Running Power Definition and

Utility

Dr. Kristine Snyder

with scientific advisors Shalaya Kipp, M.S.
and Prof. Wouter Hoogkamer

“All models are wrong. Some models are useful.” - George E. P. Box

1 Motivation and Organization

There has been significant confusion lately as to what different running power values mean,
how they relate to power values in cycling, and what metrics are consistently useful. One
of the fundamental issues in using power as a surrogate for effort in running is that the
physically defined metrics may not be the ones that a runner may find useful, whereas the
useful metrics may be more difficult to calculate. The purpose of this article is to define
the various types of power involved in running, to describe the different models used to
estimate power and their various scales, and then to discuss the potential utility of each of
these power measures.
Creating this explanation was challenging. The goal for this document was for everything
to be written in a way that someone who just wanted the basics could get what they
needed without being overwhelmed, while also including the kind of detail an academic
would require and the informed, intelligent runner might really want and appreciate. So,
this first piece is an overview of the types of power and other metrics involved in running,

1
with more detailed pieces to follow on specific topics where necessary.

2 What is Power?

metabolic
energy/power
energy loss

mechanical
energy/power

Figure 1: Conversion of metabolic power to mechanical power, with energy loss.

In running, there are two main definitions of power:

1. Mechanical power- The work/energy output per second. The mechanical power the
body exerts on the outside world and to move the limbs relative to the body, or the
output power produced by the muscles to move the body in running. This is usually
measured in Watts (Joules/second) and normalized by mass, giving Watts/kg.
2. Metabolic power- The work/energy input of the locomotor muscle per second.
The amount of metabolic energy per second the body has to use to run. This is also
measured in Watts/kg, though is also often converted into calories.
Mechanical power tends to be useful for analyzing the biomechanics, especially joint-by-
joint mechanical power, whereas metabolic power is often more useful for evaluating overall
effort.

2
3 Mechanical Power
3.1 At the organismal (body) level

weight

horizontal
vertical ground
ground reaction reaction force
force

Figure 2: Forces involved in a whole body mechanical system during running.

The simplest way to model the body in running is as a point mass that acts upon and
is acted upon by the outside world. To determine overall body mechanical power, the
body is typically assumed to be just a point mass, the potential and kinetic energy of the
center of mass are summed to get the total work, and that value is then divided by time.
Alternatively, power can be determined by multiplying forces acting on the center of mass
times the velocity. However, this model only takes into account the interaction of this
assumed point mass with the environment, and does not capture the power necessary to
move the limbs relative to the body.
In level steady-state running, ignoring losses, this value is zero, so, when using body me-
chanical power as a correlate for effort, often only the positive power contributions and not
the negative ones are included.
∆EKIN + ∆EGP E+
P = (1)
∆t
Further note that this equation only includes the mechanical power involved in interacting
with the environment, not with the limbs moving relative to the body.

3
3.2 At the joint-by-joint level

hip angle

knee
angle ankle angle

Figure 3: Definition of lower limb joint angles during running. Extension is considered positive,
and flexion is considered negative.

The body is not a point mass, but a series of connected segments, so instead mechanical
power in running is examined at a joint-by-joint level. Rather than assuming the body is
a point mass, this way of determining mechanical power includes all joints and calculates
power at each joint using the forces under the feet, the masses of the body segments, and
the angular velocities of the joints and adds these quantities altogether. This allows for
more specificity on where the power is being produced, as well as allowing for changes that
occur in swing as opposed to only stance. Note however, that biarticular muscles (muscles
that cross two joints) allow for the transfer of power between joints.
To calculate power on a joint-by-joint basis, the process starts at the foot. The moment
(torque) around the ankle due to the force of the ground on the foot, the movement at the
foot, and the weight of the foot, along with the ankle angular velocity, are used to calculate
ankle power. Similar calculations are performed at the knee and the hip. These values are
summed together to get mechanical work, with power being calculated as work/time. To
calculate joint power in a lab, optical motion capture is used to track the motion of the
limbs, and force plates are used to determine the forces under the feet.

4
4 Metabolic Power
In a lab, metabolic power (PM ET ) is generally calculated via indirect calorimetry, which
uses expired gas measurements, assumptions or knowledge of substrate utilization, and
formula(e) to determine steady-state energy consumption. While a person is running,
the amount of oxygen they consume and the amount of carbon dioxide they produce is
measured by using a mask placed over their nose and mouth. Metabolic power is then
calculated from these steady-state oxygen and carbon dioxide values. The following
commonly used equation comes from (Péronnet and Massicotte, 1991).

PM ET = 16.89V O2 + 4.84V CO2 (2)

Indirect calorimetry only accounts for aerobic processes. Therefore it alone is not accurate
when people are working hard enough that they’re also using anaerobic metabolic processes.
The true gold standard would be doubly labeled water, in which subjects are given water
labeled with deuterium (2 H) and 18 O, which are then used to determine the elimination
rate of CO2 , or direct calorimetry, in which the energy expenditure is determined by
measuring how much heat is produced by an individual in a tightly enclosed measurement
chamber, but these measurements would be more prohibitively difficult.
Estimating metabolic power without indirect calorimetry involves using knowns about
running conditions, environmental conditions, and user anatomical, biomechanical, and
physiological data to estimate the in lab values as accurately and precisely as possible.

