Impactor design and development for in-pipe sewer inspection robot

Impactor design and development for in-pipe sewer inspection robot

J.H. (Jun Hong) Choo

BSc Report

Committee:

H. Noshahri, MSc dr.ir. E. Dertien dr.ir. L.L. Olde Scholtenhuis

July 2019

031RAM2019 Robotics and Mechatronics

EE-Math-CS University of Twente

P.O. Box 217

7500 AE Enschede

The Netherlands

iii

Contents

1 Introduction 1

1.1 Problem description . . . . 1 1.2 Goal and requirements . . . . 1 1.3 Outline . . . . 2

2 Feasibility study of possible impactors 3

2.1 Possible designs . . . . 3 2.2 Conclusion . . . . 5

3 Analysis 6

3.1 Theory of impact between steel ball and concrete . . . . 6 3.2 Modeling . . . . 7 3.3 Inertia of the rod . . . . 7

4 Experiments 10

4.1 Free falling steel ball experiments . . . . 10 4.2 Impactor experiment . . . . 15 4.3 Impact measurements using accelerometer . . . . 18

5 Conclusion and recommendations 22

5.1 Conclusion . . . . 22 5.2 Recommendations . . . . 22

6 Appendix 24

Bibliography 26

1

1 Introduction

1.1 Problem description

One of the main objectives of Technology Innovation for Sewer Condition Assessment - Long- distance Information-system (TISCALI) project is to arrive at an objective detection and quan- tification of defects in sewers and to determine the constructive strength and stability of sewers locally by means of an in-pipe robotic inspection. In order to reach this objective, a possible approach would be to use the impact-echo method that utilizes the acoustics properties and vibration of relatively large non-reinforced concrete sewer pipes. In this case, an in-pipe robot will be equipped with an impactor and sensors to perform inspection inside the concrete pipes.

However, different thicknesses of concrete requires different impact time to create the right ex- citation. Based on the article "Acoustic Condition Assessment of Concrete Sewer Pipes using a Particle Velocity Sensor" (Pleijsier, 2019), one of the main challenges is that thinner concrete has a higher thickness frequency, which would mean that a shorter impact time is needed to create a high frequency wave. A common way to excite the concrete is to use a steel ball at- tached to a rod by hand.

1.2 Goal and requirements

The goal of this project is to study the impact between a steel ball and concrete and to design an autonomous impactor that could generate the right impact to excite the sewer pipe with prescribed energy and duration. As mentioned above, the main challenge is to excite the thin concrete. However, in this project, different thickness of concrete tiles will be used instead of concrete pipes. And the thinnest concrete sample available in the lab has a thickness of 41mm.

According to Pleijsier (2019), the required contact time (impact time) is approximately 20 µs, and the author had only managed to achieve the approximate impact time of 500 µs using a linear solenoid impactor. Furthermore, it is important to study the energy of the impact gen- erated by the impactor so that the excitation can be recorded by the sensors. Moreover, the energy of the impact should not be too high that it will cause damage to the concrete. Ac- cording to Konstantin Kovler (2018), energy level of approximately 2.207J is the borderline, any higher than that would damage the concrete. Last but not least, it is crucial for data processing in the impact-echo method that the impactor could produce consistent impacts, and only hit the concrete once to avoid multiple excitations in one measurement.

A summary of the goal of this project is to study the impact and design an impactor. The re- quirements of the impactor are listed below:

1. The impactor should be able to produce consistent impact.

2. To avoid multiple excitation, the impactor should only impact the concrete surface once.

3. The excitation generated by the impactor should be large enough for the sensors to mea- sure but the impact energy should not be too strong and cause damage to the concrete.

4. The impactor should be able to excite the thickness frequency of 41mm tile.

5. The impactor should be able to excite the concrete in different orientations.

In this project, the first 4 goals listed above will be the main focus. However, when designing

the impactor, the last goal should be taken into account as the inspection of pipe requires the

impactor to work in different orientations.

1.3 Outline

This paper starts off with comparing potential designs of impactor. After choosing an impactor,

the analysis and modelling of the impactor will be demonstrated in chapter 3. Thereafter in

Chapter 4, multiple experiments will be conducted to examine the theory developed through

the analysis and modelling and to evaluate the performance of the impactor. Last but not least,

the final chapter provides the author’s conclusion and recommendations of the whole project.

3

2 Feasibility study of possible impactors

In this chapter, multiple impactors will be discussed and eventually one will be chosen for this project. Four possible impactor designs that may be able to achieve the goal of this project as stated in subsection 1.2 will be discussed in the section below.

