A Comparison Of Five Mechanical Work Algorithms For Different Footstrike Patterns And Speeds During Distance Running

  • M M. Slavin
  • R. A. Hintermeister
  • J. Hamill
Keywords: distance running, algorithm, energy calculation


The mechanical work done by a runner during an average stride cyde has been calculated with a variety of algorithms that generate values that may vary by an order of magnitude. The application of different algorithms to the same data set is uncommon, and does not seem to have been used at all to compare different foot strike patterns (FSP) during distance running. Average stride cycle values from five work algorithms for forefoot strike (ffs) and heel strike (hs) running at three different running speeds are presented. In general order from most to least restrictive: Wn allows no transfer between segments; Ww, within-segment transfer only; WwbAS, transfer within and between adjacent segments only; WwbLT, within and between segments of the same limb and the trunk; and, Wwb, within- and between-segment transfer with no restrictions. The primary difference in these algorithms is the amount of energy transfer they permit between and among body segments. Twelve highly skilled, male distance runners each ran with both FSP at three speeds ranging from 3.58 to 4.58 m-s-l. High-speed video (200 Hz) was used to track eight segment endpoint markers in the left sagittal plane. An ll-segment model was used with symmetry assumed to generate right side values. Among the algorithms, the no-transfer method (Wn) produced the highest work estimates. An absolute difference of -300 joules-stride-1 (-15-20%) existed across speeds between the no-transfer and within-transfer algorithms. There was then a relatively large decrease to the span of values generated from the other three algorithms. WwbAS was slightly higher than the remaining two algorithms, moreso in relative terms as speed increased. WwbLT increased slightly over speed (-40% slow->fast), while Wwb, the least restrictive, demonstrated almost no change across speeds (-1 % slow->fast). On average, these differences converged absolutely (75->20 joules-stride-1) and relatively (9.8%->2.5%) with increased speed; i.e., differences between the two .FSP decreased as speed increased. At all speeds for each algorithm, hs was lower than ffs. Collapsed across speeds, hs as percentage of ffs was 96.7 (Wn), 96.5 (ww)- 96.7 (WwbAS), 95.8 (WwbLT) and 89.4% (Wwb). Wwb across speeds consistently showed the largest relative differences between FSP, due perhaps in part to low absolute values. However, FSP differences still decreased with increased speed. This algorithm, therefore, appears to preserve the ordinal relationship and the trend in relative change between FSP across speeds reflected in the other four algorithms. Overall, the consistency across all algorithms of absolute and relative decrease between FSP with increased speed suggests variations in actual kinematics, not algorithms, are responsible for observed differences.