Minimasi Jarak Tempuh Order Picking pada Gudang Distribution Center dengan Karakteristik Two-Cross Aisle Layout Pada Perusahaan Distributor Makanan

Rismawati, Aulia (2024) Minimasi Jarak Tempuh Order Picking pada Gudang Distribution Center dengan Karakteristik Two-Cross Aisle Layout Pada Perusahaan Distributor Makanan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5010201027-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

PT X adalah anak perusahaan dari produsen makanan terbesar di Indonesia yang fokus pada sektor distribusi. Cabang terbesar PT X di Surabaya memiliki gudang seluas 4,5 hektare, dengan lebih dari 900 SKU, dan menyumbang sekitar 15% dari total pendapatan 25 cabang. Permasalahan yang terdapat pada DC Surabaya adalah waktu tunggu yang tinggi, dimana disebabkan oleh aktivitas pergudangan yang kurang efektif dan efisien khususnya dalam proses order picking. Penelitian ini mengatasi permasalahan pengambilan barang, penumpukan, dan lembur pekerja dengan metode machine learning, khususnya Reinforcement Learning menggunakan algoritma Q-Learning, untuk mengoptimalkan proses order picking dalam layout gudang Two-Cross Aisle. Hasil penelitian menunjukkan perbandingan antara empat metode heuristic dan metode Reinforcement Learning. Metode heuristic yang digunakan adalah S-Shape, Largest Gap, Aisle-by-Aisle, dan Combined. Metode S-Shape memiliki jarak tempuh 353 meter dengan waktu tempuh 1270,8 detik. Largest Gap menempuh jarak 350 meter dalam 1260 detik. Aisle-by-Aisle dengan jarak 325 meter membutuhkan 1170 detik. Combined heuristic menempuh 305 meter dalam 1098 detik. Metode Reinforcement Learning menghasilkan jarak tempuh 300,9 meter dan waktu tempuh 1083,24 detik, menunjukkan efisiensi yang lebih baik dibanding metode heuristic dengan pengurangan jarak tempuh sebesar 5,9 meter dan waktu tempuh sebesar 14,76 detik dibandingkan metode combined heuristic yang paling optimal di antara metode heuristic lainnya. Dengan parameter learning rate 0,5, Reinforcement Learning dapat menghasilkan rute optimal yang dimulai dari lorong terjauh dan mengambil barang berdasarkan nilai q maksimal sebelum berpindah ke titik berikutnya. Hal ini mempercepat proses pengambilan dan mengurangi waktu idle, sehingga meningkatkan efisiensi keseluruhan operasi gudang. Penerapan Two-Cross Aisle layout di gudang memungkinkan optimasi rute pengambilan barang. Layout ini membantu dalam meminimalkan jarak tempuh dan mengurangi waktu operasi, karena picker dapat berpindah antar lorong dengan lebih efisien yang secara langsung mengurangi waktu dan jarak tempuh dalam proses order picking. Pengaturan ulang posisi pallet dan penambahan area aisle sebesar 3.457 m² mencerminkan usaha peningkatan alur sirkulasi tanpa mengorbankan kapasitas penyimpanan. Hasilnya, area gudang yang efektif meningkat menjadi 7.993 m², yang menunjukkan peningkatan signifikan dalam penggunaan ruang. Secara keseluruhan, integrasi metode reinforcement learning dengan tata letak two-cross aisle menghasilkan peningkatan efisiensi yang signifikan, mengurangi waktu dan jarak tempuh, serta meningkatkan keselamatan dan efektivitas operasional gudang.
====================================================================================================================================
PT X is a subsidiary of the largest food manufacturer in Indonesia that focuses on the distribution sector. PT X's largest branch in Surabaya has a 4.5 hectare warehouse, with more than 900 SKUs, and contributes about 15% of the total revenue of 25 branches. The problem found in DC Surabaya is high waiting time, which is caused by ineffective and inefficient warehousing activities, especially in the order picking process. This research addresses the problems of picking, stacking, and worker overtime with machine learning methods, specifically Reinforcement Learning using the Q-Learning algorithm, to optimise the order picking process in a Two-Cross Aisle warehouse layout. The results show a comparison between four heuristic methods and the Reinforcement Learning method. The heuristic methods used are S-Shape, Largest Gap, Aisle-by-Aisle, and Combined. The S-Shape method has a distance of 353 metres with a travel time of 1270.8 seconds. Largest Gap travelled a distance of 350 metres in 1260 seconds. Aisle-by-Aisle with a distance of 325 metres takes 1170 seconds. Combined heuristic travelled 305 metres in 1098 seconds. The Reinforcement Learning method resulted in a distance of 300.9 metres and a travel time of 1083.24 seconds, showing better efficiency than the heuristic method with a reduction in distance by 5.9 metres and travel time by 14.76 seconds compared to the combined heuristic method which is the most optimal among other heuristic methods. With a learning rate parameter of 0.5, Reinforcement Learning can generate an optimal route that starts from the furthest aisle and retrieves items based on the maximum q value before moving to the next point. This speeds up the picking process and reduces idle time, thereby improving the overall efficiency of warehouse operations. The application of the Two-Cross Aisle layout in the warehouse enables the optimisation of picking routes. This layout helps in minimising the travel distance and reducing the operation time, as pickers can move between aisles more efficiently which directly reduces the time and travel distance in the order picking process. The re-positioning of pallets and the addition of 3,457 m² of aisle area reflected efforts to improve circulation flow without compromising storage capacity. As a result, the effective warehouse area increased to 7,993 m², which represents a significant improvement in space utilisation. Overall, the integration of the reinforcement learning method with the two-cross aisle layout resulted in significant efficiency improvements, reduced travelling time and distance, and improved the safety and effectiveness of warehouse operations.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Order picking, distribution center warehouse, reinforcement learning, two-cross aisle layout.
Subjects:	T Technology > T Technology (General) > T57.62 Simulation T Technology > T Technology (General) > T57.84 Heuristic algorithms. T Technology > T Technology (General) > T59.7 Human-machine systems.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26201-(S1) Undergraduate Thesis
Depositing User:	Aulia Rismawati
Date Deposited:	25 Jul 2024 04:38
Last Modified:	25 Jul 2024 04:38
URI:	http://repository.its.ac.id/id/eprint/108809

Actions (login required)

View Item