Impactor Design for In-pipe Impact-echo Inspection Method: A Morphological Analysis

(1)

Impactor Design for In-pipe Impact-echo Inspection Method:

A Morphological Analysis J. (Jeffrey) Stroet

BSC ASSIGNMENT

Committee:

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

February, 2020

008RaM2020 Robotics and Mechatronics

EEMCS University of Twente P.O. Box 217 7500 AE Enschede The Netherlands

(2)

(3)

Introduction

1.1 Problem description

Damaged sewer pipelines can cause many problems like sinkholes in urban areas and are very ex- pensive to repair once broken. TISCALI, short for Technological Innovation for Sewer Condition Assessment - Long-range Information, is a project which focuses on finding new ways of localiz- ing damage in sewer pipelines before severe damage is caused. One of these methods is called the impact-echo method, which is a method that releases stress waves in concrete to detect flaws (Carino 2001). The impact method is often used in civil engineering to assess, for example, bridges.

However, these tests are manually operated by a person. Small sewer pipelines are not accessible for human operators and thus is a manual operation of the impact-echo method not possible. TIS- CALI is designing a crawler robot equipped with an impact system and sensors which should be able to perform measurements remotely.

1.2 Goals and requirements

The goal of this bachelor thesis is to design and realize an impact system which will release stress wave in the concrete according to the impact-echo system. This impactor has several requirements, which are listed below.

• system must have an impact time of 40 to 80 µs. With this impact time, a frequency range from 12,5 kHz (80 µs) up to 25 kHz (40 µs) can be released into the concrete. This frequency range is higher than the thickness frequency of concrete in the sewer pipelines which is needed to assess its condition (Pleijsier 2019).

• The energy upon impact must be between 0.7 - 2.2 J to be able to measure the signal over the noise but not to destroy the concrete pipeline (Kovler, Wang, and Muravin 2018).

• To measure a clear signal, the impactor should be retracted directly after the first impact and before a second impact is made.

• The system itself should be as silent as possible in order to measure the condition of the sewer pipeline clearly.

• The system could be to be able to make an impact throughout the whole cross-section of the sewer pipeline. If this requirement is satisfied, the condition of the whole cross-section of the sewer pipeline can be assessed.

Some impactor designs have already been realized and analyzed by RaM, however non of these designs have satisfied all of the requirements. To explore new design ideas, a morphological analysis is performed. At the end of this bachelor thesis, one of the generated ideas is realized according to the listed requirements.

The main focus for the impact-echo system during this thesis will be on the first three requirements.

However, the last two requirements should also be taken into account.

(5)

1.3. OUTLINE CHAPTER 1. INTRODUCTION

1.3 Outline

After describing the problem context and the goals and requirements of this project, the morphological analysis will be explained and performed. From this morphological analysis, multiple design ideas are generated. After a brief description of previous impactor designs, one generated design idea will be further analysed, and its mechanical model will be presented. In chapter 3, these parameters are specified and implemented, and a short description of the measurement setups will be given. Finally, chapter 4 will present the results of the measurements followed by a conclusion and recommendations for future research.

(6)

Chapter 2

Analysis

2.1 Morphological Analysis

As stated shortly in the introduction, the morphological analysis is a structured method for gen- erating new design ideas to solve a non-quantified, multi-dimensional problem. It starts with a design problem that is divided into some key features for the design. Possible ways to realize these key features should be written down. Finally, generate the new design ideas, the realizations of the key functions are put together in a multi-dimensional table. From this table, the illogical design ideas can be eliminated to be left with realizable design ideas.

In this thesis, the design problem is to design a system capable of exciting concrete. There are three key features that belong to this problem.

• The system should make an impact using an impactor.

• The impactor should be retracted to be able to perform multiple measurements afterwards, and to avoid making multiple consecutive impacts.

• The system should be able to make an impact throughout the whole cross-section of the sewer pipeline.

To realise the first key feature, actuators and storage elements have been analysed and are listed below in table 2.1. Both electrical and mechanical actuators can have linear and rotary motion.

Table 2.1: Different types of actuation.

Types of actuation Linear motion Rotary motion

Electric -Solenoid -Servo motor

-Stepper motor -DC motor

Mechanic -Screw -Gears

-Cam -Wheel and axle

Hydraulic -Hydraulic linear actuator - Pneumatic -Pneumatic linear actuator -

Storage Element -Spring -

Linear mechanical actuators are left out of the realizations of the key features because they are not able to deliver short impulses with the required energy for this thesis. Since the second key feature, retracting the impactor, could also be realized with the form of actuators and a spring, its realizations are the same as the first key feature. For the final key feature, the only possible realization which was thought of was to use any form of electrical motor. Since this was the only idea to realize the third feature, it was left out of the multi-dimensional table. Table 2.1 is the result of the realizations of the first two key functions put together in a table. Since multiple combinations resulted in an illogical combination of actuators, these boxes are left blank.

