A heuristic based batching and assigning method for online customer orders

Modern advanced planning systems offer the technical prerequisites for an allocation of “available-to-promise” (ATP) quantities—i.e. not yet reserved stock and planned production quantities—to different customer segments and for a real time promising of incoming customer orders (ATP consumption) respecting allocated quota. The basic idea of ATP allocation is to increase revenues by means of customer segmentation, as it has successfully been practiced in the airline industry. However, as far as manufacturing industries and make-to-stock production are concerned, it is unclear, whether, when, why and howmuch benefits actually arise. Using practical data of the lighting industry as an example, this paper reveals such potential benefits. Furthermore, it shows how the current practice of rule-based allocation and consumption can be improved by means of up-to-date demand information and changed customer segmentation. Deterministic linear programming models for ATP allocation and ATP consumption are proposed. Their application is tested in simulation runs using the lighting data. The results are compared with conventional real time order promising with(out) customer segmentation and with batch assignment of customer orders. This research shows that—also in make-to-stock manufacturing industries—customer segmentation can indeed improve profits substantially if customer heterogeneity is high enough and reliable information about ATP supply and customer demand is available. Surprisingly, the choice of an appropriate number of priority classes appears more important than the selection of the ATP consumption policy or the clustering method to be applied.

It is often believed that the square root formula called the EOQ only applies to situations where customers demand small quantities at a fairly constant rate. This note shows that essentially the same formula can be derived for lumpy customer orders occurring at a variable rate. Also, as the EOQ is derived assuming that a fixed is always ordered for replenishment it does not allow for the over-shoot when customer orders exceed the available stock. This note shows that when the EOQ is large relative to the average customer order the overshoot problem can be ignored. However, when the EOQ is small some tentative results involving exponentially distributed customer order quantities indicate that substantial savings can be made using a minimum order quantity policy.

In manual order picking systems, orders received from internal or external customers are collected by pickers that walk or ride through the warehouse. Generally, orders are grouped into several sub sets, i.e. batches to reduce picking time and cost. This paper considers the on-line order batching problem (OOBP) in an order picking warehouse with multiple pickers in which the maximum completion time of all batches (customer orders) has to be minimized. At first, a mathematical model is introduced for the off-line version of the problem. Then the on-line version of the problem is considered in which customer orders become available dynamically over time. Since the proposed model is NP-hard, a rule-based heuristic algorithm was proposed to solve the on-line problem. The main contributions of the present work is to propose a mathematical model for the off-line order batching problem with multiple pickers considering makespan minimization and to present a novel heuristic algorithm for solving the on-line version of the problem. To validate the proposed algorithm, it is proved that its competitive rate is equal to 2. Finally, the solution algorithm is evaluated through a series of experiments and the most appropriate routing, batching and selection policies are introduced.

In an ideal e-business environment, when a customer order is scheduled and a ship date is computed, the availability should immediately be reserved and not be available for future orders. However, in reality the availability data that are used for the scheduling the orders are not real time availability (physical availability), but they are availability information stored in an IT system (system availability). The availability data in the IT system (static view of availability) is typically refreshed (synchronized with real time availability) only periodically since it is very expensive to update the database in real time. Due to this potentially inaccurate view of the availability, some orders cannot be shipped on the promised ship date. Therefore, for certain customer orders, products are shipped later than the promised ship date resulting in customer dissatisfaction. Therefore, one of key decisions in order fulfillment process is to properly balance IT system (e.g., IT expense) and accuracy of promised ship date. In this work, we study how availability fresh rate (IT system) impacts customer service level. The simulation model we develop helps making critical business decision on refresh rate of availability, and avoiding expensive IT investment.

A queueing model with variable arrival rates

PurposeThis purpose of this paper is to present a methodology for optimally planning long‐haul road transport activities through proper aggregation of customer orders in separate full‐truckload or less‐than‐truckload shipments in order to minimize total transportation costs.Design/methodology/approachThe model is applied to a specific Italian multi‐plant firm operating in the plastic film for packaging sector. The method, given the order quantities to be shipped and the location of customers, aggregates shipments in subgroups of compatible orders resorting to a heuristic procedure and successively consolidates them in optimized full truck load and less than truck load shipments resorting to a Genetic Algorithm in order to minimize total shipping costs respecting delivery due dates and proper geographical and truck capacity constraints.FindingsThe paper demonstrates that evolutionary computation techniques may be effective in tactical planning of transportation activities. The model shows that substantial savings on overall transportation cost may be achieved adopting the proposed methodology in a real life scenario.Research limitations/implicationsThe main limitation of this optimisation methodology is that an heuristic procedure is utilized instead of an enumerative approach in order to at first aggregate shipments in compatible sets before the optimisation algorithm carries out the assignments of customer orders to separate truckloads. Even if this implies that the solution could be sub‐optimal, it has demonstrated a very satisfactory performance and enables the problem to become manageable in real life settings.Practical implicationsThe proposed methodology enables to rapidly choose if a customer order should be shipped via a FTL or a LTL transport and performs the aggregation of different orders in separate shipments in order to minimize total transportation costs. As a consequence, the task of logistics managers is greatly simplified and consistently better performances respect manual planning can be obtained.Originality/valueThe described methodology is original in both the kind of approach adopted to solve the problem of optimising orders shipping in long‐haul direct shipping distribution logistics, and in the solution technique adopted which integrates heuristic algorithm and an original formulation of a GA optimisation problem. Moreover, the methodology solves both the truckload assignment problem and the choice of LTL vs FTL shipment thus representing an useful tool for logistics managers.

