This is especially true, says Liu, a Ph.D.
This streamlined flow, says Liu, is made possible by machines that process large quantities of data in real time and make optimal decisions.
Liu and his colleagues, Martin Takáč, assistant professor of industrial and systems engineering and Liu's Ph.D. adviser, and Jakub Mareček of IBM Research, have made considerable progress in solving optimal power flow problems in the past two years.
For his contributions, Liu recently received IBM's Ph.D. Fellowship Award, one of the industry's most competitive sources of funding for Ph.D. students.
A power system, say the researchers, often contains many power stations, which produce electric power, and a central entity, called the transmission system operator, which coordinates the production and transmission of power.
The goal of the operator is to transmit electricity from power stations to customers with maximum efficiency and minimal losses. If a wire carries too much electric current, it incurs losses and could overheat, possibly causing a blackout.
Operators of power plants and transmission systems cannot choose where power will flow, say the researchers, because power follows the laws of physics. But they can decide where to generate power and how to set the transformers along the way.
The physics of the alternating-current model of power flows makes these decisions difficult to make, says the group. But they correspond to polynomial optimization problems (POPs), which are a hot topic today in the field of mathematical optimization.
Until recently, researchers could apply the so-called Newton method to POPs to obtain a solution quickly. Or, as suggested by Mareček and his colleagues, they could solve a sequence of surrogate problems to obtain the best possible solution.
