Design Development

DESIGN DEVELOPMENT

1.1         Overview

A detailed design process was followed to develop the Shape Memory Alloy (SMA) actuated gripper as the novel actuator. The customer requirements were first obtained from advisor by Dr.Y.W.R. Amarasinghe. Benchmarking on those areas was completed once the specifications were determined. Brainstorming was then conducted to generate possible ideas for the SMA robotic hand. Applying a design selection matrix then reduced the multitude of ideas that were generated by the brainstorming. With the final design ideas selected, the pertinent calculations, solid modeling and preparations were completed in order to make the Critical Function Prototype of the gripper.

1.2         Benchmarking

Journals, books and internet sources were consulted to find information relating to the structure and strength of the gripper. In addition, robotic grippers’ structures and SMA wire characteristics were researched. These sources determined what products were currently available, research that was previously completed, current SMA wire applications and other pertinent information relating to the project.
4.3. Brainstorming
In order to develop a wide range of possible ideas concept mind maps were completed for each major component of the project. This enabled a wide range of ideas to be explored before selecting on a final design. Mind maps were completed for the gripper structures, the transmission-actuator methods and the holder. Several possibilities came to light from the mind maps concerning structures and shapes of the robotic gripper. The possible gripper shapes included both open and closed structures.

1.3         Experimental Process

The main purpose of this effort was to conduct original research and provide previously unavailable temperature dependent performance data of SMA specimens. With this data, design guidelines for implementing SMAs as deployment actuators can be developed. A unique characteristic of SMAs is that “all properties change significantly at the transformation temperatures Ms, Mf, As and Af” meaning the transformation temperatures of the material must be known in order to draw meaningful correlations between the structure of the material and its fracture when heat is applied.



1.3.1        Heat Treatment Process

First we will try to understand the behaviour of our purchased nitinol sample; at the very beginning we tried to observe whether our received sample has any reaction with temperature changes. So we used hot water to increase its temperature form ambient to 1000C and then cool down it from ambient to below zero degrees of Celsius valves. In above cases we didn’t see any shape memory effects. Then we wanted to undergo in a heat treatment process to achieve our objective. First we used blacksmith fire and increased its temperature for about 1400C for 15 minutes and then quenched it. But after this process also it did not show any changes with temperature.  So we used the oven in mechanical engineering production lab of the university to do our heat treatment experiment process by varying temperature and heat treatment time period.

1.3.2        Heat Treatment Methodology

In heat treatment process first we give heat and increase the temperature of the specimen to a relevant value and then remain in given temperature for a specific time period. After that we got the specimen from the oven and quickly quenched the specimen while cut offing the heat supply.
Table 41 Heat treatment details and relevant transient temperature values
Temperature  (0C)
Time (hrs.)
Transient temp. approx. (0C)
300
1
Below the ambient
400
1
Below the ambient
500
1
30-40

1.5
40-50
550
1
45-55

1.5
> 55
650
1
> 60

1.5
> 60         















After the heat treatment experiment we observed transient temperature increase with heat treatment temp and time. By considering above experimental details we selected transient temperature around 35-45 C (5000C, 1hour). Because it is the easiest achievable temperature near to the ambient temperature.
Figure 42 Testing of transient temperature values

1.3.3        Induce heat for specimen

Our second problem was how to achieve the controlling capability just above the room temperature (35-45 0C) of specimen to get the relevant shape memory effect. So we identified several techniques that can be applicable for our application.



Table 42 Several techniques to control temperature of specimen
Heating method
Consequences
Hot water channel
bulky, no quick response
Resistive heating through NITINOL
selected specimen has low resistance (4)
Heating element
compact, easy controlling, quick response

Since we are focusing on applying our shape memory sample for robotics applications, using of hot water chance can be form many drawback like bulky in size and high response time for operation. When consider resistant heating technique it also not applicable due to low resistant of our selected sample. This technique is widely used for Nitinol wires that are having very narrow cross section (micro meter level). But our selected sample should have considerable cross-section to obtain required full force in the application. So we decided using the heating element is the best option to increase the temperature of our sample. Because we saw some comparative advantages in it like compact in size, quick response than other techniques and we can easily control the temperature by changing the amperage across the heating element. These heating element must have higher resistant and stable in higher value of temperatures. So heating elements like Tungsten & Molybdenum would be ideal for the application.



