Leakage diagnosis method of injection-production string in underground gas storage wellbore based on cooperative monitoring of multi-physical fields

Leakage of wellbore injection-production string is one of the important risks affecting the safe operation of underground gas storage reservoirs, and its early monitoring and diagnosis are crucial to safeguard the integrity of the reservoirs. However, the current monitoring technology for injection-production string micro-flow leakage still has the problem of insufficient accuracy. In this study, a wellbore injection-production string integrity failure diagnostic model is proposed based on the multi-physics collaborative monitoring technology of distributed acoustic sensing (DAS) and distributed temperature sensing (DTS) to improve the efficiency and accuracy of leakage diagnosis. Through theoretical modeling, experimental device development, and leakage simulation experiments, the characteristic patterns of temperature and acoustic vibration fields under different internal and external pressure differences (5 MPa, 7 MPa, 9 MPa) and leakage rates (0.25 L/min to 10 L/min) are systematically analyzed. The results show that under high leakage rate and high differential pressure, the temperature and acoustic wave signals near the leakage point change significantly. Under the condition of small leakage (≤0.5 L/min), it is difficult to effectively capture the leakage characteristic signals by the multi-physical field cooperative monitoring technique. For this reason, this study proposes a data fusion diagnostic model based on the convolutional neural network (CNN) and D-S evidence theory and successfully realizes high-precision identification of tiny leaks (support are higher than 0.99). Compared with the traditional model, the method shows significant advantages in terms of sensitivity, specificity and accuracy. The multi-physics field cooperative monitoring method in this study provides a new technical approach for integrity monitoring of wellbore injection-production string in gas storage reservoirs. The proposed wellbore leakage failure diagnostic model provides a reliable theoretical basis for improving wellbore safety and early warning monitoring capability.