Proposed Plan to Arrive at a Corrected Design for the Tracker Module Flexure Mounts
R.P. Johnson, T. Borden
July 11, 2002
Re-run the FEA to derive loads: The analysis will follow a similar procedure as used in that presented at the June 19 meeting. The loads on the fasteners for the flexure mounts will be obtained from the FEM. Franz Beihl, who built the model but was on vacation prior to the June 19 meeting, will run the analysis, so the exercise will be a useful crosscheck. Confidence that these loads are reliable comes from the white-noise plots of Section 9.2 of LAT-TD-788. Whereas the large-amplitude transverse random vibration was affected by the remnants of the polymeric thermal gasket, the small-amplitude white noise was not. By the time that the thrust-axis vibration had been completed, the gasket was sufficiently well smashed that it no longer influenced significantly small-amplitude vibration. Therefore, in Section 9.2 it makes sense to compare the white-noise data against FEA done with no gasket. The plots show that prior to the 0-dB run (during which the sidewall fasteners backed out) the model does a good job of predicting the frequencies of the two lowest and most significant modes.
Infer stresses: Hand calculations will be done to infer from the FEA loads the stress in the carbon-carbon material surrounding the fastener and insert. The resultant shear force is calculated using the direct shear component, combined with the two in-plane moments, to get a net tensile force in the material.
Include assembly pre-stress: An analysis will be carried out to estimate the stress already present in the material from the assembly. The existing top/bottom trays will be examined to see how far the closeout walls must be pulled when tightening the fasteners to bring them into alignment with the flexure bracket. From this a pre-stress in the carbon-carbon will be calculated and added to the result from the FEM.
Consider the loss of a fastener: The FEA and stress calculations will be repeated with one, then two, fasteners in a single corner flexure completely removed to see how much the loads are increased on the other corners. From this we will see how well we can understand the propagation of the failure to all 4 corners.
The safety factors and stress factors used in calculation of the margin with respect to the allowable stress, as well as the allowables themselves, will be reviewed:
Since the stress concentration and safety factors are not very rigorous, it is desirable to do something more if we are to understand better whether the failure was to be expected given the design. Rather then develop a detailed FEM of the bottom tray, we propose to make a direct measurement. Using extra closeout material already in hand, we will fabricate several coupons, with each coupon representing a closeout wall with a single insert installed exactly as in the bottom tray and at the same distance from an edge. A bolt will be passed through the insert and attached to a clevis. Using this setup we will apply shear loads independently (i.e. one direction at a time) in directions parallel to the edge as well as toward the edge, according to the values predicted by the FEM (plus 3-sigma). If no cracking is observed between the insert and edge, then the loads will be increased until the material does fail. The results will be compared against the allowed stresses.
Samples from the lots of carbon-carbon material used in the prototype tray fabrication were tested to be consistent in density with the coupons used in the materials testing. This was done, of course, before machining the blanks into closeout pieces. The blanks were sent back to the ovens for additional processing if the density was too low. We will check that the densities of the particular machined carbon-carbon pieces use in the failed article are up to specification. If they are, then the allowables derived from the coupons should be applicable.
We do not believe that looking for specific defects in the region of the actual failure is likely be a fruitful exercise. Therefore, we are not inclined to pursue that course of action unless the analysis in Section 1 strongly suggests that the material should not have failed.
Our goal for the redesign will be to ensure enough strength in the flexure mount that
This calls for use of conservative safety factors plus metallic reinforcement for the corners.
The light-weighting pockets are being removed from the design of the closeouts of the bottom tray. Also, an extra 5 mm of material is being added between the inserts and the tray edge, making the bottom trays (and the Tracker module) that much taller. These modifications result in the bottom and top tray panels being different, which adds a small but manageable additional complication to the Tracker manufacturing. We have already changed the order for the flight carbon-carbon material (for which the procurement is in progress) in order to obtain sufficiently large pieces for the bottom trays. Ongoing work on the Tracker fabrication drawings already assumes these changes.
The inserts for the flexure mount attachment will be modified to have a much larger head, in order to better distribute the load onto the carbon-carbon when the fasteners are tightened. The insert material will be changed to titanium or steel. We have looked at the suggestion to add a helicoil inside the metal insert and have concluded that the insert would have to be much too large to fit in our design. An alternative is to remove the fasteners and inserts altogether for the flexure mounts, relying only on bonding (see the list of issues in the following section). This would have the virtue of removing stress concentration around the fastener.
A metallic bracket will be designed to reinforce the corner of the bottom tray where the flexures mount. The objective will be to spread the load from the flexure mount and reinforce the tray closeout with a material of higher strength and more assured mechanical properties than we have in the carbon-carbon closeout. The load spreading will be accomplished by the bracket design and also by bonding it directly to the carbon-carbon. Several sketches have already been made, but a number of issues need to be resolved and agreed upon before proceeding:
After a specific design concept has been chosen, the FEM will be updated and analysis carried out to verify that the design is robust, with large margins and good redundancy.
The design will be validated prior to random-vibration of the engineering model by running load tests on a second equivalent bottom tray fabricated in parallel with the engineering-model tray.
We are beginning to research secondary locking mechanisms for the fasteners. Our idea since long ago has been to use an adhesive, and our philosophy was that the design should be demonstrated to pass the testing first without the secondary locking (i.e. it truly was considered to be secondary insurance). As suggested by the ART, we are taking a closer look at using helicoils. We know already that helicoils are a disaster if used directly in the carbon-carbon material, so they would have to be threaded into the metal inserts. This does not appear to be feasible given the existing size constraints. A method for secondary locking will be settled on before completion of the engineering model.
Beryllium-Copper Gasket: We now are convinced that using a polymeric gasket leads to serious problems with interpretation of test results, quite apart from the difficultly of finding a material that doesn’t flow too much. We like the suggestion from Lou Fantano at GSFC to try a springy metallic gasket. A commercial beryllium-copper gasket that clips in place (into holes in either the Grid or the Tracker closeout) has been found. We will start by making some calculations to estimate whether this will give use high enough conductivity. If so, then we will design some thermal tests based upon samples of this product.
Modified Grid Interface: If we go with this gasket concept, then we will also consider revising the interface of the flexures to the Grid, going back to the original design of bolting the flexure brackets from the side directly into the Grid. The design was changed to one in which the flexure bracket is bolted with vertical bolts into extensions protruding out from the grid. This is a softer mount, but it allowed us to use the vertical bolts to compress the polymeric gasket. Going back to the original design would simplify the interface and raise the lowest transverse-axis resonance frequency of the module from about 90 Hz to 125 Hz.
To hold to our schedule for the engineering model, and hence the CDR, we would like to finalize the design of the bottom-tray closeout by the end of August. The carbon-carbon material in the new size will be available by then, and this will allow the Italians to proceed with fabrication of the bottom tray. Delays beyond then will reduce the testing time available prior to CDR.
Engineering of the Tracker mechanical structure has already significantly overrun our initial plans, so finding resources to carry out this program is problematic. About $45,000 remains in a contract is already in place with Hytec to carry out engineering and testing related to the change in sidewall material to one with higher thermal conductivity. Since that work has a lower priority than correcting the existing design, the contract is being modified to redirect some of these funds to be used to carry out analysis in support of the ART. The remaining funding issues will be tackled once the ART has completed is deliberations and we are in position to put into place a definitive, approved plan.