Tag Archives: graphs

The Green March Madness Tournament: Setting the 2018 NCAA Tournament Field According to Sustainability

With the Super Bowl in the rearview window and the calendar about to turn to March, the attention of the sports world is about to be completely focused on college basketball and the annual NCAA Basketball Tournament (March Madness). Every year, this 68-team tournament captures the attention of people across the country, whether they are diehard fans or non-sports fans who  are simply participating in the office pool.

Not only does the NCAA Basketball Tournament serve as fodder around the water cooler, with billions of dollars of productivity lost in the American workplace every year, not only in watching the games but also in the various (sometimes unconventional) methods people use to pick the winners in their bracket. You may have seen people choose winners based on which team’s mascot would win in a fight, by choosing the schools with the superior academics, or even by choosing winners based on who has the most attractive head coach (shout out to my alma mater, University of Virginia, that AOL astutely points out would win in this last scenario).

Source

So with the Selection Committee currently watching the last few games of the regular season as teams try to bolster their chances of making the NCAA Basketball Tournament, I thought I’d take a look at how March Madness would look if the field was selected based on each school’s efforts towards sustainability, energy efficiency, and environmentalism– call it the 2018 Green March Madness Tournament!

This article will take all eligible NCAA schools and create the field of 68 for a tournament, but playing it out won’t be all that interesting because the top seeds will obviously ‘win’ each match up until the Final Four. So keep reading to see the 68 teams that make the tournament and find out which top seed comes out on top– but stay tuned once the NCAA puts out the actual bracket for the NCAA Basketball Tournament because I’ll do a follow-up article and revisit this concept to see who would win each of those real-life matchups based on who rated higher on sustainability!



Metrics used

After extensive research, I found three different measurements and rankings that look at the efforts of colleges and universities across the United States to incorporate sustainable practices, energy-saving measures, and environmentally-friendly practices. The latest version of the data for these measures, which are explained in detail below, were pulled to serve as the metrics of who would participate in the 2018 Green March Madness Tournament.

The Sustainability Tracking, Assessment & Rating System

The Association for the Advancement of Sustainability in Higher Education (AASHE) uses its Sustainability Tracking, Assessment & Rating System (STARS) to measure how successfully institutions have been performing in sustainability matters. The mission statement of STARS details how it “is intended to engage and recognize the full spectrum of colleges and universities- from community colleges to research universities- and encompasses long-term sustainability goals for already high-achieving institutions as well as entry points of recognition for institutions that are taking first steps towards sustainability.”

Source

STARS is completely voluntary, transparent, and based on self reporting. Dozens of different metrics are included in the STARS measurements, including in the categories of curriculum (e.g., whether the institution offers sustainability-focused degree programs), campus engagement (e.g., whether sustainability-related outreach campaigns are held on campus), energy use (e.g., availability of clean and renewable energy sources on campus), transportation (e.g., inclusion of alternative fuel or hybrid electric vehicles in the institution’s fleet), and many more that are found in the credit checklist.

Based on performance based on these metrics, each school can earn up to 100 points and a corresponding rating of STARS Reporter, STARS Bronze, STARS Silver, STARS Gold, or STARS Platinum. Because STARS is self-reported, institutions can continually make improvements and resubmit for a higher score. However for the sake of this Green March Madness Tournament, the latest scores for all schools playing Division I NCAA basketball were pulled as of the beginning of February 2018, with any schools not participating in the STARS program receiving a score of zero.

The Cool Schools Ranking

The Sierra Club publishes an annual ranking called the Cool Schools Ranking to measure which schools are doing the most towards the Sierra Club’s broader sustainability priorities. The data for the Cool Schools Ranking largely comes from the STARS submissions as well, though with some key changes— the Sierra Club identifies the 62 questions of the STARS survey that they consider the most crucial to their definition of sustainability and put that data in a custom-built formula, they only use information submitted or updated to STARS within the past year, and they asked institutions to also detail what moves they have made to divest their endowment from fossil fuel companies (a question not asked by STARS).

Source

As with STARS, participation in the Sierra Club’s rankings is completely voluntary and transparent, ultimately resulting in a numeric value on the 1000-point scale to use for the rankings.  For the scoring towards the Green March Madness Tournament, all eligible teams had their Cool Schools Ranking score pulled and divided by 10 (so it would be on a 100-point scale like the STARS rating), while schools that were not included in the ranking were given a score of zero.

SaveOnEnergy Green Score

The last of the three rating systems used for the Green March Madness Tournament is the 2017 Green Score given by SaveOnEnergy.com. The goal of this scoring system is to give credit to institutions making “noteworthy progress in eco-friendliness and sustainability.” The SaveOnEnergy Green Score takes the top 100 schools in the U.S. News & World Report and awards them scores based on their Princeton Review Green Score, as well as state data on farmers markets, local public transportation options and walkability scores, density of parks in the area of the school, state data on clean and renewable energy options, and availability of green jobs.

Source

The data for the SaveOnEnergy Green Score is a mix of voluntary data (e.g., data submitted to the Princeton Review Green Score) and mandatory statistics (e.g., state data on energy options and green jobs). In the end, SaveOnEnergy takes all of these factors to create a final score out of 100– though the score is only published for the top 25 schools, and the remaining schools are ranked without their score displayed. To account for this, a best-fit equation was used to correlate ranking with the score of the top 25 schools and extrapolated that equation to determine a score for the remaining ranked schools. Schools that did not make the SaveOnEnergy Green Score list were given a score of zero.

Final Green March Madness Tournament score

In the end, all 351 schools that participate in Division I basketball (representing 32 different athletic conferences) were given a final score that was the average of the STARS score, the Cool Schools Ranking score divided by 10, and the SaveOnEnergy Green Score, so that the final score is also on a 100-point scale (the final scores for all schools can be found in this article’s accompanying Google Spreadsheet).

Before moving forward, let’s make clear that this ranking system is mostly just for an overview of sustainability scores among schools based on publicly available data, and it should by no means be considered comprehensive. Indeed, each of the three ranking systems make clear that there are many more schools that care about energy and the environment and are also making great strides that do not appear on these lists. These schools might not have the time or resources to submit their data, the submission of the data to these third parties was not a priority, or they simply weren’t included on the U.S. News & World Report Top 100 Universities list and so their data was not included in the SaveOnEnergy Green Score list.

Source

That being said, schools that take the time to report their sustainability are showing that doing so is a priority to them and demonstrating a commitment to the cause that should be applauded and recognized. While there are many schools that didn’t report their data that are certainly still environmentally friendly (indeed, about half of the schools in Division I basketball ended up with a score of zero for not appearing in any of the three lists, but it would be foolish to believe that none of those 178 schools are working towards energy efficiency and sustainability), the submission of data can be considered a sign that transparency regarding sustainability is important to those in charge and thus the reporting schools earn a well-deserved place in the Green March Madness Tournament scoring. For that reason, the rest of this article will unapologetically use the Green March Madness Tournament Score as the definitive factor to determine sustainability rankings of the schools.

Quick facts and figures

Before moving on to selecting which teams made the prestigious Green March Madness Tournament, let’s take a look at a few quick facts from the scoring:

  • 173 out of 351 teams registered a score greater than zero on the Green March Madness Tournament Score, meaning over 100 schools who registered a non-zero score will still find themselves on the outside looking in.
  • Even rarer, though, are teams that have scores in all three scoring metrics used. Only 33 teams have a non-zero score in all three metrics, while only 112 teams have a non-zero score in two or more metrics.
  • As shown below in the table of conferences and conference champions, the highest score went to American University of the Patriot League with 73.4, while the lowest non-zero score went to South Dakota State of the Summit League with 9.2.
  • Looking at each of the 32 conferences:
    • 4 conferences (Pacific-12, Big Ten, Ivy League, and Atlantic Coast) had 100% of their teams score greater than zero.
    • 2 conferences (Atlantic Sun and Northeast) had only a single team score greater than zero, thus making the crowning of a conference champion rather easy.
    • 5 conferences (Big South, Metro Atlantic Athletic, Mid-Eastern Athletic, Southland, and Southwestern Athletic) didn’t have any teams score greater than zero.

Selecting the field

Even though this is mostly a silly exercise, I still wanted to follow the protocol of the real NCAA Basketball Tournament Selection Committee when determining who should make this ‘Big Green Dance’ (and, in doing so, gained some respect for the massive amount of puzzle pieces they must juggle!). The process is famously intense, with 10 committee members spending countless hours keeping up with the college basketball landscape during the year, only to convene for a five-day selection process that requires hundreds of secret ballots.

The entire process is very detailed, but it can be boiled down as follows:

  1. All 32 conference champions receive an automatic bid into the tournament
  2. The next best 36 teams are then chosen as ‘at-large bids’ to bring the total field to 68 teams
  3. All 68 teams are ranked from top to bottom, regardless of their status as a conference champion
  4. The top four teams are ranked as number one seeds in each of the four regions, then the next four are two seeds, the next four are three seeds, etc.
  5. While placing teams into each region, care is taken to ensure that each of the four regions is fairly equally balanced and that teams that played each other during the season are prevented from  having a rematch in the tournament until the later rounds (teams can be bumped up/down by a seed or two to assist in these requirements)
  6. The last four teams to make the tournament in at-large bids and the last four teams to make the field altogether are paired off to compete in the First Four games, with the winners advancing to the remaining field of 64.

While the criteria used to rank teams for the Selection Committee include resources such as the Rating Percentage Index (RPI), evaluations of quality wins based on where the game took place and how good the opponent was, and various computer metrics, things are easier in the Green March Madness Tournament Selection Committee as we only need to use the single number result of the Green March Madness Tournament Score.

The 68-team field

The bracket

For the full suite of teams, conferences, and scores, refer to the accompanying Google Spreadsheet of final figures. Using these numbers and sticking to the above selection guidelines as much as possible, the following bracket is the official result for the 2018 Green March Madness Tournament Bracket:

Click to enlarge

Breaking it down by each region for ease of reading:

The East region

The West region

The Midwest region

The South region

Note that the five conferences that didn’t produce a single team with a non-zero score would still get the automatic bids for their conference champion (four as play-in teams for the First Four and one more as a 16 seed without a play-in game), so perhaps they’ll draw straws to see who gets to go into the tournament. Regardless, they are in the bracket and labeled as that conference’s champion (placed in no particular order), just waiting to be beaten soundly by their respective sustainable opponents.

Analysis of the field

In terms of conferences, we see big winners come from the Pacific-12 (8 tournament teams) and the Big Ten (7 tournament teams), but in third is the surprise conference of the Ivy League (6 tournament teams) who is rarely in the conversation for getting more than a single team in the NCAA Basketball Tournament. On the other end of the surprises, the Big East and the Southeastern Conference (both major conferences that typically nab a handful of bids each) were kept to only one team each in the tournament.

For individual teams, we find some other surprises. A number of perennial stalwarts of the college basketball scene find themselves in the unfamiliar position of being on the outside looking in– 7 out of the 10 teams with the most NCAA Tournament appearances failed to receive a Green March Madness Tournament big (Kentucky, Kansas, UCLA, Louisville, Duke, Notre Dame, and Syracuse). On the other side of the coin, five teams (Denver, New Hampshire, William & Mary, UC Riverside, and Bryant University) that have never made the NCAA Tournament have finally found success with the Green March Madness Tournament.

Another common exercise leading up to the announcement of the NCAA Basketball Tournament teams is looking at the bubble teams, those that are just on the edge of making the tournament but find themselves potentially falling just short.  The most painfully close bubble teams for the 2018 Green March Madness Tournament were the five teams that fell less than one point shy of an at-large bid: Louisville, Northern Arizona, Ohio State, IUPUI, and Arkansas. Most painful was Louisville who fell just 0.12 points shy of being the last team in (though maybe it was serendipity– who knows if Louisville would have had to vacate that appearance, too).

What did the top performing schools have in common?