2.1 Possible designs 1. First impactor design.

The first impactor is developed by Pleijsier (2019). In the paper, the author attached a steel ball to the magnetic solenoid and by carefully changing the distance from the solenoid to the impact surface and the power of the solenoid, the author was able to gen- erate an impact with the impact time as low as 500 µs. However, the author was not able to achieve impact time lower than 500 µs using the magnetic solenoid because the inertia of the rod is too large and causes extra force to act on the concrete after the acceleration.

The steel rod that is used in the solenoid cannot be easily substituted with a thinner steel rod as it will result in the reduction of the magnetic force of the solenoid. Therefore, this design is not suitable for this project.

2. Second impactor design.

The second design is a pneumatic linear impactor that consists of one double-acting pneumatic cylinder controlled by two individual 3/2 valves to create a linear impact movement. The reason for using two individual valves instead of one 5/2 valve is due to the fact that a 5/2 solenoid valve would normally have around 20ms of switching time (time taken to turn on one valve and turn off the other). The 5/2 valve limits the impact time to be at least 20ms if not higher. Free movement can be simulated when both in- dividual 3/2 valves on the double-acting cylinder is in an open state. As a result, when impact occurs, the rod can naturally bounce back instead of having to wait 20ms for the 5/2 valve to switch. However, pneumatic linear impactor is hard to control and difficult to be modelled as the inward and outward airflow would fluctuate and might not be con- stant throughout the whole operation. Furthermore, given that air is compressible, it will be hard to predict the exact time of the rod hitting the concrete surface and there will be less control of the impact.

An example of the operating steps can be seen in table 2.1 and the schematic of the pneu-

matic linear impactor inspired by Tameson (2019) is shown in figure 2.1.

Step 1st NC valve 2nd NC valve Piston movement

1 1 0 Accelerating to a

2 0 0 Free movement from a

to a

(where impact happens)

3 0 1 Accelerating back to a

Table 2.1: Piston movement with steps (0: not actuated, 1: actuated), NC: normally closed valve

Figure 2.1: Rough model of pneumatic linear impactor (Tameson, 2019), NC: Normally closed

3. Third impactor design.

The third impactor is a speed-controlled impactor. It is designed in a way that a light- weighted rod with a steel ball attached at the end of the rod will be accelerated using a servo. Before the steel ball hits the concrete, the servo will stop the acceleration and let the steel ball continue moving freely with a constant speed to hit the concrete naturally.

A rough model of impactor can be seen in figure 2.2.

Figure 2.2: Rough model for rotational speed controlled impactor

CHAPTER 2. FEASIBILITY STUDY OF POSSIBLE IMPACTORS 5

While the bigger gear is driven by a servo, the smaller gear increases the rotational speed and accelerates the freely rotating rod with its extension rod. The relation between gear ratio and the impact speed can be calculated using the equation 2.1.

v = ω ∗ r ∗GR (2.1)

Where GR is gear ratio, r is length of the rod and ω is the rotational speed of the servo and v is the impact speed.

After the impact, the gear will rotate in the opposite direction and the other extension rod from the smaller gear will catch the freely rotating rod, preventing it from hitting the concrete again.

4. Fourth impactor design.

The fourth impactor design is created by Sivasubramanian et al. (2016). A spring will be attached to the steel wire and the frame. When the motor raises the steel ball, the spring will act as an external force to pull the steel ball down at a higher speed. The resting position of the steel ball and the length of the rod have to be carefully measured and calculated so that the external force from the spring will be negligible during the impact.

The main reason for this is that the extra force will increase the impact time, which is undesirable to achieve the goal of this project.

Figure 2.3: Fourth impactor design (Sivasubramanian et al., 2016)

2.2 Conclusion

After comparing all the potential design of impactors, third and fourth design both have the

potentials to achieve the goals as stated in section 1.2. Nevertheless, the third impactor is a

relatively simpler design compared to the forth impactor and it can be developed and built

within the time frame of this project. In conclusion, the third design will be developed for this

project and the detailed impact time calculation will be shown in section 3.1.

3 Analysis

3.1 Theory of impact between steel ball and concrete 3.1.1 Elastic contact of steel ball and concrete

As the impact time is mainly caused by the elastic contact of the steel ball and the concrete, Hertzian theory of non-adhesive elastic contact by Negrea and Predoi (2012) is applied in the model to study the behaviour of the impact. Since the elastic modulus (Young’s modulus) of steel (200GPa) is approximately 10 times greater than the elastic modulus of the concrete (20GPa), most of the deformation will occur on the concrete so the displacement of the steel ball will be approximately equivalent to the parameter d. The equations to describe Hertzian theory of non-adhesive elastic contact is shown below.

