Best practices for analyzing the direct energy use of blockchain technology systems: Review and policy recommendations
Nuoa Lei, Eric Masanet, Jonathan Koomey
Published in: Energy Policy
Blockchain technology is becoming increasingly relevant to energy analysts, policy analysts, and decision-makers. Blockchain can be applied to energy and environmental problems such as energy trading, electric vehicle charging, demand response, sustainable supply chain management, green certificates, and renewable energy promotion.
Lei et al. (2021) argue that the direct energy use of blockchain technologies is poorly understood and that the field would benefit from the greater adoption of best practices in analyzing these systems. The researchers note that current studies of blockchain’s energy consumption focus nearly exclusively on cryptocurrency applications, which are not indicative of the technology’s energy usage more broadly. Their work is intended to guide energy analysts in the production of accurate assessments of the energy implications of blockchain technology and to inform policy makers who are the end-users of these assessments.
The authors begin by outlining the key factors that determine the energy consumption of a blockchain system. Every blockchain system consists of IT devices, data flows, and electricity flows. Understanding the direct energy use of a blockchain system requires an understanding of the IT devices and the underlying conditions that govern their electricity consumption. Other factors that shape a blockchain system’s energy consumption include the scale of transactions, the applications, and the consensus algorithms that are employed.
The defining feature of a blockchain system is how it enables consensus to be reached regarding the validity of transactions across a peer-to-peer network, without mediation by a centralized authority. To achieve consensus, a blockchain system is governed by a consensus algorithm. The consensus algorithm that’s implemented has a tremendous impact on a blockchain system’s energy consumption. Lei et al. highlight some of the most popular algorithms including Proof of Work (PoW), Proof of Stake (PoS), Practical Byzantine Fault Tolerance (PBFT), Federated Byzantine Agreement (FBA), Proof of Authority (PoA), Proof of Capacity (PoC), and Proof of Burn (PoB).
Lei et al. describe 10 best practices that, if adopted by the research community, would lead to better estimates and modeling of the direct energy consumption of blockchain systems. While these best practices should be adopted by researchers, they also signal to decision-makers the types of questions they should be asking before drawing on the results and conclusions of any given study.
Properly include the full system. Blockchain systems include a combination of computing, storage, and communications hardware. Furthermore, this hardware is dependent on other auxiliary systems such as power supply and cooling systems. Researchers should transparently describe the boundaries they have established in their modeling and the reason for their approach.
Build from the bottom up. Bottom-up analysis is generally preferable to top-down analysis.
Use measured/surveyed power data. Modeling can sometimes rely on generalized or manufacturer-provided power data. These values correspond to specific system conditions or device configurations. It’s best to derive this power data through direct measurements or surveys of real-world operators of blockchain system components.
Use time-period appropriate technology data. Blockchain energy analysis should ensure that the technology data they use in their modeling is consistent with the time period of their analysis.
Account for capacity utilization. The energy consumption of a device does not always scale one-to-one with its degree of utilization. In these instances, energy analysts should clearly state and account for the relationship between utilization capacity and energy usage in their modeling.
Account for locational variations. The energy required for cooling computer centers can vary significantly by climate zone. A blockchain system’s energy use is also dependent on the underlying communication systems of a region, i.e. local network technologies, mobile station fuel types, and local network configurations. Analysts should clearly state the geographical boundaries of their analysis.
Properly account for uncertainty. Blockchain systems are rapidly evolving and analysts should address and communicate the inherent uncertainty contained within their analysis to decision makers.
Consider retrospective time series analysis. Decision makers benefit from retrospective analyses that span multiple years because these studies highlight historical changes and provide a useful framework for future scenario planning.
Avoid simplistic extrapolations. Overly simplistic extrapolations can lead to large errors in estimated future energy use. The best-practices stated with the paper can help prevent simplistic extrapolations.
Create open and complete documents allowing replication. Analyses should fully document the data sources, modeling equations, and analytical assumptions that produced them in order to ensure their replicability.
The paper outlines future research agendas that would produce higher quality research for policy-makers. Through increased public research funding, policy-makers and philanthropies have a role in spurring the research that would address these gaps. Possible research agendas that would advanced the field include:
Gathering measured energy use data for the IT technologies that undergird blockchain systems.
Developing a deeper understanding of capacity utilization effects.
Developing estimation methods for, and datasets on, the installed base of blockchain technologies. The most accurate IT energy models are built in a bottom-up fashion based on installed base data. However, a lack of available data makes it hard to perform this bottom-up approach to modeling. The research community should seek out partnerships with device manufacturers, blockchain application start-ups, and market analysts to derive the base estimates needed.
Understanding equipment lifespans, stock turnover, and generational improvements. Greater attention to these considerations will help in selecting an appropriate time period of technology data for a given analysis, performing retrospective time series analyses, and for future scenario planning.
Analyzing how spatial variations affect the direct energy use of blockchain systems. Local climates can affect the energy requirements for proper cooling of data centers, network connection types and technologies can affect communications energy use, and local electricity grid mixes will determine the environmental impacts of electricity usage.
Investigating the relationship between consensus algorithms, difficulty adjustments, validation node stock evolution, and validation node energy use.
Compiling and maintaining databases of blockchain applications.
Developing methods for proper treatment of uncertainty which include sensitivity analysis, uncertainty distributions for key parameters, and bounding scenarios where best and worst-cases are evaluated.
Citation: Lei, N., Masanet, E., & Koomey, J. (2021). Best practices for analyzing the direct energy use of blockchain technology systems: Review and policy recommendations. Energy Policy, Volume 156. https://doi.org/10.1016/j.enpol.2021.112422