Global Production Network Optimization
The production network design process developed in Chapter 2.4 is split into a global network optimization phase at the country level and a site selection phase focusing on the evaluation of individual production sites within a country. In this chapter the mathematical optimization model required to support the global network optimization phase of specialty chemicals production networks is developed. To this end Chapter 3.1 briefly introduces general research in the field of quantitative location analysis. The literature specifically focusing on production network design is reviewed in greater detail in Chapter 3.2. Alternative approaches to model major elements of production networks and their applicability to specialty chemicals industry are discussed in Chapter 3.3 and tailored modeling approaches are developed. Chapter 3.4 contains the resulting Mixed-Integer Linear Programming (MILP) model and possible extensions to include features that were not required in the pilot application underlying this work (cf. Chap. 5). Finally, Chapter 3.5 presents the results of numerical performance tests to demonstrate the applicability of the model to problem instances of realistic size.
KeywordsCash Flow Network Design Planning Horizon Transfer Price Production Network
Unable to display preview. Download preview PDF.
- 22.cf. Krarup and Pruzan (1990); Francis et al. (1983).Google Scholar
- 25.Some authors further distinguish between decision-making under uncertainty and decision-making under risk with the former referring to a situation where no quantitative information on the “uncertain” parameters is available and the latter referring to a situation where a well-determined probability distribution is available (cf. Wets 1974, pp. 309–310). Here, this distinction will not be used.Google Scholar
- 26.Production network design by definition needs to include distribution decisions. For the purpose of this classification a model is considered to include distribution network design only if at least one distribution echelon (warehouses, distribution centers, etc.) is explicitly modeled. The term “distribution center” is not defined uniformly in literature (cf. Higginson and Bookbinder 2005). Here, distribution centers where a production step takes place are considered to be a production echelon whereas those performing picking or packaging operations are considered to be a distribution echelon.Google Scholar
- 27.The reader interested in a detailed discussion of inventory management is referred to Tempelmeier (2006) and to Graves and Willems (2003) for a discussion of how to spread safety stocks across the supply chain.Google Scholar
- 28.For an overview of investment appraisal calculation methods see for example Götze and Bloech (2002) or Perridon and Steiner (2006).Google Scholar
- 29.Assumptions on temporal allocation of cash flows and discounting period need to be consistent (e.g., beginning, middle or end of period) which is sometimes not the case in mathematical optimization models (cf. Erlenkotter 1981, p. 134). Generally, it is assumed that cash flows are realized at the end of a period. Alternatively, continuous payments can be assumed but the error caused by the year-end assumption is limited (cf. Brealey et al. 2006, pp. 46–48).Google Scholar
- 31.As Luss (1982, p. 935) points out, capacity planning in multi-facility networks is closely related to network design problems.Google Scholar
- 32.This chapter is based on McDonald (1997), pp. 65–79 and Jackson (1997).Google Scholar
- 33.For further details on transfer pricing refer to Feinschreiber and Kent (2003), Choi et al. (2002, pp. 491–505) or Abdallah (1989). Feinschreiber (2004) provides details on the different transfer pricing methods and Feinschreiber (2000) discusses transfer pricing regulations of more than 40 countries.Google Scholar
- 35.The problem might also entail multiple stages if it is intended to model multiple design decision periods. See Birge and Louveaux (1997, pp. 59–60) for a discussion of the issue.Google Scholar