Conceptual Design of Fluidic Bending Actuator

Get Complete Project Material File(s) Now! »

Chapter 3 Conceptual  Design and Fabrication

As aforementioned in the literature review, combination of different materials into the same entity would cause FBA to bend towards its more rigid side. In practice, however, this concept of design would be hard to be implemented due to the fact that breakage along the borderline between different materials are easy to occur after cyclic operation. This issue could be solved by combining same material but of different mechanical properties. As a unique feature of fibre-reinforced soft actuators, the type of actuation is programmable by varying the angle between fibre braids [59], which could possibly provide a solution to this issue. According to the research in [41], fiber winding is also reported as an effective approach to reinforce soft elastic material, endowing it with attributes such as fast response, high endurance and excellent stability. In this chapter, research was focused on the development of FBA using fiber-reinforced rubber-like material. Various design schemes were explored and evaluated to find the optimal method of producing FBAs that have excellent force output, bending capability and high robustness against swelling and abrasion.



The very initial step of making a soft actuator is shaping the flexible main body of FBA, which is achieved through the design of mould. As shown in Figure 3-1, the mould was first fabricated in the form of a cylindrical comb, then variations were made based on it to evaluate the effectiveness and feasibility of different design schemes. Considering the fiber-reinforced actuator needs to be vulcanized at high temperature in a later stage, the mold was created out of iron. To provide good air sealing, initially both ends of pneumatic bending actuator (PBA) were sealed by serrated aluminum cap (Figure 3-2), which is easy to be pushed into the rubber tube, but hard to be drawn out. For the ease of assembly/disassembly and integration of PBA into a mechatronic system, the aluminum cap could be replaced by rubber, which is achieved through structuring of mold. The iron rod could be made into two parts screwed to each other. After rubber slices were wrapped around it, one part could be detached from the other, where superfluous fiber reinforced rubber could be used to seal the actuator. Replacing the aluminum cap with rubber would reduce the weight of PBA, and provide soft contact with targeted objects using its rubber-like tip.


As major ingredient to construct FBA, rubber displays hyper-elastic behavior and generates large nonlinear deformation. Its strength and modulus could be adjusted by varying the mixture ratio between the rubber base and curing agent. In order to construct the fiber-reinforced actuator, fiber yarns were wound around the outer shell of actuator to prevent rubber from bulging. As mentioned in Section 2.7.5, fibre-cutting through the soft material is reported as one of the most common sources of FBA failure. To prevent fiber cutting but retain high flexibility, the property of rubber and fiber should be well-combined to ensure its strength and elasticity is well-balanced. On one hand, a stiff FBA might be able to withhold high pressure input, but have very limited elastic energy to be converted into mechanical output. On the other hand, a soft FBA might be easy to deform, but prone to explode once pressurized. According to Figure 3-3, a large variety of fiber material was tried and tested to see the confinement effect of fiber yarn on rubber tube, such as polyester, glass, aramid and activated carbon fiber. Among them, aramid, Vectran® and Vectran UM® were selected accredited to their thermal stability at high temperature (330 °C), high modulus and low creep. Furthermore, since vulcanization of rubber requires high baking temperature (normally 130 °C ~ 190 °C), thermal stability becomes the most prominent aspect in selecting fiber materials. The best temperature and time of vulcanization was determined by comparing and analyzing preliminary experimental results.


The material winding process is the key procedure in manufacturing fiber-reinforced actuator, therefore automated and repeatable manufacturing technique needs to be introduced to replace unpredictable or imprecise manual assembly. A numerical controlled filament winding machine was employed when materials were wound around the iron comb (Figure 3-4). As such, the distance between adjacent braids, braiding angle of fiber threads, tension force in fiber threads, starting/ending position of winding, winding speed and direction of winding could all be configured systematically to ensure quality and consistency during fabrication.

Fabrication methods

Repetitive preliminary try-outs were run to explore the material and craftsmanship appropriate for production. As illustrated in Figure 3-5., an iterative method was employed to guide the design process of proposed actuator. The method of high feasibility and good repeatability was then selected to produce the proposed FBA. The fabrication of PBA was initiated with a molding process, which would determine the profile, size and structure of proposed design. Then fiber threads were wound to reinforce the soft rubber structure and control the type of actuation. Then the mixture of rubber and fiber went through the vulcanization process, which would then strengthen the bonding between different materials. The hollowed rubber tube was sealed with auxiliary components and clamped to prevent air leak. After that, the proposed PBA would be sent to go through a series of preliminary tests, such as robustness, repeatability and fatigue tests. In the end, the optimum design would be modelled, parameterized and characterized to verify its customizability and validity. Such a trial and error process were executed systematically and iteratively until a satisfactory result was achieved.

