Cost optimizations linear models are a common tool for long term energy planning models, required to balance the different costs and resources associated to energy transition. However, these models, based in mean values, struggle to evaluate high shares of variable energy resource and the performance of energy storage services, which are strategic elements of the transformation.

The aim of this work is advance in the adequate evaluation of high shares of intermittent renewable resources in the long-term energy planning tools, specifically in the presence of long term storage facilities.

Complementary to previous efforts, whose focus was the classification of days depending of the level of resource, this study will focus in the availability pattern of the resource, which can have a great impact in the storage requirements and performance. The challenge will be addressed defining a new set of corrections and constraints based in data-driven parameters.

As first step, the state of charge of the storage system will be simulated and analysed for a wide number of sets of resource sequences and a range of storage capacity values. The set of resource sequences will be fabricated depending of a range of parameters: fluctuation intensity, frequency and abruptness of its change, and autocorrelation metrics. The pattern study focus will be in the diary and hourly time-scales. Previous analysis will allow linking the resource characterization to storage performance metrics: number of cycles, percentage of the energy demand delivered by the storage service, usage capacity factor of the storage and the percentiles of the storage level, and unserved energy. Based in this association, constraints and boundaries in the storage performance will be established depending of the resource characteristics. IA tools, linear regression or k-means approaches will be used with that purpose.

In a second step, previous information will be used to introduce new correction parameters in the optimization linear model, ensuring an adequate estimation of the storage capabilities for each specific resource pattern. The performance will be dependent of the investment in new storage capacity, affecting to the optimization. An iterative approach will be used if required to ensure convergent solution.

The proposal will be tested and compared with the usage of the original data sequence and other available approaches, evaluating the adequacy of the results and the storage performance. Main test will consist in the optimization of the sizing of a system based in thermal generation, renewable resources and one long-term storage service. Evaluation will include the investment and energy supplied by every element and their deviation respect the full data sequence. If possible, systems with several intermittent sources and storage devices will be tested to evaluate the applicability range of the approach.