Category Archives: Uncategorized

Precipitation in Yosemite Valley, 1894-2013

California's severe, state-wide drought owes a significant chunk of its origins to reduced snowfall in the Sierra Nevadas. Much of the state is semi-arid, and over the last century an extensive series of dams, canals, and pipes was put into place to service the water needs of both agriculture and urban centers. The Carbon Dioxide Information Analysis Center maintains and hosts a great database of temperature and precipitation data from stations across the United States. Many of these stations have records which go back over a hundred years, and it's a phenomenal resource to quickly eyeball trends or confirm/deny a Saturday afternoon's idle theorizing.

First, here's an 18-month rolling average of precipitation from 1894-2013.

yosemite rolling average precip


From glancing at that, you might notice that the right-side tail looks a bit odd, let's zoom in on the records since 1980 and add a linear trendline (auto-calculated by Excel).

yosemite rolling average precip 30 years


This provides a more detailed picture of the reduced recent snowpack, but it's important to remember that this is only a single location and barely scratches the surface of the complete story.

Venture Capital in the United States (1998-2012)

The National Science Foundation (NSF) publishes a series of science and engineering indicators that includes numbers on VC dollars and number of deals by state. I've pulled together some graphs from this data. Not a lot to explain, except that on the NSF site the $ is listed as current dollars, and it was published in 2014.

The two easiest trends to see are the dot-com bubble in the late 90s-early aughts, and the great recession in 2009. Compared to many sectors of the economy, it seems VC disbursements recovered to previous levels with surprising alacrity.

top ten states by average deal $ with at least 200 deals top ten states by average deal $ top ten states by number of deals 1998-2012 top ten states by number of deals ca removed 1998-2012 top ten states by total CA removed VC 1998-2012 top ten states by total VC 1998-2012 VC per capita CA gone VC per capita

A Hidden Cost at the Pump - Iranian Sanctions

Taxes and Subsidies

Energy markets are prone to an obscure, confusing, and even contradictory web of subsidies and taxes. Often, the extent of this web depends on how one defines a subsidy or tax, and how far you are willing to dig. Certain energy installations, like wind, are extremely dependent on subsidies. Wind installations drop to almost nothing every time the production tax credit lapses. Utility-scale nuclear benefits from federal disaster insurance guarantees. It is probable that utilities could not be persuaded to build nuclear without this safety net in the face of catastrophe.

To oversimplify, a tax is a charge on an individual or corporation (legal entity) that is paid to the government, while a subsidy moves dollars in the other direction. This is of course not entirely accurate, and ignores instruments such as fees. Essentially, taxes allow the government to collect revenue to pay for goods and services that would not be covered by a fee-based system (like passports or the DMV). Roughly, subsidies can be indirect or direct. An example of a direct subsidy is the production tax credit for wind, while the nuclear insurance guarantees are more of an indirect subsidy, in that the program makes nuclear development more palatable as a whole, without a direct outflow of cash.

Background aside, my goal here is to identify a hidden energy tax* on American drivers. In this case, I had the notion that US-imposed sanctions on Iran may be acting as a subsidy for oil-producing nations, and a tax on American drivers. This is possible due to the global nature of oil markets and the fungible nature of crude oil itself.

Current Events

In case you missed it, there was an announcement of a deal with Iran to formally make a deal in 6 months. While the details of this current arrangement seem somewhat amorphous, with John Kerry and Iranian officials issuing some conflicting statements,what I'm interested in was the $3 price drop in Brent crude following the announcement of the deal. Bloomberg quoted analysts who referred to this price drop as a " 'knee-jerk' " reaction. While the price did recover by the end of the day, it is reasonable to view that $3 drop as something of a price premium that futures traders are allocating to increased Iranian production, and re-involvement with OPEC.

Despite the price stabilization, there is still a long road ahead. If all goes well on the deal-making front, Iran could be involved in high-level OPEC discussion by next summer, and is expected to push back on Iraqi exports while ramping up its own. However, Iran faces a series of technical challenges in restoring capacity and re-opening old production. Additionally, Iranian wells have been producing for longer than many Iraqi fields, and it is possible that they will never reach 3 million barrels a day, but remain between 1-2 million. With that background out of the way, let's move on to my original idea- that US government sanctions are costing American drivers at the pump.


