Bart Verlaat

Dimensioning of CO2 cooling pipes in

detector structuresPipe dimensioning

&Flow distribution

Detector Mechanics Forum Oxford, 20 June 2013

Bart Verlaat 1

Thermal chain in detectors

• The design of the cooling is the whole chain between heat source and heat sink

2

Heatload

Silicon C-foam Pipe wallCO2 in tube ManifoldCF-sheet

Th. pa

ste

Glue

Glue HTC

ΔP

Typical example for IBL

Thermal chain in detectors

• The design of the cooling is the whole chain between heat source and heat sink

3

Heatload

Silicon C-foam Pipe wallCO2 in tube ManifoldCF-sheet

Th. pa

ste

Glue

Glue HTC

ΔP

Load variations give gradients w.r.t the common sink => Outlet manifold!

Typical example for IBL

0 5 10 15 20 25-2

0

2

4

6

8

10

12

14

16

IBL temperature and pressure profile. MF=0.8g/s, Tsp=-40ºC, Q=101.8, xend=0.41

Branch length (m)

Del

ta T

(`C

) & D

elta

P(B

ar)

1 2 3 4 56 78 9 10 11 12

Stave TFoM: 13ºC*cm2/W

Pixel maximum temperature:

-24.4ºC

dT Tube wall (ºC)dT Fluid (ºC)dP Fluid (Bar)dT Pixel Chip (ºC)

Design of cooling: From source to sink

• So the reference should not be at a pipe wall, nor at the liquid temperature as it is generally approached. – This is similar then taking a reference in the middle of the

structure.

4

Pressure drop ~20%

Heat transfer ~ 19%

Stave conductance ~61%

Loaded stave temperature: -24.4°C(0.72 W/cm2)

Unloaded stave temperature: -39°C

Atlas IBL example

Outlet Manifold = temperature reference

Inlet manifold

How to optimize the cooling as part of the whole thermal chain?

• The best thermal solution is not only related to small temperature gradients.– If so our detectors will be made of copper…– We have to find the balance between the important parameters

• Generally thermal gradients vs mass (rad. Length)• How can we find the optimum cooling tube dimension?

– Depends on the real criteria: • Lowest mass (Radiation length)?• Smallest pipe?• Minimum amount of pipes?

– We should not only look at the pipe but also at the structure around• A smaller cooling tube is replaced by other material, when embedded.

• Where do we have to look at:– Pressure drop and heat transfer

• As part of the thermal chain – Flow distribution

• For the proper fluid conditions 5

6

First we need to understand what happens inside a cooling tube?

Heating a flow from liquid to gas

Super heated vaporSub cooled liquid 2-phase liquid / vapor

Enthalpy (J / kg)

Pres

sure

(Ba

r)

Dry-out zoneTarget flow condition

Temp

erat

ure

(°C)

Low ΔT

- 300

306090

Liquid

2- phase

GasI sotherm

Increasing ΔT (Dry-out)

Liquid Superheating

Understanding detector evaporator tubes

• In a 2PACL the capillary inlet temperature is a function of the outlet saturation pressure.

• The detector inlet is close to saturation.– But can be liquid due to pressure drop– Usually ambient heating is enough to

overcome sub cooled entry state– Detectors with high dP have to be

designed to cope with liquid at the inlet • FE: pre heating by electronics (CMS pixel)

• Pressure drop of the evaporator tube and outlet tube is part of the thermal resistance chain from heat source to sink!

7

2m4m

Liquid

2-phase

Heat exchanger Inlet capillaries

Outlet tubes

Manifolds

Detector staves

Outlet manifold: Pressure = fixed

Inlet manifold:

Temperature = fixed

Gas

Pre

ssur

e

Isothermal line

Enthalpy

Liquid2-phase

(Evaporation)

Temperature exchange

Outlet line dPDetector dP

Inlet capillary dP

Inlet manifold

outlet manifold

Long branch thermal profile

2m4m

Liquid

2-phase


Outlet tubes

Manifolds

Detector staves

Outlet tubeInlet tube Evaporator tube

Offset of evaporator temperature due to outlet pressure drop

Temperature gradient inside detector due to pressure drop and heat transfer

Liquid 2-Phase

Manifold temperature = common reference of all branches

HTC

dP

Liquid entry into evaporator

Inlet: 2mm x 4m, Detector: 2mm x 4m, Outlet: 2mm x 4mHeatload on detector: 200 Watt