(7)

2.1.MORPHOLOGICALANALYSISCHAPTER2.ANALYSIS Table 2.2: Idea generation

Retractors

Actuators Electric motor Solenoid Pneumatic actuator Hydraulic actuator Spring

Electric motor Electric motor design - - - Inverse lever design

Solenoid - Solenoid

Pneumatic actuator - Double-acting pneumatic impactor Air rifle design

Hydraulic actuator - Double-acting hydraulic impactor

Spring Lever design BB gun design -

RoboticsandMechatronics6JeffreyStroet

(8)

2.1. MORPHOLOGICAL ANALYSIS CHAPTER 2. ANALYSIS

Though many combinations are left blank in the table, the morphological analysis has generated 8 design ideas to solve the design problem. Down below, these ideas are given together with a short description of the idea behind it and a small sketch.

• Electric motor design - A design where an electric motor accelerates a mass m on a rod with length l. The rod will get a certain angular acceleration ¨θ, which accelerates the mass towards the surface. The electric motor is also used to retract the mass after hitting the surface.

Figure 2.1: Sketch of the electric motor design

• Solenoid - A solenoid will create a magnetic field with a certain strength which accelerates a mass with corresponding velocity. Before impact, the polarity of the solenoid will turn, which will retract the mass.

• Lever design - A design where a mass m is connected to a rod with length l, which turns around a pivot point. An electric motor turns a disk with angular velocity ˙θ, which rotates the beam around its pivot point, tensing spring k. Upon release, the spring accelerates the mass towards the concrete.

Figure 2.2: Sketch of the lever design

• Inverse lever design - This design idea is similar to the lever design however, in this design, the motor drives the impactor to make the impact, and the spring retracts the impactor.

• Double-acting pneumatic/hydraulic impactor - A design where either compressed air or fluids is fed into either position P₁ (to make the impact) or to position P₂(to retrieve the impactor).

(9)

2.1. MORPHOLOGICAL ANALYSIS CHAPTER 2. ANALYSIS

Figure 2.3: Sketch of the double-acting pneumatic/hydraulic impactor design

• Air rifle design - A design where, after a short force F is applied, some pressure will leak from pressure tank P , which will accelerate mass m. Connected to this mass is a small spring, which will retract the mass after impact has been made.

Figure 2.4: Sketch of the air rifle design

• BB gun design - A design where a compression spring k is tensed by compressed air P , which is led into the system. After the spring is compressed, it is released and forces the impactor towards the surface. At the moment of impact, compressed air is led back into the system to retract the mass.

(10)

2.2. PREVIOUS DESIGNS CHAPTER 2. ANALYSIS

Figure 2.5: Sketch of the BB gun design

Out of these 8 design ideas, one idea will be chosen and analyzed in section 2.3. First, previously designed and tested impact systems will be discussed.

2.2 Previous Designs

Some of the design ideas generated by the morphological analysis have already been designed and tested in previous research. In this section of the report, these techniques will be discussed, and afterward, one design idea that has not yet been analyzed will be chosen to analyze for this bachelor thesis.

2.2.1 Rebound hammer

The most used impact system for the assessment of concrete is the rebound hammer. This device is mainly used in civil engineering for the assessment of concrete used in any type of building. It operates by pushing the rebound hammer to a surface which tenses a spring. After the spring is fully tensed, it releases and accelerates a mass towards the surface. The height at which the mass rebounds, is a measure for the stiffness of the concrete. The rebound hammer is similar to the BB gun design idea, which also uses a spring in a linear motion to accelerate a mass towards the surface.

The rebound hammer could have some advantages regarding the assessment of concrete in a sewer pipeline. For example, the rebound hammer is a compact design, has a linear stroke, and, with the right spring, has enough impact energy. However, because of its long impact time due to the large inertia of the hammer, the rebound hammer is unable to assess the concrete in sewer pipelines.

2.2.2 Solenoid

A solenoid has also been tried to serve as an impactor. Adrian Pleijsier tried this setup in his master thesis regarding the acoustic condition assessment of concrete sewer piper using a particle velocity sensor (Pleijsier 2019). A non-magnetic core was used in the solenoid to get a uni-directional impact. Springs were used to retract the impactor. A benefit of this system is that it was able to deliver sufficient impact energy. However, the best impact time achieved was 500 µs, which is around 10 times more than the required impact time.

