Steady as She Goes: Reducing Flow-Induced Vibration in Hard Disk Drives

Making smaller and less expensive disk drives is critical for success in many consumer electronic devices. One aspect that is under constant examination for improvement is steady flight of the disk heads. They fly only nanometers above the disks which are rotating at 7200 rpm or more. While the spinning of the disks and the viscous entrainment of the air within the drive allows the heads to fly, it also creates dynamic secondary flows that may buffet the head. This may necessitate the design of more sophisticated controllers to combat induced vibration or limit how narrow a track may be. It seems reasonable that one can design the internal structure of the disk drive to mitigate the effects of these secondary flows. The flow in the drive had been investigated to some degree already using direct solution of the Navier-Stokes equations using proprietary software at Hitachi. While it seems to be a very good model giving good understanding of the flow in the drive, it is rather computationally expensive, requiring weeks to run. The purpose of this problem was to investigate simpler models for use in designing an appropriate flow. Perhaps the main idea is compute a good solution for a diverter that causes a decreased flow past the read heads and thus reduces buffeting of the heads; Figure 1 illustrates the basic idea. A number of ideas were proposed that are discussed below. These ideas included: a lumped parameter model for the flow inside the drive; change the head/suspension design to prevent shedding of vortices; insert a fluttering blocker to prevent shedding vortices from the head/suspension assembly; computing the results of diverter shapes using compressible Euler equation models; apply shape optimization algorithms to the solvers to try to design an optimal diverter. All but the first tried to simplify the problem by studying only a local area near where flow would be diverted from within the spinning disks, just upstream of the heads; this is the darkened region in Figure 2. The first approach modeled the whole drive with a small ode system calibrated with some known quantities from the drive. What results we have will be presented in the following sections, followed by Discussion and Summary sections.