Flow distribution:Inlet tube reduction

9

50 W100 W

150 W

200 W

250 W

300 W

50 W100 W

150 W

200 W

250 W

300 W

1.16

g/s

1.61

g/s

1.16

g/s

3.75

g/s

1.15 bar

2.18 bar

2.52 bar

3.54 bar

Dry-out zone Dry-out zone

Figures from: DESIGN CONSIDERATIONS OF LONG LENGTH EVAPORATIVE CO2 COOLING LINESBart Verlaat and Joao Noite, GL-209, 10th IIF/IIR Gustav Lorentzen Conference on Natural Working Fluids, Delft, The Netherlands 2012

2m4m

Liquid

2-phase


Outlet tubes

Manifolds

Detector staves

Inlet: 2mm x 4mStave: 2mm x 4mOutlet: 2mm x 4m


Pumping energy is flow x dP. Adding capillaries can save pumping power (in example 1.61*3.54/2.18*3.75=0.70 => 30% saving)

Pressure drop and dry-out calculated using CoBra

Which flow do we need when 200 Watt to a single stave is applied?

Influence of the in and outlet-lines on thermal performance

10

2m4m

Liquid

2-phase


Outlet tubes

Manifolds

Detector staves

0 2 4 6 8 10 12-1

0

1

2

3

4

5

IBL temperature and pressure profile. MF=1.61g/s, Tsp=0ºC, Q=200, xend=0.53

Branch length (m)

Del

ta T

(̀C

) & D

elta

P(B

ar)

1 2 3 4

dT Tube wall (ºC)dT Fluid (ºC)dP Fluid (Bar)

0 2 4 6 8 10 12-1

0

1

2

3

4

5

IBL temperature and pressure profile. MF=3.75g/s, Tsp=0ºC, Q=200, xend=0.22

Branch length (m)

Del

ta T

(̀C

) & D

elta

P(B

ar)

1 2 3 4


0 2 4 6 8 10 12-1

0

1

2

3

4

52mmID x 5m tube temperature and pressure profile. MF=1.61g/s, Tsp=0ºC, Q=200, xend=0.54

Branch length (m)D

elta

T (`

C) &

Del

ta P

(Bar

)

1 2 3 4






Dealing with environmental heat pick-up

• Three important statements: – Expose return tube to ambient heating

• There is usually enough cooling power left– Connect as much as possible the inlet to the outlet

• Outlet boils first (lower P), so will take care of heat absorption– Avoid boiling before the inlet manifold

• Flow separation will feed some channels with vapor only!

11

Gas

Pre

ssur

e

Isothermal line

Enthalpy

Liquid2-phase

(Evaporation)

Pre capillary heat pickup

Inlet manifold

outlet manifold

Gas

Pre

ssur

e

Isothermal line

Enthalpy

Liquid2-phase

(Evaporation)

Manifold boiling

Outlet transfer line

outlet manifold

Outlet transfer line

Capillary dP makes manifold liquid

To keep this problem simple: Have the manifold right after the heat exchanging transfer line.

Remaining cooling power for ambient

0 5 10 15 20 25-2

0

2

4

6

8

10

12

14

16


Branch length (m)

Del

ta T

(`C

) & D

elta

P(B

ar)

1 2 3 4 56 78 9 10 11 12

Stave TFoM: 13ºC*cm2/W

Pixel maximum temperature:

-24.4ºC


IBL: A detector with very long in and outlet lines

• The IBL detector is only 800mm long, but has about 15m long in and outlets.• dT due outlet line pressure drop significantly (ca 3ºC)• Ambient heat load in same order as detector load

12

Atlas IBL example

Ambient heating

Heat exchange

Inlet Outlet

IBL

2

49

6 7

12

Ambient

Cooling tube temperature profile (HTC & ΔP)

• In detectors the aim of a cooling tube design is:– Low mass or small diameter– Low temperature gradient (hottest point wrt outlet reference)

• For efficient heat transfer: – ΔT(ΔP+HTC) and tube diameter or mass as small as possible

• To quantify the optimal diameter we can look either to the mass or tube volume involved

13

ΔT(ΔP)ΔT(ΔP)(Reduced diameter) ΔT(HTC)

(Reduced diameter)

ΔT(HTC)ΔT(ΔP+HTC)(Reduced diameter)

ΔT(ΔP+HTC)

Tube length

Tem

pera

ture

Fluid temperature

Tube temperature

Vtube*ΔT(ΔP+HTC))

QVolumetric heat transfer =(W/m3*K) M*ΔT(ΔP+HTC))

QMass specific heat transfer =(W/kg*K)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50

2

4

6

8

10

12

14

Diameter (mm)

Tem

pera

ture

Gra

dien

t (ºC

)V

olum

etric

hea

t tra

nsfe

r (W

/cm

2 K)

Volumetric heat

transfer coefficient

Optimal Diameter?

