8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
GL2008
8th
IIF/IIR Gustav Lorentzen Conference on Natural Working Fluids
7-10 September 2008
Copenhagen, Denmark
Paper number: xxx
CO2 COOLING FOR THE LHCB-VELO EXPERIMENT AT CERN.
B. Verlaat*, A. Van Lysebetten, and M. Van Beuzekom.
*Author for correspondence
National institute for subatomic physics (Nikhef),
Kruislaan 409
1098 SJ Amsterdam,
The Netherlands,
E-mail: [email protected]
ABSTRACT CO2 as evaporative coolant has gained interest for the use in high energy particle physics
experiments. It is applied as coolant in the Vertex Locator (Velo) of the LHCb detector, which is
an experiment at the Large Hadron Collider at CERN in Geneva.
Silicon particle detectors like the Velo have special requirements on cooling. The sensors which
are spread over a large volume must be kept at a stable cold temperature (-7 ºC) at all times while
the attached electronics generate a substantial amount of waste heat which has to be taken by the
cooling system. The cooling infrastructure in the detector needs to be of low mass and the
construction materials including the coolant need to be radiation resistant. CO2 as coolant is a
good option for this application as it can withstand a large amount of radiation and has excellent
thermal behavior in small diameter tubes.
The CO2 cooling system for the Velo uses the 2-Phase Accumulator Controlled Loop (2PACL)
method. This method supplies low quality CO2 into the evaporator at a constant pressure and
requires no active components inside the detector.
This paper describes the design of the Velo detector and the Velo Thermal Control System
VTCS). It explains the 2PACL method and test results of the VTCS are presented and experience
gained during VTCS commissioning are described.
1. LHCB AND THE VERTEX LOCATOR.
A novel CO2 cooling system is developed for cooling the Vertex Locator (VELO) of the LHCb
experiment. LHCb is a particle detector studying Cp violation which must give the answer to the
question why the universe exist only out of matter, and why anti matter has disappeared. The
LHCb detector (LHCb Coll., 2003) is one of the 4 new build particle detectors constructed
around the collision points of the new Large Hadron Collider (LHC) at CERN in Geneva.
8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
1.1 The Vertex Locator
The VELO (LHCb Coll., 2001) is the sub detector closest to the collision point in LHCb. It
exists of 21 double and 2 single silicon wafer layers which are situated approximately 1 cm away
from the LHC proton beam. The silicon wafers are mounted on a module containing the read out
electronics and mechanical support. The silicon modules are situated in a vacuum which is
separated only by a 0.3mm foil from the LHC beam vacuum. The maximum pressure difference
between the 2 vacuums is 5 mbar, which results in a very complicated system for pumping down
and inflating the vacuum systems simultaneously. The reason for a secondary vacuum for the
silicon modules is the out gassing of the silicon modules, and the ability of installing detector
hardware without exposing the LHC beam vacuum volume to the outside air. The 0.3 mm
aluminum foil around the silicon stations acts also as a faraday cage protecting the silicon and
electronics from the electromagnetic interference of the proton beam.
The silicon stations suffer from a high dose of ionizing radiation induced by the LHC proton
beam. The radiation causes damage to the silicon crystal structure resulting in an increase of
leakage current (The ROSE coll. 1999). Permanent cooling of the sensors is needed to avoid the
outcome of this damage. A silicon temperature less than -7ºC is sufficient to minimize the effects
of radiation damage.
1.2 Velo module design. The silicon modules consist of a carbon fiber TPG
1 laminate at which on both sides an electronics
hybrid with a silicon sensor are glued. At the bottom of this laminate the CO2 cooling evaporator
is mounted on one side and the carbon fiber support paddle on the other side. The paddles are
mounted on a stiff aluminum base frame. Figure 3 show a picture of a constructed Velo half with
the discussed items clearly visible. The beetle read-out chips are located on the hybrid at the edge
of the silicon and generate a substantial amount of waste heat which needs to be taken away by
the CO2 cooling system. The aluminum base is the positional reference of the modules and must
therefore be maintained on room temperature. This is achieved by heaters since the thermal
connection via the paddles will otherwise cause the base to cool down by the CO2 evaporators.
