Unconventional resources, such as shale and tight-gas sandstone reservoirs, have been one of the key components of energy supplies in North America. Although research on tight and ultra-tight rocks has made significant progress, there exist questions about their physical and hydraulic properties that remain unanswered. For instance, the distribution of pore throats and its mathematical form in unconventional reservoir rocks are still unknown. The main objective of this study, therefore, is to investigate the potential probability density function representing the pore-throat size distribution in tight gas sandstones. For this purpose, we analyzed more than 100 Mesaverde tight gas sandstone samples collected from Western US basins under ambient and confining pressure conditions. We converted the measured capillary pressure (Sw-Pc) curves into the pore-throat size distributions by plotting \(\Delta\) Sw/ \(\Delta\) ln(Pc) versus log-transform pore-throat diameter, log(d), and observed non-Gaussian trends for most samples. More specifically, we detected a heavily tailed (left-skewed) distribution, in contrast to mostly observed trends in conventional reservoir rocks. For the first time, we applied the generalized normal probability density function to characterize the log-transformed pore-throat size distributions and demonstrated that its unimodal and/or bimodal forms fit the distributions from tight gas sandstones reasonably well. For the ambient samples, the generalized normal distribution was the corrected Akaike information criterion (AICc) preferred model in 56% of cases, followed by the bimodal generalized normal (28%), bimodal normal (12%), and normal (4.2%) distributions. For the confined samples, the generalized normal distribution was the AICc preferred model in 46% of cases, followed by the bimodal generalized normal (32%), bimodal normal (18%), and normal (3.8%) distributions. For both unimodal and bimodal samples, we found that the median of the pore-throat size distribution was correlated to the logarithm of porosity (with \(R>0.63)\) and to the logarithm of permeability (with \(R>\) 0.74) for which the correlations were stronger. Results also showed that the (log)permeability was exponentially correlated to the porosity with R2 = 0.83 for the samples under the ambient conditions and R2 = 0.82 for the samples under the confined conditions.
