1. • As expected, reduction in runtime and subsequent losses in accuracy were obtained with higher degrees of nodal reduction.
• Across projects, temperature differences were larger for the Hot cases (versus the Cold cases) between the detailed and reduced model. This may be brought about by the higher variations in environmental heat flux and internal power dissipations in the Hot case.
• As each project instrument model received higher degrees of nodal reduction, it resulted in a larger magnitude of error. For less simplified models (i.e. ICESat-2), the amount of error was minimal; whereas, over-simplified models (i.e. GEMS and LADEE) encountered a drastic increase in error. This would indicate that there may exist a threshold of reduction for models such that they should not be reduced beyond a certain proportion before incurring large errors.
• Though losses in accuracy are inevitable when simplifying a thermal model, there are occasions when this may be neglected in favor of quick results. During the course of its mission, it may be required to have results at hand in order to respond to some immediate flight scenario.
• The production of reduced models should be considered on a case by case basis for each mission.
Investigation on the Practicality of Developing Reduced Thermal Models
Giancarlo Lombardi Florida International University
Engineering Summer 2014
Kan Yang GSFC-545
The Thermal Engineering Branch at NASA Goddard is responsible for the design, development, and implementation of spacecraft thermal control systems. Through application of thermal design principles and heat transfer analysis, thermal engineers work alongside other subsystems to ensure that each spacecraft design conforms to its required thermal parameters. To aid in analysis, thermal engineers generate thermal models, mimicking the spacecraft’s geometry and on-orbit environment. These models are produced with immense detail to capture realistic on-orbit behavior of the spacecraft. However, this comes at a great cost: to obtain physically accurate results over long orbital periods, it may take up to several hours, or even days, of simulation time. Hence, reduced models are often generated in tandem with the detailed model to reduce runtimes and quickly obtain results, usually at the cost of accuracy. A thermal model is composed of surfaces and nodes, for which ray traces are performed to obtain radks (radiative couplings) and heating rates (environmental and internal heat sources/sinks). A model’s complexity is then determined by the number of nodes and surfaces it contains. Thus, a highly “detailed” model (i.e. large amount of nodes) holds an exponentially larger amount of calculations per run than a “reduced" model (i.e. fewer amount of nodes). This then leads to an increased amount of runtime necessary to complete a simulation. The generation of a reduced model is often desired due to the reduction in runtime. However, when simplifying a model, it is challenging to identify which surface can be modeled with less nodes such that results remain fairly consistent with the detailed model. Considering the additional effort that is required to produce the reduced model, this study focuses on the practicality of commissioning reduced models. In particular, it addresses the question: does the decreased runtime to obtain results compensate for the additional effort placed on producing the reduced models? For this study, thermal models of several projects (namely: GEMS, LADEE, and ICESat-2) were used to obtain a general comparison between detailed and reduced thermal models on runtime vs. accuracy. The worst-case "hot” and “cold” environmental loading and heat dissipation scenarios were run for each model to grasp the spacecrafts’ behavior at both ends of its operational spectrum. Overall, it is hoped that a general trend will emerge from the data to determine the practicality of generating reduced models.
