Modeling with ANN and effect of pumice aggregate and air entrainment on the freeze-thaw durabilities of HSC


CONSTRUCTION AND BUILDING MATERIALS, vol.25, no.11, pp.4241-4249, 2011 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 25 Issue: 11
  • Publication Date: 2011
  • Doi Number: 10.1016/j.conbuildmat.2011.04.068
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.4241-4249
  • Keywords: Freeze-thaw, Pumice aggregate, High strength concrete, Compressive strength, Ultrasound pulse velocity, Artificial neural network, HIGH-STRENGTH CONCRETE, FROST-RESISTANCE, LIGHTWEIGHT AGGREGATE, COMPRESSIVE STRENGTH, CEMENT, DETERIORATION, DAMAGE, ZONE
  • Inonu University Affiliated: Yes


The objective of this work is to calculate the compressive strength, ultrasound pulse velocity (UPV), relative dynamic modulus of elasticity (RDME) and porosity induced into concrete during freezing and thawing. Freeze-thaw durability of concrete is of great importance to hydraulic structures in cold areas. In this paper, freezing of pore solution in concrete exposed to a freeze-thaw cycle is studied by following the change of concrete some mechanical and physical properties with freezing temperatures. The effects of pumice aggregate (PA) ratios on the high strength concrete (HSC) properties were studied at 28 days. PA replacements of fine aggregate (0-2 mm) were used: 10%, 20%, and 30%. The properties examined included compressive strength, UPV and RDME properties of HSC. Results showed that compressive strength, UPV and RDME of samples were decreased with increase in PA ratios. Test results revealed that HSC was still durable after 100, 200 and 300 cycles of freezing and thawing in accordance with ASTM C666. After 300 cycles, HSC showed a reduction in compressive strength between 6% and 21%, and reduction in RDME up to 16%. For 300 cycles, the porosity was increased up to 12% for HSC with PA. In this paper, feed-forward artificial neural networks (ANNs) techniques are used to model the relative change in compressive strength and relative change in UPV in cyclic thermal loading. Then genetic algorithms are applied in order to determine optimum mix proportions subjected to 300 thermal cycling. (C) 2011 Elsevier Ltd. All rights reserved.