BIM BASED CLASSIFICATION OF BUILDING PERFORMANCE DATA FOR ADVANCED ANALYSIS

Abstract

Data density has increased dramatically through utilization of wireless sensors and smart metering devices. While smart meters are commonly monitored in 15 minute intervals, wireless devices are usually polled at much higher rates. Due to these frequent reading intervals, the high data granularity creates a need for improved analysis and categorisation of data. This research aims to extract descriptive information from BIM (Building Information Modelling) models in order to classify and structure monitored data. This is achieved through application of the open BIM standard IFC (Industry Foundation Classes). The extracted information gets combined with current DWH (Data Warehouse) methodologies to support superior and efficient analysis. The DWH model is in sync with the IFC standard to enable a straight forward mapping of IFC objects into the DWH. Several KPIs (Key Performance Indicators) are be applied that support the combination of monitored fact data with the extracted descriptive information. The analysis of monitored data is realised through DWH cubes. These cubes are modelled with selected IFC objects in order to maximise their significance. The envisaged DWH allows the categorisation of data per (i) building system, (ii) building tenant, (iii) building space and (iv) time. Moreover, the DWH allows any combination of afore mentioned categories based on various time intervals. This enables various analysis scenarios, e.g. the calculation of the average electricity consumption per tenant or the peak temperature in a building level. The benefit of reusing selected objects from BIM models over existing solutions is the elimination of repetitive work. In the past, descriptive data was acquired multiple times. This work demonstrates that data extracted from BIM models can be successfully used to enrich data analysis in a DWH system. The major achievement of this work is the elimination of the need to populate static information in the DWH through other means. This concept is evaluated with data sets obtained from real buildings. A proof-of-concept DWH is presented in this work. The main obstacles during the ETL (Extract, Transform and Load) process of individual IFC objects and their conversion into DWH compatible elements are highlighted. Lastly it is also demonstrated that the combination of both technologies does not negatively reflect on the speed of database queries.