The first half of the year was the hottest ever recorded in the contiguous US. July 2012 became the hottest July on record. And in communities across the country, more than 27,000 daily high-temperature records have been broken or tied so far this year.

Across the river from Opower headquarters, Washington DC faced 11 consecutive days that hit or exceeded 95 degrees…it was the city’s most intense heat wave in 141 years of keeping temperature records.

Thankfully, air conditioners keep us cool and comfortable during these sizzling summer days. More than 60% of US households now have central air conditioning, up from 23% in the late 1970s. But the luxury of A/C comes at an obvious cost: higher electricity consumption.  Exactly how much higher? We found that the average home’s electricity usage on a 103°-high summer day is up to 40% higher than during a typical summer day.

How do we know? We cracked open our data warehouse to examine anonymized energy usage data across 18,000 homes from 3 different cities in the western part of US, which has faced blistering temperatures this summer.  In particular, we explored a few key questions about energy consumption on hot days:

• What exactly happens to electricity usage as the outdoor temperature rises?
• When does home electricity usage reach a peak on hot days?
• How does a dramatic increase in air-conditioning affect the power grid and the risk of blackouts?

Our analysis shows that increased air-conditioning during the hottest days of summer 2012 caused residential electricity demand to shoot through the roof. And as we’ll see, when everyone tried to keep cool on the hottest afternoons, it put severe pressure on the regional power grid — in one case even causing the average wholesale price of electricity to double compared to a typical summer day.

Let’s dig into the data and start piecing together some energy impacts of this summer’s record-breaking temperatures…

Cranking up the A/C: home electricity consumption increases by 20-30% overall on the hottest summer days

The relationship between temperature and residential electricity usage is extremely strong, as we can see in the below graph. It displays average daily home electricity consumption on 45 summer weekdays (each dot represents a day), plotted against the average local temperature for each of those days.

The steady increase in home electricity usage as temperature goes up suggests that air conditioners are the driving force of the upward trend. The graph indicates that the typical home consumes 20-30% more electricity overall on the hottest days, relative to an average summer day (e.g. the days in the middle of the graph).

And as we’ll see, most of the increase in electricity usage on hot days takes place at a very specific time of day – a phenomenon that has major implications for the stability of the regional electricity grid.

In the late afternoon of the hottest days, home electricity consumption increases to 40% above average

A common pattern emerges in how home electricity usage fluctuates over the course of a hot summer day.

In particular, the “peak” (i.e. highest) demand in electricity usage comes somewhere between 3pm-7pm. That’s the time window when the day’s temperature is highest and also when many people are returning home from work or school. The line graph below shows average hour-by-hour home electricity usage during 4 different days in July and August. The temperature scale indicates the maximum temperature for each day.

Charting this data makes two things clear: 1) the hotter the day, the higher the electricity usage throughout the day, 2) on all these hot days, usage is highest between 3pm-7pm. Comparing a 103°-high day with an 83°-high day, we see that residential electricity demand during the late afternoon was nearly 40% higher on the hotter day. 

High A/C usage in the late afternoon means that 3-7pm on a hot day is effectively “rush hour” for electricity consumption.  But a big challenge looms during energy rush hour: how will power plants supply enough electricity to accommodate the huge increase in electricity demand?

Increased electricity demand on super-hot days means that more power plants need to be turned on, and that’s not cheap

To accommodate exceptionally high electricity usage on super-hot days, power companies typically obtain electricity from a system of backup power plants, called “peaker plants.” These plants, which otherwise sit idle, can be flipped on quickly to satisfy “peak” power demand (e.g. everyone blasting their A/C on a 100 degree day between 3-7pm).  However, electricity from peaker plants is expensive.

That’s because peaker plants are often old and inefficient, using gas-turbine or steam-turbine technologies that have low fuel efficiency. And as a result, flipping them on entails high operational costs that can be six times higher than a comparable power plant.  Peaker plants also have a poor environmental reputation — they are considered to be more polluting than similar power plants that operate throughout the year.