2.2.3 Double-acting pneumatic linear impactor

Though this design has never been realized, it has been analyzed to determine if this design was suitable for the assessment of concrete. It was clear that the system was able to deliver a high

(11)

2.3. INVERTED PENDULUM - SPRING DESIGN CHAPTER 2. ANALYSIS

impact energy. However, the impact time of this design was limited by the switching time of the fast switching valves. It takes at minimum 20 ms for the impactor to switch direction, which is much slower than the impact time requirement. The same holds for hydraulic switches.

2.2.4 Robotic manipulator

Edwin van Boven used a robotic arm as an impactor during his master thesis ”Localization of structural flaws in concrete sewer piper by physical interaction inspection with a robotic arm”

(Boven 2019). A Franka arm was used to exert forces uniformly in various configurations but could not be achieved due to deflection in the joints. Also, due to high inertia, the impact time was much larger than required.

2.2.5 Spur-gear design

This design is very similar to the electric motor design idea discussed in the morphological analysis.

The spur gear design uses an electric motor combined with two gears to drive an impactor with an angular motion towards the concrete. The impactor consists of a thin rod with a small steel ball at the end and was designed by Jun Hong Choo during his bachelor thesis, Impactor design and development for in-pipe sewer inspection robot (Choo 2019). This design was first to meet its impact time requirement of 20 µs. However, had a very low impact energy (0.05 J ), and was not very compact due to a large gear which was connected to the servo motor and a smaller gear which had 2 extension rod attached for actuating and retracting the impactor.

2.2.6 Conclusion

From all the previous designs, it can be concluded that achieving a low impact time and high impact energy simultaneously is most difficult. This is mainly because large masses are used to deliver a high impact energy. However, a large mass has a large inertia and thus thus a larger impact time. From the 8 design ideas stated in the morphological analysis, most design ideas have already been analysed and/or tested in the past. For this thesis, only three design ideas are left to be analyzed. These are both the lever and the inverse lever designs, and the air rifle design. From these 3 design ideas, the decision was made to proceed to analyze the lever design. The reason for this is because many combinations of springs and electric motors can provide the system with sufficient energy. Moreover, the lever design has many more parameters that could be specified (as illustrated in figure 2.8) compared to the air rifle design. Also, using a spring for impact and a motor for detraction is more logical than reversed. In the following section, this lever design (called Inverted pendulum - spring design) will be analyzed.

2.3 Inverted pendulum - spring design

As a result of the morphological analysis, the inverted pendulum - spring design appeared to be best suited for within the time frame of this thesis. Before any analysis of the inverted - pendulum spring design was done, a Lego prototype of the design was built. The reason for a prototype out of Lego is because Lego is easily adjustable, which opens up the possibility to discover favorable changes to the system early on in the thesis. The spring in the Lego prototype is made out of a rubber band. Figure 2.6 shows the Lego prototype.

(12)

Figure 2.6: The Lego prototype.

When building the Lego prototype, it was discovered that a rear spring could prevent the impactor from making multiple consecutive impacts. This rear spring is located behind the pivot point and also keeps the impactor off the ground in resting position. Making only one consecutive impact was realized with a ratio of 5:1, which means that the front spring was 5 times stronger than the rear spring.

Figure 2.7: Angular acceleration of the Lego prototype, measured with the gyroscope.

After this discovery was made, a gyroscope was mounted on the pivot point of the impactor to determine its acceleration. From the results of the measurement, it became clear that a gyroscope is not able to measure the acceleration nor the impact velocity of the system due to its low sampling rates (as can be seen from the gyroscope output in figure 2.7).

(13)

Figure 2.8: Overview of the system.

To meet the requirement of delivering sufficient impact energy, the system should contain a high amount of kinetic energy upon impact. The kinetic energy of the impactor depends on the mass and velocity of the impactor. The mass cannot be too large since a this will result in a large impact time. Thus, a desired impact velocity should be specified. This will be further discussed in section 3.1).

As the impact velocity is a result of the acceleration from releasing the system until the system makes its impact, an equation for the acceleration should be found. Before this equation was derived, one assumption was made. That is that there is no deformation in the beam. In practice, this will occur. However, for the simulations, this was not taken into account. Assuming no deformation in the rod, the direct relation between torque (τ ), inertia (I) and angular acceleration (¨θ) can be used:

τ = I ¨θ (2.1)

The total torque at the pivot point is the sum of the torque delivered by the gravity acting upon the ball and rod, and the torque delivered by the spring.

