"LESSONS LEARNED"

TRMM MECHANICAL & DEPLOYABLE SUBSYSTEMS

THAI PHAM, CODE 543

MECHANICAL ENGINEERING BRANCH

Last Revised: 10/28/98



TABLE OF CONTENTS

I.   DESIGN
II.  FABRICATION
III.  ASSEMBLY
IV. TESTING
V.  I&T
VI.  ALIGNMENT
VII. DEPLOYABLES
VIII.LAUNCH SITE OPERATIONS
IX.  OTHERS


 

 

DESIGN

1. Harness weight has the tendency to be grossly under estimated. Be prepared to accommodate more harness weight and volume. Also, get a hold of harness size and routing early so that time is available to design efficient harness support stands and hinge crossing supports.

2. Keep in mind that s/c batteries are likely to be the last component to be installed on s/c. Design GSE and flight hardware components to easily and conveniently accommodate battery integration at the last minute if possible.

3. Don't blindly believe FEM output. Perform hand checks to verify results. Draw a Free Body Diagram of the element which is being analyzed.

4. When possible, design multiple use GSE to cut down cost, training time, and things to take to launch site. Only disadvantage to consider possible schedule conflicts.

5. Be careful designing parts with compound angles. It’s very different to go 30 deg in X direction then 60 deg in Y than it is to go 60 deg in Y first then 30 deg in X.

6. When designing MGSE or flight structure, be sure to know all electrical interfaces (flight and non-flight safing plugs/backshells, an attachment that only shows up for testing, etc...) to design for or around to allow access room or prevent future interferences.

7. Beware of cutting tool accessibility for under cutting threaded items. The harder the material, the large the radius should be.

8. Place material hardness test location on the part where it won’t deform the part during the test. If part is sensitive, locate hardness test location on drawing.

9. Request piece part identification/drawing number to be placed/stamped on the part at location where it will still be visible after final assembly

10. Generate good instrument and subsystem MICD's. Ensure that the interfacing subsystem signs.

11. Hold frequent peer reviews. Encourage attendance from interfacing subsystems and instruments.

12. Use a non-critical weld philosophy in designing MGSE hardware. This will reduce the costly weld inspection over the life of the hardware.

13. When match drilling through dissimilar materials (i.e. aluminum and CRES or titanium), it is very difficult to achieve an uniform hole diameter in the tolerances needed for press-fit pins. This led to extra cycles of measuring holes, and fabricating customized, stepped pins to achieve the tight fits required in the deployment hinges.

14. For critical components, employ an informal team design review process regularly. This is when a design is presented to the whole team who then gets to evaluate it for what it’s worth. We held this review process on a weekly basis right after our weekly team meeting.

15. A good thorough layout incorporating all the parts to be designed is a must before detailing any parts. This is best done by the lead designer after which the smaller piece parts can be assigned to other designers.

16. Develop and distribute a drawing tree as you go, not at the end because its usefulness would be less and there is a tendency to avoid doing it at the end.

17. Involve the fabrication planning office, machinist, assemblers, and any other fab shop people who will be involved in producing the part in the design process for comments and ideas and to sieve out any unnecessary complications.

18. We experienced a fair amount of pain trying to abide with the metric charter of the agency especially in regards with metric fasteners. We took the requirement seriously while it seems that everyone else took the easy way out and waived the requirement. This is the type of requirement that can bury a program... and it almost did.

19. When designing a truss structure, use sections which are symmetric ( "I" , tubes, etc) so that moments don't have to be carried through the pinned end fittings.

20. Beware of phosphor bronze helicoils. While performing our extensive torque tension program, the data showed that the silver plated CRES helicoils clearly outperformed the phosphor helicoils while using lubricated A286 bolts. While performing the test we contacted Helicoil with our results and they were not surprised. They informed us that they only sell the phosphor bronze to the marine industry and to GSFC.

21. 7050-T7451 performed extremely well as a high strength aluminum alloy for sections over three inches where 7075-T7351 would move too much as a result of machining. Also, while machining 7075 sections under three inches in thickness, we found that stress relieving was not beneficial for our parts. In fact, stress relieving while machining tended to excessively distort the hardware.

22. Question all requirements!!

23. Use "max radius" whenever possible and mass isn't critical. It allows the machinist more choices in the tools in which he uses to fab the part.

