Nonlinear Regression Algorithm for Uncertainty Analysis in Production DataMeasurements,Estancia Cholita,Santa Cruz,Argentina

摘要

Measured production data always has a degree of uncertainty which is not considered generally.Uncertaintyin rates is defined as the size of its margin of doubt.It's crucial to characterize this uncertainty for variousreasons; Reporting accurate rates while monitoring projects leads to better management decisions.It willgive us a practical tool to evaluate the quality of a given measurement technology applied.Finally,if anyoperating condition is changed in a mature high water cut field like Estancia Cholita,it's crucial to determineoil percentage to evaluate the new operations success.Oil rates are usually quantified by measuring totalliquid rates with test separator and then taking a sample to determine the oil percentage.There are so manyfactors causing uncertainty in these procedures(duration of the oil sampling,flow regimes,skills of theoperator performing the measurement,etc.)that is difficult to fully describe error from a theoretical pointof view.A new comprehensive uncertainty method is proposed.First,a table of error as a function of water cut ispresented by using error propagation theory.Second,real field data from forty different wells was analyzed.A well-defined period without any changes in operating conditions is chosen per well and an Arps declinecurve is fitted.Then a histogram was created to establish an error distribution function(the error is definedas the difference between the measured rate and fitted curve prediction which is taken as the real state forcalculation purposes).Finally,a correspondence between error and confidence interval was highlighted torepeated or monitored specifically.A new frequency proposal and test priority is presented.The wells areclassified into type I type II and type II according to the relationship between flow and uncertainty.Thisclassification is key to define a new measurement prioritization system Impact of workovers and interventions in high water cut wells,like Estancia Cholita,are now monitoredaccurately.This allows improved control over production leading to better decision making.Specially,whenmonitoring a pilot,such as polymer injection where the accuracy of the devices used becomes very useful.Frequency and prioritization of rate measurements are specifically described for individual wells.Using this novel approach,unnecessary measurements and operations are reduced.Having clean datacould lead to a successful data mining analysis.Workovers and intervention impact on high water cut wells can be monitored more accurately.By knowing the error in the production rates,future projects will be welldefined managed and evaluated.