Show simple item record

dc.contributor.advisorWhitson, Curtis Hays
dc.contributor.authorYounus, Hafiz Muhammad Bilal
dc.date.accessioned2019-09-11T09:01:11Z
dc.date.created2017-06-09
dc.date.issued2017
dc.identifierntnudaim:17218
dc.identifier.urihttp://hdl.handle.net/11250/2615088
dc.description.abstractThis thesis work is in the continuation of TPG4560 specialization project work done on fluid characterization of Eagle Ford shale reservoir. The objective of the work is to develop a single equation of state (EOS) representing the entire Eagle Ford field. The EOS should be capable of predicting the reservoir and surface fluid properties for the subject field with reasonable accuracy. The industry software used in developing the EOS is PhazeComp written by Aaron Zick (Zick Technologies). A detailed insight of the available data and data quality issues are discussed in TPG4560 project report. Detailed step wise approach to develop an initial EOS (without any regression on experimental data) is also discussed in the project report. The thesis work is mainly focused on developing a good initial EOS and use it to initiate the tuning on the experimental data. There are 65 samples available for the EOS development. PVT experiments on 59 samples are conducted by FESCO. Only one sample is from Schlumberger and five are from GeoMark lab. The available samples cover the entire range of fluid types; dry gas, wet gas, gas condensate, near critical fluids, volatile oil and black oil. The molecular weight (M) specific gravity (γ) relationship is developed by fitting the Soreide correlation to the lab reported M-γ data. The best selected initial EOS model is different from what is proposed in the TPG4560 project report in two ways; (1) It contains more detailed component description such that BTX components and isomers for C6, C7, C8 and C9 are included individually instead of being lumped as single carbon number in earlier proposed initial EOS (2) Full matrix of BIPs is used such that BIPs between all pairs of components are non-zero. This initial EOS predicts the right saturation type for all samples before any modification to the EOS or sample compositions. The properties of non-library components in this EOS characterization are estimated in the same way as done in TPG450 project work. For compositions, SLB and GeoMark report moles and mass but FESCO lab only reports molar amounts. Lab reported mass amounts are conserved for Schlumberger and GeoMark samples. For FESCO samples, reported moles are converted to lab measured mass amounts using lab molecular weights, conserving masses. All mass amounts are converted to moles using consistent molecular weights of the EOS characterization. For separator gas, lab reported mass amounts are honored for all components whereas for separator oil, mass amounts of C10 and heavier fractions are replaced by gamma molar distribution results. Reservoir fluid compositions are calculated by mathematical recombination in PhazeComp using separator oil and separator gas compositions and lab reported gas molar fraction (Fg). The results of initial EOS predictions are studied in detail. Saturation pressure predictions are widely scattered which is improved by regressing on BIPs and slight modification of C10+ gamma average molecular weight to modify slightly the distribution of C10 and heavier amounts. Liquid relative volumes of almost all FESCO samples (gas condensate and oil) are predicted higher than lab reported values. For GeoMark, the predictions are on the lower side. This suggests that the EOS is predicting most of the gas condensate samples richer and oil samples leaner, compared to lab data. If it is believed that liquid relative volume data is accurately measured then a consistent change in EOS variables is needed to shift the critical point of all samples to the left (of their current critical point location) to make gas condensate samples leaner and oil samples more critical. Curiously, for most of the very lean gas condensate samples, the reported CVD gas compositions are predicted with great accuracy by the EOS, especially the C7+ content. This is unexpected due to practical difficulties in the accurate measurement of very small amounts of flashed liquid volumes dropping out of CVD removed gases for lean samples. Another important observation is that even though reservoir fluid compositions are calculated from reported separator oil and separator gas compositions, when this reservoir fluid is flashed to reported separator conditions, the calculated recombination GOR is often not the same as the reported value. After a detailed investigation, it was found that many of the reported separator oil and separator gas compositions are not in equilibrium. Even though the reported data look straight line on the Hoffman KPF plot, the sample compositions may not be in equilibrium. The K-values of the two sample compositions (oil and gas) should be close to Standing low pressure K-values if they are in complete equilibrium at reported separator pressure and temperature. The K-values from the EOS