Modeling skiers flows via Wardrope equilibrium in closed capacitated networks

Demyan Yarmoshik, Igor Ignashin, Ekaterina Sikacheva, Alexander V. GasnikovarXiv 2025 · DOI · arXiv

Introduction

Traditional traffic modeling, such as that used for urban commuting, typically focuses on "open networks" where agents seek the shortest path between a fixed origin and destination. However, ski resorts present a unique challenge: they are "closed networks." In these environments, skiers do not aim to exit the system but rather move in continuous cycles—taking a lift up and skiing down—to maximize their total utility, or "fun-per-hour," within a limited timeframe.

The Closed Network Approach

The core innovation of this research is the shift from path-based modeling to cycle-based strategies. In a ski resort, the network is composed of nodes (lift bases and summits) and directed edges (lifts and slopes). A skier's strategy is defined as a closed loop within this graph.

A synthetic graph representing a closed network of ski lifts and slopes.

Unlike open networks where the total number of trips is often fixed, the number of trips in a closed network is an emergent property of congestion. As more skiers choose a specific cycle, the queue at the associated lift grows, increasing wait times and decreasing the cycle's overall utility.

Mathematical Formulation

The researchers define a utility function $\tau_c$ for each cycle $c$, calculated as the ratio of the "value" of the slopes in that cycle to the total time spent (skiing time plus lift wait time). To find a stable state where no skier can improve their experience by switching cycles, the problem is framed as a Variational Inequality (VI).

A key mathematical breakthrough in this work is proving that finding steady-state waiting times in these complex queueing networks is equivalent to a primal-dual convex optimization problem. This allows for the use of efficient solvers like CVXPY to handle the non-linear and non-smooth nature of queue formation.

Experimental Validation

The model was tested using synthetic networks to demonstrate its robustness and convergence properties. These included simple "toy networks" to analyze specific behaviors like parallel slope choices.

A toy network model showing parallel paths between two nodes, used to test non-linear queue behavior.

Numerical experiments on a 5-cycle synthetic network showed that the "Gap function"—a metric indicating distance from equilibrium—decreased monotonically to zero. Furthermore, the system consistently reached the same unique equilibrium distribution regardless of the initial skier distribution, highlighting the model's stability.

Conclusion and Future Directions

This study provides the first static user-equilibrium model tailored for the unique dynamics of ski resorts. While the current model assumes a homogeneous user class and steady-state conditions, it lays the groundwork for more complex simulations. Future work aims to incorporate multi-class formulations (accounting for different skier skill levels), dynamic temporal changes (like peak morning hours), and validation against real-world RFID gate data.

Summary updated

Abstract

We propose an equilibrium model of ski resorts where users are assigned to cycles in a closed network. As queues form on lifts with limited capacity, we derive an efficient way to find waiting times via convex optimization. The equilibrium problem is formulated as a variational inequality, and numerical experiments show that it can be solved using standard algorithms.
BibTeX
@misc{yarmoshik2025modeling,
  author = {Demyan Yarmoshik and Igor Ignashin and Ekaterina Sikacheva and Alexander V. Gasnikov},
  title = {{Modeling skiers flows via Wardrope equilibrium in closed capacitated networks}},
  year = {2025},
  eprint = {2509.13392},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2509.13392},
}
Figure