Figure 43 After attaching tungsten wire to a Nitinol strip



1.3.4        Pull Force Calculation Practical

Our next target was calculating the pull force which can be delivered from various cross sectioned strips. So various cross sectioned Nitinol strips where cut from EDM wire cutting having length of 40mm and cross section of 0.5mm × 1mm, 1mm×1mm, 2mm×1mm and 4mm×1mm. from that we prepared four strips form each sizes for work. Those pull force values were taken into consideration when deciding the required size of Nitinol strip for the gripper assembly.



Figure 44 Prepaid Nitinol strips from EDM wire cutter

1.3.5        Insulation requirement

As our selection of method for the heating is the application of the heating element around the Nitinol strip wire. Since, Nitinol has very low thermal resistivity around 0.76x10-6 Ωm, it is needed to insulate the Nitinol strip to avoid the current passing through it. Our primary requirement is to facilitate the heating element for the smooth current passing along its only.
Typical insulators are not suitable in this application as the most of the insulators are electrical and thermal insulators. Though we want to insulate the Nitinol strip from the electrical flow conducting through the heating element, we want to absorb the heating conducting through heating element. Following insulators are identified as electrical insulators but not thermal insulators that are being used for heating element applications.



Table 43 Comparison of applicable insulation techniques
Insulator
Suitability
Disadvantage
Mica Insulator
Thermal conductivity of 0.528 W m-1 0C-1
Mica surfaces can be cracked in the long run as the reduced fatigue life.
Tungsten Insulated Wire
Have high thermal conductivity of 1.7 W m-1 0C-1 and lengthy fatigue life.
Too much expensive.
Rubber Pad
Excellent thermal performance and rework capabilities.
Bit hard to use in a user defined application.



Figure 45 After attaching tungsten wire to a Nitinol strip with rubber pad insulation


1.4         Design Specifications

A robotic gripper is being constructed in order to showcase an application of a Shape Memory Alloy (SMA) actuation system. These SMA’s will be placed into a groove to create a muscle array. This SMA is actuated by using the heating element in order to flow the heating and cooling around the SMA wire. Once heated, the SMA wire undergoes a phase change causing the wire to contract. The degrees of freedom of the SMA wire is one, as the SMA wire is contracted then extracted as the temperature around the SMA wire is varied with the time. The SMA wire’s initial force is greater at the beginning of its contraction then at the end, technically referred to as reverse biasing. Therefore, the robotic gripper will incorporate a mechanical device to compensate for the reverse biasing of the SMA wire.
In order for the force to be transmitted from the SMA wire to the gripper, mechanisms that will function similar to that of a human hand will be employed. These mechanisms are similar to the animal behavior seen in the environment and such mimicking of the behavior is used in the existing SMA based robotic actuators as we have seen in the literature review too.

1.4.1        Primary approach of 3D modelling


This was our first design and it was developed in order to get an idea about how a robotic gripper can be actuated through a Nitinol strip. But in this attempt what we have learned was the heating element would generate high heat in the Perspex assembly. Therefore, as a secondary attempt we have decided to install a Nitinol strip outside the Perspex assembly. That would eliminate the contact of Perspex and heating element so heat generation in the heating element would not affect the Perspex assembly.




1.4.2        Secondary Approach

In this design we have avoided the defection of the primary design which was the contact between heating element and the Perspex. The contraction of the gripper tip is generated by the actuation of the Nitinol strip with application of current. The extraction of the gripper tip is caused by the spring return mechanism supported at the both sides of the handle.
As we have eliminated the defection in the primary design, in the secondary approach we can analyze the variation of the force generated at the gripper tip according to the variation of the temperature. 






