This paper describes the calibration process of a land use and transport integrated (LUTI) Tranus based model of the urban agglomeration of Grenoble (France). In particular, for several parameters related to floor-space, we applied a semi-automatic process for calibration. We begin with an overview of the model's structure and next we define data and calibration methodology of the land use part. Further, we describe and place in the public domain the algorithms and methodologies that we developed. We are particularly interested in demand functions for housing and " substitution " functions that capture the trade-offs that households make between housing type and price. We created tools based on non-linear optimization and curve fitting that make calibration easier, more robust, faster than typical manual approach and transferable to other study areas. Finally, we discuss the trade-off between over-fitting and minimizing error in predictions of the observed data on housing consumption.