Looking at the teams that scored particularly high and scored the best seeds in the Green March Madness Tournament, a couple of trends appear:

  • Sustainability-focused schools: It’s worth noting that every team that was ranked in all three metrics ended up with a good enough score to make the tournament. As previously noted, such commitment to ensuring data is delivered for all three metrics shows the cause of sustainability is a priority and these schools are naturally rewarded by being guaranteed to make the Green March Madness Tournament.
  • City schools: A common theme found in the upper half of the schools that made the Green March Madness Tournament is that the are located in or near major U.S. cities (including one seeds American University and George Washington, three seed Northwestern, four seed Columbia, six seed Boston University, seven seed Denver, and eight seed Miami (FL)). The reason an urban setting might help schools score well in these rankings is because cities are more likely to have local sustainability organizations to partner with the school, access to effective public transportation, high walkability scores, and other nearby resources from the community that can be used for the school as well. Each of these factors positively effects the ratings that go into the Final Green March Madness Tournament Scores.
  • Green states:  Outside of the city in which a school is located, the state a school is in (and the state’s relative ‘green-ness’) has significant impact. The top of the tournament seeding is populated with teams from states often considered particularly green by various metrics. For example, the annual state scorecard rankings from the American Council for an Energy-Efficient Economy (ACEEE)  shows heavy representation from the top five states in the ACEEE scorecard in the Green March Madness Tournament: Massachusetts (Boston University, Harvard, Massachusetts), California (UC Santa Barbara, Santa Clara, UC Riverside, San Jose State, UC Irvine, Cal State Northridge, California, San Diego), Rhode Island (Brown, Bryant University), Vermont (Vermont), and Oregon (Oregon State, Portland State, Oregon, Pacific). Together, those five states account for over a quarter of the teams that made the Green March Madness Tournament, reflecting the benefits to institutions in states that commit to green jobs, renewable energy development, and other sustainability initiatives.

The National Champion

The downside of filling out our bracket based on the Green March Madness Tournament Scores is that by continuing through with the tournament, we won’t find any upsets and the top seeds will always win (again, we’ll revisit once the real NCAA Basketball Tournament bracket is released to see which of those teams would win based on sustainability). In the end, our Final Four is made up of all one seeds, as shown below, with the final champion being…

Drumroll…

 

American University! In the three times appearing in the NCAA Basketball Tournament, the Eagles have gone winless– but once the Green March Madness Tournament comes along they go all the way! Congratulations to them, and best of luck to all schools in the ‘real’ tournament in March, to all schools looking to improve their sustainability scores before next year’s Green March Madness Tournament, and to all of you in finding the best way to fill out the brackets for you office pool this year!

Sources and additional reading

Cool Schools 2017 Full Ranking: Sierra Club

March Madness bracket: How the 68 teams are selected for the Division I Men’s Basketball Tournament: NCAA

SaveOnEnergy 2017 Green Report: Top Universities in the U.S.: SaveOnEnergy

The Sustainable Tracking, Assessment & Rating System: Association for the Advancement of Sustainability in Higher Education

About the author: Matt Chester is an energy analyst in Washington DC, studied engineering and science & technology policy at the University of Virginia, and operates this blog and website to share news, insights, and advice in the fields of energy policy, energy technology, and more. For more quick hits in addition to posts on this blog, follow him on Twitter @ChesterEnergy.  

Debunking Trump’s Claim of “War on Beautiful, Clean Coal” Using Graphs

In President Trump’s first State of the Union Address last week, a wide range of topics in the Administration’s agenda were covered extensively while energy was largely pushed to the side. Trump did include two sentences on his self-described push for “American Energy Dominance,” and these two sentences sent wonks in the energy industry into a frenzy on social media:

“We have ended the war on American energy. And we have ended the war on beautiful, clean coal.”

My Twitter feed lit up with various energy journalists and market watchers who noted the impressiveness that just 18 words over two sentences could contain so many misleading, or outright false, claims.

Source

As one of those energy reporters who immediately took to Twitter with my frustration, I thought I would follow up on these statements last week with arguments why the claims of ‘clean coal’ and the supposed ‘war’ on it do not reflect the reality the Trump Administration would have you believe, and I’ll do so with just a handful of graphs.



What is ‘clean coal’?

As a pure fuel, coal is indisputably the ‘dirtiest’ energy source in common use in the power sector, accounting for about 100 kilograms (kg) of carbon dioxide (CO2) per million British thermal unit (MMBtu) of energy output. This output is notably larger than other major energy sources, including natural gas (about 50 kg/MMBtu), petroleum products like propane and gasoline (about 60 to 70 kg/MMBtu), and carbon neutral fuels like nuclear, hydroelectric, wind, and solar. In the face of the scientific consensus on CO2’s contributions to climate change, many have noted that one of the best actions that can be taken in the energy industry is to shift away from coal to fuels that emit less CO2— which has definitively given coal a dirty reputation.

The premise of ‘clean coal’ is largely a PR push (literally invented by an advertising agency in 2008)– an ingenious marketing term, but one that does not have much in the way of legs. When you hear politicians talking about ‘clean coal,’ it is usually referring to one or more of the following suite of technologies:

  • Washing coal before it’s burned to remove soil and rock and thus reduce ash and weight of the coal;
  • Using wet scrubbers on the gas generated from burning coal to remove the sulfur dioxide from being released;
  • Various carbon capture and storage (CCS) technologies for new or existing coal plants that intervene in the coal burning process (either pre-combustion or post-combustion) to capture up to 90% of the CO2 produced from its burning and then sending it miles underground for permanent storage instead of releasing it into the atmosphere; or
  • Anything done to the coal-fired power plant to increase the efficiency of the entire process of generating electricity (e.g., the 700 Megawatt supercritical coal plant in West Virginia that is so efficient it reportedly releases 20% less CO2 than older coal plants) and reduce the overall emissions.

Source

When most in the energy industry discuss ‘clean coal’ technology, they are typically referring to CCS. However it should be noted that Trump did not mention CCS by name in this (or any) speech. Some analysts have noted that the White House’s attempts to cut CCS funding and send the Secretaries of the Department of Energy (DOE) and Environmental Protection Agency (EPA) to supercritical coal plants are not-so-subtle hints that the Trump Administration’s preferred type of ‘clean coal’ is improving the efficiency of coal-fired generation. Even Bob Murray, the influential coal magnate, has written to the President to indicate his contempt for CCS, calling it a ‘pseudonym for no coal,‘ echoing the concerns of many proponents of coal that CCS is being pushed as the only ‘clean coal’ option so that if/when it fails (due to economic impracticalities) it would be the death knell of coal-fired generation altogether.

So regardless of which ‘clean coal’ technology the Trump Administration supports, issues remain. With regard to wet scrubbers, coal washing, and general plant efficiency improvements, the reductions in CO2 emissions are not nearly enough to compete with cleaner fuels. Even if all coal plants could be made 20% more efficient (and less reduce CO2 emissions by about 20%) like the West Virginia supercritical plant, which would be a massive undertaking, it would still result in coal generation being among the dirtiest energy in the country.

With regard to CCS, not only is the cost one of the biggest issues (which will be looked at in more detail later), but it does not remove all the pollutants from burning coal. Even with the most effective CCS capturing 90% of CO2 emissions, that leaves 10% of CO2 making its way into the atmosphere along with the other notable pollutants in coal gas (including mercury, nitrogen oxide, and other poisonous contaminants). When compared with the carbon neutral energy sources increasingly gaining ground in the United States, coal plants with CCS still hardly seem clean.

Again, the Energy Information Administration’s (EIA) listing of carbon dioxide emissions coefficients shows the CO2 emissions associated with different fuel types when burned as fuel. As previously noted, coal is the far-away leader on CO2 emissions coefficients as a pure fuel. In DOE analysis of future-built generation (an analysis that focuses on the costs and values of different types of power plants to be built in the future, which will come up again in more detail later), the only type of coal generation even considered is coal with either 30% or 90% carbon sequestration, with 90% being the technological ceiling and 30% being the minimum example of new coal-fired generation that would still be compliant with the Clean Air Act. The below graph, our first in demonstrating the issues with claims of a ‘war on beautiful, clean coal,’ plots the CO2 coefficients of major fuel sources in the U.S. power sector, including coal using no CCS, 30% CCS, or 90% CCS. Existing power plants do not have the same requirements under the Clean Air Act, so they might still be producing CO2 at the far right of the ‘coal’ bar (indeed, last year almost 70% of U.S. coal was delivered to power plants that are at least 38 years old meaning they are likely far from the most efficient coal plants out there). Coal plants that are touted as ‘clean’ because of their up to 20% increases in efficiency would still find themselves in the same (or greater) range of emissions as 30% CCS coal plants, while 90% CCS coal plants appear to the be the only ones that can compete with other fuels environmentally (though it comes at a potentially prohibitive cost, which will show up in a later graph).

Note that the data for these CO2 emission coefficients come from this EIA listing. The lines for 30%/90% CCS are not just drawn 30%/90% lower, but rather account for the presence of CCS requiring more energy and thus cause a dip in efficiency– this graph uses the rough efficiency drop assumed for CCS plants in this International Energy Agency report

These numbers paint a scary picture of coal and are the source of what causes many energy prognosticators to scoff at the utterance of ‘beautiful, clean coal,’ though it is important to be clear that these numbers don’t tell the whole story. While nuclear and renewable energy sources do not emit any fuel-related CO2, they are not completely carbon neutral over their lifetimes, as the building, operation, and maintenance of nuclear and renewable generation plants (as with any utility-scale generation source) all have their own non-zero effect on the environment. However, since fuel makes up the vast majority of carbon output in the electricity generation sector, any discussion of clean vs. dirty energy must return to these numbers.

Further, the separation of dispatchable vs. non-dispatchable technologies (i.e., energy sources whose output can be varied to follow demand vs. those that are tied to the availability of an intermittent resource) shown in the above graph is important. Until batteries and other energy-storage technologies reach a point technologically and economically to assist renewable (non-dispatchable) energy sources fill in the times when the energy resource is unavailable, dispatchable technologies will always be necessary to plug the gaps. So regardless of what drawbacks might exist for each of the dispatchable technologies, CO2 emissions and overall costs included, at least some dispatchable energy  will still be critical in the coming decades.

Who is orchestrating the ‘war on coal’?

Even with the knowledge that coal will never truly be ‘clean,’ the question then becomes why haven’t the advancements in coal energy that is cleaner and more efficient than traditional coal-fired plants become more prominent in the face of climate and environmental concerns? The common talking point from the Trump Administration is that there is a biased war on coal being orchestrated, and the actions of President Trump to roll back regulation is the only way to fight back against this unjust onslaught that the coal industry is facing. But again, from where is this onslaught coming?

The answer to this question is actually pretty easy– it’s not regulation that is causing coal to lose its place as the king of the U.S. power sector, it’s competition from more affordable energy sources (that also happen to be cleaner). The two charts below demonstrate this pointedly, with the left graph showing the fuel makeup of the U.S. electric power sector since 1990 along with the relative carbon intensity of the major CO2-emitting fuel sources, while the right graph shows what’s happened to the price of each each major fuel type over the past decade. The carbon intensity shown on the left graph is even more indicative than the first graph above in detailing the actual degree to which each fuel is ‘clean’ as it factors in the efficiency of plants using the fuel and indicates the direct CO2 emissions relative to electricity delivered to customers.

