Plant operators should be able to rely on process alarms to provide essential information for running their units safely and steering them efficiently through critical operations such as start-ups or upsets. Unfortunately, many incident investigations have shown that quite often this isn't the case. Missed or misinterpreted alarms can contribute to the occurrence of incidents.
Proper management of an alarm system by operations personnel is crucial to achieve quick and accurate detection, assessment and resolution of abnormal operating conditions. To be a viable tool for operators in these situations, alarms must be defined and meticulously configured according to the guidance of EEMUA publication 191 and ISA SP 18 (see: "Avoid the Domino Effect"). This implies that a site has defined an alarm philosophy for its units and rationalized its alarms according to this philosophy. When these alarms are subsequently implemented on the plant's distributed control system (DCS) in many cases management feels it has done its job in providing a well performing alarm management system for operators. That's a mistake!
Human error and a constantly changing plant environment (quality, throughput, expansion, equipment alterations, etc.) can contribute to incidents. So without regularly monitoring alarm performance and making necessary adjustments to alarm set-up, the value of alarms will diminish and alarm performance will deteriorate. This is especially true where there's no ownership by the main stakeholder — operations. A good way to address such concerns and get ownership is to appoint an alarm champion. This person's primary responsibility is to identify and resolve outstanding alarm issues. The champion's job isn't necessarily to correct these issues but to get appropriate personnel involved in addressing the problems.
Recognizing a Need
LyondellBasell values effective alarm management and the importance of monitoring the performance of the alarm management process. In late 2006 the company implemented Logmate, an alarm management application from TiPs, across its manufacturing sites. After reviewing the first months of data for alarm loading and key performance indicators (KPI), the company began discussing the need for a resource to lead and promote the alarm management process. The Corpus Christi, Texas, site was one of the first locations to recognize this need; a control specialist was tasked with monitoring the KPIs and presenting summary information of bad actors and alarm floods during regular meetings. Over the course of the next year, the scope was refined using input provided by personnel from several sites that had adopted the alarm management initiative. Staff now serves as alarm champions at these production sites and the company has begun rolling out the program to other sites.
In many cases the first alarm champions were people from the local control group because they were involved in the initial implementation of the alarm monitoring application and the initial bad actor mitigation. Nevertheless, the program is more effective if sites recruit the alarm champion from operations to promote ownership. A process or production specialist, or even an experienced lead operator, can handle the role.
The initiative does not create another full-time position. Instead, the role of alarm champion is combined with an employee's other responsibilities; this is made possible by providing the necessary infrastructure for easy alarm performance monitoring, along with automated reporting, easy drill-down and troubleshooting. Establishing a consistent alarm philosophy with support from central engineering also helps this effort. Initially, alarm champion duties will require additional time and effort to establish the program and to address shortcomings and flaws of existing alarm management. Depending upon the conditions of the existing alarm system, responsibilities of an alarm champion should easily fit into the daily routine after completion of the initiation period.
Assigning ownership of alarm management to one person is essential. Ideally, the person should become involved early in the process of establishing proper alarm management at the production unit. However, this isn't a requirement for success.
Establishing alarm management at a production site essentially involves three phases: implementation of alarm monitoring and documentation; maintenance and support; and continuous improvement. Involvement of the alarm champion increases with each phase.
Initial implementation of an alarm management application requires activities that typically are performed only once, including:
• Collecting data and benchmarking current alarm system performance;
• Establishing standard reporting, KPIs and easy-to-use drill-down tools for troubleshooting; and
• Validating the current alarm database and documentation for import into an alarm knowledge base (AKB) under management of change (MOC). The AKB is the master alarm database for a unit's alarm settings.
The local control systems group, with support from the global center of excellence, usually handles these tasks. If an alarm champion already has been selected, this is an excellent opportunity for that person to review and become familiar with current alarm management philosophy and company standards. Efficiency in this phase is achieved by using best practices and standardized tools through central engineering.
The alarm champion is much more involved in maintenance and support of the alarm management system. However, software upgrades or resolution of communication issues between the DCS and alarm database aren't the alarm champion's responsibility, but are managed by the local control systems group or central engineering. (The smaller the site, the more support from the corporate center is needed.) The alarm champion should take the lead in maintaining alarm documentation and auditing, and enforcing implemented alarms on the DCS.
