REG NASA-LLIS-1777-2007 Lessons Learned - COPV Propellant Tank Failure on the Dawn Spacecraft.pdf

1、Lessons Learned Entry: 1777Lesson Info:a71 Lesson Number: 1777a71 Lesson Date: 2007-03-07a71 Submitting Organization: JPLa71 Submitted by: Leslie Calluma71 POC Name: John Brophya71 POC Email: John.R.Brophyjpl.nasa.gova71 POC Phone: 818-354-0446Subject: COPV Propellant Tank Failure on the Dawn Spacec

2、raft Abstract: Anomalies and failures during testing brought into question the safety and flightworthiness of the flight COPV propellant tanks for the Dawn project. The tank development problems demonstrate the importance of annealing the tank liners, and completing the full flight tank qualificatio

3、n process prior to the acceptance of the flight units.Description of Driving Event: The JPL Dawn spacecraft design uses ion propulsion to obtain the velocity to reach two of the solar systems largest asteroids, Ceres and Vesta, during a nine-year voyage. The electric propulsion system ionizes xenon

4、(Xe) gas, which is stored in a pressurized Composite Over-wrapped Pressure Vessel (COPV) tank. The Dawn COPV tank is composed of a titanium liner, made from two welded domes, that is then over-wrapped with a graphite fiber-based composite. Figure 1 shows the flight tank just prior to integration wit

5、h the spacecraft. Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Figure 1 combines two similar color photographs displaying a top view and bottom view of the flight xenon tank. Both images show the dark copper colored, roughly spherical, completed t

6、ank. The liner is not visible as the outside of the tank is the outer surface of the composite wrappings. Both photos are labeled with arrows pointing to heaters incorporated into the configured item, and the crown-shaped skirt. However, the bushings on the skirt are only visible in the bottom view

7、of the tank.Figure 1. Flight Xe tank prior to blanketing and installationTwo anomalies occurred during fabrication and testing of the initial, or development tank (S/N 002). First, tank distortion was caused by each of two unsuccessful annealing cycles, and each time the liner was pressurized to ret

8、urn it to its original shape. Second, the tank failed a leak check following six proof cycles, and the liner had to be coated with polymer sealant to permit development testing to continue. However, the tank design was viewed as successful because (1) liner welding without annealing was believed to

9、be adequate and (2) the eventual burst of the tank at over 3000 psig significantly exceeded the design burst pressure. Note, however, that this development liner had actually been annealed twice, making it non-representative of the flight liners, which were not annealed. The planned progression from

10、 fabrication and testing of the development tank, then a qualification tank, and finally two flight tanks (one flight tank and one flight spare) was changed to accommodate the redesign of the tank skirt. The tank skirt was redesigned because after fabrication and testing of the development tank it w

11、as discovered that the mounting bosses used to attach the tank to the core spacecraft structure had been located incorrectly by the spacecraft vendor. It was determined that the most appropriate and least expensive way to fix this problem (in terms of acceptable performance, funding, and schedule) w

12、as to redesign the xenon tank skirt to accommodate the incorrectly placed core spacecraft bushings. This change caused a five-month delay in the availability of the flight tank, Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-but only if the flight t

13、ank would be fabricated ahead of the qualification tank. This decision was viewed as an acceptable risk based on the apparently successful results of the development tank. The flight tank (S/N 007) was fabricated and passed all acceptance tests. The flight spare (S/N 009) was subsequently fabricated

14、 and failed during its first pressurization to the proof pressure (Reference (1). The qualification tank (S/N 008) was then fabricated and subjected to the full qualification test program. This qualification test included four times the maximum number of pressure cycles that the flight tank is expec

15、ted to see at each pressure level. The qualification tank was subjected to 12 cycles to 2188 psig, 16 cycles to 1750 psig, and 24 cycles to 1250 psig. After the pressure cycling the tank successfully passed a leak test at 1250 psig. The tank passed the qualification quasi-static proof load and vibra

16、tion tests. The tank assembly was installed in a flight-like mechanical structure for the dynamics testing. The last test necessary to complete the qualification testing was to demonstrate capability in excess of a minimum burst pressure of 2625 psig. The tank did not reach this pressure, but instea

17、d burst at 2453 psig, 6.5% below the required minimum burst pressure (Reference (2). Post failure photographs of S/N 009 and S/N 008 are provided as Figures 2 and 3, respectively. Figure 2 is a color photo of the completed flight spare tank. The wrappings appear silvery in color in this image, and s

18、ome damage to the over-wrap (due to internal liner leakage) is visible.Figure 2. Post-failure condition of the S/N 009 tank. Note: Label reads, “Local Over-wrap Damage Due To Internal Leakage“Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Figure 3 c

