A global optimization for sustainable multi-domain global manufacturing

Abstract

A multi-period and domain nonlinear optimization model is developed in this article. The model incorporates the design of forward–reverse manufacturing networks topology, product platform and operation capacity planning. The model takes into account the lead times and costs for each period of planning and is formulated as mixed integer nonlinear programming (MINLP). A two stages branch and bound (B–B) with cutting planes and under-estimators is proposed, which exploits the problem structure by solving problem relaxation at the first stage upper bound (UB) and generates cutting planes and under-estimators at the second stage lower bound (LB). The application in a three-echelon forward–reverse global manufacturing network shows that the proposed algorithm is capable of efficiently handling large scale and non-convex problem formulation in order to achieve a global optimum. Some important results from the model are presented in terms of their impacts on the sustainability of global manufacturing.

Introduction

Global manufacturing is an alternative solution to parent enterprises that enables them to reduce costs, increase revenues and improve reliability. However, substantial geographical distances in these global situations not only increase transportation costs, but complicate decision making due to inventory cost trade-offs that are the result of increased lead-times in the supply chain. A lack of infrastructures in developing countries diminishes the effectiveness of business processes and also potentially raises environmental problems such as carbon dioxide emissions into the atmosphere, as well as water contamination that are related to the capacity of proper waste reuse or the disposal thereof [13]. From time to time, the rate of consumption of non-recyclable raw materials continues to increase; in fact, at some levels such consumption happens to be higher than its replenishment. Therefore, a global environmentally responsible solution should be implemented as countries׳ environmental initiatives. However, environmental initiatives in different countries provide different levels of potential success, which can affect said countries׳ economic indicators [47].

In terms of problem complexity, environmentally conscious global manufacturing is more difficult to manage than local manufacturing [9], [27]. In addition to economic dynamics across the globe, which includes the variability and uncertainty of currency exchange rates, economic and political instability [9], network topology is a crucial factor for reducing carbon foot-prints and at the same time, improves manufacturing competitiveness. Internally, such variability and uncertainty is likely to affect the entire decision making process, including those pertaining to product prices, delivery lot size, order interval and capacity investment. However, managing the mentioned risks will give a global manufacturing company a competitive advantage.

The research theme of this article is that of developing a dynamic and stochastic model of global manufacturing. The model aims to complement previous contributions related to integrated multi-echelon supply chain design with inventories under endogenous and exogenous uncertainties [45] by comprehensively approaching product design, manufacturing and logistics as a whole, rather than separate parts. The practical contribution of the model is illustrated by exhibiting the following properties: 1) multi-domain; 2) dynamic (multi-period); 3) has several sources of endogenous and exogenous uncertainties; 4) large scale and involves thousands of variables and decision making parameters. The solution methods available for these types of problems are still at relatively early stages of development [5] and their capabilities still limited, due to their potentially significant computational expenses [35]. Therefore, the solution method that combines branch and reduce technique [38] and interior point method [42] in order to handle a large scale model is the theoretical contribution made by this article. The linear assumptions of design, quality and product pricing are the gap between the current model and practice.