Study parameters
Checking inputs
Width alone does not rank accuracy. A and B use a common density target; C uses a separate perception target. B includes a whole-lane implementation rule.
Minimum-width constraints and whole-lane increments can produce equal final widths for different LOS profiles. LOS F is diagnostic, not a recommended design.
No averaging of model outputs. No automatic claim of global optimality, safety improvement or legal compliance. The search is one-dimensional and limited to the declared width grid and ceiling.
Effective width = clear walk zone minus declared edge loss. Total cross-section = clear walk zone + frontage + furniture/utility zone + curb. Physical obstructions inside a walk zone need a segment-level survey; this calculator does not simulate them.
v(k) = vf[1 − exp{−γ(1/k − 1/kj)}]; We = Q / [k × v(k)]. The width expression is a continuity-based derivation, not a new published equation.
Weidmann (1993), Transporttechnik der Fussgänger, IVT report 90, ETH Zürich. Equation and reference values reproduced in Hänseler et al. (2013), An aggregate model for transient and multi-directional pedestrian flows in public walking areas, EPFL report 131219, Equation 1, printed p. 7.
EPFL research reportThe published Equation 10 supplies queue-based speed. This implementation combines it with Q = kWv (Equation 2), enumerates integer lane counts, and sets W = b × s. It is not a direct use of Equation 15, and its service rate R is not substituted for observed demand.
Rahman et al. (2013), Modelling Pedestrian Travel Time and the Design of Facilities: A Queuing Approach. PLOS ONE, 8(5), e63503. DOI: 10.1371/journal.pone.0063503. Its steady-state, uninterrupted-flow and virtual-lane assumptions need review for each application.
Equivalent Erlang-B/Erlang-C recursion avoids factorial overflow. Unused fractional lane width is not credited. The lane parameter is a movement allowance, not an anthropometric body diameter.
I = −1.7592 − 0.2299P + 1.3240N0.1 − 0.0196We2 + 0.5447C0.1. Solve for We using the selected I target. Equation 6.1 is used because the input is a fixed sidewalk count N per five minutes, not Equation 7.5's street-wide allocation.
Carvalho (2006), Um modelo para dimensionamento de calçadas considerando o nível de satisfação do pedestre. Doctoral thesis, University of São Paulo. Equation 6.1, printed pp. 108–109; service mapping, Table 7.9. P = 0 peak / 1 off-peak. I = 0.49 means 49% in the original coding.
Uncalibrated transfer to Philippine sidewalks is not established. Its illustrative 1–4 m nomogram is not an independently validated applicability range. Satisfaction LOS is not HCM operational LOS.
University thesis| LOS | Space target (m²/ped) | Separate C target | Role |
|---|---|---|---|
| A | 6.00 | 32% dissatisfaction | Planning profile |
| B | 4.00 | 40% | Planning profile |
| C | 2.50 | 49% | Planning profile |
| D | 1.60 | 58% | Planning profile |
| E | 1.00 | 69% | Planning profile |
| F | 0.65 | 75% | Diagnostic only |
Space values are project-selected representative points within historical HCM 2000 space bands, not the full LOS method. Context: FHWA-HRT-05-137 (2006), Chapter 2, Table 4. The source describes severely restricted movement at F; it is not accepted here as a design recommendation.
C's separate A–E dissatisfaction targets follow the limits reported in Carvalho's Table 7.9; 75% for F is an explicit illustrative project choice. Coupling both targets to one selector is a workflow choice, not evidence that operational and perception LOS are equivalent. Changes in model widths can be masked in the final width by minimum clear-width constraints or lane increments.
These profile assignments are not calibration. Local calibration needs observed performance data and parameter fitting; independent validation remains necessary.