Zonal isolation is vital for safety operation of the well. Loss of zonal isolation can lead to sustained annulus pressure (SAP), which requires stopping production or even abandon of the well. The worst result of losing zonal isolation is the contamination of fresh water-table. This is irreversible and detrimental to all living creatures. However, cement sheath is exposed to pressure and temperature changes from time to time during the whole life of well. The temperature cycles can be raised up from completion process as well as the production process. The pressure cycles are caused by different operations, such as hydraulic fracturing. The objective of this study is to diagnose the fatigue behavior that caused by cyclic changing of pressure and temperature during different well operations. The main scope of the study is based on the fatigue failure cases in previous literature laboratory data. Artificial neural network (ANN) is used to predict the fatigue failure of the cement sheath.
The dataset is retrieved from 6 papers of One-Petro, which includes 314 cement fatigue failure cases. Inputs of the ANN were selected to be seven cement parameters (cement material, additive type, uniaxial compressional strength,curing pressure, curing temperature, curing time, and Young's modulus) and seventesting parameters (highest loading, loading frequency,stress increment rate, testing temperature, contain confining rock layer, confining pressure, and fatigue failure cycles). The programming work is based on R language. We looked at 150combinations of features and selected the neural net with the best performance. Back-propagation tuningwas used to fine-tuning the weights of the neural networks. The final predictoris occurrence of fatigue failure or not.
The best accuracy of the ANN method is up to 90.78%, which indicates that ANN can be an efficient method for cement sheath fatigue prediction. The fatigue model we proposed is different from previous study, because we collect the data for real wellbore geometry (hollow concentric cylindrical geometry). This makes the model more applicable to practical situation.
This paper intends to provide a more convenient and accurate model to predict cement sheath fatigue failure under cyclic pressure/ temperature change. The result will benefit the cement system design and wellbore operating strategy design such as hydraulic fracturing.