Kudos and Caveats on The 2018 New Frontier Data Study

A document entitled “The 2018 Cannabis Energy Report”, published by New Frontier Data (NFD), is another effort to understand the role of cannabis in the broader energy picture. This is one of the first reports to summarize in the public domain the site-specific measured energy data (collected by the Resource Innovation Institute, RII) from actual grow facilities. It also makes a contribution by providing data on energy used in commercial greenhouses as distinct from the "windowless-warehouse" types of facilities, and estimates of energy uses associated with fully outdoor cultivation. The data provided also suggest that there is an enormous variation in energy use and a negligible correlation between energy use and yields -- suggesting significant inefficiencies in current practices. Kudos to the authors for estimating the relative contributions of legal and illicit activity, the latter being responsible for three-quarters of total energy use from cannabis production.

There are some broad commonalities between NFD’s estimates and those in my original 2012 study (Mills 2012). NFD estimates substantial energy use for producing cannabis in artificially conditioned indoor environments (and projects their number to more than double by 2025). They find that indoor-grown Cannabis results in hundreds of times its weight in emissions of the greenhouse-gas pollutant CO2 to the atmosphere. However, these values are far lower than those found in the 2012 study. While there may well have been some progress in energy efficiency in the intervening years (and much more being produced in somewhat less energy-intensive greenhouses), data are lacking with which to know with any confidence whether this is the case or not . For example, there are trends that could have driven energy use either up (e.g., larger space volumes to heat and cool) or down (e.g. more greenhouses, which average 25% lower energy intensity, per data in the NFD report). Meanwhile, annual cannabis production and consumption has increased due to the wave of legalization across the United States. A major study by latest work by the University of Colorado (Summers et al., 2021) arrived at range of carbon-footprints overlapping Mills (2012). Measured data by Leichleiter et al., (2018) are up to three-times higher.

I have published a more detailed comparison of the NFD estimates to others in the literature here.

The NFD study illustrates the many challenges and pitfalls encountered in producing such estimates. The 2012 study was based on “bottom-up” engineering estimates of cannabis production energy based on typical practices--validated by comparison to some actual facilities--while the NFD study admirably sought to utilize reported grid-based electricity use by a small number of actual producers. Both studies extrapolate estimates of energy per unit weight to aggregate national energy consumption based on volumes of cannabis produced. Measurement and modeling each have their strengths and weaknesses. Models can have errors or assumptions that do not reflect actual practice while raw data can also have errors as well as various forms of bias or poor characterization of the broader phenomena being characterized.

Opportunity for in-depth assessment of the NFD study is limited because the underlying sources and assumptions for much of the information are not documented, and much of the underlying data are deemed proprietary or otherwise not disclosed.

Multiple other factors confound comparison of the results of these studies.

The scope and "boundary conditions" of the NFD estimates are significantly narrower than used in Mills (2012), which, in turn, doesn't cover all potential emissions sources. While the 2012 study included all forms of energy used, including electricity (grid and off-grid), natural gas, propane, CO2 production, etc., NFD's estimates include only grid-based electricity, noting that this creates a particularly significant caveat in the case of greenhouses, which rely heavily on natural gas for heating. (The report’s title unfortunately uses the term “Energy” rather than “Electricity”, which will create the wrong impression for some readers.) Survey data shown in the NFD report indicate that non-grid sources of electricity and other forms of energy are used (but not measured or included in the analysis) at 11 out of 27 of the indoor sites, 11 of 13 greenhouse sites, and 15 of 17 of the outdoor sites. On-site diesel power generation is particularly common in the illicit market, and is associated with high CO2 emissions compared to most utility grids. Likely the most important caveat is the exclusion of direct fossil fuel use. As an indication of the importance, Canopy Growth, one of the largest cannabis producers (with over 10 million square feet in cultivation at the beginning of 2020), reported that only 32% of its total GHG emissions were from electricity use (Canopy Growth 2021).

A number of factors are baked into the NFD analysis that can be expected to result in underestimates of national electricity use and greenhouse-gas emissions. Some of these are appropriately noted in their report.

• The Mills (2012) study converted all major end uses to electricity equivalents, while the NFD study counted only metered electricity in actual sites.

• The NFD study relies on a small sample of sites providing measured data collected by RII. Only 24 sites report “productivity” intensity, i.e., electricity used per quantity of cannabis yielded for sale. This is the information needed to scale-up estimates of national energy use. The profile of this sub-sample (e.g., the use of energy-efficiency strategies) is not provided in the source document. Samples ranging from 57 to 81 sites are the source of ancillary data and metrics.

• Given the efficiency-focused nature of the data-collection efforts, contributors may be skewed towards those with better-than-average practices, and/or that sites self-selected for inclusion by contributors could be cherry-picked.

• As an example of the previous point, the surveyed sites seem to over-represent the use of LED lighting (12 of 59 sites, a whopping 20%), presumably resulting in lower energy use than average facilities. The nationwide use of LED lighting in cannabis grows was no doubt far lower, especially at that time given the high price points and high hurdle rates of cultivation-facility developers which rendered LED systems not cost-effective, together with ambiguity about their possible negative impact on yields.

