Petroleum reservoir performance evaluation based on compositional grading models

  • Ikechi Igwe

    Student thesis: Doctoral Thesis

    Abstract

    without compositional grading (CG) models on realistic reserves estimates and reservoir performances prediction have been considered. The mathematical framework for the compositional grading modelling is based on one-dimensional zero-mass-flow stationary state assumption. Computer Modelling Group’s WinProp, was used for the fluid modelling while Computer Modelling Group’s compositional reservoir simulator, GEM, was used for the reservoir modelling and simulation. In the absence of historical production data, Computer Modelling Group’s CMOST was used to perform uncertainty assessment for the validation of models results and for sensitivity analysis. The effect of various equations of state on the performances of initialised reservoir models and the effect of changes in temperature gradient on the performances of applied nonisothermal CG initialised reservoir models are also reported.
    The research results suggests that inadequate account or complete neglect of compositional grading effect in reservoir simulation model initialisation has significant technical consequences. The results shows that constant composition (without CG) initialised reservoir model overestimated the original oil in-place by 13.86 % more than the isothermal model, 24.37 % more than the zero thermal CG model, 24.44 % more than the Haase;s thermal diffusion CG model, and 24.41 % more than the Kempers model. However, it underestimated original gas in-place by 12.73 % less than the isothermal CG model, 21.24 % less than the zero thermal diffusion CG model, 21.35 % less than the Haase’s thermal diffusion CG model, and 21.31 % less than the Kempers thermal diffusion CG initialised reservoir model. The results of the sensitivity analysis shows that for all the initialised reservoir models, water saturation (which indirectly accounts for hydrocarbon saturation) with 82-88 % main effects is the most sensitive input parameter responsible for the estimated reserve volumes. The observed differences in in-place volumes estimated by the various CG models are due to the influence of the CG models on the reservoir fluid formation volume factor, which made the hydrocarbon saturation (compositions of various components) to either increase or reduce in the gas and oil phases, respectively. Analysis of the effect of various equations of state (EOSs) on the performances of initialised reservoir models suggest that, although, the choice of applied EOSs is not very critical to the efficient performances of reservoir models initialised without CG models, each EOSs had noteworthy different impact on the performances of reservoir models initialised with CG models. Therefore, to effectively predict the performances of compositionally sensitive reservoir, it is important that a sensitivity analysis be carried out to determine the EOS that will guarantee optimal performance. Increasing the temperature gradient in nonisothermal CG initialised reservoir models from 0.002 oF/ft to 0.5 oF/ft caused a 30.69 % decrease in the OOIP estimated by zero (passive) thermal diffusion CG initialised reservoir model; 34.14 % increase in OOIP estimated by Haase’s thermal diffusion CG initialised reservoir model; and 38.34 % decrease in the OOIP estimated by the Kempers thermal diffusion CG initialised reservoir model. A further increase in the temperature gradient from 0.5 oF/ft to 2.5 oF/ft caused a 22.48 % increase in the OOIP estimated by the Kempers thermal diffusion CG initialised reservoir model; 61.43 % increase in OOIP estimated by Haase’s thermal diffusion CG initialised reservoir model; and 44.73 % decrease in the OOIP estimated by zero thermal diffusion CG initialised reservoir model. Therefore, temperature gradient and its associated thermal diffusion factor can significantly influence the performances of nonisothermal CG initialised reservoir models. Hence, need to adequately account for thermal diffusion factor during reservoir simulation model initialisation.
    The results of this work provides a new insight into the impact of neglecting compositional grading in field development studies and should encourage field managers, and in particular those of Niger Delta, to put more weight into the investigation of compositional grading effects and to support research works in this area.
    Date of AwardSept 2020
    Original languageEnglish
    Awarding Institution
    • University of Portsmouth
    SupervisorJebraeel Gholinezhad (Supervisor) & Mohamed Galal Hassan Sayed (Supervisor)

    Cite this

    '