This paper addresses a scheduling problem aimed at minimizing makespan in a permutation flow shop with two machines and an inspection process that must be conducted at least once every certain number of outcomes from the first machine. A mathematical programming approach and a genetic algorithm, incorporating Johnson's rule and a specific mutation process, were developed to solve this problem. Johnson's rule was used to generate an initial population, while the mutation process ensured compliance with the inspection constraints. The results showed that within a computation time limit of 300 seconds, the mathematical programming approach often failed to provide optimal or feasible solutions, especially for larger job sets. For instance, when the process times of both machines were similar and the inspection time was longer, the mathematical programming approach failed to solve all 10 experiments with just 15 jobs and only had a 50% success rate for 100 jobs. In contrast, the proposed genetic algorithm solved all instances and delivered equal or superior results compared to the mathematical programming approach.

As various types of products are produced in a single production system, it is important to determine a scheduling policy that selects one of the different types. In addition, the failure of processes in a line need to be considered due to machine failure, raw material supply and demand, quality issues, and worker absence, etc. Therefore, we studied production systems with various product types, dedicated buffers for each product type, Bernoulli equipment, and WIP-based scheduling or cyclic scheduling. To analyze such system exactly, we introduced a method to analyze the performance such as production rate, WIP level, blocking probability and starvation probability based on Markov chains and derived various characteristics. Especially, assuming that equipment does not need to select the type it just tried, the flow rate is no longer conserved and increasing buffer capacity does not guarantee increase production rate. The performance comparison between WIP-based and cyclic policy is studied as well.

This study develops a model to determine the input rate of the chemical for coagulation and flocculation process (i.e. coagulant) at industrial water treatment plant, based on real-world data. To detect outliers among the collected data, a two-phase algorithm with standardization transformation and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) is applied. In addition, both of the missing data and outliers are revised with linear interpolation. To determine the coagulant rate, various kinds of machine learning models are tested as well as linear regression. Among them, the random forest model with min-max scaled data provides the best performance, whose MSE, MAPE, R2 and CVRMSE are 1.136, 0.111, 0.912, and 18.704, respectively. This study demonstrates the practical applicability of machine learning based chemical input decision model, which can lead to a smart management and response systems for clean and safe water treatment plant.

Many manufacturers applying third party logistics (3PLs) have some challenges to increase their logistics efficiency. This study introduces an effort to estimate the weight of the delivery trucks provided by 3PL providers, which allows the manufacturer to package and load products in trailers in advance to reduce delivery time. The accuracy of the weigh estimation is more important due to the total weight regulation. This study uses not only the data from the company but also many general prediction variables such as weather, oil prices and population of destinations. In addition, operational statistics variables are developed to indicate the availabilities of the trucks in a specific weight category for each 3PL provider. The prediction model using XGBoost regressor and permutation feature importance method provides highly acceptable performance with MAPE of 2.785% and shows the effectiveness of the developed operational statistics variables.

To meet rapidly changing market demands, manufacturers strive to increase both of productivity and diversity at the same time. As a part of those effort, they are applying flexible manufacturing systems that produce multiple types and/or options of products at a single production line. This paper studies such flexible manufacturing system with multiple types of products, multiple Bernoulli reliability machines and dedicated buffers between them for each of product types. As one of the prevalent control policies, priority based policy is applied at each machines to select the product to be processed. To analyze such system and its performance measures exactly, Markov chain models are applied. Because it is too complex to define all relative transient and its probabilities for each state, an algorithm to update transient state probability are introduced. Based on the steady state probability, some performance measures such as production rate, WIP-based measures, blocking probability and starvation probability are derived. Some system properties are also addressed. There is a property of non-conservation of flow, which means the product ratio at the input flow is not conserved at the succeeding flows. In addition, it is also found that increased buffer capacity does not guarantee improved production rate in this system.

This research examines deep learning based image recognition models for beef sirloin classification. The sirloin of beef can be classified as the upper sirloin, the lower sirloin, and the ribeye, whereas during the distribution process they are often simply unified into the sirloin region. In this work, for detailed classification of beef sirloin regions we develop a model that can learn image information in a reasonable computation time using the MobileNet algorithm. In addition, to increase the accuracy of the model we introduce data augmentation methods as well, which amplifies the image data collected during the distribution process. This data augmentation enables to consider a larger size of training data set by which the accuracy of the model can be significantly improved. The data generated during the data proliferation process was tested using the MobileNet algorithm, where the test data set was obtained from the distribution processes in the real-world practice. Through the computational experiences we confirm that the accuracy of the suggested model is up to 83%. We expect that the classification model of this study can contribute to providing a more accurate and detailed information exchange between suppliers and consumers during the distribution process of beef sirloin.

Many small and medium-sized manufacturing companies process various product types to respond different customer orders in a single production line. To improve their productivity, they often apply batch processing while considering various product types, constraints on batch sizes and setups, and due date of each order. This study introduces a batch scheduling heuristic for a production line with multiple product types and different due dates of each order. As the process times vary due to the different batch sizes and product types, a recursive equation is developed based on a flow line model to obtain the upper bound on the completion times with less computational complexity than full computation. The batch scheduling algorithm combines and schedules the orders with same product types into a batch to improve productivity, but within the constraints to match the due dates of the orders. The algorithm incorporates simple and intuitive principles for the purpose of being applied to small and medium companies. To test the algorithm, two case studies are introduced; a high pressure coolant (HPC) manufacturing line and a press process at a plate-type heat exchanger manufacturer. From the case studies, the developed algorithm provides significant improvements in setup frequency and thus convenience of workers and productivity, without violating due dates of each order.