at low pressures are always in good agreement with Standing K-value correlation even though Standing never used EOS prediction data for the development of his correlation. Different meetings were conducted with FESCO to understand about the laboratory procedures for conducting the PVT experiments. Results of EOS calculations regarding the magnitude of liquid volumes being produced by flashing the CVD gas from very lean samples was shared with the lab. The conclusion of all the meetings is that the reported CVD compositional data is highly uncertain. Liquid relative volume data is also uncertain but it can be assumed that the uncertainty in the reported data is less for oils compared to leaner gas condensates. Two types of regression variables are used for EOS tuning; (1) composition variables that affect each sample individually (2) EOS variables that affect all samples in a similar way. Compositional adjustment used recombination GOR and C10+ gamma average to modify the reported compositions slightly because of the inherent uncertainties in the reported values. For the selection of the most effective EOS variables for EOS tuning, a detailed sensitivity study was made to study the effect of each EOS variable on saturation pressure (PSAT) and liquid relative volume (LRVOL) predictions of the selected samples, to cover the entire range of fluid types. The study suggests that not a single EOS variable can improve both PSAT and LRVOL data simultaneously and there is strong negative correlation between the two types of data. Another study was made to see which EOS variables are most effective in affecting the critical transition, such that shifting the critical point would improve liquid relative volume data. The most near critical oil and gas condensate samples from FESCO lab were chosen for this study. It was found that any effort to improve the liquid relative volume data makes the saturation pressure worse. Also, LRVOL of the richest gas condensates can be improved very slightly even if the near critical oil sample is forced to transition into near critical gas condensate (i.e. a large shift in the critical point). The initial EOS predicts the saturation types of even the most near critical samples correctly. This suggests that the critical point predictions with the initial EOS are reasonable and minimum changes in EOS variables would be required. Final variables for EOS regression were BIPs, recombination GOR and gamma average. For EOS tuning, the weight factors are selected such that saturation pressure is always given higher priority over liquid relative volume data. Since, gamma average affects both saturation pressure and surface liquid densities, a Pipe-It project was made to determine the optimal weight factors on liquid API relative to saturation pressure so that liquid API predictions remain within ±2API unit range. Manual weight factors for liquid relative volume data are chosen in a systematic way to give more priority to liquid relative volume data of rich samples over the lean samples. The EOS is regressed first on experimental data, except liquid viscosities, with BIPs, C10+ gamma average molecular weight and recombination gas molar fraction (Fg). The final regressed EOS meets all criteria of good quality check. The final EOS honoring all vapor liquid equilibrium type data with reasonable accuracy is then used for independent liquid viscosities regression using ZcVis as variables. The final EOS predicts almost all saturation pressures within 4% deviation. Liquid API of most samples are within ±2 API. Liquid relative volume of most samples are reasonable. The final compositions of reservoir fluid suggest a maximum change of about 1 mol-% in methane content compared to lab reported compositions. The lab separator test GOR of all oil samples are reasonably predicted. The EOS also predicts the surface vapor liquid equilibrium data with good accuracy. The quality of liquid viscosity match with the final EOS predictions is excellent. Development of a single EOS representing the entire range of fluid types was a great challenge. The type of data available for such a detailed EOS model development is fairly complete. However, the main challenge is to have most of the data reported from a single lab that is FESCO. It is believed that there is some kind of systematic error in the liquid relative volume measurements from this lab which is a key vapor-liquid equilibrium data used in EOS development. The EOS should be used to predict the PVT experiments from other labs and can be improved using new available data if found necessaryen
dc.languageeng
dc.publisherNTNU
dc.subjectPetroleum Engineering, Reservoir Engineering and Petrophysicsen
dc.titleEOS Characterization For (the) Eagle Ford Shale Reservoiren
dc.typeMaster thesisen
dc.source.pagenumber465
dc.contributor.departmentNorges teknisk-naturvitenskapelige universitet, Fakultet for ingeniørvitenskap,Institutt for geovitenskap og petroleumnb_NO
dc.date.embargoenddate10000-01-01


Files in this item

Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record