Heavy oil fields that produce through cyclic steam stimulation often demands one very recurrent, time consuming and important task during its lifetime: stimulation scheduling along the hundreds of the existing wells. Installed steam generation capacity is usually lower than the demand and is diminished even more by maintenance stoppages. This scheduling could be defined using numerical simulation, but beyond demanding a computational infrastructure, the time needed to complete the task is usually much greater than the time available.

Computational infrastructure is not usually an issue but time availability in order to run the numerous simulations necessary to complete the optimization task, as it is commonly recurrent as circumstances change as time passes. Therefore, the most common option has been to make use of the skills and experience the reservoir engineer and geoscientists gathered during their careers. Using historical production data and reservoir parameters in a subjective analysis, taking into account local experiences extrapolated to the adjacent area or even to the whole field, the reservoir team defines which order the stimulation should be done. That method has shown to be effective, but it takes time to be done and depends on the professional intuition.

Many academic references describes analytical models that tries to represent the cyclic steam stimulation and using those models, some attempts were made in order to maximize the results of the stimulation of each well, regardless the of the critical resource usage optimization. 

This research proposes an unseen methodology versing about how to combine the simplicity of the numerous analytical models available and the amazing modern computational power in order to complete a probabilistic analysis of the optimum financial return that could be reached with the installed steam generation facilities. It can also be used to sustain decisions regarding either investing in capacity enlargement or transfer between fields or production areas. 

Analytical models theoretical aspects are discussed, including application limitations and in which circumstances one can have better confidence using them as well as which adaptations can be done to diminish lacks of fitting. Instead of defining a model and strategy combination that would fit every scenario, it is more likely a suggestion of a methodology that aims better results. Therefore, it is up to the evaluator to recognize which the characteristics of the real case are and choose a model and a strategy that better complies with the case.

Finally, results from the methodology are compared with those obtained from an intuitive approach for a standard case through numerical simulation using Computer Modelling Group’s thermal simulator STARS®.