Comparative Assessment of Forecasting Techniques and Impacts of 3D Printing on Ports Container Traffic

Abstract

The start of 3D printing technologies may cause major changes in the manufacturing industry by altering the supply chains in different industrial sectors, and this would in turn affect logistics in the maritime industry. Products that could have been manufactured and shipped physically in containers would now be produced using 3D printing causing significant changes in operations of the maritime industry especially in port container traffic. Despite the risks that this new technology may bring, it is likely that 3D printing may lead to a decrease in physical trade and an increase in virtual trade which would result in a decrease in production costs. However, it is uncertain whether traffic of port containers will decrease or increase after implementing 3D printing technologies. This research thus attempts to find an appropriate forecasting method to predict how 3D printing technology may change the port container traffic volume. Holt' as well as Linear regression techniques of forecasting were chosen in this study because of their applicability to port container traffic data which is short term and non- seasonal in nature. An assessment and comparison of these methods has been made so as to find the right method which helps to predict and understand the likely impact of 3D printing to port container traffic. To conduct a meaningful research, top ten countries with the largest port container throughput globally were selected namely Belgium, UAE, Singapore, Netherlands, Malaysia, South Korea, USA, Germany, Japan, and China. Using available traffic data for each country from 2000 – 2018, the two techniques were used to estimate port container traffic for 2020- 2050 after implementing 3D printing. The data was modelled using R-code software which models such quantitative data. Comparison between results of the two methods was carried out using various measures of assessing forecasting errors (RMSE, MASE, MAPE and Theil's- U statistics). From the calculated errors for the forecasted results, the linear regression model performed better than Holt's model in many countries. However results for other countries were inconclusive under both methods making neither model better than the other. Some forecasted results of the top ten countries indicated a significant reduction in port container traffic with total elimination of traffic being reached by 2050. In reality this significant drop in port container traffic could be unrealistic due to non 3D- printablegoods that will still require shipping in containers. Suggestions were made for the use of forecasting techniques such as Neural networks which use many explanatory variables for further research studies to be more accurate.

Date of Award	May 2020
Original language	American English