You are on page 1of 22

Torsion Lab

By

Charisse Lewis 0729543 ME 354 AC

Date of this Lab Exercise:

February 11, 2010

Date of Lab Report Submission:

February 25, 2010

Executive Summary:

Torsion testing is one of the basic mechanical tests that are conducted to obtain material properties, such as the elastic modulus, strength, shear modulus, and shear strength. During this torsion lab the Technovate model 9041 Torsion Tester was used to conduct the testing. The materials used in this lab were 6061-T6 Aluminum, and A36 Steel. The setup for the lab took a lot of attention to detail, making sure that the specimen was tightened into the grip just right, and time, if the specimen was not clamped properly or failed inside the grip the test would have to be redone. The main purposes of the lab were to observe how the specimens of two materials that were used reacted under torsion, and to verify the accuracy of the Power-Hardening and Ramberg-Osgood applied torque predicting models. While observing the specimens of Aluminum and Steel data was collected as well, so as to compare the two samples numerically to each other and to the torque predictor models. When the data for aluminum and steel are compared, the aluminum sample is more ductile than steel is, which is shown in Figure C.1 in the results. This is evident because the aluminum sample underwent more angle of twist per unit length than the steel specimen did. The results show that the Power-Hardening and Ramber-Osgood torque predicting models work well for Aluminum, which was the only material that had data applied to the models. The maximum percent error, shown in Table C.3 in the results section, % Error, for the PowerHardening model was 0.706 % and for the Ramberg-Osgood it was 1.40 %. These two models had the % Error maximums at opposite ends of the data set. The Power-Hardening model had the greatest % Error when /L was a maximum, and the Ramberg-Osgood had the greatest % Error when /L was a minimum.

A. Introduction Torsion testing is widely used for evaluating the elastic modulus, strength, shear modulus, shear strength and other properties of materials. The main difference between torsion testing and tensile testing, as discussed in previous labs, is that the stress is not uniform over the cross section of the test specimen. The only useful exception to that is a round thin-walled tube, when the wall is sufficiently thin, but this lab did not use thin walled specimens, so that is that will be said about that.

In this lab the apparatus that was used was the Technovate model 9041 Torsion Tester (Appendix A) with the specimen being held in by grips. The force that was applied was measured at the force gage via a lever arm. A threaded horizontal drive rod was used to adjust the angle of twist in small increments, and when it was necessary to adjust the angle in large increments there was a lever arm, that could adjust the angle up to 90 degrees. The specimens that were used for this lab were constant-diameter gage section torsion specimens of 6061-T6 Aluminum, and A36 Steel. The diameters of the specimens were approximately 4.77 mm (about 3/16 inch).

There were many purposes for doing this lab, the most important ones being to examine the responses of metals, like the steel and aluminum that were used, to applied torsion, and to verify the accuracy Power-Hardening and Ramberg-Osgood models in predicting the applied torque. Also, this lab was conducted so that students learned about torsion testing, the apparatus that is used, and the information that can be obtained from the tests.

At the beginning of the experiment a straight line had to be drawn on the specimen. This was done so that as the lab was progressing the students were able to tell if the set of data that was being collected was a good set or not. If it was a good set, then the line that was drawn on the specimen at the beginning of the lab would wrap around the sample in a uniformly spaced progression. The less uniform the spacing the worse the data set was. This is known to hold true because a radial line which is straight before loading remains a straight radial line after loading.

B. Procedure The first thing that was done in the lab was the measuring of the diameter of the specimen gage section using calipers, and to record that value for later calculations. After that, it was necessary to draw a straight longitudinal line on the specimen so that the angle of twist of the specimen can be observed during the test.

When installing the specimen in the machine it must first be installed in the bottom grip, and the specimen has to be very tightly camped in the grip to ensure that it will not slip. Next, the pin behind the lower grip that engages the ratchet mechanism must be located and make sure the pin is pulled forward (this ensures that the ratchet is engaged). Then, the horizontal drive rod must be unthreaded until the end of the threaded region has reached the threaded nut, just before the threaded rod is disconnected from the base. Then the lever arm must be rotated counterclockwise, moved to the right, as far as possible.

