Abstract: Well production rate is commonly predicted based on the permanence principle of traditional technical economy prediction. However, such a prediction principle has been challenged by the emergence and development of chaos theory. Unfortunately, there is little work has been done on the chaos identification and prediction of acidized well production rate. This paper presents techniques of phase space reconstruction and correlation dimension extraction, by which the 60-day time series of oil production rate from an acidized well in one oil filed in west China is investigated. It is discovered that the correlation dimension of the time series of oil production rate is 11, and the attracter's dimension is 2.64. Thus, the attracter is a strange one and the production rate varied chaotically. Furthermore, it indicates that the number of essential variables of the production variation ranges from 3 to 11. The results present quantitative reference to presenting accurate dynamic model of the production system, especially to defining the correct number of input layer nodes in ANN (artificial neural network) models.

Key words: oil production rate, acidizing, chaos, correlation dimension, time series