Single-layer braids sandwiched by two layers of rubber

The most intuitive method was implanting fiber braids directly in between rubber layer so as to prevent rubber from bulging; then the elastic potential energy stored in rubber would be converted into mechanical output when FBA is pressurized. The iron rod was secured on the winding machine with its two ends clamped tightly to the rotating shaft. Rubber slice of 0.5 thickness was manually wrapped around the iron rod without overlapping. Then fiber braids of around 0.5 mm diameter were wound around the internal layer of rubber. This size was selected by trial and error to find a balance between its toughness and flexibility. In the preliminary test, fiber braids were wound on the top of rubber back and forth repeatedly 5 mm apart. Then the second layer of rubber was attached to the fiber layer without overlapping. The triple-layer fiber reinforced rubber tube was coated with a layer of woven glass fabric to strengthen the bonding between different material and avert unwarranted bulging when fluidic pressure was supplied.
Notably, two types of technique were involved in winding the rubber along iron rod: overlapping and none-overlapping (Figure 3-6). Winding with overlap means overlying half of the current winding with the other half of previous winding width-wisely. Without overlapping, the rubber pieces have to be carefully aligned along the iron rod, and the tightening process with fiber-glass cloth was regarded as a must to make sure there is no seam in rubber layer. The none-overlap technique would minimize the overall weight and inertia of PBA, whereas the overlap technique would effectively reinforce the bending actuator and effectively prevent radial expansion when internal air pressure is applied. It was also observed that a PBA of thinner wall would allow PBA easy to recover to its original shape, hence a better response without hysteresis.
After winding process was completed, the fiber reinforced rubber tube underwent vulcanization process by encapsulating the complete mixture into a mold made out of stainless steel and baking at a temperature of around 160º. Alteration in fiber braiding angle could concentrate pressure to produce specific type of actuation. For instance, at a braided angle of 30º, the actuator contracted when it was pressurized; while it elongated at a braiding angle of 60º. This property of the fiber reinforced material was taken to control the output expanding force and angular deflection. In the manufacturing process, tension force of fiber winding is also important in determining the property of fiber-reinforced actuator. It has twofold influence on the property of the fiber-reinforced actuator: increase in tension leads to higher elongational stress in fiber, and hence a greater strength of FBA against internal pressurization; yet higher stress is easier to cause fiber-cutting through rubber. Iterative experiments showed that the optimum tension force was between 5.5 and 6.3 N. Notably, the distance between adjacent fiber threads and braided angle are correlated. As the braided angle decreased, the distance between adjacent threads would increase, and vice versa. Setting-up one parameter would automatically change the other. To avoid emergence of unwinding or overlapping of fiber, control of braiding angle was prioritized over distance between adjacent threads. The fiber and rubber winding techniques are illustrated in Figure 3-7. – Figure 3-11..

Double-layer braids sandwiched by multiple layers of rubber

To investigate the influence of fiber winding on PBA, multi-layer structure of PBA was designed by winding additional layers of fiber threads and rubber, and compared with the single-layer fiber reinforced samples (Figure 3-12). Significant improvement in its stiffness and toughness against pressurization was found. The multi-layer fiber reinforced actuator could withstand internal pressurization up to 1.2 MPa, which is noticeably increased compared to the triple-layer configuration (0.8~0.9MPa maximum). Yet, improvement in producing actuation was not noteworthy, a multi-layer structured actuator generated almost the same amount of contraction/elongation as their triple layered counterpart, provided the braiding angle remained the same. Moreover, the multilayered assembly brought undesired larger volume and heavier weight to design, which ended up with a bulky design when applied in soft robots.

Double-layer braids with rubber embedded in between

In the previous two types of design, delamination was found when the distance between adjacent fiber threads was decreased. A possible explanation is that the fiber layer would inhibit adhesion of rubber during vulcanization, thus causing the bond between rubber layers easy to break. To prevent delamination, an additional rubber layer was embedded in between fiber layers, as such the rubber layer would adhere to each other more tightly. Instead of winding in the form of braids, fiber was first wound in one direction with rubber layer wound above, and then another layer of fiber wound in the opposite direction. Such a technique would endow the bending actuator with more flexibility but still sufficient constrain force. Since the winding machine could only perform back and forth winding repetitively, multiple strands of fiber were attached to an iron comb to wind fiber in a single direction. To maintain the wall thickness of rubber layer constant, each layer was tightened by slices of etamine for around 10 to 15 minutes before the next layer was applied.

