REFLECTIONS ON CO2 COOLING FOR THE CMS PIXEL DETECTOR
CO2 cooling such as for instance implemented in the LHCb VELO detector could be an interesting solution for CMS Pixel cooling.
LHCB has chosen CO2, with the following arguments:
–Radiation hard
–Has excellent thermodynamic properties for micro-channels.
•Low dT/dP
•Low mass
•Low liquid/vapour density ratio
•Low viscosity
•High latent heat
•High heat transfer coefficient
Some or all of these advantages might also apply for CMS Pixel cooling.
Boundary conditions:
For an upgrade of the CMS Tracker, it is very important to re-use the existing cabling and pipe-work since re-installing tracker and EB servicesseems like a virtual impossible task. For the Pixel upgrade it is notentirely excluded to modify some pipe-work but there is a major advantagein using the existing system. By designing a system capable of using theexisting pipes, it can also serve as a prototype for the CMS upgrade.
CO2 has a critical temperature of 31 degrees C and a critical pressure of
73 bar.
Below this point, the following two-phase conditions can be found:
Temp. 20 C, 57 bar
Temp. 24 C, 63 bar
Temp. 27 C, 67 bar
It is possible that the installation in the cavern could reach such temperatures when the system is not running, thus requiring pipe-work that can safely operatewith these pressures.
The installed cooling pipes are of soft copper material. The CERN TS-MME group has tested this material and found an elastic limit at 106 MPa. This corresponds to a pressure of 150 bar at the elastic limit. So in principle there is enough margin for the required safety factors.
On the other hand, it is possible to keep system pressures much lower by using relatively simple means.
The chilled water system that is available in CMS produces water at 5 C.
The CO2 boiling point of 5 C corresponds to a pressure of 40 bar.
When the cooling system is not operating, simply cooling the main storage tank with chilled water will lead to a stable pressure of 40 bar,even when other parts of the system are warm.
The detector cooling tube:
The preferred cooling liquid temperature during operation is around -12 C, which corresponds to a pressure of 25 bar.
The preferred Pixel Barrel cooling tube is specified as ID=1.5 mm, wall thickness 50 microns
These dimensions can most likely be produced but have not yet been found as stockmaterial.
We found at eagletube.com the dimensions OD=.065" wall thickness 3 mill which converts to ID=1.496 and a wall thickness of 76 microns.
The material is stainless steel 304
Using a specific resistance of 0.7 ohm mm2/m for SST 304 we get a resistance of
1.86 ohm/m
Layer 1 of the pixel detector dissipates 144 W over a cooling pipe length of5.5 m, 26.2 W/m.
A current of 3.75 Amp in the pipe will dissipate this power, requiring 7 V/m,a total voltage 38 V over the 5.5 m length.
In order to avoid that the current passes in other parts of the system, there is an electrical break atone end of the test tube. This consists of a piece or Rilsan (polyamid 11) tube. According to the CERNstore catalogue, this tube has an operating range from -40 to +80 degrees C. The ID of 2.5 mm has a service pressure of 50 bar, ID 4 mm has 32 bar. The material is thus well suited for this purpose.
In order to avoid heat transfer to the environment, which greatly complicates the interpretation of the measurement data, the detector pipe can be locatedinside a vacuum chamber. Only temperature measurements along the tube are required since the pressure in a boiling liquid is very well indicatedby the temperature.
Heat transfer coefficient:
The total amount of heat distributed over a length of 5.5 m is 144 W.
The tube has an inner surface of 259 cm2 for a length of 5.5 m.
Q = 5560 W/m2
Bart Verlaat has measured heat transfer coefficients around 6000 W/m2K
Delta T between liquid and wall is thus about 1 degree C.
