NASA OIG: NASA’s Parts Quality Control Process
“Although NASA has a number of initiatives in place to help ensure the selection of quality parts from reliable suppliers, Centers generally manage their parts quality and supplier assessment data unilaterally rather than collaborating through a comprehensive, integrated, Agency-wide parts and supplier information system. Specifically, the Agency does not maintain a centralized parts quality history database or facilitate the integration of individual Center systems, track all relevant supplier performance history, or enforce requirements that Centers participate in Agency parts quality management systems. Without these control mechanisms, it is more difficult for NASA to mitigate the risk of nonconforming parts entering its project hardware supply chain. …
… In addition, NASA policy requires project managers to consider risk factors when preparing Program/Project Quality Assurance Surveillance Plans for critical and complex acquisitions. These plans document contractor operations that need Government oversight and the activities, metrics, control mechanisms, and organizations that will conduct quality assurance functions for the project. We found the Agency’s current policy does not provide sufficient surveillance and audit planning guidance for project personnel to analyze and select contractor surveillance activities commensurate with the level of risk of nonconforming parts being incorporated into a product.”
Notice the IG didn’t investigate the risk is due to what The NASA mission center program managers refer to as “complicit liability”. The program managers direct the contractors, without going through procurement official. These managers intentionally give contractors direction to cause the legal liability and risk for failing to meet specifications onto the government. Contractors love this and try to get the civil servants in the DOD to give direction beyond what they are supposed to do. We were always warned not to give direction without the procurement officer. Being a level III certified procurement for the DOD, At first I thought the NASA civil servants were being duped by the aerospace contractors. The government oversight mechanism at NASA flight centers is so out of wack and in bed with contractors that the civil servant openly engage in what they refer to as “complicit liability”. They protect companies by purposely and willfully causing the government to incur the liability for the company mistakes, errors, incompetence, and intentional lies.
I don’t see how NASA could do what the IG recommends for reducing program risk -to get away from the current unilaterally managed system- without dis-engaging from this “complicit liability” practice.
I am completely on board with a centralized database for approved vendors and parts. Each center has their own, and they are all in different states of disrepair. At ARC we’ve had to reference JPL’s as Ames hasn’t had the manpower to keep their’s updated and the vendor audits current.
There should be one database where any center’s qualified QA personnel can list vendors for all centers to reference, and the cost should be covered by all the centers according to their contribution and need so each doesn’t feel like they are getting sponged off of by a center that doesn’t want to keep a properly staffed quality branch.
I agree the centers should work together on a database, but who will pay? It would logically be “cross program support”, but with total cost accounting there is little funding for anything that isn’t paid for directly by a “program”.
The report focuses almost exclusively on administrative procedures i.e. “perform mandated product assurance actions (e.g., product examination, process witnessing, and record review)”. There is little or no discussion of the actual principles of quality control. It isn’t the paperwork that keeps you alive at Mach 25. It’s digging your hands into the engineering, the design, the components, the manufacturing process, and the basic principles of statistical process control.
For example, the report lists the Glory and OCO launches but doesn’t even mention the launch vehicle. The fact that the Taurus XL had two consecutive identical failures shows pretty clearly that the problem was an unanticipated deterministic failure mode in the design rather than a quality control problem in manufacturing. This type of failure is discussed in Chang’s seminal paper “Space Launch Vehicle Reliability”. http://aerospace.wpengine.n… (page 22)
Conversely, the IG report does not list the CRS-7 failure, which was apparently due to to the fracture of a procured part in the F9 upper stage at well below its design strength, due to a manufacturing defect in many of the struts which could have been detected by traditional statistical process control (testing a sample of parts to failure) but not necessarily by NDE or any of the administrative procedures discussed in the report. http://spacenews.com/33811o…
One challenge, though, is that the EEE parts databases (which are the subject of the report, they’re not talking about 6-32 machine screws) and “approved parts lists” tend to be full of old parts that are either several generations behind current state of the art, are no longer manufactured, or otherwise unsuitable for new designs.
Typically these databases are not “forward looking” – they have parts in them that have already been selected and reviewed and used in flight, so they are not a good resource for a design engineer looking for modern parts – Sure, I’ll probably use 2N2222A transistors, which are in the database, but I’d also like to see parts that are brand new in the mfr databook. A design process that has, as a going in philosophy, “use only parts on the approved parts list” will wind up developing copies of ancient systems and not advancing the state of the art.
What is needed is a change in design and development philosophy, much like the DoD change away from MIL-SPEC parts to COTS referenced in the OIG report. There’s no significant economic incentive for manufacturers to make special parts since the 800 lb gorilla of DoD isn’t buying them any more. This is especially true in an era of increased on-chip integration with more functionality crammed into one part – Modern integrated circuit designs don’t use discrete functional blocks – rather the chip includes dozens of functions all combined in a way to be of maximum utility for the commercial market. Consider the “radio” in a cellphone – It’s a single chip these days, not a dozen, and it supports multiple frequency bands, multiple modulations, all to accommodate regulatory and industry requirements around the world.
Equipment designs have to tolerate the behavior of COTS parts, and parts engineers need to focus more on how do we characterize that behavior, particularly with respect to soft failures. (i.e. the part doesn’t meet spec, but still works fine. IF your circuit requires that the part works at 1 MHz, and you have a 100 MHz part that has “failed” and only works at 90 MHz, your circuit still works.)
This is different that the more traditional process documentation and inspection, which is more along the lines of “did you build these exactly the same way you built the last batches, so we can use the test data from 10 years ago”