مدلسازی پیشرو؛ توانمندساز تغییرات پویا در برنامه ریزی تولید

The prevalence of change propagating from the market conditions to the lowest level of functional activities of Manufacturing Planning and Control (MPC) systems, and vice versa, urges a commensurate supportive modeling approach. Progressive Modeling – a novel and integrated problem solution approach – is introduced to handle the new challenges resulting from the dynamic nature of today's manufacturing environment. It analyzes and decomposes the problem into several interacting components, builds a change-ready mathematical model and optimizes its solution. Aggregate production planning problem with a numerical example is used for demonstration and illustration.

The prevalence of change and how it propagates form the outermost scope of business strategies to the lowest level of functional areas of Manufacturing Planning and Control (MPC) systems, and vice versa, requires more dynamic and adaptive modeling and analysis approaches. Progressive Modeling (PM) is proposed to handle a class of ill-formulated industrial production planning and control problems and new challenges arising due to the dynamic nature of today's manufacturing environment. The proposed approach adopts the concepts of Component-Based Software Engineering (CBSE) [1] to analyze MPC problems (or more precisely systems) and decompose them into several fundamental interacting components. Every MPC system is analyzed from this perspective to define its components and define their functions and models, link the overall model to its solver and control the whole process. It presents the developed solution(s) in an appropriate format to the decision makers to help them to monitor, promote and optimize the whole system performance. For every system component, a set of interfaces, which represents sub-set of the specifications of the system, or problem under study, are defined. The componentized nature of developed system emphasizes the model design, functionality and modularity, and de-couples their detailed implementation. This allows implementations to be updated to reflect model changes to be commensurate with variations in the MPC system. The mathematical model specifications go beyond what is known as model assumptions by introducing the concept of assumption relaxations. This represents one of the basic requirements to make developed models more realistic and ready to be re-modeled or updated in the future as conditions or boundaries are changed. A set of objectives should be defined a priori, regardless of the subsequent evaluation methods (e.g. linear or non-linear). Similarly, constraints and their formulation may be added, modified or removed readily, and variables can be integer, binary, or real numbers. Non-linear, rather than linear, modeling is used as the default. Intelligent optimization techniques, such as Genetic Algorithms, Artificial Neural Networks and Tabu search, are the typical solution algorithms. Unlike exact methods, these techniques are loosely coupled with the problems and their assumptions and their capabilities can be independently up-graded as needed as better solution algorithms become available. The underlying premise behind Progressive Modeling is that developed MPC system or sub-systems weave developed models and their linked solution algorithms into several interacting components which should be useable for a wide range of problem variants and capable of optimizing the modeled system performance rather than optimizing a problem solution. 1.1. Propagating the balance: PM Governing philosophy Changeable manufacturing [2] and [3] was introduced as an umbrella concept that embraces the ability of manufacturing systems to change and react to changes. Changes in manufacturing propagates from markets to products, manufacturing system, process planning, Manufacturing Planning and Control (MPC) and enterprise organization. The changes on these multiple fronts do not occur in isolation but are often interdependent. The real challenge is to reach and maintain a balance among all criteria to stay competitive in today's turbulent manufacturing environment. Companies strive to excel at the strategic scope and strategic strength dimensions in order to achieve a competitive advantage. The strategic scope focuses on the composition and size of the target market and strategic strength considers the core competencies of the manufacturing enterprise. There is a clear shift from taller hierarchies to flatter and matrix-like organization structures to improve responsiveness and autonomy and increase the ability of manufacturing enterprises to address these changes. Product-wise, adopting economies of scope (versus economies of scale) places certain constraints on the design of manufacturing systems and their production control strategies. Mass-customization is growing rapidly with serious attempts to lower prices. Companies now compete on being both responsive and efficient. A mix between agile and lean practices is essential to fit these new requirements. Advances in manufacturing technologies move the changeability boundaries and its limits forward, i.e. reconfigurable manufacturing systems (RMS) with its incremental change of functionality and capability versus Flexible Manufacturing Systems (FMS) with built-in abilities to change its functionality within a pre-defined scope. The future changes and evolution of RMSs is by definition uncertain at the outset; it changes based on market and products requirements, and needs a co-evolving MPC system to effectively address its needs. MPC systems represent a gateway between the manufacturing system resources or supply side and its environment (i.e. market or demand). The ability of an MPC system to capture and achieve the balance between those competing goals is a real challenge. Maintaining the balance at all fronts (strategies, organization structure, products, technologies and MPC systems) and under varying conditions governs the driving philosophy of progressive modeling. The goal is to remove the restrictive and problem- or solution-specific constraints and embrace modular component-oriented design to provide future possibility for modifying or replacing any function or module without changing the pre-designed and streamlined system structure and components’ interaction protocols and specifications. This approach maximizes the flexibility and changeability of MPCs in light of changes in objectives, models, solution methods and data. This newly developed progressive modeling methodology has been implemented and is applied in this paper to aggregate production planning as an illustrating example.

A new progressive modeling approach was introduced and applied in the domain of production planning and control. There are two main premises that underscore the importance of having progressively adaptable planning system in the currently experienced changeable environment for manufacturing. The first is that the objectives for optimizing manufacturing production plans are varied and do change according to the prevailing conditions and scenarios. Today's tough financial conditions worldwide clearly demonstrate the changing emphasis and trade-off between the products, facilities, capacities, work force and profitability in the industrial companies struggle for survival. The ability to achieve a balance between the conflicting objectives in such a turbulent environment is a challenge and could be a real competitive edge. The second is that new modeling methods and technological solutions are continuously emerging and having MPC systems that are designed to allow modularity, interchangeability and upgradeability of all modules while preserving the overall system structure and interaction relationships is definitely an asset. This is particularly important to preserve the company specific data and knowledge that get built into an adopted MPC system, hence, the proposed incremental and progressive modeling approaches that integrate several scientific disciplines such as software engineering, artificial intelligence and operation research. The application of the developed progressive modeling approach was illustrated using an example of aggregate production planning system and verified using a published case study. The problem was divided into several components and their interaction was described. A progressive model was developed and solved using evolutionary algorithm. Black box modeling and function templates were used for ill-defined functions such as lost sales evaluation (back orders objective), workforce variability, hiring and firing costs etc. The results show that handling the problem from a system perspective and propagating the balancing across many fronts produces a set of trade-off solutions to consider; and that financial considerations are not the only good measure of manufacturing system health and its ability to survive in the face of change.