5 Relating Mechanical to Metabolic Power

5.1 Efficiency
Efficiency is the ratio of the mechanical power production divided by the metabolic power
consumption. On the surface, this seems simple, but it is not as simple in running as it can
be in other activities. In positive power production, when the muscles are producing power
by contracting concentrically (shortening), efficiency is ∼25%. Note: the word contraction
just implies muscles are active. They can contract concentrically (while shortening), iso-
metrically (while staying the same length), or eccentrically (while lengthening), Efficiency
values can be significantly higher in movements that involve negative power (via eccentric
contractions) or energy recycling, but can vary, particularly as incline increases.

5.2 Efficiency is simple in cycling

In cycling the mechanical system is constrained, because 1) force is produced and not
absorbed (via shortening muscles through a concentric contraction), and 2) no mechanical
energy is recycled. The constrained mechanical system during cycling allows for a strong

5
Figure 4: How mechanical (external) power and metabolic rate (power) compare in cycling.
Figures taken from the review paper by (Etterna and Loras, 2009).

correlation between mechanical and metabolic power. Because positive muscle power has
an efficiency of ∼25%, the metabolic power is essentially just four times the mechanical
power. The correlation between the two power values is 0.97, very high (Ettema and Loras,
2009). Further, because cycling is a highly constrained system with a known, consistent
foot trajectory, mechanical power is easy to measure. Consequently, it is common to
measure mechanical power and accurately use it as a correlate for metabolic power.

5.3 Why is the relation between mechanical and metabolic power more
complex in running?
There are a number of differences in running that make the relation between mechanical
and metabolic power more complex than cycling.
1. The mechanical system is less constrained. The amount of force produced
and the metabolic energy required to produce that force each depend mainly on two
things: the length and velocity of the muscle fiber producing that force. When the
system is less constrained, these can vary more and have a larger effect on the change
in metabolic energy required.
2. Forces are both produced and absorbed by the body. Because the efficiency
for force absorption is higher than the ∼25% for positive force production, the overall
efficiency is no longer ∼25%. However, the actual efficiency varies based on how much
of each force is produced.
3. Energy can be recycled. In running, elastic energy can be stored in the tendons
and other elastic tissues, and then returned. Additionally, there is a metabolic benefit

6
 to shortening a muscle after it has been stretched. These changes also increase the
efficiency above 25%, but it can be difficult to determine how significant a role these
factors play.
Because of these differences, mechanical power and/or efficiency are not consistently the
best indicators of effort in running across conditions, so an alternative metric that correlates
better with effort is often used instead.

5.4 An alternative to efficiency for a more complex system: economy/cost

of transport
Running economy is not the same thing as running efficiency, though they are related.
Running economy is traditionally determined by measuring the rate of oxygen or energy
consumed at a given speed. When expressed relative to body mass, running economy can
have the units of oxygen (mlO2/kg/min) or energy (J/kg/min or W/kg). Your running
economy could be better because you do less mechanical work to run at a given speed
but have a typical mechanical to metabolic efficiency. It could also be better because
you are more efficient, in that you perform the typical mechanical work, but it incurs
less metabolically due to other factors, such as more effective elastic energy storage and
return.
One of the most common ways to represent economy is by cost of transport, or the energy
cost per unit distance. Cost of transport is calculated by taking the mass-specific energy
cost per second (as described in the previous paragraph), and dividing by the speed, and
is measured in J/kg/m. You can think about cost of transport as the energy it costs to
run per meter or mile at a given speed; it is essentially the inverse of miles per gallon in a
car.

6 What metric is most meaningful for a runner?

6.1 Metabolic importance
Generally, people, largely subconsciously, choose to move in a way that minimizes metabolic
energy use. People use the stride length (Hogberg, 1952; Snyder and Farley, 2011), stride
width (Donelan, 2001; Arellano and Kram, 2011), stride frequency (Hogberg, 1952; Snyder
and Farley, 2011), speed (Inman et al., 1981), and gait (Cavagna et al., 1987) that minimize
metabolic variables, though sometimes it is metabolic rate (energy/time) that matters and
sometimes it is cost (energy/distance) that matters.
Metabolic energy expenditure is also the predominant determinant of performance in en-
durance running races and a critical element of optimizing training (Noakes, 2003; Daniels,
2013). In racing, the goal is to have minimal energy at the finish line, not to burn out
before and not to have a lot left afterwards. To do this effectively, you want to have as

7
Figure 5: A graph of stride frequency vs. metabolic cost adapted from (Snyder and Farley, 2011)
shows how humans use metabolic rate/cost to optimize their running form. Note that because
these data were all collected at 2.8 m/s, the approximately parabolic shape is the same for stride
frequency vs. metabolic rate/power.

good an estimate as possible of how much metabolic energy your body consumes at each
point of the race. Essentially, what most runners want from a power number is a better
estimate of effort than they can determine themselves and a way to log this value over
time. By having the best estimate possible of metabolic power, runners can consciously
make decisions using data their body already unconsciously uses.

