Future Smart Grids will need to integrate thousands of independently operated intelligent energy devices into one organizational unit. Distributed ledger technologies can provide the means for a business-driven consensus between supply and demand of electrical power. This approach is limited by the computing power of the participating devices. We describe a way to increase the scalability of this approach by a) aggregating power consumption and generation, thus reducing the computational load on the blockchain and by b) applying the concept of edge computing to increase reliability and reaction time. We propose a hierarchical clustering approach for smart grid power balancing, which aims to advance our understanding of architectures for future real-world smart grids.