Arrival rate is the number of passengers arriving for elevator service in a certain period of time. Arrival rate is fundamental in expressing the heaviness of the traffic. Hence, it is vital for determining the required number of elevators and the specifications of each elevator such as the speed, capacity, and sector sizes. The passenger arrival process is a random process that is full of noise, and a processing step is required to extract the arrival rate from recorded arrival times of passengers. This work develops a real-time estimator and a benchmark for estimating the arrival rate. There are three contributions in this work; the first is suggesting a benchmark for estimating arrival rate; singular spectrum analysis extracts the arrival rate from noisy data. Hence, singular spectrum analysis is suggested as a benchmark for evaluating the performance of other algorithms. Even though singular spectrum analysis is powerful in extracting the arrival rate, it is not convenient for updating the arrival rate in real time. The second contribution is developing a real-time estimator for the passenger arrival rate that updates its parameters dynamically; dynamic exponentially weighted moving average was developed to provide instantaneous arrival rate updates. The third contribution is introducing exponentially weighted moving average as a linear model for passenger arrival, which opens the door to a large number of model-based algorithms in control theory; Kalman filtering was developed in this work on the top of the EWMA linear model. The results of applying Kalman filtering and DEWMA to real-life data show them as efficient methods for estimating passenger arrival rate to the elevators in real time. Practical application: The methods presented in this paper would allow an elevator controller designer to detect the intensity of the passenger arrival rate. By doing this, it is possible for the elevator controller to switch between different group control algorithms. For example, it could decide to switch from conventional group control to sectoring control and vice versa.

This paper considers a manufacturer with a two-station make-to-stock (MTS) and make-to-order (MTO) serial production system. The MTS facility processes a single type of raw material and provides these materials for the MTO facility. The facility in the second stage produces end items using the material from the order of firms. The materials processed by the MTS facility are either stocked for production in the second stage or consumed to satisfy exogenous demand from the material market. Revenues are generated in each stage by satisfying the material demands and the end item demands, respectively. The manufacturer can either accept or reject some of the exogenous material demands according to the status of the system and the rejected exogenous material demands incur no penalty. The orders of firms can be rejected with penalty. Holding costs are incurred for the materials and the end items in inventory at each stage. Demand processes and production times of the component and the end item are stochastic. The set of decision epochs in our model corresponds to the epochs of customer order and material demand arrivals, MTO production completion, and material process completion. At each decision epoch, the manufacturer determines whether or not to process materials. An the epochs corresponding to a customer order arrival or a material demand arrival, a decision must be made whether to accept or reject it. This paper studies a joint order admission, demand control, and material process rate control policy which maximizes the manufacturer’s profit. Using the Markov decision process model, we investigate the structural properties of the optimal policy using a numerical example. The numerical investigation indicates that the optimal policy can be defined by the monotone stitching curves. Specifically, the optimal policy will accept more customer orders as material inventory increases and accepts fewer material demands as the number of outstanding customer orders increases. Furthermore, it is more profitable to increase the material production rate as the number of outstanding customer orders increases. We also examine the impact of the material process rate control option on the manufacturer’s profit. The study confirms that the material processing policy with a controllable production rate has a great percentage improvement in the average profit over the one with a fixed production rate. The study also show that the performance improvement due to using the controllable production rate rather than the fixed one is more prominent for the cases with a lower MTO customer arrival rate, a lower unit MTO sales revenue, a lower MTO production rate, a lower unit material sales revenue or a lower material holding cost. It is also observed that the impact of the material demand rate on the performance difference between the policies with a controllable production rate and a fixed one can vary depending on the value of other parameters.