Overall temperature gradient

Heat transfer temperature gradient

Pressure drop temperature gradient

Cooling tube performance example

L=3m, Q=400W, T=-20°C

Models used: HTC and dP, Thome 2008

Comparison of fluids• Volumetric heat transfer is also a good method to compare different fluids.

– How can we put as much heat into a small as possible cooling tube??

• Interesting: Performance almost linear with fluid pressure. 15

0 1 2 3 4 5 6 7 8 9 100

0.002

0.004

0.006

0.008

0.01

0.012

0.014

Tube diameter (mm)

Vol

umet

ric h

eat t

rans

fer c

ondu

ctio

n (W

/mm

3 K)

Overall volumetric heat transfer conductionL=3 m, Q=400 W, T=-20 °C, VQ=0.35

CO2 (19.7 bar)

Ethane (14.2 bar)C2F6 (10.5 bar)

Propane (2.4 bar)C3F8 (2 bar)

Ammonia (1.9 bar)R134a (1.3 bar)

0 1 2 3 4 5 6 7 8 9 100

20

40

60

80

100

120

140

160

180

Tube diameter (mm)

Hea

t tra

nsfe

r con

duct

ion

(W/K

)

Heat transfer conductionL=3 m, Q=400 W, T=-20 °C, VQ=0.35

CO2 (19.7 bar)

Ethane (14.2 bar)C2F6 (10.5 bar)

Propane (2.4 bar)C3F8 (2 bar)

Ammonia (1.9 bar)R134a (1.3 bar)

Models used: HTC-Kandlikar and dP-Friedel

“Drawback” of smaller pipes

• In an embedded structure the smaller pipe is replaced by other material. – ‘Heavy” pipe has a lower weight, but light vapor is also replaced by “something”

• The bottleneck in most cooling structures is the glue layer around the pipe.– Small area– Bad conductance

• Better to judge the whole thermal chain from a fixed volume

16What is now an optimal pipe diameter?

D=10mmD=10mm

Mass related results

(Same case as previous)• Calculation with IBL-like properties:

– T_tube=0.1 mm– T_glue=0.1 mm– k_tube=7.2 W/mk– k_foam=35 W/mk– k_glue=1.02 W/mk– d_glue=2400 kg/m3– d_foam=198 kg/m3– d_tube=4400; kg/m3

170 1 2 3 4 5 6 7 8 9 100

0.05

0.1

Tube diameter (mm)

Mas

s (k

g)

Mass contributionL=3 m, Q=400 W, T=-20 °C, VQ=0-0.35

mtotal

mfoam

mglue

mtube

mf luid

0 1 2 3 4 5 6 7 8 9 100

1

2

3

4

5

6

7

8

9

10

Tube diameter (mm)

Tem

pera

ture

gra

dien

t (`C

)

Temperature gradient contributionL=3 m, Q=400 W, T=-20 °C, VQ=0-0.35

dTtotal

dTfoam

dTglue

dTtube

dThtc

dTdp

0 1 2 3 4 5 6 7 8 9 100

500

1000

1500

Tube diameter (mm)

Hea

t tra

nsfe

r (W

/kg*

K)

Mass relative heat transfer (dP & HTC) L=3 m, Q=400 W, T=-20 °C, VQ=0-0.35

CO2 (19.7 bar)

Recalculating the IBL

• Selected IBL tube is 1.5mm => good choice!

• Next to do: Make similar analyses wrt radiation length

18

0 1 2 3 4 5 6 7 8 9 100

200

400

600

800

1000

Tube diameter (mm)

Hea

t tra

nsfe

r (W

/kg*

K)

Mass relative heat transfer (dP & HTC) L=0.8 m, Q=70 W, T=-40 °C, VQ=0-0.35

CO2 (10 bar)

0 1 2 3 4 5 6 7 8 9 100

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

Tube diameter (mm)

Mas

s (k

g)

Mass contributionL=0.8 m, Q=70 W, T=-40 °C, VQ=0-0.35

mtotal

mfoam

mglue

mtube

mf luid

0 1 2 3 4 5 6 7 8 9 100

1

2

3

4

5

6

7

8

9

10

Tube diameter (mm)

Tem

pera

ture

gra

dien

t (`C

)

Temperature gradient contributionL=0.8 m, Q=70 W, T=-40 °C, VQ=0-0.35

dTtotal

dTfoam

dTglue

dTtube

dThtc

dTdp

1.5mm

1.5mm

CO2 heat transfer and pressure drop modeling

• Nowadays good prediction models of CO2 are available.

