We investigate computationally the effect of randomness as well as the effect of fiber aggregation on the hydraulic permeability (Keff) of unidirectional fiber arrays. For this we carry out extensive viscous flow computations in various random fiber arrays which are generated through an NVT Monte-Carlo process, starting from uniform square arrays or from regular arrays of fiber clusters. Up to 10,000 individual fibers are included in each simulation. The extent of aggregation in these arrays is quantified using an integral measure of Ripley’s K-function and is shown to correlate with the dimensionless temperature used in the Monte-Carlo algorithm. This metric expresses the deviation of the K-function of any given fiber arrangement from that of a Poisson process at the length scale range of interest. The results demonstrate that the extent of aggregation correlates with the computed permeability as well as with the scatter in the computed values. The latter observation points to a criterion for the selection of properly sized computational unit cells.