Modelling and Simulating a PC with FloEFD (Part 2)

Modelling and Simulating a PC with FloEFD (Part 2)

Hello again,

Finally I was able to find some time for the project again. In the first part I showed you the CPU heat sink from Noctua and in this part I’ll show you how I gained the properties that I used for the simplified model of the heat sink. Instead of meshing the narrow space between the 36 metal sheets with a huge amount of cells in addition to the already lots of cells in the whole PC, I will use a porous media as a simplified model of it. Therfore I will need the properties of the porous media which is rather simple to do for some values but will need some calculations for others.

The first thing was to model a box around the heat sink to get an internal space for an internal analysis. Then I set up a project where I changed the computational domain to just a small cross-section perpendicular to the flow so I don’t have to create a lot of cells over the whole model, just a small sample frame is enough, and changed the sides to a symmetry or periodic boundary condition. Noctua was so kind to provide me for this blog Project with the fan curve which I will need later in the project to simulate the fan and also now to determine the flow rate that passes the porous media. Many thanks to Noctua again at this point! So I used just a small cross-section of 15×15mm and some area in front and behind the heat sink for an undisturbed flow. I applied the first volume flow rate to the inlet side of the box and the goals to receive mass flow rate and pressure drop between inlet and outlet by an equation goal. After meshing and adjusting the mesh settings to get an good mesh I cloned the project with results so I don’t have to mesh each cloned project again and save some time. This is very useful if you just change boundary condition values which doesn’t necessarily need re-meshing, except changing materials. So I cloned the project several times and with the help of the parameter editor I changed the volume flow rates so I have several values between the minimum and a little more than the maximum of the fan curve. With all projects set up I ran an batch run using one core (of an quad core computer) per project and with enough licenses you can run several projects simultaneously, in my case four simulations at the same time.

Preparing Recommendations

Now with the mass flow rate and the pressure drop I had the data I needed for the porous media definition from simulations, the others are just measured from the geometry. The porous media is defined by a porosity which basically is the ratio between the volume of the pores to volume of the whole block. So in my case the sheets are 0.5×15x58mm and I have fife of them in my little computational domain where the whole block that will become the porous media is 15×15x58mm so a ratio of 0.8333. Furthermore I set it to axisymmetrical since in the plane of the sheets (x and y-direction) the pressure loss will be the same but in z-direction I have the solid sheet metals so a infinite pressure loss. The reference length and area are needed to scale it to the later much larger size of the porous media. The reference length in radius (in-plane) direction has the 58mm and the area is 15×15mm, for the normal reference length I used 1mm and for the area 1×1mm so make it very little so the scaled values will get even larger and therefore relatively infinite. With values for the normal direction of 100bar for 0 and 100 kg/s volume flow rate so it is definitely very high and for all flow rates that will appear. It is also possible to use heat conductivity for porous medias as we will need it later in the project, otherwise it wouldn’t be a heat sink. The density of the porous matrix is therefore the ratio of the mass of all the sheet metals to the whole porous media volume. As material for the sheet metal I used Aluminum from our database also for the thermal properties in the porous media. The last property is the matrix fluid heat exchange definition which I left to zero for now since I don’t know it yet. We will find this out later when simulating the whole PC until then no heat exchange between solid and fluid is taking place in the porous media.

With the properties of the porous media I can now replace all the heat sink sheet metals with one block of the same size that represents the heat sink as porous media.

So that’s it for today and next time I’ll show you how I modelled the motherboard with chipsets and their heat sinks and PCI slots etc.

More Blog Posts

Preparing Recommendations