Smart Plants have emerged from the diversity of possibilities enabled by the fast-changing pace of modern technology. Having more data, and richer data, at their fingertips offers plant engineers a real-time snapshot with which to optimise profitability and plan maintenance.
If we take a look at the tools of everyday plant operation, there have been significant advances in applications such as plant asset management and predictive maintenance. Automation vendors have also made their control systems and historians very scalable, addressing the need for increasing IO, tags and data handling.
Modern plants present two distinct problems for the busy control room operator:
- Information. Valuable information is spread throughout disparate databases, belonging to different applications and available through a vendor's proprietary software tools. How can an operator make informed decisions, when the information they need is spread across multiple and often inaccessible databases?
- Operator overload. Having more data doesn't guarantee better efficiency and reliability – in fact, the increasing volume of measurements can overwhelm the operator if not carefully managed and presented. Raw data must be transformed into information, by adding both context and usefulness, in order to truly assist an operator.
In recent years, automation vendors have remained competitive by offering additional software applications that hold valuable information proprietary to the vendors' systems. To help the operator achieve a more complete view of the processes, information from various systems and databases must be made available to them at the right time, and in a familiar and optimised interface.
Intelligent software applications can combine, filter, and give important context to data from multiple sources. The operator can then make better decisions, improving what we term "human reliability".
When people are given the tools to do their job more effectively, human error and fatigue are reduced and people are empowered to make the decisions that a machine cannot. Here are just a few of the ways plants are benefiting from intelligent software applications.
The problems resulting from modern alarm systems are very real. Many control systems (SCADA and DCSs) have been rolled out over a number of years and had no consistent approach used to define the alarms. The common result is a vastly over-alarmed system, where truly important alarms are often buried in the midst of trivial status annunciations that do not require operator action.
High alarm rates often exceed the capability of an operator to manage, and important alarms are missed. This common scenario has contributed to a number of industrial accidents over the years.
PAS's vendor-neutral PlantState Suite (PSS) software provides many capabilities to intelligently improve alarm systems. It captures alarm settings and occurrences from all major vendors' automation systems into a consolidated database, logging both the alarms and the operator's response to those alarms over time.
A variety of analyses of both alarm occurrences and alarm configuration make it possible, for example, to identify the most frequent alarms on a system so that their worth or nuisance behaviour can be addressed.
Simple work practices to address as few as 30 to 50 alarms can usually cut overall alarm rates by up to 80 percent. A properly operating alarm system is a key factor in supporting operator performance.
Figure 1: This illustrates important process boundaries normally isolated in several different databases.
Intelligent software makes it possible to identify situations that may be very difficult to ascertain using a control system's standard toolset.
For example:
- Identifying exactly which alarms are suppressed, and when and under what circumstances that suppression occurred. It is common to find significant alarms that have been improperly suppressed for long periods of time.
- Identifying alarm setpoints or priorities that have been changed compared to their appropriate rationalised settings stored in the master alarm database. Failure of Management of Change systems is common.
- Identifying which control loops are in MANUAL rather than Automatic mode (and thus not fulfilling their design function.) Many loops simply will not operate in the Auto mode, often coming as a surprise to the control engineers.
- Identifying which controllers have the most manual setpoint or mode changes, indicating failure to operate as designed or in need of redesign.
Proper software assists plant engineers in identifying the most significant issues with the control system, and helps managers prioritise the response of limited automation resources. Such software applications help establish best practices, such as facilitating a proper handover between shifts with reports that clearly show the current control system status, and highlight otherwise hidden changes such as suppressed alarms.
The appropriate alarm management software can perform in both an analysis and supervisory role. Proper Management-of-change (MOC) to certain control system settings should also update the master database. The software can both audit the settings of multiple automation systems against these documented and proper settings, and also, when necessary, enforcing the master settings back onto the system, ensuring compliance with MOC.
Boundary-aware
Alarm management is an important area where "joined up thinking" can provide operators with crucial information when they need to make safety-critical decisions. Often, equipment design data, alarm settings, and equipment trip settings are located in totally different databases, unavailable to the operator and certainly difficult to use.
By aggregating all of the alarm settings and other operational boundary information, such as SIS trips and equipment design constraints from their respective plant databases, equipment status relative to important boundaries can be clearly depicted.
Using the context of a boundary hierarchy, vulnerable areas are exposed and the operator has better information to prevent the system tripping and to keep productivity optimal. Such techniques can also visualise to plant management whether the process is being operated within acceptable, or allowable, or safe boundaries.
By depicting the process conditions relative to these limits, problems with the operation of both the control system and the safety system can be revealed at a glance. Some or all of these settings may need automated adjustment based on the dynamics of the process. Intelligent software can further help by managing these dynamic adjustments across platforms.
By supervising alarms across heterogeneous plant automation systems, software can also automatically detect and report any deviations to this boundary hierarchy. This provides additional assurance to keep configuration parameters such as alarm limits and instrument ranges within the safe operating boundaries of the plant.
Operators are provided with an HMI to monitor and interact with the control system. However, most process graphics used by operators today are simple P&ID views covered with hundreds of numbers, a depiction that provides for the operator quite poor situation awareness and abnormal situation detection capability.
Figure 2: Depicting information rather than raw data.
In Figure 2, a single glance resolves a compressor's operational status, clearly and quickly identifying the normal, abnormal, and alarmed conditions using the boundary hierarchy previously described to provide context. Such depictions make it possible to for an operator to monitor hundreds or thousands of individual sensors and quickly identify problems before alarms even occur.
The knowledge of experienced operators and engineers is embedded into the depiction. It is clear that there are merits in integrating pertinent information from multiple sources for the specific purposes of alarm management and high performance HMI, but the next big opportunity for smart plants is to consolidate the information housed in disparate databases throughout the plant so that knowledge is captured and shared.
In tomorrow's smart plants, operators will be able to add and link information about upsets, malfunctions, incident reviews, procedures, work orders, and similar experience-based knowledge to the equipment or asset.
The information housed in multiple disparate databases will become searchable as if it were in a single database, and the search will be filtered so as to be relevant to the task currently being performed. The new technology that enables this capability is a context-based process knowledge and decision-support system.
[Mark Tibbitts is Director of Operations Effectiveness, PAS.]