F

= 4

3 E

R

d

(3.1)

u = d

d t d (3.2)

a = p

Rd (3.3)

Where R is the radius of the steel ball, and F

is the force stored by the elastic deformation, a is the radius of the contact area of the steel ball, d is the deformation of the concrete and u is the speed of the steel ball.

The combined elastic modulus of the system (E

) is calculated using the equation 3.4.

1

E

= 1 − ν

E

+ 1 − ν

E

(3.4) Where ν

, ν

are the Poisson’s ratios of steel and concrete respectively, E

, E

are elastic moduli of steel and concrete respectively.

A conceptual model of the impact theory can be seen in figure 3.1.

Figure 3.1: Contact of an elastic sphere with an elastic half-space (Commons, 2014).

F is the force on the steel ball, a is the radius of the contact area of the steel ball and d is the deformation

of the concrete.

CHAPTER 3. ANALYSIS 7

3.2 Modeling

Using the equations stated in section 3.1, the modelling of the behaviour of impact has been done via 20-sim (Controllab Products, 2008). This model aims to simulate the steel ball with an initial impact speed and with no other external force other than gravitational force. The purpose of this model is to have a better understanding of how the impact happens. This model simulates only when the impact happens so an impact speed will be set in the initial parameter of 20-sim model. The gravity and damping effect will be added in the model during the development of the impactor if necessary. However, to excite the concrete tile with thick- ness of 41mm, the diameter of the steel ball is limited to a maximum of 4.8mm according to Pleijsier (2019). Since the size of the steel ball is very small, the gravity is neglectable. Besides that, the damping effect of the steel ball is not the main concern, as only the first period of the impact is the main point of interest. The damping effect is only significant when multiple impact periods happen.

A picture of the model can be seen in figure 3.2:

Figure 3.2: Impact modelling in 20-sim

This model consists of the mass of the steel ball and a spring. The spring in the model has been modified with Hertzian theory (3.4, 3.1) to simulate elastic contact between the steel ball and concrete during impact. The code of the spring is shown in appendix 6.1:

3.3 Inertia of the rod

The impactor has an extra element compared to the 20-sim model - the rod. The presence of the rod would increase the inertia of the system so the model needs to be modified accordingly.

The inertia of the steel ball is:

I = I

+ I

(3.5)

I

= 2

5 ∗ m

∗ r

(3.6)

I

= m

∗ L

(3.7)

The inertia of the whole system (including the rod) is:

I

= I

+ I

+ I

(3.8)

I

= 1

3 ∗ m

∗ (L − r )

(3.9)

where m

is mass of the steel ball, m

is the mass of the rod, L is the length of the rod and r is the radius of the steel ball.

In order to match the increased inertia caused by the rod, the mass of the steel ball in the model has to be adjusted accordingly. The effective mass of the steel ball will now be:

m

= I

2

5

∗ r

+ L

(3.10)

For instance, when using a rod with a length of 0.12m and linear density of 0.0062kg/m (top rod in figure 3.3), the effective mass of the steel ball will be 1.0022 times the original mass. Another example would be to use a same length but thicker rod (0.0158kg/m, bottom rod in figure 3.3), where the effective mass of the steel ball in this scenario will be 1.0057 times the original mass.

Therefore, the influence of the rod in both cases is negligible.

Figure 3.3: Rods used in the experiments

3.3.1 Modelling results

As the size of the steel ball and initial contact speed are factors that could greatly affect the

impact time and energy, multiple simulations have been conducted using steel ball with a di-

ameter of 3mm and 6mm, as those are available in the lab. The simulation results are shown in

figure 3.4:

CHAPTER 3. ANALYSIS 9

Figure 3.4: 3mm(left) and 6mm(right) diameter steel ball with different initial impact speed

Initial impact speed (ms

)

Impact time ( µs)

1 22.4

2 19.6

3 17.9

4 17.1

5 16.3

Table 3.1: Impact time of steel ball with diam- eter of 3mm with different initial impact speed

Initial impact speed (ms

)

Impact time ( µs)

10 28.1

15 25.9

20 24.5

45 21.0

60 19.8

Table 3.2: Impact time of steel ball with diam- eter of 6mm with different initial impact speed As shown in table 3.1 and 3.2, the steel ball with a diameter of 3mm could reach 20 µs impact time with only 2ms

impact speed. On the other hand, the impact speed required for the 6mm diameter steel ball to achieve 20 µs contact time is 30 times higher than the steel ball with a diameter of 6mm (60ms

). Therefore, it is more feasible to achieve 20 µs impact time with the steel ball with a diameter of 3mm than of 6mm.

