2025年第106期(总第1147期)
演讲主题：外卖平台的配送设计
主讲人：詹东远 中国科学技术大学教授
主持人：关旭 供应链管理与系统工程系主任、教授
活动时间：2026年01月07日（周三）14:30-16:00
活动地址: 管院大楼205教室
主讲人简介：
詹东远是中国科学技术大学管理学院教授，曾任伦敦大学学院（UCL）管理学院助理教授、副教授。他在南加州大学获得博士学位，在清华大学自动化系获得学士及硕士学位。他的研究聚焦服务运营，重点关注大规模服务系统中服务人员和顾客的策略行为，并包括行为经济学的考虑。他目前担任《Decision Sciences Journal》期刊副主编，POM期刊编委。他曾获多项学术荣誉，包括INFOMRS BOM最佳工作论文第二名、CSAMSE最佳论文奖第二名、NET Institute夏季研究基金，以及国家优秀自费留学生奖学金，Management Sciece和MSOM的优秀审稿人奖。
活动简介：
This talk presents two studies on the design of on-demand food delivery (OFD) systems.
The first paper investigates how restaurants should prioritize service between dine-in and OFD channels. We examine the impact of OFD quality on the performance of three prevalent service policies: (1) First-Come-First-Serve (FCFS), (2) PI, which prioritizes dine-in customers, and (3) PO, which prioritizes OFD customers. While introducing OFD expands demand by boosting overall sales, it also cannibalizes dine-in customers, shifting customers to the lower-margin OFD channel. We show that OFD should only be adopted when its quality is moderate. In the short term, with prices held constant, the widely used FCFS policy is often suboptimal. We propose a novel partial priority mechanism that substantially outperforms conventional policies.
The second paper explores hybrid delivery mode design. Using a spatial game theoretic model where vendors sell through a platform and choose among offered delivery modes, we analyze how each participating party sets its delivery boundary to maximize profit. Beyond pure self-delivery and pure platform delivery, we study two distance-based hybrid modes: SP, where vendors serve nearby customers and the platform fulfills distant ones, and PS, which reverses this allocation. We address the operationalization of hybrid delivery and demonstrate that the seemingly inefficient PS mode can maximize market coverage, especially under low-demand conditions. To alleviate misaligned incentives and promote adoption of hybrid delivery, we design a coordination contract that attains system optimality, where self-delivery becomes a complement rather than a substitute to platform delivery.
