An optimization approach for the design of chemical plant geometric layout is presented. The task is decomposed into a sequence of subproblems that are solved using mixed integer linear programming models. The 3D layout problem with noninterference constraints is computationally difficult, and simultaneous treatment for placement of equipment and routing of pipes for the number of components comprising a typical plant is as yet impractical. In the approach presented, the optimization of the layout is computed first, and then is followed by the piping layout and routing. The models allow incorporation of practical design constraints. An example is presented that show the resulting design appear both reasonable and near-optimal. Different techniques are studied in order to do an automatic routing of a process plant layout. The procedure is based on a (partially arbitrary) selection of node positions where pipes can be routed, respecting restrictions of safety and minimum distances from components. These nodes are connected by arcs that avoid the physical space of components. The graph defined by the nodes and arcs are used to find the minimum cost for the pipes. Two methods may be used: a mixed integer linear program for the global minimization of the cost of all pipes, or a shortest path algorithm for the minimization of the cost of each pipe individually. The MILP approach can include capacity constraints, but it can be applied only for small problems. The shortest path approach is better for larger problems and can include pipes with branches and stress analysis. (c) 2005 Elsevier Ltd. All rights reserved.