1.4.3        Controlling Circuit

Both voltage and amperage can be altered to achieve the differencing of the temperature of the heating element. As manipulating the voltage is easier normally, we have decided to build a controlling circuit based on voltage differencing.
Supplying heat around the is controlling through a voltage regulation using a mosphate based drivers with Pulse With Module. PWM signal is generated through Atmel microcontroller.
                              
                                        Figure 48 Control circuit block diagram


1.4.4        Nitinol behaviors analysis with input voltage


Heat transfer through Tungsten coil is calculated using following equation

dE/dt = W –H
E = thermal energy stored in wire
W= power generated by Joule heating
H = heat transferred to surroundings 

    E = CwT

                                                Cw = heat capacity of tungsten wire

W = I2 ×Rw

Rw = Residence of tungsten coil

Heat transferred to surroundings is taken as H,
H = Σ (convection to fluid+ conduction to supports + radiation to surroundings)

For equilibrium conditions the heat storage is zero:

dE/dt =O

W=H, and Joule heating W equals the convective heat transfer H

To find the temperature in the Nitinol, we have taken in to consider the following assumptions,
- Radiation losses neglected
- Steady state condition arrived
- Tw uniform all over the length of the wire
- Fluid temperature and density constant (Air)
 



Figure 49 Thermal conductive pad insulated Nitinol strip wound with heating element - tungsten coil


Resistivity of tungsten element (~300K), ƥ = 52.8 nΩm      
Diameter; D = 0.025 mm
Length; L= 6.6 cm
Resistance of tungsten element, R;

R = 7.099 Ω

Power generated by Joule heating, W;
W= (V2)/R = mC(Tw-Ta)

Surface temperature of Tungsten coil, Tw ;
Tw = V2/(R×m×C) + Ta              

Tw = V2/{7.099 Ω [  ( 0.00252 6.6) cm3 ×  19.3 ×           10-3  kg/cm3]  ()}+280C
      = V2 /0.237765074 +28
      = Temperature of tungsten element at given voltage value

Considering the temperature of tungsten element equal to temperature of thermal conductive pad;
To = Outside temperature of thermal conductive pad = Tw
T= Inside temperature of thermal conductive pad
    = temperature of Nitinol strip at given voltage



            Figure 410 Temperature changes through the Nitinol strip-heat element assembly

                h2= Thermal conductivity of thermal conductive pad = 5 W/mK (0-1000C)
                t2= Thickness of thermal conductive pad = 0.5 mm
                A2 = effective area = {(0.5 +1) × 2 mm  4cm} = Surface area of Nitinol strip

By using Fourier’s law
Qx= -k dT/dx
V2/R = -k (Ti-Tw)/dx
V2/7.099 Ω = -5 W/mK (V2 /0.237765074 +28- Ta) K/ 0.5mm
 Ti = V2 /0.237765074 +28 K
     ≈ Tw  
Because, insulation thickness is so smaller though insulation has low thermal conductivity
Table 44 Variation of input voltage vs. Nitinol strip temperature
Input Voltage
Temperature of Nitinol strip
1
34.73
1.5
43.14
2
54.92
2.5
70.06
3
88.56
3.5
110.4
4
135.7
4.5
164.3
5
196.2
5.5
231.5
6
270.2





                                   Figure 411 Input voltage vs. Nitinol strip temperature
 

                                     Figure 412 Nitinol strip variation in Ansys Nitinol material model

Since Nitinol temperature vs. deflection can be get through a coupe field analysis in Ansys, we have model the Nitinol strip in there. Direct couple field analysis has been taken into account and we have developed a material model for Nitinol properties. As the temperature varies with the voltage differencing in the nitinol the apparent deflection at the tip of the Nitinol simulated in Ansys model.


                                                       Table 45 Ansys simulation data   
Temp (C)
Deflection(mm)
45
0.713
55
1.024
65
1.335
75
1.645
85
1.955
95
2.265
105
2.576
115
2.886
125
3.197

                          Table 46 Temperature dependence of modulus of elasticity of NiTiNOL [31]
Temperature (C)
Young's Modulus (Gpa)
45
42
55
50
65
61
75
65
85
65.1
95
67
105
67.5
115
68
125
68.5

The experimentally determined values of modulus of elasticity are presented as a function of temperature in the above reference. Those values are taken into account when we are going to do the force calculation with temperature.