Despite that the steel ball with a diameter of 3mm could achieve 20 µs impact time with a rela- tively low speed as compared to the 6mm diameter, the impulse J = R F d t of the steel ball with a diameter of 3mm is 200 times lower compared to the steel ball with a diameter of 6mm. To reiterate, the impulse generated by the impact greatly decreases when using a smaller steel ball as shown in table 3.3.

Diameter of steel ball (mm) Theoretical impulse (Ns)

3 0.0005

4 0.0042

4.5 0.0112

6 0.1065

Table 3.3: The theoretical impulse of steel ball with different diameter and a 20 microsecond impact

time.

4 Experiments

In this chapter, 3 different experiments will be performed to examine the theory, compare the experiment results with the simulation results and evaluate the performance of the impactor.

4.1 Free falling steel ball experiments

The purpose of this experiment is to verify the impact time of the impact theory that is dis- cussed in chapter 3.1.

4.1.1 Experiment setup

First of all, a force sensor is required to measure the force of the impact of the impactor and calculate the impact time. Through calculations and simulations, the range of the force during impact will be within 1000N(max). The force sensor that will be used in this experiment is

"FlexiForce A201 100lbs, Tekscan" as shown in figure 4.1. Although the sensor is rated at 100lbs (445N), with a slight adjustment with the amplifier shown in figure 4.1, it can measure up to 4448N as stated in the datasheet. However, if a steel ball with a diameter of 3mm or 6mm directly impacts on the force sensor, the force sensor will break after a few impacts. Therefore, an external setup to protect the force sensor is necessary.

Figure 4.1: Flexiforce A201 100lbs, Tekscan (left). Amplifier product from Tekscan (right)

One way of measuring the impact force of the concrete without damaging the force sensor is by placing a small piece of thin concrete cylinder on the sensing area of the force sensor. Since thin concrete will be too fragile to achieve that, Acrylic (PMMA) and Polyoxymethylene (POM) cylinders will be used in this experiment. An example can be seen in figure 4.2:

Figure 4.2: PMMA cylinder on the force sensor (left), cylinders used in this experiment (right)

CHAPTER 4. EXPERIMENTS 11

4.1.2 Theory and Expectation

This experiment will not involve the use of the rod, but instead, the ball is given an initial impact speed with no other force except the gravitational force. A steel ball with 3mm diameter will be dropped on the cylinders and the force that acts on the cylinders will be measured. According to SpecialChem SA (2019), PMMA has a Young’s modulus ranging from 1 to 3.5 GPa and POM has an Young’s modulus ranging from 1.4 to 5.5 GPa depending on the manufacturer. Even though the Young’s modulus is not exactly close to the ideal Young’s modulus of the concrete (20GPa), it is still a good approach to test the impact theory. The impact speed will be controlled by dropping the steel ball from a different heights. It is expected that the impact time will increase when the impact speed decreases. Although the impact force amplitude will not be the same as the measurement from the force sensor because of the energy loss in the deformation of the material, the response of the impact will be the same.

4.1.3 Results

Firstly, the experiment is done using the POM cylinder with two different heights (0.45m and 0.28m). As the theory is valid only if the steel ball is hitting a relatively large surface compared to the impact area of the steel ball, the steel ball has to precisely hit the center of the cylinder otherwise the force will not act in accordance to the theory. Hitting the side of the cylinder will cause the impact respond to behave differently. An example of the measurement of the force sensor is shown in figure 4.3:

Figure 4.3: Normalized force sensor measurement by dropping a steel ball with diameter of 3mm on a POM cylinder from 0.45m height(left) 0.28m(right)

As shown in figure 4.3, the measurements do not start from zero because the oscilloscope con-

tinuously records the signal and triggers only when the signal reaches a certain threshold. In

some measurements, due to the bounces of the steel ball on the surface or the reflection of

the excitation wave, there are multiple pulses in the measurements. The impact time will be

the first pulse of the measurement of the force sensor. In the left plot of figure 4.3, a half-sine

shaped pulse can be clearly seen in the figure so the impact time is directly equal to the du-

ration of the first pulse. However, in the right plot of figure 4.3, the first pulse is not a perfect

half-sine shape. The force slowly increases in the beginning and slowly decays at the end. In-

stead of taking the duration of the pulse, a sine wave will be fitted using MATLAB (2019) and the

impact time will be half of the period of the fitted sine. The reason of fitting a sine wave instead

of taking the whole duration of the pulse is because the wave excited by the impact will not be

related to the slowly increasing phase and decreasing phase. The excitation wave is generated

by the sine-shaped impact force. An example of the Matlab code for sine fitting can be seen in

appendix 6.2.

The result after filtering the noise and sine fitting is shown in figure 4.4.

Figure 4.4: Noise filtering and sine fitting the measurements in figure 4.3

After sine fitting, the impact time is measured by taking half of the period of the fitted sine. The contact time of the measurements is shown in table 4.1.

