The present work investigates turbulent, buoyant fire-induced flows using an experimental scaling technique known as saltwater modeling — a methodology enabling quantitative analysis of fire plumes built upon the analogy with saltwater (plume) flowing into the ambient water (air). The investigation, conducted by means of velocimetry (PIV) and concentration (PLIF) laser-based techniques, concerns unconfined plume mixing and transport, characterization of ceiling jet flows under sloped ceilings and activation of suppression devices in these sloped configurations.

Flow imaging provides detailed measurements of velocity and saltwater concentration within the entire spatial and temporal domain of a planar section of the plume. In analogy with low-pass filtering in large eddy simulation (LES), a virtual, pixel-binning grid of varying size is overlaid on images to compute statistical moments representative of the larger and smaller scales. By leveraging actual measurements, converged statistics (first, second, higher-order) enables selection of cutoff resolutions, useful for validation and development of computational fluid dynamics (CFD) simulations.

The saltwater plume's subsequent impingement onto a sloped plate generates a ceiling jet flowing both streamwise (up- and downslope) and spanwise with respect to the impingement point. Such flow is investigated to first build correlations predicting velocity and temperature along a sloped ceiling and second to analyze slope-related suppression device (sprinkler) activation. For the first task, single-planar, streamwise measurements are employed; for the second, multiple orthogonal laser sheets crossing the plate are used to generate a virtual grid of measured points. Transport characteristics are implemented into an activation model, modified to predict a dimensionless response time spatial distribution. At increasing slopes, the delay in the activation between upslope (faster) and downslope (delayed) devices progressively increases at increasing ceiling angles. This also occurs between sprinklers symmetrically located upslope and spanwise. From the response spatial distribution, the streamwise-to-spanwise correlation for the delay time (thermal responsiveness) is determined using the saltwater front arrival times. The analysis for the lag time reveals that the delay in thermal responsiveness between two sprinklers with the same activation time located up- and downslope, respectively, increases exponentially compared to that found for sprinklers located spanwise, at a quadratic rate with increasing angles.