Combustion Dynamics

Here at Dynamo we’ve taken a look at this problem of fuel flexibility, and built a power solution that is truly fuel agnostic.  While the product we are building requires a lot of engineering and years of experience (our technical advisory team spans 100+ years of combined turbine development experience), the solution itself has several key features that allow us to tackle this challenging technical problem.

The first thing we decided was to build a gas turbine engine, as they are renowned for their fuel flexibility.  In many ways a gas turbine is just a set of compressors and expanders set around a combustion tube.  As long as a combustion chamber can be made to reliably burn fuel, a turbine can be built around it.

We then developed a combustion chamber that can accommodate a wide range of BTU contents.  The challenge here was to ensure complete combustion and low pressure loss for a variety of fuel mixtures at both startup and steady-state operation.  The combustion chamber that we have developed has achieved all of this.

 

Although the combustion chamber is great, we do not rely on it 100% to ensure the reliability of our engine.  To that end we’ve included a specialized fuel conditioning system that is closely monitored by our supervisory control system.  The fuel conditioning system serves as a buffer between the wellhead and the combustion chamber, such that the fuel quality does not vary drastically over short periods of time and reduces the amount of work needed for the control system to regulate fuel flow.

Deploying our product in the oilfield adds additional complexity.  As discussed above, on the fuel supply side the consistency of the fuel can vary significantly over a few hours, and it is challenging to quantify that fuel a priori.  Additionally, on the demand side, pump jacks and other field equipment have a variety of duty cycles which change the amount of power required at any given moment.  To meet these needs, our solution has to be more than a combustion system.  It is tasked with the double duty of converting a variable energy of one type [fuel] while trying to meet varying output demands—all within very short time frames.  To enable smooth operations, this is achieved with several features, including a proprietary control system and a sophisticated custom power electronics package.

We can talk all day about how we do things, but our customers care about results.  In the lab to date we have verified the ability to operate on fuels ranging from 500-2045 BTU / scf in a single unit.  Across this range we were able to start the engine, bring it to power, and sustain operations as the fuel content was varied.  We were also able to do this with liquid water injected into the fuel lines—we were able to do this with a water cut of 80% by mass.  This effective range and the ability to handle liquids in the combustion system show that we can sustain combustion in virtually any oil field.   A more technical summary can be found in our whitepaper here.

With all this discussions on fuel flexibility, we would be remiss if we did not talk about what makes fuel flexibility difficult.  There are many factors that affect how something burns, such as the fuel being a liquid or a gas (or even a solid), the structure of the underlying hydrocarbons, the amount of oxygen present, and the geometry of the flame zone.  These are the big ones, but there are many other smaller factors which I will not have the chance to dig into here.

The state of a fuel has a lot to do with its combustibility.  Ultimately, for fuel to burn it must mix with oxygen (or some other oxidizer).  Gases mix very well and very evenly, which makes them easier to control in the combustion process.  Liquids on the other hand do not mix well, and often need to be premixed (such as with a carberator) or aerosoled into little droplets (like a high pressure fuel injector).  These processes take much more tuning to get right. Lastly, there are solids, which implicitly do not mix well.  Solids usually need to be pulverized into little bits, much like aerosoling, for them to be good combustion candidates.  More often than not, solid fuels are mixed with solid oxidizers in even proportions to enable more complete combustion.  This is most commonly seen in gunpowder or APCP found in solid rocket motors.

With fuel, you need oxygen to react with hydrocarbons to enable combustion.  However, it may not be intuitive that having some of each is not enough to enable combustion.  Even with a spark, there will be no flame—this is a phenomenon known as the flammability limits.  Generally combustion is most efficient when there is just sufficient oxygen to burn with the fuel; as the amount of fuel gets cut in half the mixture ceases to combust properly.  Likewise, if the amount of oxygen is cut to a third, the mixture will fail to combust.  To contend with this, many modern engines will have sensors to balance and meter out fuel to match the air in the system—however even with careful tuning, poorly mixed fuels may have spots that lie outside the flammability limits.

As you can imagine, combustion is a very complex process.  Energy is released as complex hydrocarbons are reduced to simple carbon dioxide & water.  The steps to get there can be rather complex. The chemical bonds between atoms are formed and reformed during the combustion process—with many intermediate molecules formed during the process.  The result is that combustion takes time, on the order of several milliseconds.  This isn’t a lot when you consider a car engine rev’s up to 7,000 RPM, each stroke (which must include compression, combustion, expansion, and evacuation of the gasses, can only take 8.5 ms.  The result for some systems that operate on these time scales is that combustion becomes hare to control, which can prevent it from coming to completion.  Generally, however, simpler hydrocarbons burn “faster” than more complex ones.

To make this even more complex, different hydrocarbons carry different amounts of energy, independent of volume or weight.  To try and describe this effect, engineers developed something called the Wobbe Index, that allows for relatively simple mathematical scaling of fuel injection rates for a given fuel—assuming that fuel is known ahead of time.   Unfortunately, this makes simple fuel metering systems, like a carburetor, a poor solution when the fuel is unknown.   The ability to handle different fuels requires a more advanced fuel delivery system that is capable of providing fuel at different rates.

For reciprocating engines, an important factor to consider is the “knock” rating of the fuel.  It is essentially a measure of when a fuel will self ignite (assuming it is heated from the compression stroke of an engine).  Reciprocating engines become more efficient and more powerful if they have more compression, however the fuel limits the amount of compression that can be practically achieved.  To further complicate matters, the flammability limits of fuels change as they are compressed.  The result is that adding fuel flexibility often comes at the cost of engine performance and emissions.  As a real world example of this complication, many diesel manufacturers are trying to build engines that can run on both natural gas and diesel.   As it turns out natural gas has a much higher anti-knock index rating than diesel—due to this quirk of nature, manufacturers have developed bi-fuel generators.  In many of these solutions the products have to run on a 50/50 diesel & natural gas mixture, such thatthe diesel is compressed to ignition, burns, and in turn burns that natural gas in the combustion chamber.  While this is a good technique for reducing diesel dependency, it is not true fuel flexibility.

The last thing that really drives combustion is the physical location where combustion takes place.  The geometry, or shape can drive the local mixing of fuel and air; it also provides the combustion constituents with the time and space to burn to completion.  The combustion chamber and its aerodynamic interactions with the rest of the engine define a lot of how a combustion process performs.

Last time we looked at the major contributors to the cost of fuel—which is generally tied very closely to the marginal cost of producing a kWh.  If we take a look at the major factors playing into the cost of energy, we can pretty easily determine that the fuel cost per kWh is a pretty simple function:

Pretty simple; but both costs have complex components that can cause them to range widely.  Let’s take a look at the cost of fuel.

I think the above is pretty explanatory—the commodity price of a fuel is what you pay for it at your local gas station.  Getting your fuel to site obviously introduces a level of cost.  Your delivery company will charge you more if they have to go significantly off the beaten path, or if they have to use special equipment to get to you.  Similarly, if your fuel isn’t just plain old gasoline, but requires special fuel tanks (CNG) or handling expertise because it may be hazardous (methanol), there is an additional cost associated with that as well. The factor ε on the end is a coefficient to capture things like taxes, discounts, and other miscellaneous costs that should be taken into account.

Lastly, and most importantly, we divide the cost of fuel by an “Availability” factor.  Availability approaches zero as a fuel gets scarcer; it gets bigger than 1 as it becomes bountiful.  A value of “1” is a “nominal” value when compared to the other parts of the equation.

In some respects, how this factor is realized is routine.  If all the gasoline facilities are off line in your region, there is no amount of money you can spend to buy a drop of gas—such as what happened on the east coast during Hurricane Sandy.  Conversely, if there is too much fuel, it either is wasted, burned off, or you have to pay to have it shipped or stored elsewhere—and the cost of the fuel will plummet.

Looking at this another way, you can consider how the supply and demand for a fuel will influence its price—they call this the elasticity of demand.  When gasoline gets more expensive, by 10% for example, demand decreased by 2.6%.  Demand can also effect supply; for example, a 10% increase in demand may increase prices by 38%.  What does this mean for us as modelers of future fuel prices?  It means that we can use the availability factor to analyze the risk associated with fuel prices (knowing how things can change) and try to take it into account when comparing different fuel sources.

Efficiency modelling is a whole different ball game, and very much changes from generator to generator. Generally, each generator has an operating point where it is most efficient, but generally the loads the power do not always allow them to operate at this point.  The resultant duty cycle can significantly influence the efficiency of the underlying system.

