Order Dispatching Research Articles

Order review and release in make-to-order flow shops: analysis and design of new methods

Increased customization has strengthened the importance of make-to-order companies. The advent of lean management and the introduction of smart and flexible technologies has enabled many of these companies to create flow shop routings. Order review and release (ORR) research, which originally focused on job shops, has started paying attention to flow shops. However, the results have not provided clarity on the best ORR method for flow shops. This study aims at developing such a method by applying a modular design approach. It identifies the relevant elements of ORR methods for flow shops, combines them into new methods and evaluates them in a simulation study. The simulation results demonstrate that performance in pure flow shops can be strongly improved by applying the right combination of workload measures, load balancing, and order dispatching. Specifically, the results show that (1) classical workload measures are still as effective as novel measures that have been suggested for flow shops, (2) balancing workloads explicitly through optimization at the order release stage strongly improves performance, and (3) shortest processing time dispatching is highly effective in flow shops as it avoids starvation of stations. In-depth analyses have been executed to unravel the reasons of performance improvements. As such, the article provides clarity on the improvement potential that is available for ORR in flow shops, while the new modular methods provide a first step in exploiting this potential.

Designing an adaptive production control system using reinforcement learning

Modern production systems face enormous challenges due to rising customer requirements resulting in complex production systems. The operational efficiency in the competitive industry is ensured by an adequate production control system that manages all operations in order to optimize key performance indicators. Currently, control systems are mostly based on static and model-based heuristics, requiring significant human domain knowledge and, hence, do not match the dynamic environment of manufacturing companies. Data-driven reinforcement learning (RL) showed compelling results in applications such as board and computer games as well as first production applications. This paper addresses the design of RL to create an adaptive production control system by the real-world example of order dispatching in a complex job shop. As RL algorithms are “black box” approaches, they inherently prohibit a comprehensive understanding. Furthermore, the experience with advanced RL algorithms is still limited to single successful applications, which limits the transferability of results. In this paper, we examine the performance of the state, action, and reward function RL design. When analyzing the results, we identify robust RL designs. This makes RL an advantageous control system for highly dynamic and complex production systems, mainly when domain knowledge is limited.

Reinforcement learning for an intelligent and autonomous production control of complex job-shops under time constraints

Reinforcement learning (RL) offers promising opportunities to handle the ever-increasing complexity in managing modern production systems. We apply a Q-learning algorithm in combination with a process-based discrete-event simulation in order to train a self-learning, intelligent, and autonomous agent for the decision problem of order dispatching in a complex job shop with strict time constraints. For the first time, we combine RL in production control with strict time constraints. The simulation represents the characteristics of complex job shops typically found in semiconductor manufacturing. A real-world use case from a wafer fab is addressed with a developed and implemented framework. The performance of an RL approach and benchmark heuristics are compared. It is shown that RL can be successfully applied to manage order dispatching in a complex environment including time constraints. An RL-agent with a gain function rewarding the selection of the least critical order with respect to time-constraints beats heuristic rules strictly by picking the most critical lot first. Hence, this work demonstrates that a self-learning agent can successfully manage time constraints with the agent performing better than the traditional benchmark, a time-constraint heuristic combining due date deviations and a classical first-in-first-out approach.

Real-time order dispatching for a fleet of autonomous mobile robots using multi-agent reinforcement learning

Autonomous mobile robots (AMRs) are increasingly being used to enable efficient material flow in dynamic production environments. Dispatching transport orders in such environments is difficult due to the complexity arising from the rapid changes in the environment as well as due to a tight coupling between dispatching, path planning, and route execution. For order dispatching, an approach is proposed that uses multi-agent reinforcement learning, where AMR agents learn to bid on orders based on their individual observations. The approach is investigated in a robot simulation environment. The results show a more efficient order allocation compared to commonly used dispatching rules.

