You can imagine Ineos selecting a parts supplier and ordering a pilot batch of parts for testing. It would not be a stretch to imagine parts supplier hand-assembling and hand-picking the parts before sending them to the customer during the testing/supplier selection stage, and ... relaxing the tolerance a little once the main hurdle is behind.
This.
I love the whole "this is all Ineos' fault" approach that some people take. That narrow view diminishes the role of the component supplier in ensuring the design of the vehicle remains sound during full rate production. I'm sure IA are just as pissed as we are about these faults. That said, after all these years GM are still finding
new ways to build a bad engine so what's IA's excuse, right?
In situations like this I often say "It's not what happens, it's what happens next". Things will go wrong in any system. How the provider addresses it is what really matters. IA need to break the chain on this issue. It's affecting their customers and brand reputation and no doubt hurting future sales. According to feedback provided through our Forum Ambassador
@Tom D, I think they understand the need to act. We're just waiting to see the actions.
_________________
Long waffle about engineering test and evaluation programs and design compromises (Goodnight, sleep well
).
On commencement IA had zero history with auto industry suppliers so de-risked the project by going with proven brand names, in this case Dana-Spicer for the driveshafts. There's no way the test team would have allowed a substandard part through to production if a fault was detected during testing.
Anyone that has been involved in Acceptance Test and Evaluation programs (ATE) knows there are detailed test serials designed to examine every element of a system before production sign-off. Once those test serials are completed they're not revisited unless a fault arises during ATE or there is a design change after ATE that triggers retesting.
ATE also assesses reliability and durability. This is relatively easy for established mature products because there is empirical data available (but if that data is supplied by the seller, it may not be accurate!). For new or modified designs or proven designs placed into a new application it's less straightforward. That places a high importance on the ATE program to prove the product is not only fit for purpose but will be reliable and durable during the target service life. The million miles of testing was the field component of IA's ATE program.
Cost engineering and supplier quality changes aside, did something else change on the vehicle between prototyping, ATE and full rate production?
We know from the Jeep world that this CV joint doesn't like running at high shaft angles. IA would have noted that (probably, possibly), but with input from Dana-Spicer likely determined they were within a reliable operating angle range. Any number of late changes could have secondary effects that impact the driveshaft (CV), but did not arise during ATE; for example castor angle, axle geometry, ride height, transfer case position, etc. Or, problems did arise during ATE but IA thought they had resolved them.
Case in point: The castor angle final setting is done through an eccentric cam built into each lower control arm rear attachment bolt. If IA had to limit the maximum castor angle to mitigate a problem elsewhere the easiest way to do that is to change the bolt cam radius. Reduce the castor angle and there can be more tendency towards bump steer and steering shimmy (hello shopping carts). Add a stiffer steering damper and it might become a manageable compromise but there is little margin left for changes, like increased spring heights, or deviation in individual component specifications.
But you gone and ruined it. 