• For detector analyzes we use mainly the models of J. Thome from EPFL Lausanne (Switzerland)– Cheng L, Ribatski G, Quiben J, Thome J, 2008,”New prediction methods for CO2 evaporation inside

tubes: Part I – A two-phase flow pattern map and a flow pattern based phenomenological model for two-phase flow frictional pressure drops”, International Journal of Heat and Mass Transfer, vol 51, p111-124

– Cheng L, Ribatski G, Thome J, 2008,”New prediction methods for CO2 evaporation inside tubes: Part II– An updated general flow boiling heat transfer model based on flow patterns”, International Journal of Heat and Mass Transfer, vol 51, p111-124

• Models are flow pattern based and are reasonably well predicting the flow conditions and the related heat transfer and pressure drop. Dry-out prediction is included.

• The Thome models are successfully used to predict the complex thermal behavior of particle detector cooling circuits.

• A simulation program called CoBra is under development at Nikhef/CERN-DT to analyze full detector cooling branches.

19

Experimental heat transfer data (measured at SLAC)

20

Interesting research on heat transfer is done at SLAC in a joint effort with Nikhef.M. Oriunno (SLAC) & G. Hemmink (Nikhef)

CoBra Model(CO2 BRAnch Model)

21

R1xR2x

R3x

R5x R2y+1

R3y+1

R4x R4y+1

R1y+1

R1X+1R2X+1

R3X+!

R5X+1 R2y

R3y

R4X+1 R4y

R1y

Px+1,Hx+1,Tx+1

Px,HxTx

Py+1,Hy+1,Ty+1

Py,Hy,,Ty

T2

3

4

1

Px+1=dPx+1+Px

Hx+1=dHx+1+Hx

dH=Q1/MFQ1 is calculated in the thermal network

2

3

4

The thermal node network calculates the heat influx in the cooling pipe based on:• Applied power Q3 on node 3• Environmental heating from fixed temperature T4 on node 4• Heat exchange with another pipe section via R5 between nodes 2 and 2 of the connected

sections

CoBra example calculation

• CoBra is able to analyze complex thermal profiles of CO2 in long tubes

22

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-1

0

1

2

3

4

5

6

7

2mmID x 5m tube temperature and pressure profile. MF=0.3g/s, Tsp=-30ºC, Q=90, xend=0.99

Branch length (m)

Del

ta T

(`C

) & D

elta

P(B

ar)

1 2 3 4


0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

50

100

150

200

250

300

350

5mx1.5mmID tube (4m heated), Mass flow=0.3g/s, Q=90Watt , T=-30ºC

Dry-out

MistAnnular

Stratified Wavy

Stratified

Bub

bly

Slu

g

Process path

Dry-out

Liqu

id

SlugAnnular

Cobra example:Node network for IBL

23

Tenvironement

R4≈ HTCair

R2 +R3 ≈ TFoM

R1 ≈ HTCCO2

TCO2

Tenvironement

TCO2 TCO2

R4 +R3 ≈ Insulation+HTCair

R5≈ Heat exchange

R1≈ HTCCO2

R2 ≈ Tube wall

Q3 ≈ Applied power

Tenvironement

TCO2

R4 +R3 ≈ HTCairR1≈ HTCCO2

R2 ≈ Tube wall

Tenvironement

TCO2

R1≈ HTCCO2

R2 ≈ Tube wall

TCO2

R1b≈ HTCCO2

R1a≈ HTCCO2

R4 +R3 ≈ Insulation+HTCair

1. Concentric line

3. Bare tube

2. Bundled lines

4. Stave

Internal heat exchange and ambient heating

24

Figure 7: CoBra calculation example of the IBL cooling tube. The branch has a 1mm inlet (1-4), a 1.5 mm cooling tube (4-8), a 2mm outlet (8-9) followed by a 3mm outlet (9-12). The dashed temperature profile is the actual sensor temperature taking into account the conductance of the support structure. The graph on the left has no internal heat exchange, the right graph takes internal heat exchange of the in and outlet tube into account.