Conclusion
GEMS - Hot Case
Submodel
Temperature [°C]
Detailed
Reduced
Δ T
MOBDECK
-56.8
-57.6
-0.8
PMIRR1
-61.6
-45.5
16.1
PMIRR2
-61.8
-45.9
15.9
SS_DECK
-56.6
-52.6
4.0
TOP_TS_1
-60.4
-44.3
16.1
TOP_TS_2
-60.7
-44.7
16.0
LADEE - Hot Case
Sub Model
Temperature [°C]
Detailed
Reduced
Δ T
BSPL
37.4
-16.1
-53.5
CPNL
-21.3
-21.3
0.0
INTP
12.9
3.8
-9.1
MEB
-42.7
-21.9
20.8
QMS
14.9
-27.0
-41.9
LADEE - Cold Case
Sub Model
Temperature [°C]
Detailed
Reduced
Δ T
BSPL
-23.9
-19.5
4.4
CPNL
-45.3
-22.4
22.9
INTP
10.1
10.8
0.7
MEB
-61.9
-22.8
39.1
QMS
-29.3
-30.2
-0.9
ATLAS - Hot Case
Sub Model
Temperature [°C]
Detailed
Reduced
Δ T
LRS_ORAD
13.9
13.9
0
LTCS_RAD
-15.6
-16.2
-0.6
MEB_RAD
25.9
26.1
0.2
PBC
19.9
20.2
0.3
PDU_RAD
23.7
23.6
-0.1
STARTPD
25.2
23.4
-1.8
Case
Complexity
Time (min)
Time Reduction
Hot
Detailed
106.4
61.7%
(65.7 min)
Reduced
40.7
Cold
Detailed
62.5
48.3%
(30.2 min)
Reduced
32.3
Case
Complexity
Time (min)
Time Reduction
Hot
Detailed
57
21.1%
(12 min)
Reduced
45
Cold
Detailed
61.7
14.1%
(8.7 min)
Reduced
53
Case
Complexity
Time (min)
Time Reduction
Hot
Detailed
444.3
29.1%
(129.3 min)
Reduced
315
Cold
Detailed
416.3
32.7%
(136 min)
Reduced
280.3
- Mission: Measure ice cap elevation and thickness
• Advanced Topographic Laser Altimeter System (ATLAS) stand-alone model results were compared individually
• Reduced Model Development Time: 120 hr
• Nodal Count: - Spacecraft bus: 3,149 - ATLAS Detailed: 11,737 - ATLAS Reduced: 6,392
- Mission: Observe strong gravitational fields around black holes and magnetic fields around pulsars
• GEMS Mirror stand-alone model results were compared individually from spacecraft bus
• Reduced Model Development Time: 80 hr
• Nodal Count: - Spacecraft bus: 6,080 - Mirror Detailed: 17,025 - Mirror Reduced: 654
- Mission: Analyze the Moon's thin exosphere and the lunar dust environment
• “Detailed” Neutral Mass Spectrometer (NMS) sub model was integrated onto “Reduced” spacecraft bus
• Reduced Model Development Time: 80 hr
•Nodal Count: - Spacecraft bus: 14,750 - NMS Detailed: 1,040 - NMS Reduced: 35
Lunar Atmosphere and Dust Environment Explorer (LADEE)
Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2)
Gravity and Extreme Magnetism SMEX (GEMS)
Introduction
Methodology
Acknowledgements
Reduction: 96.6% Max Error, ΔT [°C]: - Hot: 53.5 - Cold: 39.1
Reduction: 45.5% Max Error, ΔT [°C]: - Hot: 1.8 - Cold: 0.9
Reduction: 96.2% Max Error, ΔT [°C]: - Hot: 16.1 - Cold: 9.3
I would like to thank Kan Yang for his valuable advice and mentoring. Thanks also go to Hume Peabody, Juan Rodriguez-Ruiz, and the entire Thermal Engineering Branch at GSFC for aiding and instructing me on the use of the software and providing useful thermal knowledge.
