Development and Validation of Threaded Fastener Test Rig

Development and Validation of

Threaded Fastener Test Rig

Fredrik Wirström

Mechanical Engineering, master's level 2019

Luleå University of Technology

A

BSTRACT

Threaded fastener is one of the most common ways to join components. Having a method to test threaded fasteners is key when designing a joint and even more a nutrunner. A joint is often tightened by a torque applied on the head. The applied torque is equal to three quantities in the joint, the thread torque, the underhead torque and the clamp force. To measure these quantities are the propose of a friction test rig. A test rig was built last year to be able to do that in a flexible and effective way. But the test rig built last year does not have the possibility to measure the underhead torque. A new transducer is constructed to add that possibility. The transducer is constructed by using methods such as concept generation, strength calculations, FEM simulations and a comparison of sensors are made. An easy way to change the stiffness is also investigated but no possible solution could be found. Some modifications and trims are also done on the test rig and also a comparison with the older BLM rig. A validation is made of how good the rig perform is also done. The results shows in favour for the FTR but precise results is not possible to determine.

P

REFACE

This master thesis is the final part of the Master of Science in engineering degree at Lule˚a University of Technology and a part of the course E7011T, Degree project in Mechanical Engineering, specialisation Machine Design, Master of Science in Engineering.

The project has been of great value to my education and experience. It has been really interesting and exciting to use the knowledge gained during my studies and see how much of what we learn can later be used. I am very thankful that I had the chance to do this project.

I would like to thank all of who have helped me make this master thesis. Thanks to Mayank Kumar who was always there with guidance and all the time he put into making this. Thanks to Erik Persson that made the master thesis project possible. Thanks to Jan-Olov Aidanp¨a¨a for the wise advice and guidance. Thanks to Sergei Glavatskih for valuable input on the concepts and for questioning of everything. Thanks to Andreas Rydin for help with ANSYS and FEM simulations. I would also like to thank; Johan Nasell, Roodabeh Afrasiabi, Per Gren, Adam Jonsson and Rene Westbroek.

Fredrik Wirstr¨om

C

ONTENTS

1.5 Requirements and Limits . . . 2

1.6 Previous Work . . . 2 1.7 Test Rigs . . . 3 Chapter 2 – Theory 4 2.1 Strain Measurement . . . 4 2.1.1 Resistance Gauges . . . 4 2.1.2 Piezoelectric Gauges . . . 5 2.1.3 Optical Measurement . . . 5 2.1.4 Electrical Circuit . . . 6

2.2 Force and Torque Measurement . . . 7

2.3 Threaded Fasteners . . . 7

2.3.1 Thread Mechanic . . . 7

2.3.2 Joint Stiffness . . . 8

2.3.3 Load and Strengths . . . 8

2.3.4 Standard Measurement . . . 8 Chapter 3 – Method 9 3.1 Sensor Selection . . . 9 3.2 Testing of FTRs . . . 9 3.3 Validation of FTRs . . . 11 3.3.1 BLM Initial . . . 11 3.3.2 BLM Reference . . . 11 3.3.3 FTR Initial State . . . 11

3.3.4 Wiring and Connectors . . . 11

3.3.5 Shank Torque . . . 12

3.3.6 Load Cell Crosstalk . . . 12

3.4 Improvement of FTR . . . 12

3.6 Concepts . . . 13 3.6.1 Selection of Concepts . . . 13 3.7 Detailed Design . . . 13 3.7.1 Rigidity . . . 13 3.7.2 Strain . . . 14 3.7.3 Material . . . 14 3.7.4 Simulation . . . 14

3.7.5 Surface for Gauges . . . 15

3.7.6 Crosstalk . . . 15

Chapter 4 – Result 16 4.1 Sensor Evaluation . . . 16

4.2 Behavior of FTRs . . . 16

4.2.1 Validation of Shank Torque on the FTR . . . 20

4.2.2 Improvements of FTR . . . 21 4.2.3 FTR Final . . . 21 4.3 Concepts . . . 23 4.3.1 1 - Front shear . . . 23 4.3.2 2 - Sensor in key . . . 23 4.3.3 3 - Tension bars . . . 25 4.3.4 4 - Torque transducer . . . 25 4.3.5 5 - Double disc . . . 25 4.3.6 A - Rotating blocks . . . 25 4.3.7 B - Half plates . . . 25

4.3.8 C - Blocks with steps . . . 25

4.3.9 Stiffness Increase of Rig . . . 29

4.4 Concept Evaluation . . . 29

4.5 Final Design . . . 30

4.5.1 Strain gauge . . . 30

4.5.2 Rigidity . . . 32

Chapter 5 – Discussion and Conclusion 33 5.1 Sensors . . . 33

5.2 FTR . . . 33

5.3 Underhead Torque Transducer . . . 34

5.4 Stiffness . . . 35

5.5 Conclusion . . . 35

Appendix A – Concept of Stiffness Increment of the FTR. 39 Appendix B – Summarized Result of Hapticas Report on the Flats Size

Nomenclature

Ttot Total torque [Nm]

Tp Pitch torque [Nm]

Tt Thread torque [Nm]

Tb Underhead torque [Nm]

F Clamp force [kN] p Pitch [mm] µtot Total friction

µt Thread friction coefficient

µb Underhead friction coefficient

a Thread profile angle [°]

dm Average thread friction diameter [mm]

db Average underhead friction surface diameter [mm]

δEo Output voltage [V]

Ei Input voltage [V]

GF Gauge factor  Stain [%] σ Stress [MPa]

δ Displacement [mm] E Young’s modulus [GPa] ν Poisson’s ratio

k Stiffness [N/m]

C

HAPTER

1

Introduction

1.1

Atlas Copco

The company founded in 1873 is today a global supplier of compressors, vacuum solutions, generators, pumps, power tools and assembly systems. [1] This master thesis is written in cooperation with Atlas Copco Industrial Technique AB that handles the assembly systems and nutrunners.