Click to enlarge

Note that the costs are taken from this EIA chart, with coal taken from fossil steam, natural gas taken from gas turbine and small scale, and wind/solar taken as the gas turbine and small scale price after removing the cost of fuel. Electric power generation and carbon emission data taken from this EIA source

Just from analysing these two graphs, a number of key observations and conclusions can be made about the electric power sector and coal’s evolving place in it:

  • In 1990, coal accounted for almost 1.6 million Gigawatt-hours (GWh) of power generation, representing 52% of the sector. By 2016, that figure dropped to 1.2 million GWh or 30% of U.S. power generation.
  • Over that same time period, natural gas went from less than 400,000 GWh (12%) to almost 1.4 million GWh (34%); nuclear went from less than 600,000 GWh (19%) to over 800 GWh (20%), and combined wind and solar went from 3,000 GWh (0.1%) to over 260,000 GWh (6%).
  • While the coal sector’s carbon intensity hovered around 1.0 kg of CO2 per kilowatt-hour (kWh) of electricity produced from 1990 to 2016 (even as CCS and other ‘clean coal’ technologies began to break into the market), natural gas dropped from 0.6 kg CO2/kWh to less than 0.5 kg CO2/kWh, while nuclear, wind, and solar do not have any emissions associated with their generation (again noting that there are some emissions associated with the operation and maintenance of these technologies, but they are neglible compared with fossil fuel-related emissions). The drop in natural gas carbon intensity combined with coal losing ground to natural gas, nuclear, and renewable energy led the electric power sector’s overall average carbon intensity to drop from over 0.6 kg CO2/kWh to less than 0.5 kg CO2/kWh.
  • While the narrative some would prefer to push is that coal is getting replaced because of a regulatory ‘war on coal,’ the real answer comes from the right graph where the cost to generated a kWh of electricity for coal increased notably from 2006 to 2016. Meanwhile, natural gas (which started the decade more expensive than coal) experienced a drastic drop in price to become cheaper than coal (thanks to advances in natural gas production technologies) while the low cost of nuclear fuel and ‘free’ cost of wind and solar allowed these energy sources to start and remain well below the total cost of coal generation. This natural, free-market competition from other energy sources, thanks to increasingly widespread availability and ever decreasing prices, is what put pressure on coal and ultimately led to natural gas dethroning coal as the predominant energy source in the U.S. power sector.

What these two graphs show is that the energy market is naturally evolving, there is no conspiratorial ‘war’ on coal. The technologies behind solar and wind are improving, getting cheaper, and becoming more prolific for economic, environmental, and accessibility reasons. Nuclear power is holding strong in its corner of the electricity market. Natural gas, more than any other, is getting cheaper and much more prominent to the U.S. power sector (while having the benefit of about half the CO2 emissions of coal), which is what has made it the natural ‘enemy’ of coal of the past decade or two. All that’s to say, the only ‘war on coal’ that’s been widespread in recent memory is a capitalistic, free-market war that will naturally play out when new energy sources are available at cheaper prices and contribute significantly less to climate change.

Will Trump policies reverse the course of coal in the United States?

Going back to the statement from Trump’s State of the Union Address, he claimed that his Administration had ended the war on clean coal. As stated previously, there was never an outward war on coal that was hindering the fuel. Even still, the main policy change from the Trump Administration with regard to coal was to repeal the Clean Power Plan (CPP) that aimed to cut carbon emissions from power generation.  However, many analysts predicted that would not change the current trends, as repealing the CPP does nothing to reverse the pricing pattern of the fuels. Indeed, this week EIA released its Annual Energy Outlook for 2018 and confirmed the tough future that coal generation has compared with natural gas and renewables– both with and without the CPP. While the CPP reduces the projections of coal generation, it doesn’t move the needle all that much and natural gas and renewables are still shown to surpass coal.

Source

So the major policy decision of the Trump Administration with respect to coal generation doesn’t appear to reverse the course of coal’s future. Again, this conclusion isn’t terribly surprising considering the economics of coal compared with other fuels. EIA projects the Levelized Cost of Electricity (LCOE) for different type of new power generation (assumed to be added in 2022) which serves to show the relative costs to install new power generation. In the same analysis, EIA projects Levelized Avoided Cost of New Generation (LACE), which can be thought of as the ‘value’ of the new generation to the grid (for more detailed description in the calculations and uses of these measures, read through the full report). When the LACE is equal or greater than the LCOE, that is in indication of a financially viable type of power to build (evaluated over the lifetime of the plant). So by looking at the relative costs (LCOE) of each power type and whether or not they are exceeded by their values (LACE), we can get a clear picture of what fuel types are going to be built in the coming years (and to continue the focus on whether coal or other fuels are ‘clean,’ let’s put the economics graph side-by-side with the CO2 emissions coefficients):

Click to enlarge

Note that the source of the data on the left graph is the EIA Levelized Cost of Electricity analysis, with the ends of the boxes representing the minimum and maximum values and the line in the middle representing the average– the difference in possible values comes from variations in power plants, such as geographic differences in availability and cost of fuel. Also note that, counter-intuitively, EIA’s assumed costs for 30% CCS are actually greater than for 90% CCS because the 30% CCS coal plants would ‘still be considered a high emitter relative to other new sources and thus may continue to face potential financial risk if carbon emissions controls are further strengthened. Again, the data for the right graph takes CO2 emission coefficients from this EIA listing by fuel type

Looking at these graphs, we can see that the cost of new coal generation (regardless of CCS level) not only exceeds the value it would bring to the grid, but also largely exceeds the cost of natural gas, nuclear, geothermal, biomass, onshore wind, solar photovoltaic (PV), and hydroelectric power (all of which emit less CO2 than coal). Thus even in the scenario where 90% of carbon is captured by CCS (which allows it to be ‘cleaner’ than natural gas and biomass), it still comes at a significant cost premium compared with most of the other fuel types. These are the facts that are putting the hurt on the coal industry, not any policy-based ‘war on coal.’ Even the existing tax credits that are given to renewable energy generation are minor when looking at the big picture, as the below graph (which repeats the above graph but removes the renewable tax credits from the equation) shows. Even if these tax credits are allowed to expire, the renewable technology would still outperform coal both economically and environmentally.

Click to enlarge

The last graphical rebuttal to President Trump’s statement on energy and coal during the State of the Union that I’ll cite comes from Tyler Norris, a DOE adviser under President Obama:

Source

As pointed out by Norris and other energy journalists chiming in during the State of the Union address, if the goal were to expand ‘clean coal,’ then the Trump Administration’s budget is doing the opposite by taking money away from DOE programs that support the research and development of the technology. In fact, at the end of last week a leaked White House budget proposal indicated even further slashes to the DOE budget that would further hamper the ability of the government to give a leg up to the development of ‘clean coal’ technology. Any war on energy is coming from the Trump Administration, and any battle that coal is fighting is coming from the free market of cheaper and cleaner fuels.

Sources and additional reading

20 Years of Carbon Capture and Storage: International Energy Agency

Annual Energy Outlook 2018: Energy Information Administration

Average Power Plant Operating Expenses for Major U.S. Investor-Owned Electric Utilities, 2006 through 2016: Energy Information Administration

Carbon Dioxide Emissions Coefficients: Energy Information Administration

Did Trump End the War on Clean Coal? Fact-Checking the President’s State of the Union Claim: Newsweek

How Does Clean Coal Work? Popular Mechanics

How much carbon dioxide is produced per kilowatthour when generating electricity with fossil fuels? Energy Information Administration

Is There Really Such a Thing as Clean Coal? Big Think

Levelized Cost and Levelized Avoided Cost of New Generation Resources in the Annual Energy Outlook 2017: Energy Information Administration

Trump touts end of ‘war on beautiful, clean coal’ in State of the Union: Utility Dive

Trump’s Deceptive Energy Policy: New York Times

What is clean coal technology: How Stuff Works

About the author: Matt Chester is an energy analyst in Washington DC, studied engineering and science & technology policy at the University of Virginia, and operates this blog and website to share news, insights, and advice in the fields of energy policy, energy technology, and more. For more quick hits in addition to posts on this blog, follow him on Twitter @ChesterEnergy.  

How Much Power Is Really Generated by a Power Play?

As a huge sports fan who works in and writes about the energy industry, stumbling across this article that compared the kinetic energy produced by the high velocity projectiles in different sports got my creative juices flowing. By the estimates in that article, shooting a hockey puck produces the highest kinetic energy in all of sports.

Not only does it appear that hockey can take the ‘energy crown’ in sports, but a common occurrence during a hockey game is a ‘power play.’ A power play occurs when the referee determines that a player has committed a foul and that player is sent to spend a set number of minutes in the penalty box. During that time in the penalty box, the opposing team has the advantage of one additional player and are said to be on a power play– and if they score during that time then it is called a power play goal. While this power play has absolutely nothing to do with power plants or power generation, the idea that hockey pucks have the most kinetic energy in sports got me to wondering about what sort of power generation could be harnessed by power play goals in the National Hockey League (NHL).



If we wanted to harness the power of power plays (why would we want do do that? Maybe it’s the part of a plot by a wacky cartoon villain!), how much would that be? Why don’t we sit down and do the math!

Energy from a hockey puck

To start, we need to determine what the energy of a single hockey shot should be assumed to be (as the previously mentioned article does not include all of the necessary assumptions for academic rigor). High school physics class taught us that the kinetic energy is determined by taking one half times the mass of the object times the square of the speed of that object.

Source

Official ice hockey pucks weigh 170 grams, so we just need to figure out what to assume as the speed of the puck. Obviously every shot of the puck comes at a different speed depending on who is shooting, what type of shot is used (e.g., slap shot vs. wrist shot), how fatigued the player is, the condition of the ice, and many other factors. But for the sake of this back-of-the-envelope calculation, we can look at a couple of data points for reference:

  • The official NHL record for shot speed is 108.8 miles per hour (MPH) by Zdeno Chara in the 2012 All-Star Skills Competition;
  • Guinness World Records recognizes the hardest recorded ice hockey shot in any competition as 110.3 MPH by Denis Kulyash in the 2011 Continental Hockey League’s All-Star Skills Competition;
  • When discussing the benchmark of a particularly strong slapshot, 100 MPH is often used as the benchmark of a player getting everything behind a shot;
  • Finding benchmarks for the wrist shot is not as prevalent (people like to discuss the hardest shots possible, hence data on slap shots and not wrist shots), but some estimates show that wrist shots can reach speeds of 80 to 90 MPH; and
  • Estimates put wrist shots as accounting for 23 to 37 percent of all shots taken in professional hockey.

Given those figures, a rough estimate of average NHL shot speed can be determined by assuming slap shots are about 100 MPH and account for 70 percent of shots, while wrist shots are about 85 MPH and account for 30 percent of shots:

For the sake of this exercise, we’ll call the speed of a NHL shot 95.5 MPH, which equals about 42.7 meters per second (m/s). Plugging that speed and the 170 gram weight of the puck into our kinetic energy equation leaves us with an assumed ‘Power Play Power’ of an NHL power play goal of 154.9 Joules (J)– just over 0.04 kilowatt-hours (kWh).

For the rest of this article, we’ll refer to the energy gathered from power play goals, 154.9 J at a time, as ‘Power Play Power’– though please keep in mind the cardinal rule that power is the rate of energy over time, while the Joules and kilowatt-hours we’re talking about is total energy

Source

How much power can be harnessed from power plays?

The next step in reality would be to figure out how exactly you intend to extract ‘Power Play Power’ into actually generated energy, though that can be left up to the hypothetical cartoon villain who would be using such odd methods to create energy for his evil plots, as he did with the champagne bottles on New Year’s Eve (Side note, if I continue to write articles about the bizarre energy sources only thought up by a misguided cartoon villain, he needs a name– so in the spirit of villains like Megatron, Megamind, and Mega Shark, the energy-obsessed villain will be named Megawatt!)

But ignoring the question of how or why we would be extracting energy from ‘Power Play Power,’ let’s just look at what type of power will be generated based on 154.9 J per power play goal. Also note that there’s nothing special about the energy generated by a power play goal compared with a regular goal or even a shot that misses the goal– but where would the fun be without wordplay? POWER play goals only!

Most individual power play goals in a season

Note that all of the statistics pulled for this analysis are current as of January 1, 2018. Any power play goals scored after that date will not be accounted for in these statistics and calculations.

Pulling the top 10 individual player seasons with the most power play goals in NHL history, and assuming each of those power play goals account for 154.9 J, gives the following results:
Despite an impressive 34 power play goals in the 1985-86 season, Tim Kerr’s NHL record season would only generate enough ‘Power Play Power’ to run a large window-unit air conditioner for one hour at almost 1.5 kWh.

What about considering single players over their entire career?