Table 1 and 2 show the temperature requirements and the heat dissipation of the Velo
components.
1. Thermal Pyrolytic Graphite, High conductive material λx,y=1800
W/mK , λz=10
W/mK
Fig 2: Schematic side cut of the VELO
experiment (beam vacuum white
/ detector vacuum yellow)
Figure 1: Artists impression of the
VELO experiment the
Secondary
Vacuum
LHC Beam
Vacuum
Silicon
8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
Table 1: Operational and survival temperature limits of VELO key components
Survival temperatures: Operational
temperatures: Before irradiation: After irradiation:
Silicon Wafers -12°C / -5°C
(Nominal tip = -7°C )
-30°C / 100° Long-term: -30°C / 0°C
Short term: 0°C / 24°C
(100 hours)
Hybrid -20°C / 80°C -40°C / 50°C
RF-Foil <20°C NA
Module base frame 18°C -22°C (Stable) NA
Table 2: Power dissipation of VELO hardware inside the detector vacuum
Per VELO
module
Per PU
module
RF-Foil Module
Base
Total for
VELO
Nominal power 21.8 W 13.3 W 8 W 70W 1048 W
Maximum power 27.5 W 16.2 W 8 W 139W 1394 W
1.2 CO2 evaporator design. Each silicon module is connected to a
dedicated parallel evaporator branch.
One branch consists of a 1 meter long
stainless steel capillary of 1.5x0.25mm,
which is embedded in an aluminum
plate connected to the carbon fiber/ TPG
laminate. The aluminum plate is casted
around the tube in a special developed
procedure by melting the aluminum
around the tube in a vacuum oven.
Melting the aluminum under vacuum
will cause the aluminum to join to the
stainless steel forming inter metallic
phases. The melting procedure will give
a lot of freedom in pipe geometry inside
the aluminum and a perfect thermal
contact between the pipe and the aluminum plate. The cooling block shape is achieved by the
shape of the casting mould. The way the casting is done looks similar to baking of cookies, this is
the reason why the aluminum blocks are called cooling cookies. Each evaporator branch has a 1.3
meter restriction capillary of 1x0.2 mm at the inlet for a good flow distribution over all the
evaporator branches. The presence of the evaporator in a vacuum system gives high constrains to
the leak tightness therefore the complete evaporator assembly (figure 4) is made of stainless steel
tubes all joined together with vacuum brazing or orbital welding. No connectors are present
inside the vacuum system. The inlet manifold connected to the inlet capillaries is outside the
vacuum vessel and is accessible. This way it is possible to connect cooling to individual channels,
a feature which is used by module commissioning. During laboratory commissioning the cooling
was achieved by a CO2 bottle blow system.
Figure 3: Assembled Velo half with silicon
modules, cooling, and module base
8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
2. VELO THERMAL CONTROL SYSTEM
The Velo Thermal Control System
(VTCS) (Van Beuzekom, 2007) is
a cascade system of three
hydraulic systems. A chiller
condenses the evaporated CO2
generated in the Velo module
evaporators back to liquid and
reject the waste heat to the cold
water system of CERN. This
chiller is a commercial type
chiller with a gas compressor,
water condenser, evaporators and
expansion valves. The water
system is called the primary
cooling system, the chiller the
secondary and the CO2 loop the
tertiary cooling system. Figure 6
shows a schematic block scheme
representation of the cascade systems, with the main heat flows and the system temperature
distribution.
The tertiary cooling system is a hybrid 2-phase / single phase mechanically pumped loop using
CO2 as working fluid. It can operate in a single phase or in a two-phase cooling mode. During
start-up it operates in single phase mode, once cooled down it can be set to a 2-phase loop for a
more efficient heat transfer and accurate temperature control.
Figure 4: VTCS CO2 evaporator assembly Fig 5: Velo Module with alumi-
num cooling cookies.
