The New Methods for Alarm Rationalisation

The Current Problem

Only when you recognise that false alarms are viewed as a fact of life, do you comprehend how serious the issue is. Operating under widened alarms not only has an impact upon control room safety but also significantly impacts production, resulting in financial burdens. Whilst false alarms may seem 'the norm' they are in fact a dangerous space to play in and tolerating their existence has a corrosive effect upon users' belief in the truth of alarms. Danger exists in environmental terms, financial terms and, of course, safety. Companies who don't seek to eliminate false alarms are leaving themselves open for future hazards.




Current alarm setting methods are based on glorified guesswork – meaning annunciations are eventually moved to a tolerable level and do not take into account any business objectives. The GPC method is fast, easy and scientifically based. Alarms can be set to be reliable and actually work in cooperation with business objectives, not against them (click image to enlarge)

But how do you calculate the values at which to set alarm limits?

There has never been a general scientific method to relate alarm limit values back to the process. The EEMUA Alarm Systems Guide states that "alarms on variables reflecting operating limits should be set on the boundary between the normal and upset states of the plant." And whilst this is fine in theory, it neglects to identify how one should actually define or measure 'normal'.

The Current Solution

Available in today's market are software packages which use statistical methods to summarise alarm log data and extract parameters that help understand the scale of the alarm rationalisation problem such as annunciation rates and standing alarm counts. They give no assistance in setting alarm limit values other than to show, one variable at a time, which variables have annunciated most often in the past. There is no relationship to the process itself. So it is all too easy for someone in an alarm rationalisation review meeting to wonder (and wonder) whether and by how much the alarm limit value should be altered to give timely warning of some imaginable event that just hasn't happened yet. And if two people start wondering the meeting can descend into a long opinion-based debate that consumes the time of everyone present and still can't give a 'right' answer.

In practice the operators complain of alarms that annunciate too often. The only solution available is to move the alarm limits 'outwards' so they annunciate less often. But then another alarm becomes the 'too often' culprit and in time its limits are also moved outwards. This leap-frogging continues until it is eventually realised that many of the alarm limits have been moved so far outwards and are annunciating so late that there is too little time remaining for the operator to diagnose and respond to the alarm before the safety systems intervene. Here, an alarm rationalisation project is undertaken to reset alarms and start over. But because the method used is exactly the same as before it is inevitable that the same result will be obtained and plants find themselves repeating this cycle every 5-7 years.

A Much Better Solution

Investigating alarms concludes that the root of all the problems is the lack of any rational method for deciding the values at which alarm limits should be set to define the boundary of the 'normal' state of the plant. The GPC method of identifying the 'Best Operating Zone' is based on past performance and achievement of business objectives and automatically provides a definition of 'normal' and identifies values that define the boundary between 'normal' and 'extreme'. You can quickly obtain much better values for fixed alarm limits (known as high/low and high-high/low-low or warning and emergency alarm limits) than ever before. This process can be completed with several hundred variables at a time so it is very fast and the new alarm limits are consistent with chosen business objectives.

The strength and popularity of GPC is that it is a visual method and is therefore easily understandable and usable by anyone involved with a process plant. It requires no maths on the part of its users. The time savings are considerable when performing the task of finding alarm limit values. Additionally, there are more significant time-savings in subsequent alarm rationalisation review and HAZOP meetings. The systematic use of an easily-understandable methodology for finding new alarm limit values allays the doubts and concerns that previously fuelled rationalisation meetings. Documentation time for COMAH or OSHA compliance is also reduced. Best of all the GPC method optimises your process and has financial and safety benefits for your process.

How Does It Work?

Geometric process control (GPC) is the patented technology which creates scientifically based alarm limits by taking into account process variables and interactions and then mapping the 'Operating Envelope' of the process. The GPC method is immediately different from all previous methods in that it begins by using historical process data that meets or improves desired performance. The GPC method has two steps. The first step finds better, tighter and justifiable values from process historians such as PI, PHD and IP21 for the fixed HiLo and HiHiLoLo alarm limits. Before you put any alarms in place, the system allows you to interactively assess each set of alarm limit values being considered, meaning your alarm rationalisation review meetings become fact-based instead of opinion-based and will be much, much shorter. The much-reduced numbers of false alarms mean that your new alarm limits will be tighter so that they annunciate earlier giving the operator more time to respond and catching process upsets while they are smaller, needing smaller corrections. This first step corresponds to the scope of most of today's alarm rationalisation projects.

The second step in the GPC process is for plants where variable interactions are particularly problematical or where there are many modes or phases operated in a repeating sequence. The method involves creating a better approximation of the operating envelope to generate dynamic alerts for the operator inside the space formed by the fixed HiLo limits (located in step one). This envelope is easily found by wrapping a 'skin' around the desired behaviour and will alert and correct the process anytime it attempts to stray from this desired envelope (and before it can cause a HiLo alarm). This results in improved process safety and economic performance.

Obviously, by having scientifically calculated alarms, false alarms are significantly reduced, meaning that when an alarm does sound it is much more likely to be a genuine alarm and much more likely to be taken seriously.

Business benefits

All process operations are financially driven and you will find that these new alarms help achieve business objectives by keeping the plant operating more frequently inside the 'normal' range. Because there is business value in achieving 'normal' more often, the new alarms and improved plant safety will pay for itself! The business benefits of implementing this technology:

- Significant reduction in false alarms
- Time saving method
- Improves safety
- Improves quality
- Higher process yields
- Less waste
- Achieves business objectives
- Breaking the 5-7 year alarm rationalisation cycle

Alarms are a Journey not a Destination

So you have applied GPC and are pleased with your new alarm limits. If you never changed your products or process again you could say you are finished but in reality you will continue, intentionally and unintentionally, to make small changes to your process so that gradually the definition of the 'normal' operating envelope will change. You will also find that the new tighter alarm limits will encourage better operation so you may eventually be able to further tighten the alarm limits to capture this improved operations capability. Alarms and alerts need to be continuously monitored and revised to match the continuously evolving operating envelope. Who does this? It is usually the unit process engineer within his 'Process Stewardship' role. GPC assists him in process stewardship generally and in tracking the performance of both the process and the alarm system and, because alarm limits can now be considered and change-managed as a set, substantially reduces his burden of administrative work in the change management of alarm limits.