Over the last three decades, I've worked as counsel for several clients in the mining, oil & gas and battery industries. Because of that experience, I've always understood that battery manufacturing is extremely energy intensive. After all, the entire value chain from mining and purifying metals through component fabrication and final assembly requires massive fossil fuel and electric power inputs. Until last week, however, I didn't truly understand the magnitude of the energy inputs required to make a battery.
Last week I spent a few days digesting a paper from Stanford University's Global Climate and Energy Project titled "On the importance of reducing the energetic and material demands of electrical energy storage" that was published in the Royal Society of Chemistry's Journal of Energy & Environmental Science.
While the paper's conclusions were predictable, I was tremendously impressed with its "cradle-to gate" analytical framework that started with black earth, estimated total energy inputs for mining and processing raw materials, and then estimated additional energy inputs for manufacturing batteries from those materials. The result was a highly informative series of "embodied energy" values that represent the total amount of energy invested in manufacturing a particular type battery.
This graph from the Stanford paper ignited an insatiable curiosity because it uses "megajoules," an apples-to-apples measurement standard, to compare the total embodied energy in a battery with the total energy storage capacity of that battery. I've never seen anything like it and I believe it may be the most important concept I've ever discussed.
While the graph used megajoules for its basic measurement unit, it could have just as easily used "watt-hour equivalents" or "kilowatt-hour equivalents" since those measurement units are interchangeable in fixed conversion ratios. As long as the same measurement unit is used for both sets of data, it really doesn't matter what the unit is.
At the high end of the scale, the Stanford graph shows that it takes 694 units of primary energy to make one unit of storage capacity using vanadium-redox technology. At the low end, the graph shows that it takes 73 units of primary energy to build one unit of storage capacity using compressed air technology.
In every case, the total embodied primary energy is a huge multiple of the total storage capacity.
Since I was shocked by the sheer magnitude of the embodied energy values, I called Dr. Charles Barnhart, the principal author of the Stanford paper, to develop a better understanding of his methodology and data sources. During that conversation, Dr. Barnhart pointed me to a 2010 study from Argonne National Laboratory titled "A Review of Battery Life-Cycle Analysis: State of Knowledge and Critical Needs" that was a real eye-opener.
The goal of the Argonne study was simple - collect and summarize the available data on the cradle-to-gate energy inputs for five battery technologies. For some technologies like lead-acid batteries, detailed and reliable embodied energy data was readily available. For others like lithium-ion batteries, Argonne made well-informed estimates and then qualified their conclusions with dire warnings about the pressing need for better and more complete data.
This graph from page 22 of the Argonne report summarizes the cradle-to-gate primary energy inputs for a watt-hour of battery capacity using five different chemistries. Unfortunately, it expresses the embodied energy in megajoules while expressing the energy storage capacity in watt-hours.
While the presentation format is technically correct because any type of energy can be stated in megajoules but only electricity can be stated in watt-hours, the importance of the data is lost unless the reader knows that a megajoule is equivalent to 277.78 watt-hours of electricity. For mere mortals like me, the key information in the Argonne graph is about as useless as speed data stated in furlongs per fortnight.
Because of my conversation with Dr. Barnhart, I knew it was critically important to do the energy unit conversions and get an apples-to-apples comparison. When I did, a shockingly clear picture emerged. The following table takes the Argonne data and converts the primary energy inputs from megajoules to watt-hour equivalents, "Whe." The end result is an accurate, scalable and understandable comparison for normal people who simply want to understand the energy balance of storage batteries.
The implications are staggering, but they've been completely ignored by clean energy advocates, policy-makers and investors alike. Everybody focuses on what batteries can do, but nobody pays attention to the embodied energy costs.
In the last century people used small amounts of battery capacity for high value applications, so nobody really cared if the manufacturing process used 2 kWhe of energy to make a cellphone battery, 20 kWhe to make laptop battery or even 100 kWhe to make a starter battery.
In this century, people want to use huge amounts of battery capacity for low value applications. Advocates blithely discuss kWh-scale batteries for electric cars and MWh-scale batteries for renewables integration without even considering the massive front-loaded primary energy investment that battery manufacturing requires. They also ignore the dreadfully inconvenient truth that mining and transportation are powered by diesel, metal processing is powered by coal and natural gas, and battery manufacturing is powered by natural gas and grid-mix electricity.
In the Stanford paper, Dr. Barnhart concluded that every decision to deploy stationary energy storage should be based on a comparison between the primary energy embodied in the storage system during manufacturing and the energy that will cycle through the system over its useful life. He then proposed this general equation to calculate energy stored per unit of primary energy invested, or "ESOI."
Round-trip efficiency x Depth of discharge x Cycle life
Using the ESOI formula and a variety of operating assumptions, Dr. Barnhart then created this graph that compares the 30-year ESOIs of seven model load balancing systems using two geologic and five electrochemical storage technologies. While there's room to quibble over Dr. Barnhart's assumptions, it's clear that using battery systems for load balancing and solar power integration will be a long and difficult road.