-40°C -20°C 0°C 20°C
Tertiary
system
(CO2)
Secondary
system
(R404a)
Primary
system
(Water)
Work
(Pump)
Detector
waste heat
Environmental
heat leak
Work &
heat
(Compressor)
Environmental
heat leak
Qt1
Qt2
Qt3
Qs2
Qs3
Qs1=Qt1+Qt2+Qt3+Qt4
Control
heat
Qt4
Qp1=Qs1+Qs2+Qs3+Qs4
Control
heat
Qs4
-40°C -20°C 0°C 20°C
Tertiary
system
(CO2)
Secondary
system
(R404a)
Primary
system
(Water)
Work
(Pump)
Detector
waste heat
Environmental
heat leak
Work &
heat
(Compressor)
Environmental
heat leak
Qt1
Qt2
Qt3
Qs2
Qs3
Qs1=Qt1+Qt2+Qt3+Qt4
Control
heat
Qt4
Qp1=Qs1+Qs2+Qs3+Qs4
Control
heat
Qs4
Figure 6: Block scheme and heat balance
representation of the VTCS
8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
The main unit of the VTCS is
placed 50m away from the Velo
detector behind a thick concrete
shielding wall. This wall is shielding
the system from the radiation of the
LHC. Behind this shielding wall it is
also possible for people to work.
The VTCS is designed such that all
active hardware is located in this
safe zone. The cooling hardware in
the experimental area consists only
of tubes and passive devices such as
restrictors and check valves. Local
sensors (pressure and temperature)
are only for monitoring and are not
important for the cooling system
operation. Malfunctioning of
inaccessible vital hardware is so
reduced to an absolute minimum.
The CO2 evaporators in the detector are connected to the VTCS via 50 meter long concentric
transfer lines. The liquid feed transfer tube (1/4”x0.035”) is situated inside the vapor return tube
(16mx1mm). The concentric construction acts as a long counter flow heat exchanger needed for
conditioning the evaporator inlet flow to a low vapor quality.
The flow in the evaporator must be of low vapor quality to achieve a stable liquid expansion in
the flow distribution capillaries. The cooling system must maintain the silicon wafers cold,
powered or unpowered. The evaporator outlet vapor quality is therefore variable. The evaporator
temperature to achieve the silicon temperature requirement is between -25ºC and 30°C (Verlaat,
2005). This temperature can be set in the system by the accumulators and is called the VTCS’ set-
point temperature.
2.1 The 2PACL principle. To achieve liquid expansion over the capillaries and a low vapor quality together with a remote
controlled pressure in the evaporators the 2-Phase Accumulator Controlled Loop (2PACL)
principle is developed at NIKHEF (Verlaat, 2007). This principle is controlling the system
pressure and hence the
evaporator pressure using
a 2-phase accumulator.
This vessel is parallel
mounted to the system and
contains per design always
a content of liquid and
vapor. The guaranteed
presence of a saturated
mixture in the accumulator
makes the system pressure
to be a function of the
accumulator temperature. In this way the system pressure can be tuned independent from the
freon chiller temperature cooling the CO2 condenser.
Co
nd
en
se
r
PumpHeat exchanger
Flooded evaporator
Restrictor
2-Phase Accumulator
He
at
in
He
at
in
He
at
ou
t
He
at
ou
t
1
3 59
1013
Co
nd
en
se
r
PumpHeat exchanger
Flooded evaporator
Restrictor
2-Phase Accumulator
He
at
in
He
at
in
He
at
ou
t
He
at
ou
t
1
3 59
1013
Figure 8: Simplified representation of the 2PACL method
2-phase
gas
R404a
chiller22
33
6677
11
88
44
2-phase2-phase
liquid liquidliquid
2-phaseCondenser Evaporators
Concentric tubePump
Restriction
AccumulatorCooling plant area
Transfer lines(~50m) VELO area
55
liquid
Evaporator :• VTCS temperature ≈ -25ºC
• Evaporator load ≈ 0-1600 Watt• Complete passive
Cooling plant:• Sub cooled liquid CO2 pumping• CO2 condensing to a R507a
chiller
• CO2 loop pressure control
using a 2-phase accumulator
Accessible and a friendly
environment
Inaccessible and a
hostile environment
R507a
Chiller
2-phase
gas
R404a
chiller22
33
6677
11
88
44
2-phase2-phase
liquid liquidliquid
2-phaseCondenser Evaporators
Concentric tubePump
Restriction
AccumulatorCooling plant area
Transfer lines(~50m) VELO area
55
liquid
Evaporator :• VTCS temperature ≈ -25ºC
• Evaporator load ≈ 0-1600 Watt• Complete passive
Cooling plant:• Sub cooled liquid CO2 pumping• CO2 condensing to a R507a
chiller
• CO2 loop pressure control
using a 2-phase accumulator
Accessible and a friendly
environment
Inaccessible and a
hostile environment
R507a
Chiller
Figure 7: Block scheme and heat balance
representation of the VTCS
8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
The 2PACL system works as long as the freon chiller is colder than the accumulator temperature.
