While clearly crop dependent, even in a glass building:
"The environmental writer George Monbiot calculated that the cost of providing enough supplementary light to grow the grain for a single loaf would be about $15" [1]
A quick search reveals that typical crop photosynthetic efficiency is around 1-2% [2], and I'm guessing that's likely calculated based on the light that falls directly on the leaf.
The sun delivers >1 kW per square meter in the middle of the day. That's a whole lot of power to replace, even accounting for savings through focusing on the most important wavelengths.
[1] https://en.wikipedia.org/wiki/Vertical_farming#Energy_use
It's not clear whether this is from Monbiot or the person who provided the numbers, but either way I invite you to consider the changes in quality, availability and pricing for standard LED bulbs as replacements for incandescent bulbs in the six years since that article was published.
If you're going to generatethe light yourself, you can save a lot of energey by only emitting the correct spectrums of light.
https://www.youtube.com/watch?v=2nFQOkzEjxQ http://www.skygreens.com/
This is the country that fined a Tesla owner for indirect emissions. http://www.roadandtrack.com/new-cars/news/a28417/singapore-g...
You're right that irrigation is a significant energy cost: for example [2], irrigating 135 acres in Nebraska will use 50,000 kWh over the course of one season.
But an acre has 4046 square meters. Based on the global average, that same 135 acres will collect 3,277,260 kWh of solar energy over the course of a single day. That's the energy an indoor farm has to consume from power stations, just for lighting.
Optimizing spectrum can cut the lighting energy in half, and other measures like packing plants more tightly can help, but even if you reduce it by 90% it still dwarfs the energy expense of regular farms.
It may be a reasonable expense for particular crops that don't want full sun anyway, but energy usage isn't likely to ever be an advantage for indoor farms. An acceptable disadvantage in certain cases, maybe.
[1] https://en.wikipedia.org/wiki/Solar_irradiance#Earth
[2] http://www.loup.com/docs/energysvc/energywise/EW-Tip-Irrigat...
To reproduce natural sunlight, but is that level actually needed for adequate plant growth?
Seems to me greenhouses are a better way to go: free lighting, while still using less water and chemicals.
Even a low-tech method like mirrors ought to do it. As long as all the wavelengths that the plants need are reflected by the mirror. Why not?
Now, there may be bacteria in the regolith which we don't know about, and we'll find out why they are neededin certain proportions... but even so, we can be collecting dirt from all over the place for the next crop...
It seems that sunlight hitting plants that are "closer together" and stacked vertically will be far more efficient. Besides, some plants don't even need the full sunlight to grow!
3 terawatt-hours? No, something is wrong there. Peak solar energy at the earth's surface is about 1 KW per square meter.[1] Over 24 hours, you average maybe a third of peak. 135 acres is 54.5 hectares, or 545,000 square meters; about half a square kilometer. So we get about 4360 kWh over a day. That's raw sunlight; solar cells are maybe 15-20% efficient in conversion. (Corn plants are about 1.5% efficient in conversion)
[1] https://livingonsolarpower.wordpress.com/2013/03/23/basic-so...