HEFAT2010 7 th International Conference on Heat Transfer, Fluid Mechanics and Thermodynamics 19-21 July 2010 Antalya, Turkey DESIGN, MANUFACTURE AND TEST RESULTS OF THE VTCS CO 2 EVAPORATOR FOR THE LHCB EXPERIMENT AT CERN. Verlaat B.*, Roeland E., Van Beuzekom M. and Van Lysebetten A. * National Institute for Subatomic Physics Mechanical Engineering, Science Park 105, 1098 XG Amsterdam, Netherlands, E-mail: bverlaat@nikhef.nl ABSTRACT The VELO Thermal Control System (VTCS) is a CO 2 based evaporative cooling system developed for the temperature control of the VErtex-LOcator (VELO) subdetector in the LHCb experiment at CERN. The CO 2 evaporator is located inside the VELO vacuum system and within the detector acceptance and radiation zone. This unique environment requires the evaporator to be leak tight, low mass, and resistant to ionizing radiation. To achieve this goal special manufacturing techniques were developed. The VTCS system uses the so-called 2-phase accumulator controlled loop (2PACL) method to control accuratly the evaporator pressure and inlet vapour quality at a distance. This way a passive but accurate evaporator was achieved, fulfilling the strict thermal requirements of the VELO. This paper describes the development trajectory of the evaporator. Special construction techniques will be discussed such as orbital welding and aluminium casting of cooling connections around the stainless steel tubes. Thermal test results of the evaporator such as heat transfer, pressure drop and flow distribution are being presented. INTRODUCTION The Vertex Locator (LHCb-VELO) is the central detector of the LHCb experiment [1]. LHCb is one of the four experiments of the newly constructed LHC proton collider at CERN in Geneva. The VELO detector is designed to make a precise measurements of vertex positions from high energy particle decays like B-mesons. The B-mesons are made by the protonproton collisions of the LHC collider. The requirements for precise detection of the B-meson vertices are the highest of all 4 LHC experiments. Therefore the silicon sensors have to be placed very close to the collision point. To achieve this the VELO detector is situated inside the LHC vacuum only 8 mm away from the proton beam. The active detection surface is a disc shaped silicon strip sensor with an outer diameter of 80mm and an inner diameter of 16mm. The inner diameter hole is for the proton beam to pass through. The detector is split in two halves. Both halves can be moved 30mm away from the beam. The increased beam space is necessary as the beam trajectory can vary during beam injection. Cooling in- and outlet Electrical cables Rectangular bellow Detector base frame Beam Cooling pipe Motion control RF-foil Silicon sensor Figure 1 Inside look of 1 VELO detector half Figure 1 shows an inside look of 1 detector half. The blue half circles are the detection silicon sensors. In total there are 21 VELO modules and 2 Pile-Up modules per half. The VELO modules are used to measure the particle tracks, the pile-up modules are used for triggering. All modules are assembled on an electronic substrate (green plate) housing the beetle [2] read-out chips (red dots around silicon periphery). The modules are mounted via stand-off paddles to a precise machined aluminium base frame. Each detector half is situated inside a 300 micron thick aluminium foil box. The foil acts as a Faraday
cage to protect the sensitive silicon from the beam interference. It is also a gas barrier to protect the ultra high accelerator vacuum from the out gassing of the silicon modules. A maximum pressure difference of 5 mbar may occur between the detector and the accelerator volumes. A complex vacuum system is designed to maintain a lower pressure difference at all times [3]. THE VELO THERMAL CONTROL SYSTEM The silicon sensors are exposed to radiation. Cooling of the sensors to sub zero temperatures is needed to minimize the effect of radiation damage. The beetle chips on the modules are also dissipating heat