τ_total= τ_spring+ τ_mass (2.2)

The torque delivered by the gravity acting on the mass and rod is given by the gravitational force perpendicular to the beam (Fg,⊥) times the distance from the pivot point to the center of mass (l):

τ_g= F_g,⊥l (2.3)

The position of the center of mass is determined from the length of the rod (lrod) the mass of the rod (mrod) and the mass of the ball (mball) and is given by:

l =(l_rod/2)m_rod+ (l_rodm_ball mrod+ mball

) (2.4)

The gravitational force perpendicular to the beam is given by:

F_g,⊥= F_gsin(θ) (2.5)

The torque delivered by the springs is more difficult to find since this force is not always acting straight downwards like the gravitational force. As a result, the angle θ is not always the same as the angle α. To be able to express the system in state-space equations, the system should be expressed in the same variable θ, which means a relation between θ and α should be found. For this, the distances from the pivot to where the springs are connected (d₂, d₃ and d₄) are chosen to be exactly the same such that there are 2 triangles with 2 even sides, and thus 2 even angles.

From this, it can be derived that:

(14)

2α = θ (2.6)

By finding the relation between α and θ, the expression for the torque delivered by the front spring (τsf) as a function of the angle θ and the force delivered by the front spring (Fsf) could be found.

τsf = Fsfd2sin(θ/2) (2.7)

The torque delivered by the rear spring is opposite of that of the front spring and the gravitational force. Also, the argument of the sinusoid decreases with an increase of θ. The torque delivered by the rear spring (τsr) as a function of the angle and the force delivered by the rear spring (Fsr) can be described as follows:

τ_rs= −F_srd₄sin(π − (θ/2)) (2.8)

Substituting the expression for the torque delivered by gravity and by the springs in expression 2.2 and by substitution the expression for the spring force and gravitational force, the total torque can be described as a function of all known variables and the angle:

τ_total= k_fx_fd₂sin(θ/2) + mglsin(θ) − k_rx_rd₄sin(π − (θ/2)) (2.9) Where kf and kr are the spring constants of both springs and xf and xr are the deflections in both springs.

From this relation, an expression for the angular acceleration can be found.

θ =¨ kfxfd2

I sin(θ/2) +mgl

I sin(θ) −krxrd4

I sin(π − (θ/2)) (2.10)

The total inertia of the system is the sum of the inertia from the ball and the inertia of the rod.

The inertia of the ball can be found by applying the parallel axis theorem and is written as:

Iball= 2

5mr²+ ml² (2.11)

where l is the length from the pivot point to the center of mass of the sphere, m_ball is the mass of the ball, and r is the radius of the ball. The inertia of the rod is also found using the parallel axis theorem and can be written as:

Irod= 1

12ml²+ m∆l² (2.12)

Where ∆l is the length from the center of mass to the pivot point of the rod and mrodis the mass of the rod. By substituting the equation for the total inertia in 2.10, the final equation for the angular acceleration can be found. This equation was written in state-space equations to make a block diagram in Simulink. The state-space representation can be found below, together with the block diagram in Simulink (figure 2.9). The Matlab function used in Simulink can be found in the appendix.

y(t) = ˙θ(t) = x2(t) (2.13)

˙

x(t) = ˙x₁(t)

˙ x₂(t)

=

x₂(t)

mgl

I_rod+I_ballsin(x1(t)) +_I^k^f^x^f^d²

rod+I_ballsin(^x¹₂^(t)) −_I^k^r^x^r^d⁴

rod+I_ballsin(^2π−x₂¹^(t))

(2.14)

(15)

Figure 2.9: Block diagram in Simulink.

Further explanation and the results of the model in Simulink are given in section 3.4.

(16)

Chapter 3

Design and Implementation

In this section of the report, the design choices and the implementation of the design will be presented. All the separate parts of the design will be assessed, and each design choice will be supported by simulations and/or theory.

3.1 Translation of requirements to design parameters

To make sure the impact energy requirement is met, the system should contain sufficient kinetic energy before impact. The aim of the impactor is to reach a kinetic energy of at least double the required impact energy of 0.7 J . The mass of the impactor should be as low as possible for the inertia to remain small and is estimated at 15 g. To reach the desired kinetic energy before impact with a mass of 15 g, an impact velocity of 13.6 m/s should be achieved. The desired impact velocity will be used to find the right spring constant for the impactor, which will be discussed in section 3.4.

3.2 Ball

At the end of the impactor, a ball was placed to make the impact. A ball is used instead of any other shape is because using a ball will show a clear half sine wave when measuring the force over time (Sansalone and Carino 1986). From this half sine wave, the impact time can be obtained by fitting a sine curve and retrieving its frequency and period.

