Fracture of model concrete: 2. Fracture energy and characteristic length C. Rosselló, M. Elices ⁎ , G.V. Guinea Departamento de Ciencia de Materiales, Universidad Politécnica de Madrid E.T.S.I. Caminos, Profesor Aranguren s/n. 28040-Madrid. Spain Received 12 April 2005; accepted 16 April 2005 Abstract The specific fracture energy G F was measured in six types of simple concrete: all from the same matrix. The aggregates were spheres of the same diameter (strong aggregates, that debonded during concrete fracture, and weak aggregates, able to break); three kinds of matrix–aggregate interface (weak, intermediate and strong) were used. All in all, 55 test results are reported. These results are intended to be used as an experimental benchmark for checking numerical models of concrete fracture. A meso-level analysis of these results showed a correlation between the measured G F values and the properties of the matrix, aggregates and interfaces, particularly with the actual area of the fracture surface. The strength of the matrix–aggregate interface correlates quite well with G F , and concrete ductility, measured by means of the characteristic length, correlates also with the strength of the matrix–aggregate interface. © 2006 Elsevier Ltd. All rights reserved. Keywords: Concrete; Fracture; Fracture energy; Toughness; Debonding 1. Introduction This paper is a sequel to a previous one by the same authors [1]. The purpose of both papers is to offer experimental results that can be profitably used for concrete research and design, and they are intended to serve as an experimental benchmark for checking numerical models of the mechanical behaviour of concrete. This second paper provides experimental values of the specific fracture energy G F for very simple concretes: all made with the same matrix, with two types of spherical aggregates and three kinds of matrix–aggregate interfaces. It is hoped that these results may provide hints for the design of concretes with higher G F values [see, for example Ref. [2]]. There is an increasing awareness that in some cases tensile strength and modulus of elasticity are not enough to characterize the mechanical response of concrete, particularly when ductility or toughness are of concern [see, for example, Refs. [3–7]], and suggestions in this direction should also be welcome. The paper addresses the fracture energy of concrete as defined within the context of the cohesive crack model. However, the results are far reaching as this property can be defined not only within the framework of cohesive models and it has become a reference parameter in the field of concrete fracture. The paper is structured in the following way: First, a report of the experimental results of the measurements of G F for the different concretes; then a mesoscopic analysis of these results based on the roughness of the fracture surfaces, the type of aggregates, and the role of the matrix–aggregate interface, and finally, some recommendations to improve G F and concrete toughness. 2. Measurement of the fracture energy This section provides experimental results of the specific fracture energy, G F , of a simple concrete in the hope that these values could be used for checking numerical models of concrete failure. All the concretes were made with the same matrix, two kinds of aggregates and three matrix–aggregate interfaces, as detailed in a previous paper [1]. 2.1. Theoretical background The ideal direct procedure to measure G F is by means of a stable uniaxial tensile test. Unfortunately, it is difficult to perform stable and representative tensile tests [8], and a simpler Cement and Concrete Research 36 (2006) 1345 – 1353 ⁎ Corresponding author. E-mail address: melices@mater.upm.es (M. Elices). 0008-8846/$ - see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.cemconres.2005.04.016