Can DCS HMI impact operator performance
Many recent accidents have clearly identified the contribution of bad alarm management practices as a major contribution. Action has been taken by regulators, standards bodies and customer forums to provide good guidance on alarm improvement, targets have been set through organizations like EEMUA who have effectively raised the bar in all industrial sectors. However, many struggle still with alarm management, especially alarm floods, and will continue to, until they address their HMI issues.
Little has been published on the contribution of poor HMI design but the significant contribution to major accidents and losses is evident in the accident reports. Common themes being loss of the big picture, data overload, missed information or alarms, operators being reactive and waiting, they are operating by alarms.
The continued use of crowded graphics with no color restrictions producing visual noise is still not on many managers radar. Even though best practices and use of grey-scale graphics have been identified as a major step forward and have demonstrated the ability to improve an operators performance to detect, diagnose and recover from abnormal situations.
A new topic to our industry but not too many others is Situation Awareness and how alarms are just tool to help an operator become aware of change, this topic looks at our past and demonstrates how our operators' HMI used to be less dependent on alarms through the continuous monitoring of plant and process trends and the ability to see the "big picture".
We have the technology today to address these issues, our DCS systems are very capable but are being abused by over use of color and fancy three dimensional graphics that are compromising situation awareness.
We will never resolve alarm management issues until we get the balance right and focus attention on good HMI practices, many now have a written alarm philosophy, how many have a HMI philosophy. To mitigate major accidents and losses we need high performance HMI that truly impacts an operators performance. What we have today does not work and has proven that it exposes us to unacceptable risk.
- Login or register to post comments
- 5003 reads
- Permalink
- 14 comments
The need for immersion
My experience with plant operators is that they fall into two ends of a spectrum and whats in between are well 'in-between'. On one side are a few operators when asked whats going on in the plant would tell you, everythings ok and under control but will take their eyes of the screen for only short periods of time. They are the ones you will see flipping thru the screens endlessly and making slight adjustments here and there. On the other end of the spectrum are operators that would tell you a whole bunch of parameters and instructions and will fall into a deep conversation on the politics of the day if prodded. I call the first group of operators "immersed". These are the ones who know, no thats not the right word, who can almost feel the plant. One such operator i was discussing with called his unit (a series of hydrocarbon distillation columns) a "baby unit", simply because he needed to coax it all the time to get things optimum. Or another operator who after I had implemented an advanced control application on a deethaniser tower, switched off my application and stabilised the unit to an extent that not I, nor any APC application or any other operator has been able to till today. Sadly he left for greener pastures. These are operators who are immersed in their unit that should it get a mild temperature, they would know immediately and provide the appropriate medication. Sadly, with high attrition rates in the industry, these type of personnel disappear. What we need is a means via a very strong study into the subject matter to come up with a means to develop such immersion attitudes. I believe that DCS HMIs are one of the key elements but not the only one. The number of such HMIs, the use of trending and graphics, the use of actual plant pictures (maybe), the use of inferred properties (even in the presence of analysers), the use of online operator guidelines/procedures, etc and even down to the use of psychology i.e. the right attitude, once you sit at the panel, all of these and more would create that "immersion". Frankly, HMIs are only one cog in the whole machine.
Function follows form
You've all probably seen the old saying, form follows function. The inverse is true as well. Function follows form. If the only tools an operator has to use are P&ID type graphics, and alarm banners and lists, the operators will run loops, and respond to alarms. Maybe a better analogy is "if all you have is a hammer, everything looks like a nail".
If you want operators to optimize process performance, give them process performance information. If you want operators to optimize process economics, give them economic data.
The other important aspect to improving operator performance is feedback on the effectiveness of their actions. In addition to current performance or economic information, they need to know if things are getting better or worse, and they need to know if their actions are driving the change. Finally, they need to know if their actions are impacting performance, economics, safety, or other key criteria for other parts of the plant. We should probably stop thinking of them as operators and start thinking of them as key performance managers.
Thoughts?
TomW
Tom, I think you've nailed
Tom, I think you've nailed it (pun intended). Engineers, in designing HMI, need to be talking to the process engineers and plant managers and to understand what production, economic, and safety factors are important - along with proper metrics and guidelines for each.
