From: Michael R. Kalish Sent: Friday, January 28, 2005 10:50 AM To: Wayne T. Reiersen Cc: 'Brad Nelson'; 'leonard.myatt@myattconsulting.com'; Steve Raftopoulos Subject: RE: Testing Criteria Wayne, Yes, Yes, & Yes I think we are on the same page.... Here's my take with detail added... pretty close to what you asked for: Test Purpose: 1. To verify and benchmark the mechanical properties used in the analysis 2. To verify that the mechanical integrity of the system after a prescribed number of cycles. The criteria for this are that the stiffness of the beam does not become less stiff than the analysis predicts and that there is no other visible mechanical degradation. 3. To verify the electrical integrity of the system before during and after cycling. Method: 1. HiPot specimen before testing 2. Measure beam properties at room temperature 3. Measure beam properties at cryo temperature 4. Measure beam properties at room temperature 5. Hi Pot specimen 6. Thermally Cycle beam 15 times? (again difficulty must be weighed against requirement to determine number cycles actual number of thermal cycles at life is 140) 7. Cycle at cryo temperature 130,000 cycles at 8,200 lbs. To evaluate the degradation in stiffness (if any) we will slow the test and take data at first every 500 cycles loosening up to every 13,000 cycles after the first 26,000 cycles. (Is this data available for every cycle anyway?) 8. Hi Pot specimen 9. Continue cycling to a total of 260,000 cycles 10. Hi Pot specimen & measure beam properties at room temperature Note: 260,000 cycles provides us with a 20x life test at stress for the .5 Tesla requirement (note .5 Tesla cycle requirement = 10% of total = 1,300 cycles). With respect to material stress allowables in general Len has addressed these in his design analysis memos. Separate criteria for the insulation and the copper are set out there and references are provided. The verification of these values will be part of the calculation check. Mike Michael Kalish Princeton Plasma Physics Laboratory PO Box 451 Princeton, NJ 08543 Phone: (609) 243-2277 Fax: (609) 243-3248 You can visit the home page of the Princeton Plasma Physics Laboratory at http://www.pppl.gov -----Original Message----- From: Wayne T. Reiersen Sent: Friday, January 28, 2005 10:12 AM To: Michael R. Kalish Cc: Brad Nelson; leonard.myatt@myattconsulting.com Subject: RE: Testing Criteria Folks, We should start by crisply defining the purposes of the test, not labeling it as a "prototype test" or a "material test". I would have thought that the testing should serve the following purposes: 1. Verify that we are using the right material properties. Measure deflection versus load to verify that using the properties we assumed in our structural models, we predict the correct deflection. We had a major surprise here when modeling the modular coils in that the shear modulus of the composite beam appeared much lower than we predicted from our FEA models. For this purpose, we only need a few cycles at modest stress levels. 2. Verify that our design allowables are appropriate. Verify that over the life of the device, operating at the maximum allowable stresses, the mechanical response is stable (plateau-ing might be OK) and the electrical insulation properties do not degrade unacceptably. For this, I would think we would want to start at modest stress levels, run for 2x (?) life, run a load-deflection curve again, test the voltage standoff, and crank up the load, repeating the process until the beam starts pooping out (hopefully, at stress levels above our maximum allowable). Your thoughts? I would also appreciate clarification on how we are setting our design allowables. Do we have separate allowables for the copper and glass-epoxy insulation? What are they? If so, how were the allowables determined? How were the peak stresses in these materials determined? If this is all documented somewhere (and it needs to be), just point me to it. Thanks, Wayne -----Original Message----- From: Michael R. Kalish Sent: Friday, January 28, 2005 9:26 AM To: 'Nelson, Brad E.' Cc: 'leonard.myatt@myattconsulting.com'; Wayne T. Reiersen; Steve Raftopoulos; Irving J. Zatz Subject: Testing Criteria Brad, Brad, Missed catching up with you yesterday. One thing I wanted to go over with you was this testing criteria. I had a phone conversation with Len about this subsequent to the emails and we came to a consensus. We believe that the load arrived at in Lens analysis (8200 lbs for .5 Tesla) is representative of actual run conditions. That is the stress and deflection developed initially by this load may indeed be lower than the max. stress calculated for the softest beam assumption.... but ... because the load was arrived at using the same modeling technique (slip planes for the Kapton) on both the TF Global model and the sample test bar model we feel that the load is correct and the stresses and deflections will reflect actual operating conditions. In the test as in the actual TF coil