While ESOI is a good general formula and it will certainly work for planning purposes in stationary applications if you can nail down all the variables, for applications like electric vehicles the equation is a complication that hides the facts instead of clarifying them. In the case of electric vehicles, it's easier to calculate the embodied energy in the battery and compare that value with the amount of energy that will flow through the battery during its useful life.
I've prepared Excel worksheet with my megajoules to Whe conversions and my calculations for the following examples. Readers who want to see the detail can download the worksheet from my Dropbox.
For my first example, I'll use a Model S Performance Edition from Tesla Motors (TSLA). Based on an embodied energy of 472 kWhe per kWh of battery capacity, the Tesla's 85 kWh battery pack will have 40,120 kWhe of total embodied energy. Over a ten-year useful life, assuming 15,000 miles per year of driving and an average efficiency of 3.5 miles per kWh, 42,850 kWh of purchased electricity will flow through the battery, yielding an ESOI of 1.1.
For my second example, I'll use a Leaf from Nissan Motors (OTC:NSANY). Based on an embodied energy of 472 kWhe per kWh of battery capacity, the Leaf's 24 kWh battery pack will have 11,328 kWhe of total embodied energy. Over a ten-year useful life, assuming 15,000 miles per year of driving and an average efficiency of 4 miles per kWh, 37,500 kWh of purchased electricity will flow through the battery, yielding an ESOI of 3.3.
For my third example, I'll use a Prius from Toyota Motors (TM). Since the Prius is an HEV the assumptions get a little more complex, but the logic is easy to follow. The Prius has an EPA fuel economy rating of 50 mpg, which is significantly better than the 2013 model year CAFE standard of 34.2 mpg for passenger cars. Over a ten-year useful life assuming 15,000 miles per year of driving, the Prius will save 1,385 gallons of gasoline when compared to a CAFE compliant new car. Since a big part of the mpg performance in a Prius is attributable to mechanical design and great aerodynamics, I'll assume that 700 gallons of the fuel savings are attributable to the electric drive and the balance should be ignored for ESOI purposes. Since internal combustion engines are inherently inefficient and only deliver about 20% of the energy in a gallon of gasoline to the wheels, I'll work with a net fuel savings of 140 gallons of gasoline. Based on an embodied energy of 750 kWhe per kWh of battery capacity, the 1.4 kWh battery pack in the Prius will have 1,050 kWhe of total embodied energy. At 36.6 kWhe per gallon of gasoline savings, the energy equivalent of 5,124 kWh will flow through the battery, yielding an ESOI of 4.9.
For my fourth example, I'll use a typical mild micro-hybrid that uses a simple dual-battery system to reduce fuel consumption by 10% compared to a CAFE compliant new car. Over a five-year battery life assuming 15,000 miles per year of driving, the mild micro-hybrid will save 219 gallons of gasoline. Once again, since internal combustion engines are inherently inefficient and only deliver about 20% of the energy in a gallon of gasoline to the wheels, I'll work with a net fuel savings of 44 gallons of gasoline. Based on an embodied energy of 167 kWhe per kWh of battery capacity, the 1.5 kWh dual-battery system in the mild micro-hybrid will have 251 kWhe of total embodied energy. At 36.6 kWhe per gallon of gasoline savings, the energy equivalent of 1,605 kWh will flow through the batteries, yielding an ESOI of 6.4.
For my last example, I'll venture out a limb and evaluate the NS-999, a battery powered switching locomotive that Norfolk Southern (NSC) is presently renovating with an array of 864 PbC batteries from Axion Power International (OTC:AXPW). Based on data from a 2004 presentation by Argonne National Laboratory, a battery-powered switcher should save between 68,000 and 86,000 gallons of diesel fuel over the course of a 250-day work year. For the sake of simplicity, I'll assume the NS-999 will save 50,000 gallons per year, or 250,000 gallons over a five-year battery life. Since the thermal efficiency of diesel engines is about 30% compared to the 20% value for gasoline engines, I'll use a 30% efficiency factor and ignore the additional efficiencies of a series electric drive over a mechanical drive. Based on an embodied energy of 167 kWhe per kWh of battery capacity, the 432 kWh of PbC batteries in the NS 999 will have 72,144 kWhe of total embodied energy. At 40.7 kWhe per gallon of diesel fuel savings, the energy equivalent of 3,052,500 kWh will flow through the batteries, yielding an ESOI of 42.3.
In 1883 Thomas Edison said, "The storage battery is one of those peculiar things which appeals to the imagination, and no more perfect thing could be desired by stock swindlers than that very selfsame thing. Just as soon as a man gets working on the secondary battery it brings out his latent capacity for lying."
The ESOI metric proposed by Stanford's Global Climate and Energy Project is a tremendously effective way for serious investors to separate sensible battery-enabled energy conservation from unconscionable waste and pollution masquerading as conservation. It highlights the insanity of investing as much primary energy in a battery as the battery will store over its useful life. It can also help investors identify compelling economic opportunities that are not always obvious. In the final analysis, unless the fundamental economics and energy balance work, story stocks can't work.
Additional disclosure: I am a former director of Axion Power International. I have no long or short positions in any other stocks mentioned, and no plans to initiate any positions within the next 72 hours. To the best of my knowledge I have no clients, family members or personal friends with long or short positions in any other stocks mentioned.