1.2

Background

One of the most used ways to join two components is to use a threaded fastener as a screw or bolt. These joints require to be pre-tightened so the fastener does not get loose or fail. To tighten a joint is the most common way to use a wrench and measure the torque to twist the bolt. But what is actually desired in the joint is the clamp force and the relationship between the torque and force is dependent on friction which is always difficult because friction is system dependent and is affected by multiple factors. To better understand this a Friction Test Rig, will be abbreviated as FTR in the rest of this thesis, is used. This FTR was built as a master thesis project last year. Currently it can only measure the shank torque and clamp force of the joint. The torque from the bolt head against the underlying surface, often a washer, is not measured. Another copy of FTR is under construction, the already existing one is named Atlas FTR because it is used by Atlas Copco and the new will be named KTH FTR as it will later be used at KTH. The benefits of the FTR and the reasons it was build are that in can do measurement in a flexible way, the transducer can be changed and calibrated separately. It can also test different tightenings methods. There is knowledge of the FTR in-house and it is also easy to modify.

There is also an older test rig named BLM that measures the total torque, the thread torque and clamp force. Exactly how the BLM works is unknown and because the test

1.3. Purpose 2 rig is form an external manufacturer it is kept secret and it have no support. The BLM has a known issue with the total torque reading.

1.3

Purpose

The purpose is to create a well defined friction test rig for research purposes in order to investigate how different factors effect the tightening procedure and particularly the friction. Many factors affect the tightening such as coating of the parts, speed of tightening and the hardness of the joint, all to be tested in the rig. It is also interesting to see how the environment in other parts of the world can affect the tightening as different humidity and temperatures.

1.4

Tasks

The main part of this master thesis is to add the possibility to measure the torque generated by the bolt head on the existing FTRs. Adding a possibility to change stiffness of the test rig is desired because the tightening and friction parameters change with the stiffness. Validate the rigs and due to the current design only shank torque can be validated on both rigs when comparing torque but clamp load is also possible to measure on both. And see how the choice of sensors to a transducer affect the design.

1.5

Requirements and Limits

The new design must be able to withstand the high axial clamp force without risk of permanent deformation. The device must also be able to measure the torque at a high accuracy. The accuracy should be at least as the shank torque sensor. The new transducer should be designed to fit the current rig or can be fitted with so small changes as possible. The FTR will also later be tested in different environments to see how it behave and later to simulate tightening in different areas of the world. Therefore all modifications should be suitable to use in a wide range of humidity and temperature difference.

1.6

Previous Work

1.7. Test Rigs 3

1.7

Test Rigs

There are various test rigs on the market for testing friction in threaded joints. They are used by companies who want to test the fasteners that are to be used at their products. The market for friction test rigs are small and there are few built of each type and they are often customized for each user. However because the FTR was already built and the master thesis purpose is the develop that little research was made on other rigs.

C

HAPTER

2

Theory

2.1

Strain Measurement

Strain is the ratio of current elongation over the original length. Strain is related to the tension of the material and material properties. If the material properties are known, the geometry of the part and if the material stays in the elastic region the force can be calculated. The elastic region is when the object is loaded without any permanent deformation of the material and when it behaves linear.

2.1.1

Resistance Gauges

Strain can be measured by converting it to a change in resistance. This can be done by a metallic foil gauge or a semiconductive material. The ratio between change in resistance versus strain is the gauge factor GF . The first type consists of a metallic foil that is mounted in a pattern on an insulated back plate. This metal has a resistance that will change with its length. The gauge is mounted on the specimen and will deform equally, then the change of resistance is measured. [3, 4, 5]

The length of the gauge is of importance due to the sensor will measure the average strain over its length. [6] This value can be lower than the peak strain, see figure 2.1 The gauge should not be longer than tenth of the length of a segment of where the measurement is, like a radius, fillet or notch. [7] A short gauge causes inaccuracy because the resolution increases with the length thus a greater change in resistance. Also the influence of transverse strain increases with a short gauge because the amount longitudinal wire or foil decreases while the transverse amount remain the same.

Another method to measure strain by resistance is by a semiconducting gauge. The fundamental idea of this is that a semiconductive material changes its resistance when loaded, this is the piezoresistive effect. [5, 8] Theses gauges have a greater gauge factor than the foil gauges. They can be made stiffer too, have higher natural frequency and they are less self-heating because of small power dissipation. The semiconductor gauges

2.1. Strain Measurement 5

Figure 2.1: Strain level along the sensor. Indicated strain may be different from peak strain depending on gauge length. Image courtesy of Vishay Precision Group.

have increased error with temperature compared to metallic foil gauges and they also have less linearity. [9] The gauge factor for a metallic foil gauge is about 2 and up to 200 for a semiconductive sensor. [3]

2.1.2

Piezoelectric Gauges

Do instead of piezoresistive create a charge upon loading proportional to strain. They are mostly used for dynamic measurements due to the charge drains with time. Different material can be used as quartz or PVDF to change the properties of cost, accuracy and drain time. [5] The frequency response is higher for a piezoelectric quartz than resistance gauges. The piezoelectric gauges have a very wide range of measurement and can very precisely detect a small force in presence of a large background force. [8] According to one source the use of piezoelectric strain sensors today is almost zero. [10] This is indirectly denied by another source that gives examples of different types of piezoelectric sensors with pictures from a manufacturer. [11]

2.1.3

Optical Measurement

Strain can be measured by optics. This can be done by an optical strain gauge where a laser beam is transmitted into the gauge and some is reflected before the gauge. The difference in the phases of the reflection is measured, the elongation can be determined and then the strain. [5]

2.1. Strain Measurement 6

Figure 2.2: Wheatstone bridge arrangement with four gauges, Ei voltage supply and δEo

output voltage

light into the material the change of it can be measured. [3, 5]

Another optical method is by Moir´e interference. This method uses gratings and the amount of light that passes the gating depends on how stretched the bars are. Usually one is attached to the specimen and the other is used as reference, then it is also possible to determinate rotation by the light passing through them. This method measures displacement but the stain can be calculated from that. [5]