• This small sample of convenience is not nationally representative, yet is extrapolated to create national energy and carbon estimates. Relatively mild climates such as those of Oregon and California dominate the sample Intense heating and cooling locations thus seem to be under-represented. Further confounding estimates of national greenhouse-gas emissions, these states also have much “cleaner” grids than the national average (lower levels of CO2 released per unit of electricity generated). Summers et al., (2020) show that the carbon footprint in those states is far lower than the geographic national average).

• Given that contributors to the RII database are presumably legal operators, the energy use for producing a given amount of cannabis appears to be assumed the same for illicit production. It is not demonstrated that these two types of operations are indeed equivalent, and their are reasons to believe illicit production is less energy-efficient.

• Some factors may understate CO2 emissions even more so than energy use. These include the use of off-grid generators that often have a significantly higher emissions factors (CO2/kWh) than blended grid electricity. The NFD study also did not estimate emissions from transportation energy. Energy used to manufacture CO2 for "carbon fertilization" (to enhance plant growth) was also not included.

There are some additional ambiguities:

• • There is a risk of intentional or inadvertent reporting errors in the underlying proprietary RII data used to characterize energy intensities for individual sites. Some of the reported values seem implausible, and, although their weight is reduced by employing averages of the sample, they exert a degree of influence on the aggregate results.

  • It is not clear whether the NFD study included cannabis drying/curing process energy (which is included in the Mills 2012 study).

  • It is not clear whether the sample was calibrated by strain choice, which can lead to a factor of two or more in energy intensity (Leichliter et al. 2018 and Backer et al. 2019).

  • Residential and other smaller-scale non-residential indoor production facilities for commercial cannabis production seem to be excluded (or if included were presumably assigned energy efficiencies of large, commercial operations). These smaller facility types were dominant in 2012, and their current role is no doubt still substantial.

  • While the energy used to grow products in legal facilities that are lost due to failed harvests may be counted, it is unlikely that the same holds true for illicit facilities.

  • It is not clear whether the energy use associated with interdicted illicit production or destroyed as the result of testing failures is included, versus only amounts making it to market.

Bottom Line:

The chart below compares The Mills, NFD, Summers et al., and Leichleiter et al., studies. The NFD study is a significant outlier in terms of its very high yield density (grams/m²-y) and correspondingly low energy intensity (kWh/gram of finished flower produced).

CAPTION: Indoor-grown cannabis energy intensity vs. yield intensity for four studies. Model estimates by Mills (4) and Summers et al., (18) with measured data from New Frontier Data (6) and a Leichleiter et al., (2018). The NFD study is an outlier in terms of its very high yield density (grams/m²-y) and correspondingly low energy intensity (kWh/gram of finished flower produced) which is twice that of a large meta-analysis by Backer et al., (8). The Leichleiter et al., sample is instructive in that for three genetic strains tested in side-by-side trials, LED lighting achieves 5%, 9%, and 31% lower energy intensity compared to less-efficient high-pressure sodium lighting, although strain choice is a far more significant determinant of the wide range of observed energy intensities. Heating fuel in Summers et al., is converted to electricity equivalents for comparative purposes, assuming an efficient heating COP of 3.25. Summers et al., include some emissions embodied in inputs and transport, but exclude black-market facilities. NFD results are electricity only (excluding fuels potentially used for heating). For Leichleiter et al., we used the canopy area rather than the total area, as the test facilities were largely unutilized. All sites are indoor facilities, absent daylight.

* * *

In summary, the aforementioned caveats notwithstanding, both studies have made contributions to understanding the cannabis-energy problem. Certainly much more data (and modeling) are needed to get a strong handle on national energy use associated with indoor cannabis production, and to understand the trends going forward as well as the potential for improved energy efficiency and greenhouse-gas reductions.

References

Backer R.; Schwinghamer T.; Rosenbaum P.; McCarty V.; Eichhorn Bilodeau S.; Lyu D.; Ahmed M. B.; Robinson G.; Lefsrud M.; Wilkins O.; Smith D. L. 2019. “Closing the Yield Gap for Cannabis: A Meta-Analysis of Factors Determining Cannabis Yield.” Frontiers in Plant Science 10:495 [link]

Canopy Growth. 2021. “Improving Lives, Ending Prohibition, and Strengthening Communities: 2021 ESG Report.” Ontario, Canada: Canopy Growth Corporation, 70pp. https://www.canopygrowth.com/investors/news-releases/canopy-growth-issues-environmental-social-and-governance-report (accessed February 23, 2022)

Leichliter, K.; Bisbee, D.; McGregor, M. 2018. “Amplified Farms 2017 Indoor Horticulture Lighting Study.” Sacramento, CA: Sacramento Municipal Utility District, 36pp. [link]

Mills, E. 2012. "The Carbon Footprint of Indoor Cannabis Production," Energy Policy 46:58–67. [link]

NFD. 2018. “The 2018 Cannabis Energy Report.” Washington, DC: New Frontier Data, 63pp. [link]

Summers, H.M.; Sproul, E.; Quinn, J.C. 2021. “The Greenhouse Gas Emissions of Indoor Cannabis Production in the United States.” Nature Communications. [link]

Last Updated: May 8, 2021