The Universe abounds with things that we think we might understand but aren’t really sure about. What it often comes down to is our ability to compute: to answer the question of whether models based on the physics we know about generate the behavior we see around us. In response to that question, researchers have turned their computation gun on a long-standing problem: why do luminous blue variable stars exist?
Luminous blue variables are very big, very bright stars, but their temperature (color) varies quite a bit. Scientists were pretty sure that the variability came down, somehow, to a combination of radiation pressure, shock waves, and convection. But until now, no one could confirm that.
Emotional stars lose their equilibrium
In particular, luminous blue variables go from brightness and temperature ranges that are in equilibrium to brightnesses that are far out of equilibrium.
What do we mean by that? At the center of any star, nuclear reactions generate light. The light exerts a pressure on the matter in the star that prevents further gravitational collapse. At equilibrium, the radiation pressure balances gravitational collapse, and everything should be nice and steady (as it is with our Sun). The temperature and brightness of the star reflect this balance.
For stars that are much hotter, like the blue end of luminous blue variables, the radiation pressure exceeds gravitational collapse. During these moments, matter is blown off the star like so many dandelion seeds.
The strange thing about luminous blue variable stars is that they irregularly transition between equilibrium and non-equilibrium states. How and why that happens is a long-standing question.
When in doubt, code
The issue is not one of physics but solving difficult equations. For instance, if you want to understand a star from a mathematical point of view, a few lines of equations solve the problem. But turning those equations into predictive models is difficult.
The common approach is to simplify: stars are spherical, so maybe we only need to care about their depth. By reducing the equations from 3D to 1D—a line from the star’s core to its surface—the math becomes simpler, and we can generate predictions. Unfortunately, things like turbulence (convection) require a full 3D model. And to calculate results with a 3D model of a star, you need some serious computational power.
How much, you ask? Try 60 million CPU hours on an 8.59 petaflop supercomputer. That was enough to generate three datasets for three different brightness and temperature combinations. The values were chosen to allow two stars to be in equilibrium and one to be out of equilibrium. Their evolution was tracked for between 15 and 35 days (at least that is how much data was shown in the paper).
The helium trap
It turns out that convection and shockwaves do play an important role. When the star is cooler, the density of helium and iron is high. These materials trap radiation in the star, heating it and causing the outer envelope of the star to expand. As the envelope expands, the speed of sound slows. The density waves that are driving the expansion suddenly find themselves going supersonic, which generates a shock wave. The shock wave drives off material at a relatively high rate (the peak is about five percent of the mass of the Sun per year).
The expansion, combined with the loss of material, reduces the optical density of the star. Light begins to escape at a higher rate and the radiation pressure reduces again, allowing the star to collapse back to a cooler state. As this is happening, the heating process has already begun again. The vast distances mean that the density waves that drive these changes take about 10 years to reach the surface.
Strong fluctuations in temperature seem to require that helium absorb a lot of light. Iron helps, but, without helium, the star will be much more stable. The 3D model also highlighted the importance of timescales. The main variations that we have already observed (and that are predicted by the model) seem to be about what we would expect, given the time it takes for the opacity of helium and iron to vary.
However, the envelope has its own timescale. Convection waves and density waves take about a week to travel from the inner part of the envelope to the outer part of the envelope. The researchers’ model predicts that these waves, which mainly depend on the opacity of iron, will lead to temperature fluctuations on the order of a week as well.
There are some observations that seem to confirm this. However, it looks like some telescope time will be needed to observe these short-term changes properly and see how they fit with the model.
It just goes to show what some serious computing power will let you achieve.