PubMed Health⌕ Search

Biomedical subjects

E H Bristol

Publications and source records attributed to E H Bristol.

2 recordsLinked to original sources

Intent-based process control configuration models.

Process control application engineering would be substantially less expensive if supported with readable self-documentation. An important advance would be models and languages designed to more clearly represent application Intent (as formalized herein), without the usual implementation obscurity. This paper analyzes how Intent, in the sense later defined, can be usefully defined and expressed. Process control has traditionally been defined in terms of a number of automation levels. This supports the intended Intent concept in one way, defining the implementation of higher application goals in terms of lower level ones. But a different, even more useful model is the leveling of physical sciences where each level of problem is best addressed by an appropriate kind of theory, itself dependent on lower level theories. Each theory must be proved by more fundamental theories even though those theories are much too complex to address the higher level problem directly. But more than leveling is needed. At each application level, the associated language should support concepts that make its normal usages clear. It should ensure that appropriate application practices can be expressed transparently in terms of their Intent. It should allow the engineer to clearly relate the result to the expected implementation, allowing him full control over the application. The paper will expand on prior papers to show more generally how these concepts can be developed.

Journal Article↗

Improved process control alarm operation.

Alarms are the main connection from the automation to the operator, when addressing process operation outside of its normal function. They are often as much a source of operator overload and consternation as help. Better engineering of the relative role of the operator and automation would materially help overcome the difficulties. Expert systems have been proposed as a solution. But Expert systems are really another form of automation. There remains that aspect of the alarms, which must address our inability to cover and understand a possibly larger domain of the operation not appropriate to traditional controls or present-day automation. Appropriate tools for this domain must support operator discretion and initiative. The paper suggests a set of such general, computer science based, tools requiring only the most basic configuration. They are viewed as implemented on top of those properly detailed alarm displays and interlocks, which reflect the more formal plant operating policies. They include: (a) Various forms of alarm logging and trending; (b) Short, automatically generated, word summaries of alarm activity, which allow low level data to propagate to the highest levels, including: one word and priority summaries; (c) Causal alarm pattern analyses that help the operator to predict or diagnose alarm behavior; (d) Automatic adaptation of alarms and alarm limits to varying process situations; (e) Uniform use of alarm policies to simplify alarm configuration.

Journal Article↗