# Dispersion of LPG from spherical storage tanks: Power-law scaling and comparative analysis of LFL vs. 50% LFL

## TL;DR

This study simulates how LPG leaks from spherical tanks disperse under different conditions, showing that leak size is the main factor affecting dispersion distance.

## Contribution

The paper introduces a comparative analysis of LFL and 50% LFL dispersion distances using power-law scaling and PHAST simulations for various LPG mixtures and leak sizes.

## Key findings

- Leak diameter is the primary factor influencing dispersion extent, with strong correlations at both LFL and 50% LFL thresholds.
- Butane-rich mixtures result in longer dispersion distances at LFL, while meteorological factors have limited influence under typical site conditions.

## Abstract

Liquefied Petroleum Gas (LPG) is typically stored in pressurized spherical tanks, where accidental leaks can create dense, flammable vapor clouds. This study used PHAST to simulate LPG dispersion from a refinery-scale spherical tank, considering various leak diameters (5–805 mm), leak locations, seasonal meteorological conditions, and three propane–butane mixtures (15/85, 30/70, and 50/50 by volume). Dispersion distances were evaluated at both the Lower Flammability Limit (LFL) and 50% LFL thresholds. The findings indicate that leak diameter is the primary factor influencing dispersion extent, showing strong correlations for both LFL (ρ = 0.89) and 50% LFL (ρ = 0.91). Predicted dispersion distances downwind ranged from approximately 20–60 m for small leaks to around 400–800 m for larger releases, depending on the concentration threshold and release conditions. Distances at the 50% LFL were consistently greater than those at the LFL. Power-law regression revealed nearly linear scaling between dispersion distance and leak diameter (b = 0.94 for LFL and b = 0.96 for 50% LFL), explaining over 80% of the observed variance. Butane-rich mixtures resulted in longer dispersion distances at the LFL, while compositional effects were not significant at the 50% LFL. Meteorological and temporal factors had limited influence under typical site conditions. Overall, the results emphasize comparative scaling behavior rather than pointwise concentration prediction and demonstrate deviations from ideal D² scaling due to turbulence, buoyancy, and atmospheric entrainment. Using both LFL and 50% LFL thresholds provides a conservative and practically relevant basis for hazard zoning, quantitative risk assessment, and emergency planning at LPG storage facilities.