Just how expensive is it to operate peaker plants on the rare occasions when we need them? It’s been estimated that 10-20% of the overall cost of providing electricity each year in the US comes from just 100 hours of the year — largely corresponding to the highest-usage hours when air conditioners are relentlessly gobbling up electricity.

For the above reasons, when a larger share of electricity generation is coming from peaker plants on a particularly hot day, it becomes substantially more expensive for power companies to obtain electricity. This trend is readily apparent in the below bar graph. Although the price that most homes pay for their electricity stays constant, the wholesale price that their power company must pay for obtaining electricity on a super-hot day is double that on a typical summer day.

It’s clear that power companies pay a dear price for electricity from peaker plants.  And arguably, it’s a price worth paying: in ensuring that the supply of electricity keeps pace with electricity demand, peaker plants serve as an important last line of defense from power outages.

It turns out, however, that firing up peaker plants is not the only possible approach to managing peak electricity demand on hot days. Fortunately, there exists a cleaner and more cost-effective approach: reducing the need for peaker plants in the first place.

The cheapest and greenest kilowatt-hour of electricity…is the one that is never produced

If each household could reduce their peak electricity demand even slightly on hot days (e.g. by delaying laundry/dishwasher loads until after 7pm, or using window blinds to prevent unnecessary work for the air conditioner), the aggregate benefits for the electricity grid and households themselves could be enormous.

First and foremost, reducing peak demand relieves pressure on power plants, and so protects us all from power outages. But slimming down peak demand is also in the direct best interest of homes and businesses because it helps keep energy costs low: that’s because in the long term, many power companies are likely to incorporate the high cost of infrequent peaker-plant operations into customers’ everyday energy bills.

On a positive note, there are now often immediate cash benefits for customers who reduce their home electricity demand at peak times.  Many power companies have found that, instead of using expensive peaker plants, it is cheaper and more environmentally friendly to encourage customers — through incentives — to reduce peak demand in the first place. For example, a number of utilities actively reward customers who agree to slightly turn down their air-conditioner or electric water heater for a handful of short intervals on hot days.

As the summer of 2012 winds down, the number of days with extremely hot temperatures—and the intense air-conditioner usage that follows—may yet crank up further.  But if we can each trim down our usage just a tad during the hottest hours of these days, we’ll all benefit—lower energy bills, fewer outages, less pollution, and perhaps even extra cash payments from our power company.

Check out these easy-to-do tips, as simple as setting your thermostat to 78°, to get ideas on how to decrease your own peak electricity demand as the mercury rises.

Special thanks to David Moore, Ashley Sudney, and Yoni Ben-Meshulam.
–
Methodology:

For the purposes of comparability, days considered for this analysis are confined to weekdays between June 1-August 11, 2012 (excluding July 4) that were above 55° and did not have precipitation.  Results are based on averages across 18,000 anonymous households in 3 weather-comparable western US cities. Prices reported for wholesale electricity are Locational Marginal Prices in the day-ahead market for the 4-5pm interval, as reported by the California Independent System Operator. Weather data is sourced from Anything Weather and Weather Underground.

The estimated sixfold difference in operational costs between peaker plants and an average power plant is based on Monitoring Analytics (2012) and a related discussion in Alcott (2012). The former reports that the variable operation & maintenance cost of a combustion turbine peaker-plant facility is $7.59/MWh, which is approximately six times higher than the $1.25/MWh variable operation & maintenance cost associated with a combined-cycle gas power plant.

We looked at the correlation between email address and electricity usage across 2.8 million American households.  Of those, we focused on Yahoo Mail and Gmail because they were the most popular domains in our dataset. The email addresses correspond to the member of the household who manages the electricity bill.

We realize that Yahoo has had a rough stretch lately:

• 2 CEO’s fired within 8 months
• 14% of its workforce getting laid off 
• News that Yahoo Mail users have bad credit
• Implications that Yahoo Mail users are technological laggards 

Now we have more bad news for Yahoo…

We found that the average Yahoo Mail household uses 11% more electricity per year than a Gmail household.  It’s a sizeable, statistically significant difference (see Methodology).

Consider this: the aggregate difference in annual electricity use between 1 million Yahoo households and 1 million Gmail households is…equal to the entire annual electricity consumption of Barbados, a reasonably well-off country of 287,000 people.