inside the vacuum, so an active cooling system is needed to remove the heat and keep the silicon sensors cold. The base frame is the positional reference of the modules, this frame must therefore stay at room temperature. The stand-off paddles have therefore a low thermal conductance. Additional heaters control the base frame to maintain at room temperature. This full thermal control of the detector (module cooling and base frame heating) is done by the VTCS (VELO Thermal Control System). The VTCS is an evaporative cooling system using CO 2 as coolant. The system uses the 2PACL method (2- Phase Accumulator Controlled Loop) for circulating and conditioning the CO 2 [4],[5]. The 2PACL method condenses and sub cools the CO 2 with an external chiller. The evaporator pressure is controlled by a 2-phase accumulator, the evaporator inlet enthalpy by a heat exchanger between in- and outlet. Due to the accumulator control, the evaporator has a constant temperature which is independent of the heat load and temperature of the chiller. The inlet quality of the evaporator is close to 0%, so the full range of evaporation can be used. Table 1 and 2 show the thermal requirements of the main VELO components. Table 1 VELO temperature requirements Detection Silicon CO 2 evaporator (Stainless steel tube casted in aluminum) Beetle chips Figure 2 Velo module with CO 2 cooling connection THE VTCS EVAPORATOR The CO 2 evaporator exist of 23 parallel branches of 1.5x0.25mm 1115mm long CRES321 stainless steel tubes. Each tube is coiled to prevent thermal expansion. A 457mm long section of the tube is embedded in casted aluminium cooling blocks for heat exchange. The cooling blocks are nicknamed cooling cookies as the production method of casting showed similarities with backing cookies in an oven. The thermal interface of the cookies to the modules is achieved with THERMFLOW T710 phase-change Thermal Interface Material (TIM). This TIM was chosen because of its low melting temperature and low clamping force. Figure 2 shows a picture of a cooling cookie assembly mounted to a module. Each Velo module consists of a double sided hybrid (printed circuit board) with a silicon sensor. The core of the hybrid is a thermally conductive Carbon fiber-tpg laminate supporting the module and conducting the heat to the cooling cookies. The cooling cookies are mounted with M2.5 screws to the module using the stiff paddle structure on the opposite side as contra clamp. Each evaporator branch has a PT100 sensor glued to the last cookie in flow direction to monitor possible dry-out phenomena of tube. Silicon Wafers Operational temperatures: -12 C / -5 C Maximum temperatures: -30 C / 0 C (+30 C short term) Hybrid -20 C / 80 C -30 C / 150 C Beetle chips -20 C / 80 C -30 C / 100 C RF-Foil <20 C NA Module base 18 C -22 C (Stable) NA Table 2 VELO power specifications Per VELO module Per PU module RF-Foil Total for VELO Nominal power 21.8 W 13.3 W 8 W 978 W Maximum power 27.5 W 16.2 W 8 W 1255 W Absolute maximum power* 30.2 W 17.5 W 8 W 1617 W *Absolute maximum power is the highest possible dissipated power of the Beetle chips in case of wrong settings. A power limiter will prevent higher dissipation. Figure 3 Layout of the VTCS evaporator Figure 3 shows the layout of the evaporator assembly. The whole assembly is situated inside the VELO vacuum vessel. The requirements for leak tightness is extremely high. All the wetted parts are made of stainless steel and only reliable joining techniques as orbital welding and vacuum brazing (Nicrobraz EL-36) are applied. The 1.5x0.25mm cooling tubes from the aluminium cooling cookies are connected via 4x0.7mm tubes to the support manifold. The support manifold connects the outlets
