Computer-aided design and implementation of interlock control code

The design of the basic control of a chemical plant is mainly based on the P&IDs (pipe & instrumentation diagrams) of the plant. These diagrams, which are complemented by further documents, contain the information which is needed to specify the sensors, actuators, and the binary and continuous control functions in between to run the plant at steady state and to ensure its safe operation. In the past, P&IDs have been available for control engineers only in the form of paper drawings, which required tedious manual analysis before the control specification and implementation could start. Today's computer-aided engineering tools, as used by the process and plant design engineers, provide P&IDs in electronic format, but the control engineering workflow has not changed. However, the CAE tools' underlying database provides the basis for a computer-aided and more efficient control engineering process: the contents of the database can be parsed by computer algorithms which search for well-known patterns, e.g. certain combinations of plant objects (tanks, pipes, sensors, and actuators), and which assign well-known control solutions to each pattern found in the particular plant. Thus, repetitive, tedious, and error-prone engineering tasks can be automated. This paper describes an approach how this can be achieved, based on a standardized plant description model called CAEX and on rule-based algorithms. The concept is described taking the specification and implementation of interlocks as an example. Interlocks are pieces of control code that ensure the safety of a plant. Faulty or "forgotten" interlocks may have a severe impact on the integrity of the plant and the process running therein, thus their development requires significant engineering and test effort. Since CAEX contains all necessary plant details to find and implement a certain set of interlocks automatically, an algorithm parses the CAEX file and generates a table of causes (process events, detected by sensor signals) and effects (process actions, executed by actuators), which can be automatically converted into control code. The automatically-generated C&E table can also support HAZOP studies. The authors have designed and developed a knowledge-based system which implements these above mentioned functions