Flip a switch and your blender whirrs. Push a button and your electric car starts. As long as what happens is what you expect to happen, you don’t give it another thought. But that start, that ignition and operation that you expect from an electric motor but don’t think about, only happens because people long before you have given it a whole lot of thought. Mihai Comanescu, associate professor of electrical engineering at Penn State Altoona, is one of those people.
Comanescu works with motor drives, which are used, for example, “in transportation, to improve the efficiency of vehicles and in renewable energy applications like windmills,” he says. No matter the type of application, “the electric motor or generator needs to be controlled and must act in a certain way in order to obtain the end result.” For example, in a hybrid electric vehicle like the Prius, the electric motor is responsible for the torque boost required during acceleration and for the energy regeneration that occurs during braking. Taking this task away from the internal combustion engine (which uses a lot of gas during the speed transients) and using the electric drive instead helps increase the overall fuel efficiency tremendously. For a windmill, the speed of the turbine needs to be adjusted constantly as a function of the wind speed in order to maximize the energy harvesting. Using algorithms, the motor drive has to be designed to operate at maximum efficiency — even a mere 1 percent off over the course of the year can end up costing a lot of money.
The key to designing such algorithms is “control theory,” which is, as Comanescu explains, “a general direction of study in engineering related to the control of dynamic systems.” In principle, any object, physical system, engineering product, or sometimes even a chemical process may be described by a system of differential equations. Control theory allows the design of algorithms that will make the system achieve certain performance objectives.
Comanescu’s work begins in the "theoretical stage, usually by developing an algorithm on paper,” he says. At the moment, he is working on an algorithm that “estimates the speed and load torque of the permanent magnet synchronous motor,” a type of drive used, for example, in the Prius and Tesla cars. But the use is much broader, as he says, “the algorithm could go into anywhere electric motors are used.” In June he presented the paper at the 9th International Conference on Compatibility and Power Electronics (CPE) in Caparica, Portugal on that very subject.
After creating an algorithm, the next stages in his research are the simulation and implementation of the method. “In order to control these motors, there are various schemes,” he explains. “Some things need to be measured and some things cannot be measured or it is inconvenient to measure them — as a result, they need to be estimated. So, a lot of research goes into constructing algorithms that estimate the variables of interest associated with these motor drives. The estimators need to be validated mathematically first, then by simulations, and, finally, in a practical microcontroller implementation.”
Comanescu has been teaching for 15 years, initially as a graduate student at Ohio State University and now at Penn State Altoona. In the classroom, he says, “the secret to teaching is to be adaptive. Teaching is different everywhere you go. You have to change what you’re doing depending on the audience.”
Always aware of his audience, Comanescu is very careful to be understood when explaining what he does. “As an engineering professor you have to be precise. If somebody asks a question and you don’t know the answer exactly, I typically say, ‘I don’t know.’ Yeah, it’s a funny moment, the students are a bit shocked at first and you may get a few laughs; however, it’s much better than saying something that is not true. Also, my students understand that it’s okay to not know everything and, for a little bit, you put yourself in the same boat with them.”
In both the algorithm development and in the modeling stage of his research Penn State Altoona students have been and continue to be involved. Past students who worked with Comanescu have traveled to England and Japan, and he hopes that his current ones will have a chance to go to Scotland in spring 2016.
Creating algorithms and simulations and teaching might seem like enough to make a full schedule, but Comanescu also writes. He has authored more than 50 papers on estimation and control of motor drives and currently has a few more under review. “It’s doable,” he acknowledges, but with characteristic clarity he adds, “you need to be aggressive with the deadlines.”