Measurements impact time ( µs) Measurements impact time ( µs)

1 42.20 1 47.31

2 45.23 2 48.78

3 40.15 3 52.82

mean 42.53 mean 49.64

standard deviation 2.56 standard deviation 2.85

Table 4.1: The measured impact time by dropping a 3mm-diameter steel ball on a POM cylinder from a height of 0.45m (left) and 0.28m (right).

The same experiments are performed also on the PMMA cylinder. The results are shown in figure 4.5.

Figure 4.5: Normalized force sensor measurement by dropping a 3mm-diameter steel ball on a PMMA

cylinder from a height of 0.45m (left) and 0.28m (right)

CHAPTER 4. EXPERIMENTS 13

The measured impact time is shown in table 4.2.

Measurements impact time ( µs) Measurements impact time ( µs)

1 43.29 1 47.88

2 45.24 2 48.97

3 41.31 3 47.34

mean 43.28 mean 48.06

standard deviation 1.97 standard deviation 0.83

Table 4.2: The measured impact time by dropping a 3mm-diameter steel ball on a PMMA cylinder from a height of 0.45m (left) and 0.28m (right).

4.1.4 Discussion

From table 4.1 and 4.2, it is clear that the impact time increases when the impact speed de- creases. In addition, using the same impact time tables as a parameter in the 20-sim model as shown in section 3.2, Young’s modulus of POM and PMMA results in the expected range of Young’s modulus of the material, which are both around 1.3GPa to 1.6GPa. For example, the Young’s modulus of both POM (Ultraform N 2640 Z4 BK140 Q600) and PMMA (DRT Frosted 18) are rated in that range.

The way of controlling the impact time will be first, to use different sizes of steel ball. Second, to use the impact speed to fine-tune the impact time. The initial plan for this experiment also includes the testing of the steel ball with a diameter of 6mm. However, the size of the sensor is too small for the steel ball with a diameter of 6mm to have enough contact area to simulate an infinite plane. Furthermore, the steel ball with a diameter of 3mm has to hit the center of the cylinder to have a good impact respond. This increases the difficulty of performing the experiment and reduces the accuracy of the result. Other than that, the amplifier that is used in this experiment has noise in the measurement as shown in figure 4.3. A custom amplifier as can be seen in figure 4.6 has been made after this experiment and the issue has been solved.

The comparison of the measurement using different amplifier can be seen in figure 4.7.

Figure 4.6: Customized amplifier (left), schematic of the amplifier circuit (right)

Figure 4.7: The measurement using amplifier that has noise (left) and measurement using custom made amplifier (right)

4.1.5 Conclusion

From this experiment, it can be concluded that the impact outcome matches the impact theory

as discussed in section 3.1. The measurements show that the impact time decreases when the

impact speed increases as expected. In addition, the Young’s modulus of POM and PMMA

obtained by the experiment are both within a reasonable range.

CHAPTER 4. EXPERIMENTS 15

4.2 Impactor experiment

The purpose of this experiment is to prove that the influence of the rod of the impactor to the impact time is negligible as explained in section 3.3, and that the impactor can produce consistent impact.

4.2.1 Experiment setup

The setup of this experiment is similar to the previous experiment. The only difference is that instead of dropping the steel ball from a distance, the impact will be performed by the impactor prototype shown in figure 4.8. Since the length of the rods remain unchanged, theoretically the impact speed of the steel ball will be the same given that the servo rotates at a constant speed.

This is proven by the use of a high-speed camera. Using the high-speed camera, the time is measured from the start of the rotation of the gears until the steel ball that is at the end of the rod hits the cylinder. In both cases, the total time are the same hence prove that in both cases have the same impact speed. Besides that, the large steel cylinder used in this experiment serves two purposes. The first purpose is to provide a hard base to support the sensor. Another purpose is to provide extra height to the impact point. It is important to ensure that the impact is perpendicular to the surface to match the impact theory. The complete setup can be seen in figure 4.9.

Figure 4.8: Impactor prototype (left) and two different rods attached with a steel ball with 3mm- diameter (right)

Figure 4.9: Complete setup of the impactor experiment

4.2.2 Theory and Expectation

Since the effective mass of the steel ball is only 1.0022 and 1.0057 times higher than the original mass of the steel ball as explained in section 3.3, both rods are expected to have very similar impact time. Moreover, given the impact speed is constant, it is also expected that both rods have consistent impact strength.

4.2.3 Results

The measurement results can be seen in figure 4.10. In addition, the estimated impact time and maximum amplitude can be seen in table 4.3 and 4.4.

