Modeling of frost salt scaling Oğuzhan Çopuroğlu , Erik Schlangen Delft University of Technology, Faculty of Civil Engineering and Geosciences, Micromechanics Laboratory (MICROLAB), Delft, The Netherlands Received 8 February 2007; accepted 3 September 2007 Abstract This paper discusses the numerical modeling of deterioration in cement-based materials due to frost salt scaling (FSS). Several aspects of FSS are investigated such as carbonation, microstructure, mechanical properties and testing conditions. Mainly blast-furnace slag cement (henceforth slag cement) systems are of interest in this paper since several reports have been indicated that cementitious materials bearing slag-rich cement are critically vulnerable under combined attack of frost and de-icing salts. In the first part, the paper deals with the effect of carbonation on the micromechanical properties and FSS resistance of 1-year-old slag cement and ordinary Portland cement pastes with W/C 0.45. The micromechanical properties were evaluated by the nano-indentation technique and the results are used to evaluate the behavior of these pastes under frost salt attack. FSS damage on the paste samples is modeled according to the glue- spall theory with the aid of Delft Lattice Model. Additionally, the carbonated cement paste microstructures are characterized by ESEM/BSE. In the second part, parameters that are varied in the investigation are the salt concentration in the external water layer and ice-layer thickness on the surface. Again the lattice type model is used to simulate the mechanism in which the material structure is implemented using digital images of the real material. Both experiments and the simulation with the model show that the amount of scaling increases with increasing thickness of the ice layer on the surface. Furthermore it is shown that with the model the well known pessimum effect for salt concentration in the water (which causes maximum damage at 3% salt) can be reproduced. The outcome of the model indicates that glue-spall theory can successfully explain FSS. © 2007 Elsevier Ltd. All rights reserved. Keywords: Frost salt scaling; Freezing and thawing (C); Micromechanics (C); Modeling (E); Granulated blast-furnace slag (D) 1. Technical background The modeling of FSS has been a difficult issue due to its complex physical and chemical mechanisms [1]. Plain frost action has attracted relatively more attention and thanks to its less complicated mechanism, we increased our knowledge during the past few decades. The works of Powers, Litvan, Pigeon, Marchand, Setzer [25], and many other researchers have drawn the frame of the issue substantially. Unfortunately, similar arguments could not be made for frost salt attack. There have been a number of questions, which could not be answered by a single theory [6]. Due to having insufficient knowledge about the phenomenon, modeling attempts have been restricted to black-box type [7]. However, recently an interesting theoretical explanation of the mechanism of frost salt attack was introduced by Valenza and Scherer [8,9]. The researchers proposed a mechanism called glue-spall. Accord- ing to this theory the cracking of the ice/brine layer is the origin of FSS. They put forward a theoretical explanation for the greater damage of pessimum salt concentration under frost which has been known as a mystery so far. The principle idea is that following the ice formation on top of the concrete surface, ice starts to shrink due to further cooling. The shrinkage creates tensile stresses in the ice and causes three consequences depending on the solute concentration of the liquid. These are: 1. Weak salt concentration (i.e. 0.1%): Due to the ice formation and further cooling of the ice, the exerted tensile stress cannot exceed the tensile strength of ice; hence no cracking occurs (in ice and concrete). 2. Pessimum salt concentration (13%): Due to the ice formation and further cooling of the ice, the exerted tensile Available online at www.sciencedirect.com Cement and Concrete Research 38 (2008) 27 39 Corresponding author. E-mail address: o.copuroglu@citg.tudelft.nl (O. Çopuroğlu). 0008-8846/$ - see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.cemconres.2007.09.003