Right away I had to see if there was a clear-ish relationship between the price of crude and the price of gas at the pump. Brent crude prices are the standard which other hub crudes are priced against, so I grabbed the EIA's weekly data on Brent spot prices as well as the data on average gasoline prices. The history on gasoline prices averaged across all blends doesn't start until April, 1993 so I cut off the earlier Brent prices to match this 20.5 year span. I chose to use the averaged prices because I'm looking for broad trends, nothing fancy or subtle. A double y-axis plot would show if there was any hint of a relationship, so I started there.

Immediately apparent is that the price of Brent Crude and the price of gasoline track extraordinarily well against each other. From this point I tried to find $3 price fluctuations to match the price premium that could be attached to free-flowing (free as it gets with OPEC) Iranian oil. I can't look directly at the most recent change for two reasons.

  1. The last week's data isn't even out yet (it releases tomorrow I think).
  2. This price change occurred in a single day, and as traders came to understand the deal the price returned to nearly the starting value.

In order to generate more data to work with I opened it up to a 50 cent range.  I found 38 weeks using change ranges from +2.75 to +3.25. Before I continue, I should come clean: This is a pretty terrible way to carry this analysis out, but I'm 2 weeks from graduation and should probably not spend days developing a more comprehensive model. There is no paucity of nuanced modeling of oil markets and the effects of regulations and treaties. This is more of a thought experiment on my part, and a rough one at that.

So with these 21 Brent price changes in hand it's time to take a look at gasoline prices in the weeks around each change. Here it's important to note that the Brent and gasoline weeks are slightly offset. Each Brent week starts 3 days earlier than the corresponding gasoline week, i.e., the gasoline data begins April 5th and the Brent data starts April 2nd. This 3-day gap isn't necessarily a bad thing, because it's easy to take the gasoline prices from the 1st and 2nd weeks relative to the single Brent week. Gas prices should lag at least slightly behind crude changes, so the second week is more likely to capture change in gasoline prices as a result of changes in Brent prices. To meet my expectations the 1st week can have essentially any change, and as the price signals reach gasoline the 2nd week should display increased price. Those last few sentences are a bit confusing, so here are the results in graphical form.

Remember, neither of these charts is a time series. Rather, they're displaying the individual averages of the 21 1st and 2nd weeks of gasoline prices that overlap the 21 weeks of Brent price changes that fit my range of values.The 1st week roughly matches my expectations, but what's really interesting is the 1st in comparison to the 2nd and the 2nd itself. The set of 1st week overlaps only sees a handful of price increases, but these increases match the few decreases in the set of 2nd week values. The set of 2nd week values roughly conforms to my expectations, with price

The mean price change in the first week is -0.0338.

The mean price change in the second week is +0.0637.

The 2nd week average is the value of note here. Based on my assumptions (of which there are many) and these calculations (which are admittedly rough), a +~$3 swing in Brent crude prices averages out to a $0.0637 increase in gasoline prices per gallon. According to the EIA,

In 2011, the United States consumed about 134 billion gallons of gasoline.

(134,000,000,000 gallons/year)x($0.0637/gallon) =  $8,535,800,000/year

Granted, this price is mostly borne by drivers. Those who don't own cars and bike, use public transit, or just walk (my preferred method of locomotion) will avoid most of this cost. Most goods are transported via diesel, and public transit is normally powered by natural gas, diesel, or electricity. With those assumptions, how much does each of those drivers pay?

Going back to the wellspring of EIA data (table 58) we find that there are 120.77 million gasoline-powered internal combustion engine vehicles in the United States. These vehicles account for the vast majority of gasoline use so for this (again, very simple!) analysis we'll say they're the total consumers.

($8,535,800,000/year)/(120,770,000 vehicles) = $70.678 per vehicle per year

Of course, not all ICE vehicles are created equal, and the actual amount of this Iranian Sanctions tax charged depends on the amount of gas someone purchases each year. Another way to look at this is through the best-selling cars and their mpg ratings.

This data is from (sales) and (mpg). In some cases I attempted to roughly average the mpg of vehicles reported as a series (eg. the F-Series). For all starting mpg values I used the combined rating of the 2013 models, although I know 2012 sales does not necessarily mean 2013 models. Finally, I used Department of Transportation numbers for the average annual mileage of American drivers (13,476 miles) to create an index of tax for these cars and trucks, and find the annual cost to that group of sales.

and for easy-to-view values here are bar charts of the last two columns.

I find it intriguing that despite the differences in mileage, the top-selling cars in America are close enough that it doesn't make a huge difference in the annual price to the driver, with the range only slightly exceeding $20, the cost of a somewhat cheap meal out for two.

Here, the trucks stand out because of their poor mpg. The Dodge Ram series may sell less than the Camry, but a Ram's mileage is sufficiently worse to ensure a higher total Iranian Sanctions tax burden. Trucks paying a higher at equal mileage levels is interesting in light of survey results from Strategic Vision which found that Republicans were most likely to own trucks, and recent polling has found more Republicans in favor of continuing sanctions than other political affiliations.

While I haven't proved anything conclusively, I think there is value in demonstrating the potential for hidden taxes in energy. Particularly when it comes to a charged foreign affairs issue, and the rapid way in which prices can react and change to news. It might be the case that there is no longer a price premium for Iranian oil sanctions, but the potential effects are still worth investigating. There are a lot of fun things to do with the price of oil and oil derivatives due to both its ubiquity and international nature. Hopefully, I can start integrating more current events into these posts.

By the way, the $8,535,800,000/year  number I came to is not just a tax on American drivers, but also a subsidy to oil producers (which the US is as well), in that if it does exist it keeps the price (and profits) artificially higher.

It was fun to take a step away from the EIA-860 form data, but I'm still going through energy sources there. Next on my plate is geothermal.

* Yes, this isn't exactly a typical tax. I was thinking in terms of national oil producers. This cost premium keeps prices higher for them, and so in a way they are extracting a tax on gasoline consumption. Obviously a large chunk of this is captured by corporations, but this was my attempt at being evocative. There are also cases where specific taxes go to pay for specific programs. I think this could be viewed as a similar case, the money's just skipping a step.


UPDATE: Changed Tax in title to Cost to be more accurate.

The State of our Solar Generation

After the last post on coal generation, I thought it would be nice to cut a few steps out of the photon processing, avoiding all the dirty gunk that gets picked up along the way, and talk about straight-up sunlight. You probably know by now (also the title's a total spoiler), but this post is going to be about solar generation.


So, I think a small discussion of technologies is useful to get some background before diving into the delicious numbers. First, 4 prime movers are represented in the data: PV, ST, OT, CP.

  • PV - You were probably able to guess this one, it's photovoltaic. Photovoltaic installations are the typical depiction of solar panels on roofs. Slabs of silicon that directly convert photons into electricity. This property that is unique to this form of generation, not just amongst solar technology, but all of our generation techniques for any fuel source. What I find particularly amazing is that the photovoltaic effect was first noted in 1839. It's fun to think about an alternate history in which we were able to produce workable solar panels in the 1800s.
  • ST - Steam turbines. Steam turbines + solar can also be referred to as solar steam. This is probably the other technology you've heard of, at least in passing. The general idea is to concentrate heat using reflective plates. This heat is then used to drive a traditional steam turbine.
  • OT - Other. There's only one of these, and it's not listed explicitly in the data, but thankfully the operating utility and year is, and a little googling turned up this article. Based on that piece, I believe this 1 MW installation to be a concentrated solar power demonstration plant utilizing parabolic troughs.
  • CP - Energy storage, concentrated solar power. Like OT, there's only one of these in the EIA's data. This generator is one of the Solar Energy Generating Systems (SEGS) in California. Specifically, I believe it is SEGS I. Fun fact, a nearby solar installation inspired a side-mission in the video game Fallout: New Vegas.

With a little bit of background, let's get into the number. As always, I'm rocking the data off of that October-fresh EIA-860 form. A lot of the data in this post would have been wildly different a year ago, so it's really nice to have the updated numbers. Let's start with a basic breakdown of the generation types I just discussed.

