Overselling a vegetarian diet, underselling utility cycling

Why bother to check the math on an exciting and counterintuitive message? “Bicycles less efficient than SUVs!” Or, why you should never blindly trust a column by John Tierney of the NYTimes, and Someone is Wrong on the Internet!

It is widely known that it takes far more energy to produce meat (of various sorts) than grains, fruits, and vegetables (of various sorts). Some time ago someone made the calculation that eating beef to fuel bicycle riding is actually less efficient than driving an SUV the same distance, once you include the energy needed to produce the beef. Of course, this was never reported with numbers, but it was widely quoted, for instance by John Tierney of the New York Times. One site he links to makes the estimate that it takes 200 68 times as much energy to produce beef, as it does to produce potatoes.

However, this leads to an incorrect conclusion. The main problem is that potatoes and beef have a different energy content by weight. A secondary problem is that the energy costs for beef are widely quoted as “per kilocalorie (kcal, or food calorie) of PROTEIN”, and not per kcal of beef (including FAT). In 85% lean hamburger before cooking at least half of the food energy is fat. How we prepare that food and what we do with the fat afterwards matters quite a bit. A sedentary person “watching their weight” will prepare beef differently than a hungry bicyclist. A third problem is that there may be a math or transcription error of some magnitude in the calculations leading to the 200:1 energy cost claim; that does not appear to agree with other estimates that I read. (The largest factor appears to be confusion of weight and energy cost; by weight, 85% lean has about 2.8x the energy content of potatoes (215 kCal vs 77 kCal, for 100 grams of each). Correcting for this converts 200, to 71.)

The best reference for the energy required to produce food is Food, Energy, and Society by David and Marcia Pimentel (I have the 3rd edition, I will abbreviate as FES3). All the other estimates of relative energy cost start with data from Pimentel. The USDA provides information about the energy content of food by gram, though the weight of food is not relevant.

FES3, page 148, calculates the energy input:output cost of potatoes raised in the US as 1:1.3; for every unit of fossil fuel energy input, 1.3 units of food energy are produced.

FES3, page 69, estimates the energy input:output cost of beef cattle PROTEIN raised in the US as 40:1. If someone eating beef diligently trims all the fat, and carefully cooks the meat in a way to remove the embedded fat, and then discards that fat, this is the energy cost of the calories that they will get. If, on the other hand, the fat is retained and either consumed directly or converted into gravy, more calories are available; in 85% lean hamburger, this would change the cost per consumed kcal from 40:1, to 20:1. (Don’t forget that most of the beef is water.)

What does this mean for bicycles versus SUVs as transportation? A common estimate for the food energy needed to propel a bicycle is 50 kCal per mile. Much more is needed for higher speeds where wind resistance dominates (a good source for this information is Bicycling Science), but that does not describe ordinary cyclists, including commuters. If we could eat gasoline, it supplies about 30,000 kCal per gallon (we can eat nut oils, which have a similar energy density), so converting this to “miles per gallon” gives about 600 miles per gallon of energy input.

From this it’s a simple matter of applying ratios to get the end-to-end MPG of a bicyclist. If the fuel is 100% lean beef, then the cost is 600/40 = 15mpg. This is about as inefficient as driving an SUV, but it is more efficient than the worst SUVs (city mileage, since that is common case for cycling). However if we assume a cheap-hungry-cyclist (CHC) model of food consumption, the fuel is 85% lean beef and the fat is also consumed, then the full-cycle mpg is 30mpg. Not exactly wonderful, but far better than non-hybrid SUVs. And remember, this is reported as “eating beef and biking is less efficient than driving an SUV”, not “eating beef protein …”.

Eating potatoes increases the system mpg, instead of 600, it becomes 780. Notice, however, that the meat-protein:potato ratio is not 200:1, but 52:1. It’s large, but not that large. If we use the CHC consumption model, the ratio is only 26:1.

