Appendix R: Lessons Learned On Modelling of Cryogenic Systems

285
Appendix R
Lessons Learned on Modelling of Cryogenic Systems
Moritz Branco
(ESA/ESTEC, The Netherlands)
25th European Workshop on Thermal and ECLS Software 89 November 2011

286 Lessons Learned on Modelling of Cryogenic Systems
Abstract
The use of ESATAN-TMS as a thermal modeling tool for systems in the cryogenic domain (< 120 K),
gives rise to specific issues on model convergence and the results analysis. This presentations purpose
is to present some of the issues found and solutions considered while working on a model for a compact
cryostat with a full cryogenic chain from 300 K to 2 K.

Lessons Learned on Modelling of Cryogenic Systems 287
Lessons Learned on Modelling of

Cryogenic Systems with ESATAN-TMS
25th European Workshop on Thermal and ECLS Software
Moritz Branco
ESTEC, Noordwijk, The Netherlands

8-9/11/2011
Lessons Learned on Cryogenics Modelling

Contents
1. Context:
The XMS Cryo-Chain
2. Problems Encountered /
Solutions Proposed
3. Results Analysis:
Specific Issues
Lessons Learned on Modeling of Cryogenic Systems with ESATAN-TMS | Moritz Branco | 14/04/2011 |
Noordwijk | Pag. 2


The XMS Cryo-Chain
Brief Description:
X-ray microcalorimeter spectrometer

(XMS)
Instrument aboard ATHENA, previously

called IXO
Requires cooling down to 50 mK with 1 W

cooling power available
No liquid cryogens (5-10 years life)
Depictions of IXO, whole and XIM

The XMS Cryo-Chain
The ESATAN-TMS model:
Russian Doll type configuration (different T

stages, 100 K, 15 K, 2K..)
Performance Data from available cryo
coolers
Data on MLI, harness, mechanical supports
from previous studies and missions
Detailed analysis of the optical baffle
Study on flexible thermal links
Modelling of interdependent behaviour of

cryocoolers


Problems Encountered/Solutions
Problem: TABS = 0.0
In the model.d file the node block

defines all nodes with a T=0.0, no matter
the TABS considered.
Solver crashes
Solutions proposed:
Setting an initial temperature

boundary condition to all nodes: $INITIAL
not flexible and didnt work. C set initial temperatures
Setting an initialization routine CALL SETNDR( , T ,10.0D0, CURRENT)
in the template file with SETNDR.
Worked.

Problem: MLI modelling
MLI is modelled as having an

effective emissivity and a
temperature dependent pseudo-
conductivity
Solution found:
Additional shell added, and a conductor depending
on average T (TMLI and TSHELL) is entered between
each face
Not flexible:
- nodal breakdown analysis
- geometry study
Possible request feature: Temperature dependent through-

conductance in a shell


Problem: MLI modelling
MLI is modelled as having an

effective emissivity and a
temperature dependent pseudo-
conductivity
Solution found:
Additional shell added, and a conductor depending
on average T (TMLI and TSHELL) is entered between
each face
Not flexible: MLIcond(T)=aT+b T=Th/2+Tc/2
- nodal breakdown analysis

GL(MLIi, shelli)=MLIcond(TMLIi/2+Tshelli/2)
- geometry study
Possible request feature: Temperature dependent through-

conductance in a shell

Problem: High GLs for small Ts
Highly conductive thermal links (100 W/K) for small

temperature differences (10-1 K) are typical in
cryogenic systems (e.g. copper straps).
Numerical instabilities:
Solver not solving or
Diffusive node (shell node) with a very high
relative heat inbalance
Solution found:
Using SOLVFM / in cases transient
Applying a damping factor 0.1-0.5
Initial Temperatures boundary setting to

start with a very low T


Cooler Modelling
First Approach: Boundary nodes, T constant
Diffusive nodes with an balanced QI,

given by the cooler performance data.
Boundary node with T changing every

iteration,
T given by the cooler performance data.
A damping scheme had to be applied.
MPTC
LPTC

Cooler Modelling
First Approach: Boundary nodes, T constant
Diffusive nodes with an balanced QI,

given by the cooler performance data.
Boundary node with T changing every

iteration,
T given by the cooler performance data. Qcool = FLUXGL ( Tcooler, Tnode)
Ti+1 = Performance Curve (Qcool)
A damping scheme had to be applied. Tcooler = Tcooler + DAMPF * (Ti+1-Ti)


