Batching Policy Research Articles

Characterization of Strategic Deconflicting Service Impact on Very Low-Level Airspace Capacity

European airspace is poised for significant transformation as it prepares to accommodate a new class of unmanned traffic that will reshape the transport of people and goods. Unmanned aerial vehicle traffic will introduce a new level of services, but it remains unclear how safety and operators’ time flexibility in flight planning will impact capacity. This study focuses on the impact of strategic deconflicting services on the capacity of the very low-level airspace, a critical area in the future management of unmanned aerial vehicle traffic. The results validate the assumptions regarding the roles of airspace managers and drone operators through simulation studies; highlight the limitations of the first come, first served policy; and propose a batch policy as a potential optimization strategy for future airspace capacity management. The forecasting model developed using regression techniques provides a general method for predicting airspace capacity under specific conditions, contributing to the safe and efficient integration of unmanned aerial vehicles into European airspace.

Dockless bike-sharing systems with unusable bikes: removing, repair and redistribution under batch policies

Dockless bike-sharing systems with unusable bikes: removing, repair and redistribution under batch policies

BATCH POLICY LEARNING IN AVERAGE REWARD MARKOV DECISION PROCESSES.

We consider the batch (off-line) policy learning problem in the infinite horizon Markov Decision Process. Motivated by mobile health applications, we focus on learning a policy that maximizes the long-term average reward. We propose a doubly robust estimator for the average reward and show that it achieves semiparametric efficiency. Further we develop an optimization algorithm to compute the optimal policy in a parameterized stochastic policy class. The performance of the estimated policy is measured by the difference between the optimal average reward in the policy class and the average reward of the estimated policy and we establish a finite-sample regret guarantee. The performance of the method is illustrated by simulation studies and an analysis of a mobile health study promoting physical activity.

Evaluation of the Joint Impact of the Storage Assignment and Order Batching in Mobile-Pod Warehouse Systems

The rapid development of e-commerce has impacted the intelligence level of warehouses. The robotic mobile fulfillment system (RMFS) demonstrated that mobile-pod warehouse systems could process a significant number of order-picking operations within the same amount of time as other systems, saving money and improving operational efficiency. The order-picking mode in such a system is “parts-to-picker,” which is becoming both a prevailing operation mode in industrial warehouses and an encouraging research field during these years. One of the most significant fundamental factors in RMFS is the operational efficiency that is affected by storage assignment and order batching. This paper considers the joint impact of the storage assignment policies and order batching policies on order picking process. Our goal is to minimize the moving times of robots, which reflects the order-picking cost or efficiency. We propose using order similarity and item similarity to batch orders and assign item storage locations, respectively. Both the order batching and item grouping are tackled by a clustering model which is an integer linear program. We also develop a policy evaluation model to measure the order picking cost. We conduct numerical tests on six order batching and storage assignment policy combinations. A comparative analysis and an ANOVA analysis are then performed on the test results to compare the performances of these policy combinations. We find that the Weighted Support Count-based storage allocation combining with the correlation-based order batching achieves the best order-picking performance. Also, the more accurate information about the items and orders we can get from the historical data, the more order-picking workload we can save by exploiting the similarity features.

Enhancing in-store picking for e-grocery: an empirical-based model

PurposeThis paper identifies, configures and analyses a solution aimed at increasing the efficiency of in-store picking for e-grocers and combining the traditional store-based option with a warehouse-based logic (creating a back area dedicated to the most required online items).Design/methodology/approachThe adopted methodology is a multi-method approach combining analytical modelling and interviews with practitioners. Interviews were performed with managers, whose collaboration allowed the development and application of an empirically-grounded model, aimed to estimate the performances of the proposed picking solution in its different configurations. Various scenarios are modelled and different policies are evaluated.FindingsThe proposed solution entails time benefits compared to traditional store-based picking for three main reasons: lower travel time (due to the absence of offline customers), lower retrieval time (tied to the more efficient product allocation in the back) and lower time to manage stock-outs (since there are no missing items in the back). Considering the batching policies, order picking is always outperformed by batch and zone picking, as they allow for the reduction of the average travelled distance per order. Conversely, zone picking is more efficient than batch picking when demand volumes are high.Originality/valueFrom an academic perspective, this work proposes a picking solution that combines the store-based and warehouse-based logics (traditionally seen as opposite/alternative choices). From a managerial perspective, it may support the definition of the picking process for traditional grocers that are offering – or aim to offer – e-commerce services to their customers.

