We consider the operation of an energy harvesting wireless device (sensor node) powered by a rechargeable battery, taking non-idealities into account. In particular, we consider sudden decrease and increase of the battery level (leakage and charge recovery consequently) due to the inner diffusion processes in the battery. These processes are affecting the stability of the device operation. In particular, leakage accelerates the depletion of the battery, which results in inactive periods of the device and, thus, potential data loss. In this paper, we propose a simplified self-control management of a battery expressed by restrictions, which could be used for an efficient operational strategy of the device. To achieve this, we rely on the double-queue model which includes the imperfections of the battery operation and bi-dimensional battery value. This includes both apparent, i.e., available at the electrodes and true energy levels of a battery. These levels can be significantly different because of deep discharge events and can equalize thanks to charge recovery effect. We performed some simulation and observed that we can diminish the models variable number to predict possible unwanted events such as apparent discharge events (when the areas near electrodes are depleted while other areas of the battery still contain some energy) and data losses. This observation helps to achieve sufficiently effective self-control management by knowing and managing just few parameters, and therefore offers valuable directions for the development of autonomic and self-sustainable operation.