we expect the stresses and deflections to start at a lower level and as (or if) the Kapton layer breaks free of the copper conductor stress and deflection will increase with the number of cycles plateauing at the max. values obtained in the calculation. I'm still developing what the number of cycles and test levels should be. I would argue that because this is a prototype test and not a material test we are not strictly limited to the 2x Stress at Life Criteria. In the end though I want this test to provide enough conservatism that we are confident in the insulation scheme. Mike -----Original Message----- From: Michael R. Kalish Sent: Wednesday, January 26, 2005 4:42 PM To: 'leonard.myatt@myattconsulting.com'; 'Nelson, Brad E.' Subject: RE: Testing and TF Design View Graphs Ok.... I'm working through this correct me when I'm wrong. Len in your first email you said ... " If the WP is stiffer than my conservative frictionless slip-plane model predicts (and I expect it will be), then you might need more force than the estimated 8200 lb to reach the 196 MPa" ... but isn't the stress a function of only the geometry and the force or does that not apply for a composite material that goes nonlinear as it unzips? If the stress is constant and the deflection is smaller due to the insulation adhering I would argue that we are mimicking what the real coil will do in service. It will begin life at smaller deflections and possibly become less stiff in service with the deflection / strain increasing with time. So the strain over life may be more representative with a constant load test than if we try to match the maximum strain by increasing the load. Hopefully we will approach some representative worst case simulation of the strain on the insulation during the life of the project. Then if the sample passes HiPot we can assume the dielectric properties of the insulating system survived a representative fatigue test. ...Or is it indeed that without a strain measurement we can not know for sure what the stress is. Michael Kalish Princeton Plasma Physics Laboratory PO Box 451 Princeton, NJ 08543 Phone: (609) 243-2277 Fax: (609) 243-3248 You can visit the home page of the Princeton Plasma Physics Laboratory at http://www.pppl.gov -----Original Message----- From: Leonard Myatt [mailto:leonard.myatt@myattconsulting.com] Sent: Wednesday, January 26, 2005 9:22 AM To: 'Nelson, Brad E.'; Michael R. Kalish Subject: RE: Testing and TF Design View Graphs I suppose the test will show if the system is linear right away. But, I don’t expect it to be linear over the entire anticipated load range. Do you propose to simply cycle it between say 1000 and 8200 lb regardless of it’s performance? As a load controlled structure, that may be OK. I suppose you could argue that the 196 MPa came from a zero friction global model and a zero friction beam model. And even if it sticks in real life, then the load doesn’t change, and the 8200 lb is right. OK. I don’t feel so bad about the approach. We should just be aware that the model may be overestimating the deflections and stresses. Regards, Leonard (Len) Myatt Myatt Consulting, Inc. 8 Eric Road Norfolk, MA 02056 leonard.myatt@myattconsulting.com 508-520-4590 (tel.) 508-813-6843 (cell) 877-883-4961 (e-FAX & v-Mail) www.myattconsulting.com -----Original Message----- From: Nelson, Brad E. [mailto:nelsonbe@ornl.gov] Sent: Wednesday, January 26, 2005 7:42 AM To: leonard.myatt@myattconsulting.com Subject: RE: Testing and TF Design View Graphs Len Isn’t the strain a geometry thing for a linear system, such that a given deflection produces a given strain at the extreme fibers? I agree that if our system is non-linear, then we cannot infer the strain from the deflection. We cannot infer the stress from the strain unless we know E, and for our beam the shear modulus may also be important. We are testing some fiber optic strain gages to see if they work, maybe we can stick one of those on the bottom of the beam. Brad Brad Nelson Oak Ridge National Laboratory P.O. Box 2008 Oak Ridge, TN 37831-6169 nelsonbe@ornl.gov voice: 865-574-1507 fax: 865-576-7926 ________________________________ From: Leonard Myatt [mailto:leonard.myatt@myattconsulting.com] Sent: Tuesday, January 25, 2005 8:26 PM To: Nelson, Brad E.; Mike Kalish Subject: RE: Testing and TF Design View Graphs Having strain gauge produces results that can only be inferred otherwise. I also appreciate that it is probably a nasty thing to get right (embedded in the VPI cure, operating at 77K). Backing out stress from load and deflection might be tricky. Maybe I just need to think about it some more. Maybe I could integrate the poloidal forces across the WP X-section (where the stress is a maximum) and determine the effective bending moment from EM + Preload. Then, we could design the three point test load to produce that moment. >>Load controlled, not stress-controlled. Regards, Leonard (Len) Myatt Myatt Consulting, Inc. 8 Eric Road Norfolk, MA 02056 leonard.myatt@myattconsulting.com 