MULTI-LEVEL COMPLEXITY IN TERMS OF SPACE SYNTAX: a case study

This paper is inspired by a wayfinding study by Holscher et al. (2005), who find serious wayfinding difficulties in a complex multi-level building and identify architectural properties related to the difficulties. The present study re-analyzes the qualitative results in terms of Space Syntax measures and thus ties them down to formal properties of the architectural structure. The analysis carefully models the spatial interconnections between different floors, as stairs are considered to cause many of the usability issues. Axial maps of the separate floors were interconnected via additional staircase axes using the manual connection feature of Depthmap (Turner, 2004). With respect to Visibility Graph Analysis (VGA) staircases were represented by “widgets” – additional space with representative spatial properties. These were connected to the staircases in the floor plan by merging visibility graph nodes. Considering the building as a whole, the poor intelligibility score of 0.15 is remarkable. Analyzed as separate systems the floors' intelligibility ranges from .09 (second floor) to .71 (basement). With respect to usability issues, several techniques revealed valuable results. Along a typical trajectory through the entrance hall the primary isovist changes rapidly. At no point all relevant navigation choices are visible simultaneously. The lack of survey is best demonstrated by the distribution of integration and connectivity: E.g. the entrance hall neither contains the most connective nor the most integrated part of the system. For the analysis of dead ends in the basement we regarded them as blockages. The visual step depth from one side to the other quantifies the complexity of the detours to overcome the dead ends. (17 and 8 turns respectively). A similar technique was applied to measure the amount of turning occurring in the staircases. As all staircases are offset from the main axis one needs to travel along a minimum of 7 axial lines from the entrance hall to the corresponding main intersection in the basement. To validate our findings we constructed two layout re-designs. Eliminating the dead ends in the basement increases integration in a formerly segregated region. Also, the distribution of spatial measures becomes more similar between different floors. The second re-design removes visual clutter near the entrance where a local Integration maximum emerges together with the most connective area. Although the analyses are post-hoc at this stage the re-designs point at the potential of Space Syntax as a predictive tool for wayfinding design.