The resolution is better in optical methods than resistance gauges. The optical can measure a change of 0.0002 % while the limit for the metallic foil resistance gauge is 0.0015-0.0020 %. [12]

2.1.4

Electrical Circuit

For the resistance type gauges a bridge arrangement of a Wheatstone is commonly used as in figure 2.2. The bridge can increase the output signal and also remove unwanted strain components depending on placement of the bridge. The bridge consist of four resistors where at least one but up to all four can be strain gauges. The resistors are connected as two parallel two series configuration, the excitation voltage is applied over this. The output voltage is then measured over the two midpoints between the serial resistors. [3] A bridge with four equal gauges will have an output as

δEo

Ei

= GF

4 (1− 2+ 4− 3) (2.1) where δEo is the output voltage and Ei is input. The gauge factor GF is the ratio between

2.2. Force and Torque Measurement 7

2.2

Force and Torque Measurement

Force measurement methods are often related to strain. A basic force transducer is basically a solid with known stiffness, usually metal. When a force is applied the strain is measured with a strain gauge and the force can be calculated. [13]

A method that directly measure force is the piezoelectric. In a piezoelectric material when deformed an electric potential is created within. This charge is proportional to the force and the polarity will change with the direction of the force. [13]

2.3

Threaded Fasteners

The widely used threaded fasteners that is also a part of this thesis are the bolt, screw and nut. A threaded fastener is used to clap components together and doing so by transforming a rotating torque to a axial clamp force.

2.3.1

Thread Mechanic

The pitch of a thread is the axial movement of one revolution or the distance between the peaks of the thread. The pitch is one of the contributor of how the relationship between torque and force behave. But the friction is the largest contributor and hard to determinate. The torque to fasten a joint can be divided into three parts

Ttot = Tp+ Tt+ Tb (2.2)

where Tp is the pitch torque, Tt is thread torque and Tb is the underhead torque. [2, 14]

The underhead torque is also known as bearing surface torque. The torques can be approximated as Ttot = F  p 2π + µtdm 2 cosa₂ + db 2µb  (2.3) where the first term corresponds to pitch torque and so on. [2, 15] The clamp force is F and p is the pitch of the thread. The thread friction is µt, dm is the average diameter

of the thread friction interface and a is the angle of the thread profile, db is the average

diameter of the underhead contact surface and µb is the corresponding friction. The pitch

2.3. Threaded Fasteners 8

2.3.2

Joint Stiffness

When all parts in a joint are together and the bolt will start to deform the components and/or elongate the fastener the snug point is reached. It is when reaching this condition a preload is applied to the joint.

The stiffness of the joint is dependent of the stiffness of the components that are clamped. A stiff joint will not allow high deformation and therefore the fastener will reach its preload in a small angle after the snug point. A stiff joint that is a hard joint will go from 50 % of the total torque to 100 % in 0-30°. In a soft joint the components will deform more and 100 % of the total torque from 50 % is reached in 30-360°. [16]

2.3.3

Load and Strengths

Proof load is the maximum load a joint can handle without any plastic deformation. The yield strength is the uniaxial stress a joint can handle before plastic deformation. The proof load is where the actual elastic region ends because the plastic deformation of the joint will start. This is before the yield point due shear forces and is defined as 85 % to 95 % of the yield point. Proof load for standard bolts are specified in ISO 898-1. [17] Ultimate tensile strength is the point where a joint member breaks.

2.3.4

Standard Measurement

A ISO standard exists for total torque, clamp force testing. Including calculation for friction, the clamp force should be measured for all tests and for total friction the tightening torque should be measured. For thread friction the thread torque and equally underhead torque for underhead friction should be measured. The total friction is then calculated as

µtot = Ttot F − p 2π 0.577dm+ 0.5db , (2.4)

the friction between threads

µt= Tt F − P 2π 0.577d2 (2.5) and the underhead friction

µb =

Tb

0.5dbF

. (2.6)

C

HAPTER

3

Method

3.1

Sensor Selection

To determinate which sensor to use a selection matrix is made, this selection matrix is made with similarities of the concept of Screening. Screening is a quick and approximate evaluation aimed at producing a few alternatives. The process a includes reflection of the results before proceeding. [19] For this task some criteria is determined to find the sensor that will fit best for the application. This rig is for laboratory measurement, so accuracy is important and one criteria. The expense of the sensor is not a crucial parameter because of the accuracy and only two are made. To ensure that the torque is only measured in the desired direction crosstalk has to be minimized and that is one criteria. Another criteria is the range, the sensor must be able to measure a torque up to 100 Nm and with no risk of damage or overload so a margin to the torque is added. The rig will also be placed in a environment chamber so it have to withstand different temperatures and humidity. One more criteria is that the new device for measuring the underhead torque makes as small changes as possible on the current rig therefore the size is important but also flexibility to set up the rig in different configurations and test various joints.

The sensors that are included in the matrix is first the resistive types of strain measure-ment, the metallic foil sensor and the piezoresistive. Followed by the piezoelectric sensor and all optical methods are considered as one. Because the ultimate target is to measure torque a force sensor can also be used to measure the torque and is also included in the matrix.

3.2

Testing of FTRs

The test rig is attached to a rigid steel table by screws. A nut runner is set up in a adapter in front of the test rig. It is also attached by screws to the table. On the shaft from the nut runner there is an external torque transducer placed before the socket. The

3.2. Testing of FTRs 10

Table 3.1: Standard bolts for test Property Hex Hex with flange Unit

Length 80 70 mm Head size 17 14 mm Clamp length 58 58 mm

Socket type Hexagon Size M10 Strength class 8.8

Coating Zinc-Iron black waxed

torque transducer is a inline rotary torque transducer and will be abbreviated as IRTT. A support to the IRTT is also attached by a screw to the table. The test joint is tightened by hand until it is only 360° or less to the snug point. The nut is always placed with the marking pointing outwards from the joint hence visible when tightened. The distance between the nut runner and test rig is adjusted so the socket engages the bolt and that no axial force is exerted on the joint. A tightening sequence is defined in the software to the nut runner. The signal from the sensors at the FTRs and from the IRTT is received by the DAQ Dewe-43A and processed in the software Dewesoft X3.