Figure 4.10: Force sensor measurements with 0.0062kg/m (left) and 0.0158kg/m (right) rod

Measurement Impact time (µs) Maximum amplitude (V)

1 24.89 6.578

2 25.72 6.734

3 25.97 6.578

4 25.77 6.031

5 25.32 6.109

6 24.23 6.265

7 24.01 5.933

mean 25.13 6.318

standard deviation 0.78 0.312

Table 4.3: Impact time and maximum amplitude of the measurements using a rod with linear density of

0.0062kg/m

CHAPTER 4. EXPERIMENTS 17

Measurement Impact time ( µs) Maximum amplitude (V)

1 25.23 6.910

2 24.97 6.851

3 25.32 6.773

4 25.05 6.988

5 24.11 6.949

6 23.77 6.949

7 24.30 6.890

mean 24.68 6.901

standard deviation 0.61 0.072

Table 4.4: Impact time and maximum amplitude of the measurements using a rod with linear density of 0.0158kg/m

4.2.4 Discussion

Table 4.3 and 4.3 shows that the use of different rods (in figure 3.3) does not significantly change the outcome of the impact time as the different in mean is ±0.45µs which lies within the stan- dard deviation of both cases. On the other hand, the low standard deviation of the maximum impact strength and the impact time show that the impactor could produce very consistent impacts.

Initially, this experiment was also intended to prove that using a rod will not cause a significant difference in impact time as compared to dropping a steel ball with the same impact speed.

However, due to the limitations of the high-speed camera, it only showed that the rotational speed of both rods are the same but failed to obtain the data on the actual impact speed of the steel ball. Without the exact impact speed of the steel ball, it is impossible to prove that the use of a rod does not affect the impact time as compared to the impact time achieved by dropping a steel ball. Besides that, the value of the maximum amplitude in table 4.3 and 4.4 are obtained directly from the sensor’s measurements, which contains a constant offset of approximately 2V in every measurement. An example can be seen in figure 4.11.

Figure 4.11: Unprocessed force sensor measurement

4.2.5 Conclusion

This experiment proves that a slight difference in the linear density of the rod does not have sig-

nificant effect on the impact time and that the impact generated by the impactor is consistent,

which corresponds to the expectations and theory.

4.3 Impact measurements using accelerometer

The purpose of the last experiment is to evaluate the performance of the impactor on the con- crete. This experiment will be divided into two small experiments. The first sub-experiment will be the concrete excitation’s waveform measurement and impact time measurement. The second sub-experiment will be the impact by hand measurement.

4.3.1 Experiment setup

An accelerometer with a signal conditioner will be used for this experiment. For the first sub- experiments, the impact will be generated by the impactor equipped with a rod with linear density of 0.0062kg/m and a steel ball with a diameter of 3mm (top rod in figure 3.3). The impact of the second sub-experiment will be generated by hand using the same rod. It will be referred to as the handheld impactor. The setup can be seen in figure 4.12.

Figure 4.12: Setup of the impact measurements using accelerometer

4.3.2 Theory and Expectation

According to the theory in section 3.1, the impactor should be able to excite the thickness fre-

quency of the concrete tile with a thickness of 41mm. Other than that, according to paper

Carino (2015), impact time can be extracted through the measurement’s waveform by taking

the duration of the half-sine of the highest peak that is located in the beginning of the wave

form. Using this method, the impact time of the impact can be estimated and it is expected

to be lower than 20µs. Lastly, the impact generated by the impactor is expected to be more

consistent than the impact generated by the handheld impactor.

CHAPTER 4. EXPERIMENTS 19

4.3.3 Results

For the first sub-experiment, the accelerometer is placed 2cm away from the impact point, the measurement’s waveform and the Fourier transform of the measurement are shown in figure 4.13.

Figure 4.13: Measurement of accelerometer (left, zoomed) and Fourier transform of the measurement (right)

Impact time estimation can be done by fitting a sine to the highest peak at the beginning of the waveform.

Figure 4.14: Impact time estimation

By varying the distance between accelerometer and the impact point and repeating impact for seven times, the average and standard deviation of impact time are shown in table 4.5.

Distance between impact point

and the accelerometer (cm) Mean of the impact time ( µs) Standard deviation of the impact time (µs)

1 10.82 1.158

1.5 12.46 0.095

2 12.03 0.120

2.5 11.76 0.068

3 10.00 0.723

Table 4.5: The average and standard deviation of impact time extracted from measurements using ac-

celerometer with different distances

And the measurement’s waveform of the impact generated by handheld impactor with 2cm distance between impact point and the accelerometer is shown in figure 4.15. The measured impact time is shown in table 4.6.

