Tag Archives: electricity

Super Bowl Sunday and Electricity Demand: What Happens in Cities with Super Bowl Teams and Host Cities?

The Super Bowl is upon us once again, along with all the fun sideshows that go with it. There are few events in American culture that bring as many people collectively around a single event quite like the Super Bowl, with even non-football fans gorging on fatty foods, enjoying the commercials, and relishing in the excuse to attend parties on a Sunday evening. The grip that Super Bowl Sunday has on our group consciousness allows for some interesting analysis of data (how much food do we collectively eat?) and myths (despite what I heard on the schoolyard growing up, the simultaneous flushing of toilets during halftime has not actually caused damage to sewer systems).

Thinking of the ‘everyone flushing the toilet at the same time’ myth got me to wondering about how electricity demand as a whole is affected by the Super Bowl, particularly in the regions whose teams made the big game (and presumably cause even more of the population to tune it) and the region hosting the Super Bowl. Indeed, grid operators like ISO New England recognize that ‘even when the game is thousands of miles away, the Super Bowl can have a big impact on regional electricity demand with spikes and dips throughout the game,’ requiring them to monitor the demand closely throughout the day.

So just as I started this football season analyzing the sustainability-ranking of each team, I’ll end it by analyzing the championship game in energy terms. Going into this analysis, I expected that a city/region having their team in the Super Bowl, or hosting the festivities, would lead to a definitive increase in power demand– but keep reading to see why I was surprised to find that assumption was misguided.



Graphical Results

We’ll jump right into the graphical results of this analysis– if you’re interested in reading the methodology, head down to the Methodology section now. The methodology section will also answer where the data came from and why for Super Bowls from 2015 and earlier there isn’t data available for each of the three relevant cities (two participant teams and the host city).

Super Bowl 51

Starting with the most recent Super Bowl and working backwards, first up is Super Bowl 51. This game saw the New England Patriots defeat the Atlanta Falcons in Houston, Texas, in the largest comeback in Super Bowl history. Below is the graph of electricity use in the power regions that are home to Boston, Atlanta, and Houston compared with a typical Sunday of comparable weather (note that all times displayed in this and other graphs are in Eastern Standard Time even when the region in question is in a different timezone):

For this specific Super Bowl, the electricity demand in all three regions is mostly lower than a normal Sunday for the whole day, though the demand of the Atlanta fans drops even lower than normal come game time and we see the New England electricity demand increase compared with normal as the games continues. As will be discussed later, this difference in how the two cities reacted over the course of the game likely reflects the attitudes and activities each fan-base had to what was looking like a blowout victory for Atlanta.

Super Bowl 50

Super Bowl 50 featured the Denver Broncos defeating the Carolina Panthers in Santa Clara, California. Comparing the electricity use this Super Bowl Sunday with a typical winter Sunday in the power regions that contain Denver, Charlotte, and Santa Clara gives the following visual:

In what we’ll find is a more typical effect of Super Bowl Sunday, the electricity use in both Denver and Santa Clara saw an increase from normal use early in the day, only to fall below average during the time when the game was on. Panthers fans, however, set an unparalleled increase in power demand compared with a normal Sunday all day, but especially high during the afternoon lead up to the game and notably dropping during the game.

Super Bowl 49

Working backwards, Super Bowl 49 is the first instance where we find that data is not available for all three regions (see Methodology section for an explanation). In this game, which found the New England Patriots defeating the Seattle Seahawks in Glendale, AZ on a game-ending interception in the end zone, we only have data from the New England power region to consider:

In looking at the New England electricity demand, we find a peak compared with normal early in the day and a general increase compared with normal over the course of the game.

Super Bowl 48

For Super Bowl 48, where the Seattle Seahawks dominated the Denver Broncos all game in East Rutherford, NJ, the only available data was for the power system that is home to East Rutherford.

Here we again find a peak in electricity demand compared with normal early in the day, which dissipates and eventually leads to lower electricity used during the actual playing of the Super Bowl compared with a normal day.

