ABSTRACT
The computer control systems are ubiquitous in process plants, thus facilitating the collection and storage of data from plant instrumentation. Such data, however, are subject to variation in quality. This is inherent in all process data collection as the sampling and testing equipment, schedules, and techniques are exposed to a wide range of influences. Such influences will always give rise to a raw data set containing missing points, gross errors, and outliers, all of which need to be eliminated to obtain a useable data set. This chapter discusses issues associated with the treatment of data and their reconciliation to obtain consistent estimates. The formal data reconciliation problem is formulated and various available techniques are explored for solving it. Furthermore, the data reconciliation problem is studied in the presence of gross errors as well as the stages to follow for the treatment of gross biased data. Different strategies for testing a set of data are also described and a serial elimination strategy is discussed for identifying sources of gross errors.