Most individual power play goals over a career

As of January 1, 2018, the top 10 power play goal scorers for an entire career are as follows (note that as of writing, Alex Ovechkin is still active, as is Jaromir Jagr who is only two power play goals behind him in 11th place):
Looking at Dave Andreychuk, the individual with the most career power play goals in NHL history, his career ‘Power Play Power’ accounts for almost 11.8 kWh. Despite being an incredibly impressive number of power play goals, it’s only enough to power an energy-efficient refrigerator for about a week and a half. That’s a useful amount of energy to use in your home, but when it takes 274 career power play goals that that might be more work than it’s worth…

However looking at these first two charts, one aspect really jumps out– players who come from Canada appear to dominate ‘Power Play Power’ generation! Let’s dig into that a bit more.

Most power play goals by country of origin in the NHL

To start, Quant Hockey’s data shows that there are only 25 different home countries across all the players who have ever scored a power play goal in NHL history. Those 25 countries are listed in the below chart with their respective ‘Power Play Power’ totals generated:

Now we’re talking about some real energy. Canada, as predicted, dominates with almost 2,250 kWh of ‘Power Play Power’ since the beginning of the NHL. This amount of energy equates to about 20% of the average annual electricity used by an American household in 2016.

So that’s a pretty significant amount of energy on a micro-scale, but because we’re talking about the total ‘Power Play Power’ generated by all Canadian NHL players over nearly a century of play it is still not terribly impressive. For reference, the smallest nuclear power plant in the United States has a generation capacity of 582 Megawatts, meaning the 2,250 kWh of ‘Power Play Power’ of Canadian NHL players would be generated in under 14 seconds by the smallest U.S. nuclear plant operating at full capacity. Even if we included all power play goals scored by players of any nationality, the total ‘Power Play Power’ would only reach 3,339 kWh– or almost 21 seconds from the smallest U.S. nuclear plant.

Source 1, Source 2

Obviously the actual energy generation of each of these 25 nations will be much greater than the ‘Power Play Power’ generated by their respective NHL players– but is there some sort of correlation between ‘Power Play Power’ and actual energy production of the nations? Using the silly initial premise of this article as an example of the type of information available from the Energy Information Administration (EIA), a part of the U.S. Department of Energy, and how to find that data, we can pull the total primary energy production for these 25 countries and get a rough idea! While the NHL started recording power play goals in the 1933-34 season, EIA’s country-by-country energy production data dates back to 1980 (measured using quadrillion British thermal units, or quads), but we’ll still use these two complete time frames for the comparison’s sake. Putting the two energy figures on one graph for a relative comparison provides the following:
This graph presents a couple of interesting points:

  • Among the 25 eligible nations included in the survey, Canada, the United States, and Russia all find themselves in the top 4 countries in terms of both ‘Power Play Power’ and Total Primary Energy Produced by the nation;
  • In an interesting coincidence, when the two types of energy being measured here are put on comparative scales, Canada and the United States appear to be almost mirror images of each other, swapping relative strength in ‘Power Play Power’ and Total Primary Energy Production;
  • In another similarity between the two measures of energy, the totality is dominated by the top three nations, and the relative scale of any nation after about the halfway point shows up as barely even a blip on this graph.

But other than that, it can be considered fairly unsurprising that NHL power play success doesn’t directly translate to Total Primary Energy Produced by nation. And even if Canada saw their NHL power play prowess as their opportunity to increase energy exports (which would only serve to increase the fact that Canada is the largest energy trading partner of the United States), translating ‘Power Play Power’ into real energy, their 2,250 kWh over NHL history would only translate to 0.00000004% of Canada’s primary  energy produced in 2015 alone. Unfortunately, I do not think I’ve discovered a viable energy to be harnessed by the villainous Megawatt.

Source

More benevolently, it would also appear that ‘Power Play Power’ will not serve as a reliable new renewable energy source for hockey-crazed areas (in this scenario, are we to consider penalty minutes a source of renewable energy?? If so, Tiger Williams might be the most environmentally friendly player in major sports history). However, at 419 billion kWh of renewable generation in 2015, Canada is the fourth largest renewable energy producer worldwide (with the United States and Canada being the only nations this time to finds themselves in the top four of of renewable energy and ‘Power Play Power,’ as North America accounts for majority of NHL players and has collectively agreed to generate 50% of electricity from clean sources by 2025). Following the link for EIA international renewable energy data to bring this back to educational purposes, you’ll find other top-15 ‘Power Play Power’ nations that also account for the top-15 in global renewable energy production, including the United States, Germany, Russia, Sweden, and the United Kingdom.

Coincidence? Probably.

Interesting and informative, nonetheless? Definitely!



Sources and additional reading

Appliance Energy Use Chart: Silicon Valley Power

Comparing Sports Kinetic Energy: We are Fanatics

How much electricity does a nuclear power plant generate? Energy Information Administration

How much electricity does an American home use? Energy Information Administration

Iafrate breaks 100 mph barrier: UPI

International Energy Statistics: Energy Information Administration

Most Power-Play Goals in One Season by NHL Players: Quant Hockey

NHL & WHA Career Leaders and Records for Power Play Goals: Hockey Reference

NHL Totals by Nationality – Career Stats: Quant Hockey

Now You Know Big Book of Sports

Ranking the 10 Hardest Slap Shots in NHL History: Bleacher Report

Saving Electricity: Michael Bluejay

Scientists Reveal the Secret to Hockey’s Wrist Shot: Live Science

Score!: The Action and Artistry of Hockey’s Magnificent Moment

Sherwood Official Ice Hockey Puck: Ice Warehouse

Slap Shot Science: A Curious Fan’s Guide to Hockey

Total Renewable Electricity Net Generation 2015: Energy Information Administration

Wrist Shots: Exploratorium

About the author: Matt Chester is an energy analyst in Washington DC, studied engineering and science & technology policy at the University of Virginia, and operates this blog and website to share news, insights, and advice in the fields of energy policy, energy technology, and more. For more quick hits in addition to posts on this blog, follow him on Twitter @ChesterEnergy.  

Talking Turkey: Thanksgiving Dinner Energy Use and Carbon Dioxide Emissions

Thanksgiving is one of the most wonderful time of the year, when families gather and spend time together while the smell of turkey seeps in from the other room. You’ve probably never given much thought to the energy use or environmental impact behind that intoxicating turkey smell coming from the kitchen, and in fact the country’s overall energy use drops on Thanksgiving because the increase in kitchen power use is offset by the drop in energy use from office and commercial buildings that are closed for the holiday.

However it’s always interesting to look at the actual energy numbers behind various regular activities and consider if there’s a way to do it better. Especially these days when online cooking forums and the Food Network is constantly making it trendy to cook your Thanksgiving turkey in new and novel ways. Your grandmother’s recipe isn’t the only one in town anymore (though I’m sure it’s still the best). Those cooking the turkey now have deep fryers and smokers, while Turducken is being eaten by NFL players after the Thanksgiving Day games.



With so many new cooking methods for Thanksgiving dinner, it got me to wonder what the energy cost was to cook turkey using these different methods. While there were investigations on the total energy use across the country to cook Thanksgiving dinner (linked later in this article), I could not find anything about the energy cost or associated carbon dioxide (CO2) emissions of an individual turkey cooked using different methods, so I thought I’d run through them myself!

Recipes

After searching across the Internet, I settled on seven different methods to cook your Thanksgiving turkey– the traditional roasting of a turkey and six newer and trendier options that the hip or contrarian chef might utilize. These seven methods are the following:
  • Roasting;
  • Braising;
  • Deep frying;
  • Grilling;
  • Smoking;
  • Spatchcocking; and
  • Sous vide.

For each of these cooking methods, I’ve sought out a recipe either from a well-known chef of repute or directly from the manufacturer of the turkey or the cooking apparatus in question. By using these recipes, ideally these authorities will have an air of authority to them. Because each recipe offers cooking times based on various size turkeys, this analysis will normalize each recipe for a standard 15 pound (lb) turkey as the size recommended for a dinner of 12 people.

If you want to skip the details of the recipes and the calculations, click here to go straight to the results!
 
Note: For all of the below recipes, there are additional energy consuming steps that are not going to be included in the calculations. These steps include removing turkey from the oven to baste, pre-refrigeration, sauteing after the turkey is fully cooked to get crispy skin, etc. The point is the calculations below will focus on the energy needed to fully and safely cook the turkey, and any energy used before or after that process will be ignored for simplicity and uniformity. Of course you will be making side dishes and putting on finishing touches, so your mileage WILL vary compared with what is calculated here. The goal of this exercise is just to get a back-of-the-envelope approximation for how the different cooking methods affect the energy required– they are definitely not going to be exact or completely robust. You’ve been warned! 

Roasted turkey

Since this is the traditional cooking method, it seemed criminal to use a recipe other than the one championed by Julia Childs. Her traditional recipe for a 10 to 13 pound turkey calls for the oven to be preheat to 450oF and then the turkey roasted for 30 minutes before reducing the oven to 350oF and roasting for another 2 to 2 hours 30 minutes.
Normalizing for a 15 lb turkey, we’ll use the higher time estimate and add 15 minutes for the extra weight and say the turkey will be cooked in the oven for 3 hours 15 minutes.

Braised turkey

 Braised turkey is a great segue from the traditional to the more novel turkey-cooking methods, as it doesn’t stray too far from the original whole turkey roasting method. You are still cooking the turkey fully in the oven, but with the main difference that the turkey is sitting in a pan of vegetables and stock to bring in more moisture to your turkey.

For the braised turkey, we’ll stay with household names and use Bobby Flay’s recipe for herb roasted and braised turkey. This recipe calls for an oven preheated to 450oF with the 17 pound turkey and a bed of vegetables cooked for 45 minutes before the temperature is reduced to 350oF and cooked an additional 2 to 2 hours 15 minutes longer (while basting with warm chicken stock). After the whole bird is cooked, the legs are removed and braised in a roasting pan with stock for an additional 1 hour at 350oF.

To normalize to a 15 pound turkey, we’ll say the braised turkey cooks in the oven for a total estimated cook time of 3 hours 30 minutes.

Deep fried turkey

If you can manage to get it done without an explosion or trip to the hospital, deep frying turkey has become one of the more exciting and talked about cooking alternatives. Bobby Flay’s colleague at Food Network, Alton Brown, has one of the most used deep fried turkey recipes for those who love the science and Internet-trends of cooking.

For a 13 to 14 pound turkey, Alton has you heat up a 28 to 30-quart pot of oil to 250oF, add in the turkey and raise the temperature to 350oF, and once at that temperature cooking for 35 minutes.

To account for the weight of a 15 pound turkey, we’ll say this recipe cooks with a propane heater for a total of 40 minutes.

Grilled turkey

The grilled turkey recipe chosen comes straight from Butterball, the turkey supplier that accounts for 20 percent of total turkey production in the United States. Among grilling aficionados, the debate to grill by charcoal or by gas is one of the most heated. In addition to differences in taste, ease, and convenience, the choice of grill type also affects the end energy use to cook. Luckily for us, Butterball provides instructions for both a charcoal and gas grill.

Butterball’s recipe for charcoal grilling says that the 10 to 16 pound turkey will be cooked over 50 to 60 charcoal briquettes (after those initial briquettes have been burned for 30 minutes). At that point, the turkey is to be placed on the grill for 2 to 3 hours, with 12 to 16 briquettes being added every 45 minutes to 1 hour. To normalize at the 15 pound turkey, we’ll estimate that initially 60 charcoal briquettes will be used and, during the cooking process, 50 more briquettes will be added for a total cooking fuel of 110 charcoal briquettes on a charcoal grill over the course of 3 hours.

Butterball’s recipe for gas grilling says the same 10 to 16 pound turkey is cooked over indirect heat (after 10 to 15 minutes of preheating) at 350oF for 2 to 3 hours. For the 15 pound turkey we’ll assume the turkey is cooked on a gas grill at 350oF for the whole 3 hours.

Smoked turkey

Where deep frying or grilling the turkey may have once held the title as the ‘macho’ way to prepare a Thanksgiving turkey (whatever that may mean), smoking the meat might just have taken that crown. Using lower heat over longer periods of time, smoking turkey evokes the expert barbecue pit masters of the country to impart full flavor without drying out the turkey. Butterball once again provides authoritative guidance to smoking your Thanksgiving dinner, again allowing the consideration of two different fuel types.