Super Bowl 47

Last but not least is Super Bowl 47, featuring the Baltimore Ravens defeating the San Francisco 49ers in New Orleans, LA. I was particularly looking forward to gathering the data from this one (and was disappointed to find only the Baltimore area data available) because this is the game that infamously featured a power outage in the stadium that delayed the game by over half an hour. I was hoping specifically for the New Orleans data to see what the electricity demand looked like before and after the blackout, but it was not meant to be.

However we can see from the Baltimore data a peak in electricity use compared with normal early in the day and a distinct drop off as the game is set to begin and throughout the course of the game. Because the data provided is hourly, it’s not clear if there was any effect during the half hour delay in the Super Bowl, but it looks like people in Baltimore continued whatever it was they were doing during the power outage in New Orleans, rather than decide to use the break in action to start up the dishwasher or the clothes dryer.

Conclusions

General trends

Interestingly, we don’t find one iron-clad trend that weaves its way through the entire data set analyzed, though there are some patterns.

  • For regions with teams in the Super Bowl, four out of six (Baltimore in 2013, New England in 2015, Denver in 2016, and Carolina in 2016) of them show an increase in electricity use during the lead up to the game, while four out of six (Baltimore in 2013, Denver in 2016, New England in 2017, Atlanta in 2017)  of them show a decrease in electricity use during the game.
  • For the regions hosting the Super Bowl, a similar trend is found. Two out of three host regions (East Rutherford in 2014 and Santa Clara in 2016) showed an increased electricity demand in the hours preceding the game, while all three host regions showed a general decrease in electricity demand during the game.

While these data are not complete or detailed enough to make definitive conclusions (in addition to the lack of more years of historical data, the issue of controlling for the weather is difficult to do since some of the wider regions will have more varied temperatures throughout the region and make it more difficult to ensure the weather is not causing electricity fluctuations as a whole), they do generally follow the results of U.S.-wide studies. A study by Outlier found, through working with utilities during Super Bowl 46, the following:

More specifically, versus a typical Sunday afternoon/evening in the winter, home power usage was 5 percent lower during the Super Bowl, with big consequences for overall energy use:

Source

Going further, ISO New England’s minute-by-minute graphical analysis during Super Bowls 49 and 50 show the types of effects the big moments like the start, halftime, and end of the game have on the total demand load (and also serve to solidify that the effects are more pronounced when a region’s local team is in the game!)

Source

 

Explanation of the trends

The conclusion of less electricity usage over the course of the Super Bowl may sound surprising at first, given that it’s an event centered around an electronic device in the TV, but when you break it down it really makes sense. While it’s true that Americans gather around the television, they are often doing so collectively– going to parties or bars. So while the Super Bowl is often uncontested as the most watched television program of the year, that does not necessarily lead to an increase in the number of television sets being powered as people congregate around TVs together. These effects are going to be even more drastic if a local team is in the game (drawing in the more casual viewer) or if the game is being played locally (meaning more people will be in or near the stadium to enjoy the festivities).

Further, just because people are turning on their TVs does not indicate that household energy use is going up. That is because TVs require less than 400 Watts (W), and sometimes as few as 20 W, compared with most energy-sucking appliances like the vacuum (650 W), washing machine (2,500 W), or water heater (4,000 W). During the Super Bowl when the TVs are on, households are significantly less likely to be using these more electricity-consumptive appliances (not to mention many households would regularly have their TVs on during these hours anyway) and thus overall electricity demand noticeably drops.

That combination of people gathering together as opposed to being in separate households and using TVs instead of other appliances satisfyingly explains the drop in power use during the game. We could also conjecture that power use goes up before the game as people are getting the energy intensive chores (washing clothes, vacuuming, washing dishes, etc.) done earlier in the day before heading out to their Super Bowl gathering. They might also be preparing food to enjoy during the big game using their ovens/microwaves/stove tops in these early afternoon hours when they would not normally be in the kitchen.

