Effects of wheels and tires on high-strength lightweight wheelchair propulsion cost using a robotic wheelchair tester

Figures & data

Figure 1. The robotic anatomical model propulsion system wheelchair testbed.

The robotic wheelchair tester is seated on a manual wheelchair with its arms positioned to grip the push-rims. Motors at the end of each arm have small pinions; these are meshed into custom toothed push-rims on each drive wheel. Large plastic encoders are fastened to the centre of each drive wheel.

Table 1. Wheel properties.

Wheel name	Hub view	Diameter [cm]	Tire width [cm]	Mass [kg]	Tire hardness (Shore A)
8"×1" Unbranded [Default]		20.3	2.4	0.38	72
6"×1.5" Primo		14.7	3.7	0.33	89
24"×1" Solid Mag [Default]		62.0	2.5	2.07	82
24"×1-3/8" Primo XPress		60.0	3.5	2.06	77

Figure 2. Cross-sectional tire profiles.

Cross-sectional profile views of the tires highlight differences in the tread depth and overall contact patches between components. The default casters have pronounced ridges along the perimeter of the tire. Primo casters are wider with a gently-rounded profile. The default drive tires have a trapezoidal shape with a flat, tread-less contact patch. Primo tires have a rounded profile with small grooves.

Table 2. Configurations used to assess performance.

Config. name	Drive wheel	Caster	System mass	Weight distribution over drive wheel axle (%)
“A”	Solid Mag 24"×1"	Drive Medical 8"×1"	96.1 kg	63.8
“B”	Solid Mag 24"×1"	Primo 6"×1.5"	95.8 kg	63.5
“C”	Primo Xpress 24"×1-3/8"	Drive Medical 8"×1"	95.8 kg	64.9
“D”	Primo Xpress 24"×1-3/8"	Primo 6"×1.5"	95.5 kg	64.0

Figure 3. Torque profile for the Straight manoeuvre on tile.

Line chart of wheel torques over time for one trial of the Straight manoeuvre. Each of the seven pushes are applied to both wheels simultaneously. The first push applies 14 Newton-meters of torque over 1.05 s, followed by two pushes with progressively smaller torques and shorter durations. The remaining four pushes have lower torques of 8 Newton-meters applied over 0.55 s durations. Total duration of the manoeuvre is 20 s including a deceleration period.

Figure 4. Torque profile for the Slalom manoeuvre on tile.

Line chart of wheel torques over time for one trial of the Slalom manoeuvre. The first of five pushes applies 6 Newton-meters of torque to both wheels. The remaining four pushes apply 12 Newton-meters of torque to only one wheel at a time, changing sides each push to create the serpentine-like slalom path. The total duration of the manoeuvre is 20 s including a deceleration period.

Table 3. Coast-down deceleration values over tile and carpet.

		“A” solid DW; 8" caster	“B” solid DW; 6" caster	“C” Primo DW; 8"caster	“D” Primo DW; 6" caster
Mean decel. (m/s^2) Tile surface	Mean (± StDev)	0.156 (±0.0186)	0.109 (±0.0098)	0.123 (±0.0098)	0.089 (±0.0099)
	Effect size	–	3.12	2.19	5.13
	% Change	–	−29.7	−20.8	−49.0
	p-Value	–	0.00*	0.00*	0.00*
Mean decel. (m/s^2) Carpet surface	Mean (± StDev)	0.229 (±0.0118)	0.215 (±0.0140)	0.195 (±0.0233)	0.187 (±0.0071)
	Effect Size	–	1.05	1.86	4.35
	% Change	–	−5.9	−15.0	−18.5
	p-Value	–	0.00*	0.00*	0.00*

*Significantly different (p < 0.05) from the mean of the “A” configuration.

Table 4. Cost and distance during straight manoeuvre on tile.

	“A” solid DW; 8" caster	“B” solid DW; 6" caster	“C” Primo DW; 8"caster	“D” Primo DW; 6" caster
Cost (J/m)
Mean (± StDev)	22.80 (±1.82)	17.23 (±0.67)	17.57 (±0.89)	13.51 (±1.14)
Effect Size	–	4.10	3.70	6.10
% Change	–	−24.40	−22.94	−40.72
p-value	–	0.00*	0.00*	0.00*
Distance (m)
Mean (± StDev)	2.11 (±0.63)	4.10 (±0.41)	4.28 (±0.39)	6.98 (±0.07)
Effect Size	–	3.73	4.13	10.82
% Change	–	+93.99	+102.37	+230.17
p-value	–	0.00*	0.00*	0.00*

*Significantly different (p < 0.05) from the mean of the “A” configuration.

Figure 5. Superiority test results for the Straight manoeuvre over tile.

Propulsion cost differences between the default and up-charged configurations are plotted as 95% confidence intervals on a forest plot. The equivalence interval is shown as ±1.14 Joules per metre. All three intervals of propulsion cost differences fall completely below the lower equivalence threshold. Configurations B and C have lower costs than A by at least 4 Joules per metre. Configuration D is lower by at least 7 Joules per metre.

