This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.

Baseefa Ltd

EEMUA 191: Implications of Revision 3 on KPIs

22 August 2014

EEMUA 191 Revision 3 has revised requirements in terms of generating and representing KPIs. Alarm Management professionals and engineers responsible for control systems on process plants should therefore be aware of these changes and how to interpret them, says Bruce Nicolson of Intelligent Plant.

On the 11th of September 2013 the 3rd Revision of EEMUA’s (Engineering Equipment & Materials Users’ Association) highly influential “Alarm Systems, a guide to Design, Management and Procurement” was released at a seminar in Manchester. This revision has revised requirements in terms of generating and representing Key Performance Indicators (KPIs). This article explores what those changes are and how the visualisations are affected.

Alarm Management professionals should be aware of these changes and how to interpret them. Users can verify the visualisations with a small set of their own data by visiting: 

www.intelligentplant.com/Pages/Distinguish.aspx or by using Intelligent Plant’s Alarm Analysis.

What is the problem?

Prior to the computing revolution industrial plants would be controlled from large panels fitted with gauges, switches and indicating lights. Often the lights (accompanied by an audible signal) would show an alarm condition, that is, a measured value, which has surpassed a limit and requires action to correct. These panels had limited space on which to place indicators, and there was a cost in adding new ones, therefore careful consideration would be made before additional indicators were added.

With the advent of computerised control systems the cost of adding alarms dropped effectively to zero; every analogue signal could have high and low alarms at the press of a few buttons. Unfortunately, full use was made of this new alarming capability and many unneeded alarms were created. This resulted in operators having to deal with a steady stream of irrelevant alarms even while the process was steady, and floods of alarms when the plant was upset, with important alarms being obscured.

EEMUA 191 addresses the monitoring of these two alarm system states, as described below.

Figure 1 – Steady State Performance Level
Figure 1 – Steady State Performance Level

Performance Level by State

From section 6.5.1: “As noted above, there are two separate alarm situations that define the overall performance of the system:

•  Plant in Steady State operation

•  Plant in Abnormal/Upset condition

“Mechanisms for improving performance in these two situations do differ and it is undoubtedly the abnormal situation that is the harder to address.

“It is sensible therefore to consider performance levels relating to the KPIs for each situation separately”.

The Steady State is measured by the mean average alarm rate per 10 minutes, and displayed in a simple diagram where the alarm system state is indicated [See Figure 1].

Figure 2 – Upset State Performance Level
Figure 2 – Upset State Performance Level

Similarly, the Upset State can be measured by taking the maximum 10-minute period per day and averaging it over the month [See Figure 2].

This KPI is usually displayed in the KPI table, but there is now context as to what this means and how much it needs to change.

High alarm rates during Steady State are usually caused by “chattering” alarms; this is when alarms repeatedly annunciate because the measured variable is close to the alarm setting, either because the setting is inappropriate, or often, when it belongs to a piece of equipment that is currently shut down. De-bounce timers, filtering and dead-band changes can often cure this type of alarm.

High Upset rates happen because most alarms settings are appropriate for normal running, and become inappropriate when there is a plant trip. Low flow, for example, is what we would expect on a pump that is stopped. Generally, logic is required to remove these expected alarms.

System Performance by State

EEMUA 191 Rev 2 defined a grid used to determine an alarm system’s performance [See Figure 3]

Figure 3 – Performance State Scatter Charts
Figure 3 – Performance State Scatter Charts

In Rev 3 this grid has been considerably redefined to better cater for steady and upset behaviour [See Figure 4].

The Steady State chart plots % time > 1 alarm per 10 minutes against the average alarm rate. 1 alarm per 10 minutes is considered manageable and this should be the target we are attempting to achieve.

The upset chart plots % time > 10 alarms per 10 minutes against the maximum alarm rate. 10 alarms per minutes is considered manageable for a short time and again, this should be our target.

The same underlying alarm data is used for both pairs of grids. As well as an overall monthly average, daily values are plotted to give a feel for the range of behaviour that the system exhibits. We have used two charts instead of one to make the pattern of the daily values clearer. From the daily values chart in Figure 4, we can see the system performing well on the many days that have less than one alarm per operator per ten minutes, although there are couple of days where performance is poor and there are seven or eight.

There are now four levels of behaviour compared with the previous five, so for this data the classification goes from mostly robust to stable. This needs to be taken into account if comparing current performance with that documented in the past; it will only be meaningful if the past performance is measured in the same way.

We can also see that the terms “Robust”, “Stable”, “Reactive” and “Overloaded” used in the scatter chart are more or less synonymous with “Acceptable”, “Manageable”, “Over-demanding” and “Unacceptable” used in the Performance level chart.

Figure 4 – Revised Performance State Scatter Charts
Figure 4 – Revised Performance State Scatter Charts

The steady state scatter chart appears to follow a definite arc. This makes sense because as the average alarm rate rises, so will the % time above the target of 1 alarm per 10 minutes.

It should also be noted that in the Upset chart, the first column can only have scatters along the bottom line. This is because the maximum alarm rate is less than the target of 10 alarms per 10 minutes, therefore there is 0% of time above 10 alarms per 10 minutes.

Conclusions

EEMUA 191 rev 3 redefines both the calculation and visualisation of Alarm System KPIs to give more accurate results in a more useful format. Engineers responsible for control systems on process plants should be aware of these changes and apply them as soon as possible to their systems.

About the author:

Bruce Nicolson is Senior Control Systems Engineer at Intelligent Plant. He is an alarm management and audit specialist and has spent 28 years working on control systems across many industries, mainly in the North Sea.


Print this page | E-mail this page

CSA Sira Test