A New Separable Piecewise Linear Learning Algorithm for the Stochastic Empty Container Repositioning Problem

Table 2

Network parameters for three problem instances.

Test instance	P1	P2	P3
Number of ports	9	30	56
Total services	5	17	34
Across-region services	0	0	17
Intraregion services	5	17	17
Size of container	1	2	4
Planning horizon	2 weeks	6 weeks	12 weeks

Ports: Bangkok, Boston, Cai Mep, Charleston, Colombo, Colon, Dafeng, Dalian, Deason, Fuqing, Hai Phong, Halifax, Hochiminh, Hong Kong, Houston, Inchon, Jakarta, Kaosiung, Kobe, Kwangyang, Laem Chabang, Lianyungang, Long Beach, Los Angeles, Miami, Mobile, Moji, Nagoya, Nansha, New Orleans, New York, Ningbo, Norfolk, Oakland, Osaka, Penang, Port Kelang, Pusan, Qingdao, Savannah, Seattle, Shantou, Shanghai, Shekou, Shimizu, Sihanoukville, Singapore, Taipei, Tampa, Tocoma, Tokyo, Xiamen, Xingang, Yantian, Yokkaichi, and Yokohama. Services: Asia-North America: GME (9 weeks), GME2 (12 weeks), AWE (10 weeks), AWE2 (11 weeks), AWE3 (10 weeks), AWE4 (11 weeks), AWE5 (11 weeks), SEA (7 weeks), SEA2 (9 weeks), AAC (6 weeks), AAC2 (6 weeks), AAC3 (6 weeks), AAC4 (5 weeks), AAS (6 weeks), AAS2 (6 weeks), AAS3 (6 weeks), and AAS4 (6 weeks); Intra-Asia: AK12, AK2, AK47, AK49, AK5, PA1, JSM, KTX7, KTX3, CVT2, NCT, CHL-V, JCV, CTJ, CVT-VN, RBC2, and JVT. The context in the parentheses shows the rotation time for the across-ocean service routes.