Optimal Evacuation Control in Large Urban Networks with Stochastic Demand
Alexander Hammerl, Corresponding Author [email protected] Wen-long Jin [email protected] Ravi Seshadri [email protected] Thomas Kjær Rasmussen [email protected] Otto Anker Nielsen [email protected] · 2025
Read Optimal Evacuation Control in Large Urban Networks with Stochastic Demand by Alexander Hammerl, Corresponding Author [email protected] Wen-long Jin [email protected] Ravi Seshadri [email protected] Thomas Kjær Rasmussen [email protected] Otto Anker Nielsen [email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-06797.txt.
Introduction
Alexander Hammerl and four co-authors design evacuation control for large cities in a 2025 paper. Their framework plans releases of cars into a road network when demand is uncertain. A bathtub-style model tracks how trips complete over time across the whole network. Control takes the form of staged departure orders or ramp metering at entry points.
They prove that under a common condition on trip distances, the best release pattern falls steadily over time. With an entry cap it becomes a single switch from full release to throttling. A flood case study on a low, dense island in Copenhagen tested the policy on thousands of random demand cases. Controlled release cut expected queue buildup clearly.
Catalog record
| CREATOR | Alexander Hammerl, Corresponding Author [email protected] Wen-long Jin [email protected] Ravi Seshadri [email protected] Thomas Kjær Rasmussen [email protected] Otto Anker Nielsen [email protected] |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 68 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-06797
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Optimal Evacuation Control in Large Urban Networks with Stochastic Demand Alexander Hammerl, Corresponding Author [email protected] Wen-long Jin [email protected] Ravi Seshadri [email protected] Thomas Kjær Rasmussen [email protected] Otto Anker Nielsen [email protected] Abstract \AuthorHeaders Hammerl, Jin, Seshadri, Rasmussen and Nielsen 1 Abstract We develop a risk-aware Model Predictive Control (MPC) framework for large-scale vehicular evacuations. Traffic dynamics are captured by the Generalized Bathtub Model, which describes the network-wide trip completion rate by tracking the time evolution of the distribution of remaining trip distances. We model evacuation inflow as a stochastic inflow process, and employ origin gating as the control policy, implemented through staged departure orders or adaptive ramp metering. A convex objective integrates total evacuation delay with a generic hazard-exposure term which can embed any spatial risk field (e.g., flood depth, fire intensity). We prove that if the residual-distance distribution exhibits non-decreasing hazard rate, then the optimal origin-gating profile is necessarily monotone decreasing and, under an inflow cap, bang–bang (single switch). This result supplies a closed-form seed for numerical optimizations and clarifies why early heavy release followed by throttling is optimal. Furthermore, we demonstrate that the assumption of a non-decreasing hazard rate is always satisfied when the origins of evacuation movements are uniformly distributed over a convexly bounded evacuation zone—a property that is fulfilled in the vast majority of real evacuation scenarios, at least approximately. The framework is demonstrated through a flood evacuation scenario on Amager Island, a densely populated area of Copenhagen that faces significant flood risk due to its low elevation and coastal exposure. The Generalized Bathtub evacuation model is coupled with a lightweight shallow-water model parameterized using real bathymetric and topographic data from Amager Island. Across 10,000 stochastic demand scenarios, the MPC policy reduces the expected area-under-queue by an average of 27% compared to a no-control scenario. Keywords : Evacuation, Disaster Response, Model Predictive Control, Generalized Bathtub Model, Risk-Averse Optimization 2 Introduction Evacuations involving 1000 or more people occur approximately every two to three weeks in the United States ( [ 1 ] ), far more frequently than commonly perceived. Given the diverse range of hazards that trigger evacuations and their frequent occurrence, communities must be prepared to implement traffic control and management measures with minimal advance warning across virtually any location. Effective management of large-scale evacuations is critical in mitigating the risks posed by natural disasters such as hurricanes and flash floods. A sudden mass departure of vehicles can overwhelm transportation networks, leading to excessive congestion and dangerous delays. Trying to evacuate everyone at once can generate harmful gridlock that impedes evacuation progress, whereas carefully scheduling departures can dramatically improve