![Research Inventy: International Journal of Engineering And Science
Vol.7, Issue 1 (January 2017), PP -12-16
Issn (e): 2278-4721, Issn (p):2319-6483, www.researchinventy.com
12
Application of Kennelly’model of Running Performances to Elite
Endurance Runners.
Vandewalle Henry, Md, Phd
Faculty of Medicine, University Paris XIII, Bobigny France
Abstract: The model of Kennelly between distance (Dlim) and exhaustion time (tlim) has been applied to the
individual performances of 19 elite endurance runners (World-record holders and Olympic winners) from P.
Nurmi (1920-1924) to M. Farah (2012) whose individual best performances on several different distances are
known. Kennelly’s model (Dlim = k tlim
) can describe the individual performances of elite runners with a high
accuracy (errors lower than 2 %). There is a linear relationship between parameters k and exponents of the
elite runners and the extreme values correspond to S. Coe (k = 15.8; = 0.851) and E. Zatopek (k = 6.57; =
0.984). Exponent can be considered as a dimensionless index of aerobic endurance which is close to 1 in the
best endurance runners. If it is assumed than maximal aerobic speed can be maintained 7 min in elite
endurance runners, exponent is equal to the normalized critical speed (critical speed/maximal aerobic speed)
computed from exhaustion times equal to 3 and 12.5 min in these runners.
I. Introduction
In 1906, Kennelly [1] studied the relationship between running distance (Dlim) and the time of the world records
(tlim) and proposed a power law of fatigue for the different types of exercise in humans and horses:
Dlim = k tlim
Equation 1
The value of exponent in power law is independent of scaling [2, 3]: the value of is independent of
the expression of tlim and Dlim. It is possible that is an expression of the endurance capability. Indeed, it is
likely that the curvature of the tlim-Dlim relationship depends on the decrease in the fraction of maximal aerobic
metabolism that can be sustained during long lasting exercises. The tlim-Dlim relationship is linear when is
equal to 1. When Dlim is normalized to the value of Dlim at maximal aerobic speed (MAS), it can be
demonstrated that the slope of the tlim-Dlim curve is equal to (Fig. 1A) for tlim equal to the exhaustion time at
MAS (tMAS).
At a running speed equal to MAS,
Dlim = k tMAS
= MAS tMAS and therefore, k = MAS*tMAS
1-
Equation 2
Figure 1: in A relationship between tlim normalized to tMAS and Dlim normalized to Dlim at MAS: the
slope (red dotted line) of Kennelly curve (black curve) is equal to for tlim/tMAS = 1. In B, the value of the
critical speed (red continuous line) computed from tlim1 and tlim2 is equal to the slope of Kennelly curve at tMAS
(red dotted line).](https://image.slidesharecdn.com/b07011216-170214112337/85/Application-of-Kennelly-model-of-Running-Performances-to-Elite-Endurance-Runners-1-320.jpg)
![Application of Kennelly’model of Running Performances to Elite Endurance Runners.
13
In 1954, Scherrer et al. [4] proposed a linear relationship between tlim and the total amount of work
performed at exhaustion (Wlim) for a local exercise (flexions or extensions of the elbow or the knee) performed
at different constant power outputs (P) with tlim rangeing between 3 and 30 minutes. This linear tlim-Wlim
relationship corresponded to a hyperbolic relationship between tlim and P:
Wlim = a + b tlim and tlim = a/(P- b)
In this model, parameter a was equivalent to a finite energy store whereas parameter b had the meaning of a
power output which can be sustained during a long time and was called critical power. In 1958, Scherrer [5]
proposed to apply the critical power concept to running or swimming exercises (critical speed) and to interpret
the individual distance-time relationship as previously done for world records [6]. Thereafter, the concept of
critical speed was applied to world records in different sports [7].
