1.9: Geothermal

Eli Dourado gives a great overview of geothermal here.Geothermal energy has arguably been a kind of neglected Cinderella-like stepchild of the renewable energy space — but the fact remains, it is carbon free, baseload, and dispatchable.Iceland is technically “hydrothermal”, with natural permeability and replenished water close to the surface.Right now, geothermal energy is only accessible in a tiny number of regions where there is easily accessible high-temperature heat close to the surface. For example, Iceland has a unique geography that allows this. Currently geothermal in the USA constitutes a small fraction of a percent of all energy generation.

If you could drill deep for cheap, down to a few kilometers say, then you could access 250 degrees C geothermal heat cost-effectively in many more places — here is a temperature map at 4.5 km subsurface depth

4.5km geothermal heat map

A recent techno-economic simulation from Princeton (not yet peer-reviewed) studied the effects of applying a few steps of industry “learning curve” to enhanced geothermal technology using today’s drilling costs (much reduced of late due to progress made in drilling technology during the shale boom)

LCOE = levelized cost of energy, this basically refers to how much each unit of energy would have to cost to make a certain return on the building of a power plant given how long it operates for and how much energy it can extract

EGS = Enhanced Geothermal Systems

“development of EGS in regions with temperatures higher than 250 degC and depths above 3 km can be economically feasible at costs less than 7.2 cents/kWh in the near future. The building of these initial resources can yield subsequent improvements to drilling cost, power plant cost, and production well flow rate, further driving down costs and allowing for development of additional temperature/depth intervals. Depending on assumptions of initial LCOE thresholds and learning rates, this learning process could potentially unlock several hundred gigawatts-electric (GWe) of EGS electricity generation potential in the contiguous United States”

That’s pretty exciting — average US power consumption is a few terawatts, and we want variable renewables to make say 80% of that, with the rest being baseload and dispatchable to help with grid stability. So, 20% of 3 terawatts would 600 gigawatts… and this thesis is saying they think economical clean geothermal power could provide several hundred gigawatts in the near future! It is therefore important to invest to get the industry started down these learning curves ASAP.

Further, and this is a bit more speculative, if you can drill really really deep into hot dry rock, ultimately down to say 10 km, much of the USA has high-temperature heat that could be used for geothermal. Therefore, the theoretically-available geothermal capacity is huge, actually larger than oil and gas combined. Anything hotter than the light orange in the below plot would be good quality heat for geothermal.

10km geothermal heat map

How can we access this huge untapped capacity? There are a couple key technical issues:
  • Actually drilling wells that deep, in hot dry rock
  • Making sure that the overall system cost is low enough that the resulting energy is cost-competitive with other sources

Then there are the logistical and financial issues of how we could get those technical problems to be solved, and soon, e.g., by incentivizing talent from the oil and gas industries to “pivot” (as was the title of a recent conference) to pushing geothermal tech instead.

Let’s consider each of these in turn.

Drilling cost

First, the drilling. From this old MIT report we have a cost model for well drilling in the geothermal and oil/gas sectors:

cost of geothermal drilling graph

The Wellcost Lite model (solid red curve), from Sandia National Lab, suggests about $30M drilling cost for a 10 km well. In practice, most oil and gas drilling is <3 km or so, which is much shallower. So most of the equipment for drilling as well as for sensors and such in the wells has not been optimized for the higher temperatures and pressures. To get to ~10 km level wells in hot dry rock, the equipment is going to need to be hardened. Also, there is likely room for innovation on the underlying drilling methods and costs. The company Strada Global thinks they can do both things.

According to this blog by ethernet developer Bob Metcalfe:

“I told Ben it was our aim at GEO to cost effectively reach 10km and 350℃ for the prize of enabling globally available baseload electricity production. He answered, “No problem.” And he added there is a lot of low-hanging fruit on our way to baseload electricity generation.”

Of course, that is a startup CEO saying that, so take it with a grain of salt. But drilling this deep should likely be possible. They also say on their website they think they could reduce drilling costs by 70%, basically by drilling faster if I understand correctly.

Metcalfe estimates a $20M, 10 km deep geothermal well in the near future in the above post.So let’s say a reasonable cost range, which we’ll use later, could be between say $10M and $30M to drill a 10 km well into hot dry rock, if new technologies and industry learning curves start kicking in. Getting there, especially at the lower cost level, will involve a concerted amount of technology development and iteration.

Energy cost

Second, there is what energy cost you can actually deliver. Being cost-competitive is crucial for geothermal to make sense as a business.