We've approached this in the HMI by creating overview displays with symbols/shapes/dynamos that look like a road (where you have to keep the # between the ditches) and like a speedometer (where you're trying to maximize the value). The operator needs to know when he's doing well and when he's pushing his limits.
Operator feedback to actions
Yes, Tom has nailed it. And it is about time that the DCS suppliers understood and translated this to tools off-the-shelf to help us with this. For instance...
How many plants have situations when the Operator has made a mistake on data entry – a recurring theme I would think? Sometimes there might be follow-up with learnings and recommendations, but more often than not, unless there was a material impact to production, the incident is quietly ignored. The recommendations that I have seen typically talk about being very careful when making SP, and OP entries before hitting enter to ensure that the value being entered is what is intended (and the controller is the correct controller). In general, with the volume of # moves made on the plants, the % of errors in data entry are very low… but can we do more?
My idea is that when an Operator makes a move in SP/output/etc. that this tag/controller gets stuck on a sort of bulletin board/trend tool such that when they move on to do another task, there is a list or history of recent actions and trends associated with those variables. In this way, while the Operator has moved on to do other tasks on other displays, there is a mechanism to continue to monitor what they did in recent history. ie: SITUATIONAL AWARENESS.
It would be ideal if we could link the action to the process variable that they were hoping to see the response on…(eg: a manual increase on reboil to increase tower bottoms temp and therefore the separation in the column and economics of the slippage that they are really after), but I have no idea how to make these connections.
I believe that it would be technically feasible to get the DCS systems to manage a list of the last “X” actions on a bulletin board/trend concept via faceplate activity. We just need someone to expend the effort. I know in the facilities that I am responsible that have graphics that are set up based on ASM principles we are well positioned to do this now. Sadly, the off-the-shelf tools that I know of are not.
As an aside – as the screen technology moves toward 16:9 aspect ratios instead of 4:3, and forces our hand to move with it… maybe the extra screen real-estate could be leveraged for this concept…instead of distorting the image to make everything seem fatter like everyone seems to think is acceptable on TVs.
I like that idea of seeing
I like that idea of seeing process history of each point embedded in the faceplate or the sidebar. I think I could implement that on Experion or DeltaV relatively easily. We're still deciding what to do with that extra space on my current project - maybe I'll see how easy that is to flesh out.
I'm hoping as well that some of this functionality gets baked into products. These could be widgets, essentially, where you pick and choose the functionality that you want. Could be something like this, could be trends or "small multiple" trend groups as well - where the operator can change graphics without losing track of a slower process. There's a lot of power in that sidebar space.
Prevention is Better than Cure
An operator on a plant that I used to work on once rubbished the great new grey-scale graphics that we had just implemented following the best ASM Consortium guidelines.
"These screens just make me go to sleep. Then is all lights up like a Christmas tree when it's too late!"
Don't get me wrong - being able to respond quickly and effectively to a plant upset is a crucial element of an operators job. But its not the ONLY part of his job.
I would love to see the industry wake-up and realise that it is the process operators who make (or loose) all the money that the process industry lives from. We need to start helping them do their every-day job of optimizing the process and making money. The time is right for a second revolution in HMI design, this time focussing on "Normal Operation Optimization" - there are some great visualization techniques available that could transform the effectiveness of operator display and interfaces for normal operations.
The great work that has been done on abnormal situation management has undoubtedly made plants safer and reduced major production losses, but if we become too obsessive about it we forget what the process operators are really there for!
New overview article on HMI best practices
Doug Rothenburg just penned an excellent overview of HMI design best practices in the August 2009 issue of Control magazine. Here's a link to the article on the website:
http://www.controlglobal.com/articles/2009/PlantOperatorsEyes0908.html
New overview article on HMI best practices
Doug Rothenburg just penned an excellent overview of HMI design best practices in the August 2009 issue of Control magazine. Here's a link to the article on the website:
http://www.controlglobal.com/articles/2009/PlantOperatorsEyes0908.html
See also the thread on the Implementation section
There us a thread called "Best practices for deriving operator graphics from P&IDs?" on the Implementation section that is very much related.