clearance times. Our work introduces a risk-aware Model Predictive Control (MPC) framework tailored for city-scale evacuations under various types of disasters. Unlike traditional evacuation strategies that ignore evolving threat dynamics, our approach explicitly accounts for changing risk information and uncertainty in demand in a modular fashion that can accommodate any type of natural or man-made hazard. Traffic flow theory forms the analytical foundation for understanding vehicular dynamics and assessing the effectiveness of control measures during evacuations. A commonly used approach in traffic flow theory describes the spatiotemporal dynamics of vehicles on unidirectional road segments. This class includes macroscopic frameworks ( [ 2 ] , [ 3 ] , [ 4 ] , [ 5 ] , [ 6 ] ), which treat traffic as a continuous medium, as well as car-following approaches ( [ 7 ] , [ 8 ] , [ 9 ] , [ 10 ] ), which capture the behavior of individual vehicles. An alternative class of traffic flow models consists of reservoir frameworks, which describe the dynamics of spatially aggregated traffic quantities at the network level [ 11 , 12 , 13 , 14 , 15 ] . This approach typically relies on the assumption of a time-invariant relationship between vehicle accumulation and average flow in a homogeneous network—a relationship commonly referred to as the Network Fundamental Diagram (NFD). In some cases, these frameworks also explicitly describe the relationship between network accumulation and the number of completed trips; see, e.g., [ 16 ] , [ 17 ] , [ 18 ] , [ 19 ] . The approach in [ 20 ] (see also [ 21 ] , [ 22 ] ) extends this concept by incorporating the distribution of remaining trip lengths of individual vehicles, thereby introducing a spatial structure into the reservoir. This extended framework, known as the Generalized Bathtub Model, enables a more realistic representation of network dynamics—particularly in situations where trip lengths are highly heterogeneous. Such heterogeneity is especially relevant in evacuation scenarios, where all vehicles are directed toward a limited number of shelters or exits, resulting in strongly inhomogeneous trip distance distributions that must be explicitly accounted for. See also [ 23 ] for a discussion of the challenges associated with measuring such distance distributions. A parametrization of network-wide traffic behavior based on observations under normal traffic conditions—as is common in conventional NFD approaches—is therefore unsuitable for our application. [ 24 ] classify evacuation literature into three categories: demand modeling, traffic assignment and route selection, and evacuation strategies. Demand modeling estimates evacuation behavior by who evacuates, when, where, and by which mode. This encompasses evacuation decision-making, departure timing ( [ 25 ] , [ 26 ] ), destination choice ( [ 27 ] , [ 28 ] ), and mode selection( [ 29 ] ; [ 30 ] ). Beyond primary evacuation trips, demand models incorporate background traffic, intermediate trips such as family pickups ( [ 31 ] ), and shadow evacuations—unplanned evacuations by populations outside designated risk zones ( [ 32 ] ; [ 33 ] ). Traffic assignment and route selection address evacuee distribution across transportation networks and path selection behavior. Individual route choice typically reflects perceived safety and availability of real-time information, with evacuees preferring major highways and known routes ( [ 34 ] ; [ 35 ] ). Simulation-based assignment tools such as DYNASMART, DynusT and other evaluate evacuation performance under various scenarios ( [ 26 ] ). However, these tools rely on behavioral assumptions difficult to validate in rare evacuation contexts. The authors emphasize hybrid and en-route choice models to better capture dynamic driver responses. Empirical evidence indicates that maximum traffic flows during evacuations may differ substantially from normal conditions, necessitating adjustments such as Maximum Sustainable Evacuation Traffic Flow Rates [ 36 ] . Evacuation strategies encompass high-level policies and operational measures designed to optimize clearance times and safety. Common approaches include access-based gating, also known as staged evacuation [ 37 , 38 , 39 ] , and contraflow operations that reverse inbound lanes to increase outbound capacity ( [ 40 ] ). Research has examined integrating real-time traffic management tools, such as signal retiming and variable message signs, into evacuation operations [ 41 , 42 ] . Furthermore, an extensive body of literature exists on pedestrian flow evacuations. Representative studies on pedestrian evacuation and crowd disasters include [ 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 ] . Our work extends the existing literature reviewed above by the following key contributions and novelties: First, our framework operates in real-time