Figure 4.15: Measurement’s waveform

Measurement Impact time ( µs)

1 11.52

2 9.15

3 10.44

4 9.09

5 10.04

6 11.21

7 10.71

mean 10.31

Standard deviation 0.94

Table 4.6: Measurement of the impact by generated handheld impactor with 2cm distance between impact point and the accelerometer

4.3.4 Discussion

This experiment encapsulated the conclusion that the impactor is able to excite the thickness

frequency of a 41mm tile and it is 7.83 times more consistent than a handheld impactor. How-

ever, the accelerometer has an increasing sensitivity from 15kHz to 100kHz as shown in figure

4.16, which is within the range of the Fourier transformed signal as shown in the right sub-plot

of figure 4.13. Due to the increased sensitivity, the signal with higher frequency will have a

lower amplitude in reality compared to the amplitude in the Fourier transform. In addition, as

the frequencies higher than 7kHz have higher gain than the frequencies between 4kHz to 6kHz

and the Fourier transformed signals that are higher than 7kHz have a very low and constant

amplitude, it can be concluded that the Fourier transform in figure 4.13 is not caused by the

frequency response of the accelerometer.

CHAPTER 4. EXPERIMENTS 21

Furthermore, the impact energy should also be calculated by measuring the speed of the steel ball before impact and the height of the steel ball after impact using a high-speed camera. The impact energy can be calculated using equation 4.1.

E = 1

2 mv

− mg h

(4.1)

where E = impact energy, m = mass of steel ball, v

= speed of steel ball before impact and h

= vertical distance of steel ball from the impact surface.

However, the high-speed camera available in the lab is not powerful enough to accurately mea- sure the speed of the steel ball. Therefore the impact energy can not be calculated. After ten trials of impact using the impactor at the same impact point, there is no visible damage on the concrete.

Figure 4.16: Frequency respond of the accelerometer

4.3.5 Conclusion

As can be seen in figure 4.13, while the thickness frequency of the 41mm tile is estimated to be

42kHz using formulas derived from Pleijsier (2019), the impactor is able to excite frequencies

as high as 51kHz. The impact time estimation in table 4.5 is shown to be changing in accor-

dance to the distance between the impact point and the accelerometer. Although the impact

time is changing in accordance to the distance, the impact time estimation with fixed distance

(ranging from 1.5cm to 2.5cm) is very consistent proven by the low standard deviation. Also,

the impactor is proven to be more consistent than the handheld impactor by comparing the

results of the impact time excited by handheld impactor. The standard deviation of the impact

time by the impactor is 7.83 times better than the handheld impactor. Lastly, the impact en-

ergy is high enough to be recorded by the sensor but not too strong and cause damage to the

concrete.

5 Conclusion and recommendations

5.1 Conclusion

As mentioned in section 1.2, the goal of this project is to study the impact between a steel ball and concrete and to design an autonomous impactor that can generate the right impact to ex- cite the sewer pipe with prescribed energy and duration. First, in section 4.3, the impactor is proved to be able to excite the concrete tile with a thickness of 41mm. The impact generated by the impactor has an impact time of less than 20 µs and has managed to excite frequencies up to 51kHz.

Second, the impactor experiment proved that the impact generated by the impactor is consis- tent. The ability to produce consistent impacts can ensure the accuracy in determining the condition of the concrete by processing the data obtained by the sensor; which is crucial for the purpose of the TISCALI project that was mentioned in the introduction section.

Third, this impactor can also successfully limit the number of impacts to only one impact to avoid multiple excitations in one measurement. It makes the result of all of the experiments more than satisfactory as this is one of the most essential features of the project. Failure to achieve this goal will result in multiple excitation waves being overlapped and will jeopardize the accuracy and reliability of the wave reading.

Lastly, although the prototype impactor is not suitable in different orientation as the freely ro- tating rod is not secure, it can be easily modified to be used in different orientations by mount- ing a spring in between its resting position and one of the extension rod. The spring will restrict the freely rotating rod’s movement as shown in figure 5.1.

Figure 5.1: An example of improved version of the impactor

5.2 Recommendations

The first recommendation put forward by the author is to resize the impactor by redesigning

the bigger gear. The current design of the impactor is such that the freely rotating rod with

the steel ball can hit the concrete without having the smaller gear to rotate for more than one

rotation. As a result, during the rotation of the gears, only part of the bigger gear is needed to

complete the impact. As such, depending on the situation and requirements of the inspection,

the unused part of the bigger gear can be cut in accordance with the gear ratio as shown in

figure 5.1.

CHAPTER 5. CONCLUSION AND RECOMMENDATIONS 23

Moreover, the use of a larger force sensor is also recommended as it will provide a better impact force measurement as the impact theory assumes an infinite impact area when comparing to the impact area of the steel ball as mentioned in section 3.1.

