This paper presents a new method to deal with uncertainty mitigation by using observed data, integrating the uncertainty analysis and the history-matching processes. The proposed methods are robust and easy to use, and offer an alternative to traditional historymatching methods. The main characteristic of the method is the use of observed data as constraints to reduce the uncertainty of the reservoir parameters. The main objective is the integration of uncertainty analysis with history matching, providing a natural manner to make predictions under reduced uncertainty. Three methods are proposed: (1) probability redistribution, (2) elimination of attribute levels, and (3) redefinition of attribute values. To test the results of the proposed approach, we investigated three reservoir examples. The first one is a synthetic and simple case; the second one is a synthetic but realistic case; and the third one is a real reservoir from the Campos basin of Brazil. The results presented in the paper show that it is possible to conduct an integrated study of uncertainty analysis and history matching. The main contribution of this work is to present a practical way to increase the reliability of prediction through reservoir-simulation models that incorporate uncertainty analysis in the history period and provide reliable reservoir-simulation models for prediction forecast.

This paper was prepared for presenlal~at lhe 19S6 SPE Annual Tect'mual Conference and Exh!b!tlon held m Oenver, Colofado, U S A S-9 October 1s9STh!s paper was selected for prescnlalion by an SPE Program Commdlee follming review of !nfcfrnalim ccmlamed m an abstr~oubmifted by the author(s) Ccmtents of the pap+r, as present~, have not been rev$ewed by the .%aety d Petroleum Engineers and are subjecl to corrediin by the atkhw(s) The material, as presented -s not nacessariiy reftect any prtmn of the Soaety of Petroleum Engineers, tis Mars, of members Pafmrs presented at SPE mee!hngs are sub@t to publicatmn rewew by Edtiorial Commttteas of ttw Soaety of Petro!.Wm Engmaers Permrssmn to copy #s mstncted to an abstract of not mofe than 303 words Illustratmns may not be cc.pted The abstracl stm"ld ccmtam canspwmus scknotwedgment of where and by whom the paper was presented Wnfe Ltbreman, SPE, P 0 Box S33836 Rr.hardson, TX 75083-38SS, U S A , fax 01-214-952.9435Abstract Sucker rod pumping system is a very important artificial lift method. The surface dynamometer card (SDC) is a plot of the load versus the pumping cycle measured at the polished rod. The SDC shape is assumed to reflect the actual pumping conditions. However, SDC is a composition of the actual downhole pump dynamics and the noise added during the information transmission along the sucker rod. The difficulty in recognizing specific SDC shapes augments as the amount of noise increases mainly as a function of the well depth. Filtering algorithms may be developed with the purpose of obtaining the downhole dynamometer card (DDC) from the recorded SDC. Artificial Intelligence may provide the adequate tools for DDC classification. The present paper describes an intelligent system which automatically calculates DDC by SDC algorithm filtering, and classifies DDC taking into consideration a set of patterns associated to the most frequent abnormal pumping operating conditions, This system is actually running at different Brazilian Oil Fields. The paper describes the main characteristics of the filtering algorithm; it shows how linear mathematical programming and neural nets are used for DDC classification, and presents results collected at different pumping conditions. These results show that the system is highly reliable in correctly identi&ing the actual pumping conditions through DDC classification.