This key issue of cost is why, at one level, it is super interesting that major oil and gas execs, like the former chief scientist of Shell, Lance Cook, are getting into new geothermal ventures today — they are seeing a path to favorable economics.

Let’s digress to see what Cook is up to. His company is called Sage Geosystems. Why is it an exciting economic play? Because it proposes to re-purpose the existing oil and gas wells, which are getting left behind as stranded assets. How do they rescue those wells? They go to the bottom of them and then frack purely downward (a technical feat, but this is former CSO of Shell, so let’s believe it), to access heat that is held deeper below.They then propose to extract that heat with a so-called single-well closed-loop (or ‘pipe in pipe’) system, where the fluid being heated can circulate down and then back up inside the single original oil well.

We’ve established above that it is not totally unreasonable to expect that future 10 km drilling itself could cost $30M. Let’s assume we can have that price include the steel wellbore casing, too. That presupposes some cost improvement, but it doesn’t seem crazy. What about the CAPEX costs other than the well itself, i.e., the steam turbines and cables and such. From the old MIT report we can see that, depending on flow rate, a geothermal plant operating at 350℃ can cost on the order of $1/Watt to $2/Watt, and can output tens to low hundreds of MegaWatts (MW).

Cost of geothermal with utilization graph
Anyway, let’s say it is doing 30MW, at a specific plant cost of $2/Watt. Then, adding in the well drilling cost, which is $30M for that same 30MW, and thus a $1/Watt well cost, we have a total CAPEX of around 2+1=$3/Watt.

This interest rate accounts for the borrowing cost as well as the premium the owners hope to get over risk free assetsHow does this work out for the consumer? We can plug K=$3/Watt into the following formulas for the levelized cost of energy (LCOE), assuming an interest rate of r=7%, a capacity factor of CF=70%, and N=30 year plant lifetime:

CRF is the capital recovery factor which basically aims to give us the present value of a set of cashflowsLCOEK=KCRF8760CRFLCOE_K=\dfrac{K*CRF}{8760*CRF}


This gives about 4 cents per kWh. Compare this with John Platt’s “miracle” threshold of $2.2/W (baseload).

Given various inefficiencies in the system depending on how much of that supposed 30MW we can actually capture as electricity using which kinds of turbines with which efficiencies, it seems we might be significantly worse than that, depending on the details, but there is clearly a lot of wiggle room. Although new technologies may allow greater efficiencies than in the past when dealing with lower temperature heat, as well, and see the notes here about use of supercritical CO2 as a working fluid.)

Metcalfe talks about cost innovation potentially getting the field to 1 cent per kWh:So, we’re not looking too bad, perhaps, with the crudest and most out of date imaginable outsider’s cost calculations (which, for drilling cost data, for example, pre-date the fracking boom), but regardless, there almost certainly can be and needs to be some major further cost reduction here through improved technologies and methods.

According to Jamie Beard, a key person leading the Texas GEO center (who strongly informed the above narrative), core tech development projects needed here include higher temperature and pressure directional drilling hardware, making drilling and overall well construction faster and cheaper (using techniques like pad drilling and managed pressure drilling, which would significantly lower the cost versus something like the Wellcost Lite model above, as well as coatings rather than casings to seal the wells), and incorporating exploration/surveying and monitoring technologies, e.g., sensors, that work at the high temperatures involved, e.g., 350C. Here are two videos from the Pivot conference that explore this. Using technologies like new and more efficient supercritical CO2 turbines (GreenFire is already using supercritical CO2 in their system), which would allow operating at lower temperatures and hence depths, could be important too.

There is also the issue that a lot of this could start much more easily and cheaply in sedimentary rock, but drilling in igneous rock is going to lead to more possibilities for scale. And of course, the field needs to converge on overall system designs that optimally integrate all these primitives. These are all technology development and integration challenges that seem surmountable, if intensive.

Long term, you can perhaps imagine, for instance, a 5-10 km deep supercritical CO2 system with novel highly-efficient turbine design, drilled quickly and at reduced cost with next-gen drilling and lower cost well casing/coating (like the ones from Eavor) methods — a system that is pretty simple, robust, can be put almost anywhere, and generates a large and constant baseload power supply.

Unlike solar and wind, this would not require major changes to storage and transmission infrastructure. It would also be quite physically compact, as well as low maintenance. Could that get close to 1 cent per kWh LCOE? Maybe it could!

