An integrated approach to logistics strategy planning using visual interactive modelling and decision support

Abstract

The research in this thesis relates to the use of mathematical models and ièigputer-based modelling tools for supporting the Logistics Strategy Planning (LSP) process. conceptual modelling framework and a computer-based modelling and decision support system are developed to address practical LSP problems and improve the level of decision support currently reported. A LSP process T is described, its h complexity recognised, e and the problem domain defined. The evolution of the logistics strategy concept is addressed and defined a integrating procurement, production and distribution aspects. The need for decision support is also identified. A comprehensive review of models and modelling techniques from Management Science / Operations Research (MS/OR), and computer based tools in the LSP context is carried out. The appropriateness of the various models and types of computer tools is assessed. Gaps and drawbacks in current approaches to LSP are also identified. This revealed that past efforts have been directed towards producing more efficient solving techniques and tools for limited aspects of LSP, rather than developing models and tools that could address more realistic problems, recognising a integrated view of LSP. Hence, current approaches to LSP are fragmented in their handling of the procurement, production and distribution aspects. A conceptual modelling framework is proposed to support the LSP process. It includes a planning process, the logistics elements and key drivers required to define a model/representation of the LSP problem, and a selection of models/techniques to address various classes of LSP (sub-) problems. The framework provides a integrated view of all elements involved and contributes to formalise the knowledge necessary to address LSP problems. A modelling and decision support system is developed in order to demonstrate the framework and assess the approach here proposed to address practical LSP studies. STRATOVISION combines Visual Interactive Modelling (VIM) and Knowledge-Based (KB) techniques with "traditional" MS/OR models and modelling techniques. Additionally, the system implements a problem centred approach, combining various MS/OR solving techniques (e.g. simulation, heuristics and optimisation) into a unique modelling environment. A comparative analysis and discussion of functionality supports the view that STRATOVISION overcomes most limitations found with other modelling systems and provides better functionality to address LSP problems. The discussion covers the modelling phase, the options generation and the detailed evaluation of scenarios. Special emphasis is given both to the use of the Visual Interactive ( I) functionality for modelling and problem solving, and to the use of models/techniques included in STRATOVISION's model base. Several case studies are used to illustrate STRATOVISION's integrated approach to LSP and validate the model design. Comparison with fragmented approaches to LSP is carried out and the use of STRATOVISION in practical LSP studies including procurement, production and distribution decisions is reported. The analysis provides supporting evidence of the benefits achieved by using STRATOVISION's integrated approach to LSP. Finally, contributions of the approach are discussed and areas of further work pointed out.