Dlim = a + btlim = a + SCrit tlim Equation 3
The value of SCrit is significantly correlated to the running speed at the 4 mmol blood lactate (8) and the lactate
steady state running speed [9]. The SCrit/MAS ratio is considered as an index of aerobic endurance, that is, the
ability to sustain a high percentage of MAS during a long time. Therefore, exponent has been compared with
SCrit/MAS.
II. Methods
The individual power laws between Dlim and tlim were determined by computing the regressions between the
natural logarithms of Dlim and tlim:
ln(Dlim) = ln(k) + ln(tlim) and k = e ln(k)
The short distances (< 1500 m) were not included in the present study. Indeed, it has been showed that the
slopes of the regressions between velocity and the logarithm of time were different for races under and beyond
150-180 s [10]. This difference was the expression of the switch from the anaerobic metabolism that is needed
for sprints to the aerobic metabolism used to supply energy for long distance races [10].
Figure 2: in A, three examples of the tlim-Dlim relationships in elite runners. In B, relationships between the
logarithms of tlim and Dlim in two runners.
As the value of SCrit depend on the range of tlim [11], the individual values of SCrit were estimated for
the same ranges of tlim by computing the values of Dlim corresponding to the same pairs of tlim (tlim1 and tlim2)
from the individual power laws. The values of SCrit were computed for different pairs of tlim (3-12.5 min, 3-14
min and 3-16 min). Thereafter, it was assumed that MAS corresponds to the maximal velocity that can be
sustained over 7 min (420 s) in elite endurance runners [12]. Therefore, the individual values of MAS in the
present study were assumed to correspond to S420 and were computed from the individual Dlim-tlim power laws:
MAS = S420 = k (420) - 1
Finally, the values of SCrit were normalized to MAS:
SCrit/MAS = SCrit/S420
III. Results
The errors between the actual values of Dlim and their predicted values from Kennelly’s model were
lower than 2%. As an example, the data of Gebrsellassie are presented in Figure 3. The higher error (0.99 %)
corresponded to the performance on 10,000 m. However, when the data of half-marathon and marathon were not
included in the computation of the power law, the results of Gebrsellassie were slightly different: = 0.95
instead of 0.94 and k = 9.12 instead of 9.72.](https://image.slidesharecdn.com/b07011216-170214112337/85/Application-of-Kennelly-model-of-Running-Performances-to-Elite-Endurance-Runners-2-320.jpg)
![Application of Kennelly’model of Running Performances to Elite Endurance Runners.
15
2. The effect of the underestimations of the performances in either the shortest or the longest distances on the
linear ln(tlim)-ln(Dlim) relationship (Fig. 2B). An underestimation of the shortest distances induces a decrease
in k and an increase in . Inversely, an underestimation of the longest distances induces an increase in k and
a decrease in .
Figure 5: comparison of the running performances of Ovett and Coe (A) and their parameters k and (B).
The study of the performance of world elite runners is interesting. Indeed, the interpretation of the
Kennelly’s model assumes that the running data correspond to the maximal performance for each distance. The
best performances of world elite runners generally correspond to the results of many competitions against other
elite runners. The motivation is probably optimal during these races. However, it is assumed that the
performances correspond to the same training and the same fitness level. Therefore, all the performances must
be achieved within a few years. The differences in the results concerning Gebrselassie when half-marathon and
marathon were not included suggest that this model is not perfect and cannot describe a very large range of
distances. However, these differences could also be explained by the effects of age and ground (track vs road,
slopes ...). The comparison of Ovett and Coe is also an illustration of the limits of the Kennelly’s model. Indeed,
the differences between Ovett and Coe for the performances in 1500 and 2000 m are around 1 second but the
inclusion of longer distances (3000 m and 5000 m) induces large differences in the values of k and (Fig. 5). In
fact, Ovett and Coe were the best runners on 800 and 1500 m but, as suggested [10], the performances in 800 m
were not included in the model because they largely depend on anaerobic metabolism. The best performance for
a given distance is maximal if the elite runner has run this distance many times, which was probably not the case
for the distances equal to 3000 m and 5000 m, for Ovett and Coe. However, the difference between Ovett and
Coe was not very high (0.907 vs 0.851) and corresponded to equal to 0.879 ± 0.028, i.e. ± 3.2%.