Cooler Modelling
Problem:
Cooler temperature was calculated using Qcool = FLUXGL ( Tcooler, Tnode)

FLUXGL
Ti+1 = Performance Curve (Qcool)
Temperatures result in NaN Tcooler = Tcooler + DAMPF * (Ti+1-Ti)
At first iteration, when calling

$VARIABLES1, GLs arent yet calculated
Solution found:
To manually calculate the GL between cooler and shell node
Qcool = CNDFN1(Tcooler, Tnode, KThermalStrap, 1)*(Tcooler-Tnode)

Results Analysis
Results Analysis
Heat flows are the most

important quantity
Importance of different contributions for

a general heatflow
Accuracy of heatflow results is

critical, since it could drive the whole
system


Results Analysis
Results Analysis
Importance of heat flow accuracy
Cryogenic models are very

sensitive to very small heatflow
variations
At 20 K, every mW counts!
At 2 K, one mW can drive the

whole cryostat configuration

Results Analysis
Results Analysis
Critical Factors:
Numerical uncertainties
Modelling parameters sensitivity
Low accuracy of results high

engineering margins
More importance given to empirical

knowledge from previous cases


Conclusions
Conclusions
Specific Issues were tackled
System level cooler modelling
Important factors in cryogenic model results analysis
Empirical knowledge still most valuable
THANK YOU
Moritz Branco,
Work undergone at TEC-MT
[email protected]

Appendix R: Lessons Learned On Modelling of Cryogenic Systems

Uploaded by

Copyright:

Available Formats

Appendix R: Lessons Learned On Modelling of Cryogenic Systems

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Appendix R: Lessons Learned On Modelling of Cryogenic Systems

Uploaded by

Copyright:

Available Formats

285

Lessons Learned on Modelling of Cryogenic Systems

25th European Workshop on Thermal and ECLS Software 89 November 2011

25th European Workshop on Thermal and ECLS Software 89 November 2011

Lessons Learned on Modelling of

ESTEC, Noordwijk, The Netherlands

Lessons Learned on Cryogenics Modelling

25th European Workshop on Thermal and ECLS Software 89 November 2011

Lessons Learned on Cryogenics Modelling

X-ray microcalorimeter spectrometer

Instrument aboard ATHENA, previously

Requires cooling down to 50 mK with 1 W

No liquid cryogens (5-10 years life)

Depictions of IXO, whole and XIM

Lessons Learned on Cryogenics Modelling

The ESATAN-TMS model:

Russian Doll type configuration (different T

Detailed analysis of the optical baffle

Study on flexible thermal links

Modelling of interdependent behaviour of

25th European Workshop on Thermal and ECLS Software 89 November 2011

Lessons Learned on Cryogenics Modelling

Problem: TABS = 0.0

In the model.d file the node block

Setting an initial temperature

Lessons Learned on Cryogenics Modelling

Problem: MLI modelling

MLI is modelled as having an

Possible request feature: Temperature dependent through-

25th European Workshop on Thermal and ECLS Software 89 November 2011

Lessons Learned on Cryogenics Modelling

Problem: MLI modelling

MLI is modelled as having an

- nodal breakdown analysis

Possible request feature: Temperature dependent through-

Lessons Learned on Cryogenics Modelling

Problem: High GLs for small Ts

Highly conductive thermal links (100 W/K) for small

Applying a damping factor 0.1-0.5

Initial Temperatures boundary setting to

25th European Workshop on Thermal and ECLS Software 89 November 2011

Lessons Learned on Cryogenics Modelling

First Approach: Boundary nodes, T constant

Diffusive nodes with an balanced QI,

Boundary node with T changing every

A damping scheme had to be applied.

Lessons Learned on Cryogenics Modelling

First Approach: Boundary nodes, T constant

Diffusive nodes with an balanced QI,

Boundary node with T changing every

25th European Workshop on Thermal and ECLS Software 89 November 2011

Lessons Learned on Cryogenics Modelling

Cooler temperature was calculated using Qcool = FLUXGL ( Tcooler, Tnode)

At first iteration, when calling

To manually calculate the GL between cooler and shell node

Qcool = CNDFN1(Tcooler, Tnode, KThermalStrap, 1)*(Tcooler-Tnode)

Lessons Learned on Cryogenics Modelling

Heat flows are the most