Liverpool CDF Home Page

Liverpool CDF Group

Latest Update on the Cooling

We repeated the cooling studies on the high T Fermi fibre using more probes which allow a fuller mapping of either end, as well as the z-dependence.

7 probes were placed 5cm from the input end. 7 probes were placed 5cm from the output end. A probe was placed on the input pipe and another on the output pipe. Finally an additional 3 probes were placed on the carbon fibre rail directly above the pipe. This meant that in total there were 5 probes equally spaced in z, along the rail directly over the pipe.

This plot shows the response of the 14 probes at either end. The horizontal axis is the delta T between the input coolant and the room temperature. The vertical axis is the delta T between the probe and the room temperature. The different colours refer to the position. Full lines are at the input end; dashed at the output. We note:

  • The input end cools better than the output end for the rails above the pipe, for the 1st nearest-neighbour, for the 2nd nearest-neighbour, and for the 3rd nearest-neighbour.
  • The side with less material cools better than the side with more material. This is most obvious for the 2nd nearest-neighbour, but can also be seen for the 1st nearest-neighbout.
  • It looks strange, but the 3rd nearest-neighbour seems to cool better than the 2nd nearest-neighbour beside it. Having talked with Peter, this is probably due to the way the fibres are laid down. The 3rd nearest-neighbour has 90 degree fibres connecting it to the inner cylinder, but the 2nd nearest-neighbour has 0 degree fibres and only cools via a small overlap to the 1st and 3rd neighbours. In the real situation this isn't a problem... there will of course be another pipe under this ledge.
  • The chisquared of all these fits are excellent, giving us confidence that the linear hypothesis is valid. It corresponds to a measurement error of 0.3 degrees, which is quite a bit smaller than some of the effects we can see. Thus we believe these effects are real and due to genuine characteristics of the fibre and its construction.
  • An independent check is provided by examining the extrapolation of all lines to (0,0) point. This point is reconstructed with an error of 1.5 degrees and a good chi-squared which gives further confidence in the linear hypthesis and in the understanding of our measurement errors.

    This plot shows the response of the 14 probes when we reduced the flow by 20%. Why 20% you might well ask ? Well the pressure drop was 50% but measuring the flow directly, it only decreased by 20%. You can compare this plot directly to the previous one and you see the same behaviour. The important number here, is that the difference between the output coolant temperature and the input coolant temperature is 6.0 +- .3 degrees, while for the 20% greater flow, the difference was 5.2 +-.3 degrees. Lousy stats... the water temperature is 0.8+-.5 degrees different, roughly 20+-10% different, when the flow rate is 20% greater. This is, at best, a consistency check which agrees with thermodynamics, but one we will repeat at a genuine 50% decrease.

    Here is the plot for the delta T as a function of z, for 5 different input coolant temperatures, indicated by the different colours. We note:

  • The falloff is basically linear. The further along the fibre structure, the less well things cool.
  • However, you can see there are outliers. The 2nd and 4th points in z, do not cool as effectively as the points 1, 3 and 5. The residuals to the straight line 'best fit' are much much bigger than our measurement uncertainties. Furthermore (if you ignore the green and red second point (I will come back to this)), all the offsets are in the same direction, and of a magnitude which scales according to the size of the delta T. So this is a real effect, and your suspicions that the fibre was not being cooled uniformly along its length are correct. More work is clearly required on effective greasing.

    This is the same plot for delta T as a function of z, but with the reduced pressure. The same behaviour is observed with points 2 and 4 being worse cooled. I return now to the fact that the green and red 2nd point appear to be well cooled, and agreeing with the best fit, while all other point 2s, are bad. We puzzled over this for a while because this should be too big to be a measurement uncertainty. It turns out that these two points were the first two taken, before we went home one evening. All the other measurements were taken the next day. So 'something happened': either the grease moved or the probe detached itself somewhat. This is all good information to have. The grease can move overnight... we must include some means to ensure its stability over not days, but years.

    That's all for now. We have obtained a thermal imaging camera... so the next update will hopefully be accompanied by pretty pictures.