Full ICESat-2 Thermal Model (ATLAS in red box)
LADEE Reduced Thermal Model NMS instrument in blue box
LADEE Detailed Thermal Model NMS instrument in blue box
ATLAS Reduced Thermal Model
ATLAS Detailed Thermal Model
GEMS Mirror Reduced Thermal Model
GEMS Mirror Detailed Thermal Model
Full GEMS Thermal Model (Mirror in red box)
1. Set up Case Set in Thermal Desktop (TD) - Set up cases with both the spacecraft reduced and detailed instrument models on the spacecraft bus. When possible, integrate the spacecraft detailed/reduced model to the bus to obtain the total observatory-level model runtime - Calculate environmental heat due to spacecraft orbit (heat rate) - Specify duration of transient run (5 orbits) - Conduct ray trace calculations for radiation couplings (radks) 2. Start case set in TD to generate SINDA .inp file 3. Open SINDA/FLUINT and run .inp file - Record start and end times to solve case 4. Repeat each case set run three times to obtain an average runtime - Calculate total run time for each run (runtime = end - start time) 5. Post-process using TARP - Generate Temperature and Heat tables using Group Max/Min Parameters - Incorporate Weighting file to mitigate effect of components with low thermal masses (e.g. MLI blankets) 6. Compare Reduced model data with Detailed model data - Record change in temperature (delta T = reduced - detailed model temperature results)
Specs for Hardware Used
Processor
Intel Core i7 vPro
System
64-bit OS
RAM
8.00 GB
ATLAS - Cold Case
Submodel
Temperature [°C]
Detailed
Reduced
Δ T
LRS_ORAD
20.7
20.9
0.2
LTCS_RAD
-48.0
-48.9
-0.9
MEB_RAD
5.6
5.8
0.2
PBC
0.4
0.6
0.2
PDU_RAD
-12.0
-12.0
0
STARTPD
4.7
4.0
-0.7
GEMS - Cold Case
Submodel
Temperature [°C]
Detailed
Reduced
Δ T
MOBDECK
-74.0
-75.0
-1.0
PMIRR1
-80.9
-79.0
1.9
PMIRR2
-79.2
-69.9
9.3
SS_DECK
-82.1
-80.1
2.0
TOP_TS_1
-81.2
-80.3
0.9
TOP_TS_2
-79.4
-71.9
7.5
Break Even Run Count
Mission
Hot
Cold
GEMS
38
36
ICESAT-2
110
239
LADEE
552
400
(break even = time reduction / development time)
• To break even with the reduced model development time, the following table demonstrates the necessary amount of runs needed for the time gained from reducing the model to become beneficial. These high number of runs indicate that there was not much gain, in terms of runtime, in producing the reduced models.
Launch Date: September 2013
Launch Date: July 2016
Launch Date: CANCELLED
![• As expected, reduction in runtime and subsequent losses in accuracy were obtained with higher degrees of nodal reduction.
• Across projects, temperature differences were larger for the Hot cases (versus the Cold cases) between the detailed and reduced model. This may be brought about by the higher variations in environmental heat flux and internal power dissipations in the Hot case.
• As each project instrument model received higher degrees of nodal reduction, it resulted in a larger magnitude of error. For less simplified models (i.e. ICESat-2), the amount of error was minimal; whereas, over-simplified models (i.e. GEMS and LADEE) encountered a drastic increase in error. This would indicate that there may exist a threshold of reduction for models such that they should not be reduced beyond a certain proportion before incurring large errors.
• Though losses in accuracy are inevitable when simplifying a thermal model, there are occasions when this may be neglected in favor of quick results. During the course of its mission, it may be required to have results at hand in order to respond to some immediate flight scenario.
• The production of reduced models should be considered on a case by case basis for each mission.
Investigation on the Practicality of Developing Reduced Thermal Models
Giancarlo Lombardi Florida International University
Engineering Summer 2014
Kan Yang GSFC-545
The Thermal Engineering Branch at NASA Goddard is responsible for the design, development, and implementation of spacecraft thermal control systems. Through application of thermal design principles and heat transfer analysis, thermal engineers work alongside other subsystems to ensure that each spacecraft design conforms to its required thermal parameters. To aid in analysis, thermal engineers generate thermal models, mimicking the spacecraft’s geometry and on-orbit environment. These models are produced with immense detail to capture realistic on-orbit behavior of the spacecraft. However, this comes at a great cost: to obtain physically accurate results over long orbital periods, it may take up to several hours, or even days, of simulation time. Hence, reduced models are often generated in tandem with the detailed model to reduce runtimes and quickly obtain results, usually at the cost of accuracy. A thermal model is composed of surfaces and nodes, for which ray traces are performed to obtain radks (radiative couplings) and heating rates (environmental and internal heat sources/sinks). A model’s complexity is then determined by the number of nodes and surfaces it contains. Thus, a highly “detailed” model (i.e. large amount of nodes) holds an exponentially larger amount of calculations per run than a “reduced" model (i.e. fewer amount of nodes). This then leads to an increased amount of runtime necessary to complete a simulation. The generation of a reduced model is often desired due to the reduction in runtime. However, when simplifying a model, it is challenging to identify which surface can be modeled with less nodes such that results remain fairly consistent with the detailed model. Considering the additional effort that is required to produce the reduced model, this study focuses on the practicality of commissioning reduced models. In particular, it addresses the question: does the decreased runtime to obtain results compensate for the additional effort placed on producing the reduced models? For this study, thermal models of several projects (namely: GEMS, LADEE, and ICESat-2) were used to obtain a general comparison between detailed and reduced thermal models on runtime vs. accuracy. The worst-case "hot” and “cold” environmental loading and heat dissipation scenarios were run for each model to grasp the spacecrafts’ behavior at both ends of its operational spectrum. Overall, it is hoped that a general trend will emerge from the data to determine the practicality of generating reduced models.
