Coal Preparation » Process Control
The current advances in electronics and smart sensors, coupled with the large amount of information that modern distributed control systems can create, provide opportunities but posses some significant problems.
The potential suite of data measurements could provide plant operators, maintenance staff and supervisors with a comprehensive understanding of the current 'health of the plant'. Analysis of this issue would also provide a tool for the recognition of where important data is not or poorly (timeframe and/or quality) currently available.
The impediments to maximising the use of this information are:
· The large quantity of the available information;
· The limited time available for interpretation and analysis of this data.
The first step was to develop an approach which would allow the large amount of information to be categorised in a simple form but would assist in the process of establishing relationships between them. The project then examined the type and applicability of data generated across a coal preparation plant.
· A suitable system for categorisation of the information associated with the description of the 'Health of a Plant' has been developed.
· A relational data base model for these categories was derived.
· The process and performance information relationships were established via the use of models derived from the wide body of literature available. The main purpose of the relationships was to identify the important factors so :
o Those which require monitoring can be clearly identified and the reason and level of accuracy required for monitoring defined;
o Identify areas where the model is weak or needs validation; and
o Define research required to improve the description of the effects.
· Given the availability of the above model relationships and measurements, the best way to utilise this information in a simple intelligent manner was addressed. It involved the construction of a high level fuzzy set diagnosis chart and an underlying set of unit operation diagnostic charts.
· These charts provided the basis for the implementation of a generic diagnosis system. This was deliberately developed in EXCEL so that it can be used and/or modified to suit a particular plant by a wide range of interested parties.
· A sensing system which combines a limited set of measurements with an algorithm or logic system for optimisation of a process can be termed a smart sensor. These are vey useful in the optimisation of difficult process situations, and can be used to supplement expert systems. This is particularly the case when there a significant nonlinear behaviour between control factors and performance related factors.
· It is believed that the models developed in this project can also provide the basis for appropriate smart sensors when access to appropriate measurements is available. An example of this type of development using the spiral model developed in this report and data from an Australian plant.
The analysis contained within this report should be considered as a working document. As better understanding and technology becomes available it should be reconsidered and revised as appropriate.
The current relationships between the factors should be regarded as a working hypothesises and should be the subject of appropriate experiments to verify or improve the description of the process.