Model reduction for studying a Bike Sharing System as a closed queuing network

Bacem Samet (1, 2, 3, 4) , Florent Couffin (1, 2) , Marc Zolghadri (1, 2) , Maher Barkallah (3, 4) , Mohamed Haddar (3, 4)

1 QUARTZ - Laboratoire QUARTZ
2 SUPMECA - SUPMECA - Institut supérieur de mécanique de Paris
3 LA2MP - Laboratoire de recherche de Mécanique, Modélisation et Production
4 ENIS - École Nationale d'Ingénieurs de Sfax | National School of Engineers of Sfax

The users of a B2C Bike Sharing System are satisfied if they find always free bikes and free docks at the stations. Modeling such systems is challenging because they are large-scaled and characterized by their stochastic dynamic. To study the performance of a BSS, we rely on a queuing model solved by the Entropy Maximization. In this paper, we study the possibility of aggregating a number of stations in a single virtual station to reduce the complexity of the model and its resolution. Experiments are performed to qualify impacts of the model reduction on the accuracy of the obtained results.

Mots clés: Bike Sharing System; Performance Evaluation; Closed Queuing Network; Repetitive Service Random Destination; Entropy Maximization; Model reduction; Performance analysis; Poisson distribution; Probabilistic Analysis

Domaines: Sciences de l'ingénieur [physics]; Mathématiques [math]; Probabilités [math.PR]; Statistiques [math.ST]

Bacem Samet, Florent Couffin, Marc Zolghadri, Maher Barkallah, Mohamed Haddar. Model reduction for studying a Bike Sharing System as a closed queuing network. 8th Swedish Production Symposium, SPS 2018, May 2018, Paris, France. pp.0 - 0, ⟨10.1016/j.promfg.2018.06.055⟩. ⟨hal-01709541⟩