For the most part the end of April the Tesla factory in Fremont, California, stood. As battery factory near Clark, Nevada. On Tuesday, CEO Elon Musk has sent employees a memo, which said that the pause was necessary to lay the ground for higher levels of production in the coming weeks. Musk said that he wants all parts of the company produced 6000 Model 3 a week until the end of June, triple the rate achieved by Tesla in recent weeks. This announcement capped a nine-month period of “production hell”, as it is called Maskbecause Tesla was struggling with trying to increase production of the Model 3.
Tesla has high hopes for the production Model 3 rates. In 2016, Musk hired Manager of Audi Peter Hochholdinger for planning the production process, and Business Insider described his plans at the end of 2016: “Hochholdinger believes that robots can be a much more significant factor in the production of cars than now, largely because many of the components designed for Assembly by people, not machines.”
A year later, the Musk and he promoted advanced robotics Tesla. “We bring robots to the limit, in terms of the speed of its work, and ask our suppliers to get the robots to work faster, and they are in shock because nobody that didn’t ask for it,” said Musk in November. “As if the robots were moving too slowly.”
Now, Musk admits his mistake. “Excessive automation in Tesla was a mistake”, recently wrote Musk. “Precisely, my mistake. People undervalued”.
In short, the robots were not working.
Musk has discovered that large-scale car manufacturing is a complex thing, and that is not so easy to improve traditional car manufacturers. And while automation will certainly play an important role in the production of cars, this is not the magic bullet that Musk dreamed of a few years ago. Not to jump over the methods of traditional car manufacturers — Tesla can’t even fit in the effectiveness of their better-known rivals.
And most experts in the field of automobile industry believe that the Mask is still a lot to learn.
“Many of the errors about which we hear now, are these errors made the rest of the industry in the 1980-ies and 1990-ies,” said Sam Abuelsamid, industry analyst Navigant Research. He points to the experience of General Motors that have wasted billions of dollars on fruitless attempts to automate the production of cars in the 1980s.
At the same time, to underestimate the Mask is a bad idea. Musk always put optimistic deadlines to their companies and then could not cope with them. But Musk is persistent and learns quickly. He made a lot of mistakes, but at least he learns them and is quite capable of turning Tesla into a competitive car manufacturer.
ArsTechnica talked with two experts who have drawn a parallel with the attempts of automation of GM in the 1980-ies. At that time GM was ruled by the Chairman of the Board of Directors and CEO Roger Smith, a competitor of the company was Toyota and other foreign manufacturers. Smith dreamed of a car factory where the main part of the work was done by robots, and the company produced cars more efficiently than anyone else.
Here is how he described the results of the automation project Smith at GM plant in Hamtramck, Michigan, Paul Ingrassia and Joseph white:
“As soon as the Assembly line in Hamtramck tried to accelerate computer-controlled truck off course. Robots with spray began to irrigate each other instead of cars, resulting in the GM escorted the cars through the city on the old pyatidesyatishestiletny the Cadillac factory for repainting. When a massive, computer-controlled “robotic” welding machine crushes the car body or when the welding machine was frozen tightly, stopped the entire Assembly line in Hamtramck. The workers had nothing to do, just stood there waiting for the managers call the technicians of the contractor-robattle”.
In the 1980’s, GM has spent billions of dollars on advanced robotics, but the money was never returned.
“Instead of having to display the robots on the Assembly line a few at a time, implying the inevitable problems with debugging with redundant hardware, GM has put that whole production system Hamtramk with advanced automation will work perfectly.”
Thirty years passed, and the robots become more sophisticated. But the basic principles remain the same: automation works best when added gradually to the production process that is already running smoothly. And Musk seems to have made the same mistake that Smith: too many robots too fast — not enough time for testing and improvement process.
Robots, in theory, should produce more cars with fewer workers, but one ironic result of excessive automation is that actually robots can require more workers. Ingrassia and white report that the GM plant in Hamtramck had about 5000 workers in the mid-1980s, meanwhile, at the nearby Ford plant worked 3700 workers. And the last “out Hamtramck by a large margin”.
Today Tesla is experiencing the same problem. Tesla produces its vehicles at the plant in Fremont, California, which was formerly famous by the joint GM/Toyota called NUMMI. In 1985, NUMMI worked 2470 employees in the first year of its work, and produced a 64 764 car. By 1997, there were already 4844 employee who released 357 809 cars.
In 2016, Tesla was 6,000 to 10,000 workers, but the company produced only 83 922 of the machine. That is the Tesla factory in 2016 was two times less productive, if you count per employee than in the first year of running GM. Indicators in the heyday in the 1990s and say no.
“The number of people that Musk got dragged to it, means why he is not making money on the machines”, experts say. But Tesla makes progress. The number of work hours required for completion of vehicle, decreased by 33% since the beginning of 2016. However, Musk himself admits that criticism is appropriate. And then he concludes.