To compare different states and rigs during the tests two standard joints are used, one bolt with a regular hex head and one with a flange also, they are specified in table 3.1. The nut is a hex type M10 size that is zinc plated (EPZ). These two joints have a stable friction. Under the bolt head is a test specimen placed it is a hardened steel plate that fit on the test rigs. The standard test sequence that is used for all test when its possible consist of 2 steps. Where the first step is run in phase at a speed of 100 RPM until 5 Nm, at second phase the speed is lowered to 20 RPM until 25 kN clamp force is reached. After a relaxation phase of 5 seconds before loosening 360°. For some tests a load control tightening is not possible and then a method with angle control is used. Then an angle of 105° is used after the first 5 Nm is reached. For some later test a method with only load control tightening is used so there will not be any interrupts and pauses during the process.

To handle the sampled data two methods is used. First a method that averages the test runs is used to minimize static variations and still keep consistent notches in the graph. For the runs that have different end values on the x-axis the average end value is calculated as ¯ x = 1 n n X i=1 xi (3.1)

3.3. Validation of FTRs 11 interpolating the data. The average value of all y-values at the specific x-value are calculated and finally the new y-values are plotted against the defined x-values. The second method is least squares fitting with a polynomial of degree one to four depending on how the data looks. Between these two methods the one that fit the data best is used. The tightening for the tests are defined for a clamp force between 2 kN to 20 kN and the values outside this interval are discarded. The angle is defined as zero at the 2 kN start point.

3.3

Validation of FTRs

3.3.1

BLM Initial

Initial test was performed on the Test Rig BLM where two joint types where tested, they where tightened and lessened five times in a row to see how the friction envelopes and how the tightening curves are. The first join type consists of an oxide layer nut and bolt and the second consists of a Zinc electroplated(EPZ) bolt with the same nut as the first joint. These two joint types are not the standard test joint described in previous section. During theses test the difference between the total torque in BLM is compared to the IRTT.

3.3.2

BLM Reference

To use BLM as reference a more extensive test is done with the standard test joint. A total of 100 tightenings where made. Five tightening is made with a bolt and nut in a sequence by the nut runner before replaced, five of this runs is made at one side of a hardened test specimen. After five runs the side of the test specimen is switched and five more runs is performed. This procedures is repeated for the other bolt type, a new test specimen is then used.

3.3.3

FTR Initial State

A similar test as done on the BLM where run on the FTR. The FTR can hold different load and shank torque cells, for this test a load cell of 100kN and a torque cell of 100Nm range was used. During the first tightening of each bolt a angle of 105° is used and for the reaming four a angle of 95°.

3.3.4

Wiring and Connectors

3.4. Improvement of FTR 12

3.3.5

Shank Torque

To ensure that the FTR shows correct shank torque the FTR is twisted so that the back end is facing the nut runner. A joint is tightening from the back end so that only the shank torque transducer is in use at the FTR. No clamp force is affecting the system and the transducer is compared against the IRTT.

3.3.6

Load Cell Crosstalk

A joint is clamped around the load cell. The load cell is not sensitive to the axial force by a joint only clamped trough the center of the cell. The joint is then twisted by the nut runner only applying a torque in steps of 10, 20, 30 and 40 Nm. The force from the load cell is read with the torque from the IRTT. The crosstalk form this test will show how the force is affected by the underhead torque.

3.4

Improvement of FTR

Some improvements are done to the FTR test rig and tests with the standard hex bolt is used to see how the changes affect the tightening. The improvements are a re-machined holder for the thrust bearing, a new key with a tighter fit and ceramic balls for the thrust bearing. The holder for the thrust bearing to stay in place when the FTR is assembled is machined to a lower height and changing the fit from a tight to a loose so that the holder does not affect the torque when testing. The key is changed to a tight fit so there is no risk of tilting the key. The torsional stiffness will also increase with the larger key. To reduce the friction in the bearing the balls in the thrust steel bearing is changed to ceramic and a new grease is applied. The bearing is washed in an ultrasonic cleaner before the new grease is applied. The load cell is also connected to the nut runner so the FTR can be tightened by load control as the BLM. For the FTR, a load control tightening is used all the way without the torque step.

3.5

Shank Torque Validation with Bolts

3.6. Concepts 13 the strain is read. The bolts is also validated against clamp force, the bolt is set up in a vice with a force sensor. The vice is tightened and the signal from the bolt and force sensor are read. The strain gauges applied are metallic foil resistance gauges glued with an adhesive to the surface that is first grounded with sand paper and then cleaned by an alcohol. To align the gauges a caliper is used to scratch the surface at desired location. The gauges are, after the cables are soldered, coated with a silicon based coating.

3.6

Concepts

The concepts are generated during the project and are discussed with supervisors and manufactures. The first concepts are about measurement of underhead torque, next part is how to vary the stiffness and last part is miscellaneous of improvements on the FTR and friction testing of bolts in general. A concept for shear measurement from last year will also be evaluated. This concept is based on four piezoelectric shear sensors that sit equally spaced around a frame. The piezoelectric sensors require a compressive pre-load that is accomplished with tightening the load cell on top.

3.6.1

Selection of Concepts

The concepts where selected by calculations and discussion with supplier and supervisors. Concepts can be merged and developed to new ones along the process. The concepts must fulfill the demands for accuracy and durability.

