I gave an invited talk at the 2024 INFORMS Annual Meeting in Seattle, titled 'Modular Calibration of a Digital Twin Model for Planning-Level Decision-Making in a Semiconductor Fab's AMHS.' I appreciate Professor Park's support for this opportunity. 

I have commenced a new research project funded by the Korea Research Foundation under the Ministry of Education, with a ₩180 million research grant as the Principal Investigator. The project, titled "AI-integrated Simulation Optimization for Smart Storage/Retrieval Systems", will be conducted from September 2024 to August 2027. This funding will enable us to advance our research in AI-driven simulation optimization to enhance the efficiency of smart storage and retrieval systems.

I gave a research presentation at the Spring KIIE Conference in Yeosu, South Korea. I introduced a surrogate-based optimization approach for yard template planning in a transshipment hub. The title is "Gaussian process-based yard template planning under vehicle congestion and container rehandling: a case study of Busan Port Terminal". This research addresses the difficulties when it comes to deal with simulation-based decision-making for a high-dimensional and combinatorial problem.

I and Professor Hong participated in the joint meeting on the recent digital twin research and simulation education at the University of Washington from February 18th to 23th. We had a couple of meetings on the research topics such as the digital twin and its industry application. We'd like to express gratitude to Professor Park for his effort for organizing this workshop.

I gave a presentation at the 'Semiconductor Smart Manufacturing Working Group' during a planning session at the 2023 KIIE conference. The title of my talk was "Simulation-based optimization alternatives for large-scale material handling systems". I would like to express my gratitude to my professors Hong and Kim for their support. Furthermore, I'd like to extend a special thanks to the audience for their constructive comments and attentive engagement throughout my presentation."

"Simulation optimization of collaborative handshake operations for twin overhead shuttle cranes in a rail-based automated container terminal under demand uncertainty" has been accepted for publication in IEEE Access.

Abstract) A handshake operation can mitigate workload imbalance and interference between twin transporters in a material handling system. Terminal operators in a rail-based automated container terminal can employ the handshake operation to twin overhead shuttle cranes (OSs) under maritime demand uncertainty. Since a handshake location is critical to collaboration performance, terminal operators often rely on simulation experiments with a manual iterative design to determine optimal handshake locations. However, the simulation optimization is still challenging when a simulation execution is computationally expensive. This study proposes a Bayesian optimization-based approach to expedite the decision-making process. The approach infers the conditional outcomes of a simulation and actively searches optimal handshake locations. Our optimization results show that the proposed approach maximizes the collaborations between the twin OSs within fewer simulation runs. This study also provides extensive simulation analysis of the handshake locations. The experiments indicate that a handshake location has a significant influence on the required space for handshake operations and the workloads of the twin OSs.

"A Simulation-based Optimization for Scheduling in a Fab: Comparative Study on Different Sampling Methods" has been accepted in Journal of the Korea Society for Simulation.

Abstract) A semiconductor fabrication facility(FAB) is one of the most capital-intensive and large-scale manufacturing systems that operate under complex and uncertain constraints through hundreds of fabrication steps. To improve fab performance with intuitive scheduling, practitioners have used weighted-sum scheduling. Since the determination of weights in the scheduling significantly affects fab performance, they often rely on simulation-based decision making for obtaining optimal weights. However, a large-scale and high-fidelity simulation generally is time-intensive to evaluate with an exhaustive search. In this study, we investigated three sampling methods (i.e., Optimal Latin hypercube sampling(OLHS), Genetic algorithm(GA), and Decision-based sequential search(DSS)) for the optimization. Our simulation experiments demonstrate that: (1) three methods outperform greedy heuristics in performance metrics; (2) GA and DSS can be promising tools to accelerate the decision-making process.

I participated in an international joint research and the preprint of our manuscript is now available here. The research title is "Active Learning of Piecewise Gaussian Process Surrogates". It is such a big honor for me to join the collaborative research with Chiwoo Park, Robert Waelder, Benji Maruyama, Soondo Hong, and Robert Gramacy. I truly learned a lot from the collaboration. 

