Wind turbines (WT) are one of the main sources of renewable energy, being vital to meet sustainable goals and to increase the share of clean energy in the world energy matrix. However, during their operation, WTs are subject to adverse conditions, such as pollution, atmospheric discharges, among others, that can jeopardize their proper operation. This extreme operation environment, accelerates WTs aging, increasing the number of critical failures, adding to maintenance costs and diminishing wind farms' reliability. To prevent critical failures, predictive approaches to estimate the occurrence of these events are essential. However, most of the papers in literature focus on predicting critical failures based on diagnosing incipient ones. The main drawback of this method is the lack of information regarding when the critical failure will occur after the identification of an incipient fault. In this paper, a methodology to indicate the time remaining until the critical failure of WTs is developed. This method applies simple machine learning algorithms and SCADA data, being of easy implementation and low-deployment costs. It is expected that the method helps energy companies in operation and maintenance (O$\&$M) planning, optimizing theses processes.