The material which was chosen for the ball is steel because steel has an approximately 10 times larger elastic modulus than concrete (Choo 2019). Because of this, most of the deformation that happens upon impact, happens in the concrete surface. A steel ball with a diameter of 13 mm is chosen because this steel ball has an impact time of approximately 40 µs with an impact velocity of 20 m/s, (Choo 2019). This is the lowest impact time that meets the requirement. However, this measurement was performed using a free-falling steel ball with a low amount of inertia. The system used in this thesis will reach a lower impact velocity and will make use of a rod and a rotary motion around a pivot point, which adds more inertia to the system. Both these properties will cause a higher impact time.

3.3 Rod

The rod should be chosen carefully since many parameters of the rod influence the behavior of the system. The rod should be as rigid as possible for the rod not to deform while tensing the front spring and to not resonate after impact. However, the rod should have as little mass as possible to keep the inertia as small as possible. Also, the rod should not be too large since this also influences the inertia of the system.

The decision was made to use a carbon fiber rod for the system due to its low mass and rigidness.

There was a carbon fiber rod with a length of 30 cm, a diameter of 4 mm, and a mass of 5.8 kg

(17)

3.4. SPRINGS CHAPTER 3. DESIGN AND IMPLEMENTATION

available in the lab, which was used for the simulations. With the right spring the desired impact velocity could be reached and thus was this carbon fiber rod used for the system.

3.4 Springs

The desired spring constants are determined by simulating the equations found in section 2.3.

Instead of performing simulations of the system when both the front and rear spring are connected using equation 2.10, the simulations were performed when not taking the torque delivered by the rear spring into account. The reason for this was because the simulations showed that an impact velocity of 15 m/s could be reached, which corresponds to a kinetic energy of 1.7 J . Instead of performing simulations when both springs are connected, this larger impact velocity was considered sufficient and used as a buffer for the lost impact energy due to adding the rear spring. The reduction of impact velocity due to the rear spring will be measured and discussed in section 3.6.2.

Thus, only the front spring has been chosen based on the result of the simulations of the model in Simulink (figure 2.9). The spring constant was tuned until the desired impact velocity of 15 m/s was reached. The simulations show the behavior of the system after being released at an initial angle of 60^◦. This angle was chosen because a larger angle than 60^◦would cause for more deformation of the rod in the realization of the final system. However, the angle cannot be too large since the system must have some distance to accelerate over. The angle was limited to an angle of 113^◦. This was the maximum angle due to the length of the rod and the height of the pivot point. The distance from the pivot point to the position where the spring was mounted was set to 5 cm, which set the length of the spring in rest to 5.7 cm. The mounting distance was chosen arbitrarily but resulted in an impact velocity close to the desired impact velocity. The results of the measurements are shown in figure 3.1.

Figure 3.1: Angular velocity in m/s (left) and angle in radials (right) with a spring constant of 2500 N/m

After trying various different spring constant, a spring constant of 2500 N/m was chosen. As can be seen in the left part of the figure, the simulated impact velocity is 15.2 m/s.

Using the equations between the torque delivered by the front spring (equation 2.7) and the torque delivered by the gravitational force (2.3), a conclusion can be drawn whether or not the system is able to make an impact throughout the whole cross-section of the sewer pipeline. The torque delivered by the spring at the initial angle of 60^◦is 4800 N m. The maximum torque delivered by the gravitational force is 54 N m. From this ratio of 4800:54 is can be concluded that the requirement will be met.

(18)

3.5. REALISATION OF THE SYSTEM CHAPTER 3. DESIGN AND IMPLEMENTATION

3.5 Realisation of the system

To realize the system, several structures should be thought of to connect the whole system together.

That is, there should be a structure which connects the carbon fiber beam to the pivot point, the springs should be connected to the carbon fiber beam, the pivot point should be lifted 5 cm from the ground. and the system should be stable. In this section, these structures are explained.

3.5.1 Pivot point

Two bearings were available in the lab with a diameter of 12 mm. For these bearings, a cylindrical beam was 3D-printed out of carbon to have an as small as possible inertia. Another structure was 3D-printed which had 2 holes, which could fit both the pivot cylinder and the rod of the impactor.

A 3D model can be seen in figure 3.2. To ensure that the rod did not move, a screw was twisted in the top of the pivot structure which lightly pressed against the carbon fiber rod.

Figure 3.2: 3D-model of the pivot structure.

3.5.2 Mounting pieces

Drilling in the carbon fiber rod makes the rod weaker, so a different method should be thought of when connecting the springs to the rod. A structure was designed which could clamp itself around the carbon fiber rod and to which the springs could be connected. The design is created by Sander Smits, a technician of the RaM group, and was adjusted according to the diameter of the carbon fiber rod (4 mm). The mounting piece was laser-cut and made out of POM (Polyoxymethylene) since this is material is very rigid, which is needed to limit the vibration of the system. A picture of the mounting piece can be seen in figure 3.3.