V. Conclusions
Kennelly’s model can be used to describe the individual performances of elite runners with a high
accuracy. There is a linear relationship between parameters k and exponents of the elite runners. Exponent
can be considered as a dimensionless index of aerobic endurance which is close to 1 in the best endurance
runners. In elite endurance runners, exponent is equal to the normalized critical speed (SCrit/MAS) computed
from exhaustion times equal to 3 and 12.5 min. However, further studies should verify 1) that Kennelly’s model
can accurately describe the individual running performance for a large range of distance; 2) that this model can
be used in all the runners.
References
[1]. Kennelly, An Approximate Law of Fatigue in the Speeds of Racing Animals. Proceedings of the American Academy of Arts and
Sciences, 42, 1906, 275-331.
[2]. Katz, J.S., Katz, L. Power laws and athletic performance. J Sports Sciences. 17, 1999, 467–76.
[3]. Savaglio, S., Carbone, V. Human performance: Scaling in athletic world records. Nature. 404, 2000, 244.
[4]. Scherrer J, Samson M, Paléologue A Etude du travail musculaire et de la fatigue. Données ergométriques obtenues chez l'homme. J
Physiol Paris:46, 1954, 887-916.
[5]. Scherrer J. Applications aux épreuves sportives et à l’exercice physique des notions de travail et temps-limite. Méd Educ Phys
Sport. 32, 1958, 7-12
[6]. Hill AV Muscular movement in man: the factors governing speed and recovery from fatigue. (McGraw-Hill. New York &
London,1927)
[7]. Ettema, J.H. Limits of human performance and energy-production. Int Zeitschrift Angewandte Physiol Einschliesslich
Arbeitsphysiol. 22, 1966, 45–54.](https://image.slidesharecdn.com/b07011216-170214112337/85/Application-of-Kennelly-model-of-Running-Performances-to-Elite-Endurance-Runners-4-320.jpg)
![Application of Kennelly’model of Running Performances to Elite Endurance Runners.
16
[8]. Lechevalier, J.M., Vandewalle, H., Chatard, J.C., Moreaux, A., Gandrieux, V., Besson, F., Monod, H. Relationship between the 4
mmol running velocity, the time-distance relationship and the Léger-Bouchers test. Arch. Int. Physiol. Biochim. Biophys. 97, 1989,
355–360.
[9]. Sid-Ali, B., Vandewalle, H., Chaïr, K., Moreaux, A., Monod, H. Lactate steady state velocity and distance-exhaustion time
relationship in running. Arch. Int. Physiol. Biochim. Biophys. 99, 1991, 297–301.
[10]. Carbone, V., Savaglio, S. Scaling laws and forecasting in athletic world records. J Sports Sciences. 19, 2001, 477–484.
[11]. Hill DW The critical power concept. A review. Sports Med 16, 1993, 237–254.
[12]. Péronnet, F., Thibault, G. Mathematical analysis of running performance and world running records. J Applied Physiol 67, 1989,
453–65.](https://image.slidesharecdn.com/b07011216-170214112337/85/Application-of-Kennelly-model-of-Running-Performances-to-Elite-Endurance-Runners-5-320.jpg)
Application of Kennelly’model of Running Performances to Elite Endurance Runners.
