Abstract
Some mature reservoirs in China are encountering a sharp production decline, indicating a near-bleak production stabilizing prospect. Production fluctuation is essentially a phenomenon with a variety of affecting factors. An effective monitoring method has become a must with urgency for high water cut reservoirs. The existing development analyzing method or warning method is merely restricted to some particular indicator regarding development dynamics. However this kind of mechanism cannot be responsibly extended to continental deposited formations with strong heterogeneities for a lack of reliability and accuracy. In a reservoir engineering standpoint, all fluctuation in production can be feasibly traced from corresponding development dynamics indicators, which can change accordingly. Thus it is possible production change might be foreseen or predicted through analyzing the abnormal variations in indicators. We may start up analysis with oilfields and oil production plants owning sharp production variations, probe the causes behind and find the reflective indicators from 47 reservoirs. Our study has come up with a total of 10 perspective indicators for early warning, for instance, previous year oil rate ratio of new wells, ratio of oil rate from production enhancement, production injection ratio, enhancement efficiency and measurement Shortage, and the like. Various comparison and assessment standards are viable for different indicators. Oil production variations are further categorized into abnormal, normal, and better scenarios after evaluating uni-indicator using methods of plan comparison, trend comparison and history comparison without a neglect on individual indicator performance. Based on conventional warning strategies, a fresh early warning model with vector machine is devised. We apply this new model to 12 reservoirs in Shengli oilfield. Over 90% of abnormal production variations are successfully detected by this model through analyzing their nearly 20-year production history, yielding a practically good predicting accuracy and showcasing a big promise in effective monitoring for oil production.