# A California Solar Dilemma

After grappling with the thought experiment of replacing all of our electricity consumption with solar panels, the problem came into focus. This problem seems simple, but it isn’t trivial. As I mentioned, I have seen people approach this problem in several different ways, and after tackling it myself I believe that all of those approaches are wrong. So, I decided to produce a graph to help illustrate exactly how I see the problem:

Typical Solar Cell Power Curve vs. Actual California Demand Curve on July 12, 2003

The way I came up with this graph was by modeling the solar cell power curve based on Google’s Solar Panel Project, which they update daily for solar electricity produced. You can presume at this point some hypothetical number of panels to produce 36 GW at peak power. The reason for 36 GW is that I found a presentation that showed actual load behavior in California on a summer day in 2003, and peak power demand was 36 GW.

It became clear to me why some people are approaching this from different directions, and why neither answer is actually correct. One approach looks at peak demand, and installs enough solar panels to meet that. But as you can see, peak demand doesn’t correspond to peak output. The second approach looks at the demand for the entire day, and then attempts to produce that in 4 or 5 hours. That isn’t correct either. You need to produce the required daily output in the total area under the solar power curve. But, you need to be able to store it. And due to storage losses, you actually need to produce quite a bit more than you expect to be consumed in any particular day.

That, I believe, is the correct way to solve the problem. In all of the approaches I have seen as I have studied the problem, I haven’t seen this specific approach. Thoughts? Just eye-balling it, it looks to me – presuming you have a workable storage solution – that you would require about double the power of the peak demand number in order to produce the required energy each day. In other words, if that solar power curve topped out at 70 GW or so, that would be enough energy produced in a day to meet that demand curve.

I don’t have time to work on this any more right now, but I will come back to it. My chapter is due on August 1, and I am still tidying it up. But I think the next approach is to either integrate the area under the solar output curve, or approximate it as a square wave – and then develop the relationship to daily demand.

## 87 thoughts on “A California Solar Dilemma”

1. Anonymous says:

Errr, shouldn’t the power curve graph have GW on the Y-axis rather than GWh.

Cheers

Andy

2. Anonymous says:

Some solar is becoming more practical now. Search on (Solarwall AND photovoltaic) to see one way of getting very high efficiency if you can use thermal also. Evacuated tube solar hot water with storage tanks, can work, especially if you use annualized geo solar to cheaply store the extra summer heat. PVs are becoming more cost effective also: note that if you already have a hybrid or electrical vehicle, the amortized cost of PV recharge runs about \$1.75/gallon equivalent. However, the area of a vehicle isn’t enough. Solar in one form or another will be the future. I’ve been able to cut my space conditioning costs in an old house with some judicious modifications!

3. would it be possible to generate all of US electricity using PV solar? the answer is yes.

could we afford it? the cost would be comparable to the war effort, but yes.

But I think thermal solar costs less at the moment, PG&E just signed a contract for a 553 megawatt thermal solar electricity plant that will be online in 2011.

Currently I think wasting less energy via higher efficiency is the low cost option, so the US should pursue energy efficiency more agressively.

4. Anonymous says:

If you’re interested in seeing near-real-time data on electrical power consumption in California, the California Independent System Operator posts a graph and table at
http://www.caiso.com/outlook/outlook.html

5. Errr, shouldn’t the power curve graph have GW on the Y-axis rather than GWh.

Yes, it should. The original had it in GWh and I just copied it over. But that isn’t right. On both output and demand, the GWh should be the area under the curve. I am going to fix the graph.

6. words like output sometimes mean power and sometimes mean energy

I suggest using power and energy which are unambiguous, but you could also say “output per day”, which implies integrating over time, i.e. energy

7. Good idea. What I want to clearly convey is that the area under the solar power curve, which is in GWh, should be greater than the area under the load curve. And I believe that is correct.

8. Can’t we just cool our homes to 20C at noon and let them slowly warm up all afternoon. Once we’ve done the thought experiment that shows solar electricty is feasible, the rest is engineering details.

CFL pays for themselves in less than a year. This doesn’t preclude conservation.

9. Anonymous says:

That seems like the right approach. It seems essentially the same as sizing the system to produce the required average kwh, after accounting for storage losses.

10. Anonymous says:

Robert

You work for the petroleum industry; why are you spending all this time on solar?

On theoildrum you said that biofuels can only replace a fraction of our petroleum needs; what fraction is that? 1/3, 1/2, 7/8?

11. You work for the petroleum industry; why are you spending all this time on solar?

I have many interests. But you do realize, I presume, that BP and Shell are both heavily involved in solar?

On theoildrum you said that biofuels can only replace a fraction of our petroleum needs; what fraction is that? 1/3, 1/2, 7/8?

Realistically speaking, based on currently existing technology, probably no more than 10% on a net basis. If BTL works out, the equation will look quite a bit better because any biomass source can be converted.

In the U.S. right now, as much as people think ethanol is contributing to our energy needs, it is still less than 1% of our petroleum usage (a bit more when looking at just our gasoline usage, but still less than 1% of gasoline on a net basis).

12. Anonymous says:

Robert said

“My argument is that we won’t, as many people believe, displace large amounts of petroleum with biofuels.”

“I have long believed that our future must be electric for at least 4 reasons.” “If we are to embark on a Manhattan Project to get off of our petroleum dependence, we should direct our efforts toward an eventual electric transportation infrastructure.”

If the future is electric transportation then oil use goes down.

If oil use goes down, then the ability of biofuels do displace more than a “fraction” goes up.