- 1. Research Inventy: International Journal of Engineering And Science Vol.7, Issue 1 (January 2017), PP -12-16 Issn (e): 2278-4721, Issn (p):2319-6483, www.researchinventy.com 12 Application of Kennelly’model of Running Performances to Elite Endurance Runners. Vandewalle Henry, Md, Phd Faculty of Medicine, University Paris XIII, Bobigny France Abstract: The model of Kennelly between distance (Dlim) and exhaustion time (tlim) has been applied to the individual performances of 19 elite endurance runners (World-record holders and Olympic winners) from P. Nurmi (1920-1924) to M. Farah (2012) whose individual best performances on several different distances are known. Kennelly’s model (Dlim = k tlim ) can describe the individual performances of elite runners with a high accuracy (errors lower than 2 %). There is a linear relationship between parameters k and exponents of the elite runners and the extreme values correspond to S. Coe (k = 15.8; = 0.851) and E. Zatopek (k = 6.57; = 0.984). Exponent can be considered as a dimensionless index of aerobic endurance which is close to 1 in the best endurance runners. If it is assumed than maximal aerobic speed can be maintained 7 min in elite endurance runners, exponent is equal to the normalized critical speed (critical speed/maximal aerobic speed) computed from exhaustion times equal to 3 and 12.5 min in these runners. I. Introduction In 1906, Kennelly [1] studied the relationship between running distance (Dlim) and the time of the world records (tlim) and proposed a power law of fatigue for the different types of exercise in humans and horses: Dlim = k tlim Equation 1 The value of exponent in power law is independent of scaling [2, 3]: the value of is independent of the expression of tlim and Dlim. It is possible that is an expression of the endurance capability. Indeed, it is likely that the curvature of the tlim-Dlim relationship depends on the decrease in the fraction of maximal aerobic metabolism that can be sustained during long lasting exercises. The tlim-Dlim relationship is linear when is equal to 1. When Dlim is normalized to the value of Dlim at maximal aerobic speed (MAS), it can be demonstrated that the slope of the tlim-Dlim curve is equal to (Fig. 1A) for tlim equal to the exhaustion time at MAS (tMAS). At a running speed equal to MAS, Dlim = k tMAS = MAS tMAS and therefore, k = MAS*tMAS 1- Equation 2 Figure 1: in A relationship between tlim normalized to tMAS and Dlim normalized to Dlim at MAS: the slope (red dotted line) of Kennelly curve (black curve) is equal to for tlim/tMAS = 1. In B, the value of the critical speed (red continuous line) computed from tlim1 and tlim2 is equal to the slope of Kennelly curve at tMAS (red dotted line).
- 2. Application of Kennelly’model of Running Performances to Elite Endurance Runners. 13 In 1954, Scherrer et al. [4] proposed a linear relationship between tlim and the total amount of work performed at exhaustion (Wlim) for a local exercise (flexions or extensions of the elbow or the knee) performed at different constant power outputs (P) with tlim rangeing between 3 and 30 minutes. This linear tlim-Wlim relationship corresponded to a hyperbolic relationship between tlim and P: Wlim = a + b tlim and tlim = a/(P- b) In this model, parameter a was equivalent to a finite energy store whereas parameter b had the meaning of a power output which can be sustained during a long time and was called critical power. In 1958, Scherrer [5] proposed to apply the critical power concept to running or swimming exercises (critical speed) and to interpret the individual distance-time relationship as previously done for world records [6]. Thereafter, the concept of critical speed was applied to world records in different sports [7]. Dlim = a + btlim = a + SCrit tlim Equation 3 The value of SCrit is significantly correlated to the running speed at the 4 mmol blood lactate (8) and the lactate steady state running speed [9]. The SCrit/MAS ratio is considered as an index of aerobic endurance, that is, the ability to sustain a high percentage of MAS during a long time. Therefore, exponent has been compared with SCrit/MAS. II. Methods