Operator Performance as a Function of Alarm Rate and Interface Design||Mesa.org|
This year's Fall meeting for the Center for Operator Performance will be October 24-26 in Corpus Christi. For more information, please contact Lisa Via. Guests are always welcome!
Our summer newsletter is now available. Click here!
Take our short survey on operator span of control. Click here (new window)
David Strobhar's book, "Human Factors in Process Plant Operation," is now available in both hardcover and Kindle e-book.
Copyright © 1996-2018 Beville Engineering, Inc. All rights reserved. (937)434-1093. Beville@Beville.com