At it's most basic Thermal Rail™ provides cooling to connected servers via a simple physical interface that cools each server independently. With heat flowing from the server, through the Thermal Rail™ interface and out to the Data Center's cooling services.
This physical interface can be thought of as being simply a 'thermal' connector that the server is plugged into, in the same way that a server is plugged into a power connector or networking connection. Cooling occurs only via heat transfer between the server and the Thermal Rail™ interface and so no coolant enters the server.
To deploy Thermal Rail™ there are two primary parts.
The first is an enclosure system that's installed into the Data Center and replaces the traditional rack. The enclosure can be designed to be any width or depth and could for example be used as a drop-in replacement for a standard 19" rack and support (not cool) conventional air-cooled systems, e.g., top of rack switches.
The proof-of-concept enclosure is for instance a simple 8U enclosure with Thermal Rail™ interfaces on only one side.
The enclosure is designed to provide access to the Thermal Rail™ interfaces and provide the necessary cooling while isolating a coolant flowing within the enclosure wall from the electronic systems being cooled. In this way, it is designed to be mission-critical compatible with fail-safe operation and optional capability of full active-active redundant operation.
Thermal Rail™ proof-of-concept enclosure
View of enclosure manifolds
The manifolds and internal plumbing for the proof-of-concept enclosure can be seen in the adjacent image. There are two flow paths (green and yellow) which are independently able to provide the necessary cooling.
The manifolds split the flow evenly to each server resulting in each server receiving its own individual coolant feed and so each server is also isolated from the thermal characteristics of its neighbors.
Each enclosure is connected to the Data Center's coolant service with coolant flow within the enclosure designed to be at a temperature higher than 30°C. This means that the coolant temperature can be maintained by a simple cooling tower anywhere on the planet. Without needing expensive mechanical chillers. This also means that water use is reduced as evaporative towers are not required in most locations for the majority of the year.
Condensation is prevented by managing the coolant temperature within each enclosure.
The second component of a Thermal Rail™ deployment is the compatible server equipment. These are simply servers designed to interface with Thermal Rail™. Initially, this could be an existing server adapted to work with Thermal Rail™, a hybrid-air or hybrid-liquid-cooled system or a system designed specifically to be cooled via Thermal Rail™. The latter being fully fanless and resulting in the best cost-efficiency.
The design of a Thermal Rail™ server is limited only by imagination. Any server and component can be cooled using Thermal Rail™.
The system designed and built for our proof-of-concept demonstrates a fanless server with a minimum number of parts. This particular design has all the various components necessary for a common dual-CPU deployment including dual-CPUs, HDDs, RAM, an expansion card, and a variety of other components.
A fully operating Thermal Rail™ system can be seen in our proof-of-concept video below.
One example of a Thermal Rail™ server