@inproceedings{scholars13706,
           title = {Acoustic data driven application of principal component multivariate regression analysis in the development of unconfined compressive strength prediction models for shale gas reservoirs},
       publisher = {Society of Petroleum Engineers (SPE)},
         journal = {Proceedings - SPE Annual Technical Conference and Exhibition},
            year = {2020},
          volume = {2020-O},
            note = {cited By 1; Conference of SPE Annual Technical Conference and Exhibition 2020, ATCE 2020 ; Conference Date: 26 October 2020 Through 29 October 2020; Conference Code:164382},
             url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85095704679&partnerID=40&md5=dbcab679de7fb62600eb801c1813300e},
        keywords = {Boreholes; Compressive strength; Elastic waves; Energy resources; Forecasting; Gases; Geomechanics; Multivariant analysis; Oil field equipment; Oil wells; Petroleum reservoir evaluation; Petroleum reservoirs; Regression analysis; Shale gas, Coefficient of variation; Empirical correlations; Global energy demand; Mean absolute percentage error; Multivariate regression analysis; Principal Components; Unconfined compressive strength; United States of America, Predictive analytics},
        abstract = {Unconfined compressive strength (UCS) equally represented as geomechanical strength remains a critical mechanical property in the successful implementation of key technologies for shale gas reservoirs' development and production. Attention has been less concentrated on prediction models' development for shale geomechanical strength evaluation. Majority of the existing shale geomechanical strength correlations are dependent on single log input parameter, which is insufficient to account for the complex and nonlinear behaviour of UCS across the entire reservoir interval of interest. The high relevance of UCS has therefore triggered the need for the application of an integrated system of principal component - multivariate regression analysis in driving UCS predictive models' development for shale gas reservoirs. Generated acoustic datasets of notable shale gas reservoirs (Marcellus, Montney, Longmaxi and Roseneath) in respective countries (United States of America (USA), Canada, China and Australia) were used. Statistical test analysis was conducted in validation for wider applications of the developed UCS prediction models. Models development were driven by 21,708 datapoints of acoustic parameters, models' accuracy ratings were above 99, R-squared values had high degrees of closeness to unity, mean absolute percentage error (MAPE) values were at less than 10 and coefficient of variation (COV) at less than (1.0). UCS prediction models were all dependent on multiple direct log measured acoustic parameters in distinction to existing UCS empirical correlations; thus, a pure reflection of significant boost to the accuracy and reliability of UCS measurements for shale gas reservoirs. The developed prediction models will promote geomechanical strength accountability and lead to creation of a robust base in minimization of wellbore instability problems, optimization of wellbore trajectory and containment of hydraulic fractures. This will significantly contribute in putting gas resources of shale reservoirs with enormous potentials, at the forefront of quantitatively meeting natural gas requirements in global energy demand. Copyright {\^A}{\copyright} 2020, Society of Petroleum Engineers},
          author = {Iferobia, C. C. and Ahmad, M. and Salim, A. M. A. and Sambo, C. and Michaels, I. H.},
            isbn = {9781613997239}
}