Comparing Different Replacement Policies for Logging Machines in Brazil

Abstract

A dynamic programing algorithm to identify schedules that minimize the discounted cost (DC) of logging machines over a planning horizon including gains from technological progress was used. The identified schedules were also compared with three alternative replacement policies derived from the literature and Brazilian forestry companies. The case study used a harvester and a forwarder and a 100-year planning horizon, where the maximum replacement limit was 8 years. To apply the dynamic programing algorithm, it was necessary to generate lists from cash flows, which incorporated the possible replacement combinations of a series of machines according to the length of the planning horizon and the maximum replacement limit. The lists were formed by three descriptors: predecessor node (moment of purchase of the machine), future node (point of sale for the acquisition of a new machine), and arc value (DC information, the mean production cost and mean production). The results show that the DC identified for the series of harvester replacements was higher compared to the forwarder. It was also identified that the harvester's economic life is shorter, and with technological progress, there was a reduction in the economic life of both machines. Technological progress was also responsible for reducing the average production cost and increasing the average production of machines. When comparing the alternative schedules (AS), it was found that, although AS had a higher DC value and mean production costs, there was very little difference between them. In the harvester's case, AS01 had the highest DC value ($4.36 million). By choosing it, the decision maker would bear a DC boost of $54,000, while AS02 and AS03 would trigger an increase of $43,000 and $32,000, respectively. For the forwarder, the schedule with the highest DC value was AS03 ($3.69 million). The postponement of the replacements made in alternative schedule 01 and alternative schedule 02 resulted in an increase in the DC of $5000, while the anticipation of the replacements made in the alternative schedule 03 resulted in an increase of $48,000. The aspect that stood out the most, in relation to the results presented, was the small variation that the alternative schedules presented in relation to the schedules obtained using the dynamic programing algorithm. With a DC variation of less than 1.4%, the results lead us to conclude that the decision maker will not suffer much harm in choosing any of the alternative schedules tested.