Butterball’s recipe for preparing a turkey in a water smoker uses 10 pounds of charcoal briquettes (pre-burned for 30 minutes) to start the cooking process, adding in 12 to 14 more charcoal briquettes every 1 hour 30 minutes to ensure the temperature remains at 250oF through a total cooking time of 6 to 10 hours for a 12 to 18 pound turkey. For our 15 pound turkey, we’ll call that cooking fuel of 10 pounds plus 70 briquettes of charcoal over a cooking time of 8 hours in the water smoker.

When using an electric smoker, Butterball’s recipe calls for the smoker to be set at 225oF and the 8 to 18 pound turkey to be cooked for 8 to 12 hours. Normalizing to our 15 pound turkey, we’ll say the final cook time is 11 hours at 225oF in the electric smoker.

Spatchcocked turkey

If Julia Child was the first queen of celebrity chefs, Martha Stewart eventually took her crown, and so we have to include a recipe of Martha’s.  Martha Stewart’s magazine featured a recipe for a spatchcocked turkey, a method of cooking poultry in which bones are removed so the bird can be flattened and cooked more evenly and quickly.
Martha Stewart’s recipe has the oven preheated to 450oF, with a 12 pound and fully spatchcocked turkey roasted for 1 hour 10 minutes. For our 15 pound turkey, we’ll adjust this to be cooked in the oven at 450oF for 85 minutes.

Sous vide turkey

Sous vide cooking, or the process of cooking food that is vacuum-sealed in a plastic pouch by placing it in heated and circulating water bath, has been around for decades. The method has gained traction more recently, however, as home cooks are increasingly getting their hands on the cooking equipment necessary that was previously only available in professional kitchens. The cooking method allows meat to be cooked at lower temperatures and thus cooked more evenly, safely, and while retaining moisture.

If you are in the market for a sous vide immersion circulator, one of the first places you might go is Williams Sonoma. To aid the new owners of this equipment, they also offer up a sous vide turkey recipe by Michael Voltaggio. The water of the sous video immersion circulator is preheated to 150oF and the vacuum sealed turkey pieces then placed in and cooked for 2 hours 30 minutes.

Calculations

These recipes use a wide variety of cooking apparatuses and fuels, so the methodology of calculating the total energy use and associated CO2 emissions will vary. Much like the Halloween-themed post on the most sustainable way to light your Jack-O’-Lantern, this post will thus be calculating very rough estimates using educated choices of data and assumptions. The final numbers should be considered back-of-envelope calculations and not scientifically or rigorously tested. There are also various aspects to the cooking process that would impact the end result that will not be accounted for, as well as variables to your individual cooking efforts that would change the final result (e.g., size of oven or grill, the energy mix of your power supplier, what type of propane or charcoal you buy from the store).

All that said, if you have ideas or suggestions on how to refine any of the numbers calculated here, then please reach out and/or leave a comment! (For one, I’ve assumed an oven is using a uniform amount of power regardless of the temperature at which it is set. While the difference of power use at 350oF and 450oF is not likely that much, it is definitely measurable. However, after much digging I was still unable to find any way to estimate the power difference among different temperatures, so a uniform power consumption was chosen and used for all use of the oven.)

Regardless of fuel type, all final energy numbers are calculated in kilowatt hours (kWh) and all CO2 emissions are calculated in lbs.

If you don’t care about going through the calculations and just want to jump to the final numbers, click here to jump to the results!



Roasted turkey

We are assuming the use of an oven for 3 hours 15 minutes. The oven will also need to preheat the oven, which we’ll assume to take 15 minutes. All together, the energy use and CO2 emissions will be associated with using an oven for a total of 3 hours 30 minutes.

In the United States, ovens are commonly powered by either electricity or by natural gas (though electric stoves are almost twice as common as gas stoves). The fuel type will affect the end energy use and CO2 emissions:
Electric oven:

Electric ovens use about 2.0 kilowatts (kW) of power. Assuming this power usage for the entirety of the recipe, the energy use of roasting the turkey in an electric oven is about 2.0 kW times 3.5 hours, or 7.0 kWh.

The latest data available from the Department of Energy says that for every kWh of electricity produced in the United States, 1.096 pounds of CO2 are released. Thus for this recipe in an electric oven, the CO2 emissions are equal to 1.096 lbs/kWh times 7.0 kWh or about 7.7 pounds of CO2.

Gas oven:

Gas ovens use about 0.112 therms of natural gas per hour. Over the course of the 3 hours 30 minutes, this would result in the use of 0.392 therms. In order to convert this amount of natural gas to kWh for comparison’s sake, we use the energy equivalence of one therm being about 29.3 kWh, meaning the energy use of a gas oven for this recipe is 11.5 kWh.

The Environmental Protection Agency (EPA) has a handy carbon footprint calculator you can use to analyze the CO2 emissions of all sorts of household activities. Included among its assumptions is the emission factor of cooking with natural gas, at 11.7 lbs of CO2 per therm of natural gas (this is another place where your specific situation may vary– some gas stoves use propane or other flammable gases as fuels, but we’ll assume natural gas for the sake of this calculation). Based on this assumption, the roasted turkey recipe in a gas oven would result in CO2 emissions of about 4.6 lbs of CO2.

Braised turkey

The braised turkey recipe also uses a oven, but this time for 15 minutes of preheating and 3 hours 30 minutes of cooking for a total of 3 hours 45 minutes. Again, this process can be done in an electric or a gas oven using the same assumptions as the roasted turkey.

Source

Electric oven:

Using the same assumptions as above for 3 hours 45 minutes of 2.0 kW power usage, the braised turkey recipe uses 7.5 kWh. Using the same assumption of 1.096 lbs of CO2 per kWh results in the CO2 emissions of the braised turkey in an electric oven being about 8.2 lbs.

Gas oven:

Repeating the assumptions above again gives an approximate energy use of 0.420 therms, or 12.3 kWh, and would result in emissions of about 4.9 lbs of CO2.

Deep fried turkey

The deep frying recipe calls for a propane heater to preheat a pot of oil to 250oF, adding in the turkey and raising the temperature to 350oF, and then cooking for 40 minutes.
 

The assumptions we can make here are that a propane cooker uses 65,000 British thermal units (BTUs) per hour and preheating deep fryers takes about 30 minutes. That means the total energy use would be 65,000 BTU/hour times 1 hour 10 minutes for a total of 75,833 BTU. Converting the propane use in BTU to approximate energy use in kWh gives a final result of approximately 22.2 kWh.

To calculate the CO2 emissions from this cooking process, the EPA’s carbon footprint calculator again gives us the needed information of CO2 emissions for cooking by propane. With the EPA assumption that every million BTU of propane burned emits 136.05 lbs of CO2, the deep fried turkey’s 75,833 BTU emits about 10.3 lbs of CO2.

Grilled turkey

Charcoal grill:

When the grilled turkey recipe for a charcoal grill is used, 110 charcoal briquettes are used over the course of 3 hours (after 30 minutes of pre-burn of charcoal).

Experiment shows that the energy content of charcoal is 7.33 kilojoules (kJ) per gram, while a single briquette of charcoal weighs about 25.7 grams. All together, this means a charcoal grilled turkey takes 20,733 kJ, which is converted to about 5.8 kWh.

For the CO2 emissions of charcoal grilling, Oak Ridge National Laboratory has found that the amount of charcoal needed to operate a grill for an hour emits 11 pounds of CO2. For this recipe that uses the grill for a total of 3 hours 30 minutes, that amounts to 38.5 pounds of CO2.
Propane grill:

When prepared on a gas grill, propane is needed to preheat for about 15 minutes and then cook the turkey for 3 hours.

The rate of propane use in propane grills varies, but a standard gas grill is rated at about 36,000 BTU/hour. That means for the full 3 hour 15 minute operation, the Butterball grilled turkey recipe requires 117,000 BTU or approximately 34.3 kWh of energy.
 
As with the recipe for deep fried turkey, we can use EPA’s assumption that every million BTU of propane burned emits 136.05 lbs of CO2, meaning this propane grilled turkey accounts for 15.9 lbs of CO2.

Smoked turkey

For the smoked turkey recipe, we again have two options for cooking fuel– either a charcoal fueled water smoker or an all electric smoker.

Charcoal powered water smoker:
This recipe required the burning of 10 pounds plus 70 briquettes of charcoal for 8 hours (after 15 minutes of preheating).
Using the same assumptions as with the charcoal grilled turkey, we find that at 7.33 kJ/gram of charcoal and 25.7 grams of charcoal per briquette gives a total energy use of the charcoal for a turkey smoked with a water smoker of about 12.9 kWh.
For the CO2 emitted, we again assume that grilling for an hour emits 11 pounds of CO2 per hour, meaning for a total grill time of 8 hours 15 minutes we get 90.8 lbs of CO2.
Electric smoker:
The electric smoker will be set at 225oF and the turkey cooked for 11 hours. Common electric smokers are rated at about 800 Watts, meaning 11 hours of use would use 8.8 kWh.
Since this is all electric, we can reuse our assumptions from cooking in an electric oven. Assuming 1.096 lbs of CO2 are released for every kWh of electricity produced in the United States, the electric smoker would account for about 9.6 lbs of CO2.

Spatchcocked turkey

The recipe for spatchcocked turkey brings us back to the oven, but with the distinct (and intentional) advantage of a greatly reduced cooking time compared with the other methods. The 15 pound turkey will cook in the oven at 450oF for 85 minutes, after 15 minutes of preheating, for a total oven use time of 1 hour 40 minutes.
Electric oven:
Reusing our electric oven assumptions, 1 hour and 40 minutes of 2.0 kW power usage means the spatchcocked turkey will require about 3.3 kWh of energy. At 1.096 lbs of CO2 per kWh, that means the recipe accounts for about 3.6 lbs of CO2.
Gas oven:
If instead the spatchcocked turkey is cooked in a gas oven, which uses 0.112 therms of gas per hour, the energy use of this recipe would be about 5.5 kWh, while the CO2 emissions associated with this cooking process would be 2.2 lbs.

Sous vide turkey

Last but not least is the sous vide turkey, which requires the use of an immersion sous vide immersion circulator for 2 hours 30 minutes (after a 15 minute preheat time). Given that the power rating of a sous vide from Williams Sonoma (the source of our recipe) is 1,100 W and the total operating time is 2 hours 45 minutes, the electricity use comes out to about 3.0 kWh. At 1.096 lbs of CO2 per kWh, that means the sous vide turkey accounts for about 3.3 lbs of CO2.

Graphical results and conclusions

With all those calculations and assumptions out of the way, we can finally look at all the results in one table:

Click to enlarge

These numbers can also be displayed graphically to show the overall level of ‘green-ness’ of each cooking method:

Click to enlarge

Looking at these results, there are a number of points of interest and interesting conclusions to draw:
  •  In terms of the amount of CO2 emissions, the two options that use charcoal (smoked in a charcoal smoker and grilled on a charcoal grill) are by far the greatest emitters. This result shouldn’t be surprising, as charcoal (with anthracite coal as one of its ingredients) is one of the more carbon intensive fuels you can use in your homes. However it is interesting to note that, despite their higher CO2 emissions, they are in the same ballpark in terms of energy use as the other cooking methods. This result shows how charcoal is an efficient fuel source, it just happens to also be dirty.
  • In terms of the total energy use, the two options that use propane (deep fried and grilled on propane grill) require the greatest energy. The higher energy needed is likely due to the cooking source being less efficient than others, with gas/propane burners typically being only 40% efficient with the remaining 60% of energy output being lost to heating the surrounding air or as visible light.
  • The two best cooking methods in terms of both minimal energy use and CO2 emissions are the sous vide turkey and the spatchcocked turkey (in either a gas or electric oven). The reason these reign supreme is telling, and different for the two of them.
    • For the sous vide turkey, the turkey is vacuum sealed and cooked in heated water the size of a typical pot. The result is that a smaller volume has to be heated up when compared with a larger oven, deep fryer, smoker, or grill that needs to heat up and keep heated the larger surrounding area. By focusing the heat in a smaller area, the total energy use is greatly reduced. In all cooking, the smaller the area you are heating up the more energy efficient the cooking process will be, which is why it is actually advisable to cook using smaller, dedicated appliances (e.g., toaster ovens, panini press, etc.) than to use the oven or stovetop for everything.
    • For the spatchcocked turkey, the reduced energy use and associated CO2 emissions is simply attributed to the largely reduced cooking time. Outside of the deep fried recipe, which uses the aforementioned inefficient propane, the spatchcocked recipe is the only one that takes under two hours of cooking. Obviously, the less time you have to have your appliances working, the less energy you’ll use. So while spatchcocking may have become popular due to the convenience of reduced cooking time, the relative efficiency is also among its virtues.
  • When comparing the recipes that use either the gas oven or the electric oven, the final figures show that the gas ovens use more energy but emit less CO2. What is important to note about the CO2 difference, however, is that the numbers are based on the average U.S. figure for CO2 emitted per kWh. This number can vary greatly depending on your power company and where you live. For example, if you live in Vermont then your power likely comes from a greater proportion of renewable energy than in other states, which would reduce the relative CO2 emissions of your electric oven. Of course the opposite is true if your power company uses more coal in its fuel mix than the national average.
  • One last point is that the relative energy use here does not correlate to the relative cost to the consumer for preparing the turkey. Certain fuel types are much cheaper than others, which is part of the reason they are popular to use in the first place. For example, just because grilling by propane uses almost six times the energy as grilling by charcoal, the relative prices of the fuels actually results in grilling by gas being less costly per hour for a consumer.

