Mechanics/Cooling

U.S. Silicon Strip Upgrade MeetingM. Gilchriese

August 12, 2008

Overview

• Work since June 6 review by LBL/W. Miller• Status of stress studies/prototypes• Petal prototype, in collaboration with Valencia• Overlap with pixel mechanics/summary of

pixel plans• LBL strip mechanics plans and schedule with

currently expected resources

Since June Review

• For some months prior to the June review, almost out of $, could only maintain some design effort(W. Miller) (including for pixels) + document for review

• LBL provided some additional funds about one month ago => restarted prototype effort (also for pixels) and buying materials for future prototypes. These funds must be spent by end FY08.

• Strip work has been concentrated on stress issues (described next), collaborating with Valencia on petals and BNL/Yale interchanges.

Stave Prototype Construction• Stress concern is tube contracts as T is lowered, facing nominally expands slightly and

this can put shear stress on foam(and adhesives). Does this cause cracking/failure?• Aluminum tube coupled to foam or facing using compliant, gel-like adhesive CGL-7018

– Used for pixels. Shear modulus(G) is temperature, age(since dispense) and radiation dependent. Detailed data do not exist, simple tests for pixels decade ago suggest G ≤ about 1 MPa cold and after irradiation. Vendor spec on shear strength is < 2.1 MPa(about 300 psi) but this does not take into account temperature or radiation and concept of shear strength for this stuff is not so easy to define

• Foam to facing also used CGL, except on one side of one prototype, where we used EG7658, a compliant adhesive, with G 100 Mpa and shear strength of about 7 Mpa. But properties are temperature dependent(gets stiffer at lower temperature) and not well known. EG7658 also used for pixels, in high radiation region, to join Al to Al

CGL-7018

CGL-7018

CGL-7018

EG7658

Strip Stave Prototype Results• All three prototypes(about 34 cm long) were thermal cycled 50 times from

about 20C to about -35C• Thermal performance afterwards was same as before thermal cycling• But about all this shows is that there is no catastrophic failure. Stress

concentrated very close to end of tube(see later slides), very local failure only(in last few mm where tube exits) would not be observable in these tests.

• We have also thermal cycled a pixel prototype(about 6 cm long) that has similar construction albeit with a different type of foam and have looked directly (visually and thermally) at the edge where the tube exits. No effect seen, but very limited experience so far

• Its worth noting that there is experience outside HEP that is relevant. Space radiator panels in which metallic tubes are either brazed or hard-bonded to POCO foam in a panel structure. These see T excursions of hundreds of C. Some pictures on next page, ask for references.

Outside HEP

Figure 1: Titanium channels and titanium tube brazed to POCO foam (0.6g/cc) at 800C. Long dimension of POCO foam brazed directly to titanium is 4in (10cm). Channels were pressure tested to 300psi without separating from POCO Foam at brazed interface.

Figure 1: Brazed POCO Foam heat pipe assembly. Long dimension is 1.4m. Each POCO foam segment is 4in.

Modeling

• Measurements of POCO foam properties not in good agreement.

• Data for pixel low-density foam candidates hardly exist • Adhesive properties not well known, particularly at low T• Results for stress also depend strongly on assumed

bond-line thickness, concept of bond-line between porous foam and tube not well defined

• Will show some results but bottom line is that have to make measurements and then adjust model to agree and see if it makes sense.

Foam Properties

• We have not done an exhaustive compilation. • POCO foam Young’s modulus 0.1-0.5 GPa• Likely not uniform(remember thermal conductivity is

not uniform) and difference between compression and tension.

• Tensile strength ~ 1-2MPa• Compressive strength ~3MPa?• Plastic behaviour?

Models• Example below shows the normal stress in the z-

direction on the POCO foam with E=0.5 GPa (worst?) and with G=1.4 MPa for both tube and facing adhesive.

• To show that foam near tube is in compression and in tension near facing

• Comparison of (simplified) 60 cm and 4cm model demonstrates peak stresses near end ~ same => can use short model

Facing

Facing adhesive

Foam

Tube adhesiveTube

Example Results – Peak Values

• Values depend on bond-line thickness (4 mil here)

• Adds some confidence to prototype results but not definitive, in my opinion

• Hard to understand margin from this

Aluminum Tube/Adhesive (4mils)/ E=0.5GPa, σy=1.75MPa

New Model (POCO Foam minimum side thickness 1.9mm)

Model Length 4cm

MPa Shear (σyz)

MPa Normal

(σz)

strain MPa Von Mises

(σvm)

Foam CTE

Elements

POCO Foam increased width

0.257 0.61 0.0019/0.0027

0.916/1.36 (197psi)

4 8 node brick

Tube Adhesive G=1.4 MPa

0.34 0.0031 n/a 0.593 (86psi)

4 8 node brick (4 mils thick) 2layers

Facing Adhesive G=1.4 MPa

0.191 0.5 n/a 0.331 (48psi)

4 8 node brick (4 mils thick) 2layers

Aluminum Tube/Adhesive (4mils)/ E=0.5GPa, σy=1.75MPa

New Model (POCO Foam minimum side thickness 1.9mm)

Model Length 4cm

MPa Shear (σyz)

MPa Normal

(σz)

strain MPa Von Mises

(σvm)

Foam CTE

Elements

POCO Foam increased width

0.372 0.303 0.0017/0.0034

0.855/1.59 (231psi)

4 8 node brick

Tube Adhesive G=1.4 MPa

0.453 0.022 n/a 0.794 (115psi)

4 8 node brick (4 mils thick) 2layers

Facing Adhesive EG7658

0.284 2.44 n/a 2.44 (354psi)

4 8 node brick (4 mils thick) 2layers

Stress Samples and Testing

• Made small stress samples– Visually inspect and thermal cycle from +25 to -30 fifty times– No cracks in foam observed after T cycle….but maybe can’t see

• Will measure stress/strain directly on these samples (and others) by pulling on tube relative to facings => direct measure of strength of tube/adhesive and foam interface.