The Industry 4.0 environment enables direct communication between the manufacturer's shop floor and a customer. Thus, the manufacturer is able to respond to the customers' requests more quickly, meaning that manufacturers must now more tightly control the shop floor planning and scheduling. Here we present a simulation-based scheduling model for Flexible Manufacturing System dynamic shop-floor control. The customer's order and the processing sequence table of the products are imported into the simulation model. Experiments are implemented for the case wherein the system encounters unexpected conditions. The proposed approach represents a potential tool for manufacturers to make decisions in the real time by further connecting to the Enterprise Resource Planning and Manufacturing Execution System.

In manual order picking systems, order pickers walk or drive through a distribution warehouse in order to collect items which are requested by (internal or external) customers. In order to perform these operations efficiently, it is usually required that customer orders are combined into (more substantial) picking orders of limited size. The Order Batching Problem considered in this paper deals with the question of how a given set of customer orders should be combined such that the total length of all tours is minimized which are necessary to collect all items. The authors introduce two metaheuristic approaches for the solution of this problem: the first one is based on Iterated Local Search; the second on Ant Colony Optimization. In a series of extensive numerical experiments, the newly developed approaches are benchmarked against classic solution methods. It is demonstrated that the proposed methods are not only superior to existing methods but provide solutions which may allow distribution warehouses to be operated significantly more efficiently.

This article considers the order batching problem in steelmaking and continuous-casting production. The problem is to jointly specify the slabs needed to satisfy each customer order and group all the slabs of different customer orders into production batches. A novel mixed integer programming model is formulated for the problem. Through relaxing the order assignment con- straints, a Lagrangian relaxation model is then obtained. By exploiting the relationship between Lagrangian relaxation and column generation, we develop a combined algorithm that contains nested double loops. At the inner loop, the subgradient method is applied for approximating the Lagrangian dual problem and pricing out columns of the master problem corresponding to the linear dual form of the Lagrangian dual problem. At the outer loop, column generation is employed to solve the master problem exactly and adjust Lagrangian multipliers. Computational experiments are carried out using real data collected from a large steel company, as well as on large-scaled problem instances randomly generated. The results demonstrate that the combined algorithm can obtain tighter lower bound and higher quality solution within an acceptable computation time as compared to the conventional Lagrangian relaxation algorithm. © 2011 Wiley Periodicals, Inc. Naval Research Logistics 58: 370-388, 2011

Today's distribution warehouses often need to process a far higher volume of smaller orders of multiple products which considerably increases logistics costs. They use the so-called order picking (OP) systems where products have to be picked from a set of specific storage locations by an OP process usually driven by production batches or customer orders. The OP is often very labour-intensive and its efficiency largely depends on the distance the order pickers have to travel, which therefore needs to be minimised. Minimising this distance is affected by several factors e.g. facility layout, shape of storage area, and especially the storage assignment strategy. Products that are frequently ordered together in multi-item, less than unit load customer orders should be stored near each other: this is the correlated storage assignment strategy. This study develops, tests and compares a set of different storage allocation rules based on the application of original similarity coefficients and clustering techniques. Lastly, a case study demonstrates the effectiveness of the proposed rules in minimising logistic costs.

Influence of lot size and planned lead time on service level and inventory for a single-stage production system with advance demand information and random required lead times

Internet Protocol (IP) has brought rapid development in many applications such as video and audio streaming, Voice-over-IP (VoIP) and e-commerce. However, these applications suffer from congestion problem, which severely worsens the network performance of real time data transmissions. In this paper, we propose a fuzzy based congestion detection technique for queuing in IP networks. This technique classifies the flow as real time and non real time and the priority scheduler prioritizes the classified flows as high and low, respectively. The congestion level of high priority flows is detected by means of fuzzy logic system. The queue delay and arrival rate are considered as input for Fuzzy logic and the level of congestion is estimated. According to the congestion level of flows, they are scheduled in two different queuing mechanisms. We simulate our technique using NS-2 and simulation results prove the efficiency of our approach. It eradicates unsolicited packet drops and improves packet delivery ratio.

In IEEE 802.11 wireless LANs, channel quality, network load, as well as the protocol itself are time-varying, limiting the goodput performance improvement in wireless LANs. Therefore, it becomes critical to dynamically adjust parameters of MAC and PHY layers according to variations of channel quality. In this paper, we propose variable frame size and variable data rate schemes for goodput enhancement. We first propose two optimal frame size predictors: a goodput regulator to maintain the committed goodput for non-greedy applications and an optimal frame size predictor for maximizing the goodput for greedy applications. Then, we propose a data rate drafting scheme and develop a variable size and variable rate (VSVR) scheme for further goodput improvement. Our extensive simulation results show that the proposed VSVR algorithm can double the channel goodput of current implementations. Moreover, the proposed scheme can be easily integrated with the current implementations of the wireless LAN MAC protocol. Copyright © 2004 John Wiley & Sons, Ltd.