From: Peter Titus Sent: Saturday, August 27, 2011 7:56 PM To: Robert D. Woolley Cc: Orlando Guzman; Bob Simmons Subject: RE: Moment Coefficients Attachments: MomentInfluenceCoefficientsTitus.doc; influence9.xlsx Bob: I corrected the PF4L N-m/rad line of data. There was one cell that had two of the coefficients in it and that set off the rest of the row. As you saw, the english unit tables were OK. Our difference in the PF4 due to PF5 coefficient, I think is another case of a difference in small numbers. The FX PF4 due to PF5 coefficient and the FY PF4 due to PF5 coefficients are large. 444.4 and -368.7, so the difference in our moment coefficients of 1.6 to 3 is an error of less than a centermeter in force center vs geometric center. I should be doing better, but the mesh is still relatively coarse in my model and the modeling of the two pancakes in PF4 may be slightly different. I think that either of our sets of numbers will give adequate structural assessments. I think we should be able to sign each others calculations of the moment coefficients and defer any refinements to when the DCPS is being exercised. Again, thanks for your patience and diligence in checking my numbers. -Peter From: Robert D. Woolley Sent: Wed 7/20/2011 9:04 PM To: Peter Titus Subject: RE: Moment Coefficients Hi Peter, I took your row of in-lb/coil cefficients for PF4L torques, multiplied them by (9.8/2.2/39.37/2/PI( ) ) then substituted the result for your incorrect row of PF4LSI-unit coefficients in the spreadsheet that I was using for comparisons. Now your zero element is in its proper diagonal place and your other resulting coefficients relating PF3L. PR4L, and PR5L are closer to your coefficients relating PF3U, PF4U, and PF5U. This is a clear improvement. When I compare these coefficients to my own calculated coefficients there are discrepancies that either (a) must be tracked down and eliminated or (b) must be justified as ignorable. For the smallest magnitude coefficients, approach (b) is easy. However, the largest remaining discrepancies are for the moment interactions between PF4 and PF5, for which the numerical coefficient comparison is as follows: N-m/radian/(kA)^2 N-m/radian/(kA)^2 Woolley PF4UL ckt PF5UL ckt Titus PF4UL ckt PF5UL ckt PF4U -0.1258 +3.0808 PF4U -0.0710 +1.6309 PF5U +0.3315 +0.0119 PF5U +0.3479 +0.0163 PF5L -0.3315 -0.0119 PF5L -0.0112 -0.0158 PF4L +0.1258 3.0808 PF4L +0.2111 -1.7206 Here, my values for moment coefficients from the PF5UL circuit to PF4U and PF4L coils are almost double your values, and they are much larger than all other values in this abbreviated table. They are also larger than moment coefficients in the full table between these coils and any of PF1, PF2, PF3, or OH coils, for which approach (b) would be easy. To justify ignoring the large difference between 3.0808 and 1.6309 coupling the PF5UL circuit current with moments on PF4U, basically we would need to show that all moment effects can be safely ignored for PF4 windings. At this point, either you must redo your calculations to better estimate this coefficients for PF4 coils or an argument must be made that all torque effects can safely be ignored for PF4. Either way, the calculation write-up must be modified. I'm not sure whether torque effects can safely be ignored for PF4 since PF5 and PF4 are geometrically quite close. I believe my own 3.0808 coefficient values are accurate and that an argument justifying ignoring the moments would need to show that the maximum torque on PF4U or PF4L can be safely ignored for all coil current cobinations. Note that my upper and lower coefficient values match each other in magnitude to 5 significant digits of precision even though they were calculated separately; this accuracy is typical of the greens function method using elliptic integrals when combined with the very large number of circular filament elements that I used to represent each coil winding. It does seem mysterious that our coefficient results agreed welI for inner coils but deviated for outer coils. I do not know how to further review your calculations since I am unfamiliar with the algorithms, files, and software you used to make them. We should probably discuss this in person, perhaps tommorrow morning ?. An argument that all torques on the PF4 coil cross sectios can be safely ignored should probably be based on analyzing the protection algorithm for the PF4 coils' supports. What is the status of the protection algorithm for PF4 supports ? Best regards, -Bob Woolley ----------------------------------------------------------- At 03:53 PM 7/20/2011, you wrote: check the in-lb/kA^2 data. I think the problem was in the SI unit version. -My program prints out a row in 3 lines and I have to manually convert it to one line. I think I messed up the conversion in the SI units. -Peeter From: Robert D. Woolley Sent: Wed 7/20/2011 3:51 PM To: Peter Titus Subject: Moment Coefficients Hi Peter, After our telecon I have just now finished comparing your tabulated moment coefficients with mine, calculating the per cent deviation for each coefficient. Most deviations are less than one percent. Others which are 5 to 20% different are of very small magnitudes so their larger percent deviations don't matter. This is the same pattern which you noticed for the single-winding coils. However, all coefficients contributing to the moments on the PF4L coil body have deviations of around 100% or larger, and furthermore as I told you on the phone your self-coefficient of moment for PF4L is significantly non-zero which is impossible for physics reasons. My conclusion for now is that I don't recognize any other significant discrepancies in the tables, other than the problem with PF4L moments. After you find and fix that problem I should check the resulting numbers again just to make sure, but I don't expect to find any significant discrepancy then. After signing off the calculation review form, I will do a quick study to see how much difference the moment coefficients can make in coil hanger forces, worst-case. I expect you are right that many of the moment coefficients can be ignored as insignificant. -Bob Woolley