At the top of the apparatus identify the wire ropes that transfer the torque to the top grip and specimen, and loosen the two nuts as far as possible. Rotate the top grip clockwise, so as to remove the slack in the wire ropes, and while holding the top grip in this position as firmly as possible, tighten the grip. Make sure that the specimen is very tightly fastened in the top grip. Remove any remaining slack in the wire ropes by retightening the two nuts that were loosened before, but monitoring the force sensor while that is being done. If the nuts are completely tightened and slack still remains, tighten the threaded horizontal drive rod until an increase in load is sensed.

After the specimen is tightly clamped into the machine, measure and record the distance between the grips, but make sure to account for how far the bottom face of the upper jaw is recessed into the upper plate in the measurement. Adjust the pointer so that it indicates zero degrees, and zero the output of the force sensor.

All of the prior steps were the setup for the lab, and with those complete the data collection for the lab may begin. Start applying known angles of twist to the lower grip, beginning with 2o 4

using the threaded drive rod, and record the force displayed on the force gage. After that increase the angle of twist by 2o until the total angle of twist is 30o, recording the force shown on the force gage at the end of each increase. Next, increase the angle of twist by 5o. Continue to take measurements at this increment until the ratchet in the lower grip clicks. This usually occurs around 45o, but it can occur at a total angle of twist larger than that, up to about 60o, just keep on increasing the angle of twist until the click occurs. If the click doesnt occur until well after 60o, it may be that the specimen was not correctly installed, and therefore may not give a good data set, so it might be advisable to start over. After hearing the click, record four measurements at increased angles of twist, in increments of 45o. Record four more measurements, this time at increments of 90o, and four more measurements at increments of 180o. Finally, record measurements for increased angles of twist in increments of 360o, until specimen failure occurs.

After failure remove the broken halves of the specimen. Carefully examine the specimen and the fracture surface, record what is necessary to discuss later if it is required. C. Results The data that was collected during the torsion testing lab was force (kg), and angle of twist. Comparison of the measured values for torque (Appendix B) versus the angle of twist divided by the length, /L, is shown in Figure C.1 below:

Measured Torque vs. /L


12 10 Measured Torque (N-M) 8 6 4 2 0 0 100 200 300 400 500 600 700 /L (rad/m) A36 Steel 6061-T6 Aluminum

Figure C.1: Measured Torque vs. /L for the A36 Steel and 6061-T6 Aluminum specimens that were tested. On the next page, in Figure C.2, is an example of what one of the A36 Steel specimens looked like after fracture. This example was representative of most of the steel specimens, so it is used for determining how brittle or ductile the material was just based on the appearance of the fracture.

Figure C.2: Steel specimen at the point of failure, after fracture.

The values for force converted to Newtons, and the /L for aluminum is shown in Tables C.1 and C.2. The other data that is shown in Table C.1 is for predicting torque according to the Power-Hardening model. The data that is shown in Table C.2 other than the /L is the calculated values necessary for the Ramberg-Osgood model for predicting applied torque.

Table C.1: The Power-Hardening model (Appendix D) for predicting torque with values that are necessary for calculation
Angle (deg) 0 2 4 6 8 10 12 14 Force (N) 0 10.3 10.5 11.3 12.0 13.0 14.0 14.5 Angle (rad) (rad/m) 0 0.0349 0.0698 0.105 0.140 0.175 0.209 0.244 0 0.137 0.275 0.412 0.550 0.687 0.825 0.962 Measure Torque (Nm) 0 1.35 1.38 1.48 1.57 1.70 1.83 1.90 Predicted Torque (P-H) (N-m) 0 0.179 0.357 0.536 0.714 0.893 1.07 1.25

