Source-linked AI summary
New insights into pedestrian flow through bottlenecks
Armin Seyfried, Tobias Rupprecht, Oliver Passon, Bernhard Steffen, Wolfram Klingsch, Maik Boltes
TL;DR
Pedestrian-facility capacity estimates vary, and proposed stepwise capacity growth challenges the specific-flow concept. This study experimentally examines unidirectional flow through bottlenecks of different widths and finds that flow increases linearly with width.
Problem
Capacity-estimation procedures produce different flow values, while stepwise capacity growth has been proposed, challenging assumptions used to dimension pedestrian facilities.
Source: §1 · pp. 1–2, §4.2 · p. 13Method
The authors experimentally studied unidirectional pedestrian flow through laboratory bottlenecks of different widths while seeking to control other influences on flow.
Source: §2 · pp. 4–5Results
Flow increases linearly and continuously with bottleneck width, with no general evidence for stepwise growth.
Source: §4.2 · p. 13Takeaways & Limitations
For bottlenecks, maximum flow is only an upper limit on incoming flow that produces a jam; it does not rule out jams at lower inflows.
Source: §1 · p. 2An improved data basis is needed for a final judgment.
Source: §3.3 · p. 10
Abstract
from arXiv · showhide
Capacity estimation is an important tool for the design and dimensioning of pedestrian facilities. The literature contains different procedures and specifications which show considerable differences with respect to the estimated flow values. Moreover do new experimental data indicate a stepwise growing of the capacity with the width and thus challenge the validity of the specific flow concept. To resolve these differences we have studied experimentally the unidirectional pedestrian flow through bottlenecks under laboratory conditions. The time development of quantities like individual velocities, density and individual time gaps in bottlenecks of different width is presented. The data show a linear growth of the flow with the width. The comparison of the results with experimental data of other authors indicates that the basic assumption of the capacity estimation for bottlenecks has to be revised. In contradiction with most planning guidelines our main result is, that a jam occurs even if the incoming flow does not overstep the capacity defined by the maximum of the flow according to the fundamental diagram.
1 Introduction
The introduction frames pedestrian bottleneck capacity estimation as important for facility design but uncertain, with competing assumptions about how capacity changes with width and when queues form. It motivates a controlled experiment measuring flow, density, and velocity inside bottlenecks while a jam develops.
1 Introduction
- A jam is commonly assumed to form when incoming flow exceeds bottleneck capacity [7], but the continuity equation does not rule out jams at lower incoming flows.The introduction distinguishes capacity as an upper limit for incoming flow that produces a jam, not a necessary threshold.Source: §1 · p. 2
- Standard specific-flow reasoning implies capacity increases linearly with width, whereas Hoogendoorn and Daamen argue lane formation could make capacity rise stepwise.Their proposed zipper effect produces lanes whose spacing is independent of bottleneck width, so capacity would increase only when another lane forms.Source: §1 · pp. 2–3
- Capacity estimates differ by more than a factor of two, while measured capacities and estimation procedures can differ by more than a factor of four.The authors note that experimental measurements also vary, despite being conducted under controlled laboratory conditions with participants asked to move normally; the table compares the selected estimates and measurements.Source: Table 1 · p. 3, §1 · p. 4
- The experiment varies bottleneck width and pedestrian count under controlled initial conditions, tracking trajectories to resolve flow, density, and velocity as a jam occurs.The study uses a homogeneous group under laboratory conditions and varies no other stated setup factors.Source: §1 · p. 4
2 Experimental setup
The setup was designed to isolate bottleneck-width effects while keeping runs comparable, and to test whether results depended on the number of pedestrians.
2 Experimental setup
- The experiment aimed to keep conditions consistent across widths and account for limited resources by testing the influence of pedestrian numbers.Source: §2 · p. 4
- The bottleneck was 2.8 m long, maintained constant width from hips to shoulders, and had holding areas to set equal initial density for each run.The first holding-area center was 3 m from the bottleneck entrance.Source: §2 · p. 5
- Students and ZAM staff were instructed to walk purposefully without haste or pushing, at normal speed, and began after an acoustic signal; two cameras filmed each run.Source: §2 · p. 5
- 18 runs tested bottleneck widths from 0.8 to 1.2 m in 0.1-m steps with 20, 40, or 60 pedestrians, starting at density 3.3 m^-2.Source: §2 · p. 5
3 Data analysis
Across bottleneck widths, the analysis finds continuous changes in pedestrian-flow observables, with a lane transition but no evidence of a stepwise change. It also identifies width-related variation in specific flow and time-dependent changes in velocity and density.