Conclusion
GEMS - Hot Case
Submodel
Temperature [°C]
Detailed
Reduced
Δ T
MOBDECK
-56.8
-57.6
-0.8
PMIRR1
-61.6
-45.5
16.1
PMIRR2
-61.8
-45.9
15.9
SS_DECK
-56.6
-52.6
4.0
TOP_TS_1
-60.4
-44.3
16.1
TOP_TS_2
-60.7
-44.7
16.0
LADEE - Hot Case
Sub Model
Temperature [°C]
Detailed
Reduced
Δ T
BSPL
37.4
-16.1
-53.5
CPNL
-21.3
-21.3
0.0
INTP
12.9
3.8
-9.1
MEB
-42.7
-21.9
20.8
QMS
14.9
-27.0
-41.9
LADEE - Cold Case
Sub Model
Temperature [°C]
Detailed
Reduced
Δ T
BSPL
-23.9
-19.5
4.4
CPNL
-45.3
-22.4
22.9
INTP
10.1
10.8
0.7
MEB
-61.9
-22.8
39.1
QMS
-29.3
-30.2
-0.9
ATLAS - Hot Case
Sub Model
Temperature [°C]
Detailed
Reduced
Δ T
LRS_ORAD
13.9
13.9
0
LTCS_RAD
-15.6
-16.2
-0.6
MEB_RAD
25.9
26.1
0.2
PBC
19.9
20.2
0.3
PDU_RAD
23.7
23.6
-0.1
STARTPD
25.2
23.4
-1.8
Case
Complexity
Time (min)
Time Reduction
Hot
Detailed
106.4
61.7%
(65.7 min)
Reduced
40.7
Cold
Detailed
62.5
48.3%
(30.2 min)
Reduced
32.3
Case
Complexity
Time (min)
Time Reduction
Hot
Detailed
57
21.1%
(12 min)
Reduced
45
Cold
Detailed
61.7
14.1%
(8.7 min)
Reduced
53
Case
Complexity
Time (min)
Time Reduction
Hot
Detailed
444.3
29.1%
(129.3 min)
Reduced
315
Cold
Detailed
416.3
32.7%
(136 min)
Reduced
280.3
- Mission: Measure ice cap elevation and thickness
• Advanced Topographic Laser Altimeter System (ATLAS) stand-alone model results were compared individually
• Reduced Model Development Time: 120 hr
• Nodal Count: - Spacecraft bus: 3,149 - ATLAS Detailed: 11,737 - ATLAS Reduced: 6,392
- Mission: Observe strong gravitational fields around black holes and magnetic fields around pulsars
• GEMS Mirror stand-alone model results were compared individually from spacecraft bus
• Reduced Model Development Time: 80 hr
• Nodal Count: - Spacecraft bus: 6,080 - Mirror Detailed: 17,025 - Mirror Reduced: 654
- Mission: Analyze the Moon's thin exosphere and the lunar dust environment
• “Detailed” Neutral Mass Spectrometer (NMS) sub model was integrated onto “Reduced” spacecraft bus
• Reduced Model Development Time: 80 hr
•Nodal Count: - Spacecraft bus: 14,750 - NMS Detailed: 1,040 - NMS Reduced: 35
Lunar Atmosphere and Dust Environment Explorer (LADEE)
Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2)
Gravity and Extreme Magnetism SMEX (GEMS)
Introduction
Methodology
Acknowledgements
Reduction: 96.6% Max Error, ΔT [°C]: - Hot: 53.5 - Cold: 39.1
Reduction: 45.5% Max Error, ΔT [°C]: - Hot: 1.8 - Cold: 0.9
Reduction: 96.2% Max Error, ΔT [°C]: - Hot: 16.1 - Cold: 9.3
I would like to thank Kan Yang for his valuable advice and mentoring. Thanks also go to Hume Peabody, Juan Rodriguez-Ruiz, and the entire Thermal Engineering Branch at GSFC for aiding and instructing me on the use of the software and providing useful thermal knowledge.
