Businesses and higher education institutions are increasingly setting ambitious emissions reduction targets in response to the magnitude of the climate challenge. In 2008, Harvard set an aggressive emissions reduction goal that galvanized its community to move beyond the comfort zone in exploring opportunities for energy and emissions mitigation.

In addition to decarbonizing its energy supply and building more energy efficient facilities, Harvard’s Office for Sustainability (OFS) partners with faculty and students to envision and test new, innovative ideas and practices. In Spring 2016, OFS and its Campus Services partners in the Environmental Health and Safety and Energy and Facilities departments joined with Harvard Law School Emmett Environmental Law and Policy Clinic fellow and HLS alumnus Seth Hoedl to explore one such idea–the concept of shifting electricity consumption to times when the regional electric grid is less carbon intensive, thus resulting in lower emissions use.

Hoedl analyzed over 10 million data points from the historic emissions profile of New England’s regional electric grid (ISONE) and found that though the opportunity appeared to be not as significant as expected, shifting electricity use during less carbon intensive times could be a financially viable way to further reduce carbon emissions, especially as the grid continues to decarbonize.

Perhaps the greatest benefit of the project is a tool that Hoedl created that Harvard can use to continue to analyze opportunities as the emissions profile of the ISONE grid changes over time. As a result, the University will implement a “shadow” interval data greenhouse gas emissions inventory using this tool so it can continue to assess opportunities for making interventions to reduce emissions using this approach.

Read Seth's Executive Summary, Full Report, and Charts 


Office for Sustainability: How did you get involved with climate and emissions policy at Harvard?

Hoedl: I’m originally trained as a nuclear physicist and spent seven years as an academic at the University of Washington. I was looking for particles that probably don’t exist, so I got a little frustrated and decided to work on something practical. I joined a biotech startup for two years and helped them develop radioactive devices for cancer therapy. Those have now been used in about 10 or 15 people, and are saving lives, so that’s really satisfying.

After that, I came to Harvard Law School with the purpose of studying climate change—to gain skills and experiences to bridge divides between scientists, policymakers, and lawyers.

Here at Harvard, I specialized in energy and environmental law and policy, both through the Environmental Law Clinic and through coursework. As a student I also helped a little bit with the Office for Sustainability’s work as a member of the Complementary Mechanisms Advisory Group, a faculty, student, and staff group convened by President Faust to study and make recommendations on the use of off-campus emissions reduction projects to meet Harvard’s climate goal.

OFS: What was your initial motivation for this research project?

As I mentioned, I was a student member of the Complementary Mechanisms Advisory Group and we were looking at ways Harvard could reduce its CO2 emissions both with on-site and off-site methods.

During this process there was a thought floating around as to whether or not there was an easy way Harvard could lower its emissions simply by time-shifting its consumption of electricity—the Energy and Facilities group knew that this could result in cost savings (as electric rates differ at different times of day), but there was no data to support whether or not it would result in CO2 emission reductions. I asked if anyone had looked into it, and it turned out that nobody had. ISONE [the group that runs the New England electric grid] is a great region for testing this hypothesis because real-time data is available.

OFS: How have they developed such a robust dataset?

Hoedl: ISONE reports, on their website, the amount of electricity injected into the New England grid from all generators and nearby grids, categorized by generator type and import location.  For generators that burn multiple fuels, they ask generators to provide their fuel mix, basically their prediction for what their fuel mix will bewhich is an excellent proxy for what it actually is.  And they do that hour by hour, so that you can estimate the carbon intensity of the electric grid in New England hour by hour.

I took the hourly data from ISONE from December 4, 2014 through about the end of May 2016. You have a data point not just every hour but sometimes every five minutes so it’s something like one hundred thousand data points in time.  Because there were ten different generator types, in total, there were about a million data points to analyze. That sounds like a lot, but for a physicist it’s pretty small. I used the tools I used as a physicist to analyze it.

OFS: And what was the central hypothesis you wanted to explore?

Hoedl: For those who aren’t familiar, the electric grid is powered by a variety of different generating sources. In New England, you have a lot of nuclear power, there’s hydropower, there’s very little coal, and a lot of natural gas. There are also some imports from neighboring grids.

Those sources all have different carbon intensities, so they emit more or less carbon per megawatt hour produced than others. And they aren’t called on all at the same time. In the middle of the night the nuclear is providing a lot of our electricity and on a day like today—a very hot summer day—most electricity is provided by natural gas. That means that the carbon intensity of the electricity you take from the grid varies both by season and by hour.

