A mathematical optimization framework integrating fuzzy queuing theory and dynamic parametric programming can substantially improve traffic throughput and reduce operational expenses across China’s Three Gorges Project waterway navigation complex. Researchers developed an uncertain multi-class customer fuzzy queuing nonlinear programming model, designated as UMFQNP, specifically configured to manage vessel traffic bottlenecks under severe operational uncertainties. The Three Gorges Project, situated in Yichang, China, operates as the world’s largest hydropower installation, functioning as a vital transportation conduit along the upper Yangtze River corridor. However, rapid regional economic expansion has generated shipping demands far exceeding original infrastructure capabilities, leading to severe waterway congestion. By systematically resolving the complex mathematical interactions among variable vessel arrivals, multi-stage facility processing speeds, and customer departure behaviors, the new computational approach provides structured multi-period operational schedules that enhance navigation performance while curbing logistical costs.
The operational context of the Three Gorges facility reveals a persistent mismatch between vessel transit demand and physical channel capacity. Following its completion in 2003, navigation conditions across the Sichuan River and reservoir zone improved dramatically, triggering substantial commercial shipping expansion across upstream provinces including Yunnan, Guizhou, Sichuan, and Chongqing, connecting toward downstream industrial centers such as Wuhan, Nanjing, and Shanghai. From 2004 to 2023, annual cargo throughput crossing the dam grew from 43 million tons to 174 million tons, representing an approximate fourfold increase characterized by a 7.5 percent average annual growth rate. Although annual throughput experienced a minor temporary reduction to 159 million tons in 2024 due to scheduled maintenance closures—including a 45-day maintenance halt at the northern lock and an 18-day flood-season suspension at Gezhouba Lock No. 1—overall transit pressure remains historically elevated. Because actual annual cargo volume has continually outpaced the lock’s designed yearly capacity of 100 million tons since 2011, prolonged queuing has developed into a chronic systemic impediment.
Escalating vessel queuing times demonstrate the acute nature of these navigational constraints. Recorded field statistics indicate that average vessel waiting times rose sharply from 44 hours in 2016 to approximately 150 hours in 2018, expanding further to 191 hours in 2021 and exceeding 200 hours by 2023. These prolonged delays inflict major economic losses on shipping enterprises, while amplifying navigational safety hazards, localized environmental pollution, and administrative friction among vessel crews. Although governmental authorities formally addressed the issue in the Chinese People’s Political Consultative Conference in 2024 and approved the construction of a new dam-crossing channel, physical completion of that infrastructure project is projected to require at least 10 to 15 years. Consequently, scientific optimization and coordinated scheduling of the existing operational mechanisms represent the only viable near-term strategy to alleviate congestion before long-term infrastructure assets become operational.
The existing physical complex relies on three distinct navigation routes: a primary double-lane five-stage ship lock, a mechanical ship lift with a maximum lifting capacity of 15,500 tons designed for rapid passenger and cargo transit, and an intermodal dam-overturning highway transshipment system. Operational logs from 2021 illustrate the distribution of traffic across these three channels: the main ship lock handled 146.26 million tons across 40,379 vessels, the ship lift moved 3.65 million tons across 4,803 vessels, and the dam-overturning road system accommodated 4.75 million tons across 3,630 vessels. Dam-overturning transportation operates as a supplementary bypass mechanism involving port loading, road haulage between upstream and downstream terminals, and unloading. While the lock provides the lowest unit operating cost followed by the lift, extreme delays prompt certain operators to bypass water locks entirely. Capturing these trade-offs requires modeling structural customer heterogeneity, as vessels carry dry bulk materials, liquid chemicals, general cargo, containers, roll-on/roll-off vehicles, and passenger traffic across four primary tonnage classes ranging from below 3,000 tons to over 8,000 tons.
To capture structural variability without distorting computational tractability, the researchers classified commercial vessels into 14 fuzzy categories based on cargo types, deadweight tonnage, and origin-destination travel distances. This classification reflects stark economic disparities across ship classes; for instance, high-value container ships and scheduled passenger vessels experience waiting time penalties estimated at three to four times those incurred by ordinary bulk carriers. Furthermore, freight revenues diverge significantly depending on downstream travel destinations, with long-haul transits to coastal ports yielding revenue multiples well above short voyages. Because arrival flows, service durations, operating expenses, and transport revenues fluctuate continuously under varying hydrologic and weather conditions, parameters are represented as discrete intervals and trapezoidal fuzzy numbers rather than fixed deterministic quantities.
The mathematical architecture of the UMFQNP model formulates an objective function seeking to minimize total system expenditure or maximize net social welfare across multiple planning horizons. The formulation incorporates the direct transit costs, waiting penalties, and lock operating expenses of water transit alongside the explicit transfer surcharges and port-handling tariffs incurred when vessels divert to dam-overturning highway channels. System constraints strictly enforce physical lock and lift throughput boundaries, highway cargo capacities, non-negativity parameters, and strict total freight conservation requirements. In parallel, the model embeds a behavioral queuing game formulation: when estimated queuing delays generate financial penalties that surpass the net commercial benefit of waiting, vessel operators balk and exit the queue to enter the dam-overturning road network. Empirical vessel monitoring data enabled the authors to calibrate a high-order polynomial equation linking dynamic queue lengths directly to empirical customer departure rates.
