Discovery Channel Telescope: Cold Plate Design

Design Process

Prior to the NAU team beginning the project, Lowell and established a cooling plate concept and had proved it analytically.  This, along with the fast pace of the project, we decided to utilize the design concept and optimize is it rather than trying to completely invent a new design. 

State-of-the-Art Research

 A document was given to us done by Lowell Observatory which did a 1-D, steady-state analysis on the cooling power required to extract heat from the mirror and a Partial Differential Equation (PDE) solver which used the cooling power required as an input to verify if the mirror would be able to track the ambient temperature.  It was seen from this document that the entire cold plate assembly, if having a cooling power of 140 Watts, will be able to track the mirror to the ambient temperature.  This would effectively cool the mirror to ambient which would solve the problem of the mirror cooling naturally through convection.

Another document was found online on the Southern European Observatory this source documented a cooling system comparable to the one being developed for the Discovery Channel Telescope.  They gave a good idea of material selection; copper tubes brazed onto a rolled aluminum plate with coolant running through them.  They also used radiation cooling, but decided also to use small fans between the mirror and the plate which would increase the cooling power even further by removing heat via forced convection.  Most importantly, it verified that radiation cooling has been implemented in another telescope and has been proven to work.

Cold plate and heat exchanger design is widely used in the electronic industry to cool large servers and processors and other pieces of equipment that create a significant amount of heat.  While these are used in a smaller scale than our application, they utilize similar concepts in their design, equipment selection, and testing.  We obtained research from the University of Tennessee regarding some work done in a closed loop cooling product reliability test plan.  This research is helpful and relevant in our case as it provides valuable insight in other ways we can test the performance of the prototype and assure our clients that the overall system will be reliable for years to come.

Research was conducted to determine to physical as well as thermal properties of different types of metal that would be used in the assembly of the cold plate.   For instance copper has very high thermal conductivity and is easy to manufacture but is very heavy.  Aluminum is more difficult to manufacture and has a slightly lower thermal properties, but is much lighter than copper.    

Design Decisions

Due to the holes in each petal, the pipe geometry for each petal must be designed in order to both maximize cooling power as well as provide a uniform temperature distribution on the face sheet while leaving the holes unobstructed.  The pipe geometry was found to be the most critical piece of each petal.  This is due to the fact that the layout of the pipe beneath the face sheet directly correlates to the surface temperature distribution and the overall cooling power; critical design requirements.   From our research, copper seemed like an ideal choice given the ease of manufacturability and thermal properties needed for the pipe layout.  However, the interaction between copper and aluminum is corrosive and building the entire assembly of copper would exceed the weight limit specifications, so copper was ruled out.   We then to choose to use 1100 series aluminum alloy in order to best mimic the ideal qualities of copper while maintaining an acceptable weight for the assembly.

The face sheet was the next most critical element of the design.  The conduction through the face sheet needed to be high in order to allow for the most even temperature distribution possible.  Another necessary quality was to have a high emissivity to maximize the radiation between the mirror and the face sheet surface.   Other necessary qualities for the face sheet include being lightweight, flexible and manufacturability.   The final decision, in consideration of these qualities, was to utilize 1100 series aluminum that would be black-anodized.  Doing this maximized the heat transfer between the mirror and the face sheet.

The final two components of each petal are the panel frame and the insulation.  The panel frame is needed to provide structural integrity and give the petal its overall curvature.  We chose aluminum 6061, as it is lightweight and very strong.  The insulation is required to prevent the pipe from absorbing any external heat from the opposite side of the mirror. For satisfying this purpose, we selected semi-rigid PVC foam that had the R-value of 5.88 at one inch thick.  This insulation was chosen for its lightweight and ideal insulation properties.

Analysis

The importance of the pipe to the overall performance of the cold plate can not be overstated.  The layout determines both the surface temperature distribution, the cooling power, and the head loss experienced by the glycol.  We used Finite Element Analysis simulations to optimize the first two categories and excel calculation to select a pipe diameter that would minimize head loss while minimizing thickness as well.  The images from the FEA simulation can be seen below.  The pipe diameter was selected to be 1/2” and is detailed more in the final design page.

Preface

Shown Above: Top and bottom thermal images from heat lamp test

Shown Above: Top and bottom thermal images from ambient air test

Testing Methods:

1. Ambient air test— Glycol set at 2.5 degrees below ambient temperature

2. Heat lamp test— Utilize 250W heat lamp to simulate radiation experienced by the cold plate

 

Testing Results:     

 

Shown below in the table and thermal images are results from our testing.  In general, the results received from testing backed up our FEA simulation results.  However we identified possible sources of error that may have affected the outcome of our data.  This includes: a drafty lab room adding unwanted convective effects, resolution error in camera and thermocouples, and an incorrect application of the thermal paste that was applied to increase the contact area between the plate and face sheet but was put on in excess by Sulzer and may have hindered the contact instead. 

