Wednesday, 25 January 2017
4E (Washington State Convention Center )
Located near population centers, hydrocarbon-fueled distributed generation (DG) units (e.g., diesel backup generator and biomass heating systems) are typically have shorter stacks than central plants, causing emissions subject to entrainment inside wake zones next to buildings, a phenomenon often referred to as “building downwash”. The capability to estimate the emission rates from DG units based on near-surface environmental sensing can greatly facilitate the mitigation efforts to reduce the local air quality impact. In this paper, we present our study on assessing different source estimation methods in terms of quantifying DG emission rates in a building downwash environment created by a recent extensive wind tunnel experiment conducted at the USEPA Meteorological Wind Tunnel. Compared to field experiments, the controlled environment in the wind tunnel reduces the uncertainties caused by atmospheric conditions not fully captured by field experiments. We evaluated three methods including Tangent-linear, Adjoint, and Bayesian inference. For the first two methods, a cost function which measures the distance between the forward model with respect to the observations and with respect to the background state is established and to be minimized. The tangent-linear model is used to calculate the incremental updates, and the adjoint model is used to calculate the gradient of the cost function. Bayesian inference method takes the randomness of the measurements into consideration. The likelihood function is constructed based on the error/difference between the forward model and the measurements, or based on the time series data generated from CFD simulation at each receptor location. Cases with different building aspect ratios, source stack heights, source locations with respect to the building, and wind angles were examined. We presented the comparison results by describing the applicability and limitations of each method. Finally, we discussed the needs for improving source estimation techniques in order to be better integrated with environmental sensing, and further evaluations using field measurement data.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner