While I am a skeptic by nature, I am a problem-solver as well. I always ask the people around me – and I try to practice this at all times – if you come across a problem, or envision that you will have a problem, try to envision a solution as well. Otherwise, you have simply created an obstacle. This is important in my current job, as we have commercialized a technology that had never been commercialized before. It’s sort of like chess: Envision where you are going, the potential problems as you make your journey, and how you will cope with that. The more contingencies you have planned for, the higher your chances of success. These are principles I live by. The reason I feel the need to point this out is that some read my skepticism in some quarters as a “can’t do” attitude.
With that preface, there are a couple of things that I left unresolved in my Coskata investigation. One was my questions on the energy balance. But I also wanted to do a mass balance – or at least a carbon balance – around the process to see if those claims of 100 gallons of ethanol per ton of woody biomass are realistic. I also want to look at the logistics to get a feel for the amount of biomass required to run a 100 million gallon a year plant.
Finally, I will offer some advice to someone thinking of investing into Coskata, or any energy startup. I am not going to offer up solutions to potential problems simply because I would require quite a bit more information to do so. But what I can do is flag various areas that a prospective investor should investigate.
Let’s take the mass balance first. Woody biomass contains around 50% carbon. I have that from personal conversations with Roger Rowell, one of the world’s foremost wood experts and my part-time room-mate in the Netherlands. But if that’s not good enough, here’s another source that indicates that 50% carbon is a reasonable estimate. Therefore, in a ton of dry, woody biomass there is a half ton of carbon. The atomic weight of carbon is 12, and there are 454 grams in a pound, so the number of moles of carbon (remember freshman chemistry?) is 1,000 pounds * 454 grams/(12 grams/mole) = 37,833 moles of carbon.
Each ethanol molecule has two carbons, so if you had 100% conversion of the woody biomass to syngas and then to ethanol (of course you won’t, but I want to get the maximum theoretical yield) you would have 37,833/2, or 18,917 moles of ethanol. The molecular weight of ethanol is 46, so if we convert the ethanol into pounds we get (18,917 moles * 46 grams/mole)/454 grams/pound = 1917 pounds of ethanol. That’s the absolute maximum theoretical yield based on carbon, and assumes that we have added enough oxygen to the mix (during the gasification step). The density of ethanol is about 6.6 pounds per gallon, so this leads to 1917 lb/(6.6 lb/gal) = 290 gallons of ethanol per ton of woody biomass. (Note that in reality, depending on the specific metabolic pathway, there may be CO2 produced whenever a molecule of ethanol is produced, lowering the theoretical yield).
Conclusion from that exercise? Some of the available carbon will go into microbe production, and some will end up as carbon dioxide. Some will be lost as tail gas in the process. But if 100 gallons is converted to ethanol, that means only 34% of the carbon in the starting biomass ended up as ethanol. Therefore, claims of 100 gallons (or more) per ton of woody biomass are consistent with the chemistry.
This is where I feel like there is a problem. Let’s put all of the energy inputs and outputs out there and see.
Here is a reference for the BTU content of woody biomass. As you can see, the energy value varies quite a bit depending on moisture content and type of wood. The numbers are clustered around 12.5 million BTUs/ton, so I will use that as a standard. Coskata reports that a ton of woody biomass will produce 100 gallons of ethanol. As noted in the previous section, that is a believable statement. This much ethanol contains 8.4 million BTUs (based on the higher heating value, as was the case with woody biomass). The problem though with calculating an energy return is that there are energy inputs that go into producing the oxygen for the gasifier. And air separation units suck up a lot of energy (and capital).
But I can do a different exercise. If I have a solution that is 3.5% ethanol, as Wes told me their fermentation broth is, how much energy does it take to get it out? If I had access to a process simulator – and I don’t have one here in the U.S. (but I do in the Netherlands), then we could actually determine the break even point; that is the point at which the energy I put into the separation is equivalent to the energy of the ethanol I am separating.