24. Use a pre-existing parts/hardware list during the design phase. It reduces the amount of work and time to locate and purchase specialty parts, hardware, etc.

25. Check on the availability of a particular material during design. Order long lead materials as soon as design is mature, don’t wait until it goes into fab.

26. During the design phase, be aware of tool (Rivet gun, drills, wrenches) access for assembly.

 

FABRICATION

1. Ask the shop technicians if you can help make the part easier for them to produce and support their comment/suggestion when appropriate. This applies not only to changing the design but also to other areas such as providing the proper tools and resources for them to do their job (even though this is the shop manager’s job).

2. For critical parts or parts with tight schedule, the responsible engineer (or an engineer representative - pure scheduler not as efficient) must invest the time to visit the fab shop to interact with the planner and/or the fab techs to answer any "small" fab related questions or part related curiosity questions and to communicate the importance of the part. This not only keeps the engineer abreast of the part’s fab status but it motivates the tech to do a better job because the customer is real and he understands the importance of the part.

3. During fabrication, try and have all processes done in one location. It adds time to the fab process to go from one facility to another.

4. Check on items that are in storage for any lengthy period. Stuff happens to hardware when no one is looking.

5. For critical parts, visit the shop often to witness the work being done to find out or assess deviations first hand before the problem gets bigger and becomes too costly to correct. At times, an acceptable mistake will not have to be thrown out and redone if the engineer is involved in the assessment of the mistake. Many shop errors are assumed to be un-acceptable and therefore are re-started from scratch which can be very costly. Catching a mistake before the part comes off the fab line or get inspected can save a lot of time.

6. Be sure to regularly visit machine shops to ensure that your hardware is being processed. Phone call status updates are not always accurate.

 

ASSEMBLY

1. Inspectors catch many errors but not all errors 100% of the time. The engineer should take a cursory look at each part before it gets assembled. Sometimes the part was made correctly according to print but it turns out not to be what the engineer imagined.

2. Assemble components such that their part/drawing number is visible afterward.

3. For large and/or complex assemblies, the engineers should lead the assembly tooling design effort with the help of the assemblers themselves. This process saves a lot of time because it allows for the development of the tooling while the parts to be assembled is being designed.

4. When using hi-loks, consider performing a coupon boomer-banger test for lot acceptance. The test is a screen for the locking feature element. Normal lot inspection does not check for proper locking tightness.

5. Consider ordering spare parts for critical operations to account for boo boo's (in fab., plating, delivery, mishandling, etc…).

6. Always listen to your Techs. Even though you don’t have to agree all the time.

 

TESTING

1. Forward final performance test data (deployment time, potentiometer readings, thermistor readings etc.) to spacecraft control group so they can update their numbers and minimize confusion during launch time. (make this a formal requirement to all subsystems)

2. When possible use redundant instruments at critical locations to prevent halting/repeating test because of a bad reading. This could also help in using one accelerometer reading to check another.

3. Test article CG must be centered about shaker table vertical axis to minimize lateral loading during vertical vibration. (Facility should have a check list (Do's/Don’t) for all users )

4. When reworking electronic boxes, insist on at least a minimum workmanship random vibration and T/V tests. Thorough inspections and good procedures are not adequate replacements for these workmanship tests.

5. Be sure to use the flight level marman band preload for the Observatory sine test to ensure gapping doesn't occur at 1.25 times limit. The marman clamp gapped during our s/c simulator testing.

6. Clearly understand the signal processing electronics and the resulting output sensitivity for your instrumentation. In choosing a signal processing methodology, consider probable failure modes.

7. Involve test facility personnel as early as possible to start planning. One of their biggest gripe is lack of planning on the project’s part thus not giving them the adequate time to make the proper arrangements to proceed smoothly.

