Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 87
Edited by: B.H.V. Topping
Paper 27

Production of Chloride Ingress Profile with Neural Networks in Concrete with Various PC-PFA-MK Binder Compositions

J. Bai and S. Wild

Department of Engineering, Faculty of Advanced Technology, University of Glamorgan, Pontypridd, United Kingdom

Full Bibliographic Reference for this paper
J. Bai, S. Wild, "Production of Chloride Ingress Profile with Neural Networks in Concrete with Various PC-PFA-MK Binder Compositions", in B.H.V. Topping, (Editor), "Proceedings of the Ninth International Conference on the Application of Artificial Intelligence to Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 27, 2007. doi:10.4203/ccp.87.27
Keywords: neural networks, modelling, prediction, chloride ingress profile, metakaolin, fly ash, concrete.

Although the practicality of reinforced concrete has been recognised for over a century [1,2,3], it is still difficult to predict its performance in structures and many cases of concrete deterioration have been observed [4]. It is well recognised [1,2] that chloride penetration into concrete is affected by parameters such as mix proportions. However, the complexity of effectively tackling this problem still exists due to the uncertainties associated with the interrelationship of chloride penetration with concrete materials [5,6].

This paper reports the application of neural networks in the production of chloride ingress profiles of concrete that incorporates metakaolin (MK) and pulverised-fuel ash (PFA). The chloride ingress profile models are validated on independent data covering a wide range of mix proportions, materials and ages, and gives good prediction accuracy. The production of the profiles based on models well reflect the effect of variations in pozzolanic replacement. The results show that the neural network models can be used to generate the chloride ingress profiles across a wide range of PC-MK-PFA compositions.

Neural network models have been developed for production of concrete chloride ingress profiles with various PC-MK-PFA binder compositions. The networks established have, in general, considered major dependency parameters, including material and age at test. The NNs were evaluated against independent experimental data and the predicted values compared favourably with independent test data. The neural network model constructed provide an efficient, quantitative and rapid means of obtaining optimal solutions to chloride profiles for concrete mixtures using PC-FA-MK blends as binder.

On the basis of the models developed, the examples of chloride profiles were plotted using trained neural networks to analyse the effects of MK and FA on chloride concentration. This makes it possible for the designer to produce mixtures with various blend compositions for a given profile.

Bai J., Wild S. and Sabir B.B., "Chloride ingress and strength loss in concrete with different PC-PFA-MK binder compositions exposed to synthetic seawater", Cement and Concrete Research, Volume 33, Issue 3, Pages 353-362, March 2003. doi:10.1016/S0008-8846(02)00961-4
Bai J., Sabir B.B., et al. "Strength development in concrete incorporating PFA and Metakaolin", Magazine of Concrete Research 52(3): 153-162, 2000.
Neville A.M., Properties of concrete. Lonman Scientific and Technical, London, 779pp, 1981.
Wallbank E.J., The performance of concrete in bridges: a survey of 200 highway bridges. HMSO, London, 96pp, 1989.
Wood J.G.M. and Crerar J., "Tay road bridge: Analysis of chloride ingress variability & prediction of long term deterioration ", Construction and Building Materials, Volume 11, Issue 4, Pages 249-254, June 1997.
Clarke M.A., Parrott L.J. Spooner D.C., "A Future with Concrete", Concrete 2000, Economic and Durable Construction through Excellence, E. and F.N. Spon, pp.1749-1758, September, 1993. doi:10.1016/S0950-0618(97)00044-5

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 £62 +P&P)