The scenario sounds horrifying: A lightning bolt strikes the ground just eleven meters from a pedestrian. Whether he survives the strike will depend upon, among other things, the length of his stride. Fortunately, the scene took place in a computer, rather than outdoors, and the pedestrian involved was a computer model of a human being, rather than a living person. The lightning bolt also existed exclusively in the form of bits and bytes. A team of researchers, led by Professors Thomas Weiland and Irina Munteanu of the Institute for the Theory of Electromagnetic Fields (TEMF) at the TU Darmstadt, did the programming.
Prof. Munteanu noted that, “We want to observe what goes on in the bodies of people present near a lightning strike at the time of the strike.” Since the currents flowing through human beings cannot be measured, a computer model is necessary for assessing the currents to which their hearts are subjected when lightning strikes near them. The researchers also simulated lightning bolts striking passenger aircraft carrying a full load of passengers and automobiles equipped with sunroofs. In the latter case, for example, they found that field strengths were much higher in the vicinities of their driver’s head than was the case for automobiles having conventional, one-piece, metal roofs.
Every lightning strike consisted of hundreds of Terabytes of data
The methods that Weiland and Munteanu’s team has developed will be of great interest to engineers. As Munteanu put it, “For example, they can use them for optimizing grounding systems,” since more accurate knowledge of the currents involved means that grounding systems will no longer need to be overdimensioned. Furthermore, their simulations will contribute to effectively shielding aircrafts’ onboard electronics against lightning strikes.
They are also able, in principle, to simulate electromagnetic fields generated by other sources, such as the electromagnetic radiation emitted by mobile telephones when they are utilized inside aircraft. Prof. Munteanu added that, “If the field strengths involved were accurately known, aircraft cockpits could be effectively shielded against them.” The professor of engineering believes that her team’s simulations might accelerate development processes. She went on to state that, “Although computer simulations cannot totally replace direct measurements, they eliminate need for building numerous prototypes, most of which will turn out to be unsuitable once measurements have been made.” That alone will save a lot of time and money.
Simulation models are highly detailed
The Darmstadt researchers are treading new pathways with their simulations, since their models are highly detailed. Their models are, in a sense, assembled from virtual building blocks. They have configured sheet-metal and plastic building blocks for automobiles and bone and brain-tissue building blocks for human beings. However, in order to arrive at reliable, accurate results, the building blocks employed must be chosen extremely small if all details are to be incorporated, which means that a complex model will consist of large numbers of such building blocks. For example, a reasonable model of a passenger aircraft carrying a full load of passengers consists of around a billion such building blocks.
The institute’s computer cluster solves Maxwell’s equations for every one of those building blocks. However, even that is not enough. In order to simulate the temporal courses of the processes occurring, solutions of those same equations must be reiterated, over and over again, for each point in time. As Munteanu put it, “A total of hundreds of terabytes of data must be computed, saved, and graphically displayed.”
The researchers had to dig deep in their box of tricks in order to prevent such computations from taking years. Following every computation for a simulated object at a given point in time, the results of computations must be exchanged as efficiently as possible between the around 172 computers in the cluster, which requires some clever programming. Furthermore, the simulations are run on superfast graphics cards, rather than on slower, conventional microprocessors. On that point, Munteanu stated that, “In order to do that, we needed sound knowledge of both the hardware and software involved.” In the future, that knowledge will be utilized for simulating further, spine-tingling scenarios.