The freon chiller condenses and sub-cools the CO2 so the pumps can operate free from cavitation.
The remaining sub cooling after the pump is heated up by the concentric transfer line such that
the evaporator inlet remains in saturation. The operating condition of the evaporator is
independent of the amount of pump sub cooling. This sub cooling may vary in a wide range so a
controlled condenser temperature is not needed. The accumulator temperature is the only accurate
controlled item in the CO2 loop.
2.2 VTCS construction The VTCS chiller and CO2 loop are installed in racks on one of the LHCb service platforms. The
CO2 loop is constructed from orbital weld components from Swagelok and Cajon VCR
connectors. The CO2 pump used is a Lewa CO2 membrane pump. The accumulators are self
engineered stainless steel vessels
containing cooling spirals and a
thermo siphon liquid heater. There are
2 identical CO2 loops, each cooling 1
Velo detector half. Both CO2 loops
are cooled by 1 water cooled chiller
for normal operation and an air-
cooled back-up chiller for
redundancy. The chillers are self
engineered R507a chillers using
standard refrigeration hardware such
as Bitzer compressors, Danfoss line
components and SWEP plate heat
exchangers. The CO2 condensers are
also SWEP plate heat exchangers
reinforced for the high CO2 system
pressure. The design pressure for the
CO2 loop is 135 bar, the test pressure
170bar.
-450 -400 -350 -300 -250 -200 -1505x102
103
104
2x104
h [kJ/kg]
P [kP
a]
-40°C
-30°C
-20°C
-10°C
0°C
10°C
0.2 0.4 0.6
Tertiary VTCS in P-H diagram
1
23
4
5
67
Accumulator pressure = detector temperature
Internal heat exchanger brings evaporator pre-expansion per definition right above saturation
(3-5)=-(10-13)Satu
ration lin
e
Capillary expansion brings evaporator in saturation
Detector load (9-10)
1013
9
8
53
1
Pump is sub cooled
-450 -400 -350 -300 -250 -200 -1505x102
103
104
2x104
h [kJ/kg]
P [kP
a]
-40°C
-30°C
-20°C
-10°C
0°C
10°C
0.2 0.4 0.6
Tertiary VTCS in P-H diagram
1
23
4
5
67
-450 -400 -350 -300 -250 -200 -1505x102
103
104
2x104
h [kJ/kg]
P [kP
a]
-40°C
-30°C
-20°C
-10°C
0°C
10°C
0.2 0.4 0.6
Tertiary VTCS in P-H diagram
1
23
4
5
67
Accumulator pressure = detector temperature
Internal heat exchanger brings evaporator pre-expansion per definition right above saturation
(3-5)=-(10-13)Satu
ration lin
e
Capillary expansion brings evaporator in saturation
Detector load (9-10)
1013
9
8
53
1
Pump is sub cooled
Figure 9: VTCS operation in the Pressure-Enthalpy diagram for CO2
Tertiary CO2 unit
Secondary freon unit
Controls PLC
Accumulators
CO2 pumps
Primary water
supply
CO2 transfer line
Tertiary CO2 unit
Secondary freon unit
Controls PLC
Accumulators
CO2 pumps
Primary water
supply
CO2 transfer line
Fig. 10: The VTCS cooling plant installed at LHCb
8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
2.2 VTCS installation commissioning and testing The VTCS installation started at the LHCb
experiment in July 2007. Commissioning
work is ongoing but the system was
successfully operated for the detector
commissioning in March and May 2008.
