Self-Regulating Thermal Tiles

Mechanical Engineering

Data Center Design

A passive thermal-management tile for data center server racks — absorbing, storing, and releasing heat automatically, with no power or moving parts required.

Role

Mechanical Engineering

Timeline

1 Semester · Capstone

Tools

Thermal Analysis, CAD

Method

Material & Geometry Testing

The problem

Data center server racks generate enormous, constant heat. Most cooling solutions are active — they consume power and can fail.

The solution

A tile that cools passively: no power, no moving parts. Heat transfer happens through material and geometry alone — a honeycomb core between a pyrolytic graphite top and a waterjet aluminum base.

Why

Active cooling is one more system that can fail under load. A passive tile removes that failure point entirely — it works because of what it's made of, not what it's plugged into.

Who I worked with

A team of 5 for this mechanical engineering capstone — I contributed to material selection, thermal analysis, CAD, and prototyping alongside the group.

Engineering deep dive

Three materials, chosen for three different jobs: pyrolytic graphite to spread heat fast, a paraffin PCM to absorb it gradually, and an aluminum honeycomb base to keep dissipating it by convection.

Fig. 1 — Material properties: pyrolytic graphite, aluminum 6061-T6, RT50 paraffin PCM

A thermal resistance model (R = L/kA) was built and simulated in LTspice using an electrical-thermal analogy — temperature as voltage, heat flow as current. It verified the multi-layer configuration held up across a realistic range of inlet temperatures and heat flux.

Fig. 2 — Thermal resistance network implemented in LTspice

Design sketch of rack placement and layer breakdown

Fig. 3 — Design sketch: tile placement & layer breakdown

Fig. 4 — CAD: full assembly with graphite top layer

Fig. 5 — CAD: aluminum honeycomb base with PCM cells

Fig. 6 — Final 4″ × 4″ prototype

The prototype came down in scale from the original concept — 8″×8″ to 4″×4″ — after running into machining constraints. Cooling fins were dropped once the mounting location shifted to the rack's side wall, and the vapor barrier and thermochromic coating were cut after faculty consultation.

Testing & validation

Testing combined analytical, numerical, and experimental methods. A directed heat source simulated the rack's thermal load against the mounted tile to validate the simulation against real behavior.

Physical heat exposure test of mounted tile

Fig. 7 — Heat exposure test: tile mounted for real-world thermal validation

Next Project

Cup Carrier

→