Assessing the impacts of spatial profile on site production using clustering techniques

Owing to the scarcity of land supply, high-rise buildings in urban areas are prevailing in Hong Kong. In order to maximize the site coverage ratio, the lower levels usually occupy most of the site area. Site planners generally agree that the availability of working space prescribes the level of complexity for site layout production planning (SPLP). It is also agreed that a confined site will bring about difficulties in planning and reduce site productivity. Generally, the spatial profile construction sites is classified in generic terms, such as sufficient, confined or congested. The degree of congestion could affect the selection of plant and machinery, construction method and work sequence. Various studies had been conducted to provide the effective use of site space, improve utilization of site space and positioning of plant and machinery. It is rare to find any studies that aim to quantify site space and explore its impacts on site planning. Such information would be useful for construction managers in reviewing SPLP in the tendering stage or project planning stage. In order to establish objective views on site spatial factor, unsupervised clustering using neural network techniques is used to classify site space conditions. The layout of buildings including site space, lower floors, high headroom areas and tower blocks is superimposed onto a standardized matrix grids and the spatial ratio between site space, site areas and building areas is calculated for analysis and clustering. Unsupervised clustering techniques namely, the K-Mean Cluster Analysis and the Self-organizing Map (SOM) algorithm are used to group the sample projects in terms of site spatial information and spatial ratios. The use of unsupervised clustering techniques and the objective data could assure the classification not predetermined and unbiased. Six site spatial groups are identified distilling sites from 'manoeuvrable' to 'congest'. Sample projects are selected from the spatial groups generated and their implications and constraints on site layout planning are discussed. The study will establish a new perspective to define the degree of congestion and their impacts on site planning. The knowledge will be useful for construction managers in making strategic decisions in production planning and improving accuracy in cost estimation for temporary facilities.