16 18 20 22 24 26 28 30 35 40 45 90 135 180 225 315 405 495 585 765 945 1125 1305 1665 2025 2385 2745 3105 3465 3825 4185 4545 4905 5265 5625 5985 6345 6705 7065 7425 7785 8145 8505 8865 9225

16.8 16.9 17.4 18.1 19.6 20.4 21.6 22.4 24.5 26.5 28.4 35.6 37.5 40.2 40.7 41.7 41.7 41.7 41.7 42.7 42.9 44.4 44.4 45.2 46.2 46.6 47.1 47.1 47.2 47.6 47.6 47.9 48.2 48.3 48.8 49.1 49.1 49.1 49.1 49.1 49.1 49.5 49.3 50.0 50.0

0.279 0.314 0.349 0.384 0.419 0.454 0.489 0.524 0.611 0.698 0.785 1.57 2.36 3.14 3.93 5.50 7.07 8.64 10.2 13.4 16.5 19.6 22.8 29.1 35.3 41.6 47.9 54.2 60.5 66.8 73.0 79.3 85.6 91.9 98.2 104 111 117 123 130 136 142 148 155 161

1.10 1.24 1.37 1.51 1.65 1.79 1.92 2.06 2.40 2.75 3.09 6.18 9.28 12.4 15.5 21.6 27.8 34.0 40.2 52.6 64.9 77.3 89.7 114 139 164 189 213 238 263 288 312 337 362 387 411 436 461 485 510 535 560 584 609 634

2.20 2.22 2.28 2.38 2.57 2.67 2.83 2.93 3.21 3.47 3.73 4.66 4.92 5.27 5.34 5.46 5.46 5.46 5.46 5.59 5.62 5.82 5.82 5.93 6.06 6.11 6.17 6.17 6.18 6.24 6.24 6.27 6.31 6.33 6.40 6.43 6.44 6.44 6.44 6.44 6.44 6.49 6.47 6.56 6.56

1.43 1.61 1.79 1.96 2.14 2.32 2.50 2.68 3.12 3.57 3.93 4.81 5.01 5.12 5.19 5.31 5.40 5.47 5.52 5.62 5.69 5.76 5.81 5.90 5.98 6.04 6.09 6.14 6.18 6.22 6.26 6.29 6.32 6.35 6.38 6.40 6.42 6.45 6.47 6.49 6.51 6.53 6.54 6.56 6.58

Table C. 2: The Ramberg-Osgood model (Appendix E) for predicting torque with values that are necessary for calculation (MPa)
0 50 100 150 200 250 270 280 290 300 305 310 315 320 325 330 335 340 345 350 355

0 3.33E-14 4.64E-08 0.000182 0.0646 6.13 29.5 62.0 127 253 355 495 686 945 1280 1770 2410 3260 4390 5890 7860

0 0.00196 0.00391 0.00587 0.00782 0.00978 0.0106 0.0110 0.0113 0.0117 0.0119 0.0121 0.0123 0.0125 0.0127 0.0129 0.0131 0.0133 0.0135 0.0137 0.0139

0 0.00196 0.00391 0.00605 0.0724 6.14 29.5 62.0 127 253 355 495 686 945 1300 1770 2410 3260 4390 5890 7860

(rad/m) 0 0.820 1.64 2.54 30.3 2580 12400 26000 53200 106000 149000 207000 287000 396000 544000 743000 1010000 1370000 1840000 2470000 3300000

Predicted Torque (R-O) (N-m) 0 1.07 2.13 3.27 5.57 6.99 7.55 7.83 8.11 8.39 8.53 8.67 8.81 8.95 9.09 9.23 9.37 9.51 9.65 9.78 9.92

The comparison of the two predictive models, Power-Hardening and Ramberg-Osgood, to the measured values of applied torque is in Figure C.3. This was done so that the students could visually see how the predicted models compared to the measured values, instead of just seeing data values.