Exceptions to the trends

Though the previously discussed trends were found in a majority of the cases analyzed in this article, there were a couple that bucked the trend. Specifically, analyzing the electricity use on Super Bowl Sunday compared with a comparable Sunday found that:

  • In 2017, both New England and Atlanta, as well as host city Houston, had lower than normal electricity demand in the hours before the game;
  • In 2016, the Carolina region saw large peak in electricity use compared with normal in the afternoon leading up to the game; and
  • In 2015, New England had increased electricity demand the morning of the Super Bowl as well as during the game.

There are a number of potential reasons that these specific instances did not meet the trends found in other places. The main one could be that while the average temperature used to find a comparable Sunday was close to the temperature on Super Bowl Sunday, there could have been wildly varying temperatures in different parts of the region or in different times of the day that prompted heating or cooling systems to be ramped up. Without the availability hourly temperature data and/or the analysis of temperature data of many cities within a region, it is impossible to know for sure. Further, grid operators also monitor aspects of weather like dew point, precipitation, cloud cover, and wind to predict electricity demand– which would be significantly more difficult for me to control for here. So for aberrations outside of the expected trends, these type of weather effects are the most likely culprit.

Another interesting explanation to look for is how captivating a particular game might have been. In its analysis of Super Bowl energy numbers, MISO notes that the more captivated and the more glued to their seats watchers are during the game, the more the demand will remain steady and low. As soon as people start to get up and do other things in the house (either because its halftime or a game is uninteresting), they notice a real uptick in electricity demand. This effect could perhaps explain why electricity use started to go closer to normal levels in New England in 2017 when the Patriots were building a seemingly insurmountable deficit, and it could also explain why electricity demand started to increase compared with normal in Carolina in 2016 about midway through the game (while never down by more than 10 until the closing minutes, more casual Panthers fans might have been frustrated with their team’s lackluster offense and inability to score more than 7 points through the third quarter and tuned out to partake in more energy-intensive activities).

Methodology

Availability of data

The availability of a region’s electricity demand depends on the entities who deliver energy and how far back in time you are looking. At the suggestion of the Federal Energy Regulatory Commission (FERC), a number of regional transmission organizations (RTOs) and independent system operators (ISOs) have been established in the United States to coordinate, control, and monitor complex and sometimes multi-state grid systems. One of the results of the use of these systems is they often make hourly electricity demand data publicly available going back a number of years, which allows for us to look back on some of the regions of the participants/hosts of the Super Bowl. The cities/years where those data are available are shown in the table below.

For regions that are not a part of RTOs or ISOs, unfortunately the electric companies rarely make public the same type of data. However a proxy we can utilize the Energy Information Administration’s (EIA) Electric System Operating Data tool. While it only goes back to the summer of 2015, it does provide the same type of hourly electricity demand data for regions and utilities outside of RTOs/ISOs. So where needed, this data is used as well as indicated in the below table.

When going back to Super Bowl 49 and earlier, some data become unavailable and those cities are not included in the analysis, shown below as ‘not available.’

 

For links to each of the electricity data sources listed in the above table, go to the ‘Sources and additional reading‘ section.

Finding a reasonable day for comparison

To determine the changes in electricity demand that are attributed to each region for Super Bowl analyzed, a reasonable day for comparison was found in each region using the following criteria:

  • As pointed out in the previous post analyzing electricity usage during a federal government shutdown, nothing will affect a region’s power demand more than the weather. If all buildings and homes are turning up either the air conditioning or the heat, that will have a greater effect on electricity usage than anything else– even an event as large as the Super Bowl. With the goal of comparing electricity demand on Super Bowl Sunday with other days and controlling for other factors, the methodology used was to assure that the comparison day chosen had an average temperature as close as possible to the average temperature on the day of the Super Bowl in that region. As a rough proxy, the average temperature on the day of the Super Bowl in the major city associated with each team/region was found on Weather Underground, and the goal was to find a comparison day with an average temperature within a few degrees Fahrenheit;
  • There are also distinct patterns to electricity demand depending on the day of the week, so the comparable day chosen was always made to be a Sunday; and
  •  Lastly, the comparable day chosen was kept to be within one to three weeks of the Super Bowl (either before or after), while avoiding any Sunday that had a playoff football game for the region’s home team, to assure any other externalities are kept as constant as possible.

