Making Alarm Management Myths by using DMAIC
A friend sent me a blog posting recently where somebody was trying to allay so-called "myths" of alarm management. I once attended a sales seminar on these types of tactics they called land mines. Like the kid on the playground, the idea was to call you over to the side and mention all the bad things you might hear before somebody else pointed them out, and tell you why they were false. In this way, if another person mentioned them, it blew up in his face by the now pre-educated recipient. I don't encourage this type of practice from our own sales people. We usually give most of our clients the benefit of a doubt in recognizing such sophomoric attempts at self-aggrandizement.
The first "myth" this blogger attempts to expose is: Alarm management is all about counting alarms.
The misdirection is purposeful, and attempts to misinterpret for the reader's benefit the inherent value of logging alarms. It's an easy thing to do when you have a great shortcoming in that area. So I will attempt to clarify- What is being done by alarm logging software is collecting and measuring data-Nothing more. But it is a necessary step to a properly controlled alarm system, and only one part of the life cycle of proper management of an alarm system.
Six sigma concepts recognize that data collection is a precursory step to control of a process. That is why we at TiPS put such an emphasis on that step. If you don't get the data right, it is virtually assured the rest of the process will be more dysfunctional. DMAIC: Define, Measure, Analyze, Improve, Control. I refer you to recent issues of Control magazine where you will see a story by a Monsanto plant where they did exactly that process. See this two-part story at http://www.controlglobal.com/articles/2008/051.html and http://www.controlglobal.com/articles/2008/087.html . In this analysis, the engineer- Brent Thomas- follows the six sigma process to show how an alarm management project is undertaken in a logical and intelligent approach. Whether you're a fan of the six sigma process or not, all engineers who have any scientific understanding know that the six sigma process has its roots in what we in the US learned in grade school as the scientific method.
At Pavilion Technologies, we used to say "In God we trust- all others bring data". And Pavilion made a slogan of the saying "Turning your Data into Gold". What we understood was that in order to control anything you first have to measure it. Our above-mentioned blogger would have us believe that the proper approach is to just go in and re-design the alarm system without ever having first quantified the problem. That's the "M" in DMAIC. Collecting measurement data reveals much information:
- Firstly: Is the problem big enough to justify further efforts?
- If so, just how bad is it?
- Where and when does the problem most exhibit itself?
- Does the data reveal any other outstanding issues that were not initially considered?
- Where should I start my effort?
- What actions are the operators taking to resolve these issues on their own?
- Do I have all the data I need to analyze the problem?
What does the data yield? In Mr. Thomas's case, he shares that much of the problem actually was associated with bad data, and instrument messages. Had they tried to simply fix the alarm system, it would still be just as bad. That doesn't build respect from operators.
Once you have measured the problem, and resolved nuisance issues that stick out like a sore thumb- and these should be resolved just to clean the tables- you can set about making a plan for what you need to do to resolve the issues that made the alarm system get that bad to start with. What are some of those issues?
- Was the system poorly designed? If so- a rationalization may be in order so long as the cost of the rationalization effort does not outweigh the value it will deliver, or the risk it will alleviate.
- Do I have poor procedures for alarm management in place? If so, perhaps I need to update my alarm philosophy, and my procedure enforcement.
- Do things change constantly? If so, then I need to institute some sort of management of change procedures to stop the continuous alterations that affect performance, and eventually lead to incidents.
- Are alarms used to substitute for poor operator communication? These take the form of:
- Poor graphics design
- Poor inter-operator communications
- Poor control room layout
- Poor control panel layout
- Operator training
- etc.
Once you have analyzed the issues (the "A" in DMAIC) that led to a badly performing alarm system, you need to now institute the improvements (I) identified as necessary to resolve the current situation. These MAY come in the form of a rationalization exercise, if a system redesign is called for, or they may take other forms. Often, the information that this exercise yields allows you to make a prioritized list of improvements. If you need help in that area, we have several people who can help you get there.
In the long run, a rationalized system is extremely valuable. Many will tell you the effort itself is an education process and yields greatintrinsic value. However, such an effort runs the risk of dominating the scenery from a management viewpoint, and returning very little obvious value for its effort. In the future, regulatory issues may require a complete documentation of the system, in which case this may become a requirement, but other pressing issues should not go untreated at the expense of the effort.
In other words- use good common engineering judgment to preserve the capital to perform that task when it is absolutely required. Build respect from your management for wise use of the resources they put at your disposal. That will yield returns in the form of political capital that is necessary to succeed.
All roads lead to the logical last step of control (C). Put the proper controls in place to see that what you have fixed does not revert to its former broken state. The proper way to do this is to institute continuous monitoring (usually in the form of KPI's- Key Performance Indicators) that allow you to know when the system is getting out of control, and requires a revisit. Additional treatment is even easier, as all the methods, measurements, and tools are in place for quick updates. These ready measurements yield speedy- even automated- analysis and improvement.
The real secret here is that you don't want to look back at this problem in three years and find out you blew your entire capital on resolving only a part of the issue. Be aware that there are others who will try to convince you to take that path. Avoid that trap and make a few myths of your own.
Comments