19、ombines two similar color photographs displaying the qualification tank. The left photo shows bands of the composite overwrap pulled away from the tank, and shows fibers that are broken due to bending. The right photo shows the remnants of the qualification tank skirt, including several fragments sc

20、attered across the floor,Figure 3. S/N 008 tank after burst at 2453 psig. Many low-angle helical fibers failed in the dome region, well away from the weld area. The skirt was broken into five large pieces.The root causes of the Dawn xenon tank failures have been attributed in Reference (3) to: 1. We

21、ld process did not include vacuum annealing. In welding the tanks, the liner manufacturer was unable to control the annealing process, and the tanks would collapse. To avoid the cost and schedule impact of finding another liner manufacturer, the Dawn project decided to omit the annealing step. This

22、decision was supported by the large margins identified by the Finite Element Analysis, and it was verified in the testing of the developmental tank. Not annealing the liner results in higher residual stresses in the weld and in the heat affected zone (HAZ). 2. The ductility of the weld HAZ was lower

23、 than expected. The welding caused hourglass distortion of the liner due to radial sinking at the weld (Figure 4). This also resulted in low ductility in the HAZ. Although no heat specifications were furnished to the liner manufacturer, all of the liners used in the Dawn tanks met the hour-glassing

24、specification. However, strain-to-failure measurements made during the failure investigation indicated that the HAZ cannot tolerate nearly as much strain as suggested by the material handbook. Figure 4 is a color photo of one of the metal tank liners. The spherical tank liner exhibits a weld seam al

25、ong the equator of the globe, and the liner is clearly indented along the equator, giving the sphere a slight hourglass shape. The material appears to exhibit a band a slightly shinier metal extending a few degrees above and below the ?equator.?Figure 4. Liner hour-glassing Provided by IHSNot for Re

26、saleNo reproduction or networking permitted without license from IHS-,-,-3. There was a never-bonded region between layers of the composite over the weld. The low-angle helical composite wrap was not well bonded to the hoop wrap in the region directly over the weld. This may have been a result of th

27、e hoop wraps inadequately filling in the liner hour-glassing. The low-angle helical wraps control the axial strain in the liner. If these wraps are not adequately bonded to the hoop wraps, the axial strain control of the liner in the weld region can be compromised, resulting in larger-than-expected

28、liner strain in this region. The larger-than-expected axial liner strain, combined with the less-than-expected liner strain capability, resulted in the failure of the S/N 009 tank shell and the failure of the S/N 008 qualification test. The flight tank (S/N 007) will be used as is with a reduced pro

29、pellant load and reduced temperature limits (Reference (4), having been assessed as low risk for the duration of launch plus 250 days (and very low risk after that). Neither the qualification xenon tank (S/N 008) nor the flight spare (S/N 009) will be used in flight. References:(1) “Dawn S/N009 Acce

30、ptance Proof Test Failure,“ Problem/Failure Report No. Z85933, Jet Propulsion Laboratory, March 10, 2005. (2) “Dawn Qual Tank Burst Test Failure,“ Problem/Failure Report No. Z86896, Jet Propulsion Laboratory, June 21, 2005. (3) Dawn Project Xenon Tank COPV Flight Tank SN007 Disposition Independent R

31、eview Team Final Report, March 17, 2006. (4) “Xenon Tank Derating,“ Engineering Change Request No. 104740, Jet Propulsion Laboratory, May 31, 2006. Lesson(s) Learned: COPV propellant tanks are difficult to fabricate, and the flaws are difficult to detect and to characterize. Tank defects may pose a

32、significant cost and schedule risk late in the project lifecycle that may be exacerbated if the project performs the flight qualification process out of sequence, and if deviations from specified manufacturing processes are not identified as significant project risks.Recommendation(s): To mitigate t

33、he significant risk to project cost, schedule, and performance from COPV propellant tank defects: 1. Complete the full flight qualification process before the flight tank(s) are accepted for flight. 2. Require annealing or stress relief of welded liners prior to composite wrapping. 3. Require direct

34、 verification of the mechanical integrity of the composite overwrap.Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Evidence of Recurrence Control Effectiveness: JPL will reference this lesson learned as additional rationale and guidance supporting P

35、aragraph 4.7.1.1 (Propulsion System Design-Design and Test Requirements) and Paragraph 4.7.4.1 (Propulsion System Design (Safety) - Fiber-Reinforced Composite Over-Wrapped Pressure Vessels) in the Jet Propulsion Laboratory standard Design, Verification/Validation and Operations Principles for Flight

36、 Systems (Design Principles), JPL Document D-17868, Rev. 3, December 11, 2006. Documents Related to Lesson: N/AMission Directorate(s): N/AAdditional Key Phrase(s): N/A Additional Info: a71 Project: DAWNApproval Info: a71 Approval Date: 2007-04-20a71 Approval Name: ghendersona71 Approval Organization: HQProvided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-