3.7

Detailed Design

3.7.1

Rigidity

The concepts must be rigid enough to withstand the forces without plastic deformation. The uniaxial stress is defined by

σ = F

A (3.2)

which is the applied load F dived by the area A. [20] The stress must be lower than the materials yield strength with a desired factor. Non homogeneous body will have concentrations of stress at holes and radius. The maximum stress at the concentrations can be a lot higher than the nominal in the body and a factor defined as

Kt=

σmax

σnom

3.7. Detailed Design 14 The new part will effect the stiffness of the rig and a stiffness analysis for the individual part is made and calculated by

k = F

δ (3.4)

and F is the axial force and δ is the maximum displacement in the axial direction. The new stiffness of the entire FTR will now be

1 ktotnew = 1 ktotold + 1 k (3.5)

and because of the already low stiffness of the FTR a high as possible value is desired.

3.7.2

Strain

The gauges measure strain and the concepts need to have a desired strain in the operating range. High strain will increase the resolution of the measurement but a low strain will increase the linearity and also reduce the fatigue. The strain must balanced depending of requirements. To calculate the shear strain that the gauge will sense on a cylinder this formula is used

 = M (1 + ν)

W · E . (3.6) The torsional resistance for a cylinder with outer and inner radius b and a is

W = π 2b(b

4_{− a}4_{) (3.7)}

and E is Young’s modulus and ν is Poisson’s ratio. The standard strain used at Atlas Copco is about 0.03 % for a metallic foil gauge with a factor of about 2. This is the strain along the gauge.

3.7.3

Material

For material the steel SS-2541-03 is to be used because its high yield strength and it is easy to machine. It is similar to what was used for the transducers in last year master thesis.

3.7.4

Simulation

3.7. Detailed Design 15

3.7.5

Surface for Gauges

If the surface for the gauges is round there will be a deformation of the measurement element when bent over the surface. Therefore a flat area is preferable and how the flat effects is calculated. The effect of how the size on the flat area affect the stain is calculated by an external supplier. The linearity of this is calculated by applying five steps of torque and comparing the strain on the flat area and on the curved surface in ANSYS. A flat area of the size 2.5 x 2.5 mm is used to test the linearity.

3.7.6

Crosstalk

C

HAPTER

4

Result

4.1

Sensor Evaluation

The results from the comparison of sensors is presented in table 4.1. From the table the optic is rated highest followed by the force transducers and on third place is the standard metallic foil stain gauges. The optic has a cost of 2000 sek for each gauge and it needs an optical equipment for 100 000 sek. Compared to the metallic foil with a cost of 1000 sek for a package of ten and only need equipment that already is in-house. The force transducers have prices spanning from 2000 to about 8000 sek.

4.2

Behavior of FTRs

The result in figure 4.1 shows the three tightenings for each joint. In the figure the clamp force of the joint, the total torque to tighten read by the IRTT and the shank torque by the threads is shown by different colors. During theses tests the total torque measured by the BLM where also sampled. For the respective joint a comparison between the IRTT and the BLMs total torque is shown in figure 4.2. The total torque from the BLM suffers from noise but also from an offset that shows about 0.5 - 1 Nm less then the IRTT when the joint is tighten.

The initial test of the FTR shows that is was performing as figure 4.3 and BLM plooted as reference is also shown. The tightening angle for the BLM is 84° and 100° for the FTR. The connector is re-soldered because the excitation leads is swapped with the signal. The zero values offset of the load cell before was -0.995 and after -0.883. The value from the manufacturer is -0.868. From shank torque measurement from the back there is not any real difference between the IRTT and the torque transducer.

4.2. Behavior of FTRs 17

Table 4.1: Sensor scoring matrix Criteria A (Reference) Metallic foil B Semiconductive C Piezoelectric D Optic E Force Accuracy 0 + 0 + + Range 0 - - 0 + Temperature dependence 0 - - - 0 Humidity resistant 0 0 0 + 0 Size 0 + - 0 -Flexibility 0 - - 0 0 Crosstalk 0 0 0 + 0 Sum of + 0 2 0 3 2 Sum of 0 7 2 3 3 4 Sum of - 0 3 4 1 1 Score 0 -1 -4 2 1 Rank 3 4 5 1 2 0 1 2 3 4 5 6 7 Time [s] -20 0 20 40 60 80 100 120 140 Force [kN] / Torque [Nm] Clamp Force Tightening Torque Shank Torque

(a) Oxide nut and bolt.

0 1 2 3 4 5 6 7 Time [s] -20 0 20 40 60 80 100 120 140 Force [kN] / Torque [Nm] Clamp Force Tightening Torque Shank Torque

(b) EPZ treated bolt and oxide nut.

4.2. Behavior of FTRs 18

(a) Oxide nut and bolt. Scatter of 3 runs. (b) EPZ treated bolt and oxide nut. Scatter of

3 runs.

Figure 4.2: The difference in torque by the external IRTT and the BLM total torque(IRTT - BLM). 0 10 20 30 40 50 60 70 80 90 100 110 Angle [°] 0 5 10 15 20 25 30 35 Torque [Nm] / Force [kN] Force BLM Torque BLM Shank BLM Force FTR Torque FTR Shank FTR

4.2. Behavior of FTRs 19 0 5 10 15 20 25 30 35 40 Torque [Nm] -0.1 0 0.1 0.2 0.3 0.4 0.5 Force [kN] 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Crosstalk [%] Force from 50 kN Force from 100 kN Crosstalk 50 kN Crosstalk 100 kN

Figure 4.4: Force measured by the load cell when only a torque is applied. The percentage crosstalk is shown by the dashed lines. Least square fit of 3 runs per transducer.

0 5 10 15 20 25 30 35 40 Torque [Nm] 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Force [kN] 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 Crosstalk [%] Force Crosstalk

Figure 4.5: Force measured by the BLM when only a torque is applied. The percentage crosstalk is shown by the dashed lines. Least square fit of 6 runs.

4.2. Behavior of FTRs 20 0 1 2 3 4 5 6 7 8 9 Sample [-] 104 -2 0 2 4 6 8 10 Torque [Nm] / Force [kN]

Force Sensor Bolt Torque

(a) Value from bolt and force sensor during clamp load calibration.

0 2 4 6 8 10 Force [kN] -0.5 0 0.5 1 1.5 Torque [Nm]

(b) Fitted curve of bolt torque reading with clamp force.

Figure 4.6: Results from calibration of shank torque bolts for clamp load.

0 5 10 15 20 25 30 35 40 45 50 Shank Torque [Nm] -2 -1.5 -1 -0.5 0 0.5 1 Torque Difference [Nm] BLM FTR 50 FTR 100

Figure 4.7: Torque difference between the shank torque transducer in the test rig compared to the bolt (bolt - shank torque). Least square fit of 2 runs per case, hex bolt.

4.2.1

Validation of Shank Torque on the FTR

4.2. Behavior of FTRs 21 0 10 20 30 40 50 60 70 80 90 100 110 Angle [°] 0 5 10 15 20 25 30 35 Torque [Nm] / Force [kN] Force Initial Torque Initial Shank Initial Force Modified Torque Modified Shank Modified

Figure 4.8: Atlas FTR with modifications compared with initial state. Mean value of 50 runs per state, hex bolt.

Test rig Angle [°] Angle factor BLM 71 1.00 FTR Atlas 50 87 0.820 FTR Atlas 100 87 0.813 FTR KTH 50 89 0.801 FTR KTH 100 86 0.823

Table 4.2: Average angle to complete the tightening and a factor for comparison with the BLM.

4.2.2

Improvements of FTR

The total changes of the FTR when equipped with a new bearing holder, key and using load control is shown in figure 4.8. The angle with all the improvements is now 84° instead of 100°.

The difference between a steel and a ceramic thrust bearing with new grease is show in figure 4.9. The angle with ceramic balls is 77° instead of 84°.

4.2.3

FTR Final

4.2. Behavior of FTRs 22 0 10 20 30 40 50 60 70 80 90 Angle [°] 0 5 10 15 20 25 30 35 Torque [Nm] / Force [kN] Force Steel Torque Steel Shank Steel Force Ceramic Torque Ceramic Shank Ceramic

Figure 4.9: Steel balls compared to ceramic balls in the thrust bearing in Atlas FTR. Mean value of 10 runs per state, hex bolt.

0 10 20 30 40 50 60 70 80 90 Angle [°] 2 4 6 8 10 12 14 16 18 20 22 Force [kN] BLM FTR Atlas 50 FTR Atlas 100 FTR KTH 50 FTR KTH 100

4.3. Concepts 23

(a) Variant with a holder for the test specimen. _{(b) Variant with threads on both sides.}

Figure 4.11: Concept 1, a cylinder that is mounted front of the load cell.

4.3

Concepts

The major concepts generated during the concept generation are presented in this section.

4.3.1

1 - Front shear

The first concept is a cylinder that is attached to the load cell by the thread for the test specimen holder. This cylinder will twist when a torque is applied and that deformation can be measured by strain gauges and converted to torque. The strain gauges are positioned around the outer circumference of the cylinder. This concept can either have a holder for the test specimen as in figure 4.11a or threaded in both ends as figure 4.11b . The thread at the front is the same as on the load cell so the standard test specimen holder can be assembled on this concept.

4.3.2

2 - Sensor in key

4.3. Concepts 24

(a) View of key.

(b) Deformation of key.

(c) Key placement on load adaptor.

(d) Key placement in housing.

4.3. Concepts 25

4.3.3

3 - Tension bars

This concept is based on bars between the test rig and the load cell. The bars span between the bolt holes and will be tensioned upon loading. The strain can be measured with gauges attached to the bars. To keep the bars in position a center ring will be connected to the bars with arms. See figure 4.13 where it is also assembled on the FTR.

4.3.4

4 - Torque transducer

There already exists torque transducers on the market and this concept uses one from HBM to measure the underhead torque. The bolt holes need to be adjusted so it will fit. In figure 4.14 the transducer is shown and also how it is assembled on the FTR.

4.3.5

5 - Double disc

This concept is two discs that are connected through the center. The front disc has a design like a rim. With an outer ring that is connected to the center with spokes. The back disc a soil thinner disc. By using separated discs can measurement can be performed without having the interfaces in direct interaction. The gauges can be placed somewhere along the spokes. The concept can be seen in figure 4.15 and an exploded view with the concept mounted on the FTR.

4.3.6

A - Rotating blocks

For the variation in stiffness this is the first concept for that requirement. This concept has four blocks that rotate in the groove of a base plate. The movement is controlled by a plate that contains four slots for the blocks and this plate acts as a lid. The blocks will be under the test specimen and the distance from the center will vary the stiffness, see figure 4.16.

4.3.7

B - Half plates

The second concept for stiffness is by using two half circle plates with a hole in the center for the bolt to go through. The distance between them is controlled by a threaded rod and a smooth rod that the plates will glide on. The stiffness will be lower when the plates is far apart and higher when close, see figure 4.17. To turn the threaded rod a key is used at one of the holes at the circumference.

4.3.8

C - Blocks with steps

4.3. Concepts 26

(a) Tension bars transducer overview.

(b) The transducer mounted on the test rig in

its position between the load cell and its holder. (c) View of the transducer without the load cell.

4.3. Concepts 27

(a) Torque Transducer. Image courtesy of HBM

GmbH. (b) The transducer assembled on the FTR.

Figure 4.14: Concept 4, a torque transducer that already is available.

(a) Double disc front view.

(b) Exploded view with the transducer between load cell and the holder.

4.3. Concepts 28

(a) Base and the blocks. _{(b) With cover to hold the}

blocks. (c) With the test specimen.

Figure 4.16: Concept A, rotating blocks under the test specimen.

(a) View of double plate assembly, the black rod is threaded and the gold is smooth.

(b) Mounted on the test rig.

4.4. Concept Evaluation 29

(a) The block place in three different ways for each different stiffness mode.

(b) View of a loose block. This orientation will give the system highest possible stiffness.