Table 5. Cost distance, and yaw displacement during the Slalom manoeuvre on tile.

	“A” solid DW; 8" caster	“B” solid DW; 6" caster	“C” Primo DW; 8"caster	“D” Primo DW; 6" caster
Cost (J/m)
Mean (± StDev)	19.56 (±0.20)	17.00 (±0.25)	16.75 (±0.29)	14.54 (±0.83)
Effect size	–	11.48	11.35	8.31
% Change	–	−13.08	−14.38	−25.69
p-Value	–	0.00*	0.00*	0.00*
Distance (m)
Mean (± StDev)	1.24 (±0.07)	2.89 (±0.18)	2.96 (±0.23)	5.78 (±0.27)
Effect size	–	12.32	10.28	23.04
% Change	–	+133.68	+138.99	+367.64
p-Value	–	0.00*	0.00*	0.00*
Yaw (rad)
Mean (± StDev)	0.45 (±0.05)	1.69 (±0.10)	1.42 (±0.15)	2.66 (±0.17)
Effect size	–	15.65	8.80	17.26
% Change	–	+278.36	+217.14	+495.06
p-Value	–	0.00*	0.00*	0.00*

*Significantly different (p < 0.05) from the mean of the “A” configuration.

Figure 6. Superiority test results for the Slalom manoeuvre over tile.

Propulsion cost differences between the default and up-charged configurations are plotted as 95% confidence intervals on a forest plot. The equivalence interval is shown as ±0.98 Joules per metre. All three intervals of propulsion cost differences fall completely below the lower equivalence threshold. Configurations B and C have lower costs than A by at least 2 Joules per metre. Configuration D is lower by at least 5 Joules per metre.

Table 6. Cost and distance during the Straight manoeuvre on carpet.

	“A” solid DW; 8" caster	“B” solid DW; 6" caster	“C” Primo DW; 8"caster	“D” Primo DW; 6" caster
Cost (J/m)
Mean (± StDev)	28.04 (±1.32)	26.84 (±1.05)	24.52 (±1.07)	23.13 (±1.05)
Effect size	–	1.00	2.93	4.12
% Change	–	−4.28	−12.57	−17.53
p-Value	–	0.04*	0.00*	0.00*
Distance (m)
Mean (± StDev)	3.52 (±0.45)	4.30 (±0.70)	5.52 (±0.62)	6.21 (±0.72)
Effect size	–	1.33	3.67	4.50
% Change	–	+22.15	+56.70	+76.38
p-Value	–	0.01*	0.00*	0.00*

*Significantly different (p < 0.05) from the mean of the “A” configuration.

Figure 7. Superiority test results for the Straight manoeuvre over carpet.

Propulsion cost differences between the default and up-charged configurations are plotted as 95% confidence intervals on a forest plot. The equivalence interval is shown as ±1.40 Joules per metre. The interval for Configuration B is within the equivalence interval. The other two intervals fall completely below the lower equivalence threshold. Configuration C has lower cost than A by at least 3 Joules per metre. Configuration D is lower by at least 4.5 Joules per metre.

Table 7. Cost, distance, and yaw displacement during the Slalom manoeuvre on carpet.

	“A” solid DW; 8" caster	“B” solid DW; 6" caster	“C” Primo DW; 8"caster	“D” Primo DW; 6" caster
Cost (J/m)
Mean (± StDev)	27.94 (±1.85)	28.74 (±2.59)	26.43 (±2.75)	26.38 (±2.59)
Effect Size	–	0.35	0.64	0.69
% Change	–	+2.85	−5.40	−5.57
p-Value	–	0.27	0.05*	0.04*
Distance (m)
Mean (± StDev)	1.47 (±0.32)	1.96 (±0.68)	2.41 (±1.05)	2.98 (±1.11)
Effect size	–	0.93	1.21	1.85
% Change	–	+33.30	+63.67	+102.53
p-Value	–	0.01*	0.00*	0.00*
Yaw (rad)
Mean (± StDev)	0.89 (±0.23)	1.72 (±0.46)	1.68 (±0.65)	2.64 (±0.45)
Effect size	–	2.29	1.61	4.87
% Change	–	+93.18	+88.24	+196.55
p-Value	–	0.00*	0.00*	0.00*

*Significantly different (p < 0.05) from the mean of the “A” configuration.

Figure 8. Superiority test results for the Slalom manoeuvre over carpet.

Propulsion cost differences between the default and up-charged configurations are plotted as 95% confidence intervals on a forest plot. The equivalence interval is shown as ±1.40 Joules per metre. The range of cost differences between Configurations B and A are completely within the equivalence interval. The ranges for Configurations C and D have upper bounds within the equivalence interval. Their point estimates are outside of the interval bounds.

Table 8. Component cost comparisons from online retailers.

Component	Cost per pair ($)
8"×1" caster	23
6"×1.5" Primo caster	42
24"×1" solid tire (pyramidal)	34
24"×1-3/8" Primo XPress	48