With that criteria in mind, the following were the days used for comparison to Super Bowl Sunday in each region:

Click to enlarge

Graphical comparisons

Once the hourly data for each Super Bowl Sunday and chosen comparable dates were pulled, the hour-by-hour comparison is calculated using a simple percentage change from the regular non-Super Sunday. These percentages are what are ultimately graphed on an hourly basis, with the up to three regions (depending on how many available) on the same graph to see if there are any trends based on the cities. Similar comparison was not included for overall U.S. electricity trends because the large and varied geography of the United States makes controlling for the effects of weather on electricity demand much more complicated and difficult (however, as noted earlier, a study that looked at thousands of households during the 2012 Super Bowl found that an on overall basis, electricity demand increases on Super Bowl Sunday in the hours before the game and decreases once the game begins).

Sources and additional reading

5 Facts About Energy During the Big Game: MISO

Baltimore Gas & Electric: PJM RTO

California ISO: Pacific Gas & Electric electricity demand

Carolinas region electricity demand (EIA)

Energy Reliability Council of Texas (RTO) Coastal Region Electricity Data

How a Patriots Super Bowl affects the region’s power grid: ISO Newswire

How the Super Bowl saves energy: ABB

New England ISO Electricity Data

Northwestern region electricity demand (EIA)

Public Service Company of Colorado (EIA)

Public Service Electric & Gas Company: PJM RTO

Regional Transmission Organizations (RTO)/Independent System Operators (ISO): FERC

Southeastern region electricity demand (EIA)

Why people use less energy on Super Bowl Sunday: Washington Post

 

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.  

Federal Government Shutdown: Analyzing Electricity Demand When Government Workers Get Furloughed in Washington DC

In a dance that’s become a bit too commonplace in the federal government, threats of a government shutdown over political differences and budget issues are looming once again. After multiple continuing resolutions agreed to between Democrats and Republicans, the latest deadline for appropriation bills to fund the government is fast approaching. While a potential government shutdown would put my 9-5 job on hold until a resolution was reached, a frustrating prospect for all families who rely upon paychecks from their government jobs, there’s not much to do for those of us outside of the White House and Congress. What I can do with that nervousness, though, is ask energy-related questions!

The fact that energy and electricity use changes at regular intervals throughout the day and week is well established, and these trends are reliably correlated with the day of the week, time of the day, and weather. Knowing this led me to the question of how a government shutdown would effect the electricity demand in the Washington DC area, where over 14% of the workforce is made up of federal employees. Would a government shutdown lead to an electricity demand closer to a typical weekend day than a weekday because of the large amount of people who would no longer be reporting for work? Would the overall electricity demand go up or down? Is any of this even noticeable, given that about 86% of the workforce would be going to work as normal? We are only four years removed from the last federal government shutdown, so looking at the electricity demand surrounding the 2013 shutdown can provide some insight as to what might happen if there is a shutdown this time around.



Background

The 2013 federal government shutdown lasted from October 1 through October 16, with President Obama signing a bill to reopen the government shortly after midnight on October 17. The political football at stake in 2013 was the Affordable Care Act, as Republicans in Congress sought to defund the program while the Democrats refused to pass funding bills that would do so. As a result, nearly 800,000 non-essential federal employees across the country were out of work without pay, while about 1.3 million essential employees reported to work as normal (though they saw their paychecks delayed). At the heart of the potential 2018 shutdown is the political debate surrounding immigration policy, though the effects on government workers would likely be largely the same as in 2013.