13. Benjamin Cole says:

I second Mike Weiss on the 553 megawatt plant being built in Mojave, Ca.
Anyway, solar does not have to carry the load alone. There is wind, there is hydo, there is nuclear, there is conservation.
I am not sure about this thought experiment, which reminds me of some of my sophomore year stream of conscienceness writing.
The PHEV crowd notes that the plug-ins are largely plugged in at night, when the grid is well below max. In California anyway, we max out on hottest summer days – exactly when solar would be able to kick in and help with peak load. So, instead of building any more coal-fired plants, we should be building solar plants, or lots of rooftop panels. Happily, the solar would be the daily cavalry.
The good news in all of this is that we can move to a post-fossil oil world while improving living standards and the environment. There are some costs, but the Iraq war has cost us \$500 billion and we are not through yet. Imported oil costing a pretty penny.
The only problem with widespread adoption of PHEVs, or high MPG cars, is that it would crank oil prices down to \$25 a barrel again, and then people would go back to the V8s.

14. The storage capacity is 16 million cubic meters of water, not sure how to convert that to kWh.

You have to know how much hydraulic head you have. That’s the limitation of pumped storage; you need to have somewhere uphill to pump it to.

15. You have to know how much hydraulic head you have

the Tianhuangping article says the maximum generating head is 607.5m

as an approximation, assuming the reservoir has a uniform shape over its entire depth, the average bit of water is about 300m high, so the potential energy (mgh) is

16 million cubic meters * 1000 kg / cubic meter * 9.8 m/s^2 * 300 m

= 47 x 10^12 Joules

= 13.1 GWh (gigawatt hours)

16. continuing our back of the envelope calculation:

13.1 GWh of pumped hydro capacity * 70% round trip efficiency
= 9.2 GWh net capacity

at a cost of \$1.08 billion
= 8.5 GWh / \$1 billion

Looking at the CA ISO page, California averages about 32 GW of demand over 24 hours, or 768 GWh per day.

So one day of storage capacity for CA would cost roughly \$90 billion.

17. I checked ERCOT’s (Electric Reliability Council of Texas) 24 hour demand forcasts. Texas has an even more pronounced peak than CALISO.

For tomorrow, ERCOT is forecasting power consumption to dip to about 29 gigawatts at 5am and peak at just over 49 gigawatts at 5pm.

Texas has more AC load than California, and more heavy industry, so you would expect its peak demand to be higher.

18. If you are looking at PV, you really need to take into account the electrical energy input to refine silicon and semiconductor factory construction energy input. In theory you could build a totally automated Solar PV factory to crank out square miles of PV. The reality is that the electricity required in the arc furnace to refine semiconductor grade silicon, the factory construction cost and the storage issue will keep the price above just burning coal locally for base load electricity.

With China building old school coal power plants as fast as they can, the real energy cost of Solar PV is going to get hidden in the retail price by getting them manufactured in China with coal electricity, subsidized by grants and then plunked into a power demand curve that doesn’t match the output.

SHPEGS attempts to look at the demand and then design a system with common materials that matches the electrical/structure heating demand as close as possible.

19. I have an old copy of EDN (electronic design news) magazine. It’s an old anniversary issue (1981) looking at electronics for the next 25 years (just clicked over).

Now that I’ve got it out I’ll read it again, but one of the comments that quickly proved itself in that industry was that “engineers are overly optimistic about their ability to bring new technologies on-line, and overly pessimistic about their ability to extend existing technologies.”

(In the same magazine many predicted that optical disks would drive magnetic disks out of the market in short order. The extension of both kind of illustrates the need to second-guess our optimism and pessimism.)

We often think a tech is out of steam, and look for something brand new to replace it.

I was thinking about that in terms of solar this morning, wondering what it would mean if the future happen exactly that way.

There are no guaranties, but I think it’s interesting to note that in this scenario solar thermal would win. We know how to do it. We are making steady improvements.

And it lends itself to energy storage (as heat).

20. Anonymous says:

From the First Solar 2006 Annual Report:

“In 2006, our
average manufacturing costs were \$1.40 per watt”

21. Benjamin Cole says:

Per comments here that biofuels can’t make up more than 10 percent of liquid fuel use.
1. In two or three years, ethanol will already make up 6 percent of US gasoline by volume, and we have not even broken a sweat. The inefficient first generation corn-ethanol plants have got us this far already.
2. New cellulosic plants using wood chips and orange peels are being built now. Second-gen corn ethanol plants, the E3 plant, are now up and running, getting 3-to-1 energy retruns.
3. US corn acreage the same today as the late 1940s. That is how much yields have risen. And our population has well more than doubled in that time frame. This is hardly an environmental disaster, or if it is, no worse than 50 years ago.
4. And ethanol is hardly the best bio-fuel. There is a Massachusetts outfit planting jatropha in Panama, for diesel. Have you read up on dimethylfuran?
Amazing.
Either oil prices come down, or we see biofuels boom. You can’t have it both ways.
The extra good news is that with PHEVs, we may be able to get a much larger fraction of total liquid use from biofuels, I’d say half is doable. Hell, not only doable, but will raise living standards while reducing pollution.
As Odograph points out, the Nazis and South Africans managed to go totally synfuel when they had to.

22. As Odograph points out, the Nazis and South Africans managed to go totally synfuel when they had to.

LOL, I wasn’t trying to say that!

In fact those cases might show what a slog it is to bring some technologies on-line at a practical ROI.

But to tie them together, how does the ROI on solar thermal compare to that of coal-to-liquids?

I’d assume solar thermal is ahead (and “practical”) while CTL is not … but is that green bias?

23. Robert, you’re getting there. You should find that the area under your 12 hour wide parabolic curve of solar cell output in California is the same area as would be under a square wave about 5 hours wide and as high as the peak. Afterall, that’s how they calculate solar hour/days in the first place. (the number of KWH per square meter per day that strikes a flat plate on the ground at a given location, including all weather conditions, typically averaged over an entire year).

So yes, you need about double the power of your peak demand, which is incidentally roughly the factor between your 79 mi x 79 mi and your 50 mi x 50 mi areas.

If you want to account for storage losses, you could add in another 30% if you want to go with the most efficient storage methods such as pumping water or compressing air in underground caverns that used to hold natural gas or oil. Or double your numbers again if you want to go the hydrogen/fuel cell route.