Figure 3.3: Picture of the mounting piece.

By tightening the bolt, both ends of the mounting piece are pulled towards each other creating a deformation in the center, which clamps the structure around the rod. The lower hole was designed to hold the springs. However the metal of the springs could not be bend and thus could not fit through the lower hole. Because of this, keyrings were used to connect the springs to the mounting

(19)

3.5. REALISATION OF THE SYSTEM CHAPTER 3. DESIGN AND IMPLEMENTATION

pieces. The use of keyrings may add additional noise to the system, which will be measured.

3.5.3 Foundation

The foundation of the system is made out of aluminum beams. These beams have a large groove on each side, which allows the use of bolts and screws to attach components or other aluminum beams. Two aluminum beams of 20 cm form the bottom of the system. Two aluminum beams (30 cm) are connected perpendicularly on the bottom beams. To these perpendicular beams, the bearings which hold the pivot point are mounted. The aluminum beams and an overview of the realized system can be seen in figure 3.4. Table 3.1 shows the desired parameter values based on the simulations and calculations.

Figure 3.4: Visualisation of the final system.

Table 3.1: Desired parameter values.

Parameter Value

d1 15 cm

d₂ 5 cm

d₃ 5 cm

d₄ 5 cm

k1 2500 N/m

k2 500 N/m

m 15 g

θ 60^◦

To clarify, d₁ is the distance from the point where the front spring is connected to the mass, d₂, d₃and d₄are the distances from the points where the springs are connected to the pivot point, k_f and k_r are the spring constants of the front and rear springs respectively, m is the mass of the impactor, and θ is the initial angle from which the impactor is released.

(20)

3.6. MEASUREMENTS CHAPTER 3. DESIGN AND IMPLEMENTATION

3.6 Measurements

To verify that the system meets its requirements, three different measurements will be performed.

They are listed down below, together with a short description. To make sure every measurement was done under the same conditions, the end of the rod was released from exactly 4 cm above the ground in each measurement. To make the system more stable, it was mounted on a wooden plate. This wooden plate was much longer than the system, which left room for concrete plates to be used as counterweights.

3.6.1 Impact time

The impact time is measured using the ”Flexiforce A201” force sensor in combination with an amplifier. The amplifier was designed by Jun Hong Choo and was slightly modified by changing the variable resistor with a resistor of 6.8 kΩ to prevent the amplifier from clipping. The circuit is shown in figure 3.5. To not destroy the force sensor, a cylindrical piece of brass was stuck on top of the sensor with a piece of tape, which can be seen in the measurement setup in figure 3.6. This piece of brass should only affect the total force measured by the sensor. However, it should have no influence on the time the impact takes. After the force sensor measurements were performed, the data was transferred to Matlab, where a sine fitting was manually performed. The exact impact time could be determined from half the period of the fitted sine.

Figure 3.5: Amplifier circuit. Figure 3.6: Force sensor setup.

3.6.2 Impact energy

To measure the impact energy of the system, the kinetic energy just before impact is calculated.

This is done by measuring the impact velocity just before impact using an ”Omron EE-SG3 photomicrosensor”. The circuitry and measurement setup can be found in figure 3.7 and 3.8 respectively. A piece of tape will be mounted at the end of the impactor, which could fit through the gap of the photomicrosensor (which can be seen in figure 3.8). From the time it takes to pass the photomicrosensor, and the thickness of the piece of tape (9 mm), the impact velocity can be determined. An additional measurement of the impact velocity without connecting the rear spring will be performed to analyze if adding the rear spring has a significant effect on the impact velocity.

(21)

Figure 3.7: Circuit used to determine the impact velocity.

Figure 3.8: Measurement setup to determine the impact velocity.

To approximate the impact energy, the ratio between potential energy before and after dropping a steel ball on a concrete tile is determined. To measure this ratio between potential energy before and after impact, a steel ball will be dropped from 10 different heights. At every different height, the resulting potential energy will be measured 10 consecutive times to get an accurate approximation of the ratio. A slow-motion camera, in combination with a ruler, will be used to determine the potential energy after impact. An overview of the setup can be found in figure 3.9.

As can be seen from the figure, a rail was used to make sure the ball was dropped from the same height during the 10 consecutive measurements.

Figure 3.9: Measurement setup to determine energy relation before and after impact.