8. Sine Testing - During the Sine Test, we had automated the calculation of the base bending moment (BBM) from the acceleration time history data (G's vs. Time). We also calculated the BBM using the processed acceleration data (G's vs. Frequency) as a check. The BBM calculated from the processed data was consistently lower than that calculated directly from the time history data. Comparison of the time history data with the processed data showed significant differences in the peak responses. What we determined caused this difference was the fact that the processed data uses a tracking filter to calculate the peak G value for a given frequency while the time history data is unfiltered. The reason for the difference is that at a particular time during the sweep the input to the spacecraft is supposed to be at a single frequency, however, because of the limitations in the test setup the actual vibration input at a given time consists of several frequencies all of which excite the spacecraft and contribute to the total force at the interface. The use of a tracking filter to calculate G's vs. frequency removes this effect. The lesson learned here is that when calculating peak interface forces from sine sweep acceleration data it is necessary to use the unfiltered time history data to get the total contribution across all frequencies. When looking at acceleration response, using the processed data is ok because you are concerned with understanding the response as a function of frequency.

9. Acoustic Testing - After completion of the acoustic test we were reviewing the processed response data (PSD's) and saw that the overall response at the lower solar array corner was 18 Grms. By accident we happened to see the printout from the data acquisition system which provides a listing of the peak G values for each channel based on the time history data. The peak G value for the lower solar array corner was 160 G's. Since it is expected that the peak response data should statistically be 3 times the Grms value (3 sigma), the fact that the peak response at this location was on the order of 9 sigma indicated the need for a closer look at the time history data for this location. A review of the time history data showed intermittent, one-sided, high G spikes which indicated gapping at the lower solar array snubber. The lesson learned here is that when evaluating the processed acoustic response data it is also useful to have the peak G data for each channel to identify locations that require a more detailed review.

10. Look for non-normal distributions. Don't just look at the PSD plots for acceleration responses, look at the min/max responses also. If things are gapping or banging, the PSD plots will not pick this up. The min/max responses will. We experienced this during the Observatory acoustic test where the solar array hard snubbers gapped.

 

I&T

1. Keep a log of s/c weight each time it is lifted for future reference/support of mass properties verification, lifting sling turnbuckle adjustment, etc...

2. Photographs used for documentation are much more useful when they are dated. Programming your camera to put the date on the pictures works best.

3. Beware of multiple extension cords connected to each other. They can be moved/pulled enough times to loosen their connection and S/C grounding strap can come in contact with exposed prongs. It’s best to tape the connection together securely.

4. Having a 754 engineer leading the NSI technician team proved to be very effective and influential.

5. Beware of facility floors. They are not flat so over constraining s/c with flimsy GSE such as dollies will make s/c react GSE loads.

6. Mechanical plays a major part in I&T and needs to work closely with the I&T Manager to avoid unreal scheduling.

7. Keep a list of everyone who ever supported you and remember to include them for project awards, certificates, and freebies. A hat or a mug will go a long way. Some very capable and responsive people in certain services always get forgotten because their involvement is usually short term.

8. Record and document major activities and anomalies chronologically as you proceed so that future questions such as alignment deviations can be traced back and possibly explained.

9. Spread and distribute information up and down the line so that every one in the team knows and is aware of the status. Knowledge can help, ignorance can hurt.

10. Beware of "indoor rain", cover up hardware when it is not being worked on for a long time (overnight).

11. Remember to include transition time between activities during schedule planning (i.e. move to and from facility, setup time) because it can throw your schedule off significantly.

 

ALIGNMENT

1. Beware of alignment data obtained with relay shots. These shots can be complicated and errors can easily be introduced.

2. Beware of data obtained after a long day of staring in a scope by alignment techs. They are only human and projects have the tendency to schedule them on weekends and second or third shifts.

3. Keep a good record of alignment configurations for possible explanation of future shifts or differences.

4. Always check alignment tools in all orientation if it can be inserted in different ways to verify its reliability and repeatability. Our thruster nozzle alignment tool was found to be non-concentric.

5. Choose a stable platform for the master reference cube (MRC). This is easier said than done so really think about it. Consider what can happen to that location throughout the hardware’s life cycle (handling, natural place to step on or leave tool on, busy location or interface, etc…) Consider having two MRC’s on the same stable platform for redundancy if alignment is very critical. Our MRC actually got damaged even with a cube cover installed!

6. The Shop alignment system requires much setup time and data can be wiped out if power failure occurs. To save time and frustration, make sure system is hooked up to UPS.

7. Use a geometric alignment technique (photogrammetry or AIMS) to check for overall movement and to help correlate cube data.

8. Consider using a rotary alignment table, if possible, to save a lot of time or use a technique such as the photogrammetry.