For whoever it was who said these numbers didn’t account for future growth and moving from liquid fuels to electricity, double the numbers again and again and again, and it’s still less than the area of the Great Basin Desert of Nevada.

24. If you are looking at PV, you really need to take into account the electrical energy input to refine silicon and semiconductor factory construction energy input.

According to http://www.nrel.gov/docs/fy04osti/35489.pdf a PV module generates in 3 years as much energy as it took to make the module including refining the silicon and everything. After that, PV requires no more fuel to burn. That’s not as fast a payback as a wind turbine, but it might be better than moving hundreds of thousands of tons of coal a day on our railroads.

http://www.nrel.gov/international/china/pdfs/2005_renewable_energy_world.pdf pages 9 and 11. China found it such a problem shipping coal thousands of miles from the coal mines to Guangdong that wind power was cheaper than coal power in that coal-poor part of the country.
Think global, act local.

25. Anonymous says:

Since we are going to be awash in electricity and thus able to generate cheap hydrogen, what is the upper limit on the price of synthetic fuels using H2 and C02?

26. In two or three years, ethanol will already make up 6 percent of US gasoline by volume, and we have not even broken a sweat.

Ben, your optimism gets far ahead of itself sometimes. As I have pointed out – on several occasions – you are missing some key points. Even as it stands today, ethanol has cranked up solely because of the subsidies, which have caused some other problems. Now that corn prices have doubled as a result of demand, ethanol producers are finding their margins squeezed. So, while it is mandated that we will go to 6% by volume, it is going to be very difficult to do as producers are squeezed, and corn prices continue to rise.

But that’s not the important point. The important point – the point that you insist on missing – is that the net addition to the energy supply is very small. Because of the poor EROEI – and I am not talking about the hypothetical EROEIs that you keep stating as fact – most of the energy content in the ethanol is used up making the ethanol. Take last year. 5.4 billion gallons of ethanol were produced. If we take the generous USDA numbers, which count animal feed as BTUs just like you could burn in your car, then the EROEI is 1.3. That means that to produce 5.4 billion gallons of ethanol took 5.4/1.3, or 4.15 gallons worth. The net then is 1.2 billion gallons, less than 1% of our gasoline supply on a volume basis. Since ethanol has 67% of the energy content of gasoline, the true contribution to energy supply is actually 0.6%. Until this soaks in, you are never going to get it. (And if we presume that animal feed can’t be burned in your car, now you are down in the 0.1% range).

New cellulosic plants using wood chips and orange peels are being built now.

Just reading the EIAs Energy Outlook 2007. The mentioned a number of times how uneconomical these things were going to be. Capital costs are too high, and energy costs are too high. Just because you can do something, Ben, doesn’t mean that it will be viable long term. They were trying to build DNA computers 15 years ago. I am still waiting for mine.

Second-gen corn ethanol plants, the E3 plant, are now up and running, getting 3-to-1 energy retruns.

This is a perfect example of what’s wrong with your arguments. You don’t actually know this. They haven’t released any information on this. And I can tell you that I have spoken with them and seen their energy models (I have their energy model in this computer). I have followed them more closely than you have. They have released nothing about where they actually are. You are going on projections, which is not a wise thing to do. Again, where’s my incredibly powerful, ultra-cheap DNA computer? Right, there were some technical problems that the media reports tended to gloss over.

Either oil prices come down, or we see biofuels boom. You can’t have it both ways.

Running out of time, but this has to be addressed. This is where you do not understand the EROEI issue. For a biofuel with a poor EROEI, rising oil prices means rising biofuels prices. Always. That’s because the fuel is essentially recycled oil. So you can have it both ways. Rising oil prices can stem demand for biofuels when the EROEI is poor.

27. You should find that the area under your 12 hour wide parabolic curve of solar cell output in California is the same area as would be under a square wave about 5 hours wide and as high as the peak.

Actually, I am finding the the square wave has to be about 8 hours wide. For instance, yesterday they peaked at 877 KW, but total energy production was 7021 KWh. I have to multiply by 8 to get that. In fact, that’s been a pretty consistent theme. So, I think the 5 hour peak is too conservative.

28. If you’re interested in seeing near-real-time data on electrical power consumption in California

Anyone know where there are historical charts like this? I would like to compare summer/winter demand. Because while the winter will give us less solar energy, it may be offset by lower demand. Just trying to picture the worst case scenario; i.e., the scenario in which we need the most solar panels.

29. Google claims their system is rated at 1600KW. I’m surprised and disappointed they only peaked at 877 KW on a sunny summer day. The 5 hours I mentioned was supposed to relative to peak rated KW. (You may have noticed that PV panels are rated on the watts they produce under standard testing conditions which include 1KW per square meter of solar spectrum light. You might not have noticed from the National Solar Radiation Data Base http://rredc.nrel.gov/solar/old_data/nsrdb/1961-1990/ that the peak radiation reaching the ground in the sunniest hour of the year is almost exactly 1KW/m2.)

While solar radiation in San Francisco (about 25 miles north of the Googleplex) averages over the entire year at 5.3 kWh/m2/day for a panel facing south at a tilt of Latitude – 15 degrees according to http://rredc.nrel.gov/solar/old_data/nsrdb/redbook/sum2/23234.txt over the month of July it averages 7.3 kWh/m2/day, so I’m not too surprised you’re getting numbers well over 5, considering you’re starting your observations when the days are long.

30. Google claims their system is rated at 1600KW. I’m surprised and disappointed they only peaked at 877 KW on a sunny summer day.

If you look at the entire week, production is higher during the normal work days. Not sure why it should fall off on the weekend. I know that demand falls off, but why should production?

31. If you’re interested in seeing near-real-time data on electrical power consumption in California

Anyone know where there are historical charts like this?

The next to the last item on http://www.caiso.com/marketops/OASIS/index2.html looks promising

32. For instance, yesterday they peaked at 877 KW, but total energy production was 7021 KWh.

I just realized that I didn’t clarify who “they” are. They refers to Google Solar, linked to in the initial essay. Clee figured this out, but I certainly wasn’t clear.