Developing bulk arrival queuing models with the constant batch policy under uncertainty data using (0 - 1) variables

This paper delves into some significant performance measures (PMs) of a bulk arrival queueing system with constant batch size b, according to arrival rates and service rates being fuzzy parameters. The bulk arrival queuing system deals with observation arrival into the queuing system as a constant group size before allowing individual customers entering to the service. This leads to obtaining a new tool with the aid of generating function methods. The corresponding traditional bulk queueing system model is more convenient under an uncertain environment. The $alpha$-cut approach is applied with the conventional Zadeh's extension principle (ZEP) to transform the triangular membership functions (Mem. Fs) fuzzy queues into a family of conventional bulkqueues. This new model focus on mixed-integer non-linear programming (MINLP) tenders a mathematical computational approach is known as (0 -1) variables. To measures the efficiency of the method, the efficient solution strategy plays a crucial role in the adequate application of these techniques. Furthermore, different stages of the $alpha$-cut intervals were analyzed and the final part of the article gives a numerical solution of the proposed model to achieve practical issues.

Performance analysis of batching decisions in waveless order release environments for e‐commerce stock‐to‐picker order fulfillment

AbstractWarehouse automation is increasingly adopted to manage throughput fluctuations in e‐commerce order fulfillment. This work develops queuing network models and solution methodologies for performance analysis of a stock‐to‐picker system that connects an upstream automated storage system to a downstream pick station. We focus on the pick station process, quantifying the throughput differences between a pick station that employs a static versus dynamic batching strategy. We consider a waveless order release environment where the item totes are requested from the storage system only when an order arrives. We capture throughput performance in this environment for single as well as multi‐line orders with/without item commonality by developing closed‐queuing network models. The consolidation of multiple product‐lines for a multi‐line order is modeled using fork–join synchronization stations within the closed queuing network. For analyzing such queuing networks, we develop a network‐decomposition based solution methodology. We validate the models using a simulation model of the upstream storage and downstream order‐picking system. We find that in waveless order release environment, dynamic batching always outperforms static batching in terms of system throughput. However, for single‐line orders, the percentage gain in the throughput (by implementing dynamic batching) decreases for smaller item tote inter‐arrival times. For multi‐line orders, dynamic batching increases the system throughput by 37–43%. We also analyze the effect of batch size on the throughput performance. The results indicate that the system throughput increases with an increase in the batch size under both batching policies. But, the marginal benefit reduces as we increase the batch size.

An analytical investigation of alternative batching policies for remanufacturing under stochastic demands and returns

This article examines a fundamental lot-sizing problem which arises in the context of a make-to-order remanufacturing environment. The problem setting is characterized by a stochastic used-item return process along with a stochastic remanufactured-item demand process faced by a remanufacturer. We explicitly take into account for all relevant costs, including the fixed costs (associated with remanufacturing of used-items and dispatching of remanufactured-item orders in batches) and inventory-related cost (associated with used-item inventory holding costs and remanufactured-item order waiting costs). We propose five batching policies inspired by shipment consolidation practice (three periodic policies and two threshold policies). For the purpose of computing policy parameters, we develop analytical models that are aimed at minimizing the long-run average expected total cost of the remanufacturer. Since the underlying cost expressions are not analytically tractable, we propose easily computable approximations that lead to closed-form expressions for obtaining policy parameters. A careful numerical investigation demonstrates that the resulting policy parameters are highly effective approximations. Then, we extend the policies by considering disposal options when needed. For this extension, an effective parameter-based approximation approach is developed for computational purposes, and additional numerical experiments demonstrate the effectiveness of the proposed approach.

