For my PhD, I am working on green data center design. In particular, I am developing a measurement-based parametric model for rapid thermal assessment of a data center. With this parametric model, I have solved several important problems related to thermal design of a data center.
- The first problem is about determining optimal response time during power outages in data centers. This is an important problem for critical thermal decision-making on how long data centers can continue reliable computing services during an emergency such as a power outage. To solve this problem, I have developed a thermal prognostic model to determine reliable operational window during a power outage in a data center.
- The second problem is about reducing temperature monitoring cost in data centers. To solve this problem, I have modeled rack exhaust air temperatures as parametric functions of spatial locations. This framework improves the spatial resolution of measured temperature data; thereby, it reduces sensor requisition for transient air temperature monitoring in data centers.
- The third problem is about determining optimal cooling set points in response to dynamically-evolving heat loads in a data center. To solve this problem, I have developed a parametric air temperature model with time and heat load as parameters.
- The last problem is about determining optimal cooling set points in response to dynamically-evolving IT loads in data centers. To solve this problem, I have developed a parametric CPU temperature model with cooling resources as parameters under a given IT workload profile. This is a pertinent problem for energy-efficient design of cloud data centers.
Overall, the developed framework is useful for real-time thermal decision-making, cost-effective temperature monitoring, and dynamic cooling resource optimization in a data center.
For my master’s thesis, I worked on flow control in microfluidics systems. In particular, I designed a bio-mimetic mechanism to control particle motion inside a microfluidic system. Immersed in a viscous fluid, the bio-mimetic mechanism consisted of elastic filaments tethered to the microfluidic substrate wall and a microscopic particle. I simulated fluid flow and particle motion driven by the oscillation of elastic cilia. The simulation results showed that by controlling oscillation frequencies of these elastic filaments, the particle motion can be manipulated.
Overall, the microfluidic design is useful for building an efficient drug delivery platform and improving convective heat transfer coefficient in a microchannel-based chip cooling device.
A novel mathematical approach is devised to analyze the fluid flow through a rotating microchannel. Special attention is devoted to estimate the effects of variable hydraulic resistance over different flow regimes and the dynamically evolving contact line forces. Flow characteristics depicting advancement of the fluid within the microfluidic channel turn out to be typically non-linear, as per the relative instantaneous strengths of the capillary forces, centrifugal forces and viscous resistances, evaluated in the rotating reference frame. The centrifugal forces are found to be clearly dominating during the later transients, especially for higher rotational speeds. It is revealed that excellent cooling performances can also be obtained by exploiting these features, for CD-based microchannel heat sinks.