Wednesday, January 4, 2012

Woman vs. Machine... Whoa, Man vs. Machine... Woaoaoaoah Man!

Ifrequently read that labor represents a small portion of photoelectric panel production costs. In Japan/US/Germany (JUGs) the statement holds water but it mischaracterizes the situation in low wage countries. Counter-intuitively, the labor costs (as a percentage of production costs) in low wage countries can actually be higher than what they are in JUGs. How could this be?


Large Scale PV Module Manufacturing in India and China was written by Lily Zhao, Matthais Ruh and Ronald F.M. Lange back in 2010 for the trade magazine Solar Power. In this article they pit man against machine:

"One operator in a 9 hour shift with 1 hour break time processes 8*600 is 4800 cells per shift. Hence, a 500MWp production line, assuming an output of 4Wp per cell and 10 days holiday per year, needs 500.106 *1/4 * 1/(4800*255) = 102 operators."

In contrast, the numbers of machines, having an output of 2400 cells per hour, would be 500.106 *1/4 * 1/(2400 * 12 * 365) = 12. Assuming 1 to maximum 1.5 operators per machine leads to the need of 12-18 operators for the complete 500 MWp stringing process..."

PUNCHLINE: You need 102 manual stringing/tabbing technicians to equal 12 machines operated by 12-18 operators.

OBVIOUS QUESTION #1: How much does industrial labor in China cost.

A Boston Consulting Group document published back in December 2010 lists $2.10/hour.

OBVIOUS QUESTION #2: How much does an automated stringer/tabber cost?

Here's a current quote from a Chinese equipment manufacturer for a 600 cell/hour machine priced at 170,000 to 180,000 USD.

Here's an older 2009 quote from a German equipment manufacture for a 1200 cell/hour machine (page 27) priced at 590,000 Euro (456,000 USD)

For mathematical simplicity let's split the difference and assume a 1200 cell/hour machine costs 400,000 USD. So in the left corner we have a 500 MW factory with 24 of these machines and 15 technicians to operate the machines. I'm going to arbitrarily assume the machines last 4 years. Your rough annual costs are:

24*400,000/4+ 15*2000*2.10 = $2,463,000

In the right corner we have:

102 workers * 2000 hours/year @ $2.10 per hour = $425,000

This crude math indicates that manually stringing/tabbing is about 2 million dollars per year more cost effective than automation. Does a half-cent per watt really matter? Let's just say it's a start. What other considerations are there? Well, you have to pay for the automated stinger/tabber up front whereas you pay for labor as you go. This limits the flexibility of the automated factory considerably. If push comes to shove you can't furlough your automatic stringer/tabber and there's limited resale value. That said, the bigger consideration at work here is that I'm comparing China to China. What happens if we compare China to JUGs where industrial labor rates are more like $25/hour.

The right corner equation looks like 24*400,000/4+ 15*2000*25 = $3,150,000

So now we're up to a difference of 2,600,000 per year. OK... so what, it's still only a fraction of a cent/watt. Well, now consider the cost to build the factory? What if there's another half cent/watt in labor savings there? What if it's two cents? Consider the other labor intensive stages of production like module conversion. What if there's another penny or two there? What if there are savings associated with manually testing cells rather than automated testing?

This is meant to be an analysis but not an in depth analysis. I'm playing Devil's Advocate here. You have to ask yourself why China is using all that labor in their factories. When I looked into it the surprising answer I came up with is that labor can be cheaper than machinery. This means that when western fabs say labor is insignificant they aren't making an apples to apples comparison. They are talking about themselves. What may well be true for their manufacturing process is not necessarily true for the manufacturing process in China. I've only looked at stringing/tabbing. There's also ingots, wafer sorting, wafer testing, module conversion and so on. My gut feeling is that Chinese labor beats machinery in some of these areas as well - at least it does for now. So anyways, the next time someone says labor represents a small part of photoelectric production costs ask for proof.

Note that the stringing and tabbing process shown by Suntech is automated. In fact there's a hell of a lot of automation shown in the video. It's impressive really. At the same time you can still see plenty of labor manually picking and placing wafers. When I see this I wonder if they are saving money over fabs that use automation to do the same job. Maybe it's a penny here or there but pennies matter.

Here's our kitty cat Leelu.

No comments:

Post a Comment