Data storage: Better hard drives ready for lift-off

Researchers at the A*STAR Data Storage Institute (DSI) in Singapore have now developed a computational algorithm for studying the properties of the slider in a hard drive that is faster than existing algorithms1. Instead of taking days to finish, dynamic simulations using the new algorithm take only an hour, notes Wei Hua from the research team. "It greatly improves our simulation and research abilities," he adds.

A read head typically moves across the disk surface of a hard drive at more than 7,000 revolutions per minute. The flying height of this fast-moving head is as low as 2 nanometers from the surface of the disk, some 50,000 times less than the width of a human hair. Controlling this motion is not easy, notes Hua. "The slider housing the read/write head flies on the fast-rotating hard drive disk, owing to a very thin layer of air. This air bearing pushes the slider upward, while a suspension bearing pushes the slider down toward the disk."

Thermal effects control the distance of the head to the surface when it is being pushed down. To understand these effects, and other factors that control disk and head movements at high speed, fine-grained computer simulations are necessary.

Hua and co-workers expanded the DSI's ABSolution air bearing simulation software for faster and more precise modeling. Instead of dividing the hard drive slider into a structured rectangular mesh typically used to aid calculations, the researchers used an unstructured triangular mesh that accurately captures the geometry of the read/write head (see image). Moreover, the algorithm better implements the dynamic effects that occur in drive heads, meaning that overall the code works faster and more efficiently.

This modeling software should prove useful in the future development of drive heads, Hua notes. Modeling the interaction between the slider and the rest of the drive is also important. "Influences such as those from the air suspension and disk effects are now being considered," he explains. Hua and co-workers will use the improved algorithm to model slider properties that were almost impossible to simulate using the previous versions.