Full ICESat-2 Thermal Model (ATLAS in red box)
LADEE Reduced Thermal Model NMS instrument in blue box
LADEE Detailed Thermal Model NMS instrument in blue box
ATLAS Reduced Thermal Model
ATLAS Detailed Thermal Model
GEMS Mirror Reduced Thermal Model
GEMS Mirror Detailed Thermal Model
Full GEMS Thermal Model (Mirror in red box)
1. Set up Case Set in Thermal Desktop (TD) - Set up cases with both the spacecraft reduced and detailed instrument models on the spacecraft bus. When possible, integrate the spacecraft detailed/reduced model to the bus to obtain the total observatory-level model runtime - Calculate environmental heat due to spacecraft orbit (heat rate) - Specify duration of transient run (5 orbits) - Conduct ray trace calculations for radiation couplings (radks) 2. Start case set in TD to generate SINDA .inp file 3. Open SINDA/FLUINT and run .inp file - Record start and end times to solve case 4. Repeat each case set run three times to obtain an average runtime - Calculate total run time for each run (runtime = end - start time) 5. Post-process using TARP - Generate Temperature and Heat tables using Group Max/Min Parameters - Incorporate Weighting file to mitigate effect of components with low thermal masses (e.g. MLI blankets) 6. Compare Reduced model data with Detailed model data - Record change in temperature (delta T = reduced - detailed model temperature results)
Specs for Hardware Used
Processor
Intel Core i7 vPro
System
64-bit OS
RAM
8.00 GB
ATLAS - Cold Case
Submodel
Temperature [°C]
Detailed
Reduced
Δ T
LRS_ORAD
20.7
20.9
0.2
LTCS_RAD
-48.0
-48.9
-0.9
MEB_RAD
5.6
5.8
0.2
PBC
0.4
0.6
0.2
PDU_RAD
-12.0
-12.0
0
STARTPD
4.7
4.0
-0.7
GEMS - Cold Case
Submodel
Temperature [°C]
Detailed
Reduced
Δ T
MOBDECK
-74.0
-75.0
-1.0
PMIRR1
-80.9
-79.0
1.9
PMIRR2
-79.2
-69.9
9.3
SS_DECK
-82.1
-80.1
2.0
TOP_TS_1
-81.2
-80.3
0.9
TOP_TS_2
-79.4
-71.9
7.5
Break Even Run Count
Mission
Hot
Cold
GEMS
38
36
ICESAT-2
110
239
LADEE
552
400
(break even = time reduction / development time)
• To break even with the reduced model development time, the following table demonstrates the necessary amount of runs needed for the time gained from reducing the model to become beneficial. These high number of runs indicate that there was not much gain, in terms of runtime, in producing the reduced models.
Launch Date: September 2013
Launch Date: July 2016
Launch Date: CANCELLED](https://image.slidesharecdn.com/626a4dc7-e6ec-4d72-8d81-053cb70db5e9-141124184730-conversion-gate01/85/giancarlo-lombardisummer2014nasagsfc-1-320.jpg)
