In this paper, we show the interest of integrating a knowledge-based component in a hypertext system intended to improve technical documentation consulting by engineers performing a specific task. An experience was led in the electrical networks planning domain. We first present the Technical Documentation Consulting Systems (TDCS) by defining various access modes to text, the different modules of such a system, and the links between them. In order to design a TDCS, we combined Natural Language Processing (NLP) and Knowledge Acquisition (KA) tools and methods. The methodological aspects are the main scope of the present paper : problems arise when we try to connect a task model, a domain model, terms, and texts. We describe the solutions we adopted for our particular project. As a conclusion, we critically examine our experience and bring out future prospects mainly concerning the interaction between NLP and KA for designing TDCS.