Fernando Roumiguičre

A fundamental requirement for optimal operation of power plants is optimal water chemistry and high corrosion resistance of the systems. However, fast and comprehensive analysis of the water chemistry is only possible at a limited extent if traditional water chemistry supervision methods are used. In order to achieve an optimum water chemistry supervision, including the fast and quality-assured control under normal as well as perturbed plant conditions, Siemens has developed the DIWA diagnostic system applicable to water/steam cycles of PWR plants. This system not only performs a the chemistry data acquisition, like most of the systems available, but also an analysis of them, checks for data plausibility (quality control of the data),and by logic analysis of these data through symptoms and rules, performs a diagnosis of the current plant situation.

This system is one of the modules of the DIGEST expert system, an integrated, modular diagnostic system. The DIWA diagnostic expert system permits to perform an automated monitoring of the water/steam cycle with high data density the process and operating variables, and to detect rapidly disturbances with effect on the water chemistry, including identification of probable perturbation sources.

The system accomplishes the following tasks:

- On- and off-line data acquisition, including manual laboratory data input.
- Calculation of derivated parameters from the acquired data.
- Checking the acquired data for plausibility, consistency (data analysis)
- Immediate annunciation of detected disturbances
- Comparison of acquired data with threshold values and set points.
- Trend analysis.
- Generation of diagnosis (description, source analysis, countermeasures)
- Data compression and filing.

The required knowledge base for analysis and diagnosis is stored in an expert system based on fuzzy logic, free and easily programmable by the operator.

The latest development of the DIWA system is presented in this work, attention is paid on describing the main system features.