33. The next to the last item on http://www.caiso.com/marketops/OASIS/index2.html looks promising

So, it looks like July demand could peak at 40-50% higher than April demand. Therefore, peak summer demand may in fact be the limiting factor on the system. Even though you have fewer hours of sunlight in April, demand is significantly lower (which I presume is even lower in January).

34. Found it at
http://oasis.caiso.com/
click on System Load and enter the date you’re interested in to get actual hourly demand

35. Benjamin Cole says:

Robert-
You assume that the first-generation ethanol plants, which may get 1.3-1 energy returns (I think a little better, as yields are rising), are the best anybody is ever going to do.
But, already, E3 is saying they are getting 3-1. The plant is up and running now, so we can ask them soon. If the E3 plant works to specs, then what?
If E3 gets 3-1, will you change your tune? Why not?
What if the E3 plant becomes industry norm? That would seem a natural progression. In business history, more efficient methods drive out less efficient methods. So, 15 years from now, we have an industry-wide average of 3-1 energy returns from ethanol. Why not? Suddenly (well, over 15 years) a so-so idea becomes a good idea (for energy security reasons).
On cellulosics, who knows? Cellulosics do not require arable land. You say such plants will not work, but backers are spending hundreds of millions. In fact, the E3 plant is a type of a cellulosic plant, since it burns corn stalks for energy. The jury is still out. And, just as with corn-ethanol plants, the first generation will not be the best.
Maybe at some point we convert corn to dimethylfuran, which has 40 percent higher energy content than ethanol, and some other benefits. Even pig dung – you can get about 1/3 of a liter from each pig each day, by subjecting dung to heat and pressure. You can’t do this manually, so don’t try.
What is certain is this: If the current price regime for oil is preserved (not a given), there will be a biofuels boom. Most ethanol outfits are shooting for \$1 a gallon.
In Brazil, there is talk now that the bagasse portion of sugar cane will also be used, boosting yields by 30 percent. What a boon!
I see end game for the fossil boys. They are destined to become less and less important, a dinosaur industry. This year, maybe next year, will mark the first year of Peak Demand. Then we move to a post-fossil economy, using less fossil oil every year.
Whole nations, such as Sweden and Indonesia, are take themselves off the world oil markets. India planting 60 million hectares of jatropha. We can build solar plants or rooftops.
When oil was \$25 (we forget quickly, it was only a few years ago) none of these other technologies really made sense. In a short four years, we have almost flatlined oil demand, and had a biofuels boom, and some real advances in solar. The trends are just starting. The first PHEV is slated for 2010, but hey, so what if it is 2012 or a couple of years later? Once they work, it will mark the beginnings of radical and permanent reductions in fossil oil demand.

36. I don’t know but it is possible production falls off on the weekend because the impedence of the grid is different then the impedence of the googleplex. When they say 1.6MW, they mean the maximum power they can deliver is 1.6MW at some magic impedence. It the panels drive a dead short, they produce no power. If the panels are open circuited, they produce no power.

37. But, already, E3 is saying they are getting 3-1.

You are wrong, and repeating yourself. You don’t know what they are getting. You are going with what they thought they would get before they started up the plant. You persistently do that – take projections and state them as reality. If the world worked that way, the U.S. would have been energy independent 30 years ago. But we always tend to overestimate what we are going to do with respect to our own energy production.

If E3 gets 3-1, will you change your tune? Why not?

Ben, understand that it is you claiming that we are going to replace 6% of our gasoline without breaking a sweat. My point is that this isn’t true even if all plants retrofitted to do what E3 is trying to do. You consistently ignore energy inputs into the process. You do understand, I presume, why more plants are built on the E3 model? And you do understand that even if E3 gets 3/1, most ethanol producers aren’t going to follow suit due to high capital costs?

So, 15 years from now, we have an industry-wide average of 3-1 energy returns from ethanol. Why not?

Why not? Because nobody is going to tear down existing infrastructure to retrofit their plants. That’s why. Maybe you get the point. I don’t know.

38. JN2 says:

Another possibility is shifting the demand curve to the left, maybe through mandatory time-of-day pricing?

39. Doug says:

Those curves are bang-on. Anyone interested in the cal daily curve can go to the cal ISO’s web site.

The daily demand curve rises from a baseload of 24GW to around 38GW on a typical day, and there are a few hours of lag between max solar production and max demand. (On a really hot windless summer day peak demand could go to 2x the baseload.) The cost of storage will scale with the area under the curve that must be buffered. If you were going to go 100% renewable you might have to look beyond the daily chart to also account for the worst-case run of winter days (or maintain fossil plants as backups).

In looking at the curve it strikes me that nuclear power (plus baseload renewables such as hydro and GT) could provide the steady 24 GW, and then the amount of power storage would devolve to the area of the solar curve minus the area of the demand curve that intersects it, a much smaller number of GW hrs.

Conservation efforts might change the magnitude of the curve but are unlikely to change its shape. This is the reason I believe the grid can’t expand beyond about 20% renewable power without a storage breakthrough. 20% of a lower number is a good thing, but it still leaves the question of where the other 80% is going to come from.

Demand management might change the shape somewhat, though it’s unclear to what exactly. For example, charging of EVs during the day could help tap into excess solar capacity, but then again charging them at night could allow you to smooth out the curve and thus make baseload a greater fraction of the overall total. Besides recharging EVs, it’s unclear how much demand would yield to management techniques. People aren’t going to easily change when they cook, eat, watch TV, need air conditioning, etc. Businesses are hard to run on an as-available basis (especially if a workforce is involved).

I hope you’ll dig into the storage options in more detail in later posts. I’m afraid I don’t know much about the relative merits or potential of batteries, flywheels, thermal mass, etc. The usual references to techniques such as pumped hydro are unsatisfying because those appear to be site-specific and don’t look like they could scale – we can’t use pumped hydro to buffer the 300 GW hrs cal needs at night.