The individual power laws between Dlim and tlim were determined by computing the regressions between the natural logarithms of Dlim and tlim: ln(Dlim) = ln(k) + ln(tlim) and k = e ln(k) The short distances (< 1500 m) were not included in the present study. Indeed, it has been showed that the slopes of the regressions between velocity and the logarithm of time were different for races under and beyond 150-180 s [10]. This difference was the expression of the switch from the anaerobic metabolism that is needed for sprints to the aerobic metabolism used to supply energy for long distance races [10]. Figure 2: in A, three examples of the tlim-Dlim relationships in elite runners. In B, relationships between the logarithms of tlim and Dlim in two runners. As the value of SCrit depend on the range of tlim [11], the individual values of SCrit were estimated for the same ranges of tlim by computing the values of Dlim corresponding to the same pairs of tlim (tlim1 and tlim2) from the individual power laws. The values of SCrit were computed for different pairs of tlim (3-12.5 min, 3-14 min and 3-16 min). Thereafter, it was assumed that MAS corresponds to the maximal velocity that can be sustained over 7 min (420 s) in elite endurance runners [12]. Therefore, the individual values of MAS in the present study were assumed to correspond to S420 and were computed from the individual Dlim-tlim power laws: MAS = S420 = k (420) - 1 Finally, the values of SCrit were normalized to MAS: SCrit/MAS = SCrit/S420 III. Results The errors between the actual values of Dlim and their predicted values from Kennelly’s model were lower than 2%. As an example, the data of Gebrsellassie are presented in Figure 3. The higher error (0.99 %) corresponded to the performance on 10,000 m. However, when the data of half-marathon and marathon were not included in the computation of the power law, the results of Gebrsellassie were slightly different: = 0.95 instead of 0.94 and k = 9.12 instead of 9.72.
- 3. Application of Kennelly’model of Running Performances to Elite Endurance Runners. 14 Figure 3: application of Kennely’s model to the performances of Gebrselassie including half-marathon and marathon and percentages of errors on Dlim. There is a linear relationship between parameters k and exponents of the elite runners (Fig.4 A). The extreme values correspond to S. Coe (k = 15.8; = 0.851) and E. Zatopek (k = 6.57; = 0.984). Figure 4: in A, relationship between the individual values of k and exponent in the 19 elite runners; in B, relationships between the individual values of exponent and SCrit/MAS computed from different pairs of tlim (3- 12.5, 3-14 and 3-16 min). The correlation coefficients of the linear regressions between and SCrit/MAS were > 0.999 for the different ranges of tlim in Fig. 4B (3-12.5, 3-14 and 3-16 min). The regression line was close to the identity line for tlim1 and tlim2 equal to 3 to 12.5 min, respectively. IV. Discussion The results of the present study indicate that Kennelly’s model can describe individual performances of elite runners with a high accuracy because the errors in Dlim were lower than 2 %. The value of is close to 1 (0.984) for Zatopek and lower (0.851) for Coe who was a middle distance runner. In Fig. 2A, the tlim-Dlim curve was almost linear for Gebrselassie in contrast with Coe. The hypothesis that exponent can be considered as a dimensionless index of aerobic endurance is confirmed by the linear relationship between and SCrit/MAS that were almost equal when SCrit was computed from two values of tlim equal to 3 and 12.5 min. The negative slope of the relationship between parameters k and (Fig. 4A) has two origins: 1. physiological origins because k = MAS*tMAS 1- (2) It is likely that the best runners on “short” distances (1500 m) have larger maximal anaerobic capacities and, consequently, higher values of tMAS. On the other hand, the best performers in very long distance (20 km, marathon) have probably higher percentages of slow fibers and higher values of exponent . Consequently, these long distance endurance runners should have lower values of tMAS 1- and, therefore, lower values of k.