Now, how about a breakdown of the PV and ST installations. If you're curious, the capacity of the CP generator is 13.8 WM, and the OT generator is 1 MW.

And finally, the same breakdown, but for all the solar installations together.

If PV's dominance wasn't terribly obvious from the first table in this post, it should be now. The sheer number of PV installations compared to ST and the others allows it to define all of the operation date categories, and maintain a statistical stranglehold on the capacity values as well. Before moving away from the basic information I'd like to throw 2 maps at you to get an idea of the spatial distribution in solar generation.

A glance at some basic histograms illustrates the capacity values and installation ages above. 

Unsurprisingly, most solar generators are under 5MW, with the bulk of those actually at 1MW or less.

Mirroring the capacity graph, the number of annual installations show a sharp rise that began less than 10 years ago. As I mentioned at the beginning of this post, the 2012 numbers are dramatic, and reaffirm the trend of solar as a growing energy source in terms of installations. However, due to solar's capacity factor, conversion losses, and lack of storage available 1000 1MW generators could produce less useful energy than a 600MW coal power plant. In its current state, solar isn't all the great for peak power, so it generally is going to be contributing to baseload. But why would what appears to be 1000MW of solar generation produce less energy than a 600MW nameplate coal power plant?

I mentioned a few factors above, but the major one with solar is capacity factor. I've mentioned it several times before, but it still might not be clear exactly what a cpacity factor is. Lots of people get confused by it (as I recently found on reddit), but it's a simple, albeit very useful, metric. Essentially, capacity factor is the actual energy production over the theoretical energy production.

Let's take our theoretical 1000MW of solar, and 600MW of coal. Typical capacity factors for coal and solar are around 80% and 20% respectively.

8760 = Number of Hours in a Year

Coal: 600MW×8760 = 5256000MWh

Solar: 1000MW×8760 = 8760000MWh

20% Capacity Factor means you only operate for 20% of the theoretical max time

80% Capacity Factor means you only operate for 80% of the theoretical max time

Solar Actual Production →  8760000MWh×0.2 = 1752000MWh

Coal Actual Production →  5256000MWh×0.8 = 4204800MWh

4,204,800MWh > 1,752,000MWh

The idea of a capacity factor is a nice as an easy metric for industry folks to make quick assessments with. It's important to note that some generation technologies have low capacity factors by design because they just aren't used all that often. Natural gas peaking turbines occupy this niche, and I'll discuss that further when I get to the natural gas post. Coal and solar are not a part of that same niche. Both of these technologies either are already run as close to 24/7 as possible (coal), or that is the eventual goal (solar, some sort of satellite-based microwave solar could achieve a pretty high capacity factor[hell, with the cost of one of those systems, it better]).

So solar's capacity factor isn't great, but there have been a huge number of installations in the past few years, which should add up. Right?

Good news, the recent uptick in installations wasn't all itty bitty generators, but enough large ones to blow most other years' capacity installed out of the water. I also plotted capacity in terms of its rate of change from year-to-year and against the average. But this produced really absurd graphs with minimal usefulness. Which makes sense given how extreme the changes are in the past several years

At this point it might be interesting to look at maps of just the 2012 installations. Perhaps California is installing loads of capacity in a handful of installations, while states further back on the solar uptake curve are experienced spurts of growth in smaller generators?


So, overall, the answer is "not really." California's pace of installation and the relative capacity of those outstrips almost every state by a huge degree. However, there are two other visualizations that may present a different picture. One has to do with Renewable Portfolio Standards (RPSs), which I will save for a future post.  The other has to do with population.

California's a big state. It has 12 million more people than the 2nd most populous state, Texas. Only 6 states even have 12 million people altogether. So, what if I rolled back to the original 2 maps, but did everything in terms of a state's population?


So California gets dethroned on installations per 100k people, but the more dramatic drop is in its capacity, which isn't surprising given its population. Here are the top 10 states, sorted by capacity, then installations. This list might have been surprising 6 maps ago, but I bet you could guess most of them right now. Also, because I can't help myself, I've labeled each state with how they voted in the last election, and the political affiliation of their governor. I believe that a state's standards on renewable energy are a product of both their government and the people.


