Evaluation of Mercury Distribution Models in Process Simulators
Abstract
Mercury is a trace component that can be found in all fossil fuels. Mercury s existence in hydrocarbons raises concerns about equipment degradation, about the health and safety of the field personnel and about the environment. Equipment degradation, caused by mercury, is of great importance, especially for the gas industry, because it can lead rapidly to catastrophic failures of the aluminum heat exchangers. Both the fact that mercury exists at very low concentrations in fossil fuels and that it can cause major accidents in the natural gas industry, leads to the need of very accurate models. Therefore, an evaluation of the available models for mercury is necessary. The scope of this thesis is to evaluate different models by comparing them to solubility data and by implementing them in different process simulators, to simulate two natural gas processing plants, in order to compare the predictions given by the simulations to field data. To this purpose, a literature review was conducted to find all the available models for mercury. All the models found, use the Peng-Robinson(PR) and Soave-Redlich-Kwong(SRK) equations of state as their basis with varying expressions for the alpha function and different binary interaction parameters. In addition to these models, a model developed by Statoil, referred to as SRK-Twu(Hg), was available for evaluation. The models that were selected for evaluation included the PR and SRK models with no binary interaction parameters for mercury. Also, the PR and SRK models found in the PRO/II software package were evaluated. These two models use binary interaction parameters and therefore they improve the prediction ability of the model for the liquid phase. Additionally, the SRK-Twu and PRMC models were tested. These models use advanced expressions for the alpha function, which results in a major improvement of the predictive ability for the vapor phase. Finally, the SRK-Twu(Hg) by Statoil was evaluated. This, model uses both an advanced expression for the alpha function and binary interaction parameters. These models were selected in such a way, in order to evaluate how the models behave when vapor phase correction is used(SRK-Twu and PRMC), when liquid phase correction is used(default PRO/II SRK and PR), when no correction is used (SRK and PR) and finally when both corrections are used(SRK-Twu(Hg)). It was found that if one uses for their model only vapor phase correction or only liquid phase correction they could possibly get worse results for mercury s distribution. Therefore, it is recommended that both vapor phase and liquid phase corrections should be applied at the same time.The results of the evaluation indicate that the model developed by Statoil is the most accurate of the models tested and is considered reliable. The second best model was found to be the PR PRO/II default model, but it is recommended to be used with caution. All the other models, were found unable to predict mercury s distribution, with the worst models being the SRK and the SRK PRO/II default models.