Will you conduct only one session? It depends on your research questions and aim.

3 hours is surely not enough. You should follow the scene you're investigating at least some weeks, depending on how your method-mixture is set up.

Not knowing what the research interest is makes it difficult to answer. Is it only one specific process, what and how many people are executing this process? What potential of comparison do you want to gain, etc., might be a few questions that would help clarifying the need of time. Anyway, the time is usually a consequence of the research fouce and the means you have to fulfill the research.