I want to model a ventilated room (3m by 2m by 1.5m)with inlet and outlet. Inside the room there are 36 temperature sensors which are hanged on a thin iron stick. Do you think it is necessary to model all the thin stickes in the room so that I can get good validation result? I use CFX4.3

Tadiwos

Thin? What's thin...?

Astrid

There are thin metal roads which supports the 36 thermocouples sensors. Is it necessary to model this roads in order to have a good validation result?

Tadiwos,

I think astrid means what are their dimensions wrt the rest of the geometry.

Room dimension 3m by 1.5m by 2m, inlet area 0.13sq.m, outlet area, 0.016sq.m. Thin wire diameter 1cm (used to hang thermocouples),Height of each thin wire, 1.6m Number of thinwires in the test room, 12

Thanks in advance for your help

Tadiwos

I really don't think it is worth it.

Tadiwos,

I agree with Herve but, if you are want to be sure and you have the time to check, you could run a test case with and without the rods. If there is little difference you know that you can fairly skip the rods. If not then you might be in trouble in creating a geometry.

Good luck, Astrid,

(1). There is this problem. If the local flow coming toward the sensor happen to be along the vertical thin rod direction, then the existence of the rod will modify the flow field. (2). Then there is another problem. The conductivity of the metal rod is much higher than that of the air. So, the existence of the thin rod will definitely modify the temperature field. That is, the temperature field around the thin rod will be more uniform because of the high conductivity. So, the measurement might be incorrect. (3). So, there is a wake effect, and the high conductivity effect. This can be especially important if the mass flow is low, and the flow field is sensitive to the room temperature distribution.

I agree with John, however, modelling such a small element (here a series of small geometry elements) might be difficult from a meshing point of view due to the small scale of the rod. Therefore You ought to run a simple model first and see how it goes in comparison with your exp. data prior to investing, if needed, in a more expensive simulation.

Doesn't it depend on what kind of accuracy you want?

I can't fell a few tenths of a degree difference. So if it is for climate control it doesn't matter whether T=22.3 C or 22.35 C I would think. It also depends what kind of temperature gradients you expect. Is one side of the room 100 C and the other 10 C? Or is it more like 30-20 C?

Dear Bart,

The working temperature range is about 20-35

And what is the resolution in T that you're interested in?

it is about 0.1 degre celcious