Comparison of Measured Torque to Predictive Models for 6061-T6 Aluminum


7 6 5 4 3 2 1 0 0 200 400 Torque (N-m) 600 800 Ramberg Osgood Power Hardening Measured Torque

Figure C.3: The comparison of measured torque to the predictive models for the 6061-T6 Aluminum specimen. Data values are still an excellent way of representing the differences between the measured values and the predicted values. This is done in Table C.3 below, which shows the values for the measured torque at four different values of /L, the corresponding predicted torque for the Power-Hardening and Ramberg-Osgood, and the % Error corresponding to differences from the measured values. Table C.3: The percent error of the predictive models, Power-Hardening (P-H) and RambergOsgood (R-O), to the measured values for torque
/L (rad/m) 100 250 400 550 Measured Torque (N-m) 5.86 6.21 6.41 6.47 Predicted Torque (P-H) (N-m) 5.85 6.20 6.39 6.52 % Error 0.222 % 0.111 % 0.391 % 0.706 % Predicted Torque (R-O) (N-m) 5.95 6.23 6.38 6.45 % Error 1.40 % 0.307 % 0.596 % 0.403 %

(/L) ( rad/m)

10

D. Discussion In this experiment the main goals were to observe the metals specimens to see how they act when under torsion, and to verify the accuracy of the Power-Hardening and Ramberg-Osgood predictive models for determining torque. Observing the data collected for each of the metals in a graph of measured torque vs. /L, Figure C.1, it can be seen that the Aluminum specimen is much more ductile than the steel specimen. This is shown by the fact that the aluminum withstood a greater amount of angle of twist compared to its length, and the steel endured a greater amount of torque prior to fracture. Although, the steel was more brittle than the aluminum, the steel did behave as mostly ductile. This is shown in figure C.2 in the Results section, where the steel specimen exhibited mostly ductile behavior at fracture, which is shown by the flat fracture surface, and the small amount of tearing at the edge. The aluminum sample looked very similar to that of the steel, with the biggest difference being around the area of fracture where on the steel specimen it was course like sand paper along the sides that were originally smooth (Figure C.2 in the Results). Both samples initially had the straight line that was drawn on them before starting the testing, and by the end of the lab it formed a spiral around the outside of the sample. The more uniform the spacing that the spiral had, the better the data for the specimen would be.

As stated above the other main goal was to verify the accuracy of the Power-Hardening and Ramberg-Osgood models for predicting torque. This can be shown by Figure C.3 in the results section, which is a graph with the measured torque and the predictive models all on the same graph. The predictive models for torque appear to be very close to the measured values, and they are as shown in Table C.3, the percent error, % Error, of the predictive models as compared to the measured values for torque. The values for the Power-Hardening had an increasing error percentage as the values for /L increased, which means that this model is better for predicting the torque being applied for smaller values of /L. With the Ramberg-Osgood, the opposite thing happened, where the % Error was greatest in the beginning and then it became smaller as the value for /L increased. Even though the % Errors varied from beginning to end, the values for the % Error is still very small, with a max error for the Power-Hardening model being .706 % and for the Ramberg-Osgood it was 1.40 %. Since the max percent errors are so small, it is a 11

good assumption that both models are a good at predicting the applied torque, at least for a ductile material.

The best way to test how applicable the Power-Hardening and Ramberg-Osgood models are for predicting torque would be to test them with a lager variety of materials, and run more than one test for each material. The materials used in this lab were both mostly ductile, and this showed in the lab results, but this only gives results for ductile materials for the behavior of the applied torque predictive models. If there was a more brittle material used in the lab alongside the ductile materials it could be seen whether or not the predictive models apply. The minimum that could have been done was try to apply the predictive models to steel, which as discussed is more brittle than the aluminum.

Another problem that many groups experienced was tightening the specimens in the grips. If the grips were too tight, the material would get scratched, and would tend to fail in that region. On the other hand if the specimen wasnt tightened into the grips tightly enough, then the specimen would slip, and the data set would be poor. The best way to resolve this would be to find a way to gage how tight the specimen is in the grip, and have a specified grip strength (how tight the grip is on the specimen) so that students know how much tighter it has to be, or if they have gone too far.

