Research Papers

Effects of Bladder Geometry in Pneumatic Artificial Muscles

[+] Author and Article Information
Erick Ball

Laboratory for Intelligent Machine Systems,
Cornell University,
Ithaca, NY 14853
e-mail: ejb279@cornell.edu

Ephrahim Garcia

Laboratory for Intelligent Machine Systems,
Cornell University,
Ithaca, NY 14853

1Corresponding author.

Manuscript received December 19, 2013; final manuscript received March 21, 2016; published online August 5, 2016. Assoc. Editor: Venketesh Dubey.

J. Med. Devices 10(4), 041001 (Aug 05, 2016) (11 pages) Paper No: MED-14-1004; doi: 10.1115/1.4033325 History: Received December 19, 2013; Revised March 21, 2016

Designing optimal pneumatic muscles for a particular application requires an accurate model of the hyperelastic bladder and how it influences contraction force. Previous work does not fully explain the influence of bladder prestrain on actuator characteristics. We present here modeling and experimental data on the actuation properties of artificial muscles constructed with varying bladder prestrain and wall thickness. The tests determine quasi-static force–length relationships during extension and contraction, for muscles constructed with unstretched bladder lengths equal to 55%, 66%, and 97% of the stretched muscle length and two different wall thicknesses. Actuator force and maximum contraction length are found to depend strongly on both the prestrain and the thickness of the rubber, making existing models inadequate for choosing bladder geometry. A model is presented to better predict force–length characteristics from geometric parameters, using a novel thick-walled tube calculation to account for the nonlinear elastic properties of the bladder. It includes axial force generated by stretching the bladder lengthwise, and it also describes the hoop stress created by radial expansion of the muscle that partially counteracts the internal fluid pressure exerted outward on the mesh. This effective reduction in pressure affects both axial muscle force and mesh-on-bladder friction. The rubber bladder is modeled as a Mooney–Rivlin incompressible solid. The axial force generated by the mesh is found directly from contact forces rather than from potential energy. Modeling the bladder as a thin-walled tube gives a close match to experimental data on wall thickness, but a thick-walled bladder model is found to be necessary for explaining the effects of prestrain.

Copyright © 2016 by ASME
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Fig. 1

McKibben muscle actuation: (a) unpressurized muscle at its relaxed length and (b) the pressurized muscle expands in diameter but contracts in length

Grahic Jump Location
Fig. 2

Sleeve geometry: (a) helically wound strand forming part of the sleeve cylinder and (b) unwound strand showing how the dimensions are related geometrically. Adapted from Tiwari et al. [24].

Grahic Jump Location
Fig. 3

Normalized tension force as a function of actuator length from the volume potential energy model

Grahic Jump Location
Fig. 4

Bladder failure mode

Grahic Jump Location
Fig. 5

Calculation of hoop stress in the bladder

Grahic Jump Location
Fig. 6

Normalized force versus length, thin-walled bladder model. Length is normalized by the strand length b.

Grahic Jump Location
Fig. 7

Thin-wall model of full actuation cycle with (a) 0% prestrain and (b) 50% prestrain

Grahic Jump Location
Fig. 8

(a)-(e) Actuation curves for five different bladder lengths, from muscle length to 1/3 of muscle length, using thick-walled model. (f) The net work (area between the curves) for the five cases as a function of bladder length.

Grahic Jump Location
Fig. 9

Bladder tensile test results. Thick tube data are from the bladder type used for muscles A, B, and C. Thin tube bladder was used for muscle D. The Mooney–Rivlin curve shows theoretical stress from a uniaxial Mooney–Rivlin model with C1 = 161 kPa and C2 = 39 kPa [31].

Grahic Jump Location
Fig. 10

Selected data from tensile tests of the four test muscles A, B, C, and D. Lengths are normalized by stretched muscle length, not strand length: ((a)–(d)) unpressurized and ((e)–(h)) pressurized.

Grahic Jump Location
Fig. 11

Thick-walled model compared to data for muscle A, thick bladder with 3% prestrain: (a) muscle A: 2 kPa, (b) muscle A: 335 kPa, (c) muscle A: 555 kPa, and (d) muscle A: 671 kPa

Grahic Jump Location
Fig. 12

Thick-walled model compared to data for muscle B, thick bladder with 52% prestrain: (a) muscle B: 1 kPa, (b) muscle B: 279 kPa, (c) muscle B: 420 kPa, and (d) muscle B: 607 kPa

Grahic Jump Location
Fig. 13

Thick-walled model compared to data for muscle C, thick bladder with 82% prestrain: (a) muscle C: 1 kPa, (b) muscle C: 275 kPa, (c) muscle C: 487 kPa, and (d) muscle C: 675 kPa

Grahic Jump Location
Fig. 14

Thick-walled model compared to data for muscle D, thin bladder with 2% prestrain (a) muscle D: 3 kPa, (b) muscle D: 88 kPa, (c) muscle D: 170 kPa, and (d) muscle D: 318 kPa




Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In