Automation of ILL and EDD services at the Library for Natural Sciences of the Russian Academy of Sciences

The Library for Natural Sciences of the Russian Academy of Sciences possesses unique collections of the national and foreign science literature. It acts as a federal STI authority in the natural and engineering sciences. Therefore, one of the Library’s key functions is delivering materials (originals and fragment copies) from its collection to other libraries within the ILL system. The RAS Library for Natural Sciences heads the country’s largest centralized library system (CLS) embracing 49 affiliates of research organizations in Moscow region. The CLS collections, including those of the Central Library, make the single distributed collection to provide services to ILL subscribers. The peculiar feature of the Library’s ILL services is that it delivers materials to its subscribers (the CLS libraries) not only from the CLS collections alone but also from other Moscow and Russian major library collections. The authors describe the computerized system designed at the Russian Academy of Sciences Library for Natural Sciences and characterize the used ILL technology. The structure of the information base comprises ‘Clients’ and ‘Orders’ units. Order dispatching and cash settlements with subscribers are provided by the system’s management unit; the retrieval and statistical tasks are also accomplished. The user unit supports order generation based on union e-catalogs. Specific features of order generation based on various Internet-catalogs, i. e. journals, books and periodicals, image catalogs of national and foreign books, are examined.

Challenges and Potentials of Digital Twins and Industry 4.0 in Product Design and Production for High Performance Products

Digital twins offer great opportunities in various domains of the product engineering process. However, current approaches to the use of digital twins only focus on different separated disciplines. In contrast to that, it is expected that the holistic use of digital twin models in product development and production will dominate future product generations, because they allow to create high-performance products competitively. This paper explores important challenges and future potentials of digital twins and Industry 4.0 for the seamless integration of product specification and production. In this regard, approaches of linking digital twins to other domains open up new possibilities in tolerance allocation and production integration. Thereby, the most efficient product specifications in technical and economic terms are achieved for the manufacturer. In addition, the connectivity of Industry 4.0 broadens the scope and enables the evaluation of alternative approaches in production planning and control. Approaches at the organizational level, product functions with specifications beyond the technological limits and production control strategies (e.g. order dispatching) ensure high performance operations. Simulations with a digital production twin with integrated digital product twin allow early estimations even before the actual ramp-up of the product. The future challenge addressed in this paper is to define a consistent framework for the holistic use of digital twins in the entire product development process, which requires the integration of product designers and production planner concepts.

Design, Implementation and Evaluation of Reinforcement Learning for an Adaptive Order Dispatching in Job Shop Manufacturing Systems

Modern production systems tend to have smaller batch sizes, a larger product variety and more complex material flow systems. Since a human oftentimes can no longer act in a sufficient manner as a decision maker under these circumstances, the demand for efficient and adaptive control systems is rising. This paper introduces a methodical approach as well as guideline for the design, implementation and evaluation of Reinforcement Learning (RL) algorithms for an adaptive order dispatching. Thereby, it addresses production engineers willing to apply RL. Moreover, a real-world use case shows the successful application of the method and remarkable results supporting real-time decision-making. These findings comprehensively illustrate and extend the knowledge on RL.

Autonomous order dispatching in the semiconductor industry using reinforcement learning

Cyber Physical Production Systems (CPPS) provide a huge amount of data. Simultaneously, operational decisions are getting ever more complex due to smaller batch sizes, a larger product variety and complex processes in production systems. Production engineers struggle to utilize the recorded data to optimize production processes effectively because of a rising level of complexity. This paper shows the successful implementation of an autonomous order dispatching system that is based on a Reinforcement Learning (RL) algorithm. The real-world use case in the semiconductor industry is a highly suitable example of a cyber physical and digitized production system.

Reinforcement learning for adaptive order dispatching in the semiconductor industry

The digitalization of production systems tends to provide a huge amount of data from heterogeneous sources. This is particularly true for the semiconductor industry wherein real time process monitoring is inherently required to achieve a high yield of good parts. An application of data-driven algorithms in production planning to enhance operational excellence for complex semiconductor production systems is currently missing. This paper shows the successful implementation of a reinforcement learning-based adaptive control system for order dispatching in the semiconductor industry. Furthermore, a performance comparison of the learning-based control system with the traditionally used rule-based system shows remarkable results. Since a strict rulebook does not bind the learning-based control system, a flexible adaption to changes in the environment can be achieved through a combination of online and offline learning.

Capability Matchmaking Procedure to Support Rapid Configuration and Re-configuration of Production Systems

Rapid responsiveness in terms of processing functions, production capacity and order dispatching is required from today's production systems. This paper introduces a capability-based matchmaking procedure, which supports rapid configuration and re-configuration of production systems. The approach relies on formal OWL-based descriptions of both product requirements and resource capabilities. The base algorithm for the matchmaking is introduced together with the required rules for combined capability calculation and capability matchmaking. The presented approach supports the system designer and reconfiguration planner by automatically suggesting possible resource alternatives for certain product requirements.

