Probabilistic Flow Modelling Approach for Kick Tolerance Calculations

Abstract

Listen

Translate article

During drilling, there must be an evaluation of the maximum pressure that the formation can handle during a well kill scenario. This will depend on various parameters like fracture pressure, pore pressure, kick volume and several other factors. The depth of the next planned hole section will depend on if a kick of a certain size can be handled safely. This evaluation is often referred to as performing kick tolerances. When starting to drill a section, one will take a leak off test to get an indication of the fracture pressure at the last set casing shoe and this will be important information for the kick tolerance results. For HPHT wells the margin between pore and fracture pressures will be small, and one often has to resort to using transient flow models to perform the kick tolerances. However, there are many uncertain parameters that are affecting the results. Some examples here are pore pressure, type of kick and kick distribution. There is a need for trying to incorporate the uncertainty in the calculation process to give a better overview of possible outcomes. This approach has become more and more popular, and one example here is reliability based casing design. This paper will first describe the kick tolerance concept and its role in well design planning and operational follow up. An overview of all parameters that can affect the results will be given. In water based mud, the gas kick will be in free form yielding higher maximum casing shoe pressures compared to the situation when oil based mud is used where the kick can be fully dissolved. Then it will be shown how both an analytical and a transient flow model can be used in combination with the use of Monte Carlo simulations to generate a probabilistic kick tolerance calculation showing possible outcomes for maximum casing shoe pressure for different kick volumes. Here uncertain input parameters that can affect the calculation result will be drawn from statistical distributions and propagated through the flow model to estimate the casing shoe pressure. Multiple runs will be needed in the Monte Carlo simulation process to generate a distribution of the maximum casing shoe pressure. This will demand a rapid and robust flow model. The resulting maximum casing shoe pressure distribution will then be compared against the uncertainty in the fracture pressure at the last set casing shoe to yield a probability for inducing losses. The numerical approach for predicting well pressures and a schematic of the total calculation process will be given. Emphasis will also be put on discussing how this should be presented to the engineer with respect to visualization and communication. It will also be shown that one of the strengths of the probabilistic approach is that it is very useful for performing sensitivity analysis such that the most dominating factors affecting the calculation results can be identified. In that way, it can help in interpreting and improving the reliability of the kick tolerance simulation results.