The Power Game

Demo device communicating to Ethereum
Miner with shaking camera :P
Problem statement
A cool chart on energy consumption in Germany

Inspiration

Power provider companies use spot prices in order to keep their electricity generation balanced. They design power consumption plans for industrial power consumers, which involves a minimum and a maximum consumption restrictions which impose penalty on violation. The consumers have to keep an eye on the price changes and regulate their energy consumption (and goods production) accordingly, which sometimes leads to consumers having too low consumption rate. Since each consumer monitors its own energy consumption and the price rates individually, they all might reduce or increase their production at the same time. However, if all the consumers could handle their energy consumption in a collaborative manner, the energy consumption would be optimal. For example, when one consumer requires a low amount of energy, another one can cross his maximum limit without imposing pressure on the power provider and vice versa.

What it does

We design a game in which there are a number of industries as energy consumers and power provider companies. The game is a sort of distributed bidding system deployed on the top of a blockchain, which reads the most recent energy rates and based on other participants' bids, and controls the volume of the industrial production line. The utility function of the game is designed to optimize both the energy production and consumption (which in turn translates to goods production). Basically in the Nash equilibrium point of this game all the players win. There are no human beings involved; all IoT devices talk to each other and communicate through a blockchain.

How we built it

We deployed Ethereum on a server (a laptop for the POC). We use a STM32 Nucleo development board with NodeRed which reads the meters and communicates with a Raspberry pi (another laptop for the POC) over CoAP. The Raspberry pi is an Ethereum client which publishes the values into the blockchain and read them in order to predict the energy demand for the next day. We make use of a linear regression algorithm for the prediction.