The VTCS is operated during the vacuum
period at the end of March 2008 at several
set point temperatures varying from 0ºC to
-30ºC. In this period it was not yet possible
to switch on the detector to induce load on
the system. Figure 11 and 12 show the
temperature results of this vacuum
commissioning period. The first half of the
detector was commissioned in May 2008
with a cooling temperature of +10ºC.
During this commissioning the vacuum
system was under atmospheric pressure. To
avoid too much pressure difference
between the 2 vacuum systems the cooling
temperature was limited to +10ºC. Detector
commissioning with cooling under vacuum
is foreseen in July 2008.
Due to the absence of the detector the VTCS is tested using dummy by pass heaters. These
heaters are mounted on top of the Velo parallel to the evaporator and are a good representation of
the detector heat load. Figure 13 shows the response of the VTCS to a nominal heat load change
from 0 to 600 Watt. The system pressure is raised and compensated by cooling the accumulator.
10 15 20 25 30 35 40 45 50 55 60 65 70-50
-45
-40
-35
-30
-25
-20
-15
-10
-5
0
5
10
15
20
25
30
Time from 29 March clock time (Hours)
Mea
su
red
Tem
pe
ratu
re (
ºC)
Pump Outlet Temperature (TRTT120pvss)
C-Side RF-Foil temperature (TRTT043pvss)
C-Side Module Base Temperature (TRTT038pvss)
Sp=-10ºC
Sp=-20ºC
Sp=-30ºC
Sp=0ºC
Sp=-25ºC
Sp=-30ºC
Sp=0ºC
2 Days + 6 hours cooling under vacuum at several set point temperatures
Start-up
10 15 20 25 30 35 40 45 50 55 60 65 70-50
-45
-40
-35
-30
-25
-20
-15
-10
-5
0
5
10
15
20
25
30
Time from 29 March clock time (Hours)
Mea
su
red
Tem
pe
ratu
re (
ºC)
Pump Outlet Temperature (TRTT120pvss)
C-Side RF-Foil temperature (TRTT043pvss)
C-Side Module Base Temperature (TRTT038pvss)
Sp=-10ºC
Sp=-20ºC
Sp=-30ºC
Sp=0ºC
Sp=-25ºC
Sp=-30ºC
Sp=0ºC
2 Days + 6 hours cooling under vacuum at several set point temperatures
Start-up
Sp=-10ºC
Sp=-20ºC
Sp=-30ºC
Sp=0ºC
Sp=-25ºC
Sp=-30ºC
Sp=0ºC
2 Days + 6 hours cooling under vacuum at several set point temperatures
Start-up Condenser Out- / Pump Inlet Temperature (TRTT112pvss)
Pump Outlet Temperature (TRTT120pvss)
C-Side RF-Foil temperature (TRTT043pvss)
Condenser Out- / Pump Inlet Temperature (TRTT112pvss)
Pump Outlet Temperature (TRTT120pvss)
C-Side RF-Foil temperature (TRTT043pvss)
Condenser outlet / Pump sub cooling
Pump outlet
RF-foil
Accumultor saturation temperature (TRPT102tsat) = Setpoint
TR Evaporator Temperature (TRTT048pvss)
Condenser Inlet (TRTT101pvss)
Accumultor saturation temperature (TRPT102tsat) = Setpoint
TR Evaporator Temperature (TRTT048pvss)
Condenser Inlet (TRTT101pvss)
Accumulator saturation (Set-point)
Evaporator temperature
Condenser inlet
Figure 12: Transient temperature results of the commissioning under vacuum.
