...I´m using Floworks to simulate natural convection
heat transferon a rectangular finned surface. The heat sink is made
fromaluminum, the fluid is air, i Know the initial temperature of
airand the initial temperature of the heat sink (same as the
airtemperature). I also know the heat fluxon the lower base of
theheat sink.As indicated by Floworks help topics I check the
boxconsidering gravitational effects an started the
transientsimulation.After the simulation I´m able to get the
heat transfercoefficients on the heat sink. This is the point! How
Floworkscansolve a heat transfer problem without knowing the heat
transfercoefficient?? I tried to do the same simulation on
otherFeasoftware and some of them can not start the simulation
without agiven heat transfer coefficient...anyone knows what´s
goingon?thanksfabio

hello

convection is actually conduction inside a fluid. But this fluid can be water, air or other, and can move, too, so that' s very complex.

Usually in FE thermal codes, we simplify this phenomemon by using a coefficient and a fluid temp. And flux = H S (Tsurface - Tfluid).

This is not the case with flow simulation : it is a CFD code that takes fluid properties and velocity of the fluid into account. So there is no need to define a coefficient because this is here more accurate.But you can still check what would have been a equivalent coefficient during your post processing : you know heat flux, you know fluid temp and surface temp, there is one unknown : heat transfer coef. That's why you can plot it.

Loïc

I am in the process of calibrating flowsimulation for a heatsink application similar to the original poster. I am using a three part assembly for testing, ans we have a real world test rig of same set up in the lab as a comparison during the callibrating period. I am modeling LED light arrays pasted to an alluminium based mcpcb, which in turn is pasted to a heatsink. I am setting up the diode as the volumetric heat source inside the LED, and trying to track heat transfer through three boundaries (diode -biscuit, biscuit-mcpcb, and mcpcb-heatsink). I am then setting point goals for fluid temperature (I am measuring air temp at the same point on our test rig with a thermo couple). I have calculated the thermal conductivity for the biscuit and mcpcb (i think) and am using those values. I have set no boundary conditions ...... I am new to solidworks and flowsim .... so I rely on the thermal conductivity values on each material.

My initial results are way below the temperature measured, so I assume that the heat is not flowing at the actual rates. How should I go about getting my simulated values to be in the ball park of my test rig? Do I need to add boundary conditions? If so, how do I determaine these and what types should I use?

Ben

You need to keep in mind, that heat transfer in this instance is by convection. Convection is dependant on the fluid and solid, flow rate, temperature, time, etc. It is very difficult to have accurate convection properties in regular thermal analysis on anything other than a flat plate.

FloWorks actually computes the true convection properties for the heat sink by modeling what happens over time over all regions of the surface and accounting for the different fluid velocities at different areas.

If you think about how do people come up with heat transfer coefficients, it's typically through experimentation (when you get beyond simple geometry). It is done at a global level based on a generic fluid condition (stationary or moving at a specified speed) and then observiing the temperature change as a function of time in the part and air. FloWorks does the same thing, but you can look for local hotspots or cold spots on the heat sink to improve the design.

This is the significant benefit of using FloWorks on thermal problems compared to COSMOSWorks. When you need to look at heat transfer in complex geometry with convection, FloWorks is an excellent tool to use.

Hope this helps,

Ian