 

DEPLOYABLES

1. Trade-off the benefits and risks in performing launch site deployment tests.

2. Designing in the capability of measuring the available torque for deployment mechanisms is highly desirable.

3. Hot flash testing and inspection of solar arrays is very beneficial. When the substrate facesheet is aluminum, the effectiveness of this testing technique has been demonstrated.

4. The difficulty of inspecting for cracks in GaAs solar cells was clearly demonstrated. Orientation of the arrays with respect to the light source is very important.

5. Beware of bubbles in dampers! Vacuum X-ray dampers for bubbles.

6. When procuring honeycomb panels, allow the vendors to propose adhesives with which they have experience. For solar arrays, we called out specific adhesives on our drawings, forcing TRW to use materials new to them, rather than equivalent adhesives with which they had years of experience. This led to discomfort on their part. We had to let them do some "practicing" with the adhesive to feel more comfortable with it.

7. When building flight solar array panels, it was beneficial to install the flight interfaces onto the substrates prior to installation of the solar cells. Doing this for TRMM, we greatly reduced the risk of damaging solar cells by not having to do the drilling and pinning operations with the cells in place. This did, however, complicate somewhat the cell laydown, because the panels now had protuberances in several places, rather than the flat, bare substrates.

8. Perform thermal vacuum test on solar panel substrates to find delaminations prior to installation of solar cells. It is much less costly to find and repair prior to cell laydown.

9. Off the shelf stainless steel ball bearings are usually not passivated. They can rust once the original lubricant is cleaned off. Braycote 601 is not a good rust inhibitor. It is better in the long run to purchase "custom" passivated bearings, if possible. The initial cost may be higher, but it will save headaches in the long run.

10. The filling process for viscous dampers is critical. Any air entrapped in the damping fluid can come out of solution in vacuum and cause loss of damping for significant portions of travel. Filling should be done in a vacuum. Dampers should be x-rayed for bubbles in vacuum (Materials Branch has facilities to do this). X-ray in normal atmosphere (the current vendor acceptance test) is useless.

11. Strip heaters with aluminum backing do not conform well to convex surfaces. The aluminum tends to peel the heater off due to the springback effect.

12. When procuring multiple custom strip heaters via a tabulated drawing, take care to specify the length and width in the correct order. Otherwise you can end up with leads coming off of the long sides instead of the short sides. This happened with the SADA stator housing heaters, causing great difficulty in applying the heaters without pinching the wires on nearby surface features.

13. Question thoroughly any requirements which force you away from already designed/qualified hardware. Some believe that it may have been possible to use the XTE SADA design with slight concessions from Electrical and C&DH. This would have saved considerable design, analysis and test dollars and effort.

14. For deployable systems, mark the flexing portion of the harness at the "p" clamp locations. The marker will indicate if the harness slips through the "P" clamp during deployment.

15. Relax, the project will launch.

 

LAUNCH SITE OPERATIONS

1.Take the time to list and photograph everything you are bringing to the launch site. This can save you much time searching when you get there.

2. Beepers are a must. Cell phones are even better.

3. Bring every tool you can imagine needing.

4. Perform an inventory to assure you have all the necessary tools/parts for an upcoming operation early enough to get/remake any forgotten parts.

5. Insist on having one s/c to launch vehicle interface person during s/c to vehicle operations so that activities and planning can work out better with less confusion and miscommunication.

6. Dry run of launch site operations and accessibility will pay off.

 

OTHERS

1. Deal with possible personnel problems early as soon as bad signs show up to avoid being burned.

2. Give ownership to people and sincerely show them that you trust and count on them and you may be surprised at how far they will go for you.

3. Don’t give up or let go anybody until they have delivered their product because once they start the next job, there’s a good chance you won’t see the deliverables at all ... and if you are lucky, you will get it very late.

4. If you want high probability that the job will be done right, insist on getting an experienced person, otherwise, you will pay for the learning curve in more ways than one.

5. Question reasons for why things are not working as well as they can/should because it may be deteriorating or heading toward a bigger problem.

 


 
  Return to Mechanical Engineering homepage
 
 
Last Revised 5/23/2007
Responsible NASA Official: James Ryan
Curator: C. Lazrus
 
Privacy Policy and Important Notices