In addition, a new sensor that is capable of measuring frequency higher than the thickness fre- quency of the concrete is recommended to properly measure the thickness frequency without extra gain or attenuation. The improved accuracy of the data would allow the data collected by the sensor to be further processed to determine the condition of the concrete pipe.

Last but not least, a new high-speed camera that is able to capture the speed of the steel ball is

important to further investigate the theory, the performance of the impactor and to calculate

the impact energy.

6 Appendix

 

spring : {

parameters

r e a l b a l l _ r a d i u s = 0.0025 {m} ; r e a l E_ball = 200e09 { Pa } ; r e a l E_concrete = 20e09 { Pa } ; r e a l p s _ s t e e l = 0 . 3 0 ;

r e a l ps_concrete = 0 . 1 8 ; v a r i a b l e s

r e a l d ;

r e a l E _ t o t a l ; r e a l Energy ; equations

1/ E _ t o t a l =(1− ps_steel ^2)/ E_ball +(1−ps_concrete ^2)/ E_concrete ; d = i n t (p . v ) ;

i f d<0 then d=0;

end ;

p . F = ( 4 / 3 ) * E_total * ball_radius ^(1/2) * d ^ ( 3 / 2 ) ; Impulse = i n t ( p . F ) ;

 } 

Listing 6.1: 20-sim spring model code

 

function [ f i t r e s u l t , gof , c ] = c r e a t e F i t 3 ( time , Amplitude )

%CREATEFIT(TIME,AMPLITUDE)

% Create a f i t .

%

% Data f o r ’ u n t i t l e d f i t 3 ’ f i t :

% X Input : time

% Y Output : Amplitude

% Output :

% f i t r e s u l t : a f i t object representing the f i t .

% gof : s t r u c t u r e with goodness−of f i t info .

%

% See al so FIT , CFIT , SFIT .

% Auto−generated by MATLAB on 04−Jun−2019 14:17:59

%% F i t : ’ u n t i t l e d f i t 3 ’ .

[ xData , yData ] = prepareCurveData ( time , Amplitude ) ;

% Set up f i t t y p e and options . f t = f i t t y p e ( ’ sin1 ’ ) ;

excludedPoints = excludedata ( xData , yData , ’ Indices ’ , [ 1 2 3 4 5 6 . . . ] ) ; opts = f i t o p t i o n s ( ’Method ’ , ’ NonlinearLeastSquares ’ ) ;

opts . Display = ’ Off ’ ;

CHAPTER 6. APPENDIX 25

opts . Lower = [−I n f 0 −I n f ] ;

opts . S t a r t P o i n t = [ 4.23483196273013 78156.8477855954 0.929199428612899 ] ; opts . Exclude = excludedPoints ;

c = 1 . 9 0 7 ;

% F i t model to data .

[ f i t r e s u l t , gof ] = f i t ( xData , yData−c , f t , opts ) ;

 

Listing 6.2: Example Matlab code for sine fitting

Bibliography

Carino, N. J. (2015), Impact Echo: The Fundamentals.

https:

//www.ndt.net/article/ndtce2015/papers/257_carino_nicholas.pdf Commons, W. (2014), Contact of an elastic sphere with an elastic half-space.

Controllab Products (2008), 20-sim.

http://www.20-sim.com/

Konstantin Kovler, Fengzhe Wang, B. M. (2018), Testing of concrete by rebound method: Leeb versus Schmidt hammers.

MATLAB (2019), version R2019a.

Negrea, A. and M. Predoi (2012), The elastic contact of a sphere with an elastic half-space, a comparision between analytical and finite element solutions, Scientific Bulletin. Series A:

Applied Mathematics and Physics. Politehnica University of Bucharest, vol. 74.

Pleijsier, A. A. (2019), Acoustic Condition Assessment of Concrete Sewer Pipes using a Particle Velocity Sensor, CE, Utwente.

Sivasubramanian, K., K. P. Jaya and M. Neelamegam (2016), Virtual Edge Extension Technique to Reduce the Edge Effect in Impact-Echo Method, vol. 30, no.2, p. 04014205,

doi:10.1061/(ASCE)CF.1943-5509.0000718.

https://ascelibrary.org/doi/abs/10.1061/%28ASCE%29CF.1943-5509.

0000718

SpecialChem SA (2019), Young’s Modulus, Accessed: 21-06-2019.

https://omnexus.specialchem.com/polymer-properties/properties/

young-modulus

Tameson (2019), 3/2-Way Pneumatic Valve, Accessed: 21-06-2019.

https://tameson.com/32-way-pneumatic-valve.html