I was a project engineer at General Motors and worked quite a bit with Chinese manufacturers. My impression was they made every attempt to make parts to the drawing, but back in the day lacked the experience of problem solving issues.
An example was a problem at the GM Shanghai assembly plant around 20 years ago. They were in the process of shutting the plant down because the nuts with a captured free-spin washer that secured the drop links to the stabilizer bars were breaking. They didn't break on the initial torque rundown, but popped after 12 hours or less. Their yard was full of cars needing nuts replaced and they had no idea what to do or how to figure out what the problem was.
I got called in to their problem solving meetings. The problem sounded to me like something called hydrogen embrittlement in the captured washer. Fasteners are typically plated/coated and cleaned ahead of plating/coating. Typically an acid wash is used. However, if a steel washer is acid-washed (all acids contain hydrogen) before plating/coating, it needs to be "cooked" at ~400 F after plating to boil the hydrogen out of the steel. Otherwise, the hydrogen can migrate into the microstructure of the steel which can lead to fractures when the part is subsequently put under load. The failure surfaces within 24 hours; no crack after 24 hours under load, no problem.
I mentioned that they should look at their stock dates and see if there was another date that could be used. Their failure rate was over 25%. The failure happened to be what we called a binomial distribution; it's either okay or not okay. I don't have the chart handy, but there's a stats chart that gives the number of samples that need to be tested based on failure rate to prove within a confidence level if you have fixed the problem or not. I recall the number of samples needing to be around 12 for 95% confidence with the 25% failure rate.
Take 12 nuts from each of your manufacture dates and run the drop links down on stabilizer bars. Come back 24 hours later and see if any have cracked. If they have, quarantine that stock date. It's what we called "prequalifying material". I had to elaborate on that term with them. "Guys, if you have lug nuts that don't want to run down on the wheel studs, take the lug nuts off line and run them down on a loose knuckle (assuming you've proven the studs are okay). Prequalify the lug nuts before taking them to the line". To us in the USA who'd been doing this for years, it was "Problem Solving 101". To my Chinese colleagues who lacked the experience, it was rocket science. They were super-eager to learn, a number of their lead guys came to the USA to train with our statistical engineering group and get some reps under their belt. We'd occasionally get some of their product in our audit area to review, they did a very fine job putting their cars together.
I also had some experience with low-level Chinese vendors who cared about profit, not so much about quality. I had a problem with peel and stick emblems literally falling off the car. We did a wet-out test and the parts failed. Then came the arguments that I don't miss as I recently retired. They'd do the same wet-out test and have the parts fail, but still wanted us to use them. . . "Guys, would you expect your company to print badges for you with a few letters missing from your name and call that okay? Of course not. General Motors will not accept cars in the field with the name plate "B ick". We are selling a Buick, not a B ick".
Summarizing, the Chinese can do as good of work as anyone on the planet. They (and any other country) can put out crap too. Depends on what's important to you. We used the line, "cost, quality and timing; pick two". If you want it for really cheap/quickly, don't expect high quality.
I've learned the buy once, cry once lesson over the years, but still forget it on occasion. A lot depends on my project and the accuracy needed.
Bruce