Figure 4.18: Concept C, blocks that change stiffness with placement.

is down into the head. This gives the test specimen a total flat surface to rest on and stiffness of the rig will be at maximum and same as without a stiffness changer. To reduce the stiffness each block is turned so that the peak outside is now closest to the center hole. As the right block in figure 4.18a. The lowest stiffness is achieved by having the mid peak of the blocks pointing upwards and the other peak downwards outside of the test specimen. The head can be modified so it can be combined with concept 1 (4.3.1) to also measure torque.

4.3.9

Stiffness Increase of Rig

Due the low stiffness of the FTR a concept that will increase the stiffness is made. In the concept the load adaptor is redesigned such that the key is removed and the adaptor is directly attached to the housing, see appendix A graphically representation. The holder for the bearing is removed as the thrust bearing is clamped. The load adaptor and nut adaptor holder is then made shorter and the hole housing is re-machined shorter. The guide strips are no longer necessary as the load adapter is rigidly attached.

4.4

Concept Evaluation

4.5. Final Design 30 twist the disk results in a small slippage and that will cause the sensor to read the wrong value. Therefore is this concept discarded and that is also because some of the torque will go through the bolts instead of the sensors.

The first concept ”Front shear” is denied because of manufacturing time. The special thread is a high risk of delay in manufacturing, this thread is of type M55x6. The concept has a desired deformation behaviour because of its cylindrical shape. It would also have been easy to mount on the FTR.

The sensor in key concept is not made because of the modification required to the housing. Doing this key concept without modifying the housing would have resulted in a bad results for measurements due to movement and uneven deformation.

The third concept ”Tension bars” is denied because torque will be transferred trough the surfaces by friction, this torque will not be measured by the gauges.

After discussion with suppliers of Torque Transducers are there found that not any have one that can handle the high clamp force that will occur. To use an existing transducer is therefore not possible.

The double disc concept is after talking with a supplier a bit uncertain to use because its similarity to rim and that is used in cars when you want to have a high rotational stiffness. The winning concept is therefore none of the initial but a combination between 1 and 5.

For a concept for change of the joint stiffness is non made because their are not rigid enough and would be permanently deformed.

The stiffness of the rig increases with 49 % with the new higher stiffness design. The concept is however not something that is currently not continued with because of the high modifications needed.

4.5

Final Design

The final design consist of a cylinder with two flanges. The cylinder wall is thin so that the device will get an angular displacement but not too thin so that the transducer can handle the axial clamping force. The flanges have a bolt pattern to fit with the FTR. There are threaded holes prepared for a cable clamp. Figure 4.19 shows a picture of the final design. To be easier to assemble and disassemble do the transducer have holes on opposite flange of the bolt pattern against the load cell adaptor to be able to stick in a Allen key and tighten the bolts. The cable clamp do also has hole on the opposite flange so it can be fastened.

4.5.1

Strain gauge

4.5. Final Design 31

Figure 4.19: View of the final design.

Figure 4.20: View of one of the flat areas.

4.5. Final Design 32

Table 4.3: Linearity in strain on the flat

Torque [Nm] Strain on cylinder [10-4 _{%] Strain on flat [10}-4 _{%] Difference [%]}

20 4.073 4.171 2.424

40 8.145 8.342 2.425

60 12.22 12.51 2.431

80 16.29 16.69 2.425

100 20.36 20.86 2.426

Table 4.4: Results from strength calculations Property Value Unit Maximum von-Mises Stress 310 MPa

Buckling factor 45

Axial deformation 20·10-4 % Axial stiffness 5.0·109 N/m New rig stiffness 130·106 _N/m

Stiffness reduction 4 %

4.5.2

Rigidity

C

HAPTER

5

Discussion and Conclusion

5.1

Sensors

Even though the sensor with the highest score is the optic, the semiconductive is used for the final concept. This is due to the high sensitivity that is needed only can be achieved with the optic and semiconductive. The semiconductive however can be used with the equipment that Atlas Copco already has making it a far less expensive option. The optic is not as common in lab a environment and have more documentation of it when being used for measuring structures and buildings, making this also a more uncertain option in a lab. A point in favour for the optic is that the electric noise and disturbance disappear and make them to something to look more deeper into the future.

Using a stock torque transducer could have a resolution high enough for this application but they have a fixed size and there was none that would fit the FTR. The FTR should in that case been designed initially to fit one. A stock torque transducer is often based on strain gauges already attached to known body and can therefor directly give the torque but gauges can be applied on any body that is later calibrated.

The piezoelectric is not very common when using special transducer but are found more common in standard applications. However the other characteristics of a piezoelectric sensor is not in favour for this application as seen in the results.

5.2

FTR

For the BLM the difference in torque between the BLM and the IRTT is probably due to the bearings. Bearings are not as suitable for slow moment with high forces, it can be that the bearings stop rotating or make a sudden rotational leap. When the front torque transducer is installed the total torque by the FTR can be measured and compared with the IRTT. If a difference exists here too a more deeper study should be considered. The improvements made on the FTR show that an increase of stiffness is achieved, which is

5.3. Underhead Torque Transducer 34 desired. The stiffness can be reduced by several techniques as for example longer clamp length but it is not possible to increase it without changing the base system. But to give a exact value for the stiffness should be done with great consideration for how much a tightening can vary with different test specimens and other small impacts that give a noticeable angle difference. The cable fix, re-soldering the contact reduce the offset and make the value more similar to the calibrated but it does not change the readings otherwise. However is it good to have the connection as intended.

For the validation of shank torque and crosstalk in load cell the results can indicate trends in how the behaviour differ between the rigs but because of that the DAQ is not suitable for calibration, the values are not exact. The system is also influenced by hysteresis and creep. The test for crosstalk measured in the load cell is done with a reverse torque that is applied between the applied front torques. This gives a hysteresis in the IRTT. This way is similar to how the standard procedure is done but not in line with a true torque calibration standard.

