Adding carbon fiber to shoe soles does not improve running economy: a muscle-level explanation

Owen N. Beck, Pawel R. Golyski, Gregory S. Sawicki

Preprint posted on 29 February 2020

Article now published in Scientific Reports at

Carbon fibers added to running shoes is not enough to save energy and win the marathon

Selected by Jonaz Moreno

Categories: bioengineering


Running is one of the most common and easy ways to stay active. As a professional athlete, every single piece of advantage can make the difference between winning the gold medal or not. One of the most recent and controversial topics in long-distance running is whether carbon fibers embedded in the shoes improve performance by reducing total aerobic energy expenditure. Previous studies have indicated that the addition of carbon fiber plates to shoes causes footwear-altered biomechanics and aerobic energy expenditure that improve performance1,2. However, these studies did not provide a physiological explanation to these performance improvements. Thus, the authors of this study were interested in analyzing lower limb biomechanics and muscle dynamic changes of the soleus with changes in stiffness on the shoes by inserting carbon fiber plates.

Key findings

The experiment consisted of 15 male participants free of cardiovascular, orthopedic, and metabolic disorder, and who could run 5 km in under 25 minutes.  The running shoe prototype that was used in this test was the Adidas Adizero Adios BOOST 2. Their control shoe was no carbon fiber and they embedded 0.8, 1.6, and 3.2 mm thick carbon fiber plates. This corresponded to a 13.0 ± 1.0, 31.0 ± 1.5, 43.1 ± 1.6, and 84.1 ± 1.1 kN/m (3-point bending stiffness test), respectively. The 3-point bending stiffness test consists of placing the shoe in a frame that has two supporting bars 80 mm apart. Then, a material testing machine applies a vertical force on the top of the shoe to create a 10 mm displacement following a 2 N preload. The bending stiffness was quantified using the average linear slope of the force-displacement data within the 5 – 9 mm range.

 Limb-joint dynamics:

Authors found no significant effect of adding carbon fiber plates on hip, knee, and ankle angles or moments (average, minimum, and maximum; Figure 1).

                Figure 1: Left column: Hip, knee, and ankle angle versus time (a,c,e). Right column:  hip, knee, and ankle moment versus time (b,d,f). Vertical lines indicate the average end of ground contact time for the respective footwear condition.  Reproduction from Figure 1 of the preprint (Beck et al.)


Stride kinematics and Ground Reaction Force (GRF):  

Increased bending stiffness lead to longer ground contact time (p=0.48) but did not change the step time (p=0.956). There was no significant effect of increasing bending stiffness on average vertical, breaking, nor propulsive GRF. Furthermore, there was no change in the fraction of vertical or horizontal GRF during the first half of ground contact (Figure 2).

Figure 2: Left column: average vertical and horizontal ground reaction force (GRF), average gear ratio, and average moment versus time (a,c,e,g). Right column: average vertical and horizontal ground reaction force, average gear ration, and average moment versus shoe bending stiffness (b,d,f,h). Vertical lines indicate the average end of ground contact time for the respective footwear condition and error bars indicate standard error. Reproduction from Figure 2 of the preprint (Beck et al.).


Muscle-tendon Dynamics:

There was no significant effect on soleus muscle-tendon (MT) dynamics. Average soleus MT force, length, or velocity were not significantly different across all the different bending stiffness during ground contact time. Also, the ratio of the GRF versus soleus MT moment arms to the ankle-joint center (gear ratio) showed no significant difference across all bending stiffness (Figure 3).

Figure 3: Left column: Average soleus muscle-tendon (MT) force, length, and velocity versus time (a,c,e,). Right column: Average muscle-tendon force, length, and velocity versus shoe bending stiffness (b,d,f). Vertical lines indicate the average end of ground contact time for the respective footwear condition and error bars indicate standard error. Reproduction from Figure 3 of the preprint (Beck et al.).


Soleus Dynamics:

There was no significant effect on average of maximum soleus fascicle pennation angle, force, velocity, or length across all the bending stiffness. On the other hand, footwear bending stiffness did affect the average (p=0.026) and maximum (p=0.014) activated soleus volume during ground contact, but  there was no significant effect on the average activated soleus volume per stride (Figure 4).

                Figure 4: Average (±SE) gross aerobic power (a), and activated soleus (Sol) volume (Vol) per stride. Right axis indicates the percent difference in the respective variables from the Adidas condition without a carbon fiber plate (black dot) versus shoe bending stiffness (blue dot: 31.0 kN/mm, green dot: 43.1 kN/mm, orange dot: 84.1 kN/mm). Reproduction from Figure 3 of the preprint (Beck et al.).


Muscle Activation:

Changes in footwear bending stiffness had no significant effect on stance- or stride-averaged activation of any of the seven measured muscles (soleus, medial gastrocnemius, tibialis anterior, biceps femoris, vastus medialis, gluteus maximus, and rectus femoris).

Running economy:

The authors found no significant effect on gross aerobic power. However, the level of stiffness that minimized running economy varied across all participants. 1 participant had the optimal (minimum) energy consumption with the control condition, 4 participants showed optimal energy consumption using the 0.8 mm tick carbon fiber, 4 participants had their optimal energy consumption with the 1.6 mm tick carbon fiber, and 6 participants had their optimal energy consumption with the 3.2 mm tick carbon fiber (Figure 4).


Overall, this study demonstrates that increasing bending stiffness of the footwear does not improve running economy. All the recorded dependent variables showed no significant effect, except for soleus muscle volume activation during contact.  Some limitations of this study were 1) the placement of the carbon fiber plates in between the socks and the insole of the shoe while other footwear have the carbon fiber embedded between the foam of the shoe sole , 2) soleus was modeled as cylinder, and 3) not having a formal habituation protocol like previous studies1,2.

What I liked about this preprint

The use of ultrasound imaging is becoming something that can be used during dynamic motions. Tapping into what is the muscle response to activities such as running can be beneficial to understand muscle mechanics. Furthermore, the information obtained from the imaging can aid in understanding the link between biomechanical and physiological responses within the muscle. The muscle dynamics are parameters that have been addressed by models and simulations without having in vivo information of the muscle response to running.

Open Questions

  1. You had all participants running at fixed speed (3.5 m/s), so I was wondering if you thought about determining each participant’s preferred running speed instead. There is proven research of how running at the preferred speed optimizes energy consumption. Did you thought about increasing or decreasing the speed?
  2. Although most of your participants were characterized as heel-strike runners, you had some forefoot and midfoot runners. I was curious to know your ideas of whether foot strike patterns may affect the GRF-ankle moment arm and if that might help more one runner versus the other runner.
  3. In your discussion, you commented about how the Nike shoe improves running economy not by simply adding the carbon fiber but how it is embedded in the foam of the sole. Aside from that part of the shoe design, have you thought about modifying the carbon fiber plate design to cause changes in the metatarsophalangeal joint which is the center of rotation for the foot. Perhaps simply trying different carbon fiber plates designs and the effects on ankle moment in general.


  1. Roy, J.-P. & Stefanyshyn, D.J. Shoe midsole longitudinal bending stiffness and running economy, joint energy, and EMG. Med Sci Sports Exerc 38, 562-569 (2006).
  2. Oh, K. & Park S. The bending stiffness of shoes is beneficial to running energetics if it does not disturb the natural MTP joint flexion, Biomech 53, 217-135 (2017)


Posted on: 11 May 2020


Read preprint (No Ratings Yet)

Have your say

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here