Process analysis is a task often performed by chemical engineers, for instance to estimate a process performances or to detect dysfunctioning. Process analysis involves two successive steps, namely state evaluation and state interpretation. State evaluation consists in estimating as precisely as possible the value of process parameters, while state interpretation aims to characterize qualitatively the process operation mode from state parameters values previously estimated. This paper proposes a method combining constraint propagation over intervals and assumption-based reasoning for process state evaluation. Principles of theses techniques are described and their interest for state evaluation is discussed by taking a part of a vapour network as an example. Problems inherent to intervals calculus are discussed and solutions to overcome them are presented.