Does the Autonomation Policy Really Help in a Smart Production System for Controlling Defective Production?

This study explains about a serial smart production system where a single-type of product is produced. This system uses an unequally sized batch policy in subsequent stages. The setup cost is not always deterministic, it can be controllable and reduced by increasing the capital investment cost, and that the production rates in the system may vary within given limits across batches of shipments. Furthermore, as imperfect items are produced in long-run system, to clean the imperfectness autonomation policy is adopted for inspection, which make the process smarter. The shipment lot sizes of the deliveries are unequal and variable. In long-run production system, defective items are produced in “out-of-control” state. In this model, the defect rate is random with a uniform distribution which is clean from the system by autonomation. In addition, in the remanufacturing process, it is assuming that all defective products are repaired, and no defective products are scrapped. The main theme of developing this model is to determine the number of shipments and the optimal production lot size to adjust the production rates and decrease the total system cost under a reduced setup cost by considering the discrete investment and make a serial smart production system. A solution procedure along with an advanced algorithm was proposed for solving the model. Numerical examples with some graphical representations are provided to validate the model.

Links between kit quality and kit preparation design

ABSTRACT Kitting is a materials-feeding principle commonly used with mixed-model assembly, but literature is lacking with respect to how kit quality can be supported. The purpose of this paper is to create an understanding of the links between kit preparation design aspects and kit preparation error types, that can be useful to support kit quality. The paper draws on empirical data from a multiple case study in the automotive industry to study how typical kit errors are linked to eight kit preparation design aspects: location, work organisation, storage policy, batching policy, storage packaging, kit carrier and container, picking information system, and error communication. The findings suggest several opportunities related to kit preparation design aspects for preventing kit errors and facilitating kit error corrections. The paper extends earlier knowledge and can support kit quality of industrial kit preparation.

Gradient-Aware Model-Based Policy Search

Traditional model-based reinforcement learning approaches learn a model of the environment dynamics without explicitly considering how it will be used by the agent. In the presence of misspecified model classes, this can lead to poor estimates, as some relevant available information is ignored. In this paper, we introduce a novel model-based policy search approach that exploits the knowledge of the current agent policy to learn an approximate transition model, focusing on the portions of the environment that are most relevant for policy improvement. We leverage a weighting scheme, derived from the minimization of the error on the model-based policy gradient estimator, in order to define a suitable objective function that is optimized for learning the approximate transition model. Then, we integrate this procedure into a batch policy improvement algorithm, named Gradient-Aware Model-based Policy Search (GAMPS), which iteratively learns a transition model and uses it, together with the collected trajectories, to compute the new policy parameters. Finally, we empirically validate GAMPS on benchmark domains analyzing and discussing its properties.

Mathematical and analytical approach for the management of defective items in a multi-stage production system

Reworking defective items any in multi-stage production systems have become a common issue for production industries. Due to random defective rate of production, it may contain several imperfect items as wastes. The reduction of those wastes through reworking in each stage of a cycle or at the end of stage of a cycle by an identifying opportunity design technique. A design is developed for making such type multi-stage production system. For this, imperfect products may be reworked within each cycle while avoiding any shortages or reworked after end cycle with shortages in each cycle. The aim of the model is to diminish wastes by reworking strategy for minimizing the total cost. For this purpose, the proposed study develops a multi-stage production model for an optimum production batch policy with random production of imperfect products. Two-stage inspection is considered for detecting those faulty products and ensuring that there are no faulty items anymore. As it is a long-run production system, setup cost plays an important role, an attempt at setup cost reduction has been made to reduce it by a discrete investment for each stage along with a budget constraint. The Lagrangian method is employed to solve the model and the global minimum cost for optimal decision variables is obtained. Ten lemmas are established to obtain the global optimum solution. The model shows that a large investment to reduce the setup cost is better than a constant setup cost. It is found that the imperfect items should be reworked after the end cycle. There are eight numerical examples, eight special cases, a sensitivity analysis of the key parameters, and sixteen graphical representations to demonstrate the proposed model.