But I can do a crude illustration. If I have a pound of fermentation broth, then there are 0.965 pounds of water and 0.035 pounds of ethanol. The amount of energy in that much ethanol is 0.035 lb * 1 gal/6.6 lb * 76,000 BTUs/gal = 403 BTUs. The heat of vaporization for water is 970 BTUs/lb, so if you were going to vaporize the mostly aqueous mixture (which you would do in a conventional distillation) it would take around 940 BTUs – more than twice what you could get back in the form of purified ethanol.
In a corn ethanol plant, the fermentation broth comes off at 16% ethanol or so. For our same exercise above, there are 1840 BTUs of ethanol in the mix, which is well more than enough to justify vaporizing the mixture. That should roughly illustrate the mountain that a 3.5% ethanol mixture has to climb.
Of course that implicitly assumes that the value of the BTUs that are being used to separate the ethanol are roughly equivalent in value to those in the ethanol. That may not be the case, and there may be times where there is an economic justification. For instance, let’s say you had a bunch of waste heat that you can use. It might make sense. But as always, I would ask the question whether boiling water is the most efficient usage of those BTUs.
Coskata says they have addressed the energy problem in the distillation by using membrane technology. The claim is that it takes half the energy of distillation. This is somewhat hard to believe, as I would expect ethanol plants across the country to rush to adopt the technology. And it isn’t brand new. Here is a 2001 article talking up the benefits: Pervaporation comes to age.
Yet there have been numerous ethanol plants built since 2001. Why aren’t they being built with membrane separation technology? Without going in and checking their claims, I can’t say for a fact whether that claim of lower energy usage is valid. But there are question marks all around it. (Note: I don’t dispute the technology, because I know that it works. I would just make sure – if I were about to invest in Coskata – that I had a very close look at their claims around this area.)
Finally, what of their claims that they get “up to 7.7 times more energy than what is used in making the ethanol.” In my conversation with Wes, I had asked if this was from Michael Wang. He said yes, which then put that claim into context for me. Michael Wang has created a model that has been widely misused. The number above – 7.7 – will refer not to the energy that is used in the process but rather to the overall fossil energy used. This is the same way Brazilian sugarcane can claim an 8/1 energy return, despite the energy intensive process step of separating ethanol from water. This is a valid metric as long as the context is clear. But the context isn’t usually made clear.
Here is an illustration of the potential problems with the metric. Let’s take an extreme example, as I think they are very useful in illustrating concepts. Let’s say that I have a million BTUs of biomass. But let’s say I have a conversion process that is terribly inefficient. I use that biomass in an inefficient process to produce a trifling amount of liquid fuel: 100 BTUs. In the process, 999,900 BTUs – 99.99% of what we started with – are lost in the process because they are used to drive the process.
But let’s say I have to input a small amount of fossil fuels; say in the form of electricity to run a pump. If I used 13 BTUs of fossil fuel to produce the 100 BTUs, then the energy return based on Wang’s metric is 100/13, or 7.7. So, I could claim to have a high energy return despite the fact that almost all of the available BTUs are wasted. This is the ‘opportunity cost‘ of those BTUs. Had we used the starting biomass to produce electricity, for instance, we would have had far more BTUs at the end of the process.
Now I am not for a moment suggesting that Coskata loses most of their BTUs in the process of making their ethanol. But without a real energy accounting – which the 7.7 number is not – it is difficult to determine whether this process makes better use of the available BTUs than a competing process. A proper energy accounting should take into account the overall BTUs consumed in the process, and not just the fossil fuel usage.
David Henson, President of Choren USA (another company involved in biomass gasification), once commented to me “You know, most people just don’t understand that biomass isn’t very energy dense.” David was absolutely correct, but what does that mean? The lower the energy density of a substance, the closer it needs to be to the factory. Imagine hauling potatoes from New York to California in order to convert them into ethanol, and you get the picture. You would certainly burn more fuel transporting the potatoes than you could make from processing them into ethanol.