The hypothesis was that with targeted investments Harvard could take advantage of this variation in the carbon intensity to lower its CO2 emissions.

OFS: What’s new about this question?

Hoedl: The thought’s been out there from a financial perspective to time consumption to minimize costs. There are groups out there that provide analysis looking at ways in which people can consume at different points in time to lower emissions, but I don’t think anybody in New England has done a systematic look at the historical record to see actually when the CO2 emissions were high or low, and use that as a basis for investment.

OFS: Is this similar to when people are told to use their washing machine on a hot summer day like today at night because that’s less stress on the system?

Hoedl: In the past people would say that because they wanted to lower peak demand, which is typically the most expensive power. It was also the case that it tended to be electricity with the highest carbon emissions.

What we found in this project is that that is no longer actually the case, at least here in ISONE.  Because natural gas is such a large component now of electricity generation, and because imports are so high, it actually isn’t necessarily true that changing your consumption from peak hours to non-peak hours makes that big of a difference [in CO2 emissions] anymore, which was a surprising result.

However, it’s very geographically dependent. What we found only applies to ISONE. Other grids will have very different characteristics depending on their generation mix.

OFS: And that was one of your most surprising results, that the gap was not as a big?

Hoedl: Yes, it’s not as big as we thought.

OFS: Moving on, what were some of your other top findings?

Hoedl: The first one is sort of what we were just saying. Just to put some numbers on it, the daily variation in the carbon intensity from the least carbon intensive hour, which is about four or five a.m., to the most carbon intensive, which is around four p.m., is only about 10 percent. 

So, if you moved all of your electricity consumption to four in the morning, you’d only reduce your carbon emissions by 10 percent. At least in ISONE, it doesn’t make a whole lot of sense to make a lot of effort in that way.

Another surprising feature was that if you switch to what we call an hour by hour accounting method – in other words if you calculate an entity’s emissions by multiplying its hourly consumption by the hourly carbon intensity and then summing up all the hours in the year instead of summing up all the electricity consumed and multiplying it by the average carbon intensity for the grid that yearthe numbers aren’t that different. This was surprising because we thought that since Harvard’s consumption patterns are different than the grid as a wholethey are tied to the academic calendarthat it would make a difference. I think it didn’t make a difference because even though there’s a strong daily variation, the seasonal variations aren’t that strong and so the difference between Harvard’s seasonal consumption and the grid as a whole washes out.

The other surprising thing is that there is a substantial amount of wood that is burned in New England to make our electricity, and whether or not one considers that carbon neutral makes a big difference on the carbon intensity of the grid. It’s about a 20% effect, so if you think wood is bio-neutral you’ll think the carbon intensity of the grid in ISONE is about 20% less than someone who thinks wood is not carbon neutral. That’s surprising because wood is only about 3% of grid generation but it’s so carbon intense that the 3% makes a big difference.  It’s probably somewhere in betweenthe forests don't grow back, the soil gets degraded but, on the other hand, the carbon that is in that wood originally came from the atmosphere.

OFS: Did you have an assumption on wood generation?

Hoedl: I analyzed it both ways. My initial assumption was that it is carbon neutral, following EPA’s guidance using their E-grid database. ISO-NE takes the opposite approach, they say that wood is not carbon neutral at all—it’s as if you are taking the coal out of the ground. Our analysis found that if you take it as carbon neutral there’s a greater daily variation in the carbon intensity than if you treat it as not carbon neutral. So if you say it’s not carbon neutral it motivates this time-shifting even less but whether that’s true or not I don’t know. People write their Ph.D. theses on this topic.

OFS: What challenges did you confront when doing this project?

Hoedl: There were some real challenges. There was some data that was hard to acquire. ISONE is very open and easy to access but ISONE takes about 20% of its electricity from NY and Canada, and neither of those grids provide quite the level of detail as ISONE.

It also took some time to estimate the uncertainties in the numbers in the report. I didn't want to just approach it as an accounting project, I wanted to actually get some confidence in the numbers I put out there. That took some time because you have to ask how confident you are in the emissions factors you receive from power plants in our grid, as well as Canada and New York.

In general, less is known about the carbon intensity of the grid than people suppose. Often, when people are debating whether or not they have achieved 5% reductions here and there, I wonder whether or not it makes sense to argue at that level because I don’t have confidence at the level of 5%.