According to the National Turkey Federation, 46 million turkeys are roasted each Thanksgiving. Various outlets have attempted to estimate the actual energy use of those turkeys cooked in aggregate, with answers ranging from 48 million kWh to 792 million kWh (quite a wide range, showing just how uncertain the true number is). Using the numbers calculated here, if all 46 million turkeys were cooked sous vide then that would be 138 million kWh, whereas if they were all grilled on a propane grill then that would be over 1.5 billion kWh. Concerning CO2 emissions, the 46 million turkeys could account for 152 million lbs (sous vide) or over 4 billion lbs (grilled on charcoal grill)– for context, a passenger vehicle emits about 10,000 lbs of CO2 per year. That’s all to say, the small decisions everybody makes individually can add up to make a large difference in total energy use or CO2 emitted– even when talking turkey.

In the end, though, there isn’t too much reason for you to stress. There are plenty of methods you can use to cut down on energy use while cooking if you choose to do so(see some examples here and here, or you can even invest in a solar cooker that uses just the sun and reflectors to cook at temperatures up to 400oF!). But again, the overall energy use on Thanksgiving is lower than the average Thursday. It’s a time to relax and be grateful, not necessarily to measure out your exact briquettes to minimize energy spent. But you can come to the Thanksgiving table with some of these fun facts handy to impress your family, just be sure to praise the cooking of the chef first– he or she spent plenty of time making that dinner!

Have a happy Thanksgiving!

Sources and additional reading

About the author: Matt Chester is an energy analyst in Washington DC, studied engineering and science & technology policy at the University of Virginia, and operates this blog and website to share news, insights, and advice in the fields of energy policy, energy technology, and more. For more quick hits in addition to posts on this blog, follow him on Twitter @ChesterEnergy.  

What is the most climate friendly way to light your Jack-O’-Lantern?

The truest sign to me that Autumn has arrived isn’t the changing of the leaves, the advent of sweater weather, or even the pumpkin spice lattes everywhere you look. The real sign of the Fall season in my life is when the seasonal sections of Target fills up with Halloween costumes, decorations, candy, and trinkets. On my recent trip to this holiday mecca, I was looking at the decorations– specifically the little lights that are meant to go into jack-o’-lanterns instead of candles– and realized the sustainability factor for these decorations has not become nearly as pervasive as it has for Christmas lights (which now commonly advertise how efficient they are on the front of the package). After a bit of research, it appeared that this topic had not garnered any real investigations. Being ever the energy-conscious consumer, I could not let that stand!

What follows is some ‘back of the envelope’ type number crunching to figure out the most efficient and green option for illuminating your carved pumpkins. Very specific data is not really available, so there are certainly some liberties taken. However just for the sake of finding ballpark answers, I’ll hope this slight lack of statistical rigor is found acceptable. But if the Senate and Natural Resources Committee is looking to tackle the issue, then this will be a good starting point.



Background

One of the main differences between Christmas lights and Halloween pumpkin lights that changes how the market approaches them is surely that Christmas lights get plugged into an outlet. Families with large Christmas light displays will see a noticeable bump in their monthly power bill, making the efficiency of these lights more present in the forefront of their minds. However, jack-o’-lanterns are instead lit up with either candles or lights that use portable, disposable batteries. Not only does this fact (and the relative smallness of pumpkin lights compared with full house Christmas lights) reduce the necessity of efficiency to most people, but it also makes efficiency calculations more difficult to come by. The power of the lights come either from the candle itself or from portable batteries, the comparison becomes fairly difficult.

But wait!

The choice of what to light your pumpkin up with is tied pretty strongly with a debate that arises every Earth Hour (an event organized where everyone turns off their lights for one full hour to symbolically support climate change and energy reduction efforts)– and that debate is whether candles, as a form of fossil fuels, actually end up emitting more carbon dioxide (CO2) than the electricity to power light bulbs do. Without jumping too deep into that issue, the point is that while candles do not require any electricity, they do release CO2 into the atmosphere (depending on the specific type of candle).

That being the case, it seems that comparing the CO2 emissions tied with various jack-o’-lantern lighting sources might be the easiest and most digestible exercise in determining the ‘greenest’ pumpkin lighting method.

Jack-O-‘Lantern Lighting Options

Real candles

  • Traditionally, jack-o’-lanterns were lit up exclusively by candles. The idea of candles in jack-o’-lanterns is so ingrained in people’s minds that the artificial lights often include a ‘flicker’ to mimic the actual look of candles. Because of this, the baseline lighting source is the traditional and widely-available paraffin candle. These candles are found at virtually any store that sells candles, and are created from a by-product of the refining of crude oil (hence their CO2 emissions).

  • As people became more environment- and health-conscious, alternative types of candles that did not emit the pollutants of traditional paraffin candles became more popular. So a second alternative are more the more eco-friendly soy or beeswax candles.

Artificial lights

(for these I’ll find a sample of flameless artificial lights that are readily available on Amazon.com and cover a variety of options for the batteries that power them)

Calculations

Again I just want to stress that what’s about to take place are rough calculations that should not be taken as 100% accurate, but rather to gain a general idea of the scale of CO2 emissions for each of these pumpkin lighting sources (wow it’s hard to try and sound scientific and serious while typing that phrase…). With that said, here’s a look at the back of the envelope on which these calculations were done:

Real Candles

Paraffin candles
For paraffin candles, considered the standard and classic candle with which to light up a jack-o’-lantern, a number of sources cite a figure of about 10 grams of CO2 released per hour of candle burn, so that is the number we will go with. For these candles, we’ll also ignore the CO2 emissions associated with the production and transportation of the candles because 1) paraffin is a by-product of various petroleum refining processes, meaning if not used then the material would go into the waste stream, 2) the low cost of the product suggests that the energy used to produce and transport them (called embodied energy) will be relatively low compared with the tangible CO2 released in burning, and 3) data for such questions is not readily available.

Paraffin candle CO2 emissions: 10 grams of CO2 per hour of candle burn

Beeswax/soy candles
On the other hand, beeswax or soy candles (the touted green alternative) are often considered carbon-neutral. This assumption is made despite the fact that they do also release CO2 when they are burned, as the released CO2 was recently absorbed by plants in the atmosphere (which was transferred by a bee to the beeswax used in those candles, or was still in the soy used for soy candles). In these instances, common practice is to not count such CO2 emissions, as they used CO2 that was in the atmosphere and will cyclically release it back, as opposed to fossil fuels (such as those in paraffin) that are releasing CO2 that had long been stored in oil reserves underground. We’ll again ignore the CO2 emissions associated with the production and transportation of the candles because 1) beeswax and soy plants are both renewable sources for material, 2) the low cost of the products suggests the embodied energy, and thus associated emissions, are relatively low (especially if these candles are bought locally, as they are commonly found at farmer’s markets and the like), and 3) we don’t have such data available.

Beeswax/soy candle CO2 emissions: 0 grams of CO2 per hour of candle burn

Artificial Lights

Each of the three artificial light options found, their equivalent CO2 emitted per hour of use will be calculated based on the batteries required to run them. Making the comparison this way will require a number of generous assumptions (back of the envelope here, don’t forget!):
  • The associated CO2 we’ll look at is only coming from the batteries used to power the light, not the construction or transportation of the light itself. Again, the data to find the CO2 associated with producing/transporting the light is not easy to find– but moreso, we’ll assume that the lights will be used year after year, thus minimizing how much CO2 per hour would end up being.
  • The California government sponsored a study on the emissions associated with producing alkaline batteries, one of the conclusions of which was that the CO2-equivalent produced for primary batteries was about 9 kilograms (kg) per kg of battery produced. This figure assumes that batteries are single use (either thrown in the trash or recycled after use) and accounts for the energy needed to store power in the batteries that will eventually add a sparkle to the eye of your jack-o’-lantern. We’ll use this number, combined with the weight of the batteries for the lights and the lifetime of that battery, to find the CO2 per hour of use associated with the lights.
  • Assumptions will be made on how long the batteries will last in these lights, using either the product’s page or a best guess based on battery capacity and typical drain.
  • Additionally, we’re assuming the use of the typical disposable batteries– any rechargeable batteries would throw off the calculation, but this analysis won’t go there.

Combine these artificial lights with the real candle options, and the final values for the five options in terms of associated CO2 released per hour of jack-o’-lantern operation is as follows:

Or in graphical form:Click to enlarge.

Very obviously, the environmentally friendly soy or beeswax candles that account for no CO2 release in their production or burning are going to come out on top here. But what might surprise you is that the options that use batteries come out significantly ahead of the typical paraffin candle. While the industrial production of the batteries to power the artificial lights (and even if you add in a fudge factor to account for the production of the artificial light itself) seems like it would obviously be energy-intensive and account for greenhouse gases, the less obvious fact of paraffin candles outdoing that by direct emissions is not as clear until you look into the numbers specifically.

Conclusion

In the end, this might come across as a silly exercise– and maybe it is, just in the name of holiday fun. Releasing 10 grams per hour with paraffin candles compared with the significant reductions possible with the other options might seem like small potatoes in the grand scheme of things– in a world where a single cow can release up to 200 grams of methane (a greenhouse gas that contributes more strongly to climate change per gram of it released than CO2) through flatulence and belches, why even question the CO2 released due to halloween decorations?
I would agree that’s a fair point, but let’s keep the calculations going really quickly. In the United States, there are about 73 million children under the age of 18. Let’s just say that half of those kids have a jack-o’-lantern (some families might not celebrate Halloween with jack-o’-lanterns, some families might not see the need for one for each child, but on the other hand many adults such as myself might still find joy in carving and lighting up pumpkins– so 50% will be our randomly chosen number. Back of the envelope!). And let’s say that for the two weeks leading up to Halloween, those jack-o’-lanterns are lit up for 3 hours per night. All of a sudden, we’re dealing with 1,533 million total hours of jack-o’-lanterns being burnt. If all of those jack-o’-lanterns are releasing 10 grams of CO2 per hour with paraffin candles, all of a sudden that’s 15,330 metric tons of CO2– or the equivalent of the annual CO2 released in a year by over 3,000 cars.
The point of all of this to show how much of a difference small changes can make. Are you an environmental criminal for lighting your pumpkin up with a paraffin candle? Certainly not. But you can be an environmental warrior by noting all these small choices that surround you (during the holidays and in your everyday life). And if you want to add some more energy-efficient related fun to your jack-o’-lanterns, check out these stencils from the Department of Energy!

Sources and additional reading

About the author: Matt Chester is an energy analyst in Washington DC, studied engineering and science & technology policy at the University of Virginia, and operates this blog and website to share news, insights, and advice in the fields of energy policy, energy technology, and more. For more quick hits in addition to posts on this blog, follow him on Twitter @ChesterEnergy.  