READ  Efficient Learning of Linear SVMs with Stochastic Gradient Descent 

Unilateral strengthened by extra fibre mesh

Since the aim of design was to create a bending actuator, different techniques were also implemented to create bending motion on the current design. One way of doing it was strengthening one side of the actuator by implanting additional layer of glass fiber. Manufacturing of fiber reinforced rubber tube remains the same as specified in the first method, except an additional layer of glass fiber was mounted before fiber winding was applied. Since only half of the pneumatic actuator was attached by this fiber, it is impossible to wind fibers around it. Hence, glass fiber was attached and tightened around the rubber tube using a piece of fiber glass cloth for around half an hour. After that, another layer of rubber was attached on the top of it by using the same vulcanization technique aforementioned (Figure 3-13).
In order to reduce the overall weight of PBA, the outer shell could be wrapped with fiber threads without rubber wall (Figure 3-14). However, this external fiber reinforcement is susceptible to unwinding during cyclic operation, and easy to wear and break. Preliminary experiments reveal that such a setup would cause fiber to drop off after a few cycles of operation, which significantly reduce the lifespan of the actuator.
At a braiding angle of 30°, the unilateral fiber reinforced PBA bent with significant radial expansion when activated; while significant elongation was found at a braiding angle of 70° with almost no change in its raidus, both of which produced desirable continous motion when being pressurized. When the braiding angle was changed, the tip deflection and elongation/contraction of PBA at the same input pressure was also changed correspondingly, making the type of actuaiton programmable and customizable. These unique nature and combined types of actuation could be meaningful in some specific applications. For example, a soft robot could be built to grow and bend simultaneously to navigate its environment. However, in terms of efficency in using compressed air, the proposed design cannot concentrate pressure to produce bending and a significant proportion of energy is unexploited by creating longitudinal or radial deformation. Hence, more effective approach needs to be taken to further enhance the performance of PBA.

Variation in cross section

In order to enhance the bending and force capability of PBA, variation in cross-sectional area was also explored. An asymmetrical structure was first explored by shaping the mandrel into a semi-circle cylinder (Figure 3-15), where wrapping of rubber pieces and windings of fiber underwent the same process as the procedure specified in Section 3.3.1. The external surface of PBA still approximated a circle, with the flat side of semi-circle was reinforced by an additional layer of fiber mesh. During the manufacturing process, rubber slice was hard to be evenly distributed around the mandrel especially after the winding and vulcanizing process. Such an asymmetrical design did not produce bending motion, since the semi-circular inner cavity would approximate to a full circle when PBA was pressurized, which then could not contribute to an asymmetrical configuration. To improve it, a rubber tube in the form of half cylinder was created by molding. During the manufacturing process, rubber spilt over mold, and instantly cause the position of rubber tube deviate from its central axis (Figure 3-16). PBA manufactured under this process would exhibit a slightly better bending output than those produced in Section 3.3.4, except more uncertainties were found during manufacturing.