6.2 Utility of other metrics

Though metabolic power under given conditions comes close, there is no one silver bullet
to measure your running efficacy. Cyclists use a faster than metabolically optimal cadence
and pay a small metabolic price so that they can avoid large muscle and tendon forces
(Etterna and Lornas, 2009). In some downhill and other energy absorbing actions, muscle
moments or other measures also seem to be more of a determining factor in what is the
optimal method to use than metabolic power (Dean, 2013). Avoiding injury can outweigh
minimizing effort.

6.3 What is the solution?

Because no one metric entirely captures running efficacy under all conditions, it is impor-
tant to collect as much precise, accurate, and useful biomechanical and physiological data
during training as is feasible. It is then necessary to share the most useful metric(s) to the
runner in close to real time, as well as to determine values for other useful metrics for later
use, along with guides for interpretation of their meaning. It would be ideal to be able
to directly measure oxygen consumption during training, but, currently, even if it were af-

8
fordable, it is impractical, if not impossible, to run with a portable metabolic system every
day. Further, even if it were practical, metabolic rate would only be updated every breath
plus processing time, and would not capture anaerobic metabolism. Heart rate devices are
wearable and offer a commonly used metric, but heart rate is slow to reflect a change in
metabolic requirements and also can be affected by stress, immune response, and other
factors unrelated to instantaneous metabolic demand. Note that we define instantaneous
metabolic demand to be the metabolic power due to aerobic and anaerobic sources, similar
to oxygen demand in (Gløersen, 2019). Currently, the best option is to have a wearable de-
vice that offers an accurate, precise close-to-real-time correlate for instantaneous metabolic
demand, but also determines other metrics, biomechanical and physiological, for post-run
analysis.

7 Conclusion
The fundamental purpose of this document is to define the different types of running
power, the models used to estimate them, and to discuss each power measure’s utility.
There are two fundamental types of power in running: mechanical power (energy exerted
by the body on the environment and to move the limbs per second) and metabolic power
(energy used by the body per second). Each can be modeled with varying degrees of
accuracy/precision depending on the data and assumptions used to build the model. The
question of which value to use fundamentally comes down to utility. Which measure is most
useful for running power across different people, running conditions, and environmental
conditions? If you want to use one power number to estimate your effort and determine your
pacing, there is a preponderance of evidence showing that your body already fundamentally
uses metabolic power/cost, or more accurately, instantaneous metabolic demand, which
lacks the time delay of metabolic power. Being consciously aware of the metric your
body is subconsciously using can help you make choices to optimize your training and
performance.

References
[1] Péronnet, F., and Massicotte, D. (1991). Table of nonprotein respiratory quotient:
an update. Canadian Journal of Sport Sciences = Journal Canadien des Sciences du
Sport,, 16 1, 23-9
[2] Ettema, G., and Lorås, H. W. (2009). Efficiency in cycling: a review. European Journal
of Applied Physiology, 106 (1), 1-14.
[3] Hogberg, P. (1952). How do stride length and stride frequency influence the energy-
output during running?. Arbeitsphysiologie 14, 437-441.

9
[4] Snyder, K. L., and Farley, C. T. (2011). Energetically optimal stride frequency in
running: the effects of incline and decline. Journal of Experimental Biology, 214 (12)
2089-2095.
[5] Donelan, J. M., Kram, R., and Kuo, A. D. (2001) Mechanical and metabolic determi-
nants of the preferred step width in human walking. Proceedings of the Royal Society
of London. Series B: Biological Sciences, 268 (1480), 1985-1992.
[6] Arellano, C. J., and Kram, R. (2011). The effects of step width and arm swing on
energetic cost and lateral balance during running. Journal of Biomechanics, 44 (7),
1291-1295.
[7] Inman, V.E., Ralston, and Todd. (1981). Human Walking. Williams and Wilkins, Bal-
timore.
[8] Cavagna, G. A., and Kaneko, M. (1977) Mechanical work and efficiency in level walking
and running. The Journal of Physiology, 268 (2), 467-481.
[9] Dean, J. C. (2013). Proprioceptive feedback and preferred patterns of human movement.
Exercise and Sport Sciences Reviews, 41 (1), 36.
[10] Tim Noakes. (2003) The Lore of Running. Human Kinetics, Champaign, IL.
[11] Jack Daniels (2013) Daniels’ Running Formula Human Kinetics, Champaign, IL.
[12] Gløersen, 0. N. Gilgien, M. Dysthe, D.K. Malthe-Sørenssen, A. Losnegard, T. J. Oxy-
gen Demand, Uptake, and Deficits in Elite Cross-country Skiers during a 15-km Race
Medicine and Science in Sports and Exercise

10