40. For approximation purposes, you could do worse than 180-degrees of a cosine wave centered at high noon.

41. I have to multiply by 8 to get that. In fact, that’s been a pretty consistent theme. So, I think the 5 hour peak is too conservative.

I think that what you are seeing here (particularly with the Google data) is the effect of the long summer days. Production will drop substantially in the winter, and probably average out to 4-6 hours/day equivalent. This will be even more dramatic because the Google array is, I believe, mounted at a relatively low tilt angle. 30 degrees is the theoretical ideal, but 15 degrees is common because you get less self-shading, better aesthetics, and a minimal drop in production. But it does have the effect of shifting production more heavily towards the summer months, because of the angle between the sun and the panels. Which is fine — we need our summer production — but will throw off your estimates.

42. Something to bear in mind: The mismatch between solar production and load is region-specific. The California graph, for example, shows the influence of air conditioning (i.e. SoCal and the Central Valley). Places with different weather conditions will have different mismatches. So the basic problem class is pretty universal, but the solution will be specific to region and climate.

43. Optimist says:

Can’t we just cool our homes to 20C at noon and let them slowly warm up all afternoon.
Well, that won’t work very well, since air has a low heat capacity, i.e. a little bit of heat can increase air temperature quite a bit.

OTOH, water has a high heat capacity and requires a lot of latent heat to melt. Some are using that as a method to store thermal energy, which is one way to cut back on peak demand in areas where a lot of the demand goes to AC. And they seem to be saving a lot of money in the process.

44. doggydogworld says:

For a biofuel with a poor EROEI, rising oil prices means rising biofuels prices. Always.

Robert, this is wrong. Most biofuels require little oil to produce. If oil prices rise materially but coal and natural gas prices do not, biofuel production cost will not rise by a meaningful amount.

Biofuel prices rise with oil prices because biofuels substitute for oil. It has nothing to do with EROEI. As oil prices rise people look to buy more substitutes, and prices of those substitutes then rise. It’s Econ 101.

45. Robert, this is wrong. Most biofuels require little oil to produce. If oil prices rise materially but coal and natural gas prices do not, biofuel production cost will not rise by a meaningful amount.

That’s true in priniciple, but not realistic. Natural gas and oil are very economically intertwined, and their prices tend to track. If oil prices shoot up, you will see some demand shifting to natural gas, and those prices will come up. You are unlikely to see a sustainable divergence in price.

Also, “little oil” is relative. There are still pretty substantial inputs of gasoline and diesel into the mix. The last numbers I saw, it was around 20% of the total BTU value. So, if something that makes up 20% of the overall energy value increases rapidly in price, you are going to need big offsets in other areas to avoid following along.

46. doggydogworld says:

Natural gas and oil are very economically intertwined

They used to be true because dual-fuel power plants would buy the cheaper fuel. This kept prices near BTU equivalence (about 6:1 plus transport adjustments). In fact, it’s a perfect example of the substitution principle I mentioned. NG prices followed oil prices because of substitution by customers. It had nothing to do with oil’s role in NG production.

Dual-fuel plants now play a pretty minor role and the oil and NG markets have de-linked. The ratio is about 12x now, and there’s no sign this is temporary. Also note that more ethanol plants are using coal instead of NG. Coal prices have very little correlation with oil.

There are still pretty substantial inputs of gasoline and diesel into the mix. The last numbers I saw, it was around 20% of the total BTU value.

Which makes it about 10% of the economic value. So if oil doubles to \$150/bbl then ethanol will rise from \$2.00 to \$2.20/gal. That’s pretty weak correlation.

Of course that ethanol won’t sell for \$2/gal very long with oil at \$150/bbl and gasoline at \$5/gal. Substitution will drive ethanol up to about \$3.50/gal.

I’m not saying the price of oil as an ethanol input has zero effect. But it’s clear substitution by customers is a much larger factor. Of course substitution requires infrastructure. Today’s ethanol transport and blending infrastructure can’t handle all the new ethanol coming online, so ethanol prices have temporarily de-linked from gasoline. If the infrastructure magically upgraded tomorrow ethanol futures would jump 50%. As an investor I’m trying to find a way to play this.

47. So if oil doubles to \$150/bbl then ethanol will rise from \$2.00 to \$2.20/gal. That’s pretty weak correlation.

Without looking (because I don’t have time right now), I bet if we laid natural gas and oil on the same graph and tracked them for a year, there is going to be a very high degree of correlation. There are still a lot of dual fuel usages. Refineries are a prime example. And of course \$150 oil drives up the price of a lot besides just the gasoline and diesel inputs.

Your point about coal is accurate, and one I have noted. If ethanol producers rely more heavily on coal, you will see the price influence due to the energy inputs decrease a lot.

48. doggydogworld says:

Back on topic, US peak generating capacity is around 1 terawatt. 1 TW of solar panels would cover a 53×53 mile square and cost about \$7 trillion.

US electric consumption is 4000 TWh/year. That’d take about 3 TW of panels (2 TW if all were in the Southwest), which would cover a 93×93 mile square and cost \$21 trillion. Plus perhaps \$10 trillion for storage.

Land (or roof) area is not a problem for solar, but \$31t is a non-starter. Our current electrical system is worth \$1-2 trillion. We need an order of magnitude cost reduction:

-First Solar claims \$1.40/W, but that’s for cells (not installed panels) and Cadmium usage presents issues.

-SES (Dish Stirling solar thermal) claims sub-\$2/W in high volume, but hasn’t demonstrated this yet.

-Nanosolar continues to promise the moon (er, sun). Most recently founder/CEO Roscheisen mentioned \$0.99/W in an interview.

-Organic solar cell ivory tower types keep throwing around \$0.10/W numbers.