Delta T over the tube wall:
Assuming all heat is produced on the outer surface. (Conservative during this test, realistic in detector operation)
Thermal conductivity of stainless steel 316, 16 W/mK
Delta T: 0.026 C
Test system description:
The system consists of the following components listed in the order of the CO2 passing the components. (See the scheme CO2pixel_cooling.pdf)
- CO2 Bottle with plunger, 40 litre, CERN SCEM 60.04.15.050.7
- Swagelok design concentric tube heat exchanger, inner tube, details below
- RHEONIK RHM 015 Coriolis Mass Flowmeter data in "Data sheet RHM 015.pdf"
- Swagelok S-series metering valve, details below
- Capillary equivalent to distance PP1-TK (optional)
- 5.5 meter of detector tube with electric heating
- About 20 temperature sensors along the tube including readout system
- Vacuum chamber to isolate tube from environment
- Swagelok design heat concentric tube heat exchanger, outer tube
- Coiled tube in water bath with immersion heater, details below
- Pressure gauge, for verification of correlation with the measured temperatures
- Swagelok proportional relief valve SS-6R3A-MM, described in catalogue MS-01-141.PDF
- Vent to atmosphere
Heat exchanger:
Swagelok design using Swagelok components (see drawing high_pressure_heat_exchanger.pdf).
Primary tube for high pressure side OD=6 wall thickness 1 mm, stainless steel 316L, elastic limit 200 N/mm2 min.
Pressure at elastic limit: 1000 bar
Secondary side limited by Swagelok components: pressure rating 315 bar
Pressure rating for VCR fittings: 550 bar
Metering valve:
The Swagelok S-series metering valve SS-SVR4 is described in catalogue MS-01-142.pdf. The curve of the Flow coefficient (Cv) is given.
The Cv value gives the relation between pressure drop and flow using the following formula:
Q=Cv*sqrt(dP/rho)
Q=flow [gallons/min]
dP=pressure drop [psi]
rho=density [kg/liter]
In the test setup we expect a flow of 1 cc/sec and a pressure drop of about 40 bar (bottle at 65, detector at 25) resulting in a Cv of 0.00065 which corresponds to about 3 turns open for this valve.
The working pressure of this valve is 137 bar.
Coiled tube in water bath:
In order to avoid CO2 snow in the relief valve, it is necessary to warm the CO2 (mix of liquid and gas) to about 40 degrees C before entering the valve.
This can be achieved in simple way by immersing a coiled up tube in a water bath and keeping the water at a temperature of about 40 degrees using an electric heater.
Details:
Past experience shows that simple detail problems can cause system failures.
It seems clear from the beginning that CO2, as a gas or as a low viscosity liquid is not tolerant to systems with small leaks. From the beginning we should be conservative, use only fully adequate components and strictly follow the producer’s recommendations.
For example, Swagelok compression fittings should not be used for dismountable connections and should be installed strictly respecting the manufacturers recommendations,
For dismountable connections, VCR fittings are recommended. At NIKHEF, the experience with these fittings has been very good. A new stainless steel washer is required aftereach disconnection as well as the appropriate tightening procedure.
Primary goals:
By measuring the temperature along the detector tube we determine the pressure drop over its length. This can be compared to the various calculations.Following this we can find the optimal combination of length, diameter, temperature non-uniformity and material budget.
Questions and Answers:
Question: What is the reason for having the heat exchanger between the supply and return flow?
Answer:The supply from the bottle is warm CO2 at about 20 C. In the Enthalpy diagram one can see that if you expandto a pressure that corresponds to -20 C about 1/3 of the liquid evaporates in order to cool the liquid from+20 to -20 C. This leaves only 2/3 of the liquid to cool the detector tube, which then risks to run dry at the endand thus is not cooling properly. It is therefore better to arrive with already pre-cooled liquid in thedetector tube. In this setup, only 50% of the liquid is evaporated in the detector tube, the remaining liquid goesin the heat exchanger and evaporates there while pre-cooling the liquid. In theory, if one regulates the valveprecisely,only gas should come out of the heat exchanger. In practice, the plan is to have say 15-20% coming outof the heat exchanger, which will then be evaporated in the water bath heat exchanger.