- 4. Application of Kennelly’model of Running Performances to Elite Endurance Runners. 15 2. The effect of the underestimations of the performances in either the shortest or the longest distances on the linear ln(tlim)-ln(Dlim) relationship (Fig. 2B). An underestimation of the shortest distances induces a decrease in k and an increase in . Inversely, an underestimation of the longest distances induces an increase in k and a decrease in . Figure 5: comparison of the running performances of Ovett and Coe (A) and their parameters k and (B). The study of the performance of world elite runners is interesting. Indeed, the interpretation of the Kennelly’s model assumes that the running data correspond to the maximal performance for each distance. The best performances of world elite runners generally correspond to the results of many competitions against other elite runners. The motivation is probably optimal during these races. However, it is assumed that the performances correspond to the same training and the same fitness level. Therefore, all the performances must be achieved within a few years. The differences in the results concerning Gebrselassie when half-marathon and marathon were not included suggest that this model is not perfect and cannot describe a very large range of distances. However, these differences could also be explained by the effects of age and ground (track vs road, slopes ...). The comparison of Ovett and Coe is also an illustration of the limits of the Kennelly’s model. Indeed, the differences between Ovett and Coe for the performances in 1500 and 2000 m are around 1 second but the inclusion of longer distances (3000 m and 5000 m) induces large differences in the values of k and (Fig. 5). In fact, Ovett and Coe were the best runners on 800 and 1500 m but, as suggested [10], the performances in 800 m were not included in the model because they largely depend on anaerobic metabolism. The best performance for a given distance is maximal if the elite runner has run this distance many times, which was probably not the case for the distances equal to 3000 m and 5000 m, for Ovett and Coe. However, the difference between Ovett and Coe was not very high (0.907 vs 0.851) and corresponded to equal to 0.879 ± 0.028, i.e. ± 3.2%. V. Conclusions Kennelly’s model can be used to describe the individual performances of elite runners with a high accuracy. There is a linear relationship between parameters k and exponents of the elite runners. Exponent can be considered as a dimensionless index of aerobic endurance which is close to 1 in the best endurance runners. In elite endurance runners, exponent is equal to the normalized critical speed (SCrit/MAS) computed from exhaustion times equal to 3 and 12.5 min. However, further studies should verify 1) that Kennelly’s model can accurately describe the individual running performance for a large range of distance; 2) that this model can be used in all the runners. References [1]. Kennelly, An Approximate Law of Fatigue in the Speeds of Racing Animals. Proceedings of the American Academy of Arts and Sciences, 42, 1906, 275-331. [2]. Katz, J.S., Katz, L. Power laws and athletic performance. J Sports Sciences. 17, 1999, 467–76. [3]. Savaglio, S., Carbone, V. Human performance: Scaling in athletic world records. Nature. 404, 2000, 244. [4]. Scherrer J, Samson M, Paléologue A Etude du travail musculaire et de la fatigue. Données ergométriques obtenues chez l'homme. J Physiol Paris:46, 1954, 887-916. [5]. Scherrer J. Applications aux épreuves sportives et à l’exercice physique des notions de travail et temps-limite. Méd Educ Phys Sport. 32, 1958, 7-12 [6]. Hill AV Muscular movement in man: the factors governing speed and recovery from fatigue. (McGraw-Hill. New York & London,1927) [7]. Ettema, J.H. Limits of human performance and energy-production. Int Zeitschrift Angewandte Physiol Einschliesslich Arbeitsphysiol. 22, 1966, 45–54.
- 5. Application of Kennelly’model of Running Performances to Elite Endurance Runners. 16 [8]. Lechevalier, J.M., Vandewalle, H., Chatard, J.C., Moreaux, A., Gandrieux, V., Besson, F., Monod, H. Relationship between the 4 mmol running velocity, the time-distance relationship and the Léger-Bouchers test. Arch. Int. Physiol. Biochim. Biophys. 97, 1989, 355–360. [9]. Sid-Ali, B., Vandewalle, H., Chaïr, K., Moreaux, A., Monod, H. Lactate steady state velocity and distance-exhaustion time relationship in running. Arch. Int. Physiol. Biochim. Biophys. 99, 1991, 297–301. [10]. Carbone, V., Savaglio, S. Scaling laws and forecasting in athletic world records. J Sports Sciences. 19, 2001, 477–484. [11]. Hill DW The critical power concept. A review. Sports Med 16, 1993, 237–254. [12]. Péronnet, F., Thibault, G. Mathematical analysis of running performance and world running records. J Applied Physiol 67, 1989, 453–65.