RESUMEN
The development of numerical models for investigating the risks and impact caused by human activities to the marine environment is important. Herein, the recently developed ChemicalDrift Lagrangian dispersion model was coupled to a toxicokinetic model and applied to investigate emissions of polycyclic aromatic hydrocarbons (PAHs) discharged from oil and gas production facilities as produced water. The performance of the model was evaluated with available data from a monitoring survey conducted at two oil fields. The survey provided exposure concentrations by means of passive samplers and bioaccumulation data in caged mussels; multiple depths and locations were assessed. The study included 26 PAHs and alkylated derivatives, showing good agreement between the model and the survey measurements. The compounds dominating the scenario were naphthalenes and phenanthrenes. Model provided contamination gradients were in agreement with the survey results, with levels decreasing with distance away from the main sources and with higher concentrations at 20â¯m depth. ChemicalDrift and the toxicokinetic model provided detailed time series, showing peaks of C1-naphthalene bioaccumulation significantly higher than values accumulated at the end of the monitored period. The utilised model was able to separate the relative contributions of multiple platforms and to identify the major contamination sources, providing a valuable and versatile tool for assessing the impact of discharges at sea.
RESUMEN
When drilling an oil or gas well, well pressures may be controlled using a technology called managed pressure drilling. This technology often relies on model predictive control schemes; however, practical limitations have generally led to the use of simplified controller models that do not optimally handle certain perturbations in the physical system. The present work reports on the first implementation of a highly accurate system model that has been adapted for real-time use in a controller. This real-time high-fidelity model approximates the results of offline high-fidelity models without requiring operation by model experts. The effectiveness of the model is demonstrated through simulation studies of controller behavior under various drilling conditions, including an evaluation of the impact of sparse downhole feedback measurements.