A model of market power in electricity industries subject to peak load pricing

This paper studies the exercise of market power in price-regulated electricity industries under peak-load pricing and merit order dispatching, but where investment decisions are taken by independent generating companies. Within this context, we show that producers can exercise market power by under-investing in base-load capacity, compared to the welfare-maximizing configuration. We also show that when there is free entry with an exogenous fixed entry cost that is later sunk, more intense competition results in higher welfare but fewer firms.

Manufacturing execution systems for customized production

Mass customised manufacturing is a current tendency in many production sectors. In this scenario, the client details the desired product often using informatics means and respecting available options. This customisation context imposes systemic integration and co-operations between the concerned manufacturing entities to explore their capabilities, aiming at adaptability to the product heterogeneity. An integration element of the management system and systems related to the shop floor is the manufacturing execution system (MES). A conceivable way to allow MES supporting this customised e-manufacturing is the concept of smart-product, where each product “drives ” its own production, allowing a decoupling between production and order dispatching, as well as the consistency between physical and informational flows. A smart-product requests services from manufacturing resources and it can compete for them. These resources must co-operate based on their features and based on some established flexible co-operation logic, aiming to carry out smart-products requests. However, this is by itself another issue, firstly due to the heterogeneity of factory resources. Looking for homogenisation and integration of this diversity, resources and also smart-products may be “encapsulated” inside of communicating entities called “holons ”. However, the composition of holonic MES (H-MES) is not trivial, because of the dynamic between all holons may be complex. Previous studies presented the organisation of resource-HLs co-operation carried out by computational entities called “rules”. In this paper, it is proposed a conceptual solution of a meta-model for rules-oriented and product-driven H-MES and its application for the holonification of a design and simulation tool.

An order dispatcher for dispersed manufacturers solved by a mixed-variable hybrid genetic algorithm

This paper describes a new solution for an order dispatching problem (ODP) within a distributed manufacturing environment. A mathematical model has been developed to describe the ODP, including both assigning orders to manufacturers and determining the completion times while satisfying a set of constraints. This model has two types of variables: 0-1 decision variables and integer variables that make it a non-linear mixed-variables integer programming model. The usage rate of parts inventory is taken into consideration to avoid parts devaluation. To solve the problem, a novel hybrid genetic algorithm (GA) with a new evolution strategy has been introduced. The hybrid GA can provide more survival chances in the evolving competition for outstanding individuals and introduce potentially excellent individuals to the next generation. To find the appropriate completion times for each order, the hybrid GA embeds a mini-scheduling algorithm to arrange the executing sequence of orders for each manufacturer. The computation results have shown that the proposed hybrid GA can find more satisfactory solutions than the GA with traditional evolution strategies, particularly as the problem size increases with the number of orders.

The effects of rework on the shopfloor control in wafer fabrication

The purpose of this paper is to study the effect of rework operations on the shopfloor control’s strategies, including order release and dispatching. An attempt is made to determine the performance of various production control strategies on the system performance indicators under different levels of rework rate. Also the interaction of order release and dispatching strategies under different rework operations is investigated. Some representative shopfloor control (SFC) strategies are considered in our simulation model. It was found that the effect of rework operation on the system performance is significant. The order release and dispatching strategies’ performance is affected by the rework operation. The performance is improved dramatically if a suitable combination of order release and dispatching strategies is adopted. Under different rework rates and performance indicators a suitable combination of SFC strategies is suggested in this paper.

Order Dispatching and Labor Assignment in Cellular Manufacturing Systems

ABSTRACTAlthough order and labor dispatching in the job shop manufacturing setting have been investigated extensively over the last three decades, its representation of actual processes found in practice today is limited due to the move to cellular manufacturing (CM). Manufacturing cells have become an important approach to batch manufacturing in the last two decades, and their layout structure provides a dominant flow structure for the part routings. The flow shop nature of manufacturing cells adds a simplifying structure to the problem of planning worker assignments and order releases, which makes it more amenable to the use of optimization techniques. In this paper we exploit this characteristic and present two mathematical modeling approaches for making order dispatching and labor assignment/reassignment decisions in two different CM settings. The two formulations are evaluated in a dynamic simulation setting and compared to a heuristic procedure using tardiness as the primary performance measure. The formulations are superior to the heuristic approach and can be incorporated into detail scheduling systems that are being implemented by corporations employing enterprise resource planning (ERP) systems today.