Source

While these numbers account for the vast amount of federal employees furloughed outside of Washington DC (such as employees in National Parks across the country), they still included a large number of DC residents. Further, employees of government contractors were reportedly sent home and furloughed without pay as well, though the data surrounding exactly how many government contracts were affected is unclear. So while there are other metropolitan areas that have a larger percentage of their workforce employed by the federal government, the prominence of federal contractor workers in DC still makes it an obvious choice for examining how the electricity demand changed in the wake of the 2013 federal government shutdown. More importantly, though, this analysis will focus on Washington DC because the data from the power companies is available in a sufficiently granular way for the region. The Potomac Electric Power Company, or PEPCO, is the electric power company that serves the entire city of Washington DC, as well as the surrounding communities in Maryland, so looking at PEPCO’s data over the shutdown dates will enable insights into the effect of the shutdown. Federal workers in other regions are typically served by much larger power companies (such as Dominion Energy in Virginia serving many of the Northern Virginia communities of federal workers in addition to the rest of Virginia and parts of North Carolina), making the potential effect on the power delivery data from the shutdown less significant on a relative scale.

Data and graphics

PJM, the regional transmission organization that coordinates the movement of wholesale electricity in 13 states and DC, makes available PEPCO’s metered electricity load data on an hourly basis. This type of data is available for most U.S. power companies, which is fun to play with to get an idea of how Americans behave during certain events like holidays, the Super Bowl, or any other large-scale event. In order to get a baseline of what the weekly electricity distributed by PEPCO, we can first look at the two weeks leading up to the government shutdown of 2013:


A couple trends become clear looking at these two seemingly normal weeks. First, the weekends (with Saturday and Sunday graphed using a dashed line instead of the solid line for weekdays) appear to have less electricity demand compared with weekdays. This trend is noted everywhere, not just DC, as weekends are when typical commerce activity drops. Additionally, there are clearly patterns of high and low electricity use by time of day, regardless of weekend or weekday. Demand appears to be at the lowest late at night and early in the morning when most people are sleeping, ramp up in the morning as people wake up to begin their day, and peaks around 5 PM when people are coming back home, making dinner, turning on the TV, putting laundry in the washing machine, etc. But did any of these trends change during the 2013 federal government shutdown? Here is the same data for the three calendar weeks during which the government was shut down:


When comparing these graphs with the two weeks prior, there does seem to be some noticeable differences– though the differences vary between the three weeks the shutdown was effective:

First Week

  • To start, the peak and cumulative power use appears to have increased a significant amount during the first week of the shutdown– though that could always be caused by the weather and a need to increase air conditioning or heating in a home. Indeed, looking at the temperature (discussed more later), the average temperature during the week climbed from about 66 degrees Fahrenheit the week before to about 73 degrees Fahrenheit. A possible explanation is the higher power use coming from people turning on their AC for the first time in a while due to unseasonably warm temperatures.
  • The overall ‘shape’ of the curves remain constant, so the furloughed employees and contractors did not appear to change their daily patterns enough to shift the timing of peak and minimum electricity loads.
  • Also interesting to note is that the Sunday before the shutdown (Sep. 29) stays lower than the weekdays, as was noted to be typical of weekend days, but the Saturday following the shutdown (Oct 5) then shifts to be among the days with the greatest electricity demand. I wasn’t expecting the furloughing of employees to have much of an effect on the weekend electricity demand, as most of the furloughed federal employees presumably did not typically work on weekends, but the answer can likely be attributed to weather as the weekend of Oct 5-6 had the warmest temperatures (79 and 80 degrees Fahrenheit, respectively) of the whole analysis period.

Second Week

  • The second week is the most anomalous of the three, with Sunday and Monday having the shape of the curve significantly affected and also having much higher peaks than the rest of the week (whereas the first week increased the peaks more comparably among the days of the week). In terms of why Sunday might have shifted so significantly, a search of what might have happened in Washington DC to cause this change on October 6, 2013 turned up an article about an explosion accident on the Metro. Perhaps the emergency response to this incident caused significant effects to the electricity demand?
  • Outside of Sunday and Monday, the peaks and shapes of the demand curves were back to being comparable to pre-shutdown levels. As will be shown shortly, though, this trend looks to be attributable to the returning of temperatures to an average of 65 degrees Fahrenheit.