FERRY: access control and quota management service

Fermilab developed the Frontier Experiments RegistRY (FERRY) service that provides a centralized repository for access control and job management attributes such as batch and storage access policies, quotas, batch priorities and NIS attributes for cluster configuration. This paper describes the FERRY architecture, deployment and integration with services that consume the stored information. The Grid community has developed several access control management services over the last decade. Over time, services for Fermilab experiments have required the collection and management of more access control and quota attributes. At the same time, various services used for this purpose, namely VOMS-Admin, GUMS and VULCAN, are being abandoned by the community. FERRY has multiple goals: maintaining a central repository for currently scattered information related to users' attributes, providing a Restful API that allows uniform data retrieval by services, and providing a replacement service for all the abandoned grid services. FERRY is integrated with the ServiceNow (SNOW) ticketing service and uses it as its user interface. In addition to the standard workflows for request approval and task creation, SNOW invokes orchestration that automates access to FERRY API. Our expectation is that FERRY will drastically improve user experience as well as decrease effort required by service administrators.

Batching decisions in multi-item production systems with learning effect

The average time in the system (or average throughput time) is recognized as a crucial performance measure of manufacturing systems and, in recent years, has become even more relevant as markets have shifted towards mass-customization scenarios, since it directly affects the average inventory in the system. Modern manufacturing systems design paradigms, Industry 4.0 on top, focus on the need to achieve high responsiveness by shortening throughput time as the main lever to maintain cost efficiency and competitiveness in such market scenarios.In the contexts in which the human impact on production operations is not negligible, the learning phenomenon may reasonably affect the average time in the system. This paper analyses the learning effect on this performance measure through batching policies. Here, batching is considered as the possibility to group customized orders presenting similarities in the production processes, thus making it possible the exploitation of learning effect.Both single- and multi-item systems under Markovian arrivals are studied, also deterministic and stochastic processing times are considered. The problem that we pose consists in determining the batch size for each product that minimizes the average time in the system taking into consideration the learning effect, which is included by means of the Plateau model. A solution procedure for each case is discussed and, through numerical experiments, the sensitivity of the model to variations in parameter values is studied.

Optimizing automated sorting in warehouses: The minimum order spread sequencing problem

In warehouses, order consolidation processes are inevitable whenever picking orders are assembled under a zoning and/or batching policy. In our specific warehouse setting, bins containing partial orders picked under a zoning and batching policy are intermediately stored in an automated storage/retrieval system (ASRS) and, afterwards, released on a conveyor system supplying the consolidation area. Here, a fully-automated sorter merges partial orders dedicated to the same customer in packing stations where, finally, the customers’ cardboard boxes are packed. For sequencing the release of bins from the ASRS we derive an elementary optimization problem, which aims to minimize the spread of orders in the release sequence so that picking orders are quickly assembled at their packing stations. The resulting problem is formalized, computational complexity is proven, and efficient solution procedures are provided. Additionally, managerial aspects are addressed with the help of a comprehensive simulation study.