of all the branches together. A flexible braided metal hose connects the manifold to the vacuum feed through. The inlet of the branches are connected via 2038mm long 1x0.25mm capillaries directly to the vacuum feed through flange. The capillaries all have the same length so the flow impedance is the same for all. This ensures a good flow distribution for all the parallel channels. The capillaries are accessible from the outside, a feature which was used during the installation and commissioning of the VELO modules, where individual modules needed to be cooled. The capillaries are routed into a small channel on the outside of the support manifold. This channel also houses the PT100 cables. For beam injection the detector halves need to be retracted away from the beam by 30mm. The evaporator connection to the outside needs to be flexible to allow for this movement. The flexible return hose together with the inlet capillaries are inside a sealed vacuum bellow. The below is indicated as dotted line in figure 3. Figure 4 show a photo of the evaporator assembly. Liquid inlet and capillary Manifold Vapor return pipe Vacuum feed through Support manifold and capillary/cable channel Flexible vacuum bellow with return hose and capillaries 23 Cooling cookies Figure 4 VTCS Evaporator assembly aluminium was done under a 1 bar argon atmosphere to suppress the created magnesium vapour. The vacuum oven was heated in1½ hour to 700 C, 10 minutes later the argon was applied, and another ten minutes later the oven was switched off. Figure 5 shows the fresh backed cooling cookie prior to machining. Figure 5 Aluminium cookie still in the casting mould THERMAL VACUUM TESTING A single evaporator was tested on a dummy module using heaters instead of beetle chips. The tests were done in a vacuum vessel to simulate the VELO environment [6]. The evaporator was connected to a CO 2 test plant able to cool down to -30 C. The support paddle was mounted using the original interface to a aluminium block with a heater as a replacement for the base frame temperature control. The evaporative heat transfer and the thermal conduction of the evaporator and module was measured to verify the required cooling temperature, to meet the specifications of table 1. Figure 6 show the thermocouple (type-t) locations on the VELO module. From the measurements thermal conductances (C1 to C5) can be calculated. With these conductances a simplified thermal model (shown next to module) is made which can be used to predict the temperatures under different circumstances. ALUMINUM CASTING OF THE COOLING COOKIES The aluminium cooling blocks are casted around the cooling tube with a specially developed procedure. The evaporator must have a low mass near the modules, to minimize obstruction for the particle tracks. Aluminium has a low density and good conductive properties. Casting gives the freedom of applying material only at the place where it is needed. The thermal connection to the cooling pipe is achieved by the crimping of aluminium around it. The CTE differences between steel and aluminium clamps the aluminium around the pipe during cooling down. The shape of the cooling cookies is achieved by the stainless steel mould. The flat interface to the module is made by machining the excess aluminium away. Q BEETLE T HYBRID Cookie T SILICON C 3 T COOKIE C 2 C 4 C 1 T PADDLE T CO2 C 5 T BASE The melting of the aluminium is done in a vacuum oven. The aluminium flows perfectly in the mould under vacuum conditions (<10-3 mbar) creating a smooth finish. The view of the cookies in figure 2 is the result of the mould print. The best results were obtained with ACP 5080 (AlMg4,5Mn). The vacuum however causes the magnesium content to vaporize creating bubbles in the aluminium. Solidification of the Base Frame Q BASE Figure 6 VELO Module test set-up. Left the measurement setup, right the module in the vacuum chamber. Table 3 shows the calculated conductance values for different test cases. The average conductance is used to predict the temperatures for the required operational conditions of table
