Decision Policies for Production Networks

This chapter provides the reader with an overview of this volume from a number of perspectives. First is an overview of the business problems addressed and the decision policies required stretching from networks of machines in a factory to networks of factories in a company to networks of companies in a supply chain. Next there is a brief overview of each chapter with advice to the reader on useful sequences of study depending on individual goals and tastes. Finally there is an overview of the network of authors who contributed to this book.

In this chapter we provide a framework within which concepts from the field of systems and control can be used for controlling manufacturing systems. After introducing some basic notions from manufacturing analysis, we start with the concept of effective process times (EPTs) which can be used for modeling a manufacturing system as a large queuing network. Next, we restrict ourselves to mass production, which enables us to model manufacturing systems by means of a linear system subject to nonlinear constraints (clearing functions). These models serve as a starting point for designing controllers for these manufacturing systems using Model-based Predictive Control (MPC). Finally, the resulting controllers can be implemented on the queuing network model, and ultimately at the real manufacturing system.

Over the past 20 years organizations have put significant energy into making smarter decisions in their enterprise wide central planning and “available to promise” processes to improve responsiveness (more effective use of assets and more intelligent responses to customer needs and emerging opportunities). However, firms have put only limited energies into factory floor decisions and capacity planning and almost none into generating a tighter coupling between the factory and central planning. The bulk of the work to make “smarter factory decisions” has focused on two simple metrics: increasing output and reducing cycle time—often without accommodating the need to run lots at different velocities and without recognizing how the operating curve (trade-off between lead time and tool utilization—Appendix 3) links them. In fact, many of the recent Lean initiatives have focused on eliminating variability to induce simplicity to achieve improved output or cycle time without concern for the impact on responsiveness or capacity. The purpose of this paper is to (a) make clear the critical, and often overlooked, role of factory responsiveness with respect to central planning; (b) explain how traditional factory planning and the current application of Lean can severely impact the firm’s responsiveness; (c) elaborate on touch points between central and factory planning demonstrating simple tactical methods that can improve responsiveness and protect the factory from churn; (d) explain why smarter dispatch scheduling is critical to successful responsiveness; and (e) outline the basics of smarter dispatch scheduling. Although the focus of this paper is the factory, many of the core concepts apply to a wide range of industries from restaurants to health care delivery.

Most of the current dispatching approaches for complex manufacturing facilities like semiconductor fabs are related to due dates. They are variants of classical dispatching rules such as Critical Ratio (CR), Apparent Tardiness Cost (ATC), or Operation Due Date (ODD). Besides that there are a number of operational control policies which target the control of the inventory level of the work centers such as Kanban, Starvation Avoidance, or Minimum Inventory Variability Scheduler (MIVS). While the first set of dispatching rules does not primarily lead to low inventory levels, the latter ones do not always lead to good on-time delivery performance. We are currently developing an approach which combines both ideas, i.e., keeping a low WIP level, avoiding bottleneck starvation, and meeting the due dates. While due dates are usually given by the planning department, adequate WIP levels usually have to be set appropriately by means of pilot studies or educated guessing. As a consequence, an adaptive procedure to determine the adequate inventory levels should be implemented. In our contribution, we provide an overview of current dispatching approaches of both types and discuss their pros and cons. Then, we present our approach in detail and compare its performance with the classical approaches from the literature. Recently, we were able to outperform ODD with respect to WIP levels while having the same on-time delivery performance. The disadvantage is that the optimal target WIP levels (minimum and maximum workload level for the work centers) had to be set experimentally. In our future study, we intend to develop a back-propagation neural network for adaptive parameter setting.

A reduced model of a re-entrant semiconductor factory exhibiting all the important features is simulated, applying a push dispatch policy at the beginning of the line and a pull dispatch policy at the end of the line. A commonly used dispatching policy that deals with short-term fluctuations in demand involves moving the transition point between both policies, the push–pull point (PPP) around. It is shown that with a mean demand starts policy, moving the PPP by itself does not improve the performance of the production line significantly over policies that use a pure push or a pure pull dispatch policy, or a CONWIP starts policy with pure pull dispatch policy. However, when the PPP control is coupled with a CONWIP starts policy, then for high demand with high variance, the improvement becomes approximately a factor of 4. The unexpected success of a PPP policy with CONWIP is explained using concepts from fluid dynamics that predict that this policy will not work for perishable demand. The prediction is verified through additional simulations.

