Seat Inventory Control for Intercity Passenger Rail: How It Works with Customer Satisfaction



Seat Inventory Control for Intercity Passenger Rail: How It Works with Customer Satisfaction

Authors

S Terabe and S Ongprasert, Kochi University of Technology, JP

Description

It is proved that seat allocation can improve not only overall revenue but also improve passenger load and rejection request. The overall passenger goodwill, or customer satisfaction, does not decrease if seat allocation control is employed.

Abstract

1. Introduction
Many railway companies employ gfirst-come-first-serve (FCFS)h concept in reservation system. The gfirst-come-first-serveh concept seems fair to passengers but not effective in overall passenger load and revenue management. For example, long distance passengers cannot purchase tickets because seats are not available in some intervals, during the wanted origin and destination (O-D), which are purchased earlier by short distance passengers. Moreover, those empty seats are waste of opportunity at the departure. To eliminate the bottleneck, we propose using seat inventory control in high speed rail. Seat inventory control is a concept in revenue management, for example keeping some seats for long distance passengers who may come later, instead of selling to earlier-comer short haul O-D passenger. The network can earn higher revenue and improve serviceability (in passenger-km) and overall passenger load by maximizing the utility of existing facilities.
Even though seat allocation control seems to be beneficial to Railway Company, the company concern about the loss of passengerfs goodwill to the company because seat allocation control reject some passengers to accept other passengers in order to increase overall revenue.
The objective of this study is to prove that seat allocation can improve not only overall revenue but also improve passenger load and rejection request. Finally, we show that the overall passenger goodwill, or customer satisfaction, does not decrease if seat allocation control is employed.

2. Objective functions
In our simulations, we set 3 objective functions as linier programming; passenger load maximization, revenue maximization and number of rejection minimization. The optimization in one objective may cause unwanted in other term, e.g. passenger load maximization may cause overall revenue decrease. Moreover, we summarize the features of each objective function as follows; (1) average passenger load factor (APLF) maximization, the firm may appeal that the improvement of passenger load which means serviceability improvement in term of passenger-km, which is beneficial to passengers. (2) Total revenue maximization: this objective is really beneficial to Railway Company but meaningless to passengers. Therefore, Railway Company may loss some goodwill form passengers if this objective is employed. (3) Number of rejection minimization: this objective should be beneficial to overall passengers, e.g. number of rejection decreases form 10% to 5%. However, it is still unclear whether the overall benefit of passenger improves, e.g. number of rejection decreases because a long distance ticket is rejected for 2 or more of shorter distance ticket.
What is the best answer for seat allocation optimization? The best answer must be the seat allocation that can maximize passenger load, maximize revenue and minimize number of rejection simultaneously. However, the simultaneous optimization of 3 objectives exists in some cases. Therefore, we conclude that the next best answer is the seat allocation that can maximize both passenger load and revenue simultaneously. The number of rejection maybe ignored if passenger load and revenue are optimized simultaneously. The optimization of each factor may occur simultaneously or separately depending on the patterns of passenger demand.
The objectives of this simulation are (1) to observe how many cases that optimization of 3 objectives occur simultaneously, 2 objectives occur simultaneously, and the characteristic of the rest and (2) to examine how seat allocation can improve revenue, passenger load and number of rejection simultaneously with the real passenger demand.

3. Data
Data were taken from a railway company, route A to D during 2001 August 1st-31st, 14 trains a day. Each train, there are 195 seats for reservation seats. From station 0 to station 22, there are totally 23 stations and 22 sections. To simplify calculation, 23 stations were divided into 4 nodes; node 1 for station 0 to 4, Node 2 for station 5 to 10, Node 3 for station 11 to 17, and Node 4 for station 18 to 22.

4. Results
We can show demand and number of accepted passenger in FCFS method and optimizations by 3 objective functions of 14 trains on August 1st. The results of total revenue, number of rejection, and APLF are shown are also shown in the table. From general observations of the results, we conclude that the characteristic of 3 types of demand are summarized as
(1) Peak-time train. Peak-time train is the train that has demand higher than capacity. Most of the time, passenger requests are rejected because there is no available space. The trains in this case are train 3, 4, 6 and 7 in August 1st.
(2) Off-peak train. Off-peak train is the train that has demand less than capacity. The trains in this case are train 8, 9, 10, 11, 12, 14 in August 1st.
(3) Intermediate train. Intermediate train is the train that has demand less than peak-time but higher than off-peak train. Most of the time, requests of long OD are rejected because seats are occupied by the shorter OD passengers. The trains in this case are train 1, 2, 5, 13 on August 1st.

5. Discussion and conclusion
Seat allocation can improve not only revenue but also average passenger load factor and number of rejection, which is important to railways in order to get merit from society and passenger. Therefore, it is worth to do seat allocation in high-speed railways. However, in extreme cases, optimization causes 100% rejection in some O-D which is unfair to those passengers. For example, in train 4 on August 26th, all 3 factors are improved by the optimization; however, all demand of OD 1-4, 4667 passengers, are rejected. It is unfair to 1-4 O-D passengers to be rejected all. For this reason, the firm should provide minimum number of seats to specific O-D for social fairness reason. The minimum number of seats can be decided by various policy of the firm, for example (1) whether the substitute modes are available in that O-D, (2) competitiveness in the route, (3) proportion of passenger demand, and (4) provide some proportion for first-come-first-serve and the rest for seat allocation.

Publisher

Association for European Transport