1 and 2. Table 4 show the results for the nominal and maximum power. The corresponding CO 2 temperature is calculated to meet the operational boundaries temperatures of the silicon. At the given CO 2 temperature the temperature of the silicon is also calculated in case the power to the electronics is off. This temperature must meet the survival requirements given in table 1. For the silicon this means a minimum temperature of -30ºC. For the nominal power the CO 2 temperature needs to be around -30ºC. For the maximum power setting both the operational and survival limits are exceeded. The solution cannot be found in the cooling, it is a limitation of the module design itself. The maximum power setting is however unlikely to occur during normal operation. Table 3 Measured evaporator and module conductance at a CO 2 temperature of -22ºC for several powers and flows. C 1 CO 2 HTC (W/m 2 K) C 2 Cookie IF C 3 Module Cond. C 4 Cond. C 5 Base IF Heat (Watt) Flow (g/s) 24 0.4 5326 2.98 1.92 0.03 0.17 24 0.3 5627 2.98 1.93 0.03 0.19 24 0.26 5564 2.95 1.93 0.03 0.19 24 0.21 5394 2.91 1.95 0.03 0.19 24 0.14 Dry-out 24 0.20 Dry-out 24 0.53 5685 2.96 1.92 0.03 0.2 30.6 0.41 6200 3.05 1.95 0.03 0.22 30.6 0.3 6259 3.03 1.95 0.03 0.21 30.6 0.2 5220 2.89 1.99 0.03 0.21 Average Conductance: 5659 9.46 W/K 2.97 1.94 0.03 0.20 Table 4 Calculated temperatures Nominal Power (21.8Watt) Maximum Power (27.5Watt) Silicon at Silicon at Silicon at Silicon at upper limit (-5 ºC) lower limit (-12 ºC) upper limit (-5 ºC) lower limit (-12 ºC) T CO2 (ºC) -26.3-33.4-31.8-38.9 T COOKIE (ºC) -23.9-31.0-28.8-35.9 T HYBRID (ºC) -16.3-23.3-19.2-26.2 T SILICON (ºC) (Unpowered) -25.8-32.8-31.2-38.2 OFF-LINE EVAPORATOR COMMISIONING The evaporator assembly was tested after the installation in the base frame. For testing heaters were screwed directly to the cooling cookies. The temperature of the cooling cookies was measured using the PT100 s on the last cookie. The scope of the test was to verify the heat transfer of all the branches and to test the flow distribution over the 23 parallel channels. Figure 7 show the measured pressure drop over the branches including capillaries. In parallel evaporators the pressure drop over upstream flow resistance must dominate the total pressure drop. The downstream resistance depends on the vapour quality which is a function of the applied load. The upstream resistance is constant. A good and continuous flow distribution over all the parallel branches is achieved if the total pressure drop is not influenced to much by the applied load. The pressure drop measurement in figure 7 shows only a slight increase of pressure drop between full power and no power. This proves that the capillaries are properly dimensioned for this distributed parallel branch evaporator. Figure 7 Pressure drop of the VTCS evaporator under at several heat loads Figure 8 show a heat transfer measurement as a function of the exit vapour quality of all the evaporator branches. The heat transfer coefficient (HTC) remains constant and stable up to a vapour quality of 0.4. Above 0.4 the HTC increases but also shows large fluctuations. This change is believed to be the nucleate to convective boiling transition. At a flow of about 200 kg/m 2 s (x 0.5), two branches are at the onset of dry-out, two more follow at around 150 kg/m 2 s. The dried out branches are those at the top of the inlet manifold. The pressure drop over the capillaries was not sufficient for a proper inlet sub cooling. Phase separation in the manifold caused the two upper branch connection to be fed with vapour. The vapour formation at the inlet is also visible in the pressure drop of figure 7. Below 5 g/s (236kg/m 2 s) a higher pressure drop with respect to the expected curve is seen. Based on the above observations it was decided to set the flow in the system to 10 g/s (471 kg/m 2 s) which is twice the flow at which dry-out occurs. Figure 8 Heat transfer coefficient measurements of the 23 parallel cooling branches as a function of vapour quality. The heat load is 25.5 Watt (17.8kW/m 2 ) and the CO 2 temperature -33.6ºC. The green line is the mass flux, the blue line the theoretical prediction according to the Kandlikar correlation.