We’ll always have a mix of energy sources. Wind is already at the \$1/W ‘magic price’ and becoming a big player. Nuclear’s 20% share could easily grow to 50% and provide all our baseload. Demand management, especially ice cooling and PHEVs, could drastically reduce renewable storage needs. We can’t do any of this tomorrow, but over 30 years it’s not difficult.

49. doggydogworld says:

I bet if we laid natural gas and oil on the same graph and tracked them for a year, there is going to be a very high degree of correlation.

You’d lose that bet. I don’t have an overlay chart handy, but here are two separate graphs (open in new pages to compare side by side):

The last six months are especially divergent. NG was flat around \$7.50 until recently falling to \$6 while oil has gone straight up.

50. The last six months are especially divergent.

I would like to do the actual correlation. I think the scale is fooling you. Natural gas rose quite a bit from January, just as oil did. But without actually doing the correlation, it’s hard to know just how correlated they are. That side by side business is hard to do, especially when the scales are so different.

Maybe later. I am still trying to finish this chapter. It’s due tomorrow, and then I hope to write a bit more.

51. Anonymous says:

doggyworld said

“We’ll always have a mix of energy sources.”

Well said. I am writing a reserch paper on sustainable energy for my English class and I’ve learned that several sources “could” supply all our energy needs.

Nuclear, solar and geothermal could all supply thousands of times our needs. Wind is about five times the worlds current use, but I doubt we would want to see that many wind turbines.

Even though we could replace all fossil fuels, it doesn’t make sense to do it for a long time. We will slowly phase them out over many decades.

I’m glad that each industry is trying to expand their own source though, that means a more secure energy future.

I think the biggest dangers to a secure energy future are the extreme environmentalists and the malthusians who influence the politicians.

52. Would not 70G be overkill? Any production over real-time usage, yet under that 70G target, would be stored and then time-shifted to peaking. Storage can approach 100% efficiency with most heavy loads such as air conditioning if you store the energy as work (ice) rather than an intermermediary resource.

53. Btw, should of added this to the last post. If you look at the curve as Gigawatt hours it looks very different. Peak GW only measures the top range of electric delivered to the grid, not actual usage. This is where real-time load following storage solutions (solid state battery and “stored work” batteries like ice) shine in reducing the solutions costs.

54. “Optimist said…

Can’t we just cool our homes to 20C at noon and let them slowly warm up all afternoon.

Well, that won’t work very well, since air has a low heat capacity, i.e. a little bit of heat can increase air temperature quite a bit.”

Indeed work is being done to exploit the original commenter’s idea. Purdue U is working on load balancing logic to store cooling in air.

Researchers study energy-saving
method for small office buildings

Their idea is a mirror image of the needs of solar-energy and uses the existing regime of overnight surpluses and overnight cooling gradients to prepare the space cooling for high-density populations. Much of the chilling extends to interior thermal mass cooling, reducing the “too cold” problem deep chilling can present.

A real-time application during the day could exploit the lower gradient of cooling air versus the larger gradient of chilling ice. It’s much more efficient per degree to chill air on demand than storing coolth in high-density form. A smaller ice/ice water store could be added for longer-term back-up and system lag reduction.

55. JN2 said…
“Another possibility is shifting the demand curve to the left, maybe through mandatory time-of-day pricing?”

This already exists. \$0.02/kWh between 10PM and 10AM and \$0.17/kWh between 10AM and 10PM last time I checked for average rates schedules. Raising rates even further during the post-afternoon bump would catch too many consumers who can’t time-shift. A better solution would be Smart Appliances (A2G) and Smart Cars (V2G) hooked up to a Smart Grid. Smarter People would be nice too. 😉

56. Robert Rapier said…
“Why not? Because nobody is going to tear down existing infrastructure to retrofit their plants. That’s why. Maybe you get the point. I don’t know.”

Ouch. Using Ben as a whipping boy much Robert? 😉

15 years is a lifetime for any industrial installation, let alone a 21st century version. When the time preferences of forward-looking investors catch up to the market conditions of peak oil you’ll see large step functions of capital investment.

Also, just because God Ethanol can’t provide 100% import, let alone total resource replacement doesn’t mean bad-mouthing the offered solutions is rational. Even sub-1 EROEI makes thermodynamic and economic sense for the corn ethanol interests as most of the value is in food co-product.

57. 15 years is a lifetime for any industrial installation, let alone a 21st century version.

Companies may amortize over 15 years, but equipment in the oil and chemical industry lasts a lot longer than that – 30 years is generally what we assume. So, in reality what will happen is that if E3’s concept works as expected, the only producers who are going to consider using it are the ones who haven’t yet built plants. In 15 years, ethanol producers will still be setting on the same equipment they have now, and they will weigh the benefits of decommissioning existing equipment to upgrade, or just keeping the status quo. Unless the payback is outstanding and the risks low, they are going to sit on the status quo.

58. Robert Rapier said…
“Google claims their system is rated at 1600KW. I’m surprised and disappointed they only peaked at 877 KW on a sunny summer day.

If you look at the entire week, production is higher during the normal work days. Not sure why it should fall off on the weekend. I know that demand falls off, but why should production?”

A recent climate study found correlation between work-week emissions and heavier weekend precipitation. Perhaps this is in play?

With California’s maritime/sheltered valley climate it would be interesting to look at what effect prevailing winds have on That-Which-Will-Have-Been-Called-Google’s powerplant output.

59. Robert Rapier said…
“Companies may amortize over 15 years, but equipment in the oil and chemical industry lasts a lot longer than that – 30 years is generally what we assume.”

How much of that investment is different between systems? How much of that investment turns over in 30 years?

Robert, like most EROEI hackers you seem to slack off when it comes to boundaries. Your capital investment projections for biofuel powerplants and refineries are a perfect example, almost as bad as Pimental et al’s de-pop EROEI work. Assumptions of single outputs at a singular price point not corrected for even by inflation or scarcity let alone innovation and adaptation.

No industrial ecology thinker would ever design an installation with the sanitized input-output projections nor by assuming a static process flow design.