Solving this nonlinear interval queuing problem required a rigorous multi-stage mathematical procedure employing α-cut decomposition and parametric programming techniques. The multi-customer batch service mechanism of the five-stage lock was structured as a fuzzy queuing system with batch sizes of 4 to 5 vessels, while the ship lift was modeled as a single-unit processing facility. By sweeping α-cut confidence thresholds from zero to one, fuzzy trapezoidal input parameters were systematically translated into bounded deterministic intervals. System performance measures—specifically fuzzy queuing delay, overall residence duration, and queue length—were evaluated across these α levels. An interactive two-stage algorithm then decomposed the uncertain programming structure into upper-bound and lower-bound sub-models, resolving optimal decision vectors representing allocated vessel tonnage for each customer class across all operational routes and timeframes.
The computational model was evaluated using empirical operational data collected from field investigations and interviews with the navigation scheduling department of the Three Gorges Project during March, June, September, and December of 2019. Numerical optimization results at an α-cut level of one demonstrated marked operational gains across successive planning intervals designated as periods k equals 1, 2, and 3. During the initial period, the primary ship lock was assigned to process classes 2, 3, 5 through 8, and 10 through 13 entirely, alongside portions of classes 1, 9, and 14, yielding an annual lock volume between 7,215 and 7,432 million tons, while the ship lift processed all class 4 traffic and portions of classes 1, 9, and 14. Due to capacity thresholds, 370 to 790 kilotons of class 1 vessels diverted to the dam-overturning highway mode. As demand escalated in period 2, the model reallocated all class 1, 8, and 13 traffic to road transport, expanding ship lock throughput to between 8,493 and 8,725 million tons and ship lift volume to between 605 and 706 million tons, which elevated operational efficiency by 17.4 to 17.7 percent for the lock and 28.8 to 45.3 percent for the lift relative to period 1.
As transit demand intensified further during period 3, optimal scheduling reallocated additional vessel classes to bypass routes, directing all class 1, 2, 8, and 13 vessels plus fractions of classes 5 and 10 to dam-overturning transportation, which carried between 1,799 and 1,921 million tons. Across the entire case study evaluation, the optimized management strategy achieved cumulative throughput and operational efficiency gains of 21.8 to 22.4 percent for the primary ship lock and 44.9 to 46.9 percent for the mechanical ship lift. Concurrently, comprehensive systemwide financial performance improved markedly: the overall plan reduced total system logistics expenditures or elevated net economic benefits by 40 to 86 percent, establishing the practical value of dynamically segmenting fleet traffic and diverting high-cost classes toward complementary transshipment facilities during peak congestion intervals.
Sensitivity analyses conducted across diverse α-cut confidence levels illustrated how parameter uncertainty governs expected operational metrics. When the uncertainty parameter α was set to zero, interval fuzziness peaked, yielding wide bounds with waiting time ranges spanning 6 to 588 hours and queue lengths between 15 and 1,592 vessels. Conversely, at the highest confidence level where α equals one, performance indices tightened to their most probable values, showing expected delays of 12 to 15 hours and queue lengths of 32 to 40 vessels. Under this condition, the total system net economic outcome converged within a bounded range of minus 34.6 to positive 28.31 billion yuan. While the baseline model abstracted the navigation infrastructure into single-server queuing facilities with batch processing, the authors noted that future modeling extensions should incorporate multi-server configurations, multi-agent behavioral simulations, and reinforcement learning algorithms to accommodate real-time disruptions, providing adaptive decision support for large-scale inland waterway networks.
Subject of Research: Mathematical optimization and fuzzy queuing operations for waterway lock scheduling under uncertainty
Article Title: Optimizing ship dam-crossing operations under uncertainty: a fuzzy queuing-based approach
Article References: Zhang, H., Xie, H., Perera, S., & Nayanalochana, P. C. J. (2025). Optimizing ship dam-crossing operations under uncertainty: a fuzzy queuing-based approach. Management System Engineering, 4(1), Article 22. https://doi.org/10.1007/s44176-025-00056-z
Image Credits: AI Generated
DOI: 10.1007/s44176-025-00056-z
Keywords: Three Gorges Project, fuzzy queuing theory, nonlinear programming, waterway transportation, ship lock scheduling, dam-crossing logistics, Yangtze River shipping, parametric programming, traffic congestion mitigation, interval uncertainty, queue departure rate, vessel transit management, ship lift operations, dam overturning transport, operational efficiency, transportation economics
Cite Scienmag News
Denise Maddox. (November 20, 2026). Fuzzy Queuing Model Optimizes Ship Traffic Scheduling at Three Gorges Project. Scienmag. https://scienmag.com/fuzzy-queuing-model-optimizes-ship-traffic-scheduling-at-three-gorges-project/
Denise Maddox. "Fuzzy Queuing Model Optimizes Ship Traffic Scheduling at Three Gorges Project." Scienmag, 20 November 2026, https://scienmag.com/fuzzy-queuing-model-optimizes-ship-traffic-scheduling-at-three-gorges-project/. Accessed 20 November 2026.
Denise Maddox. "Fuzzy Queuing Model Optimizes Ship Traffic Scheduling at Three Gorges Project." Scienmag. November 20, 2026. https://scienmag.com/fuzzy-queuing-model-optimizes-ship-traffic-scheduling-at-three-gorges-project/