Without going into too much detail into efficiency modelling, I’m going to jump into a couple of different comparisons for various systems using the information we have here.   I’ve selected three technologies routinely used in the oil and gas industry for use in different types of power applications: small Diesel Generators, Fuel Cells, and Thermal Electric Generators (TEGs).  Each are used in different applications and also have different capital costs, (the impact on LCOE of which is not thoroughly analyzed here).  The following chart compares the efficiency of each generator, the relative cost of the fuel on a per kWh basis as delivered, as well as the marginal dollar cost to generate a kWh by each generator Tables for the inputs for this chart is attached at the end.

The diesel generator comes out where you would expect it to, with fuel costs at close to 40 cents per kWh.  The supply chain is relatively simple, with non-road diesel coming in at roughly $4 per gallon, and shipping contracts only adding a marginal cost.  Efficiency can range from the mid-twenties to the low thirties; I picked 30% efficiency since most of these generators run well below their prime rating in the field.

Fuel cells are often chosen for small power applications (<1kW), where reliability is essential, and where the cost of fuel is secondary to the cost of maintenance and the cost of downtime.  While many fuel cells are designed to run very efficiently on propane or natural gas, the fuel reformer found in most of these fuel cells require the fuel to be highly refined, above standards found in more traditional applications.  In this case, we built the model around a European fuel cell that is finding acceptance in the US O&G market. The fuel cell operates on highly refined methanol, which can only be provided by the manufacturer in Europe.  (Having a single source supplier imposes its own supply risks—that availability factor I described earlier, which I did not include here).  The result is a very high cost of energy, however for small applications the cost of fuel is dwarfed by the value of reduced maintenance and downtime.

TEGs are also commonly found in the O&G industry, for use powering very small loads <100W; again they are used where reliability is key, although they are very large and suffer from being very expensive.  As opposed to fuel cells, TEGs have very low efficiency’s (3-5%) but they also have the distinction of being able to run well very poor quality fuel (basically any heat source will do).  In many cases, TEGs operate on pre-pipeline quality natural gas, often found in upstream applications.  In this case the source fuel is plentiful, and cheap—often its face value is below the cost of commoditized natural gas, as it has yet to be transported to market; and in many cases, the operator doesn’t have to pay the lease holder the cost of using the fuel, which equates to an additional 10% discount on the fuel.

As this example illustrates, the cost of operating the very inefficient thermal electric generator is 1/10 the cost of operating a fuel cell, and ½ the cost of operating a diesel generator .  Unfortunately, thermal electric generators do not scale up well in size and value much above the 100W mark.

There are three primary considerations when considering a prime mover which ultimately drive the Levelized Cost of Energy:  capital cost, maintenance, and fuel expenditures.  The first two are topics of discussion for a later time, but the last item, fuel expense, is our area of focus today.

There are several significant components that play into fuel expenses (mapped out below).  Fundamentally the fuel expense is the product of the realized cost of fuel to the user and the realized efficiency of the underlying generator.

Cost Map

A logical first step in reducing the cost of energy would be building a more efficient generator, such as a fuel cell, or a complex cycle engine.   However, Fuel efficiency is pretty much set by the state of the art in technology, and will change only incrementally with time.  There are some minor enhancements that can be made with energy storage to keep the generators operating at their peak operating point, but making systems more efficient is very expensive and the development effort takes a long time.

A cost effective alternative to driving reduced costs is to have the engine operate on the lowest cost fuel of the moment.  To achieve this, a variety of engines have been developed to work on specific singular fuels.  However, in the modern environment, there is pressure on engine manufacturers to have engines that can operate on a variety of fuels.  The physical phenomena that drive combustion make this a technical challenge, but there are a variety of solutions that exist on the market to meet the challenge of reducing costs—each has their strengths and their weaknesses.

Most shifts in fuel sources are driven by an under-supply or over supply of different types of fuel.  In the past, for example, the price of natural gas would often track the price of oil for an equivalent amount of energy.  However, oil fracking and the US natural gas export restrictions have caused a spread of roughly 6x in the cost of energy between oil and natural gas.  Needless to say, many companies (Dynamo included) are looking to leverage the lower cost fuels in a reliable manner.

Next time we will look at a few corner scenarios that illustrate how various fuel expense conditions can drive the cost of energy.