4 ms·

I am not an expert on manufacturing biomedical devices, but I do know quite a bit about manufacturing safety-grade nuclear components. So take this with a grain

by _n_b_ 7y ago

I am not an expert on manufacturing biomedical devices, but I do know quite a bit about manufacturing safety-grade nuclear components. So take this with a grain of salt. However, from what I do know, fundamentally the requirements are pretty similar. There is likely nothing magical from a purely mechanical standpoint about making a ventilator. The part that is difficult is: * Have an adequate QA program for you and your suppliers, tracing safety-critical components down to the raw materials * Adjusting your tooling, making molds, etc. * Doing low-production runs followed by testing to validate your process * Getting whatever certification is required; even if there is a flexible emergency process, I think we still want some demonstration of compliance * Not having the lessons-learned of companies who have been doing this for years, and have worked through enough problems to have general domain knowledge * etc. My experience would lead me to believe that short-term returns would better come from finding ways to optimize the supply chain and manufacturing pipelines of existing manufacturers, which potentially could be done much more quickly. If you look at the example of WW2, when auto factories converted to war production, it was not a quick or easy affair. The excellent book Arsenal of Democracy describes the conversion of the Willow Run factory to producing B-24s, it took at least a year and a half to get the plant producing planes at an acceptable quality and quantity.

dehrmann 7y ago
> it took at least a year and a half to get the plant producing planes at an acceptable quality and quantity. I wonder if this would be longer or shorter today with how much of car assembly is automated. How many of the tools do one specific thing a human used to? How many are robots that can be reprogrammed? That, and cars and ventilators are completely different. I'd think a ventilator is closer to a dishwasher than a car (think appliance).
- HeWhoLurksLate 7y ago
  I've had some experience with the robotics / automation engineering side of industry. It's not easy to take super high-level automation and turn around and make it produce something else entirely. Base / core machinery- such as a 3-axis CNC machine, or a lathe, or a saw, for instance, are fairly nimble- if you have the right cutting bits (which you probably do, considering that it's industry) you can switch from milling out aluminum mounting plates to milling out enclosures for electronics fairly easily. At this level, generally humans are responsible for sticking in the material and telling the machine what to do each time an action is performed. Where it gets more complicated is at the higher levels of automation, where humans might be supervising the machines, but aren't initiating processes themselves- for example, there was a prod cell that I worked with that had a robot putting raw material into a mill and then removing it- humans program those arms, but then after that, it just goes, and goes, and goes. Programming those arms is generally an exercise in making sure the math works, and then implementing the job in a bunch of for loops and hard-coded / static adjustment variables and the like. In these instances, to make the cell produce something else, not only do you have to program the machine that actually produces parts, but you also have to reprogram the bot that puts and pulls from it, and you may also need to change the grip on the bot to support the new geometry. At the really high levels, like how car manufacturing is currently, there are processes where metal blocks go in and a whole monoframe chassis comes out. Taking that and saying "I want a ventilator assembly" is pretty much insane- besides reworking the entire line to support the new process, you've got to wire everything up, test it, inspect it, make sure everything is up to OSHA standards, etcetera etcetera, etcetera. Going from producing one thing with a high amount of automation to another that doesn't have any similar parts is insanely difficult. Not saying it isn't worth it, but it might be better to go to the factories that are already producing the stuff we need- ventilators, masks, respiratiors, tests- and seeing if there are any bottlenecks that could be removed. The best way? Have enough stuff on hand to actually handle a crisis. Too bad prepping for a rainy day doesn't really align with maxing out dividends and payoffs.

by _n_b_ 7y ago

dehrmann 7y ago

> it took at least a year and a half to get the plant producing planes at an acceptable quality and quantity. I wonder if this would be longer or shorter today with how much of car assembly is automated. How many of the tools do one specific thing a human used to? How many are robots that can be reprogrammed? That, and cars and ventilators are completely different. I'd think a ventilator is closer to a dishwasher than a car (think appliance).

HeWhoLurksLate 7y ago
I've had some experience with the robotics / automation engineering side of industry. It's not easy to take super high-level automation and turn around and make it produce something else entirely. Base / core machinery- such as a 3-axis CNC machine, or a lathe, or a saw, for instance, are fairly nimble- if you have the right cutting bits (which you probably do, considering that it's industry) you can switch from milling out aluminum mounting plates to milling out enclosures for electronics fairly easily. At this level, generally humans are responsible for sticking in the material and telling the machine what to do each time an action is performed. Where it gets more complicated is at the higher levels of automation, where humans might be supervising the machines, but aren't initiating processes themselves- for example, there was a prod cell that I worked with that had a robot putting raw material into a mill and then removing it- humans program those arms, but then after that, it just goes, and goes, and goes. Programming those arms is generally an exercise in making sure the math works, and then implementing the job in a bunch of for loops and hard-coded / static adjustment variables and the like. In these instances, to make the cell produce something else, not only do you have to program the machine that actually produces parts, but you also have to reprogram the bot that puts and pulls from it, and you may also need to change the grip on the bot to support the new geometry. At the really high levels, like how car manufacturing is currently, there are processes where metal blocks go in and a whole monoframe chassis comes out. Taking that and saying "I want a ventilator assembly" is pretty much insane- besides reworking the entire line to support the new process, you've got to wire everything up, test it, inspect it, make sure everything is up to OSHA standards, etcetera etcetera, etcetera. Going from producing one thing with a high amount of automation to another that doesn't have any similar parts is insanely difficult. Not saying it isn't worth it, but it might be better to go to the factories that are already producing the stuff we need- ventilators, masks, respiratiors, tests- and seeing if there are any bottlenecks that could be removed. The best way? Have enough stuff on hand to actually handle a crisis. Too bad prepping for a rainy day doesn't really align with maxing out dividends and payoffs.