Personally I think both threads belong in Design
Francis
Francis
www.controldraw.co.uk
COPS study on Operator Effectiveness
I took a brief look at a Center for Operator Performance study in a blog post.
One of the interesting things they noted in the study was, "The researchers also made another observation; color is probably not the most pressing problem, a bigger factor is the overwhelming number of displays and the design and presentation of the information on these displays. For example, although mimic displays such as P&ID's are simple enough to create, they are not necessarily the best way to present information to operators."
This seems to amplify Aaron's point that what's on the display, and its simplicity has a great impact on effectiveness than the colors in the display.
I have done migrations where
I have done migrations where the exact layout and overcrowdedness of the graphics is preserved but the colors have moved towards the grayscale look. That just doesn't get it done.
I think that the operators should have a lot of input in what the new graphics look like, but I would start with how they operate the plant, which units go down together, which numbers they constantly watch, and what the common problems and complaints are and go from there. The results are well worth the investment, in my experience.
Limited by knowledge
I too agree that operators should have an input to the layout and form of the graphics. After all they are the operator's eyes and ears to the process when sitting in the control room. However I would say that we as engineers are mostly keeping up with trends in HMI screen design but are the operators aware of these trends. They only know what they know......their knowledge is limiting them so they need to be made familiar with all their options. This in turn allows them to grow those ideas to suit the plant that they are operating in the industry they are working.
Each industry application has their own performance criteria, key information and unique operating conditions. Whether you are in meat processing, milk powder plant, pharmaceutical, petrochemical processing or other industry the operators need to see the most relevant information to allow them to get the best out of their plant.
Any guidelines provided for the future of HMI screen and background applications need to be generic enough and flexible enough to facilitate the best format and content creation for each industry apllication.
The necessity of HMI screen designers to show all process pipes like a P&ID is helpful if you are new to a process and do not have current copy of the P&ID to view. If however you have a screen that contained your specific key performance criteria and warned of trending that indicated plant or equipment failure then this would be more useful for operators.
This leads me into maintenance of plants. It can be common practice in certain industries that HMI screens show process information and are designed for optimising process performance but without any regard for maintenance support. Information is often available that can indicate plant and equipment failure but all the operator may see is a control loop or system that is trending away from optimal. With a few tweaks this can be corrected but this only masks the potential problem developing.
A wholistic approach to HMI screen and background system design that is adaptable over a range of industries (simple to complex) is needed. It needs to allow the integration of information from all measurement sources and the dissemination of understanding or interpretation of this information for management, operations and meintenance alike.
You may well say that you are already doing this because you have huge capital or maintenance budgets and you are following the trends in the industry. But for those whose budgets are tight or limited, any usability project must consider development of guidelines that allow its application over all industries.
The combined wisdom of all stakeholders across all industries in creating these guidelines will provide the best outcome for all who need or desire the guidance.
Migration Projects and HMI Best Practices
How about the number of migration projects where the old graphics are preserved? May users demand automated graphics conversion tools so they can get the same look and feel of their old graphics in their new system. Presumably because the operators feel more comfortable with these old graphics? It still amazes me how much energy is put into preserving the "old way" of doing things.
Operations doesn't want change
My faceplates are DeltaV analogs of faceplates from the TDC-3000 / TPS upgrade we did next door 8 years prior. These were analogs of the Bailey Net 90 faceplates from an across-the-site upgrade that was done in the 80's, whose faceplates were made to emulate the Foxboro "Microspec" faceplates it replaced . . . so my 2000-vintage faceplates are direct descendants of the 1980-vintage Foxboro system.
I was at another site where operations required the graphics of the new DeltaV system to be identical to the old Centum system it was replacing. No consideration was given to how the graphics might be improved or optimized. While there is brilliant work IMO being done at consultants and elsewhere, the problem may be that our "customer" = operations doesn't want it. Many of us are still dealing with vestiges of prior generations, and the less-than-optimal "same as it always was" is nearly always the chosen path. Even dysfunctional work-arounds that have persisted for years are preferred to any "change".
Whenever I revise a graphic to have less-meaningless-color (move toward gray scale) all I get is bitter beer face.
John Rezabek
ISP Lima LLC