Even though GHG accounting is approached from an accounting perspective, it’s really more a physics problem where at some point you have to step back and say we don’t know the emissions within 5% because there’s all these unknowns that aren’t worth our time to answer. You’d have to go to every plant at every minute of the day and measure their emissions to have better confidence.

OFS: How can your findings be applied to Harvard’s energy system specifically?

Hoedl: When entities like Harvard that are trying to make a difference in their GHG emissions are thinking about projects they may steer away from projects that include some type of electrification because they are worried increasing their electricity consumption can increase their emissions.

But we found that there are certainly times of the year where electrification is GHG positive—in other words the grid is sufficiently clean at those times that you are better off consuming electricity than burning natural gas. 

If you take advantage of those times you can increase electricity consumption and you won’t actually increase CO2 emissionsyou may not reduce it substantially by shifting to non-peak hours, but at least you can make an investment that increases your electricity consumption with some confidence you’re not increasing CO2 emissions.

The implication is that often these investments are financially favorable and you can use the savings to invest in projects that actually do have real GHG benefits that are much more significant per dollar than you would get from just time shifting. I think that applies to any entity looking to reduce its emissions.

However, these results only apply to ISONE. You’d have to redo the analysis for other grids for a broader recommendation.

I do want to note that there are people that look for the marginal carbon intensity of the grid. That’s asking the question: if you reduce your consumption of electricity how does the grid respond and what’s the CO2 emissions change due to that response. That’s a different question than we asked in our research because we were looking at it from an accounting perspective. Harvard has a GHG emissions goal and we were trying to study efforts to meet that goal. So that requires not the hypothetical question, but instead asking what was the carbon intensity when the electricity was consumed from the grid. Both are important but the conclusions you’d make may be different.

OFS: What’s an example of technology or equipment that could be used in this case?

Hoedl: We looked at two specific technologies. One is to displace a natural gas boiler with an electric boiler. If you run the electric boiler at those hours where the grid is less carbon intensive than burning natural gas then its GHG favorable. Those hours also tend to be those times when electricity is really cheap.

The other piece of equipment we analyzed is an electric chiller combined with thermal ice storage. The idea is that at the middle of the night you could essentially operate an air conditioner to make ice and that would displace the air conditioner in the afternoon when the electricity is more expensive and tends to be more carbon intensive. From a carbon perspective there’s a slight benefit of operating this device and from a cost perspective there’s a lot.

The third technology people often talk about is a heat pump. The ISONE grid is clean enough it makes sense to operate that essentially at all hours. If anything, our analysis shows that this is a more favorable investment to make.

OFS: Where there any other takeaways that are important to consider?

Hoedl: Even though we found that the hour by hour accounting method is about the same as an annual accounting method, I think it makes sense to do an hour by hour method. I structured my research to provide a set of tools that Harvard can use for years going forward to continue the analysis. I think this is important because it’s likely this hour by hour carbon intensity is going to change dramaticallyespecially if renewables become a larger fraction. It’s not inconceivable in summer months that there’ll be a lot of solar, so that, carbon low hours are not early morning, but in the middle of the day, which changes the dispatch of some of this equipment and change the investment decisions.

I think also it would be interesting to apply the analysis to other grids and compare how the investment decisions change. In markets where coal is baseload source and peaking plants are natural gas, you actually have a lower carbon intensity at peak hours. There you’d be better off operating equipment at the highest load point of the day.

OFS: What are some of the more promising levers you see for reducing emissions moving forward?

Hoedl: I think there’s a huge and often underappreciated role for solar at both the utility and consumer level.  Prices are falling so fast, about 30% a year, that analyses that were done just a few years ago are completely irrelevant. We don’t really know how long that decline will last but it’s not inconceivable that solar power will be the cheapest form of energy. That may turn a lot of the energy debates on their head.

Another big opportunity is to convert electricity into liquid fuels for either long-term storage or use in energy-intensive applications like aircraft and heavy-duty vehicles. That may have a big role in New England and other cold climates where you have to heat the buildings. So, why not burn something that you’ve created from electricity, presumably made with solar or wind? This technology is more speculative, but I bet in the long term it becomes a more critical part of our energy mix.

I’d be remiss as a nuclear physicist if I didn’t mention nuclear power. Nuclear power could be a large part of our energy system but it’s already in many contexts more expensive than solar on a per kilowatt basis. The biggest issue in my mind is the proliferation risk, which argues against using nuclear to solve climate change. The scale at which you’d have to use nuclear power to make a difference for climate change could increase the risk of proliferation of nuclear weapons.