Length of the Nitinol strip = 0.04 m
Cross section area = 0.5 mm2

                                         Table 47 Force suppressed with variable temperature
Temp (C)
Deflection(mm)
Young's Modulus (Gpa)
Force(N)
45
0.713
42
0.37
55
1.024
50
0.64
65
1.335
61
1.02
75
1.645
65
1.34
85
1.955
65.1
1.59
95
2.265
67
1.9
105
2.576
67.5
2.17
115
2.886
68
2.45
125
3.197
68.5
2.74



                                                     Figure 413 Force suppressed with variable temperature

1.5         Finite Element Analysis approach for the gripper

Typically, engineers develop working design models prior to fabrication and implementation in order to predict behaviors when the design is subjected to environmental stresses. Doing so allows the engineer to make necessary modifications before any chips are cut with a direct correlation to reduced costs. Designs also typically involve well characterized materials (e.g. aluminum) with well-defined material properties and predictable behavior.
Evident from the literature, development of an algorithm addressing large displacement and thermo mechanical behavior of SMAs is in itself an area of research beyond the scope of this effort. Thus, commercially available computational analysis tools were used in an attempt to create a Finite Element Model (FEM) that closely modeled Nitinol response when restrained and strained to failure.

1.5.1        Selection of Multi-physics Modeler – ANSYS Workbench

ANSYS Workbench Release 15.0 was chosen as the multi-physics program due to its project schematic view ties together the entire simulation process, guiding the user through complex metaphysics analyses with drag-and-drop simplicity. Given the experimentally derived behavior of the material, it was modeled as nonlinear with large displacement.

1.5.2        Select Analysis System

ANYSYS offers many different analysis systems to aid the designer in predicting concept behavior. It is anticipated that the SMA will be instituted in a pre-strained static condition with heat being applied while in that static condition. Thus, the Static Structural analysis system was selected for developing the SMA finite element model.

1.5.3        Determine Analysis Material Properties

Nitinol is not available in the martensitic shape memory form as a standard material preloaded in ANSYS and so was created. Material characteristics required to accurately model nonlinear material behavior include



 




1.5.4        Assign Material to Model

The default material assigned to the specimen by ANSYS was generic structural steel. The material was changed to the created Nitinol for both the linear and nonlinear analyses.
Nitinol possesses very unique material properties. As such, a new material model must be created in ANSYS Workbench 15.0 to obtain satisfactory results. The team understands that Nitinol can have very different properties depending on the thermal processing it is subjected to after being cut. For extremely accurate results, a sample of the Nitinol strip must be tested for the various parameters listed below:
- Density
- Coefficient of thermal expansion at different temperatures
- Reference temperature value
- Stress vs. strain data points
- Hardening parameter
- Elastic limit
- Temperature scaling parameter
- Maximum transformation strain
- Martensite modulus
- Load dependency parameter
- Tensile yield strength
- Tensile ultimate strength
Nitinol properties of the model has to be determined by carious experimental tests. However to achieve such details, we have to perform so many experiments which are mostly out of our touch. Therefore, we assumed a Nitinol model and carried out our experiment on Nitinol behavior considering following
- Density: 6500 kg/m3
- Coefficient of thermal expansion at different temperatures: 6.6 × 10-6  0C -1
- Reference temperature value: 32 0C
-- Hardening parameter: 6.894× 106 Pa
- Elastic limit: 3.4474 × 105 Pa
- Temperature scaling parameter: 26062 Pa 0C -1
- Maximum transformation strain: 0.04 mm-1
- Martensitic modulus: 3.3095 × 108 Pa
- Load dependency parameter: 0.05
By using the above material model we have obtained the figure 4-12; result for the Nitinol strip temperature with deflection variation.


2 comments:

  1. Hi
    From what sources did you get this information?

    ReplyDelete
  2. I got help to find SMA analysis from your post. But still I face some problems. I get error messages of incorrect property data of nitinol. What I have to do to solve the problem?

    ReplyDelete