Abstract) Active learning of Gaussian process (GP) surrogates has been useful for optimizing experimental designs for physical/computer simulation experiments, and for steering data acquisition schemes in machine learning. In this paper, we develop a method for active learning of piecewise, Jump GP surrogates. Jump GPs are continuous within, but discontinuous across, regions of a design space, as required for applications spanning autonomous materials design, configuration of smart factory systems, and many others. Although our active learning heuristics are appropriated from strategies originally designed for ordinary GPs, we demonstrate that additionally accounting for model bias, as opposed to the usual model uncertainty, is essential in the Jump GP context. Toward that end, we develop an estimator for bias and variance of Jump GP models. Illustrations, and evidence of the advantage of our proposed methods, are provided on a suite of synthetic benchmarks, and real-simulation experiments of varying complexity

I had a presentation at PNU Grand Performance Exchange Programme. My topic was "Simulation-based decision making in large-scale simulations".  Since many manufacturing and material handling systems have enlarged the scales of factories and warehouses to achieve economies of scale, simulation has been widely used as a promising tool to diagnose and predict their decision making for upcoming demands. In this presentation, I shared my recent research progress and got comments from other graduate students.  

"A GA-based Optimization of a Weighted Lot Targeting Rule in a Semiconductor Wafer Fab" was published in Journal of the Korean Institute of Industrial Engineers. The paper is now available here.

Abstract) Production scheduling in a semiconductor wafer fabrication (FAB) can be decomposed into two phases: lot targeting and lot dispatching. A weighted dispatching rule is a widely applied concept to obtain the production schedule in the FAB under its complex manufacturing factors. The weights of the dispatching rule should be carefully determined since the weights substantially impact the performance of the FAB. In this study, we investigate a weighted lot targeting rule considering the time-variant manufacturing factors, i.e., processing times, set-up operations, work-in-process levels, and transportation times for the bottleneck (photolithography) process. We propose a Genetic Algorithm (GA) to determine the efficient weights of the weighted lot targeting rule within a limited simulation run. Our simulation experiments demonstrate that the proposed approach outperforms widely used targeting rules under the time-variant manufacturing factors in the FAB.

"A Dynamic OHT Routing Algorithm in Automated Material Handling Systems" was published in Journal of Korean Society of Industrial and Systems Engineering. The publication is now available here.

Abstract) An automated material handling system (AMHS) has been emerging as an important factor in the semiconductor wafer manufacturing industry. In general, an automated guided vehicle (AGV) in the Fab’s AMHS travels hundreds of miles on guided paths to transport a lot through hundreds of operations. The AMHS aims to transfer wafers while ensuring a short delivery time and high operational reliability. Many linear and analytic approaches have evaluated and improved the performance of the AMHS under a deterministic environment. However, the analytic approaches cannot consider a non-linear, non-convex, and black-box performance measurement of the AMHS owing to the AMHS’s complexity and uncertainty. Unexpected vehicle congestion increases the delivery time and deteriorates the Fab’s production efficiency. In this study, we propose a Q-Learning based dynamic routing algorithm considering vehicle congestion to reduce the delivery time. The proposed algorithm captures time-variant vehicle traffic and decreases vehicle congestion. Through simulation experiments, we confirm that the proposed algorithm finds an efficient path for the vehicles compared to benchmark algorithms with a reduced mean and decreased standard deviation of the delivery time in the Fab’s AMHS.

"A Simulation-based Genetic Algorithm for a Dispatching Rule in a Flexible Flow Shop with Rework Process" was published in Korea Journal of BigData. The publication is now available here. 

Abstract) This study investigates a dynamic flexible flow shop scheduling problem under uncertain rework operations for an automobile pipe production line. We propose a weighted dispatching rule (WDR) based on the multiple dispatching rules to minimize the weighted sum of average flowtime and tardiness. The set of weights in WDR should be carefully determined because it significantly affects the performance measures. We build a discrete-event simulation model and propose a genetic algorithm to optimize the set of weights considering complex and variant operations. The simulation experiments demonstrate that WDR outperforms the baseline dispatching rules in average flowtime and tardiness.

"A Simulation Study of a Storage Policy for a Container Terminal " was published in Dynamics in Logistics. The book chapter is now available here.

Abstract) This paper proposes a storage policy for container terminals that handle large numbers of vessels and containers. The storage policy considers the estimated workload at a certain area in a given period; the partition of a storage block into subblocks; the proximities between containers belonging to the same group; the segregation between different groups of containers; and the stack heights of containers. We develop a framework for simulating container repositioning and vehicle congestion and use it to evaluate the yard crane productivity rate, amount of repositioning, and service time of a real-world port terminal. The preliminary result shows that the container terminal operates more efficiently under the storage policy with a bay as a subblock setting.