The expert systems which encompass process knowledge, act as “super operators”,assisting and performing all the mundane and highly tedious jobs, such as monitoring of thousands of plant variables, evaluating their trends and interpreting their behaviors and alerting and helping the operator where required. This technology is mature and running in many plants, improving performance and bringing many other benefits to the plant.
The problem – complex industrial processes
Industrial plants are becoming more difficult to control because they are becoming larger and their accompanying processes more complex. When we couple this to the highly competitive nature of business in general, it can be seen that to be competitive and achieve optimality is a major challenge.
The key – operators
The key to the operations of these plants is still the operators and there is an increased burden on operations staff especially in the South African context where there is a serious shortage of skilled labour and expertise. Operators are being assigned larger areas of responsibility as well as many other tasks to perform, and are thus thrust into positions where they are fighting a losing battle against the odds.
A major contributing factor to the operators’ problem is that there are just too many interacting complex processes with too many variables and accompanying issues to watch and control. This is an unintended consequence of installing powerful supervisory control and data acquisition (SCADA) systems and the many instruments available. Compounding the situation is that the processes are also run closer to or in excess of their design capacities with higher volumes as a result of scaling up to achieve economies of scale. The operator is always under pressure to ensure that designed capacities and production targets are at least met if not exceeded.
Adding to their difficulties, is that there is generally little understanding from a first principle point of view of exactly how the processes work, and the understanding becomes less as newer and bigger plants are built. Add in the “just in time” (JIT) philosophy now being applied to continuous processes, environmental compliance, safety and social issues, running a plant is a hugely sophisticated task where one mistake can now be more costly (and deadly) than ever before.
The plant interface
SCADA or distributed control systems (DCS) are in general the interface between the plants and the operators. This is the primary tool with which operators and managers try to co-ordinate and control the processes on their plant.
Although generally quite successful, especially when compared to the previous mimic screens etc, the increasing sophistication of SCADA is also contributing to the operators problems which can be summarised as information overload, where quite simply :
• There is too much information from too many variables given to the operator. For example, the new Profibus type motor drives are incredibly powerful and have a huge number of I/O which are very useful, but in fact during operations confuse the operators.
• SCADA systems generate too many events and too many alarms. The decision as to which alarms are important and which are not, is a huge area of debate in its own right, and has spawned a very interesting side business of “alarm management” or “alarm rationalisation”. The bottom line is that most alarms go unread or are ignored. Only after the fact (usuallyduring the post-mortem) are these huge databases possibly sifted for clues as to what went wrong. This itself can waste a lot of time and cost a lot of money.
• The information as arranged in SCADA graphical user interfaces (GUIs) is not well presented and actually confusing. There are too many screens, too many icons, too many colours, too many trends – most of which are hidden. Even more interesting is the tendency to use the new Windows multimedia capabilities to the extreme which is proving to be totally counter-productive. Arranging screens with trees does help but again this and other GUI mechanisms does not solve the problem of information overload.
• There are too many interacting processes that are difficult to understand. Even the experts have problems here, especially as most processes are multivariable, non-linear
and behave in a non-intuitive way. These process problems and knowledge of these idiosyncrasies are only apparent when the plants are pushed to the edge, and are inevitably forgotten when things go wrong.
• The SCADA GUI’s are naturally limited in physical size and what can be displayed at any one time, and thus the process itself becomes difficult to represent in the GUI. One can compensate with many screens or by huge wall displays, but again the question is ‘who is scanning and watching all those screens and all those wall displays which can span hundreds of metres?’
Compounding the above is the fact that operators are human – their scope of responsibility is often too large, shifts are too long, they get tired, and are not well motivated. There are too many goals – operators must maximise production and efficiencies – all within a multitude of mechanical, safety, and environmental constraints. Thus at best, decisions made to perform basic control (never mind optimise), are in general just sufficient to maintain production, and hopefully meet minimal safety rules, quality control and environmental objectives.
Fig. 1: Logic flow within an intelligent Scada system which uses an expert system to detect events and situations, to reason with the events and then come up with a response such as giving advice or taking corrective actions.
There are many solutions to try and help the operators and improve their performance. These range from better training and education, better SCADA GUI design, a holistic approach to operator wellbeing and motivation, and the use of advanced computer technology. Advanced technology solutions range from the obvious advanced process control (APC) and optimisation systems, which are a high level and provide for the optimisation and co-ordination of the various processes and in some cases the plant as a whole. Others include intelligent SCADA systems based on artificial intelligence.
by Dr. Tony Lange,
Automation & Control Technical Jornal