Page 4 - Predictive.indd
P. 4

Predictive Analytics                              Simple Economics                                  Relative Probabilities                             Well Planning



           Time / Asset Development                                                   *                          Least Likely Case      Most Likely Case
                                                         Net Cashflow +                                                 Below TC


                                                                         conducive reservoir
       3D         numerous  production  predictive                                                          Frequency
       seismic    lateral   monitoring  production                                                                                            Above TC
       & well log   completions  (IP90/  analytics
       data                 IP180)                                     shorter payout for   average well payout                On TC



                                                         0
                                                                                             Time
                                                                           non-conducive reservoir
                             are some
                what criteria  wells    what areas                        longer payout for
                determines  producing   will produce
                 where to   better than  best?
                   drill?                                                                                        0%     Relative Probability   100%
                             others?
                                                       Most lateral wells drilled into an established    For any planned location, the type curve           Once a predictive analytic production
                             action                    shale reservoir will “pay out” barring some       classification (above, on or below type            metric model is constructed, well paths
     Modified from predictiveanalyticstoday.com        gross mechanical error. There are few             curve) statistic can be extracted and used         through any bench or landing zone can
                                                       uneconomical wells. However, some wells           to determine well feasibility and/or priority.     be analyzed and production metric
     With unconventional shale reservoirs,             will pay out faster than other wells due          In addition to the classification, a relative      classification or probability of class can
     a predictive model can be built with              to a more favorable shale mineralogy.*            probability statistic is also extracted for        be measured. Type curve probability
     production monitoring from numerous               Depending on the production metric of             each type curve class. The statistics above        measurements can be reviewed and well
     lateral wells and extensive 3D seismic            choice, there can be a significant difference     show that this location has a very strong          paths optimized to maximize IP90 or IP180
     coverage. The predictive production               in production and financial returns between       probability of being an above type curve           yields.
     metric model identifies the best producing        a well yielding an “above” type curve             producer, and is less likely to be a below
     areas within an asset, and an integrated,         performance and one yielding a “below”            type curve producer.                               A relative probability of on type curve
     intelligent, in-fill drilling plan developed      type curve performance. Drilling only the                                                            performance is shown above.  This
     to provide the best financial returns for         above type curve performance wells can            Measurements such as this are performed            validation well shows the toe of the well
     the asset.                                        substantially change the financial returns        on an out-of-sample or blind population to         path having a high probability (bright
                                                       from an unconventional asset.                     test the robustness of the predictive model,       colors) of on type curve performance.  The
                                                                                                         and to predict potential returns based on          heel of the well has a high probability of
                                                                                                         previous production and returns.                   above type curve performance (not shown).
   1   2   3   4   5   6   7   8