I believe this issue of low biomass density, which I have referred to as the logistics problem of cellulosic ethanol, is a killer for cellulosic ethanol. In fact, I recently calculated that to keep a medium-sized cellulosic ethanol plant running would consume the biomass equivalent of almost 900,000 mature Douglas firs every single year.
However, the Coskata process is not a cellulosic ethanol process. I don’t consider any gasification process to be cellulosic (I explained why here). The consequence is that a gasification process can have a higher yield because it converts lignin and hemicellulose in addition to cellulose. In Coskata’s case, they promise 100 gallons (+) per ton. How much biomass then to run a 100 million gallon per year facility? A million tons per year. How much biomass is this? If we return to the Douglas fir example, it is the biomass equivalent of around 1.2 million mature Douglas firs per year.
That’s still hard to wrap my head around, but I can put that in context from my current job. In our wood acetylation plant in the Netherlands, our nameplate capacity is 30,000 cubic meters of wood per year. A cubic meter weighs half a metric ton, so we run 15,000 metric tons per year through our plant (about 17,000 short tons). Coskata proposes to process about 60 times as much biomass through their 100 million gallon per year facility. That is the sort of logistical challenge that boggles my mind, when I try to scale up our process by a factor of 60.
To put in the context of rail cars, the coal cars lined up outside of a coal-fired power plant are a familiar site. According to this, each car carries about 100 tons of coal. For a million tons of coal a year, you would have to have 1 million/(100 tons per car) = 10,000 cars per year coming into and leaving the plant. That’s more than a car an hour, 24 hours a day, 365 days a year. And of course coal is quite a bit denser than biomass, so more cars would be required in the case of biomass.
I won’t say that’s impossible, but it is going to be a significant challenge. All I can say is Coskata better have hired some very good logistical experts. They are going to need them.
So what’s the bottom line? Let’s say you are an investor with a billion dollars burning a hole in your pocket. You contact me and ask if Coskata is for real. I want to see your billion dollars invested wisely, so here is what I would tell you.
The plasma gasification piece and the membrane separation piece both need a very good technical vetting from someone who has signed a secrecy agreement and has access to the experimental data. Whether a technology works in the lab is one thing. After all, if I can kill cancer cells in the lab, have I cured cancer?
You need to know to what extent it works in conditions close to what Coskata is proposing. Has it been tested under these conditions? For how long? What were the results? What were the key challenges? How accurately were the energy inputs measured? In fact, I would probably want to park myself in their labs for a few days, and spend a lot of time talking to technicians. I would want to know – outside of the tours – what’s really going on.
Second, I would really focus in on the logistics issue. I would want some serious details on how they are proposing to handle the logistics. How is the biomass going to come into the plant? Has a calculation been done on how far away something can be transported before it becomes break even on the energy? If it is waste biomass already coming into a point source, then it isn’t as big an issue. But then I would ask if there is any location in the U.S. that is handling a million tons of waste biomass at a point source (which the gasification plant would be). I would want to see actual examples of someone handling this much biomass.
Finally, I would go over that $400 million capital estimate with a fine-toothed comb. I would ask for an example of any technology that has been piloted in the lab, and then had an accurate capital estimate done at a scale of tens of thousands of times larger than the lab scale. As I have said before, you have different problems at a pilot scale than you had at the lab scale, and the problems become even bigger at commercial scale. The capital estimate is already $400 million for a 100 million gallon per year plant – $61,000 per daily barrel. That puts it at a disadvantage to GTL or corn ethanol. Why wouldn’t I expect that capital estimate to climb as they gain piloting experience? Why would I expect them to stick with biomass, when the logistics of gasifying (partially oxidizing) natural gas are trivial when contrasted with biomass logistics?
At least that’s what I would do. But then again, I am notoriously frugal with money, and perpetually skeptical on top of that. If you are a gambler, then you may want to adopt a different strategy.
Note: As always, if you spot an error, let me know and I will gladly correct it.