Sample Priority Tube Type Foam Type Tube/Foam Adh Foam/Facing Adh Facing A 7 Al (2.85mm OD) POCO CGL 9396/30%BN 90-0-0-90 K13D2U B 1 Al (2.85mm OD) POCO EG7658 9396/30%BN 90-0-0-90 K13D2U C 3 Al (2.85mm OD) POCO 9396/30%BN 9396/30%BN 90-0-0-90 K13D2U D 6 Al (2.85mm OD) Allcomp2 CGL 9396/30%BN 90-0-0-90 K13D2U E 2 Al (2.85mm OD) Allcomp2 EG7658 9396/30%BN 90-0-0-90 K13D2U F 4 Al (2.85mm OD) Allcomp2 9396/30%BN 9396/30%BN 90-0-0-90 K13D2U

What Should be Done Next?• Ideally, careful program of measurement of samples of foam

and adhesives, including temperature dependence to extract moduli and strengths.

• Adjust model to achieve agreement with data.• Will take time (~year) and not so easy to understand but

should be done• In parallel, build prototypes and test• Full-length, thermal cycle necessary to convince people• Note for strips, could use POCO HTC (or CVDed POCO) that

has higher strength and higher K but also twice density, which would increase X0 by ~ 0.1

Connection to Pixels

• We are pursuing pixel concept that has many similarities with strip stave except that low-density, thermally conducting foam is used throughout

• Even more important to understand stress issues. Foam more fragile

Comment on ~ 3mm Strip Stave• To achieve close to 3mm thickness of stave I believe requires

tube/foam concept • Cannot cut foam to “knife edge”• Prototype made (although tube 4.9mm OD) in this way and

thermal performance good. • Did machining test with POCO foam to see what 3mm looks like

(also a trial for pixels). Could be done (a bit messy).• Tube exposed to facing. “Masking” small region on top/bottom of

tube to prevent adhesive from connecting tube to facing difficult(impossible?) = > compliant adhesive

• But analysis different than case of foam surrounding tube. Must model adhesive and contact with facing….not done by us

Petal

Petal Overview• Valencia is doing the overall design, FEA and detailed design

of prototype and tooling (or almost all tooling).• Will be conservative construction, basically same as LBL

short prototype stave but in petal shape• Our role restricted to

– Comments on the design of the prototype– Fabrication at LBL of 1 – 2 full-size prototype petals– Hand off fabrication knowledge to Valencia…– No design or FEA responsibility

• Expect Valencia to take petal core(s), instrument and test, starting ~ November.

Petal Content and Status

• Honeycomb ordered(for other prototypes as well), delivery end September.

• Make and cut plates (3 ply K13D2U, one started) by September -> facings

• Expect stainless tube from Yale. Bend at LBL• Expect foam from Yale. If we can get foam soon and know tube

diameter, can cut at LBL before end September. If delayed into October, Yale may have to do it ($ problem in FY09)

• Assemble at LBL, with Valencia personnel in attendance, at least for first petal. No earlier than October. Exact schedule TBD.

• 2nd petal fabrication at LBL depends on how all this goes($ problem in FY09). Cheapest to do both at once.

Foam Facing to Facing to Foam to Build Item Width(cm) Length(cm) Thickness(mm) Cell Size Type Ply Layup Material OD(mm) ID(mm) Material Type Honeycomb Foam/Tube Tube Location Purpose

Petal 1 8-20 ~70 5 3/16" M46J/EX1515 3.5 lbs/cu ft 3 0/90/0 K13D2U ~2.8 ~2 SS POCO/Koppers 9396 EGxxxx or CGL CGL LBL Mechanical/thermalPetal 2 8-20 ~70 5 3/16" M46J/EX1515 3.5 lbs/cu ft 3 0/90/0 K13D2U ~2.8 ~2 SS POCO 9396 EGxxxx or CGL CGL LBL Module mounting

AdhesivesDimensions Honeycomb Facings Tube

One Page Summary of Pixel Mechanics Plan• Have made and tested thermal performance of 4 small prototypes with

foam from 3 vendors and will make more small prototypes as vendors develop foam. All good so far and thermally concept looks promising

• Detailed design optimization(thru FEA) of facings(carbon fiber, orientation, thickness, is carbon-carbon better…) foam(K), tubes (including Ti, SS) to be done thermally and mechanically.

• Foam and adhesive properties, including understanding related adhesives (eg. diamond loaded instead of AlN)

• Models and small prototypes to understand stress issues• Allcomp, Inc via SBIR plans to make ~ ½ length stave with their foam

that would be instrumented and thermally tested at LBL• Significant overlap with strips in some parts (mostly stress issues) but

also a lot of work to be done, more than for strips

Current LBL Strip Mechanics Plans• FY09 mechanics(pixels+strips) budget currently ~ 70% of FY08• Continue stress calculations and prototypes. How long this will take?

Guess ~ early 2009 for initial stress understanding validated by sufficient prototypes(for strips, much longer for pixels)

• Fabricate 1, hopefully 2, petal prototypes by ~ November 2008• Hand-off petal fabrication knowledge to Valencia by ~ end 2008• Compare thermal FEA between W. Miller and Yale(Emmet) and hand-

off by ~ September.• Continue advisory role on design/prototypes ~ end 2008• Retain part of FY09 funding for pixel work, don’t spend all on strips,

which means phase out of strip work• Concentrate on pixel mechanics from 2009 onwards