The foundation of alarm documentation should be an alarm rationalization where each individual alarm is discussed, defined and prioritized by a team. This team usually consists of one or two experienced operators, a production specialist or engineer, a control specialist or engineer, and, as needed, a maintenance engineer and other specialty engineering personnel. The alarm champion's role is to participate in alarm rationalization and educate team members on the guiding alarm philosophy.
An important component of alarm rationalization is identifying causes, consequences and corrective actions for alarms. To be valuable, these must be documented in the AKB and operators must have easy access to this electronic documentation. The alarm champion is responsible for ensuring that the documentation is up-to-date and prospective changes follow MOC procedure. The alarm champion also must ensure that the current DCS alarm configuration is regularly audited to verify alarm settings match the current AKB.
An important and sometimes underestimated responsibility of alarm rationalization and of the alarm champion is proper alarm configuration. This involves defining process conditions that should trigger an alarm and consequently setting the alarm limit or set point, any delay and dead band to avoid alarm chattering. Actual implementation should be the responsibility of control systems personnel.
Another potential important duty for an alarm champion is participating in process hazard analysis (PHA), process change analysis (PCA), safety integrity levels (SIL) and other investigations to ensure new and existing alarms are properly evaluated and rationalized.
Overseeing the continuous improvement process of alarm management is the responsibility of the alarm champion. This involves very rigorous data mining and analysis, identifying and addressing problems, and following up to confirm resolution of issues.
To facilitate the monitoring of alarms, we have defined the three KPIs and associated goals for alarm management (based on EEMUA 191 and SP18 recommendations):
• Average number of alarms per hour per operator shouldn't exceed six;
• Average number of standing alarms (those active for more than 24 hours) shouldn't exceed nine; and
• Peak alarm rate per operator (defined as maximum number of alarms during any 10-minute period within a month) shouldn't exceed 10.
These KPIs are monitored 24/7 and automatically calculated.
In addition, automated standard reports and drill-downs should be established to assist the alarm champion in continuous review and analysis of alarm data for patterns and inconsistencies to focus efforts on problem areas. Upset conditions and alarm floods should receive particular attention. During those periods a wealth of information is created that should be used to improve the performance of the alarm management system. The importance of providing easy and automated drill-down cannot be overstated. Expert alarm champions use this opportunity to identify causes of alarm floods and nuisance alarms to correct these issues.
In addition, expert alarm champions should explore options for real-time alarm management provided by their specific automation environment. Many modern control systems provide capabilities such as state-based alarming, in which alarms and their set points are configured for different operating conditions (start-up, shutdown, product switches and normal operation), suppression of alarm floods by conditional alarming, or shelving of alarms because of broken sensors (see: "Consider State-Based Control", and "Adroitly Manage Alarms"). This can go as far as shelving, subsuming and predicting alarms based on real-time data and even creating new rules based on alarm data or on information stored in the causes, consequences and corrective actions of the AKB.
Adoption and implementation of these advanced applications require the alarm champion to engage in additional activities:
• Support the advanced alarm management application, specifically act as interface to operators;
• Monitor a new KPI for the reduction of alarms;
• Check for newly identified alarm patterns or predictions;
• Validate and get approval to implement (via MOC) new alarm rules and predictions;
• Activate approved alarm rules and predictions; and
• Ensure no independent protection layer alarms are being shelved or subsumed.
These are areas where an expert alarm champion can excel and make a significant difference by producing a step change in alarm management performance. Ultimately, this helps board operators effectively manage critical upsets and avoid serious incidents.
The work of alarm champions has already proven to be beneficial at LyondellBasell early adopter sites. At one of the company's Rotterdam units alarm loading and peak alarm rates have dropped by more than a factor of two. At one of its Houston units the average alarm loading has decreased by a factor of four, while at two other Houston units the number of standing alarms has halved.
The company fosters experience exchange through a specifically created Critical Condition Management user group. In addition, we use a center of excellence to provide training and ensure consistency.
LyondellBasell has learned what is required to become an effective alarm champion. The person must understand and buy into the alarm philosophy and sometimes challenge operators or engineers as needed. Good communication and leadership skills are essential for building positive relationships with operators and maintenance staff. Of course, support by local plant management is key to alarm champion success.
LOTHAR LANG, Ph. D., is a consulting engineer for Lyondell, a LyondellBasell company. E-mail him at Lothar.Lang@lyondellbasell.com.