Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 82
PROCEEDINGS OF THE EIGHTH INTERNATIONAL CONFERENCE ON THE APPLICATION OF ARTIFICIAL INTELLIGENCE TO CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING
Edited by: B.H.V. Topping
Paper 46

Evaluation of Building Performance using Artificial Neural Networks: A Study on Service Life Planning in Achieving Sustainability

J.M. Yatim+, S.H. Tapir* and F. Usman$

+Department of Structure and Materials, Universiti Teknologi Malaysia, Skudai, Malaysia
*Department of Civil Engineering,
$Sustainable Construction Research Group,
Universiti Teknologi Malaysia, Kuala Lumpur, Malaysia

Full Bibliographic Reference for this paper
J.M. Yatim, S.H. Tapir, F. Usman, "Evaluation of Building Performance using Artificial Neural Networks: A Study on Service Life Planning in Achieving Sustainability", in B.H.V. Topping, (Editor), "Proceedings of the Eighth International Conference on the Application of Artificial Intelligence to Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 46, 2005. doi:10.4203/ccp.82.46
Keywords: degradation factors, service life prediction, artificial neural network.

Summary
Demands for low cost housing provided by the government are increasing every year. The increase of demands is due to migration of population from rural areas to industrial cities. From the first Malaysia Plan that was implemented in 1966 the first formal and structured housing programs were undertaken to provide low cost housing. However, the maintenance cost of this low cost housing is imposing a great burden to the government due to financial constraints. A study of low cost residential housing in Malaysia found that most of these buildings are occupied without regular or scheduled maintenance |citeyatim:1. For an efficient operating and maintenance program to take place, the evaluation of buildings and their components performance is very important as it reflects the service life of the components.

The construction process has often been described as a highly complex because of the number of disciplines involved from conceptual design to the construction stage. Once completed, the environmental change and use of the building test both the quality of the design and workmanship as well as the suitability of material used. The degradation of buildings is influenced by a whole set of factors such as environmental degradation agents, quality of material, protective treatment, design of buildings, quality of work and maintenance. Selection of suitable materials for the building components can prolong the service life of particular building components and in certain cases require less maintenance and replacement activity. Emphasis on material characterisations at the design stage is limited because most of the time great emphasis is given on delivering with the lowest initial building cost rather than lowest life cycle cost.

Artificial neural networks have become increasingly common in diverse fields such as diagnosing, forecasting, extracting, identification, and control along with advanced computer technologies. In this study, an artificial neural network is used to assess and predict the service life of existing buildings and its components. The back-propagation learning algorithm is used as a learning model. This learning model is among the most efficient tools in engineering applications [2,3]. The environment load factors, workmanship, building materials, usage and level of maintenance are used as input variables in the training process of the neural network model. The environmental and building assessments data were collected from different locations in Malaysia. This paper focuses on the potential application of the neural network for evaluating and predicting the service life of timber as one of the building materials.

The result from the neural network output and the training target data is compared. The coefficient of correlation, r, for the training process of 0.8939 was achieved. The coefficient of correlation, r, for the testing is 0.4405. The results indicate that the neural network was successfully learning the complex relationship between input and output variables from the input patterns. However, the network model cannot achieve a high level of accuracy. It is suspected that this situation is due to the data characteristic. There is a need to improve the neural network performance for data distribution and learning capability. Further analysis is on going to achieve the best performance of the results in predicting the service life of other local building materials.

References
1
S.H Tapsir, "Final Report: Affordable Housing Research Project - Life Cycle Costing Approach for Residential Housing in Malaysia Phase-1, Ministry of Housing and Local Government", Malaysia, 65-72, 2001.
2
C.H Tsai and D.S Hsu, "Diagnosis of Reinforced Concrete Structural Damage Base on Displacement Time History using the Back-Propagation Neural Network Technique", Journal of Computing in Civil Engineering, ASCE, Vol. 16, No. 1, January 1, 2002. doi:10.1061/(ASCE)0887-3801(2002)16:1(49)
3
A.T.C Goh, et. al, "Multivariate Modelling of FEM Data Using Neural Networks", Developments In Neural Networks and Evolutionary Computing for Civil and Structural Engineer, CIVIL-COM PRESS, UK, 59-64, 1995.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £80 +P&P)