Data-centers: electricity use and demand shifting?

This data-file estimates the electricity demand from data-servers and the proportion of that demand that could be “demand-shifted”, by compiling notes and data-points from technical papers.

We estimate data-centers comprise 2% of global electricity and will likely rise to at least 5% by 2030. Around 40% of data-center energy demand can likely be demand-shifted, of which two-thirds is temporal and one-third is geographical.

This permits another 1-2% potential share for renewables in the grid, without needing to resort to curtailment, or to install expensive batteries or other backup solutions, per our recent research.

Companies buying nature-based carbon offsets?

This data-file aims to tabulate how 35 leading companies globally are purchasing nature-based carbon offsets, in order to offset their CO2 emissions.

The data-file includes the rationale for the purchases, the projects supported, how they are verified, their scale and their cost.

We find appetite for carbon offsets accelerating. 60% of the projects are reforestation projects, 2.5x more prevalent than forest conservation. 70% are undertaken indirectly through partners, versus 30% undertaken directly. 95% of indirect projects have sought third-party verification, while 60% of direct projects have self-verified.

Companies in hard-to-abate sectors, such as transport and materials were more likely to prefer carbon-offsets, while those in easy-to-abate sectors, such as tech, finance and media have been more likely to purchase nature-based carbon credits as a ‘last resort’ after exhausting other options.

CO2 from electric vehicles versus ICEs and hydrogen?

This data-file tabulates the CO2 intensity of producing and charging lithium ion batteries for automotive use, split across 10 different components, informed by the technical literature. Producing the average EV battery emits 9T of CO2 (chart below).

Electric Vehicles should nevertheless have c50% lower emissions than gasoline vehicles over their entire useful lives, assuming equivalent mileages. Although we see gasoline vehicles’ fuel  economies improving.

Manufacturing EVs has an energy deficit, which means the ascent of EVs could increase net fossil fuel demand all the way out to 2037 (note here).

This data-file can be used to calculate the crossover point, which comes after around 3.5 years and c50,000 miles (chart above). The numbers will vary as a function of grid composition, technical improvements and vehicle specifications.

Inspection costs: drones versus traditional quality control?

This data-file estimates the costs of drone inspections, for the construction and resources industries, using bottom-up numbers from technical papers.

Costs per hour can be 30% lower than for traditional quality control inspections. A single drone, including software licenses likely costs c$30k, which is disaggregated line-by-line.

Our notes from technical papers are also included in the data-file.

US Air Passenger Miles and Fuel Economy?

This data-file tabulates statistics on the US aviation sector, from the Bureau of Transport Statistics, to compute the fuel economy of US air travel, per plane-mile and per passenger-mile.

In 2019, 10M US flights carried 930M passengers 1.1 trn passenger-miles. The latest data in the file run to February-2020. The latest date in the file run through the end of 2020, and show flights down 40%, passengers per flight down 40% and total passenger miles down -65% for 2020.

Fuel economy per passenger mile has risen at a 2.8% CAGR since 2003. Flight numbers have fallen by -0.4% pa and flights have become 0.8% longer.  But load factors have improved by 0.7pp each year, spreading 0.5 plane miles per gallon across more passengers. Low load factors worsened fuel economy by c40% in 2020.

Lighting: historical costs and energy efficiency?

This data-file assesses lighting solutions throughout history, quantifying energy efficiency (in lumens per watt and in percentage terms), plus the costs of lighting (in cents per 1,000-lumen-hours). The file goes back to 1800 and covers candles, whale oil, town gas, incadescent bulbs, halogen and LEDs.

The work culminates with LED lighting, tracing the record efficiency levels announced in recent years. Overall, the best LEDs now achieve over 80% useful energy efficiency, while lighting costs have fallen 100x over the past century.

Energy efficiency of household appliances?

What is the typical range of energy efficiencies for household appliances, such as air conditioners, clothes dryers, refrigerators, dishwashers, lighting, et al? To answer this question, we have tabulated almost 20,000 data-points from the US EPA’s excellent ENERGY STAR program, and other technical papers.

We estimate a house equipped with modern appliances will likely have 60% lower energy demand versus 30-years ago. However, c30% of tyipcal household hold appliances are more than 10-years old, and 5% are more than 20-years old, so there may be very large savings still to come from upgrading.

Even today, the best appliances have c30% lower energy consumption than the worst appliances (90th percentile versus 10th percentile), so it would be helpful to prioritize efficient purchasing decisions.

Electric trucks: what battery sizes?

This data-file models the possible battery sizes in a fully electric semi-truck. Lithium ion batteries up to 15 tons are considered, which could deliver 2,500 miles of range, comparable to a diesel truck.

However, large batteries above c8-tons in size detracts around 10% from the fuel economy of electric trucks, and may cause trucks to exceed regulatory weight limits, lowering their payload capacities.

4-6 ton batteries with 700-1000km ranges and 5-8% energy penalties may be best, and would likely add $110-170k of cost at 2020 battery costs. 

Our roadmap to decarbonize trucking most prefers carbon-offset diesel, then hybridization with super-capacitors, then electric semi-trucks, and least prefers hydrogen trucks.

Costs and CO2 Intensity of fuels: a cross-plot?

We have produced a cross-plot of the costs and CO2 intensities of different fuels, in $/boe and kg of CO2 per boe, to compre the relative attractiveness of decarbonization options.

Included in the file are different oil products, gas markers, coal, wood, nuclear, biofuels, methanol, hydrogen, CO2-EOR products and the US electricity grid for comparison.

This data-file simply contains the numbers behind the cross-plot shown above, for anyone looking to interrogate the data or re-format the chart. The workings behind each number are linked in other data-files.

Managed reforestation: growth rates by tree species over time?

We have tabulated data into the growth rates of over 2,500 trees, in over a dozen locations globally, based on tree ring measurements, reported by the National Center for Environmental Information.

The objective is to assess whether growth rates are faster at younger or older trees, given that biomass accumulation is correlated with tree widths. We find that tree widths continue rising steadily over 100-500 years, with radiuses growing at an average pace around 2mm per year.

However, growth rates are fastest in early years. As a rule of thumb, they will slow down by 25% after 20-years, 40% after 40-years and 50% after 60-years. This may be an argument for forest management and sustainable harvesting as more reforestation projects are undertaken to combat climate change (note here).

Relatively faster growing species tend to remain relatively faster growing even after longer time-frames. This may be an argument for rigorous species selection, as discussed further in our data-file here.