Systems Integration / HMI

What Is the High-Performance HMI?

Operators Learn to Live With Design Flaws and Often Take the Easy Way Out and Live With the Less-Than-Perfect Systems They Grew Up With

By Ian Nimmo

The industry today is in a state of confusion regarding basic process control systems' (BPCSs) human-machine interfaces (HMIs). The industry has about 40 years experience with man- or human-machine Interfaces in one form or another. With the evolution of the BPCS, the HMI has evolved over a period of 50 years and has transitioned from physical lights, switches and annunciator panels with analog gauges and trend displays to electronic simulations and finally to computer interfaces.

But, the computer interfaces were designed with little knowledge or science added into the design, and many issues still remain because of this. To help understand the issues and the proposed solutions, we need to understand the current state of the industry. What specific problems we are trying to eliminate?

A quick survey of the current state of the HMI design using a strengths, weaknesses, opportunities and threats (SWOT) analysis reveals that the industry has fully adapted to the current state, and, in spite of known problems and limitations, is reluctant to change. This is mainly because change requires commitment to learn a new system, which involves design, implementation, testing, documentation and training.

Process control operators learn to live with design flaws, and often take the easy way out and continue to live with the less than perfect systems they grew up with. The strength of the existing system is that it evolved from panels to electronics to a first-generation faceplate equivalent and finally to a crude, graphical interface based on plant design.

The HMI many people have been using  evolved from a hard panel to group faceplate displays and then to P&ID graphics-integrating faceplates. This has been an easy solution, first taking the controllers from the panel and placing them into groups on a computer display, and then using the faceplate live values within a P&ID graphic.

The black backgrounds and bright, fully saturated colors were not designed. They were more a symptom of the technology, one that the vendors did not mind because bright, fully saturated colors are aestheticly pleasing to the eye, even though they may be 180° from the best practices learned from the science of using color. The vendors still sell their systems based on this premise, rather than on the science of using color correctly.

This statement can be proved by looking at automation vendors' websites and viewing the examples they use to promote their systems. Even though they may have policies supporting the new ASM Consortium-promoted graphics, grey-scale does not sell systems.

As we examine the weakness or problems that are created by this solution, we can read of multiple accident/incident reports that identify the HMI as a contributor to these incidents. We also discover that operators struggle with tracking information or getting overloaded with information because their graphics are not task-based, and information is scattered by the P&ID design. The old groups were faster, as they were assembled based on tasks.

Problems with the Old Way

We see navigation issues caused by lack of hierarchy, in which everything is designed at the same level with no overview. We also see inconsistencies in design because no structure was anticipated by the design. We also see operators trying to get around this issue by requesting more screens. I have actually seen a single operator with more than 24 screens, even though the new standards and guidelines recommend only four process control screens per operator based on short term memory (STM) issues and the limitations on the operator's ability to track more screens.

In additon, ergonomic design principles also help us understand that main screens should be within a 30° and 60° design.

These kinds of graphics have basic readability issues during "normal operations," let alone when data is moving fast during both abnormal operating conditions (AOCs) and emergency operations. These have been categorized as issues with clarity, consistency, too much variety, overload, visual noise and luminance contrast.

Many operators complain of eye-strain because of high-contrast color usage and the use of  >3:1 contrast ratio for colors, such as extremes of brightness from yellow-on-black backgrounds.

Why Users Like the current System

The eye has to adjust to light and dark continuously, especially with the extremes of light and dark in the environment. Most control rooms with these style graphics are kept dark to reduce glare issues. However, windows, task lights, ceiling lights and windows all create problems in these types of environments.

Graphics designed in this way suffer from poor or lacking functionality. They create an environment that supports human error. For example, they exacerbate short-term memory issues. These can inclulde operators forgetting they've made a manual move, such as opening a drain valve or bypassing an alarm. This system also generally creates high levels of cognitive workload as operators try to diagnose problems by searching for information, all the while keeping an eye on the operation. In the past, we have experienced many data entry problems, which often happen when the operator is several moves ahead of display updates.

The bigger issue with human error is one I've described as situation awareness issues. These touch on some of other areas, such as salience or misplaced salience, data overload and attention tunneling. Each of these deserve a white paper of its own. Other issues include distractions, communication breakdowns, out-of-loop syndrome, complexity creep, workload, fatigue, and working with the wrong mental model due to use of P&ID style.

Some quick fixes are available, and include reviewing  existing graphics against a formal philosophy and style guide, addressing issues around consistency, and only making graphics pop for important information. You can improve readability by addressing text font and character size or height, and by controlling color and following strict coding rules. Apply similar rules for lines and graphic objects. Address visual clutter, and follow rules regarding how much white space should be left and how many screens each operator should be following. Develop of a hierarchy of graphics views—overview, unit view, detail view and diagnostic view—instead of just using a flat P&ID view. P&IDs normally live down at the detail view. Also, training, management of change and documentation practices often need to be added to a graphic enhancement project.

Weaknesses of the Current System

All or Nothing?

Some of the biggest questions I'm often asked are, "Is it all or nothing? Can we do anything to our existing graphics without throwing away what we have? Or, do we have to redesign them totally?" Those are great questions.

I view improvement as an iterative process. Often it all depends on what else is happening. If you're going through an automation replacement, it makes sense to redesign the graphics to today's new standards. Many parts of a current design—the good parts—can be kept, but chances are the new techniques will require something that does not exist in the present design. Many of our customers testify that the new design techniques often result in a significant reduction in the number of graphics or schematics.

The important thing is to have a roadmap, in the form of a philosophy and style guide, which will ensure that the designer and users understand the rules for designing and building graphics, and address the issues of consistency, clarity, variability, etc.

In additon, the greatest lesson that designers and users need to understand about BPCS graphics is that they need to be layered. The background layer should be for fixed reference information, the next layer for variable information, then notifications, then alarms, and finally safety-critical alarms.

The pop-outs at the very front are controlled by color, thickness, brightness, contrast, etc. This allows different priorities to catch the operator's attention, provides improved situational awareness, and allows the graphic to work with the alarm system, rather than being a competitor for the operator's attention.

Finally, new graphics should be designed to the new life-cycle model promoted by ISA-101 draft standard, and follow Human Factor/Ergonomic rules outlined in the ISO 11064-5 standard. 

[Editor's note: In the October 2012 issue, Ian Nimmo will focus on taking advantages of the opportunities offered by following best practices in HMI design.]