People in South East England can try out a new system of getting hot water for free. Her name is Heata e uses the heat produced by the servers to heat the water from the water heaters in the homes of the people who have joined the project.
At first glance, the concept behind Heata remains elusive. How is it possible that the heat produced by the data processing of a server can heat the sanitary water of a house? Is it district heating? In a sense yes: the server is mounted directly on the boiler in the home of the person participating in the project, it receives data from the cloud, starts its processing, heats up and the heat produced generates hot water.
Once you understand the process, it’s natural to wonder who pays for the electricity needed to run the server. If the landlord paid for it, Heata would have found a way to relocate its servers while saving on the cost of the electric bill.
But no, the energy to run the processing units heata payindeed, the user is credited with the amount for the energy used with an increase of 10% more than the market rate.
A server at home to get free hot water
Originally a British Gas project, Heata has now been spun off from the British energy company and is living a life of its own, running cloud services focused solely on workloads running on Kubernetes.
Heata does not accept jobs that require real-time computation but, among others, heata grinds data for computational finance and risk analysis, higher education research and supercomputing, medical research, climate modeling, and 3D rendering.
These calculations generate a huge amount of heat in the machines called to solve them. According to Heata, between 20 and 60% of the energy consumed by data centers is used to run cooling systems for servers.
Ever since it was a British Gas project, Heata therefore thought of exploiting this incredible amount of lost heat to heat the sanitary water of a part of the British population, because the heat generated by a single Heata data center can theoretically heat the water of 11,000 homes.
In Heata they could have built district heating works, but the structural costs would have been very high. So they figured that transferring bits is less energy intensive than transferring heatthus the processing units to be mounted at people’s homes were created.
The processing unit is mounted on the home boiler, a parallelepiped about 40 cm long and 11 cm thick, with a large metal plate that is bonded to the domestic hot water tank using thermal epoxy resin.
The plate is connected directly to the CPU heat sinks, which in the experiment are 56 refurbished Xeon cores. Installation is carried out by British Gas, a plumber is not required and fitting does not affect the boiler warranty.
The unit receives data using a cable, 4G or 5G connection paid for by Heata, then it starts processing them, the water from the boiler is used to dissipate the heat produced by the calculation and the boiler heats the water inside it. The noise level of the unit is 20dB.
With this system, Heata says it can deliver 4.8 kWh of hot water per household per day, which is equivalent to around 80% of the average British household’s hot water demandsaving her about 200 pounds a year, equal to about 225 euros.
The boiler is therefore capable of producing hot water even without the help of the processing unit, a fundamental feature should the unit malfunction and interrupt the calculations that heat it.
Each Heata unit consumes 56% less electricity than a data center would need (for unit specific calculations) plus hot water consumption. The saving of CO2 emissions per single unit is 1 ton per year.
The project is in the testing phase and will involve a hundred processing units in as many homes for the next three to six months.
Cover image: stock.adobe.com – Valerii Honcharuk
We want to give thanks to the author of this short article for this remarkable content
Free hot water at home by mounting the servers on the boiler: the trial of a widespread Heata data center is starting in the UK
Explore our social media profiles and also other related pageshttps://prress.com/related-pages/