Another popular example is using milk for power. For milk protein, energy input:output is 14:1, but protein is the minority of calories in milk. In reduced-calorie 1% milk, 8 grams (32 kCal) of each 110 calorie serving is protein; the 1% milk energy input:output ratio is 14:(110/32) = 4.1:1. 600 divided by 4.1 is 145mpg — no car on the road, never mind SUV, gets that gas mileage. Walking is about 45% as efficient as cycling, so the end-to-end energy cost of someone walking and drinking 1% milk for fuel is about 65mpg.

There are more extreme cases. Lamb protein (p. 69) has a 57:1 cost, and oats (p. 108, Minnesota) yield 5.1 kcal for each 1 kcal of fossil energy input (that is, 1:5.1). That gives a true 290x ratio of energy cost (ignoring the fat-trimming issue for now); a lamb-protein-fed bicycle MPG would be only 10.5 mpg (still better than the worst SUV city mileage), but an oat-fed bicycle MPG is 3060. It’s unlikely that any other device can move people as efficiently; if we could find a way to burn oats in a train engine, perhaps that would do better. This has interesting implications for the “efficiency” of horses, too.

However, notice that it still takes energy to cook oats. If you heat 300 kCal of dry oats (1 cup) in 400mL (1-3/4 cup) of water from 10C (50F) to 100C (boiling point; but do not boil) it requires 90 times 400 equals 36000 calories, or 36 kCal. Since the input energy was only 59 kCal, this adds quite a lot to the energy input, and gives you a delivered-to-your-mouth energy return on energy invested (ERoEI) of only 3.15, and a “mere” 1900 mpg for the cyclist. This can be avoided by solar cooking in the summer, or by cooking on top of a wood-fueled stove in the winter, when the heat is useful in its own right. If you’re checking every detail, you know that radiant electric heat (from an electric stove, for example) captures only about 25-30% of the heat energy in its original fuel, after distribution.

Cooking doesn’t change the ERoEI of beef much because it is so bad already.

There are several sanity checks to apply to this. Very few people consume all the fat that comes with beef, and very few remove all the fat that comes with beef. The actual numbers are closer to the mushy middle. It is, however, almost insane to propose that a cyclist accumulating any useful mileage would make up the excess calories with pure protein; someone cycling 50 miles per week is burning 350 extra kcal each day on average. A gram of protein contains about four kcal of food energy — this means that a hypothetical 100%-lean-fueled cyclist would consume 90 grams of protein each day, just to drive the bicycle. The US adult RDA (FES, pp. 67-68) of all protein (plant and animal) is 56 grams, but the average American consumes 112 grams already. Boosting this to 200 grams takes you to the maximum intake recommended at a body-building site for a 220lb man. Keep in mind that all of this is in the context of non-athletic commuter and utility cycling, so “repair of muscle damage” does not really enter into it. In practice, no cyclist anywhere eats like this.

Another issue to consider is the impact of other greenhouse gasses. This increases the “cost” of eating beef or drinking milk, by what looks like a large amount. I haven’t yet been able to check the math from end-to-end on any paper claiming this, but since I have come across some cherry-picking of data, I want to be careful. It’s not a small amount, and it does matter.

Nonetheless, it does remain true that to an engineering approximation, Americans should eat much less meat. We would save energy, save money, not starve, and probably end up a little healthier. And further, bicycling really is that efficient, and you would need to spend a lot of money, prepare your food carefully, and adopt a profoundly unhealthy diet to make it even approach the inefficiency of ordinary, non-hybrid SUVs. If, instead, you prefer to use the most efficient human transport on the planet, merely ride a bike, eat more oatmeal, and cook it carefully.

2020 update: Exercise.  Duh, exercise.  Calories expended to push a bicycle are also calories expended for exercise; someone who bikes regularly for errands or commuting gets a literally healthy amount of exercise.  This puts cycling in a rather more favorable light, since spinning at the gym is a zero-miles-per-gallon activity that also burns calories.  In my case, I could claim to ride 60 miles per week on my bike for exercise, and expend zero energy on commuting.