0 5 10 15 20 25-2

0

2

4

6

8

10

12

14

16


Branch length (m)D

elta

T (`

C) &

Del

ta P

(Bar

)

1 2 3 4 56 78 9 10 11 12

Stave TFoM: 13ºC*cm2/WPixel maximum temperature:

-24.4ºC


0 5 10 15 20 25-2

0

2

4

6

8

10

12

14

16


Branch length (m)

Del

ta T

(`C

) & D

elta

P(B

ar)

1 2 3 4 56 78 9 10 11 12

Stave TFoM: 13ºC*cm2/WPixel maximum temperature:

-24.2ºC


CapillaryNo Boiling

Boiling starts

Boiling starts


Comparison to test results

25

Figure 8: Temperature and pressure test results of the CMS pixel upgrade cooling branch (left) and the Atlas IBL cooling branch (right). Comparison with the CoBra calculator showed that the calculator is a promising tool for predicting the temperature and pressure gradients over long length cooling branches.

0 5 10 15 20 25-1

0

1

2

3

4

5

6

7

8

9

IBL temperature and pressure profile. MF=1.1g/s, Tsp=-26.71ºC, Q=73.5, xend=0.36

Branch length (m)D

elta

T (`

C) &

Del

ta P

(Bar

)

12 3 4 56 78 9 10 1112


0 5 10 15-20

-18

-16

-14

-12

-10

-8

-6

Loop Length [m]

Tem

pera

ture

[°C

]

m = 1.48g/s | Qtotal = 256.87W | Pin = 26.40Bar | Tin = -17.40°C | dP = 6.33Bar | dT = 10.86°C

0 5 10 1520

21

22

23

24

25

26

27

Pre

ssur

e [B

ar]

Exp. Wall TemperatureTheory Wall TemperatureTheory CO2 Temperature

Theory CO2 Pressure

CMS detector (1.4mm)El

ectr

onics

(1.8

mm

)

Inle

t tub

e (1

.8m

m)

Outle

t tub

e (1.

8mm

)Inlet tube

(1mm)

Outlet tube(3mm)Ou

tlet t

ube (

2mm

)

Detector(1.5mm)

CMS-B-PIX Atlas-IBL


Cooling development philosophy

• Whatever the model give as a result: – Don’t trust them! the

models are empirical.– Use them as a design

guideline.• Always verify in a test!

– Not only to quantify heat transfer, but as well to filter out strange behavior

26

0 2 4 6 8 10 12 14-20

-18.4

-16.8

-15.2

-13.6

-12

Length [m]

Tem

pera

ture

[°C

]

Test 1 | m = 2.94g/s | Qtotal = 159.30W | dP = 17.77Bar | dT = 4.20°C

dTexp =6.55°CdPexp =20.41bar

0 2 4 6 8 10 12 1419

23.4

27.8

32.2

36.6

41

Pre

ssur

e [B

ar]

Exp. Wall Temperature

Theory Wall TemperatureTheory CO

2 Temperature

Exp. CO2 Pressure

Theory CO2 Pressure

Strange start-up behavior in the Velo

• In the Velo the CO2 does not always start boiling.

• It turned out that our bright idea of increasing the tube length wasn’t so brilliant after all.

27

18:00 18:30 19:00 19:30 20:00 20:30 21:00-18

-16

-14

-12

-10

-8

-6

-4

-2

0

2

4

6

8

10

0

4

8

12

16

20

24

28

32

36

40

44

48

52

Hea

tload

(W)

Tem

pera

ture

(°C

), S

troke

(mm

)

Time (hh:mm)02-Oct-2009

Pump stroke (mm) Evaporator saturation temperature (°C) Evaporator 12 begin temperature (°C) Evaporator 12 end temperature (°C) Evaporator 27 begin temperature (°C) Evaporator 27 end temperature (°C)

Tube length

At startup everything is liquid

1. The outlet starts to boil2. The good boiling heat transfer

is taking heat away from the inlet

3. As a result boiling at the inlet is suppressed.

4. Once boiling is achieved it will not go back to the liquid state

Fluid temp.

Tube wall temp.

Tem

pera

ture

Conclusions

• The common temperature boundaries of all parallel systems are:– The pressure (=saturation temperature) in the outlet manifold.– The temperature / enthalpy of the liquid in the inlet manifold.– Normally both temperatures are the same.

• Parallel channels need flow distribution by increasing the inlet pressure drop

• The in let manifold must be sub-cooled liquid.

• Overall performance of the thermal system includes:– Conductive path in detector structure– Heat transfer to the evaporative liquid– Pressure drop in evaporator and outlet tube.– Full thermal path must be considered in the design optimization

• Always verify your models with tests. 2-phase flow has sometimes strange behavior

• Avoid tube crosstalk, boiling in 1 channel can suppress boiling in the other 28

Things to keep in mind when designing CO2 cooling loops:

Documents

Bart Verlaat