Also, the machines that were used were very old, and the force gage was in kilograms (kg). The force gage being in kg, which is a unit of mass and not force, was not terrible because it is not difficult to convert kg to N, but it was also very touchy, probably due to how old it was. If anyone hit the table at all the gage usually changed values, whether it increased or decreased the value was not consistent either. This could also be a source of error between the measured and predicted values.

Other possible sources of error include; not tightening the specimen into the grip enough, tightening the specimen into the grip too much (as discussed above), the specimen being bent at all when installed, and adding angle of twist at too rapid a pace. The first two sources of error were already discussed above, but if the specimen was bent at all, this would create other forces 12

other than the shear stress and shear strain. The predictive models would not be as accurate because the initial assumption that was made about there only being shear stresses and forces would be incorrect; all of the equations would be off and would not account for the additional forces that are present. Also, if the angle of twist was added to the specimen too rapidly, then the planes in the crystalline structure of the material would not have as much time to slide past each other. Therefore the material would fracture sooner than a specimen that had the angle of twist applied at a slower rate.

13

Appendix A: The Technovate Torsion Test Machine During this torsion lab the Technovate model 9041 Torsion Tester was used to test specimens made of A36 Steel, and 6061-T6 Aluminum. In order to know how the lab works, it is necessary to know the setup that is being used. Therefore, below in Figure A.1 is a picture of the top view of the torsion tester with the key components labeled. Wire Rope

Top Lever Arm

Top Grip

Force Gage

Wire Rope

Figure A.1: To view of the Technovate model 9041 Torsion Tester, with key components labeled. There is also a schematic of the above picture in Appendix B, which is used for discussing how to calculate torque from the force reading collected. Below are two photos that display the bottom half of the testing apparatus with the key components labeled. These pictures are for reference throughout the report so that it is clear what is being discussed.

14

Ratchet Pin

Driving Rod

Bottom Grip

Bottom Lever Arm

Figure A.2: A view of the bottom half of the testing apparatus, and the key components labeled.

Bottom Grip

Angle Indicator

Bottom Lever Arm Driving Rod Figure A.3: An alternate view of the bottom half of the testing apparatus, with key components labeled.

15

Appendix B: Converting the Force Reading to Torque

The setup of the torsion test needs to be known in order to be able to calculate the torque being applied to the specimen, and this is shown in Figure B.1 below:

Figure B.1: Schematic top view of the Technovate Torsion Test Machine. As shown above the grip and specimen are held in place by the wire ropes, one that is attached to one side of the machine and one that is attached to the lever arm. The lever arm is held in place at the pivot point and the force gage, both of which cause a force at either end of the beam. The wire rope is the only other force acting on the lever arm, which is acting on the lever arm slightly off center. This is shown in the form of a Free Body Diagram (FBD) in Figure B.2, on the next page.

16

Applied Force ( ) Figure B.2: FBD for the lever arm with the three forces acting on it.

Reaction Force

During the torsion experiment the data that was collected was the force at the force gage, and the force was in kilograms (kg). First the force in kg had to be converted to Newtons (N), which is done by multiplying the force value, in kg, by and then to calculate the torque on the

specimen it was necessary to calculate the force pulling on the wire rope. This was done using static equilibrium moment equations centered at the pivot point, pp, shown here: With Equation B.2, is able to be obtained since (Equation B.2) is the measured data from

the lab that was converted to N. With the force in the wire rope the torque applied to the specimen can be calculated by multiplying the force in the wire rope times the radius of the top grip, (the distance to the application of the torque). This is shown in Equation B.3 below. (Equation B.3) The FBD for the forces acting on and the torque being applied to the top grip are shown in Figure B.3 below.

