Statistical evaluation of Lower Flammability Distance (LFD) using four hazardous release models

Abstract

Using four hazardous dispersion models, the ability to predict Lower Flammability Distance (LFD), the distance from a release through which the released gas is present in concentration below flammable range, has been evaluated using a set of large‐scale propane release field experiment data. The four models evaluated were the DEGADIS model, the SLAB model, the OME Simple Gas model and the OME Heavy Gas model (1983). The observed LFDs were compared with the predicted LFDs for three atmospheric conditions. The complete data set was broken into two parts: (1) the cyclone type releases (representing instantaneous release situations) and (2) the nozzle type releases (representing horizontal jet releases). A procedure based on the USEPA guidelines on air quality models was followed to evaluate the models. The complex models (DEGADIS and SLAB models) appear unreasonably weak in predicting LFD's compared with the relatively simple Gaussian model used in the OME Simple Gas model for instantaneous releases. The OME simple gas model is significantly better than the DEGADIS and SLAB models for cyclone type (instantaneous) releases under unstable and neutral atmospheric conditions; and the DEGASIS model is significantly better than the SLAB model for nozzle type (horizontal jet) releases under stable atmospheric conditions. Since the size of the data set is not large enough (n = 20 for cyclone type release and n = 18 for nozzle type release), it is difficult to show 95% significant differences between models.

Performance Evaluation of AERMOD and ADMS-Urban for Total Suspended Particulate Matter Concentrations in Megacity Delhi

“…To determine the reliability of the model, the criteria used is as set in a study by Kumar et al (1993) and Chang et al (2004). According to Kumar et al (1993), the performance of the model can be deemed as acceptable if; NMSE < 0.5 and -0.5 < FB < +0.5…”

“…According to Kumar et al (1993), the performance of the model can be deemed as acceptable if; NMSE < 0.5 and -0.5 < FB < +0.5…”

Development of an ANN Interpolation Scheme for Estimating Missing Radon Concentrations in Ohio

“…Kumar and Gudivaka [40] have discussed in detail the statistics relevant to model evaluation and has applied it to heavy gas models. Similarly, Kumar et al [41] has used statistical tools to evaluate the prediction of lower flammability distances. Patel and Kumar [42], Kumar et al [43], and Kumar et al [44] have indicated that Mean Absolute Error (MAE), Factor of Two (Fa2), Root Mean Square Error (RMSE), Fractional Bias (FB), and Normalized Mean Square Error (NMSE) are important parameters to assess the performance of air-quality models.…”

Development and Evaluation of a Dispersion Model to Predict Downwind Concentrations of Particulate Emissions from Land Application of Class B Biosolids in Unstable Conditions

“…As expected in the real life, the shear layer model is not a perfect model. However, the acceptable range for NMSE and FB for an air quality model suggested by Kumar et al (1993) is given as, NMSE ≤ 0.5 and -0.5 ≤ FB ≤ 0.5. The values of NMSE and FB for shear layer model in all wind conditions were within acceptable limits.…”