With a concerted effort to bring those methods to a level where for-profit investors can come in on deploying advanced, e.g., 10 km deep systems (or less deep systems using, e.g., supercritical CO2 cycles), and an associated push to draw talent from the oil and gas industries, it may indeed be possible for geothermal to make a large contribution to clean energy generation over the coming 2-3 decades. The above is why advanced geothermal may make for an important mid-stage R&D target, to bring the relevant technologies underlying cost reduction and ultra-deep drilling to a technology readiness level (TRL) where they can be taken forward further by venture investors. What this makes me think is that policymakers should be working hard to incentivize the “Pivot” by the oil and gas industry that was the subject of Jamie Beard’s above-linked conference at Texas GEO — there is a lot of headroom in this technology, but it needs a big push of exactly the kind that a big industry, if it indeed pivoted, could make.

What's the catch

Well, one issue is that many of these technologies require “stimulating” the rock to increase its heat conductivity. This is pretty similar to “fracking”, and although in many cases it can be more controlled, and of course you aren’t fracking in a location that could release hydrocarbons, there is a potential non-zero seismic risk that may have to constrain the exact sites where this occurs, and cause some slowdown in environmental approvals for such projects even where the actual risk is or can be made very low.

Is there any possibility to avoid “well stimulation” entirely? To allow this, people are developing “closed loop” geothermal systems. To explain this a bit, in a normal geothermal system today, you need to create a loop where cold fluid goes into one vertical well and comes out another. To connect the two wells, one creates a fracture network in the rock between them.

Quoting from the DOE website, they indeed define the technology as involving fracking: “During EGS development, the injection of fluid into the hot rock enhances the size and connectivity of fluid pathways by re-opening fractures.”This is what so-called Enhanced Geothermal Systems (EGS) generally do. They randomly frack and try to make a circulating flow of water from cold to hot, with access via two vertical wells. The US Department of Energy (DOE) has put in a lot of effort towards incrementally improving and widening the scope of utility of such systems, e.g., through the FORGE project

Fervo Energy is a cool company pursuing that, funded by Breakthrough Energy. It’s a smaller and more controlled perturbation to the rock, but still a perturbation. (Incidentally, Fervo is also using distributed fiber optic sensing, something Rodriques and I suggested for another application — the brain.With “directional drilling” technology, you move into what are called “Advanced Geothermal Systems” (AGS). Instead of fracking between vertical wells, they can make a turn and drill horizontally at the bottom of a well. Using this, they can make two horizontal wells on top of each other, and only frack in a controlled region in between those. What this allows is a more controlled process that will yield successful geothermal installations more reliably in more locations.

So can one get rid of fracking entirely? The company Eavor is trying to do this with a “closed loop” system. They basically drill a “radiator” using advanced directional drilling technology. No fracking required. This is pretty amazing, in theory. Here is a picture from their website where you can see the radiator:

diagram of evaor system
In the geothermal space another super interesting effort is Quaise. They are doing a new type of drilling that uses electromagnetic waves produced by plasma to blast through rock. It emerged from the work of Paul Woskov at MIT Plasma Science and Fusion Center. Here is a paper on “Penetrating rock with intense millimeter-waves“. They are talking about going to 20 km deep. That makes things very hot and would really change what one can do.So this sounds awesome. But how realistic is it to use closed loop broadly to avoid any “well stimulation”? This is where it gets unclear. Can you really access enough heat by drilling these radiators? If you think about it, drilling costs are a key part of the geothermal economics. A radiator like the one above requires a lot more drilling, but may ultimately create much less surface area for heat exchange than the random fracture networks created in rock by “well stimulation”. This means you may be paying more for drilling, but getting less energy out, which will ultimately hurt the economics. How badly? Well from this paper from Stanford it looks pretty bad — the plant cost per Watt is huge, much worse than current nuclear. Here is another paper that concludes “the simulation results show limited thermal production capacity of the downhole heat exchanger configuration”. Granted, these are specific designs — they don’t rule out every closed loop design. Maybe there will be a technology breakthrough allowing full closed loop systems with no well stimulation to access much larger surface areas for heat exchange without massively increasing drilling costs, but it seems unclear as of yet what this will be.

For further reading, David Roberts recently covered this space in an excellent article:Overall, for drilling, it looks like we should aim for drilling systems with rates of penetration of >100 meters per hour that can operate down to >10 kilometer depths in hot (>350 Celsius) dry rock with minimal need for drill bit replacement and the ability to drill both vertically and horizontally, for example.