Control loops - pleasure or plague?
A plant floor view within the process industry
Alexander Horch
 1 Cost distribution of a control loop.  2 Loop ranking overview for a typical plant in the process industry. CLCM works like a doctor’s stethoscope: it obtains a diagnosis by passively listening to the process. Typically, no more information than standard DCS tags – setpoint (SP), process variable (PV), controller output (CO) etc. – is required. In a typical production plant in the process industry, for example, there may be up to several thousand control loops. 2 shows a typical and important loop performance ranking result, including typical data for each category. The need for CLCM? In helping to resolve this issue considerable research has been carried out to develop a holistic and non-invasive methodology that can automatically assess controller performance. An overview of ongoing research is presented in [3]. The most obvious and serious control loop problem is a persisting oscillation. Reasons are manifold: bad controller settings; external problems; valve friction; equipment failure; or process-related reasons. Irregular deviations from targets are more difficult to analyse. Luckily nowadays, oscillations and poor performance can be automatically detected – with help, for example, from what is known as the Harris Index [2]. However, the main challenge of diagnosing bad performance remains. CLCM typically focuses on basic control loops that are vital in achieving the targeted product quality and plant performance. However, in situations where highly advanced control loops are used, more advanced supervision functionality is needed. Advanced control (eg, a model-predictive controller) relies heavily on the assumption that the underlying basic control loops perform satisfactorily. CLCM ensures this requirement. CLCM in industry This interest is also an effect of the more general trends affecting asset management. These trends have been recently published by the ARC group [4]:
Modern CLCM tools strive to support these trends. In fact ARC recommends the combination of control loop condition monitoring with a controller tuning tool. ABB has adopted this idea and integrated both functionalities into what it calls the OptimizeIT Loop Performance Manager (LPM) tool [5]. What’s on offer from ABB ABB offers CLCM functionality on different levels of it’s IndustrialIT automation palette 3.  3Automation architecture indicating controller condition monitoring functionality (red arrows). OptimizeIT Loop Performance Manager (LPM) ABB’s OptimizeIT Loop Performance Manager (LPM) is a general and powerful tool for controller performance condition monitoring. It combines both control loop assessment and controller tuning functionality, and runs with any automation architecture via OPC data connectivity 4. The latest version also includes a Plant-wide Disturbance Analysis module which has proven to be able to locate plant-wide disturbance root causes very successfully.  4 ABB Loop Performance Manager (LPM): loop auditing window. LPM’s control loop auditing not only indicates the best and worst performing loops in a plant section, but it also gives detailed analyses on how to remove identified problems. These problems include discrepancies in the final control element, external disturbances, and controller tuning. Controller hardware: ControlIT AC800M  5 Oscillation detection and stiction compensation function overview in the 800M controller. The process value exhibits typical stiction behavior. The controller can detect sticking valves and apply a stiction compensator algorithm to guarantee best possible controller action until the next valve maintenance. System 800xA: Asset Optimization and control loop asset monitoring ABB’s 800xA Asset Optimization System includes fully automatic loop monitoring functionality via a so called “Control Loop Asset Monitor”. By that the detection and diagnosis of control loop problems is fed into the asset optimization data handling of the 800xA system [7]. Messaging, connection to the computerized maintenance management system (CMMS), and access to historical data and other real-time plant information helps the user trace problems and initiate corrective actions. Industrial applications A simple example is to compare the high-precision position controller in a disc drive with a surge tank level controller in a paper mill. Obviously, both controllers share the same task but their respective benchmarks should be set on two different scales. Consequently, some of the control performance monitoring methodology would fit the first application, some of it the other. Since control performance monitoring traditionally originates from the process industry, most established methods focus on the problems that are typically encountered in this industry. Diagnosis of controllers in the (petro-) chemical and pulp and paper industry 6 shows a subset of data before and after a performance improvement initiative in a pulp mill. CLCM methods detected oscillatory control loops, and experiments verified the diagnoses. The subsequent improvements are obvious from the data collected later.  6 Improvement of controller performance after application of CLCM with subsequent maintenance. Diagnosis of controllers in power plants One important point is the ability to classify CLCM results by the current load situation in the power plant. Controller behavior is typically a function of the load (eg, high, low, start-up, load change) or of other properties like raw material type etc. Modern CLCM methods do consider such conditions. Diagnosis of control-relevant disturbances in cold rolling mills In the rolling mill industry, a few highly sophisticated control loops are needed for tension and thickness control. However, the application of standard CLCM methods in the rolling mill industry is perhaps not as straightforward as in the chemical industry, as recent applications of these methods have produced results that are difficult to interpret. On the other hand, where CLCM functionality has been specially designed for rolling mill applications, the results have been very encouraging. To be more specific, CLCM functionality has been successfully designed to diagnose and remove periodic disturbances which are predominant in rolling mills [9], and a typical automated diagnosis screenshot is shown in 7.  7 ABB OperateIT screen for the diagnosis of periodic disturbances. Beyond single-loop controller condition monitoring CLCM is able to detect loop performance deficiencies and can contribute to substantial gains once the appropriate maintenance actions have been taken. However, there are cases where the plant is not properly optimized even though the controllers are performing well. In such cases, it is highly probable that the current controller structure is not sufficient. A systematic and fast way of assessing the controller landscape and the prevailing automation infrastructure is by using a benchmarking service provided by ABB [10]. Conclusions  ABB’s research and product variety has enabled a flexible application of CLCM across many different industries. These applications have been adapted so that the existing hardware at a customer site is used. CLCM can be applied to any process architecture regardless of whether ABB’s 800xA System or a third-party DCS is installed. Alexander Horch ABB Corporate Research, Ladenburg, Germany References
|
||||||||||||||||||||||||||
|