Let’s dig into the numbers and find out what’s going on here…

The average Yahoo Mail household spends $110 more per year on electricity than a Gmail household.

The usage difference — 939 kilowatt-hours per year — is stark: Yahoo households consume almost a whole extra month of electricity relative to Gmail households. At an average going-rate of 11.8 cents/kWh, that’s a difference of $110 per year.

It’s as if, relative to the average Yahoo household, the average Gmailer is strictly hang-drying their laundry, forgoing high-definition TV, and hand-washing their dishes with cold water for a year.

We can establish off the bat that the “email-domain versus usage” relationship is one of correlation, not causation. So then, what underlying differences exist between Gmail and Yahoo users that can help explain their dramatic disparity in electricity usage?

Exploring reasons for the disparity between Gmailers’ and Yahoo users’ electricity consumption

There are some key demographic and lifestyle differences between Gmailers and Yahoo users that are likely to affect their household electricity usage.

We’re not the first researchers to think that Yahoo and Gmail users are different:

• Last year, Hunch.com discovered, among other things, that Yahoo users tend to lounge around in pajamas at home and enjoy sweet snacks; Gmail folks, on the other hand, are more likely to lounge in jeans and prefer salty snacks.
• The personal finance company Credit Karma has found that Yahoo users have noticeably lower credit scores (652) than Gmail users (682).
• Last week, the online-dating site Circl.es determined that Gmail users tend to have a higher “desirability quotient” than Yahoo users

But let’s explore some of the differences between Yahoo users and Gmailers that are likely at play in driving the disparity in their electricity consumption.

Do Yahoo users and Gmailers live in fundamentally different climates or use different fuels to heat their home?

No.  It’s true that climate and heating fuel vary significantly by region, and can have a large effect on energy consumption. For example, of the 1.8 million Michigan homes in our database, nearly 60% of them get through frigid winters by heating their homes with natural gas.  In contrast, the majority of the 780,000 North Carolinian homes in our database use electric heating systems (and they also face a milder climate).

But neither climate nor heating fuel should be expected to correlate meaningfully with email address domain here.

The reason is that the 1.15 million Gmail and Yahoo users in our dataset are spread out across 23 states and several distinct climate zones. There is no clear pattern that emerges wherein Gmail or Yahoo Mail is more popular in one particular state or climate zone than another.  This geographic spread of Gmail and YahooMail makes sense: anyone is free to sign up for either domain.

Do Yahoo users consume more electricity simply because they live in larger homes?

Yes and No. Yahoo users do appear to live in larger residences, which increases their total energy needs. But they also consume more electricity per square foot than Gmail users.

How do we know that Yahoo users tend to live in larger homes? Our friends at Experian and Hunch.com have found that Yahoo users tend to live in suburbs and rural areas, while Gmailers live in cities.  Data sourced from the 2009 US American Housing Survey suggest that suburban-rural residences (i.e. where the Yahoo Mailers live) are on average 7-13% larger than in cities (where Gmailers live), and also have more occupants (i.e. extra electricity-using human beings in Yahoo homes).

Our own household characteristics data similarly suggest that Yahoo users reside in larger residences: Yahoo households are approximately 10% more likely than Gmail households to live in single-family residences (as opposed to apartments and condos).

But, even controlling for home size, we found that Yahoo households are still more energy-intensive than Gmail households. Based on square footage data that we have for single-family residences, we found that the typical YahooMail household uses 12% more electricity per square foot of living space (6.84 kWh/sqft) than the typical Gmail household (6.09 kWh/sqft).

Do Gmailers and Yahoo users have different lifestyles?

Yes. Hunch.com and Experian have found that Gmailers are more likely to be younger, single people.  Credit Karma found the average Gmailer’s age to be 34, while the average Yahoo user clocked in at age 38.  Being young and single means going out more; less time at home – and fewer occupants – means less electricity usage.

By contrast, Yahoo users are more likely to be in relationships and have children. Additionally, Hunch found that Gmail users are more likely to be active travelers (having journeyed to 5 or more countries), and so might be away from home more often.

