Leak detection in liquid subsea flowlines with no recorded feed rate

This paper presents a statistical method to detect leaks in subsea liquid f lowlines when the inlet flow rate measurements are unavailable and conventi onal mass balance techniques cannot be used. Presently the only method used in these situations is to "predict" the inlet rate from a steady-state res ervoir/wellbore model for use in a transient pipeline simulator. The purpos e of this work was to examine the utility of a pressure loss model which re quires only routine production data. This technique is derived from the sta tistical theory of estimation and testing, and treats fluctuations due to t urbulence, transients, and measurement errors as statistical noise. This me thod compares two mean values of the pipeline resistance coefficient, R; th e new updated value representing the potential leak and the old value being R without leak. Precision of the new method was demonstrated using the 946 0-ft long, (3.64 in, i.d.) flowloop at LSU, with flowrates up to 10,000 bbl /day of water and pressures up to 2400 psi, The leak was simulated by an or ifice which discharged the water from the flowline into a shut-in, well. An environmental (wellhead) pressure up to 2200 psi was applied to simulate h ydrostatic pressure in the deep-sea environment. The results presented in t his paper show how detection of small leaks relates to the number of R meas urements, leak position, and statistical noise resulting from flow rate-pre ssure fluctuations, transients, and measurement uncertainties.