Regarding the shank torque validation bolts they have a high influence of the applied force and this is most likely due to misalignment of the gauges. This force sensitivity is high in comparison to the operating torque and force but as seen in the results is linear and could then be compensated. But when shank torque from the bolt is compared to the shank torque from the test rig is the error even greater with a compensation. Therefore the graph in the result is not compensated. It is unclear why this is, maybe the effect is not as great when bolt is aligned in the rig or the influence of compressive load is greater than tensile load. When taking into account the results from the back end test it can be seen that when only a torque is applied it shows the correct value. Another result had been strange since they are calibrated in that way. It is different when the shank torque transducer is also affected by the clamp load and maybe that is read by the transducer. It can at least be said that the FTR and BLM do differ and the shank torque variants of the FTR are more similar to each other than BLM. To make a good validation of the shank torque a bolt of this type should be sent away to get a better alignment of the gauges and then be calibrated with equipment only made for calibration. But that will have a high cost which have to be considered with the end purpose of the FTR.

5.3

Underhead Torque Transducer

5.4. Stiffness 35 The bearings are neither optimal for a slow application with high forces. The rigidity of the transducer is high even though the maximum Von-Misses stress is high but as the material have a yield point twice as high are there no risk for permanent deformation. The buckling factor is high and even though the calculations are for linear buckling there is no need to check for nonlinear buckling. The axial stiffness are so high that the rigs stiffness will only change with a few percent.

5.4

Stiffness

No concept for stiffness changer device was manufactured, all concepts encounter the problem with not being rigid enough. The high force makes it hard to make a small and flexible device that does not break when loaded. The same problem was encountered last year and no possible way to go around was found. Making a large device is not suitable either because it will probably have to be made major modifications on the current rig. Another aspect is the total stiffness of the rig that should be enough to test hard joints so a device that will change between soft and super soft is not doing any good. The concept of redesign the rig to increased the stiffness was not considered because of the modifications that need to be done. This concept is also a early idea and a deeper study should be made if later considered. However it can be concluded that a more stiff rig would is preferable and the modifications done during the project seems to have achieved that.

5.5

Conclusion

The first task constructing a transducer to measure the underhead torque was done. The stiffness changer device was not constructed since no concept could handle the force. However will the FTR now have at least same functionality as the BLM since it now can measure the same quantities. The study of sensor gave the final design semi conductive instead of the more common metallic foil. This will give the transducer a much higher amplification and allow for a stiffer design.

36

5.6

Future Work

R

EFERENCES

[1] “Atlas copco in brief.” https://www.atlascopcogroup.com/en/about-us/ atlas-copco-in-brief. Accessed: 2019-01-24.

[2] R. Afsharian and A. Theodoropoulos, “Modular friction test rig for measuring torque and tension in threaded fasteners,” Master’s thesis, KTH, 2018.

[3] R. S. Figliola, Theory and design for mechanical measurements. New York: Wiley & Sons, 3 ed., 2000.

[4] “How it works - strain gauge load cell.” https://www.rdpe.com/ex/hiw-sglc.htm. Accessed: 2019-01-29.

[5] C. S. Lynch, “Strain measurement,” in Measurement, Instrumentation, and Sensors Handbook (J. G. Webster, ed.), CRC Press LLC, 1999.

[6] OMEGA Engineering, Practical Strain Gage Measurements, 2017. Application Note 290-1—Practical Strain Gage Measurements, ©Agilent Technologies 1999.

[7] Micro-Measurments, Strain Gage Selection: Criteria, Procedures, Recommendations, May 2018.

[8] D. M. S¸tef˘anescu, Handbook of Force Transducers Principles and Components. Berlin, Heidelberg: Springer Berlin Heidelberg, 2011.

[9] D. Eberlein, “An introduction to measurements using strain gauges,” 2017. Webinar, HBM.

[10] K. Hoffmann, An Introduction to Stress Analysis using Strain Gauges. HBM. [11] G. Gautschi, Piezoelectrie sensories : force, strain, pressure, acceleration and acoustie

emission sensors, materials and amplifiers. Springer Berlin Heidelberg, 2002. [12] “Fiber optic sensors vs resistive strain gages,” 2015. Sensuron, rev-1.0.

38 [13] M. A. Elbestawi, “Force measurement,” in Measurement, Instrumentation, and

Sensors Handbook (J. G. Webster, ed.), CRC Press LLC, 1999.

[14] S. A. Nassar, G. Barber, and D. Zuo, “Bearing friction torque in bolted joints,” tech. rep., Fastening and Joining Research Institute, Mechanical Engineering Department, Oakland University, Rochester, Michigan, 2004.

[15] Swedish Fasteners Network, “G¨angans mekanik - l˚ang, handbok f¨or skruvf¨orband.” http://handbok.sfnskruv.se/template.asp?lank=189, 2016. Accessed: 2019-02-05.

[16] VDI/VDE-Gesellschaft Mess- und Automatisierungstechnik, Transducers for nu-trunning systems, Guideline for dynamic checking of tools according to ISO 5393, December 2005. VDI/VDE 2647.

[17] Mechanical properties of fasteners made of carbon steel and alloy steel - Bolts, screws and studs with specified property classes — Coarse thread and fine pitch thread, 2009. ISO 898-1:2009.

[18] Fasteners – Torque/clamp force testing, 2005. ISO 16047:2005.

[19] K. T. Ulrich and S. D. Eppinger, Product Design and Development. McGraw-Hill, 2012.

A

PPENDIX

A

Concept of Stiffness Increment

of the FTR.

In figure A.1 a section cut of the current FTR is shown and in figure A.2 a design proposal with increased stiffness. The new underhead torque transducer is not shown in the figures.

40

41

A

PPENDIX

B

Summarized Result of Hapticas

Report on the Flats Size Impact

on Strain Measurement.

This appendix only contain the calculated result from Hapticas report. For the flat three different sizes are tested and one test without the flat. The material in the simulation is aluminum with a Young’s modulus of 70 GPa and a Poisson’s ratio of 0.3. The calculated strain for the different sizes can be seen in table B.1.