lifted from comments by run 75441
These are comments that I found interesting describing change to manufacturing processes over tha last few decades. It made for a foil next to Martin ford’s projections and our understanding of past events. It is not a post but I learned a few things.
No one buys a stamping press that will make only one part. Stamping companies do buy stamping tooling (that die going into the stamping press) for a particular part. It is very possible to dedicate a stamping press to one part given the capacity of the machine and the volume of the part; however, different tooling can still be inserted into the stamping press dependent upon the machine’s bed. There is flexibility to buying a stamping press and it can be used for various parts, even in Henry Ford’s days. There is also flexibility to the stamping tooling by buying progressive dies which will do multiple operations in one press.
There has also been an evolution from manual machines and automatics to NC and CNC machines over the years. Let’s talk about throughput first though. Everyone is familiar with Henry Ford’s assembly line were the vehicle in some form of assembly moves down the line. At each station, another part is added until the end when the car is complete. This type of assembly is still being used. What most people didn’t see was the departments of machines (annealing, stamping, grinding, deburring, welding, turret lathes, automatics, drill presses, etc.) dedicated by function and not by flow of parts to supporting that assembly line These departments were not dedicated to the flow of a part through the plant the same as one might see in the assembly of a car. Each of the departments could be located in the corners of the building creating distances of transportation, requiring multiples of forklifts and labor to move parts from one operation to another, to stock eventually, and then from stock to final assembly (not including the multiple amounts of part inspections).
Magnify this by economic order quantities determined by Demand (months or weeks), Machine Setup (manuals and automatics setups were frequently measured in shifts), etc. It was used to determine a hopefully optimum inventory manufacturing lot based upon constants such as purchasing cost, carrying cost, fixed lead times, demand; which all of them at one time never were constant. The supply chain from customer order, to 100% of the Lot order, and going out the door could be months long . . . i.e. a casting having a 12 week lead time, through, 8 weeks long, etc. There was also a constant battle between Manufacturing, who wished to optimize setup, and typically Materials hoping to optimize inventory (unless they were into build it to be safe mode).
You have a factory setup by machine function rather than throughput and the functions scattered through the plant. Is it is any wonder why, there was a huge requirement for Labor to move parts, to operate each of the machines by function, have excessive transit distances, and manufacturing times plus lead times??? A part processing spaghetti farm. Using Pareto analysis of the flow of parts through a factory from the routing of operations to make a part, it is, and was entirely possible to achieve a much more efficient layout of a factory by eliminating machine departments by function, layout the shop floor by major part flows from receiving to shipping, and in the end reduce internal transit time and labor required, improve throughput and delivery, and reduce EOQ for a part until manufacturing whined about “setups.” All of this could be achieved without buying one CNC machine. It is the process of manufacturing that can make a huge difference.
So what did the CNC and or NC machine do for manufacturing? It combined operations such as drill, threading, boring, and lathe. No longer did a company need a drill press, lathe, etc functionality except for alternative runs, because all of those operations were resident to one machine. With tooling for each function resident to the machine; setups were reduced, and again there was a reduction of setup, machine, and forklift operators. It also cut down on the size of the plant needed as all of the operations were resident to one machine instead of taking up valuable floor space. The only thing it didn’t impact was inventory as people like that “feelie-feelie” type of safety in it rather than rely upon capacity and speed of throughput.
I am going to disagree a little with your take on implementing new product or new parts. I would suggest it happened consecutively with the running of old product. While it is true that brown field analysis and change over of the shop floor might be disruptive, it was typically planned for and the necessary inventory laid aside to handle demand. In turn the plant people were used to move machinery around. In the end, the changes were implemented on weekends, holidays, Christmas shutdown, Summer shutdowns, etc.
Is it capital equipment or is it a process called kanban impacting throughput? Kanban doesn’t have an outlay for its implementation other than to change the way we think about making parts and product, one part or product at a time, in the smallest increment of inventory possible, and correctly each time. It doesn’t even require computerization (internally) to implement as it is a pull type system rather than a push or build to forecasted demand system. In each operation, the inventory, wait time, operation time, and transit time is minimized. Is the success of Toyota based upon CNC equipment or is it in the process of manufacture itself? You give way to much credit to computerization of machines. There is something else a brewing and I liked Spencer’s explanation.
There was a time when one spent a lot of time as an apprentice to operate a lathe or to be a tool and dye maker. You are right; this is a vanishing “skilled” capability. Instead we rely upon computer cad cam to image parts and then apply the correct operations needed to shape the part in a Mazak or Cincinnati Milacron CNC. It does not occur quite a quick as you state does; but, it certainly is another consolidation point of various operations.
I consulted at a company called Miami Industries (taken over) in the glorious town of Piqua. I lived in MadCity WI and spent much of my time consulting in Ohio. They were meeting much of their customer order lead times when demanded of them; but, they had difficulties doing so. The process was simple (as I recall): slit and cut flat stock (came in rolls), draw over mandrill and weld, burnish outside to eliminate welding burr, cut to size, bend to shape, plate or coat. The company wanted to know how to improve delivery.
At 30-someting and looking like doogie-howser, I got no respect and always had an older consultant with me (colonel sanders this time). I went on the plant tour and listen to their spiel. They ran monthly lot sizes (20 day, 5 days/week) and wanted to improve deliver to weekly from whatever it was . . .
I looked at the VP of Manufacturing and said; with a 20 day manufacturing lot size, you are averaging 10 day deliveries on any order. “Oh no, we can get then out in 5 days if needed.” Yes, but the average is 10 days. Given the size of the Manufacturing Lots planned from the customer orders accumulated, the company having an average throughput of 10 days made sense. If they wanted to improve delivery, than cut the lot sizes to 10 days from 20 days which would increase the turns of the orders on the shop floor. It would be a start towards a much quicker throughput, less inventory on the floor and in stock on either end, and improved delivery.
“We can’t do that because of the setup.” Change your setup by making it more efficient with tooling resident to the machine, handles instead of nuts and wrenches, and look at the costs of inventory as opposed to setup, etc. The excuses for not doing something were endless and this change did not involve one new piece of capital equipment. I have another word for them; but for now I will settle for . . . “whims.” Oliver Wight used to call them “cement heads.”
The whole issue revolved around lot sizing. I set out to prove how silly they were and that my average of 10 days was spot-on. They had a shop floor system that collected data on each operation. I was able to look at each Mandrel and ascertain the amount of time for a lot to clear it, the setup times involved, the start of the next operation and total time, and transit times between operations. 75% was sit around time waiting for the operation, 8% was transit time, 10% was setup time and the balance was actual operational time making parts. Oh, an average lead time varied from 9.3 to 10.1 days for an order. Not only was my quoted average correct; but, there was adequate cost savings to be had in inventory reduction to also justify the change. I was hated . . . the life of a throughput analyst! Don’t chase technology for technology sake.