60. For example…

Robert Rapier said…

On theoildrum you said that biofuels can only replace a fraction of our petroleum needs; what fraction is that? 1/3, 1/2, 7/8?

“Realistically speaking, based on currently existing technology, probably no more than 10% on a net basis. If BTL works out, the equation will look quite a bit better because any biomass source can be converted.”

“based on currently existing technology”???

Meaning based on currently installed capital capacity? Surely not currently understood technique.

BTL??? How ’bout BTG? Clean natural gas from bottoming-cycle poopmass. Use hydrogen from the small BTL plant used to supply the legacy ICE vehicles to upgrade high-carbon feedstocks and then recycle the H2. Or use surplus electric from the BTG powerplant to gasify the previous day’s biomass waste accumulation?

Conservation, effeciency and exploitation = problems solved.

People that can’t work with that need to die-off and free up resources for the Brights.

61. How much of that investment is different between systems? How much of that investment turns over in 30 years?

Not that much. That’s the whole point. Money is spent on maintaining, not tearing out and replacing. That’s why I say the EROEI won’t change much if E3 is successful. It will take a very long time to creep up. In fact, I would make a bet that in 10 years the EROEI of corn ethanol won’t be that much different than it is now. The industry average – take it to the bank – will be less than 2 to 1. That is, if we haven’t collectively awaken by then and stop subsidizing this nonsense.

62. Meaning based on currently installed capital capacity? Surely not currently understood technique.

No, based on what we current understand. If you look at yields, trends, available land, energy inputs – then biofuel usage is unlikely to reach 10% of present petroleum consumption. I did leave that disclaimer off of the initial comment, but to me that’s implicit. Of course if we reduce demand by 95%, then biofuels can displace a large percentage of petroleum consumption.

63. No industrial ecology thinker would ever design an installation with the sanitized input-output projections nor by assuming a static process flow design.

What I can tell you is that I work in the real world of energy production, and I know how slowly these things turn over. I am not talking about some hypothetical BS where you have an “ecology thinker” constantly redesigning systems. I am talking about reality, where owners install equipment, and then are very slow to change it out unless there is an incredibly compelling reason to do so.

That’s not theory. That is reality that I have observed since I was 18-years old and working in a Campbell Soup factory.

64. Even sub-1 EROEI makes thermodynamic and economic sense for the corn ethanol interests as most of the value is in food co-product.

One wonders why they need subsidies then. But you are wrong. Very little of the value is in the food by-product (which incidentally, is animal feed byproduct).

The only situation is makes economic sense (in an unsubsidized world) is if natural gas and coal are very cheap, so the energy inputs are low. In that case, you can get away with turning 1 BTU of coal into 0.8 BTUs of ethanol. But it isn’t very environmentally sound.

65. Your capital investment projections for biofuel powerplants and refineries are a perfect example, almost as bad as Pimental et al’s de-pop EROEI work. Assumptions of single outputs at a singular price point not corrected for even by inflation or scarcity let alone innovation and adaptation.

Incidentally, you may want to clarify here, because I don’t know what you are talking about. What are my capital investment projections? That we tend to install equipment and use if for 30 years? That wasn’t a projection.

Can you provide more information on the meaining of your last sentence there? It isn’t clear to me what you mean. It appears to me that you are making assumptions based on things I have not said.

66. doggydogworld says:

I am right about the correlation

Your chart only goes to MAY06. Extend it to present day and the divergence shown in the chart’s last 7-8 months will stretch almost two solid years. That’s no temporary blip — the 6:1 relationship has been broken. We’re at 12:1 now. Maybe we’ll return to 6:1 some day (though futures markets think not), but for the time being the idea that oil prices are highly correlated with NG prices and thus a major ethanol cost driver is clearly not valid.

67. A recent climate study found correlation between work-week emissions and heavier weekend precipitation. Perhaps this is in play?

Interesting correlation. It didn’t precipitate that weekend in Mountain View, CA where the Googleplex is, but there were clouds part of the day, so maybe there’s an effect.

68. Jim Coombes says:

I’m not an engineer, so I’m reserved about posting alongside y’all. There has been quite a lot of talk about scaling up solar technologies, biofuels, and even nuclear.

When estimating the COST of massive scaling up of anything (energy or anything), the discussion is always more complex than taking today’s “price” (subjective or incomplete in its own right) and simply multiplying it by a bazillion.

For solar equipment, I’d note that quoted prices obviously include the effect of subsidies in the places they are sold. There is a reason that SunPower, for example, tells you that they get most of their revenue from California and New Jersey (and past/future in Germany and Portugal). One, its because of subsidies. Two, its because there is therefore risk. At a minimum, add in the current subsidy that Taxpayers are shouldering.

Moreover, as I’ve pointed out before, when you’re planning to produce hundreds of times more PV-grade silicon than has been produced in the history of man, realize that historical costs don’t remain linear. I mean, this is precisely why we’re talking about this, right? Because OIL isn’t staying linear. WSJ today repeats the story that oil infrastructure costs are skyrocketing. Same is happening for PV, and although that will “ease” in 2008, just bookmark this note for 2018 when we’re either (1) building PV farms everywhere to try to meet these projections, or (2) listening to people tell us how solar is *still* right around the corner.

Solar today is an infinitesimal contribution to our energy needs. Any projections that don’t address the deeper issues of silicon supply and cost, grid connection and upgrade costs, etc. are going to look naive even ten years from now.

This is why I ask the question: Should we scale back on solar installations, or at least subsidies, for now, and give the technology time to improve before throwing money at the problem?

It would also be helpful if the messianic global warming crowd would allow at least a little debate on the timing and magnitude of the crisis, so we can decide the TIMING of our response strategy. Can we wait 50 years for solar, or do we need nuclear plants today?

69. Robert Rapier said…

“Can you provide more information on the meaining of your last sentence there? It isn’t clear to me what you mean. It appears to me that you are making assumptions based on things I have not said.”