Correlating Energy Data Sets: The Right Way and the Wrong Way

Determining the correlation between multiple sets of data—a measure of whether data sets fluctuate with one another—is one of the most useful tools of statistical analysis. Correlating data sets can be the endgame itself, or it can be what cracks open the door on a full statistical investigation to determine the how and why of the correlation. No matter the reason, knowing what data correlation is, how to correlate data sets, what a confirmed correlation might mean are all necessary ideas to have in your tool belt.

 What is data correlation?

Generally speaking, correlation examines and quantifies the relationship between two variables, or sets of data. In statistics, data correlation is typically measured by the Pearson correlation coefficient (PCC), which ranges from -1 to +1. Whether the PCC is positive or negative indicates whether the relationship is a positive correlation (i.e., as one variable increases, the other variable generally increases as well) or a negative correlation (i.e., as one variable increases, the other variable generally decreases). The absolute value of the PCC indicates the strength of the relationship, where the closer it is to 1 the more strongly related the two variables are, while a PCC of 0 indicates no relationship whatsoever.

Source

 

How do you calculate data correlations?

The PCC of two variables can be easily calculated with a built-in function of Microsoft Excel (if you want to know how to calculate the PCC according to hand—first, kudos to you, scholar; second, see either this resource or this one for more detailed instructions on the calculation itself).



To start, list out your two variables in two columns of an excel sheet. For this example, we’ll pull the West Texas Intermediate (WTI) oil prices and the U.S. regular grade gasoline prices during a four-month period in the Fall of 2016 from the website for the Energy Information Administration (EIA) (for guidance on pulling data from EIA, see this previous blog post).

Link to Gasoline Price Data; Link to WTI Spot Price data

Note that the weekly prices here reflect the average price calculated for the week ending in the date listed. Also the Cushing, OK WTI spot price reflects the price of raw crude oil in Cushing, OK, a major trading hub for crude oil that is used as the price settlement point for WTI oil on the New York Mercantile Exchange (NYMEX).

Now to find the PCC, use the excel function CORREL. This function takes the form of the following:

=CORREL(ARRAY1, ARRAY2)

where ARRAY1 and ARRAY2 are the two data sets you are seeking to correlate.

Using this excel function, we get a PCC of 0.545. Remember that a positive PCC indicates that the two arrays tend to increase with each other,and that the closer the PCC is to 1 then the more closely related they are. This result of 0.545 would seem to indicate a fairly decent correlation between the price of WTI oil and the price of regular gasoline over these several months. Not only does a positive correlation between the prices of these two products make intuitive sense (because the price of crude oil is the largest factor in the retail price of gasoline), but we can confirm with a data visualization as well.

:

Note that the first graph is showing the change in the two prices over time, with the date on the x-axis and the prices on the two y-axes. Visualizing the data this way, we can see that the prices are climbing and falling somewhat step-in-step. The second graph shows the relationship in a different way, with the price of oil on a given week on the x-axis and the price of gas on the same week on the y-axis. Visualizing it this way, and including a trendline for that data, you again see that as one variable rises, generally so too does the other variable. However, clearly it isn’t a direct one-to-one relationship—hence why the calculated PCC is 0.5455 and not closer to 1.

As a second example, let’s now find the correlation between gas prices during this same time period with the quantity of finished motor gasoline supplied to the market—as basic economic principles give us a sense that there should be a relationship between quantity sold and price. Below we again pull the relevant data sets from EIA and use the CORREL function

Link to Gasoline Price Data; Link to Gasoline Supplied Data

Note that the weekly prices here reflect the average price calculated for the week ending in the date listed.

For these two variables, we get a PCC of -0.173. Now that the PCC is negative, this implies a negative correlation—i.e., as the gasoline price increases, the amount of gasoline sold decreases. This conclusion again makes a degree of intuitive economic sense, as when the price of something increases ,the expected consumer response would be to purchase less of it. However, with PCC so far from -1 we don’t necessarily see this as a very strong correlation. We can look at the data visualization for these data sets as well:

Looking at the first graph, we can again see visually what the PCC was indicating in general. As the gasoline price reaches local peaks, the amount of product supplied tends to reach local valleys, and vice versa. The second graph indicates that with a negative trend line, though again it’s overall just a slight, general trend and not very rigid—as indicated by the PCC being closer to 0 than it is to -1.

There’s a data correlation—what now?

So the key to answering what happens next is to know why you were looking for a data correlation in the first place. Let’s say I was examining the correlation between gas prices and oil prices because I wanted to identify the factors that best predicted gas prices going forward. For each of the two variables tested with gas prices over the four month period in 2016, the expected generally correlation was confirmed with the data, though the PCC wasn’t strong enough to definitively declare victory at having found a correlation. What would I do in this scenario?

More data

The first course of action would be to gather more information. I’ve only looked at 16 weeks of data, but it has been enough to give me a correlative hypothesis (increased gas prices correlate with increased WTI oil price and decreased gasoline supplied). You might take this hypothesis and expand your analysis to include more historical data and see if the same correlation holds or if it moves in a different direction. Further, you might reason out that there are more subtle interactions of between the data that should be explored. Perhaps looking at the price of gas and the price of oil during the same week is too simplistic, and rather you should be looking at the price of oil compared with the price of gas the following week, two weeks, or even month to account for the time needed to refine crude oil into gasoline? Or if your goal is to really find the most influential correlating factors, then it would go without saying to test many more variables to figure out the ones with the closest correlation. For gas prices, you might consider also looking at general economic data, import/export data, production and refining production data, drilling data, and much more.

Test further

Once you have exhausted the data you are looking at and determine what correlates well based on that data, it is important to make sure to test it as well to make sure any conclusions you make are based on sound correlation. As with any type of hypothesis, a correlation is essentially meaningless unless it gets tested.

A couple methods for testing the correlation are available. First, as mentioned previously, expand your data set and put the correlation to the test on a wider set of data—either by looking further in the past to see if the correlation persists, or by using the correlation as a predictive model for future data and seeing if the relationship holds when the new data becomes available.

If you have not already done so, creating a visual representation of data, as done for the two sets of variables above, is a great way to gain understanding of your correlation (and has numerous other advantages for taking in data). As you conduct your data detective work, be sure to always check yourself by creating graphs and other visualizations to confirm suspicions and/or catch some new insights. Whenever possible, as well, work with the data yourself instead of referencing the visualizations of others. In the worlds of data and statistics, it is notoriously easy to ‘make’ the data appear to say whatever you want to say to a lesser informed audience (stay tuned for a future post on this topic).

Another important ‘test’ of sorts is one we already implicitly did when selecting our examples in the previous section—reason out why a correlation might exist. For the prices of crude oil and finished motor gasoline, the reason behind a correlation is somewhat self-evident. But if you’re looking at variables that are less obviously linked, this is where you can do research or consult with experts to determine if there exists any logical rationale to explain the correlation. Otherwise you could be grasping at straws, despite the apparent correlation—discussed in more detail next.

Recognize limitations

Being aware of the limitations of correlating data is the best defense against falling victim to the shortcomings of the technique. This idea is best illustrated in another example.

Let’s say I was continuing the above effort to find factors that I could use in the future to predict gas prices. As discussed, the spot price of WTI oil, with a PCC of 0.545, is determined to be great candidate for correlation with a reasonable PCC, data visualizations that illuminate the relationship, and a very logical and rational reason for the two variables to be correlated. So if oil demand at a PCC of 0.545 is counted—then I should be excited when I stumble upon a mystery variable with a PCC of 0.592!

Link to Gasoline Price Data; Source of Mystery Variable (Spoiler Alert!)

Note—Mystery Variable had no available data for the week of November 7, 2016

With a PCC of 0.592, I could feel great that I have another factor to add to my model. Looking at the data visualizations below does nothing to dispel that notion, either.

The issue is, however, not realizing that if you wade through large enough sets of data you are virtually guaranteed to find coincidental correlations. In this example, I was able to find just such a coincidental correlative data set by looking through the only other vast set of data I spend as much (or sometimes, shamefully, more) time with than energy-related data—fantasy football! Yes, the mystery variable that appeared to correlate decently well with U.S. gas prices from September to December of 2016 was actually the standard fantasy points scored by Washington player, Chris Thompson (missing data for the week of November 7 was due to his bye week).

The man that correlates with gas prices

After revealing the actual source of my mystery variable, you would obviously have me pump my brakes on any correlation. There is no possible explanation for why these two variables would be correlated (unless perhaps you would like to make the argument that when the price of gas goes up, Chris Thompson drives less and walks to and from practice—thus improving his cardiovascular endurance and improving his performance that subsequent week; I unfortunately could find no information on his in-season transportation habits).

The fallacy of connecting my mystery variable to gas prices would almost certainly have been exposed were you to test the correlation through expanding the data set and logical reasoning, as previously discussed. Unfortunately, other factors will not always be so obvious to rule out—which is why having as large of data sets as possible is key. Even then, however, you are bound to stumble upon these coincidental correlations (for some thoroughly entertaining and statistically vigorous examples, check out the Spurious Correlations blog) when casting a wide enough net. That fact is just one of the quirky statistical truths with very large sets of data (if interested on this topic specifically, I’d highly recommend reading either or both of these two fabulous books: The Drunkard’s Walk: How Randomness Rules Our Lives & The Improbability Principle: Why Coincidences, Miracles, and Rare Events Happen Every Day)

Beyond that, even if the correlation might seem sound, keep in one of the firs things taught in introductory statistics, and also one of the first things forgotten, that correlation is not causation (credit to Thomas Sowell). So while our fantasy football to gas prices comparison is a false correlation, even a true correlation does not automatically let you leap to the conclusion that one variable must be causing the other– a topic that this section of the blog will assuredly revisit in a future post. For now, though, I’ll leave it to America’s favorite statistician to summarize:

“Most of you will have heard the maxim “correlation does not imply causation.” Just because two variables have a statistical relationship with each other does not mean that one is responsible for the other. For instance, ice cream sales and forest fires are correlated because both occur more often in the summer heat. But there is no causation; you don’t light a patch of the Montana brush on fire when you buy a pint of Haagan-Dazs.”
― Nate Silver, The Signal and the Noise: Why So Many Predictions Fail–but Some Don’t

 

About the author: Matt Chester is an energy analyst in Washington DC, studied engineering and science & technology policy at the University of Virginia, and operates this blog and website to share news, insights, and advice in the fields of energy policy, energy technology, and more. For more quick hits in addition to posts on this blog, follow him on Twitter @ChesterEnergy.  

Navigating the Vast EIA Data Sets

The Energy Information Administration (EIA) is an independent arm the Department of Energy (DOE) that is tasked with surveying, analyzing, and disseminating all forms of data regarding energy in the United States. Further, EIA is a politically isolated wing of the DOE– meaning it is there to provide independent and factual data and analysis, completely independent from the partisan decision makers in Washington or the political inclinations of those in charge of at the top of DOE. Because that is the case, you can be confident the data put out by EIA is not driven by any agenda or censored in favor of a desired conclusion.

Thus for anyone with even a passing interest in the national production and use of energy, EIA really is a treasure trove of valuable information. However, those who are unfamiliar with navigating the EIA resources can easily get overwhelmed by the vastness of the data at their fingertips. Additionally, even seasoned veterans of the federal energy landscape might find it difficult to find the exact piece of data for which they are digging within the various reports and data sets made publicly available on the EIA website. So regardless of your experience level, what follows is a brief guide to what type of information is available as well as some advice as to how to make the best use of your time surfing around EIA.gov.



Types of data available

One of the really fabulous things about the EIA data sets is that they cover every kind of energy you can imagine. The energy categories you can focus into include, but are not limited to, the following:

Within these energy categories, you can look at the trends of production, consumption, imports/exports, and carbon dioxide emissions going back years (oftentimes even decades) and also modeled as a forecast into the coming years. Most data sets will have tools to automatically manipulate the data to change between units (e.g., total barrels of oil vs. barrels of oil per day), or even manipulate data trends (e.g., go from weekly data to 4-week moving averages to 10-year seasonal averages). Depending on the type of data, these numbers are regularly updated weekly, monthly, and/or yearly. If there’s a topic of particular interest, there’s a good chance there’s a report with the data on it being released at regular intervals– some of the more prominent reports are highlighted below.

