Characterisation and Modeling of Dust in a Semi-Arid Construction Environment

This paper describes an investigation into fugitive dust patterns and characteristics at a 200 hectare oil and gas terminal in the Middle East. Dust monitoring took place as part of an environmental and socio-economic impact assessment (ESIA) for proposed construction work, with initial monitoring over three months as part of a baseline study. Once the new phase of construction had started, passive dust monitoring was undertaken for 18 months using an extensive array of sticky pad dust collectors, with samples of dust in directional flux and deposited dust obtained. Dust monitors were placed upwind and downwind of the construction site, near building activities, and adjacent to nearby receptors. Collected dust samples were subjected to basic analysis for dust coverage (AAC %) and dust soiling (EAC %) with further samples later chosen for additional elemental, mineralogical and particle size analysis. Dust was removed from the sticky pads so that elemental analysis could be undertaken using ICP-MS and mineralogical analysis could be completed using SEM-EDX. Particle size analysis was accomplished using laser granulometry. Dust composition was compared with reference to background dust and soils. Results showed that collected dusts, both from the baseline study and construction phase and from background samples and receptors, were comprised of similar major minerals (r² values > 0.78) to the calcareous sandy silts and clays that make up the majority of local soils. Similar results (r² > 0.79) were also found for proportions of major elements in background dusts, receptor-collected dusts, and soils. One notable exception to the findings of dust homogeneity was the increased presence of calcite (and thus Ca) in samples near to roads: this is thought to originate from road traffic spillages of limestone which is quarried nearby. Particle size distributions for dust samples were also similar across samples at individual monitoring periods, but seemed to vary temporally in response to differing weather conditions. Weather data from an onsite monitoring station were also collected throughout the study and temporal dust levels were investigated in relation to individual weather factors. High dust levels originating from north of the site were recorded throughout the study period, and were observed to correlate with respective high winds from the north. Dust levels were also significantly increased during summer months and after periods of low rainfall. A predictive model for background dust propagation was therefore developed midway through the study, using wind strength and frequency from the north, temperature, and antecedent rainfall. Background dusts were defined as those arriving from north of the site (330° - 30°) and collected at dust monitors unaffected by construction workings. Linear regression was used to incorporate the different weather parameters into individual weather factors which combined to create a prediction of background dust levels. The resulting model, based on the first year of data, was subsequently tested against recorded levels over the next six months with high levels of accuracy observed (r² > 0.78). The model was also tested against data collected during the baseline study with further positive results (r² > 0.67).