(22)

3.6.3 Noise of the system

The loudness of the noise of the system over time will be measured using an app on a mobile phone. The app is called ”Sound Meter and Sound pressure level meter” and is developed by

”SO LAB”. The app has a sampling rate of 5 samples/s and displays the Sound level (in dB) over time. This is used to determine the loudness of the system and the time frame in which the system makes noise. This time should be as little as possible to clearly assess the condition of the sewer pipeline. To measure the loudness and the time frame in which the system makes noise, the impactor was positioned at the edge of a table, such that after releasing the system, the impactor would not make any impact. This way, all the noise produced by the keyrings and the springs can be measured.

(23)

Chapter 4

Results

In this section of the report, the results of the measurements are discussed. The measurements are performed to determine if the impact energy and impact time requirements are met.

When realizing the system, some parameters were changed compared to the desired parameter values. In table 4.1, an overview of the real parameter values is given. A spring with a constant of 2500 N/m and a length of 5.7 cm was not available in the store and thus the closest possible spring constant (2800 N/m) with a length of 5.7 cm was achieved. Simulations with this spring constant showed that the impact velocity with this spring reached 16 m/s. Also, when realizing the system, a single impact was not acquired using the distances given in table 3.1. A single impact was acquired by positioning the steel ball 1.3 cm above the ground at rest. This was realized by mounting the front spring 6.9 cm from the pivot and the rear spring 6.5 cm from the pivot (both including key rings). The height of the pivot point remained the same. Also, the mass of the system was 13.5 g instead of 15 g. Finally, the initial angle of 60^◦could not be realized due to deformation in the rod. The initial angle that the impactor made was 85^◦if the rear end of the rod was released 4 cm form the ground.

Table 4.1: Final design parameters.

Parameter Value

d1 15 cm

d2 6.9 cm

d3 5 cm

d4 6.5 cm

k1 2800 N/m

k2 500 N/m

m 13.5 g

θ 85^◦

4.1 Impact time measurement

The first performed measurement is the impact time measurement. The impactor was positioned such that the steel ball would hit the brass cylinder precisely in the middle. As described in the section above, the steel ball was positioned 1.3 cm from the ground such that the impactor would make a single impact.

In total, 5 impact time measurements have been performed. Two results from these measurements can be found in figure 4.1 and figure 4.2.

(24)

4.1. IMPACT TIME MEASUREMENT CHAPTER 4. RESULTS

Figure 4.1: First force time measurement. Figure 4.2: Second force time measurement.

Both figures show clear half-sine pulses, but besides the first half-sine pulse, an additional half- sine pulse can be seen. This additional half-sine pulse is not the result of the impactor hitting the surface twice. This was made sure by making a slow-motion recording of every measurement. These pulses could be coming from the piece of tape, which connects the brass cylinder to the force sensor. Without this piece of tape, the brass cylinder bounces up after impact. The piece of tape keeps the brass cylinder in place but could cause a vibration on top of the force sensor.

After performing the measurements, a sine fitting was manually performed to determine the impact time of the system. The results can be seen in figure 4.3 and figure 4.4

Figure 4.3: Sine fitting on first force time measurement.

Figure 4.4: Sine fitting on second force time measurement.

The impact time is directly derived from half the sines period. The results of the five measurements are shown below in table 4.2 together with their mean and standard deviation.

Table 4.2: Impact time results.

Measurement Impact time (µs)

1 28.6

2 28.6

3 31.7

4 28.3

5 30.5

Mean 29.5

Standard deviation 1.52

An impact time of 40 to 80 µs was required for the impactor. As can be seen from table 4.2, the results from the impact time measurements are 29.5 µs on average. The impact time from a steel ball with a diameter of 13mm and an impact velocity of 11 m/s would have had a larger impact

(25)

4.2. IMPACT ENERGY MEASUREMENT CHAPTER 4. RESULTS

time than 40 µs. Also, the rotary motion around a pivot together with the axis introduces extra inertia, which would increase the impact time even more. The only reason why the impact time could be significantly shorter than expected is due to the rear spring pulling the impactor back when the impact happens.

4.2 Impact energy measurement

The first measurement that was performed to approximate the impact energy was to see if there is any linear relation between potential energy before and after contact with a concrete surface. The results can be seen in figure 4.5.

Figure 4.5: The results from dropping a steel ball from 10 different heights.

The red line is the result of fitting a first-degree polynomial onto the measured data using Matlab.

The red line shows a linear relationship between the potential energy before and after impact with the concrete surface. According to Matlab, the polynomial has a coefficient of 0.302, which means that for every Joule of potential energy before impact, approximately 30 % of energy comes out, and thus 70% of the energy is absorbed by the impact. Using this linear relationship, the system should contain approximately 1 J of kinetic energy to meet the requirement of having an impact energy of at least 0.7 J .

