Purpose - The paper aims to introduce an efficient contact detection algorithm for smooth convex particles. Design/methodology/approach - The contact points of adjacent particles are defined according to the common-normal concept. The problem of contact detection is formulated as 2D unconstrained optimization problem that is solved by a combination of Newton's method and a Levenberg-Marquardt method. Findings - The contact detection algorithm is efficient in terms of the number of iterations required to reach a high accuracy. In the case of non-penetrating particles, a penetration can be ruled out in the course of the iterative solution before convergence is reached. Research limitations/implications - The algorithm is only applicable to smooth convex particles, where a bijective relation between the surface points and the surface normals exists. Originality/value - By a new kind of formulation, the problem of contact detection between 3D particles can be reduced to a 2D unconstrained optimization problem. This formulation enables fast contact exclusions in the case of non-penetrating particles. © Emerald Group Publishing Limited.