508-520-4590 (tel.) 508-813-6843 (cell) 877-883-4961 (e-FAX & v-Mail) www.myattconsulting.com -----Original Message----- From: Nelson, Brad E. [mailto:nelsonbe@ornl.gov] Sent: Tuesday, January 25, 2005 5:50 PM To: leonard.myatt@myattconsulting.com Subject: RE: Testing and TF Design View Graphs Len We were not planning on using strain gages. Since it is a load controlled test and we measure the deflection, we hoped that the deflection would tell us how the stress was apportioned between the copper and insulation. Is that right? Brad Brad Nelson Oak Ridge National Laboratory P.O. Box 2008 Oak Ridge, TN 37831-6169 nelsonbe@ornl.gov voice: 865-574-1507 fax: 865-576-7926 ________________________________ From: Leonard Myatt [mailto:leonard.myatt@myattconsulting.com] Sent: Tuesday, January 25, 2005 10:43 AM To: 'Michael R. Kalish'; Nelson, Brad E. Subject: RE: Testing and TF Design View Graphs Michael & Brad, I took the liberty of modifying my beam model to look like the 42” prototype WP and 3-point bend test. If this is not consistent with the final test configuration, than it is an easy re-run. See attached spread sheet (tab: 3x4 Prototype Test) Looks like the present 10,000 lb test capacity might be OK; 8200 lb produces 198 MPa in this model. Comments: * Do you plan on applying strain gauges to the bottom mid-span location of the Cu bars to confirm that the test is producing the necessary stress level? * If the WP is stiffer than my conservative frictionless slip-plane model predicts (and I expect it will be), then you might need more force than the estimated 8200 lb to reach the 196 MPa (0.5 T) stress level. * I see no reason to add torsion to the test. Regards, Leonard (Len) Myatt Myatt Consulting, Inc. 8 Eric Road Norfolk, MA 02056 leonard.myatt@myattconsulting.com 508-520-4590 (tel.) 508-813-6843 (cell) 877-883-4961 (e-FAX & v-Mail) www.myattconsulting.com -----Original Message----- From: Michael R. Kalish [mailto:mkalish@pppl.gov] Sent: Monday, January 24, 2005 3:25 PM To: leonard.myatt@myattconsulting.com Subject: FW: Testing and TF Design View Graphs Len, Can you back out the force required for the test fixture to simulate the max. bending stress for the 1/2 tesla case. Lets assume that the end supports are 1.5" in at either end. Looks like the max force the test fixture can attain (without modifications) is about 10,000 lbf (see Brad's input below) Michael Kalish Princeton Plasma Physics Laboratory PO Box 451 Princeton, NJ 08543 Phone: (609) 243-2277 Fax: (609) 243-3248 You can visit the home page of the Princeton Plasma Physics Laboratory at http://www.pppl.gov -----Original Message----- From: Nelson, Brad E. [mailto:nelsonbe@ornl.gov] Sent: Saturday, January 22, 2005 5:52 PM To: Michael R. Kalish Subject: RE: Testing and TF Design View Graphs Mike See answers below Brad Brad Nelson Oak Ridge National Laboratory P.O. Box 2008 Oak Ridge, TN 37831-6169 nelsonbe@ornl.gov voice: 865-574-1507 fax: 865-576-7926 -----Original Message----- From: Michael R. Kalish [mailto:mkalish@pppl.gov] Sent: Friday, January 21, 2005 5:01 PM To: Nelson, Brad E. Subject: Testing and TF Design View Graphs you're email choked on the large attachments so I'm sending it in batches Brad, Here's the drawing of the test sample and some Power Point slides of the TF Design to think about before we meet Thursday. Is there anyway for me to work directly with the test group to determine what is possible? I'm working out a test plan and it would be most efficient I think to interface with whoever is performing the test directly. THE NAME OF OUR TEST PERSON IS RICK BATTISTE, 865-574-0736. I MENTIONED THAT WE MAY HAVE ADDITIONAL TESTING TO PERFORM, BUT I DID NOT HAVE THE DRAWING TO SHOW HIM AT THAT TIME. I CAN TAKE A COPY OF THE DRAWING TO HIM SOME TIME ON MONDAY. Some ideas I'd like to work through: * If I remember you said you do not have the capability to high pot the sample? I WILL CHECK THIS AGAIN. WE HAVE HI-POT GEAR SOMEWHERE, I JUST NEED TO LOCATE IT * With respect to the cyclical testing the ideal would be to achieve 140 thermal cycles to simulate life. I suspect without some level of automation this will be an unrealistic goal. I AGREE, THIS WOULD BE HARD TO DO. * I would like to achieve 2x stress at life with respect to mechanical cycling (design life = 130,000 cycles). At 1HZ that's only a day and a half so this seems achievable. WHAT LOADING DO YOU WANT TO PUT ON THE SAMPLE? WE CAN DO 3 POINT BEND WITH OR WITHOUT TORSION, BUT MAY BE LIMITED TO ABOUT 10000 LBS DUE TO THE LOWER FIXTURE ROD. IF WE BEEF THAT UP WE COULD GO MUCH HIGHER It sounds like the earliest you may be able to start on this 2/14? I THINK THAT WOULD CORRECT. I DO NOT THINK WE NEED THE TEST RESULTS PRIOR TO THE PDR See you Thursday Thanks, Mike Michael Kalish Princeton Plasma Physics Laboratory PO Box 451 Princeton, NJ 08543 Phone: (609) 243-2277 Fax: (609) 243-3248 You can visit the home page of the Princeton Plasma Physics Laboratory at http://www.pppl.gov