Immediately, I want to say that I feel bad about Hawaii, because I've kept it and Alaska off of all the maps I've made so far. The reason is partially for ease of aesthetics, but also because each of those states is highly anomalous compared to the contiguous states. I'll try to mix them in sometime, maybe they'll each get their own post (Alaska and Hawaii both have completely different, largely uncharted, and totally intriguing resource bases), but for now they'll just occasionally pop up in tables. Hawaii aside, there are a few things I find very interesting about this. Foremost is the myth that solar only works in the Southwest. Realistically, solar can work anywhere, it just depends on what kind of installations are being considered, and what output you expect. The Southwest is almost certainly the best bet for huge multi-gigawatt plants, but New jersey and Vermont, neither of which is notoriously sunny, make it clear that with the will to do so solar installations can happen. As for the political spectrum, I was largely thinking ahead when I added those, as I'll be doing some kind of political post in the future. For now though, I think it's interesting that the state's general populations seem to be more Democratic than the average, while the governorship is more split between Democrats and Republicans.

This post was a bit light on text, as my brain seems to be running out of fuel as it near 10pm on a Saturday, but hopefully the cavalcade of figures was enough to keep at least a few people amused. As always, if you have any questions about methods, data sources, or simply want to rant at me, hit up the comments.


The State of our Coal Generation

Last time, I wrote about our energy storage generation: why it's important, what it consisted of historically, and what it looks like today. This post is going to focus on coal. Again, I'm using the data from form EIA-860, with the finalized 2012 numbers released this month.


First up, what coal energy sources are there?

Out of these 42 "Energy Sources", 4 are coal ranks, ordered here in decreasing rank:

  • Anthracite
  • Bituminous Coal
  • Sub-bituminous Coal
  • Lignite

and two are coal that has been processed in some way:

  • Coal-Derived Synthetic Gas
  • Refined Coal

Anthracite, Bituminous, Sub-bituminous, and Lignite refer to the coal's rank. Coal rank is essentially the degree to which something has been turned into coal.  As the heat and pressure processes work on the original organic components of coal, moisture and other impurities are forced out. This process increases the heating content of the coal, and generally increases its value. High rank coal has greater energy/carbon content, and lower moisture/volatiles content.  Here's a handy chart:

Anthracite is the highest rank of coal. It has a very high energy content, low moisture, and relatively few volatiles. Anthracite commands a higher price than lower rank coals, and was generally used in applications requiring consistent high-quality heat, such as steel making. Anthracite's combination of usefulness and rarity means that it is often rapidly exploited as a country industrializes. In the United States, only 0.846% of our demonstrated reserve base for coal is anthracite, and nearly all of that is found east of the Mississippi.

Bituminous and Sub-bituminous are the main ranks of coal used for power generation for two major reasons. The first is that they are the most common, comprising roughly ~89% of our demonstrated reserve base together, split 53% bituminous and 36% sub-bituminous. Second, they're considerably cleaner than lignite. In the grand scheme of things, lignite is practically dried-yesterday peat, direct from your local swamp. Distribution of bituminous and sub-bituminous coal has a distinct geographic split in the United States, which I threw a gif together to illustrate.

It might surprise some of you to see how much coal Illinois has. Despite the state's (my home state actually) massive reserves, there is very little demand for Illinois coal. It's worth taking a brief aside to explain this; I see a few major reasons for the absence of Illinois coal flooding the market (hint: it's basically just not [economically] worth it):

  1. The coal is generally deep and somewhat tricky to extract,particularly compared to Wyoming's surface bonanza.
  2. The land it sits under is highly valued for agriculture.
  3. Coal is already available nearby  in less environmentally-stringent states (Indiana and Kentucky).
  4. The coal industry isn't doing so hot right now, so there's no reason to develop the area.

This, of course, could change. In fact, it might have had shale gas proved a bust and coal-to-liquid and coal-to-gas technologies, paired with cheap and effective carbon capture entered the market strongly. Instead, pilot plants are gasping for air, surviving on a lifeline of rate hikes and federal subsidies.

