In order to enable IoT nodes to efficiently use their energy harvesting capabilities, algorithms are used to determine a reasonable energy budget and allocate it to the node tasks, enabling energy neutral operation. However, most of these algorithms have been implemented and evaluated in simulation frameworks. In this paper, we evaluate the implementation of these algorithms to manage the energy of real-world LoRaWAN IoT nodes. We measure and compare the performance of the different energy budget estimation methods on a commercial LoRaWAN IoT platform. Results show that in this use-case, the choice of algorithm impacts the system Quality of Service by less than 15 %. This enables much simpler energy budget estimation methods to be used.