17

Torque

Figure B.3: FBD of the grip with the wire ropes as the forces causing torque. This is just the torque caused by one of the wire ropes, therefore the has to be doubled

in order to obtain the actual value for the torque applied to the specimen. In other words; (Equation B.4) This is the measured torque in the specimen that is compared to the predicted torque from the Power-Hardening model (Appendix D), and the Ramberg-Osgood model (Appendix E).

18

Appendix C: True and Effective Stress and Strain

The equation for effective stress is listed below as Equation C.1 ( ) (Equation C.1)

Since this lab was interested in finding predicted values for torque beyond yielding, the use of true stress was used in calculations. Thus the equation for effective stress becomes: ( ) ( ) ( ) (Equation C.2)

The effective stress equation can be rewritten to get the effective strain equation, as shown in Equation C.3. ( ) ( ) ( )

(Equation C.3)

During the torsion test there was only one stress, , and one strain, , applied. Knowing that there is only uniaxial stress, and uniaxial shear strain, Equation C.2 and Equation C.3 become: (Equation C.4) (Equation C.5)

Equation C.4 an Equation C.5 are important for the application of the power-hardening and Ramberg-Osgood torque predicting models as discussed in Appendix D and Appendix E, respectively.

19

Appendix D: Power-Hardening Model

Using the data collected from the torsion test that was conducted, a power-hardening model, (Equation D.1), was fit to the measured force, (assuming true axial stress and true axial strain), response for 6061-T6 Aluminum.

(Equation D.1)

The coefficients in Equation D.1 are H, the strength coefficient, which is 413 MPa, and n, the hardening exponent, which is .0633. Also, assuming the use of true stress and strain, as discussed in Appendix A, and using Equations C.4 and C.5, a new equation for developed that relates it to can be

using the coefficients H and n, as shown in Equation D.2.

(Equation D.2)

With this model the goal is to predict torque versus ( /L) response, and for this it is necessary to find equations relating the two. The total applied torque is equal to the sum of the torque acting on the elastic inner-region and the torque acting on the plastically-deformed outer-region, shown below: (Equation D.3) The elastic inner-region extends over Therefore: , and in this region , where

The plastic outer-region extends over

(Equation D.4)

, and in this region Equation D.2 becomes


) ( ) [ ]

(Equation D.5) (Equation D.6)

Using Equation D.4 and D.6 the torque vs. ( /L) response can now be predicted based on the Power-Hardening model. During this lab the Power-Hardening model was only applied to the Aluminum 061T6 sample data.

20

Appendix E: Ramberg Osgood Model An alternate approach to the Power-Hardening model for fitting data is the Ramberg-Osgood model, which uses Equation E.1 as a base equation for relating the predicted torque to the ( /L) response.

( )

(Equation E.1)

Just like the Power-Hardening model (Appendix C) the coefficients H and n are the strength coefficient and the strain hardening exponent, respectively. The values for H and n are: H= 407 MPa, and n=0.0490. As discussed in Appendix C, the use of effective stress and effective strain changes Equation E.1 into:

Where:

(Equation E.2)

(Equation E.3) The modulus of elasticity, E, for Aluminum 6061-T6 is 68.5 MPa, and the Poissons ration, , is 0.34 . Also, the equation becomes less daunting when a shear strength coefficient is defined, , as shown below:

(Equation E.4)

This changes Equation E.2 into:

(Equation E.5)

With all of these equations an equation for torque is made as shown below in Equation E.6.

Where Also:

(Equation E.6)

, meaning that the max shear stress occurs where the radius is the largest.

(Equation E.7)

21

( )

(Equation E.8) (Equation E.9)

(Equation E.10)

Equations E.6 through Equation E.10 are used to predict the torque vs. ( /L) response based on the Ramberg-Osgood model, with ( /L) being equal to: (Equation E.11) Just like the Power-Hardening model the Ramberg-Osgood model was only fit to the Aluminum 6061-T6 data set.

22

You might also like