There is one other important physical characteristic about American coal to keep in mind, western sub-bit coal has a bit lower energy content than bituminous eastern coal, but eastern coal is higher in sulfur. This imbalance in attributes leads to an interesting dance as eastern utilities attempt to maximize their dollar by constantly comparing the cleaner, but more expensive to transport western coal, with the dirtier, but closer to home and higher carbon eastern coal.

So, with those basic coal-in-america facts out of the way, what does our roster of coal generation look like in the United States?

Based on last week's release of the finalized EIA-860 for 2012, there are 1,290 generators whose first energy source is a form of coal. Here's a basic breakdown table of some pertinent facts.

While there have been some refined coal pilots plants in recent years, they didn't show up on the EIA's reporting. Per the discussion of coal rank earlier, it makes sense that there are no anthracite plants, and syngas has a single operating plant, because the ones that are supposed to be online had plenty of technical difficulties, then got knocked on their ass as the price of natural gas plummeted (I'll definitely be getting to natural gas soon, that shift was nuts). For capacity, I'm using the same methods I outlined in my last post.

From here on out, this post will be a bit graph heavy, hope you enjoy!

Again, I'm only working with currently operating coal plants, but what might surprise you is just how old some of these suckers are. While I brought up the median age of construction above, this plot makes it clear just how heavily we invested in coal in the mid-century, and what a dramatic decline has occurred since then. Possibly the most intriguing piece of this figure is the uptick in construction over the last 10 years. I would posit that these plants were planned well before the natural gas boom, and were actually constructed under decreasingly favorable economic conditions.   I left the syngas plant off of this figure, because it's only a single point.

This graph fills in a few of the details left out of the prior plot. While the number of annual generator installations decreased dramatically from the 1950's, this figure shows that the new installations were far larger, and implies that the backbone of our current baseload generation began operation 20-30 years later. Another takeaway comes from the twin peaks: bituminous and sub-bituminous. On the installations plot, it appears that bituminous coal handily beats sub-bit in generators, but what it fails to show is the average capacities of each generator, which I calculated after seeing this myself.

These numbers really surprised me when I first saw them (don't forget, that final column is in GW, so 1000x the first two). Honestly, I find the lignite numbers a bit appalling. While it doesn't make up a ton of the final portion (only 5.35%) I would endeavor to get that number even lower. I expressed some not-so-veiled disdain for lignite above while talking about coal rank for good reason. As a reminder, lignite is a brownish coal, with high moisture and ash content. Lignite also requires more processing than higher rank coals in order to be transported safely, so it is generally used close to where it's found. I've tabled the lignite generators, but perhaps you can guess where they are already (hint: check out the map above).

Unsurprisingly, it's the states that have lignite that are using it. It makes sense to use your in-house energy sources for generation politically as you can position yourself as a job creator. In fact, due to the difficulties in cleaning and using lignite in compliance with clean air standards you're going to be hiring more people per BTU of generation throughout your in-state lignite facilities, and spending more to clean each pound of coal that's also (on average) inferior in terms of heat content. So, lignite might be politically expedient, but it's probably economically inefficient. That inefficiency gets worse if you're thinking about environmental impacts, both at the local level (NAAQS), and globally (CO2). However, all of these negative side effects explain the relatively large size of lignite generators. It makes the most  sense to consolidate your inefficiencies to a small number of installations where they can be more easily controlled.

I think I'm going to end it here for now. I hope this sheds some light on coal generation in the United States, and made you think a little bit. Next time out I hope to talk about one of the non-regulatory reasons why coal is down, and maybe out (out being a relative term, it's not going to disappear or take a far smaller role in our day-to-day mix anytime soon, just taper off as plants retire): Natural Gas.

I'll leave you with another map, and one to keep in mind moving forward.

* Weighted average was calculated by multiplying each # of generators by the median year for that fuel input, then dividing by the total number of generators. This was done to reduce the influence of the 1 coal-to-syngas plant construction. If an average of the median years is taken directly, the average year is 1978.