Forest Transportation Planning Under Multiple Goals Using Ant Colony Optimization

Abstract

For more than a century, forestry practitioners and decision makers have given significant attention to forest transportation planning because transportation is one of the most expensive activities in timber harvesting (Greulich, 2003). The traditional goal of forest transportation planning has been to develop road networks that minimize costs of log hauling and road construction. However, modern forest transportation planning problems (FTPP) are not driven only by the financial aspects of timber management, but also by the multiple uses of roads as well as their social and environmental impacts such as recreation, soil erosion, and stream water quality. These additional considerations and requirements introduce multiple objectives into transportation planning problems, increasing their size and complexity. Two different problem-solving approaches have been applied to solve FTPP; exact algorithms such as mixed-integer programming (MIP) and approximation algorithms generally called heuristics (Falcao & Borges, 2001; Weintraub et al., 1995). MIP has been used to optimally solve these complex FTPP, but its application has been limited by the difficulty of solving large, real world problems within a reasonable amount of computation time (Gottlieb & Paulmann, 1998; Sun et al., 1998). On the other hand, heuristic algorithms, although might not always provide optimal solutions, have been the focus of a large number of researchers because of their high efficiency and problem solving capability especially for large and complex optimization problems (Olsson & Lohmander, 2005; Bettinger and Chung, 2004; Sessions et al., 2003; Zeki 2001; Falcao & Borges, 2001; Martell et al., 1998; Weintraub et al., 1995; Jones et al., 1991). FTPP considering fixed and variable costs form complex optimization problems that to date have only been solved efficiently using heuristic approaches (Kowalski, 2005; Adlakha & Kowalski, 2003). NETWORK II (Sessions, 1985) and NETWORK 2000 (Chung & Sessions, 2003), which use a heuristic network algorithm combined with the shortest path algorithm (Dijkstra, 1959), have been widely used for over twenty years to solve such FTPP. NETWORK 2000 can solve multi-period, multi-product, multi-origin and -destination transportation planning problems. However, it used a single objective function for either profit maximization or cost minimization without taking into account other attributes of road segments modeled as side constraints. A recently developed approach used the ant colony optimization (ACO) to solve FTPP considering fixed and variable costs as well as side constraints (Contreras et al., 2008). This approach provided near-optimal solutions for