A bit more than a year after my previous blog, it is time for an update. Since this is my second blog, following the 1-2 tradition rule, I now have a steady series going on. Since every series has a common thread, this one can’t be missing out. With these blogs I want to share my experiences as a PhD-student and give you an insight in the science which is conducted at the other end of the building. In this series I will introduce you to a field which you won’t encounter that often as a physicist.

Last time I wrote, I was at the precipice of my PhD-research, now I am already one year into my adventure. The first conclusion I can make is that time has not stood still. In a desperate attempt to remove all connections with business and management, the institute for which I work has changed names: from ITM (Institute for Technology and Management) to ENTEG (ENgineering and TEchnology institute Groningen). Whether or not this has worked, I do not know. What I do know however is that TBK (or IEM) is still here. But let’s not dwell on these minor matters.

As a recap, what do I do? I am a PhD student in building 17 at the third floor. Does that place even exist I hear you say. The answer is yes! So what is it that drives me every day to this strange place? It is the science of Control Engineering. Which is, according to Wikipedia, “… the engineering discipline that applies control theory to design systems with desired behaviors.” Explained in layman’s terms, we study a combination of a system, a goal and some sensors. Sensors measure reality and then with the help of some mathematics we calculate the optimal input to achieve the desired goal. Not too difficult, right? Let consider a small example to complete our understanding. Consider the cruise control system of an arbitrary car; the goal is a desired velocity of the car. The system is the mechanics of the car (gas pedal, engine power, etc.) and the physical interaction of the car with real life (air resistance, resistance of the tires, etc.). The sensor is the speedometer of the car. The next step is to design an algorithm to achieve the desired goal: if we drive too slowly, we step on the gas and if we drive too fast, we slow down. Got that one in pocket!

Now if we would stop there it wouldn’t be very interesting. Therefore scientists have started thinking about improvements on the cruise control systems to be implemented in self-driving cars. The results are very interesting and make use of sensing technology to measure where other cars are driving and how fast they are going. The first improvement which was made is an adaptive cruise control, i.e. a cruise control system which automatically adapts the velocity of your car depending on the velocity of and the distance between the car in front of you and your car. Fun fact: consider a column of 7 cars all using adaptive cruise control. If the first car brakes only a little bit, the last car will almost have to make an emergency stop, in technical terms the current system is not string stable (see for example here for a demonstration of string stability). So what do control engineers do in this case? They devise the mathematical representations of these cruise control systems and investigate stability properties of the algorithms, see reference [1] if you are interested in the technical details of such analyses.

The beauty of the field is that you can apply this theory to almost any situation in which you want to automate stuff. So going back to what I do. I study the thermal control and energy efficiency of data centers. Without realizing it, data centers are becoming a great part of our lives, think about storage in the cloud or the giant Google data centers which you connect to every time you search on Google. In 2012 and 2013 the combined energy consumption of data centers worldwide was about 350 TWh per year. In comparison, according to the CBS, the total energy consumption of the Netherlands in those years rests at 120 TWh. So efficiency of data centers is a hot topic nowadays.

In Figure 1, a typical layout of a data center is given. Racks of computers are oriented in hot and cold aisles. In the cold aisles (blue arrows), cool air from the cooling system enters the system. The air is heated up by the computers and in the hot aisles (red arrows) the hot air is extracted and sent back to the cooling system to be cooled down again. To break this topic down in terms introduced earlier on: the goal is to make sure that the computers do not exceed a certain temperature threshold. The system is the thermodynamic relation between the work the computers do and the heat that is generated by executing this work. The sensors are the temperature sensors attached to the computers. The control system is a scheduler that divides the work in a smart way such that temperature is kept in check and the least amount of energy is consumed. In a nutshell, the past year I have focused at understanding the smart ways people have come up with until now. The goal for the coming year is to formulate my own theories such that by the end of 2016 I can stand in front of a large audience of scientists and convince them that my ideas are simply amazing!

Next time: Why being a PhD is the best job of them all! (or maybe not the best job)

Figure 1: Typical layout of a data center. Racks of computers positioned in hot (red arrows) and cold (blue arrows) aisles

1. Oncu, S., Ploeg, J., Wouw, N. van de & Nijmeijer, H. (2014). Cooperative adaptive cruise control : network-aware analysis of string stability. IEEE Transactions on Intelligent Transportation Systems, 15(4), 1527-1537.