Effective bullwhip metrics for multi-echelon distribution systems under order batching policies with cyclic demand

A large number of problems in a distribution supply chain require that decisions are made in the presence of the bullwhip effect phenomenon. The impact of the order batching policies on the bullwhip effect is analysed in this paper, when cycle demand on a multi-echelon supply chain operating is considered. While investigating which bullwhip effect metrics are more adequate to measure the bullwhip effect in these type of systems, the optimal reordering plan that minimises the operation costs of the overall system is calculated. A Mixed Integer Linear Programming (MILP) model is developed that takes into account an inventory and distribution system formed by multiple warehouses and retailers with lateral transshipments. The bullwhip effect is measured through four metrics: the echelon average inventory; the echelon inventory variance ratio; the echelon average order; and the echelon order rate variance ratio. As conclusion the inventory metrics suggest that (i) using batching policy reduces instability; (ii) batching may reduce in general order variance if using larger batches and (iii) cycle demand length has no major impact in the bullwhip effect. A motivational example and a real word case study are used and tested.

Single-machine scheduling with learning effect and resource-dependent processing times in the serial-batching production

• Under the given resource allocation, the structural properties on job batching policies and batching sequencing are proposed. • An optimal batching policy is derived for the special case. • An effective hybrid GSA–TS algorithm is developed to solve the studied problem. In this paper, we study a single machine scheduling problem by simultaneously considering the processing method of serial-batching, learning effect, resource-dependent processing times, and setup operations. We consider minimizing the makespan as the objective of the studied problem under the constraint that the total resource consumption does not exceed a given limit. For the special case where the resource allocation is given, we first propose the structural properties for job batching policies and batching sequencing, and an optimal batching policy is derived based on these properties. Then, we develop a novel hybrid GSA–TS algorithm which combines the Gravitational Search Algorithm (GSA) and the Tabu Search (TS) algorithm to solve the general case. Computational experiments with different scales show the effectiveness and efficiency of the proposed algorithm.

Model for Order-Batching with Congestion Consideration

The paper presents the model for optimal number of merged orders that will reduce the average order-picking time. Two order batching policies were described: pick-then-sort and sort-while-pick. One-block rectangular warehouses were considered. The author studied two routing heuristics: s-shape and return for which the equations for average order-picking time can be designated based on probability distributions: binomial distribution and uniform distribution. The research takes into account the possible congestion problem. Based on simulations the analytical form of blocking time for pickers was proposed. The study shows that the advantages of each additional merged order is getting smaller. Using ABC classification for storage location assignment in warehouses with low number of aisles and large pick lists can extend the average order-picking time.

Some Properties of Batch Value of Information in the Selection Problem

Given a set of items of unknown utility, we need to select one with a utility as high as possible (“the selection problem”). Measurements (possibly noisy) of item values prior to selection are allowed, at a known cost. The goal is to optimize the overall sequential decision process of measurements and selection. Value of information (VOI) is a well-known scheme for selecting measurements, but the intractability of the problem typically leads to using myopic VOI estimates. Other schemes have also been proposed, some with approximation guarantees, based on submodularity criteria. However, it was observed that the VOI is not submodular in general. In this paper we examine theoretical properties of VOI for the selection problem, and identify cases of submodularity and supermodularity. We suggest how to use these properties to compute approximately optimal measurement batch policies, with an example based on a “wine selection problem”.

A Systematic Approach to Quality Oriented Product Sequencing for Multistage Manufacturing Systems

Product sequencing is one way to reduce cost and improve product quality for multistage manufacturing systems (MMS). However, systematically evaluating the influence of product sequence on quality performance for MMS is still a challenge. By considering the rate of incoming conforming product, manufacturing system quality transition between batch to batch, and quality propagation along stages, this paper investigates the appropriate batch policies and product sequencing for MMS so that satisfied quality performance can be achieved. A model to analyze the relationship between the product sequencing and quality performance is conducted just by using the quality inspection data and the complex engineering knowledge used in the variation method is avoided. Based on Markov Chain processes methodology, quality performance is modeled as a function of transition states jointly determined by multistage condition, product sequencing, incoming part quality, and propagation of the rate of conforming products among multistage. Quality related batch strategies are discussed for optimal quality performance. Two kinds of quality efficiency are put forward to facilitate the modeling and the discussion. The results of the model will lead to guidelines for quality management in multistage manufacturing systems.