3.1 Specific flow
- For N = 60, specific flow shows a small but systematic increase with width, while for N < 60 the variation is partly obscured by fluctuations.The authors attribute the N = 60 pattern to the zipper effect and bottleneck boundaries; at small N, there may not be enough time to reach a stationary state. Specific flow is calculated as ∆N/(∆t b), with ∆t measured between the first and last person at y = 0.4 m.Source: §3.1 · p. 6, Table 2 · p. 6
3.2 Trajectories and probability distributions
- Pedestrian trajectories were manually tracked from overhead video, then checked and corrected for perspective and pixel scaling before analysis.Markers followed the center of each head in every second frame (80 ms); the resulting trajectories were used to examine positions, time gaps, and time-dependent flow quantities.Source: §3.2 · pp. 6–7
- For b ≥0.9 m, lane formation is visible, but lane spacing and time-gap distributions change continuously with width, with no indication of a stepwise change in the observables.The transition from one to two lanes is the exception to the otherwise continuous changes; time gaps broaden and drift to smaller values as width increases.Source: §3.2 · p. 7, Figure 4 · p. 8
3.3 Time dependence of ρ, vi, and ∆ti
- Velocity decreases and density increases over time, while time-gap trends are difficult to identify because those changes largely compensate and zipper-effect jumps obscure them.The density estimate fluctuates substantially because the small observation area contains only 1–4 people.Source: §3.3 · p. 9, Figure 6 · p. 10
- Regression results suggest relaxation is almost width-independent, but for b ≥1.0 m a stationary state is not reached even with N = 60; stationary density rises with width while velocity decreases less markedly.The authors explain the density increase through the zipper effect and boundary effects, and caution that a final judgment requires more data or more test persons.Source: §3.3 · pp. 9–10
4 Combined analysis with data from other experiments
The combined analysis attributes apparent differences among bottleneck-flow measurements to different initial conditions, then examines how flow depends on bottleneck width and when jams occur.
4.1 Comparison with the data of other experiments
- Across laboratory measurements, differences in initial density explain elevated flows better than bottleneck geometry or position, which have minor effects.For b = 1.2 m, Nagai’s flow increased from J = 1.04 s−1 to 3.31 s−1 as initial density rose from ρini = 0.4 m−2 to 5 m−2; the data collection included 30 to 180 participants asked to move normally.Source: §4.1 · pp. 11–12, Figure 9 · p. 13
4.2 Linear dependence of flow and bottleneck-width
- Flow increases linearly and continuously with bottleneck width; the small edge near b = 0.7 m is associated with the zipper effect, not general stepwise growth.The authors report no evidence for constant-distance lanes, an assumption used to motivate stepwise growth.Source: §4.2 · p. 13
4.3 The criteria for jam occurrence
- Under normal conditions, bottleneck flow does not reach the capacity defined by the fundamental diagram’s maximum; jams occur well before that maximum.The authors identify flow fluctuations, irregular pedestrian organization at the entrance, and people’s preference for larger distances as possible reasons for jamming below capacity.Source: Figure 11 · p. 15, §4.3 · pp. 15–16
5 Summary
Experiments found that bottleneck flow grows approximately linearly with width, while jamming can occur below the maximum flow predicted by the fundamental diagram. The authors caution that their laboratory measurements do not directly establish design values, but say these qualitative findings can inform capacity estimates.
5 Summary
- Trajectory analysis tracked individual velocities, local densities, and time gaps across bottleneck widths, finding continuous changes in time-gap distributions and lane distance when the zipper effect acted.The authors compared velocity and density results with common capacity-estimation methods and prior experiments.Source: §5 · p. 16
- For bottlenecks wider than 0.7 m, flow increased linearly and continuously with width across bottleneck types and initial conditions; geometry had only minor influence.The exception was near b ≈0.7 m, where the zipper effect begins to act.Source: §5 · p. 16
- Raising initial density from 1.8 m−2 to 5 m−2 increased flow, indicating a rising fundamental-diagram slope in that range and capacity at densities substantially above 1.8 m−2.The measurements used young participants under laboratory conditions, so absolute flow values should be interpreted with that scope in mind.Source: §5 · p. 16
- ρ ≈1.8 m−2 characterized stationary flow inside the bottleneck when a jam formed, even though flow remained below the fundamental-diagram maximum.The authors suggest stochastic flow fluctuations, local organization, or psychological changes in access incentives as possible causes.Source: §5 · pp. 16–17
- Absolute laboratory measurements cannot be transferred directly into design recommendations, but the authors say the qualitative findings are robust to measurement scaling.They call for studying when density rises inside bottlenecks and propose improving planning-guideline capacity estimates to help prevent dangerous crowding.Source: §5 · pp. 16–17