Our own data also suggest that Gmail users may have a greater interest in energy-efficiency. Among the approximately 10 million US households that have access to utility web-based energy-efficiency advice tools that Opower manages, Gmail users are 30% more likely than Yahoo users to sign up for an in-depth analysis of how they can reduce their energy usage.

It’s not definitive, but it appears that several lifestyle choices correlated with Gmail use are also correlated with lower home energy use.

What are Yahoo Mail users to do?

If Yahoo households want to slash their energy consumption to Gmail levels, it’s probably going to involve more than just switching to a Gmail account (nice try!). For the benefit of Yahoo users and Gmailers alike, we’ll aspire to sprinkle in energy-efficiency advice in future posts that will help all folks reduce their energy bills.

Update: Some commenters were curious about AOL and Hotmail. That got us curious too. Check out those results here.

Special thanks to my data-crunching partner Jillian Cairns and our all-star designer Efrat Levush. They are proud Gmail users.
–
Methodology: Annual electricity usage of households is based on 2011 data. Of the 2.8 million household electricity customers with email addresses in Opower’s dataset, 1.15 million were Gmail or Yahoo users.  The usage difference between Yahoo users and Gmail users is statistically significant at the 99% confidence level based on a t-distribution.

Barbados’ electricity consumption in 2008 was 945 million kWh. Dryer usage based on 1.89 kWh/cycle, derived from LBNL paper. TV usage based on 40″ Digital HD model from US DOE. Dishwasher usage based on 1 kWh/cycle (Energy Star) from NRDC.

Data Privacy: All data analyzed here are anonymous and treated in strict adherence to Opower’s Data Principles.

We looked at the correlation between email address and electricity usage across 2.8 million American households.  Of those, we focused on Yahoo Mail and Gmail because they were the most popular domains in our dataset. The email addresses correspond to the member of the household who manages the electricity bill.

We realize that Yahoo has had a rough stretch lately:

• 2 CEOs fired within 8 months
• 14% of its workforce getting laid off 
• News that Yahoo Mail users have bad credit
• Implications that Yahoo Mail users are technological laggards 

Now we have more bad news for Yahoo…

We found that the average Yahoo Mail household uses 11% more electricity per year than a Gmail household.  It’s a sizeable, statistically significant difference (see Methodology).

Consider this: the aggregate difference in annual electricity use between 1 million Yahoo households and 1 million Gmail households is…equal to the entire annual electricity consumption of Barbados, a reasonably well-off country of 287,000 people.

Let’s dig into the numbers and find out what’s going on here…

The average Yahoo Mail household spends $110 more per year on electricity than a Gmail household.

The usage difference — 939 kilowatt-hours per year — is stark: Yahoo households consume almost a whole extra month of electricity relative to Gmail households. At an average going-rate of 11.8 cents/kWh, that’s a difference of $110 per year.

It’s as if, relative to the average Yahoo household, the average Gmailer is strictly hang-drying their laundry, forgoing high-definition TV, and hand-washing their dishes with cold water for a year.

We can establish off the bat that the “email-domain versus usage” relationship is one of correlation, not causation. So then, what underlying differences exist between Gmail and Yahoo users that can help explain their dramatic disparity in electricity usage?

Exploring reasons for the disparity between Gmailers’ and Yahoo users’ electricity consumption

There are some key demographic and lifestyle differences between Gmailers and Yahoo users that are likely to affect their household electricity usage.

We’re not the first researchers to think that Yahoo and Gmail users are different:

• Last year, Hunch.com discovered, among other things, that Yahoo users tend to lounge around in pajamas at home and enjoy sweet snacks; Gmail folks, on the other hand, are more likely to lounge in jeans and prefer salty snacks.
• The personal finance company Credit Karma has found that Yahoo users have noticeably lower credit scores (652) than Gmail users (682).
• Last week, the online-dating site Circl.es determined that Gmail users tend to have a higher “desirability quotient” than Yahoo users

But let’s explore some of the differences between Yahoo users and Gmailers that are likely at play in driving the disparity in their electricity consumption.