Previous comments on biomass refineries compared to fossil fuel refineries. Much of the business-as-usual has great diseconomies of scale not to mention state perversion of the free market.

Most value in the biofuel industry is indeed contained in the food co-product (and fertilizer and fiber). Just ask someone who’s hungry and looking at a gallon of biodiesel or a gallon of milk. But I guess you could drink the oil or ethanol. 😉 Biofuel is a small energy recovery and import replacement component to an integrated solution for agri-communities. 100% displacement for the Amerikkkan Prole is a strawman, an exagerated goal which distracts from the real value of biofuel.

Information science has a place in this: caching. Send once, use many times. Import diesel fuel and embodied energy in the form of machinery: make EROEI 1 100% import replacing fuels like alcohol, corn biodiesel, substitute natural gas, and fertilizer from the ash and char. Rinse. Repeat.

70. Robert, your latest blogging speaks to what I was getting at with the relative values of the biofuel process co-products.

Btw, I think you mistyped “a pound of coal” twice.

“Consider that your goal is merely to make money. You may be able to make lots of money with a process having an EROEI of less than 1. You can take a pound of coal and use it in an ethanol process to make a pound of coal. Considering only your energy inputs, you have increased the value of your BTUs by a factor of 10. So, even if you take 1 BTU of coal and convert that into 0.7 BTUs of ethanol, there may be plenty of economic incentive to do it, despite the energy returns.”

Replace “money” with exergy or emergy.

71. Robert, your latest blogging speaks to what I was getting at with the relative values of the biofuel process co-products.

But you see, I know all that, and was in the process of incorporating that into the FAQ, which is why I took exception to your EROEI hacker comment. I understand the uses and limits of and EROEI approach. It can’t be done in a vacuum.

Btw, I think you mistyped “a pound of coal” twice.

Sleep deprivation. I wrote it, and went to bed. I have to stop doing that. Thanks for catching it.

72. “But you see, I know all that, and was in the process of incorporating that into the FAQ, which is why I took exception to your EROEI hacker comment.”

HA! That’s not something you need to take exception to Robert, just something you need to accept and answer to. Economic preferences and technical expertise allow for honest disagreement on supposedly hardcore sciencific questions. If you want to play in the cesspool of statistics (and damned lies) you need to see the criticisms coming. And EROEI hacking is fun. Almost as fun as biohacking microbes to make them do your bidding. 😉

73. “You can take a pound of BTU of coal and use it in an ethanol process to make less than a BTU of ethanol.”

Another late night? 😉

I’m a free-runner so I dunno about thse late-night writing errors. I’m just regularilarilyy sloppy.

74. “Doesn’t the Ethanol Subsidy Actually Benefit Oil Companies?”

Oh, this new part to your post should include an oil industry subsidy down-to-the-cent (it’s in the the dollars-per-gallon range thanks to inflation).

75. Oh, this new part to your post should include an oil industry subsidy down-to-the-cent (it’s in the the dollars-per-gallon range thanks to inflation).

Two problems here. First, the subsidies are indirect, and everything gets lumped into 1 big pot and called an “oil industry subsidy.” If some subsidy is available for small producers, and it isn’t even used, it still gets lumped into the oil industry subsidy pot. I have seen some of the more ridiculous estimates that would amount to subsidies greater than that entire U.S. federal budget. This is the kind of nonsense that happens when all of this stuff is piled together.

Second, any subsidy that benefits the oil industry is a secondary subsidy to the ethanol industry, since they use petroleum products to produce ethanol.

You might be interested in this, from a pro-environmental website on this very issue:

http://zfacts.com/p/63.html

“Ethanol Today,” (8/’05) states “Five years ago, a US General Accounting Office report showed that ethanol had received \$11.6 billion in tax incentives since 1968, while the oil industry had received over \$150 billion in tax benefit over the same period.

Probably true. But the oil industry produced 1068 times more energy so the subsidy rate per unit energy was 54 times higher for ethanol. That’s like ethanol gets 54¢ and oil gets 1¢. Now if we had oil subsidies, and we do, and ADM is making more profit than …

You just can’t compare indirect subsidies to a direct per gallon credit.

76. Yes Robert, all subsidy to the American petroleam industry is really just a subsidy to the the local grocery store. I got it.

😉

77. And why do oil companies even need any subsidy? Protection from predatory ethanol companies?

Can’t we all just get along (without socialism?)

78. And “environmental” was a cueword for “need to shut-down critical thinking functions” I think.

79. Yes Robert, all subsidy to the American petroleam industry is really just a subsidy to the the local grocery store. I got it.

Most of the things that could be called an actual petroleum subsidy are going to small, marginal producers for things like stripper wells that would otherwise be unprofitable. They get some tax advantages and such. The ethanol subsidies are going largely to ADM and Cargill.

80. By the way, don’t think that I am defending oil subsidies. I don’t agree with any subsidy that encourages fossil fuel consumption. I am just telling you what they are. But I would prefer not only to eliminate the subsidies, I want to sharply raise gas taxes to stem some of the fossil fuel usage.

Also, if you think the answer to the question “Who benefits from ethanol subsidies?” is to discuss oil company subsidies, I encourage you to write a FAQ. Because in my FAQ, that would be the answer to a different question.

81. Anonymous says:

FWIW, the question of correlation between PV generation and system peak in California has received a lot of attention from CA utility companies and the CEC. Historical data shows PV generation as a fraction of installed PV capacity, at the time of utility system annual peak, running in the 39-46 percent range. There is some evidence that the time of daily summer utility peak demand (which averages about 4 pm) is earlier on the most extreme days of the year, with annual peak demand more likely to occur between 2 and 3 pm. So the PV performace on the extreme peak hour of the year may be more like 50% of installed capacity or a touch higher, while being closer to 40% averaged across all the summer daily peaks.

Winter peaks in California come in the 6-8 pm range, when solar generation is zero.