Different braided angle

Instead of making the embedded fiber weave of homogeneous braided angle, its contractile and extended sides are reinforced with fiber of different braided angle. The fabrication process has three main stages, which is illustrated in Figure 3-17. Initially, two separate fiber reinforced rubber tubes with different fiber braid angle were prepared independently (Figure 3-17 (a)). Each tube was composed of two rubber layers and one layer of fiber braiding. To fabricate each tube, the first rubber layer was wrapped around a steel mandrel, with its length and radius determined by the dimensions of the mandrel. Then fiber was wound around the rubber tube at a predefined braiding angle (i.e. θ1) set in the winding machine. The tension force applied during fiber winding was selected in such a way that the fiber reinforced PBA would have resistance against swelling and abrasion, but the fibers would neither cut into the rubber nor the actuator delaminate later. Then a second layer of rubber was wound around the fiber thread. This tube fabrication process was repeated to create a second tube with the other required braiding angle (θ2). Each fabricated tube was then slit lengthwise into half. In the next step (shown in Figure 3-17 (b)), these two half-tubes, one with braiding angle θ1 and the other with braiding angle θ2, were positioned around the iron rod. In the final step (Figure 3-17 (c)), an additional layer of rubber was wrapped around the half tubes to strengthen the bonding and prevent bursting along the seam. The assembly was then vulcanized by baking it at a temperature of around 190ºfor 20 minutes. One end was sealed with a serrated aluminum cap, and the other end was connected to the air hose. This technique of bonding two half-tubes with different braiding angles prior to vulcanization was found to be a facile method to create the required PBA structure. The resultant PBA was tested to bend in the form of an arc, without elongation or contraction, hence providing an efficient way of producing bending actuation only. In comparison with other methods mentioned previously, this approach provided a way of producing PBA which generates bending moment by excluding all other forms of actuation, such as elongation, expansion and twisting. As shown in Figure 3-18 – Figure 3-22, a controllable and repeatable manufacturing technique was employed to facilitate its fabrication. By altering the design of mold, one end of PBA could be sealed with rubber instead of embedding aluminum cap (Figure 3-23).

Preliminary Tests

Preliminary experiments were run on a few samples of proposed PBA, which is reinforced by fibre threads patterned at different braiding angle. As is shown in Figure 3-24, the bending displacement would gradually increase as pressure supply was increased. Notably, no radial bulging or longitudinal elongation was found, which means the energy provided by the compressed air was concentrated to produce bending actuation. The proposed actuator could be activated by different forms of fluid, no matter it is water or air. In this case, the pneumatic power supply was available in the lab of Engineering Department in Auckland University, hence made it easy to test with compressed air. In addition, applications specified in Chapter 6 and Chapter 7 would prioritize the lightweight of proposed actuator, therefore only the pneumatic powered actuator, named as PBA (pneumatic bending actuator) was made and tested.


In this chapter, a series of conceptual design schemes on PBA were explored, tested and compared in terms of their feasibility, functionality and efficiency of producing mechanical output. Design schemes were implemented and tested based on the automatic and repeatable winding technique. In the end, a bimorph-like structure by controlling the braided angle of fiber reinforcement was proved to be the most effective technique in producing bending actuation at normal pressure supply. The other methods, though might fail to demonstrate desirable bending capability, still show the possible technique to produce soft actuators for different purposes. By optimizing and combining them, they could be conducive to variation and optimization in the design of actuators to suit a large variety of applications.

Table of Contents
Table of Contents 
Chapter 1 Introduction 
1.1 Soft Robots
1.2 Soft Bending Actuator
1.3 Research Motivation
1.4 Objectives and Scope
1.5 Thesis Synopsis
1.6 Summary
Chapter 2 State of Arts 
2.1 Conceptual Design of Fluidic Bending Actuator
2.2 Characterization and Modelling
2.3 Materials and Fabrication
2.4 Drive and Power
2.4 Control
2.5 Applications
2.6 Performance evaluation
2.7 Limitations
2.8 Summary
Chapter 3 Conceptual Design and Fabrication 
3.1 Preparation
3.2 Fabrication methods
3.3 Preliminary Tests
3.4 Summary
Chapter 4 Modelling and Characterization 
4.1 Theoretical Modelling
4.2 FEA Modelling
4.3 Summary
Chapter 5 Testing and Verification 
5.1 Characterization Experiments
5.2 Static Analysis
5.3 Quasi-static analysis
5.4 Dynamic analysis
5.5 Robustness
5.6 Fatigue Test
5.7 Repeatability
5.8 Speed Test
5.9 Failure Modes
5.10 Competing Research
5.11 Summary
Chapter 6 Application 1 – Post-stroke Rehabilitation Glove 
6.1 Introduction
6.2 Design Specification
6.3 Configuration of System
6.4 Testing and Validation
6.5 Results
6.6 Discussions .
6.7 Price Breakdown
6.8 Features of Glove
6.10 Summary
Chapter 7 Application 2 – Designing, Modelling and Simulation of Soft Grippers
7.1 Introduction
7.2 Materials and methods
7.3 FEA Simulation
7.4 Experiment
7.5 Testing and Validation
7.6 Results
7.7 Discussions
7.8 Summary
Chapter 8 Conclusions and Future Work 
8.1 Research Contributions
8.2 Future Work
Design, Characterization and Application of A Pneumatic ?ending Actuator

Related Posts