Dynamics of Vapor Intrusion under Varying Source Strength

Vapor intrusion (VI) is a dynamic process characterized by strong temporal variability, as field observations show significant fluctuations in indoor air data over periods ranging from days to years. This study develops a semianalytical model to investigate transient vapor migration under varying source strengths (large-volume, diminishing, and fluctuating) and source types (groundwater and soil-gas). For large-volume sources, vertical vapor concentration profiles in soils with high moisture stabilize over 50 years, demonstrating that steady-state screening-level models may mischaracterize long-term exposure risks. The soil diffusion time scale approximates the overall time scale of the VI pathway, identifying subsurface vapor migration as the limiting process. The capillary fringe above groundwater sources slows vapor migration. For finite and diminishing sources, indoor air concentrations rise from zero to a peak value before declining back to zero. Altering the system's influx and outflux can effectively mitigate VI risks: increasing the ratio of the building volume to the source volume shortens the peak arrival time, and increasing the air exchange rate reduces the peak indoor air concentration. For periodic sources, a highly fluctuating attenuation factor indicates that the soil diffusion time scale is comparable to or much smaller than the source period, necessitating time-dependent VI analysis for long-term exposure assessments. Conversely, steady indoor air concentrations at late times suggest that the VI time scale far exceeds the source period, and periodic sources behave similarly to large-volume sources. A linear system approach is proposed to simulate diminishing or sporadically varying source strengths and is applied to a recent VI field test under a fluctuating trichloroethylene groundwater source. The approach effectively generates continuous results from sparse field measurements, with the simulated VI pathway exhibiting patterns consistent with the transient semianalytical model.