Significant Milestones:

· November 2009: Selected double-backed pipe geometry for pipe layout

· December 2009: Finalized shop drawings to be sent out for bids

· March 2010: Set-up testing apparatus and selected Sulzer Machine and Manufacturing to build prototype

· April 2010: Received prototype petal and began testing

 

Tools Used During Design:

· SolidWorks (A 3-D modeling software)

· CosmosFloWorks (A Finite Element Analysis (FEA) software that interfaces with SolidWorks)

· Microsoft Excel

 

Problems Encountered During Design:

No design team ever completes a project with out its fair share of problems and we were no exception.  Below lists some significant problems or “roadblocks” we ran into during the design process and how we dealt with them.

1. FEA Problem: It should be noted that none of the team members had taken a course in FEA, yet it was an integral part of the project as we needed to use analytical methods to show the surface temperature distribution on the face sheet.   The CosmosFloWorks has a somewhat difficult interface and can be frustrating at times.  In addition, radiation is very complex form of heat transfer and can be a difficult concept to visualize.  With these two subjects in mind, we had quite a hard time early on in the semester trying to use FEA and set-up correct boundary conditions for the analysis.  When we finally began to get results, the final problem we encountered with this was we weren’t able to save our results on the NAU network.   

FEA Solution: The above problem was solved by utilizing our resources at NAU; our technical advisor Dr. Brent Nelson, our Heat Transfer text book, and a computer lab with high performance PC’s.  This in combination with many hours spent in the lab doing trial and error analysis finally allowed us to create accurate boundary conditions and valid results.  Alec, who was the primary team member doing these analysis, also discovered that he could write the result files to his flash drive thus solving the problem of not being able save the file.

2. SolidWorks Problem: Unlike FEA, the team was experienced in SolidWorks, as well as other 3-D modeling software.  This experience allowed us to create the complex parts in SolidWorks without many problems at all.  However, without knowing it, we had inadvertently created models that had an incorrect contour in respect the primary mirrors.  This was discovered just days before the drawings were due to be turned into Lowell for final approval.  The problem, at this time, we didn’t understand why we were having this issue because it seemed that our contour should be correct. 

                 SolidWorks Solutions: The problem was potentially very severe as the deadline for completing these models was very close and we simply couldn’t send in incorrect drawings that would result in a very expensive prototype being built that wouldn’t work with telescope interface.  To solve this problem, we went back to the drawing board, literally.  We drew out on a white board the curvature of the mirror and how it is defined on SolidWorks versus how our petal was defined.  We eventually found a small error that would have resulted in a huge problem down the road, corrected it, and were able to turn in accurate drawings by the deadline. 

3. Manufacturing Problem: In the spring of 2010, we experienced a major setback where we were not getting enough bids to justify ordering the petal.  At the beginning we had only received one bid, and it was more than $60,000 over our original budget.  We were already slightly behind schedule that this point and it seemed just about every machine shop we contacted was either unable or unwilling to build our prototype.

                 Manufacturing Solution: The solution to this problem was a simple one: be persistent and be patient.  We ended up contacting nearly 25 different machine shops and finally toward March, we had received bids from 6 different companies.  From this, we were able to select Sulzer and keep our project under budget and still get a quality product for our prototype. 

                

Testing

An equally important of the analysis and design of the cold plate was the testing.  The testing would ensure that our design and analysis are sound in the cold plate would perform at or near where we predicted that it would. 

Testing Goals:

1. Locate hotspots on the plate surface

2. Determine the cooling power of the plate

3. Determine the head loss through the pipes

 

Measurement Equipment Used:

 

Inlet-Ambient Temperature (C)

Outlet – Inlet Temperature (C)

Cooling Power (Watts)

Day 1

2.6193

0.2107

8.43

Day 2

2.9508

0.3410

13.65

Combined Days Averaged

2.785

0.2758

11.04

FEA @ 30 C

2.5

0.2630

10.52

Measurement

Equipment

Thermal Imaging

FLIR i7 thermal imaging camera

Inlet & Outlet Temperatures

Type K thermocouples with DMM interface

Ambient Temperature

High accuracy mercury thermometer

Reservoir Temperature

High accuracy mercury thermometer

Head loss

Manometer