Prediction of Toxicity Effects of Oil Field Chemicals Using Adaptive Genetic Neuro-Fuzzy Inference Systems

Abstract Chemicals are used in various stages of oil production such Drilling (Drilling fluids, cementing, completion and workover fluids), Production, Stimulation and Enhanced Oil Recovery. Many research studies have shown these oil field chemicals have toxic effects on the environment. The oilfield chemicals include various additives for drilling/cementing and work-over such as Fluid loss additives, rheology modifiers, Viscosifiers, Emulsifiers, Biocides, Surfactants, Packer fluid corrosion inhibitors. Toxicity tests are crucial for the assessment of the harmful effects of complex chemical mixtures, such as waste drilling mud, hydraulic fracturing fluid on aquatic environment. The objective of the study is to develop screening protocol to assess, evaluate, and manage the inherent risks. To achieve this, it is imperative to develop models, tools and an acceptable mechanism for screening, predicting and monitoring the application of oil field chemicals. In this paper, Adaptive Genetic Neuro-Fuzzy Inference System is developed to assess the toxicity of oilfield chemicals. Several toxicological studies have shown the evidence of toxicity of some oilfield chemicals to living organisms and their potentially negative side effects on environmental ecosystems for which relatively tedious animal testing methodologies are documented for their assessment. The description of this intelligent system is provided and has proven to have better classification and regression capability and ability to handle high dimensional features. This study adopts a novel evolutionary computing approach to search and obtain the optimal Neuro-Fuzzy parameters to enhance the prediction accuracy and generalization capability of the model. The system was applied to a dataset on Oil Field Chemicals toxicity and it was found that the genetic algorithm yields optimal parameters of Neuro-Fuzzy for the given datasets. The prediction and classification of Oil Field Chemicals (toxic or non-toxic) using this hybrid intelligent system is a work that requires an in-depth study and understanding of the various underlying principles of Neuro-Fuzzy inference system and Genetic Algorithm, which is commonly applied for classification and regression purposes. The developed model based on the fuzzy rules was trained with available data set. The unseen or new data is therefore either classified into appropriate class or have toxicity predicted using gaussian membership function chosen for this application. The motivation of this approach is that it is less cumbersome than the conventional computational modeling usually adopted for chemical classification and characterization. It also seeks to eradicate the existing animal testing that are hitherto very tedious and cumbersome.