Annual planning of harvesting resources in the forest industry

A cost-efficient use of harvesting resources is important in the forest industry. The main planning is carried out in an annual resource plan that is continuously revised. The harvesting operations are divided into harvesting and forwarding. The harvesting operation fells trees and puts them in piles in the harvest areas. The forwarding operation collects piles and moves them to storage locations adjacent to forest roads. These operations are conducted by machines (harvesters, forwarders and harwarders), and these are operated by crews living in cities/villages that are within some maximum distance from the harvest areas. Machines, harvest teams and harvest areas have different characteristics and properties and it is difficult to find the best possible match throughout the year. The aim of the planning is to find an annual plan with the lowest possible cost. The total cost is based on three parts: production cost, traveling cost and moving cost. The production cost is the cost for the harvesting and forwarding. The traveling cost is the cost for driving back and forwards (daily) from the home base to the harvest area and the moving cost is associated with moving the machines and equipment between harvest areas. The Forest Research Institute of Sweden (Skogforsk), together with a number of Swedish forest companies, has developed a decision support platform for the planning. One important element of this platform is that it should find high-quality plans within short computational times. One central element is an optimization model that integrates the assignment of machines to harvest areas and schedules the harvest areas during the year for each machine. The problem is complex and we propose a two-phase solution method where, first, we solve the assignment problem and, second, the scheduling. In order to be able to control the scheduling in phase 1 as well, we have introduced an extra cost component that controls the geographical distribution of harvest areas for each machine in phase 1. We have tested the solution approach on a case study from one of the larger Swedish forest companies. This case study involves 46 machines and 968 harvest areas representing a log volume of 1.33 million cubic meters. We describe some numerical results and experience from the development and tests.