Stamping is one of the most complex operations in the automotive supply chain, providing over 400 end items to dozens of assembly plants and service facilities. This operation consists of a complex network of blankers, presses, and subassemblies. Stamping is affected by much variability, such as unexpected machine and tool down time, quality concerns, and customer requirement fluctuations. These facilities typically run a tight schedule, and supply chain visibility is a critical factor in efficient operations. The data pertaining to operations is distributed across several systems including material requirements planning (MRP), plant floor automation, and logistics management. As a result, decision makers are faced with too much data and not enough information. This leads to time loss and effort spent in consolidating and comprehending the data. This chapter describes the Just-in-time Execution and Distribution Information (JEDI) system that collects and integrates relevant data from a set of disparate systems and generates a set of spreadsheet models that represent the stamping production and supply chain status. JEDI not only presents the information in an intuitive way, but also provides what-if analysis capability and decision support for scheduling and distribution.

In manufacturing operations, a Master Production Schedule (MPS) can be used to make mid-range planning decisions that not only influence the production decisions for a manufacturing facility, but serve as input into other decision systems to determine materials ordering, staffing, and other business requirements. With the advance of computing and data acquisition technologies, an MPS can be recomputed on a more frequent basis to make the production schedule more agile in meeting customer needs. However, uncertainty in the demand forecast or production model may also increase the possibility and/or severity of “schedule nervousness”. The mitigation techniques of frozen horizon, move suppression, and schedule change suppression are evaluated to determine the robust stability margins of each approach at their performance-optimal tunings. Since an MPS is typically computed using Linear Programming these techniques are formulated in this manner, and therefore an empirical Nyquist stability analysis using Empirical Transfer Function Estimates (ETFE) is employed. The technique of move suppression is shown to provide better robust stability margins in the small-scale problem. Further evaluation is needed on scheduling problems of industrial size.

While the problem of planning production in the face of uncertain demand has been studied in various forms for decades, there is still no completely satisfactory solution approach. In this chapter we propose several heuristics based on chance-constrained models for a simple single stage single product system with workload-dependent lead times, which we compare to two-stage and multi-stage stochastic programing formulations. Exploratory computational experiments show promising performance for the heuristics, and raise a number of interesting issues that arise in comparing solutions obtained by the different approaches.

We present an overview over recent developments of traffic flow models for production networks. Particular emphasis is given to the implementation of service rules for complex systems, involving multiple product types and re-entrant loops. A rather general scheduling concept is introduced and demonstrated on some numerical experiments.

Modern production and logistic systems are facing increasing market dynamics: customers demand highly individualized goods, the adherence to due dates becomes critical and stipulated delivery times are decreasing. Particularly logistic networks, e.g. production networks or supply chains, are strongly affected by this trend. On the other hand, production networks have to deal with inherent internal dynamics, which are caused by e.g. machine breakdowns or rush orders. The concept of autonomous control, coming from the theory of self-organization, offers decentralized autonomous decision policies (ADPs), which enable logistic objects to make and execute decision on their own. Due to this kind of decision making, autonomous control aims at a distributed coping with dynamic complexity and, at the same time, at an improvement of the logistic performance. This contribution addresses the concept of autonomous control and the underlying autonomous decision policies as a novel concept for the control of the material flows in networks of coupled production facilities. Moreover, it shows different concepts of modeling and analysis of autonomously controlled networks. To achieve this goal, a dual approach including both, mathematical methods as well as simulation models, is presented. Subsequently, the possibilities to analyze the dynamic behavior of the autonomous logistic system are discussed, i.e., the system’s stability and its logistic performance. Finally, this contribution presents an exemplary case of a production network to demonstrate the practicability of the approach of modeling and analysis of autonomous control for production networks.

Production networks are usually defined as a set of processes utilized to efficiently integrate suppliers, manufacturers, and customers so that goods are produced and distributed in the right quantities, to the right locations, and at the right time and in order to reduce costs while satisfying delivery conditions. We focus on a network of suppliers or producers which order goods from each other, process a product according to orders, and receive payments according to a pricing strategy. Modeling manufacturing systems is characterized by many different scales and several different mathematical approaches. We follow a dynamic approach: we are interested in the time behavior of the entire system. Therefore we introduce a coupled system of ordinary differential delay equations, where time-dependent distribution and order strategies of individual manufacturers influence the flow of goods and the total revenue. We also allow manufacturers to face bankruptcy. All order and distribution strategies are degrees of freedom which can vary in time. We determine them as solution to an optimization problem where additionally economic factors such as production and inventory costs and credit limits influence the maximization of profit. Instead of using a simulation-based optimization procedure, we derive an efficient way to transform the original model into a mixed-integer programing problem.

Determining the production rate of a factory as a function of current and previous states is at the heart of the production planning problem. Different approaches to this problem presented in this book are reviewed and their relationship is discussed. Necessary conditions for the success of a clearing function as a quasi steady approximation are presented and more sophisticated approaches allowing the prediction of outflow in transient situations are discussed. Open loop solutions to the deterministic production problem are introduced and promising new research directions are outlined.