Regularly updated reports

EIA releases a regular stream of reports that serve to update the publicly available data at given intervals. Some of the more prominent reports are listed below, and they are typically used to update all of the energy categories previously mentioned:

  • The Monthly Energy Review (MER) is a fairly comprehensive report on energy statistics, both from the past month and historically back a number of decades. Published during the last week of every month, the MER includes data on national energy production, consumption, and trade across petroleum, natural gas, coal, electricity, nuclear, renewables– as well as energy prices, carbon dioxide emissions, and international petroleum.
  • The Short-Term Energy Outlook (STEO) is another monthly EIA report, this one released on the first Tuesday following the first Thursday of the month. The STEO includes data on much the same topics as the MER, with the inclusion of some international energy data, and it also includes monthly and yearly projections for the rest of the current year and all of  the next year based on EIA’s predictive models. The inclusions of these forecasts makes for particularly useful data sets for anyone who might be trying to stay a step ahead of the energy markets. Also of particular interest for statistically-minded people out there is a regular comparison of numbers between the current STEO forecast and the previous month’s forecast. These comparisons show which way the model shows data to be trending, with the more significant ones called out in the report and noted with reasoning behind the changes.
  • The Annual Energy Outlook (AEO), like the STEO, provides modeled projections of energy markets– though the AEO focuses just on U.S. energy markets, models these annual forecasts long-term through the year 2050, and is released every January. The other aspect of the AEO that makes it particularly interesting is that its modeled forecasts, in addition to a reference case forecast, include different assumptions on economic, political, and technological conditions and calculate how those various assumptions might affect the outlook. For example, the 2017 AEO includes projections based on high economic growth vs. low economic growth, high oil price vs. low oil price, high investment in oil and gas resources and technology vs. low investment, and a projection that assumes a complete roll-back of the Clean Power Plan.
  • The International Energy Outlook (IEO) provides forecast energy market data consistent with the AEO, but regarding the international energy market through 2040.
    • With forecasts in both the STEO and the AEO, an understanding of exactly what is meant by the forecasts is imperative. The forecasts and projections do not necessarily reflect what a human prognosticator within EIA thinks could, should, or will happen– rather it demonstrates what the predictive models calculate given the best possible and unbiased inputs available. This difference is a subtle one, but if you ever find yourself questioning “does the person behind this report really think this is going to happen?”, recognize that some nuance exists and the reason you are skeptical might have not yet been able to be statistically included in the model.
  • The State Energy Data System (SEDS) is published once annually and breaks down national energy use, price, spending, and production by sector and by individual states. Within each of these categories, you can also break down the data by energy type (e.g., coal vs. natural gas) and by primary energy use vs. electric power generation. Having this granularity is useful to further dig into if certain energy trends are regional, restricted to certain climates, or are in response to specific state policies.

While they are not necessarily releasing new and specific data on a regular basis, two other EIA articles of note are worth pointing out because of the interesting stories and analyses they tell:

  • Today in Energy (TIE) comes out every weekday and gives a quick and readable article with energy news, analyses, and updates designed to educate the audience on the relevant energy issues. TIE frequently features graphs and charts that elegantly demonstrate the data in an easy to understand but also vastly elucidating way. One of the real advantages to reading TIE each day, though, is they often include tidbits from all the previously mentioned regularly updated reports, as well as other major releases or EIA conferences, enabling you to keep up with the newest information from EIA (click here for a post on the best TIE articles of 2017 to get you started).
  • This Week in Petroleum (TWIP) is an article that comes out every Wednesday that is very similar to the TIE articles, but focuses on the world of petroleum specifically and provides crucial insights on topics such as drilling, oil company investments, retail prices, inventories, transportation of crude and refined petroleum products, and more.

If any of these regular reports are of interest to you, you can sign up to get email alerts anytime these (or a number of other) reports are released by EIA by visiting this page. If you don’t know which reports you’d want but you want to keep an eye on what EIA is putting out, you can also simply subscribe to the “This Week at EIA” list that will once a week send you an email to notify you of ALL the new EIA productions from that week.

Finding specific data

While keeping up with all the regular reports from EIA is immensely useful, what brings many people to the EIA website is the search for a specific piece of data. You might want to see a history of average gasoline prices in a certain region of the country, find the projection of how much solar capacity is expected to be added in the next few years, track how much petroleum product is being refined in the Gulf Coast, or countless other facts and figures. Below you’ll find a few strategies you can employ to track down the information you seek.

Navigating the menus

EIA.gov has a useful menu interface through which you can usually navigate to your desired dataset easily.

Source: Homepage of EIA.gov
  • The “Sources & Uses” drop down will be where you can navigate to data sets about specific fuel sources and energy use;
  • The “Topics” drop down highlights the analysis on data by EIA as well as economic and environmental data; and
  • The “Geography” drop down is where you can navigate data by state or look at international data.
Source: Homepage of EIA.gov

Navigating from these menus is fairly self-explanatory, but let’s walk through the example of finding the recent history of gasoline prices in the Gulf Coast region of the United States. Gasoline is a petroleum product, so we would click on “Petroleum & Other Liquids” under the “Sources & Uses” menu.

Once on the “Petroleum & Other Liquids” page, the information we’re interested in would be under the data menu with the “Prices” link.

Source: Landing page for EIA.gov/petroleum

You’ll then see a listing of various regular releases of petroleum product price reports and data sets. Since we’re interested in Gulf Coast gasoline prices, we’ll click the third link for “Weekly retail gasoline and on-highway diesel prices.”

Source: EIA’s Petroleum and Other Liquids Prices

Clicking on this report will bring up the below interactive table. The default view will be to show U.S. prices averaged weekly. The time frame can be adjusted to monthly or annual prices (we’ll keep it at weekly). The location of the prices can be changed to allow viewing of data by region of the country or by select states and cities (we’ll change it to the Gulf Coast). The interactive table then displays the most recent week’s data as well as the previous five weeks (note: for ‘gas prices’ as is most often reported in the media and related to people filling up the gas tanks in their cars, we’re interested in the row titled ‘Regular’).

Source: EIA’s Weekly Retail Gasoline and Diesel Prices

If you’re interested in going further back in time then shown in the interactive table, the ‘View History’ links can be clicked to bring up an interactive table and graph going as far back as EIA has data (1992, in this case), shown below. Alternatively, if you want to have the raw data to manipulate yourself in Microsoft Excel, then click the ‘Download Series History’ link in the upper left (I’ll download and keep this data, perhaps handy for later in this post).

Source: EIA’s Weekly Gulf Coast Regular All Formulations Retail Gasoline Prices

Note in the above interactive chart there is the built-in abilities to view history by weekly/monthly/annual data, to download the source data, or the adjust the data to be a moving average or seasonal analysis.

If you find a page with the type of information you’ll want to reference regularly or check in on the data as they update, be sure to bookmark the URL for quick access!

STEO Custom Table Builder

Another useful tool is the STEO Custom Table Builder, which can be found here. The Custom Table Builder allows you to find all of the data that is included in the monthly STEO report (e.g., U.S. and international prices, production, and consumption for petroleum products, natural gas, electricity, coal, and renewable energy; CO2 emission data based on source fuel and sector; imports and exports of energy commodities; U.S. climate and economic data broken down by region; and more). This data can be tracked back to 1997 or projected forward two years on a monthly, quarterly, or annual basis. All you need to do is go to the Custom Table Builder, shown below, and select the options you wish to display.

Source: EIA’s Custom Table Builder

As an example, let’s use the STEO Custom Table Builder to determine the projected of how much solar power capacity in the near term. Solar would fall under the ‘U.S. Renewable Energy’ category, so click to expand that category, then expand the ‘Renewable Energy Capacity,’ and you’ll see the STEO has data for data for the capacity of large-scale solar for power generation, large-scale solar for other sectors, and small-scale solar for other sectors.

Source: EIA’s Custom Table Builder

Select all the data relevant to solar data, select the years you want (we’ll look at 2017 thus far through the end of 2018), and what frequency you want the data (we’ll look at monthly). Then hit submit, and the following will be the custom table built for you.

Source: EIA’s Custom Table Builder

Note: The forecast data is indicated in the Custom Table Builder with the numbers shown in italics. The above data was pulled before the September 2017 STEO was published, so the projections begin with the month of August 2017.

For this example, we’ll want to then download all the data to excel so the total solar capacity can be added up and analyzed. Click the ‘Download to Excel’ button at the upper right to get the raw data, and with a few minutes in Microsoft Excel you can get the below chart:

Source of Data: EIA.gov, pulled on September 10, 2017

This graph, made strictly from STEO Custom Table Builder data, shows the following:

  • As of July 2017, large-scale solar generation capacity was only 0.3 GW outside of the power sector and 23.7 GW, while small-scale solar generation capacity was 14.8 GW.
  • Together, solar power capacity in the United States added up to 39.1 GW as of July 2017.
  • By the end of 2018, total solar power capacity is projected to rise to 53.7 GW (an increase of 14.5 GW, or 37%), according to the EIA’s August 2017 STEO.

Search function

Using a search bar on some websites can be surprisingly frustrating, but luckily the EIA search function is very accurate and useful. So, I have found that, when in doubt, simply doing a search on EIA.gov is the best option.

Perhaps I want to track the amount of petroleum products in production on the Gulf Coast. This information is not in the STEO report, so the Custom Table Builder won’t be of use. And maybe I don’t immediately see how to navigate to this specific information on the menus. I would type into the search bar the data I’m seeking as specific as possible—‘weekly gulf coast refiner gasoline production’:

Source: Homepage of EIA.gov

Doing the above search yields the below results, of which the first one looks like just what we need.

Source: EIA.gov

Click on that first link, and ta-da! We’re taken to the weekly gasoline refinery report for the Gulf Coast (referred to as PADD 3). Again, you see the options here to look at the history back to 1994 both on a weekly and a 4-week average basis, use the chart tools to analyze moving averages or seasonal analyses, or download the data to utilize in your own way.

Source: Weekly Gulf Coast Refiner and Blender Net Production of Conventional Motor Gasoline

Contact experts

As a last resort, the EIA website offers resources to contact should you have questions or issues navigating the data. The people behind the EIA data are civil servants who are intelligent and very dedicated to their job and making sure you get the accurate and relevant information you need. So in a pinch, head to the Contact Us page and find the topic on which you need help from a subject matter expert.

If you want an alternative to going straight to the people at EIA, however, feel free to contact me as well and I’d be happy to try and help you track down information on EIA.gov as well. Use any of the contact methods mentioned in the Contact Page of this site, or leave a comment on this post.

Using the data

I have found that it is not at all an exaggeration to say that the world (of energy data, at least) is at your fingertips with EIA’s publicly available data. To demonstrate, I’ll walk through a quick example of what you can find.

If we take the previously gathered weekly data for Gulf Coast gasoline prices and gasoline production, we can plot them on the same graph:

Source of Data: EIA.gov, pulled on September 10, 2017

By taking advantage of the publicly data on EIA’s website, we can notice some trends on our own. In the above, there is a drastic increase in Gulf Coast gasoline prices, coincident with a large decrease in Gulf Coast refiner production of gasoline that bucks the month-long trend of production generally increasing. This is a curious change and would prompt investigation as to the reason why. Luckily, several of EIA’s Today in Energy articles already points out this trend and offers explanation—all related to the effects of Hurricane Harvey on the Gulf Coast petroleum systems (Article 1, Article 2, Article 3). Just goes to show that one of the best way to stay abreast of trends and information in the energy world is to follow EIA’s various reports and analyses.

 

Updated on September 28, 2017

 

 

 

About the author: Matt Chester is an energy analyst in Washington DC, studied engineering and science & technology policy at the University of Virginia, and operates this blog and website to share news, insights, and advice in the fields of energy policy, energy technology, and more. For more quick hits in addition to posts on this blog, follow him on Twitter @ChesterEnergy.