-30 -25 -20 -15 -10 -5 0-50
-40
-30
-20
-10
0
10
Setpoint Temperature (ºC)
Measure
d T
em
pera
ture
(ºC
)
TR Condenser Inlet (TRTT101pvss)
TR Pump subcooling (VRTT112pvss)
TR Pump Outlet (TRTT115pvss)
TR Evaporator Inlet (TRTT047pvss)
TR Evaporator outlet (TRTT048pvss)
TR Accu Saturation (TRPT102tsat) = Setpoint control
Condenser inlet Pump sub cooling Pump outlet Evaporator inlet Evaporator outlet Accumulator saturation
Figure 11: Steady state temperature results of
the commissioning under vacuum.
8th IIR Gustav Lorentzen Conference on Natural Working Fluids, Copenhagen, 2008
-40.00
-38.00
-36.00
-34.00
-32.00
-30.00
-28.00
-26.00
-24.00
-22.00
-20.00
1:40:48 PM 1:48:00 PM 1:55:12 PM 2:02:24 PM 2:09:36 PM 2:16:48 PM 2:24:00 PM
Time (hh:mm:ss)
Te
mp
era
ture
('C
)
-1000.00
-500.00
0.00
500.00
1000.00
1500.00
Po
we
r (W
att
), L
ev
el
(‰)
Evaporator Temp (ºC)
Accu Temp ≈ Set-point (ºC)
Detector Power (Watt)
600 Watt
Accu level (‰)
Accu Cooling Power (Watt)
Pumped Liquid Temp (ºC)
-40.00
-38.00
-36.00
-34.00
-32.00
-30.00
-28.00
-26.00
-24.00
-22.00
-20.00
1:40:48 PM 1:48:00 PM 1:55:12 PM 2:02:24 PM 2:09:36 PM 2:16:48 PM 2:24:00 PM
Time (hh:mm:ss)
Te
mp
era
ture
('C
)
-1000.00
-500.00
0.00
500.00
1000.00
1500.00
Po
we
r (W
att
), L
ev
el
(‰)
Evaporator Temp (ºC)
Accu Temp ≈ Set-point (ºC)
Detector Power (Watt)
600 Watt
Accu level (‰)
Accu Cooling Power (Watt)
Pumped Liquid Temp (ºC)
Figure 13: VTCS response to a load step
The accumulator level has
increased and the
evaporator pressure is back
to normal. The pump sub
cooling is increased due to
increase of the freon
chillers’ compressor
suction pressure.
The VTCS was tested
under several heat loads
and set point temperatures.
At the nominal set point of
-25ºC the system was able
to remove 1600W. The
lowest achieved set point
was -40ºC. These numbers
are far beyond the systems
requirement, so dummy
load tests have shown that
the VTCS is able to meet
the requirements of table 1
and 2.
3. CONCLUSION AND DISCUSSION
The VTCS has demonstrated that the 2PACL method with CO2 is a good principle of controlling
the temperature of the Velo silicon modules. The benefits from this system such as small cooling
channels and no active components inside the detector has been notified by other particle
detectors as a possible future cooling system. Both the 2 large CERN experiments Atlas and CMS
are considering a similar system for their upgrade inner detectors. The challenge will be to
upgrade the discussed principle from a 2.5 kW system towards a 100 kW system.
4. REFERENCES
[1] LHCb Coll. 2003, LHCb Reoptimized Detector design and performance, CERN/LHCC
2003-30
[2] LHCb Coll. 2001, The LHCb VELO technical design report, CERN/LHCC 2001-0011
[3] The ROSE Coll 1999, R&D On Silicon for future Experiments, CERN/LHCC note, 2000-
009
[4] B. Verlaat 2005, Thermal performance testing of the VTCS evaporator and VELO
module, NIKHEF EN05-01.
[5] B. Verlaat 2007, Controlling a 2-phase CO2 loop using a 2-phase accumulator,
International Conference of Refrigeration 2007, Beijing, ICR07-B2-1565
[6] M. Van Beuzekom, A. Van Lysebetten, B. Verlaat 2007, CO2 cooling experience
(LHCb), The 16th International Workshop on Vertex detectors, Lake Placid, NY, USA,
PoS 009