Biological processes for remediation of soils contaminated by petroleum products combine a great environmental efficiency (products are transformed instead of changing their phase of occurrence as it happens in physical processes) with an extreme complexity due to the biological nature of the structural molecular transformations involved.
This research handles bioremediation in a comprehensive and multi-disciplinary perspective, although favouring the physical component of the processes in detriment of the most studied chemical and biological features.
The experimental work involves the biodegradation of soil samples contaminated by crude or diesel oil, collected in the terrains of Leça da Palmeira refinery, which has already microorganisms adapted to petroleum environments. General characteristics of the population of heterotrophic aerobic bacteria, as well its population density, were determined through liquid medium enrichments.
In a first phase the results of biodegradation were interpreted under a chemical and biological point of view using global kinetics models. A phenomenological model describing the degradation of petroleum products in soils was then developed. The model considers two main components, one related to the abiotic degradation (volatilization) and another describing the biodegradable component.
The physical component of the process was studied through the respirometry of biodegradation. The metabolism of microorganisms consumes oxygen and produces carbon dioxide. A continuous measurement of the concentrations of these gases in the atmospheric vicinity of the soil sample allows an interpretation of biodegradation at a fine scale. Several respirometric experiments were performed where oxygen consumption and, whenever possible, carbon dioxide production were quantified.
An explicit objective was the development of methodologies for processing the biological signals originated by respirometry. This area of research is virtually non-explored, having an enormous attractive power.
The respirometry data was analyzed applying techniques as robust and diverse as Fourier analysis, wavelets and directional statistics and its interpretation was also performed in terms of time series, namely using autocorrelation, partial autocorrelation and cross correlation functions.
Finally, black box type stochastic models were built using the System Identification theory