Do Yahoo users and Gmailers live in fundamentally different climates or use different fuels to heat their home?

No.  It’s true that climate and heating fuel vary significantly by region, and can have a large effect on energy consumption. For example, of the 1.8 million Michigan homes in our database, nearly 60% of them get through frigid winters by heating their homes with natural gas.  In contrast, the majority of the 780,000 North Carolinian homes in our database use electric heating systems (and they also face a milder climate).

But neither climate nor heating fuel should be expected to correlate meaningfully with email address domain here.

The reason is that the 1.15 million Gmail and Yahoo users in our dataset are spread out across 23 states and several distinct climate zones. There is no clear pattern that emerges wherein Gmail or Yahoo Mail is more popular in one particular state or climate zone than another.  This geographic spread of Gmail and YahooMail makes sense: anyone is free to sign up for either domain.

Do Yahoo users consume more electricity simply because they live in larger homes?

Yes and No. Yahoo users do appear to live in larger residences, which increases their total energy needs. But they also consume more electricity per square foot than Gmail users.

How do we know that Yahoo users tend to live in larger homes? Our friends at Experian and Hunch.com have found that Yahoo users tend to live in suburbs and rural areas, while Gmailers live in cities.  Data sourced from the 2009 US American Housing Survey suggest that suburban-rural residences (i.e. where the Yahoo Mailers live) are on average 7-13% larger than in cities (where Gmailers live), and also have more occupants (i.e. extra electricity-using human beings in Yahoo homes).

Our own household characteristics data similarly suggest that Yahoo users reside in larger residences: Yahoo households are approximately 10% more likely than Gmail households to live in single-family residences (as opposed to apartments and condos).

But, even controlling for home size, we found that Yahoo households are still more energy-intensive than Gmail households. Based on square footage data that we have for single-family residences, we found that the typical YahooMail household uses 12% more electricity per square foot of living space (6.84 kWh/sqft) than the typical Gmail household (6.09 kWh/sqft).

Do Gmailers and Yahoo users have different lifestyles?

Yes. Hunch.com and Experian have found that Gmailers are more likely to be younger, single people.  Credit Karma found the average Gmailer’s age to be 34, while the average Yahoo user clocked in at age 38.  Being young and single means going out more; less time at home – and fewer occupants – means less electricity usage.

By contrast, Yahoo users are more likely to be in relationships and have children. Additionally, Hunch found that Gmail users are more likely to be active travelers (having journeyed to 5 or more countries), and so might be away from home more often.

Our own data also suggest that Gmail users may have a greater interest in energy-efficiency. Among the approximately 10 million US households that have access to utility web-based energy-efficiency advice tools that Opower manages, Gmail users are 30% more likely than Yahoo users to sign up for an in-depth analysis of how they can reduce their energy usage.

It’s not definitive, but it appears that several lifestyle choices correlated with Gmail use are also correlated with lower home energy use.

What are Yahoo Mail users to do?

If Yahoo households want to slash their energy consumption to Gmail levels, it’s probably going to involve more than just switching to a Gmail account (nice try!). For the benefit of Yahoo users and Gmailers alike, we’ll aspire to sprinkle in energy-efficiency advice in future posts that will help all folks reduce their energy bills.

Update: Some commenters were curious about AOL and Hotmail. That got us curious too. Check out those results here.

Special thanks to my data-crunching partner Jillian Cairns and our all-star designer Efrat Levush. They are proud Gmail users.
–
Methodology: Annual electricity usage of households is based on 2011 data. Of the 2.8 million household electricity customers with email addresses in Opower’s dataset, 1.15 million were Gmail or Yahoo users.  The usage difference between Yahoo users and Gmail users is statistically significant at the 99% confidence level based on a t-distribution.

Barbados’ electricity consumption in 2008 was 945 million kWh. Dryer usage based on 1.89 kWh/cycle, derived from LBNL paper. TV usage based on 40″ Digital HD model from US DOE. Dishwasher Usage based on 1 kWh/cycle (Energy Star) from NRDC.

Data Privacy: All data analyzed here are anonymous and treated in strict adherence to Opower’s Data Principles.