EVAPORATOR COMMISIONING IN LHCB The modules were installed on the base frame at CERN in 2006. For the commissioning of the VELO modules a CO 2 blow system was built. The blow system was connected to each individual cooling channel, so a single evaporator branch could be cooled. The blow system flushes liquid CO 2 from a bottle, evaporates it at a set pressure and vents the residue to the atmosphere. The fully assembled detector halves were installed in the VELO vacuum vessel late 2007. Safety regulations do not allow vacuum in the system during detector construction. The detector was therefore filled with 850mbar ultra pure Neon. Cooling of the detector in neon affects the pressure difference between the detector and accelerator volume. At a cooling temperature of 8ºC the pressure difference was at the limit. Fortunately 8ºC was just enough to commission the detector under neon. The silicon temperatures were around 27ºC, while the interlocks were set at 30 ºC. Warm operation under neon was fine for the time being as the silicon sensors were not yet irradiated. Temperature ( C), Power (Watt), Level (vol %) 80 Temperature ('C), Liquid level (vol %), Power (%kw or Watt) 60 40 20 80 60 40 20 Evaporator Temperature ('C) Accumulator Liquid Level (vol %) Acumulator heating/cooling (% kw) Module VL11-C power (Watt) Module VL11-C cookie temperature ('C) Module VL11-C NTC00 temperature ('C) Module VL11-C NTC01 temperature ('C) Detector half heat load (x10) Module Heat load Accu Heating/Cooling Accu level 0 0 Silicon temperature -7 C SP=-5 C -20-20 Evaporator SP=-25 C temperature -40-40 16.5 0 0:30 17 17.5 18 18.5 Time (Hour) Clock 1:00 time (Hour) 1:30 2:00 1 Figure 9 VELO temperature response to a power-up, a cooling set point change and a power down. In March and June 2008 the VELO was under vacuum for a few days. In march the detector could not yet be powered up. Unpowered cooling tests were done to measure the thermal environment of the VELO detector. In June 2008 the detector was switched on for the first time in its final condition. Tests at two different cooling temperatures were carried out. Figure 9 show the response of the VELO temperatures to a power-up, a cooling set point change from -25ºC to -5 ºC and a power down. The measured temperatures of figure 9 have been verified using the thermal model of figure 6. A difference of only 1.5ºC is observed between the model and the measurement. Table 5 show the measurement and simulation results. Based on the measurements and the model a set point temperature in the cooling plant of -30ºC was decided as the default cooling temperature. The CO 2 temperature in the evaporator is approximately 2ºC higher due to the hydraulic resistance between the evaporator and the cooling plant. With a CO 2 temperature of -28ºC a silicon temperature of -10.6ºC is expected according the model. This corresponds with a plant set-point of -30 ºC. Table 5 Overview of the measured and calculated temperatures. Set Point (-25ºC) Set Point (-5ºC) Measured Simulated Measured Simulated Q BEETLE (W) 17.7 17.7 17.7 17.7 T CO2 (ºC) -23.3-23.3-4.0-4.0 T COOKIE (ºC) -20.4-21.3-2.0-2.1 T HYBRID (ºC) -13.2-15.0 6.0 4.0 T SILICON (ºC) -7.4-5.9 12.3 13.1 CONCLUSIONS A CO 2 evaporator was successfully designed, constructed and tested for the thermal control of the LHCb-VELO detector. Special techniques such as orbital welding and aluminium casting were applied. The evaporator with detector module was prototyped and final measurements show a good correspondence with prototype measurements. The design requirements were met with extra margin. The detector s default operational set-point temperature is to be -30ºC. REFERENCES [1] Augusto Alves Jr, A.,2008, The LHCb Detector at the LHC. JINST 3, S08005. [2] Lochner, S, 2006, Development, optimisation and characterisation of a radiation hard mixed-signal readout chip for LHCb, CERN-THESIS-2006-061 [3] Van Beuzekom M. et al, 2008, The LHCb VELO Vacuum System User manual, CERN EDMS 901535. [4] Verlaat B., 2007, Controlling a two-phase CO 2 loop using a two-phase accumulator, International Conference of Refrigeration 2007, Beijing, China, ICR07-B2-1565 [5] Verlaat B. et al, 2008, CO 2 cooling for the LHCb-VELO experiment AT CERN, 8th IIF/IIR Gustav Lorentzen Conference on Natural Working Fluids 2008, Copenhagen, Denmark, CDP 16-T3-08 [6] Verlaat, B, 2005, Thermal performance testing of the VTCS evaporator and VELO module, NIKHEF EN05-01