Subsurface tidal analysis requires only continuous pressure monitoring data and therefore can be a cost-effective technique for estimating aquifer properties. The tidal behavior of a well in a semiconfined aquifer can be described by a diffusion equation that includes a leakage term. This approach is valid for thin aquifers, as long as the overlying layer has low permeability relative to the main aquifer. However, in cases where the aquifer is not thin and the permeability of the overlying layer is not low, using the existing solutions based on these approximations may lead to unsatisfactory outcomes. Alternative solutions for both vertical and horizontal wells were obtained by solving the standard diffusion equation, with leakage expressed as a boundary condition. Furthermore, a nondimensional number was derived mathematically, which forms the basis for a quantitative criterion to assess the applicability of the existing solutions. In the case of a vertical well, the existing solution exhibits acceptable error only if the nondimensional number is less than 0.245. Our new solution extends this upper limitation to 0.475. However, when the number is greater than 0.475, both the existing solution and our new solution are invalid due to the invalid uniform flowrate assumption. For a horizontal well, when the number is less than 0.245, the existing solution is suitable with acceptable error. Our new solution effectively overcomes this limitation. Finally, the new solution was applied to the case of the Arbuckle aquifer to demonstrate the improved validity of the new solution compared to the existing one.

Tidal analysis provides a cost-effective way of estimating aquifer properties. Tidal response models that link aquifer properties with tidal signal characteristics, such as phase and amplitude, have been established in previous studies, but none of the previous models incorporate the skin effect. It is found in this study that the skin effect and the wellbore storage effect can have significant influence on the results of tidal analysis and should be included in tidal response models. New models are proposed with skin and wellbore storage effects fully incorporated, so that aquifer information can be assessed more accurately based on tidal analysis. The models can be applied to confined aquifers with only horizontal flow or semiconfined aquifers with both horizontal flow and vertical flow. For confined aquifers, the new model indicates that positive skin leads to larger phase lag between the tidal response the the theoretical tide, and negative skin can reduce the phase lag or even cause a phase advance. For semiconfined aquifers, both the skin effect and the vertical flow affect the phase difference between the tidal response and the theoretical tide, and with the proposed model, contribution from these two sources can be separated and analyzed independently, making it feasible to evaluate semiconfined aquifer properties considering both factors. Increasing wellbore storage causes larger phase lag or smaller phase advance for both types of aquifers. Real-world examples for confined and semiconfined aquifers are analyzed respectively to demonstrate practical applications of the proposed models.

Geothermal well log analyses consist of utilizing pressure and temperature data measured along the wellbore to predict feed zones, reservoir temperature, and reservoir pressure. Interpreting geothermal production logs can be subjective and require great expertise to achieve repeatability. In situations where there may be several log data, interpreting the logs may be time consuming for quick decision-making processes. This work discusses the implementation of a multi-layered deep learning convolutional neural network to automatically diagnose sets of temperature and pressure well logs. The algorithm achieves results similar to those of a professional engineer. This algorithm enables the interpretation of many well logs in just a few seconds. Data input for this project is synthetic well data that mimics real data. 10,000 datasets were used. The data was split as follows: 9,800 and 200 for training and validation respectively. The algorithm used takes as input three “depth-series” logs of temperature, pressure, and temperature gradient, passes the data through a convolutional neural network including a flat layer and then a fully connected layer with five output variables which are the depths of the feed zones, the reservoir temperature, the reservoir pressure and the depth at which the reservoir pressure is known. The cost function for this model was the mean squared error. The optimizer algorithm used was Adam, and the learning rate had an exponential decay. The algorithm recorded the model state that had the lowest mean absolute validation error. The architecture was implemented in Keras with a TensorFlow backend. The best model found during the process of hyper-parameter tuning was used to predict the reservoir characteristics for the validation and testing data sets. The results show a good match between the predicted data and actual data with a training error of 0.9%, validation error of 2%, and test error of 7%. Future works will involve adding more real data to the training and validation set, increasing the number of feedzones that can be identified, and performing sequential analysis using interdisciplinary data.