Third Week

  • By the time of the third and final week of the shutdown, the electricity demand curve looks to be mostly back to normal. The last Sunday of the shutdown and the first Saturday after the shutdown look like normal weekend days, while the weekday curves look normal all week, even though the furloughed government employees and contractors did not head back to work until Thursday.

Just to be complete and ensure the trends we saw before and during the 2013 federal government shutdown were not just random week-to-week variations, below are the same graphs for the two weeks following the shutdown:

These two weeks show somewhat the same general trends we saw prior to the shutdown, with the main changes being that the peak demand for each day appears to be shifted to first thing in the morning when people are waking up and the morning of Saturday Oct 26 showing a higher peak than is typically expected of a weekend day. The peak electricity demand shifting to the morning likely comes from the weather getting colder (down to average temperatures of 53 and 59 degrees Fahrenheit, respectively), while the early peak electricity demand on Saturday Oct 26 might have been caused by a rally protesting mass surveillance that attracted thousands of people to Washington DC (though it too is likely in part due to the fact that it was the first day of the season where the average daily temperature dipped to 46 degrees Fahrenheit and people cranked the heat up when they woke up shivering that Saturday morning).

In addition to the demand curves, it’s important to look at the total daily electricity consumed by day over these previously discussed weeks, while also comparing these totals to the average daily temperatures in DC as I’ve done through the previous analysis:

As these two graphics demonstrate, the total electricity demand mostly moves step-in-step with the daily weather regardless of whether or not the federal government is open. If it gets too warm or too cold, that is when you see the spikes in electricity demand– and that will always be the most significant factor.

Conclusions

In the end, there does not appear to be a significant effect on Washington DC’s electricity demand during a federal government shutdown. While having thousands of employees and contractors stay at home is certainly not trivial, there are still even more government employees who would be deemed ‘essential’ and would be in the federal buildings (who would still be operating their heating/cooling systems). Beyond that, a vast majority of PEPCO customers are not in the federal workforce, so the change in daily habits of the unfortunately furloughed employees does not move the needle in a noticeable manner in terms of electricity demand. What’s more important to consider is the weather, and perhaps any daily events such as the Metro accident or the anti-surveillance rally. So while no one, especially in DC, is rooting for a federal government shutdown this week (the 2013 shutdown cost the country $24 billion and disrupted Veterans Affairs benefits from being sent out), we can take incredibly small solace that it won’t disrupt the expected electricity demand. Despite liquor sales increasing during the 2013 shutdown, the thousands of workers who would find themselves temporarily out of work would not have their change in daily routine threatening the electrical grid’s behavior.

If this type of data is of interest to you, by the way, the Energy Information Administration has an amazing tool that allows you to track electrical demand across the country in real-time. Are there any other events you think would be interesting to investigate for their effect on electricity demand? Let me know in the comments!

Sources and additional reading

Absolutely everything you need to know about how the government shutdown will work: Washington Post

Customer Base Line: When do you use the most electricity? Search for Energy

Demand for electricity changes through the day: Energy Information Administration

Democrats face make-or-break moment on shutdown, Dreamers: Politico

Electricity demand patterns matter for valuing electricity supply resources: Energy Information Administration

Electricity supply and demand for beginners

Everything You Need to Know About the Government-Shutdown Fight: New York Magazine

Here’s What Happened the Last Time the Government Shut Down: ABC News

How Many Federal Government Employees Are in Alexandria? Patch

Metered Load Data: PJM

U.S. Government Shutdown Looms Amid Immigration Battle: Reuters

Which Metro Area Has the Highest Share of Federal Employees? Hint: Not Washington: Government Executive

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.