After determining this linear relation between energy before and after impact, the energy before impact of the system is measured. This is done by measuring the velocity before impact and calculating the associated kinetic energy of the impactor. The output of the photomicrosensor can be seen in figure 4.6.

(26)

4.2. IMPACT ENERGY MEASUREMENT CHAPTER 4. RESULTS

Figure 4.6: Output of the photomicrosensor.

The output of the photomicrosensor shows two clear square waves from which the time can be directly determined. Only the first square wave is used to determine the impact velocity of the system. From the time it takes for the piece of tape to pass through the sensor and from the thickness of the piece of tape (9 mm), the impact velocity can be calculated. Five measurements were done from which the velocities before impact are measured. From these velocities, the kinetic energy is calculated and shown in table 4.3. Afterwards, the linear relation between potential energy before and after impact with a concrete surface is used to approximate the impact energy.

These results are shown in table 4.4.

Table 4.3: Velocity and associated kinetic energy before impact.

Measurement Velocity before Kinetic energy impact(m/s) (J)

1 9.0 0.55

2 10.0 0.68

3 9.2 0.57

4 9.1 0.56

5 9.1 0.56

Mean 9.3 0.58

Standard deviation 0.4 0.05

Table 4.4: Approximated impact energies.

Measurement Approximated impact energy (J)

1 0.38

2 0.47

3 0.40

4 0.39

5 0.39

Mean 0.41

Standard deviation 0.037

The results show that the approximated impact energy is half the required impact energy. This is due to the impact velocity, which is lower than the desired impact velocity. This can be explained

(27)

4.3. NOISE OF THE SYSTEM CHAPTER 4. RESULTS

by two reasons. The first reason is that the initial angle of 60^◦cannot be reached due to the deformation of the rod. Another reason is due to the rear spring, which decelerates the system before impact. This second reason was further analyzed by removing the rear spring and measuring the impact velocity with just the front spring. The results can be seen in table 4.5.

Table 4.5: Impact velocity and approximated impact energy of the system without the rear spring.

Measurement Impact velocity (m/s) Impact energy J

1 11.6 0.64

2 11.9 0.67

3 9.7 0.45

4 9.6 0.44

5 9.8 0.45

Mean 10.5 0.53

The results show that the rear spring reduces the impact velocity with approximately 1 m/s and the impact energy with 0.1 J . This is a large change, so for future research towards this impactor design, this large reduction of impact energy should be accounted for.

4.3 Noise of the system

A small measurement was performed determining the loudness and the time frame over which the system makes noise, as explained in section 3.6.3. The measurement is performed 10 times to get a good approximation. Since the app takes a sample after every 0.2 s, the average measurement error of the app is 0.1 s. An example of the result is shown in figure 4.7. The results of the 10 measurements are shown in table 4.6.

Figure 4.7: The result of measuring the loudness (db) of the system over time (s)

(28)

4.4. SUMMARY CHAPTER 4. RESULTS

Table 4.6: Measured loudness and time frame over which the system makes noise.

Measurement Time frame (s) Average loudness (dB)

1 0.8 47.9

2 0.6 47.8

3 0.6 47.8

4 0.6 47.2

5 0.6 47.0

6 0.8 47.3

7 0.6 47.5

8 0.6 47.8

9 0.6 48.2

10 0.6 48.0

Mean 0.6 47.7

As can be seen from the results, the system makes noise over an average time period of 0.6 s (±1 due to the measurement error). The average loudness of the system is 47.7 dB. A time frame of 0.6 s is quite large and will likely distort the measurement of the condition of the concrete. The time frame is so large because springs together with key rings are used. Both the loose keyrings and the springs vibrate after making an impact, which need to dampen over time.

4.4 Summary

To summarise, 4 different measurements have been performed to assess if the system meets its requirements. To determine the impact time of the system, a force sensor was used to show that the impact time of the system is 29.5 µs, which is significantly shorter than the desired impact time.

Afterwards a measurement was performed to determine a linear relationship between the potential energy of a steel ball before and after contact with a concrete surface. From this linear relationship it became clear that the energy before impact of the system should be 1 J at minimum. Using the linear relationship, and measuring the energy the system has before impact, an approximation of the impact energy was made (0.41 J ) which is lower than the required impact energy.

Afterwards, the impact velocity without the rear spring was measured to determine the reduction of the impact energy by the rear spring. The rear spring only reduced the impact velocity with a factor 1/11. However, this results in a significant reduction in impact energy (0.1 J ).

Finally, the loudness and time frame of noise of the system were measured. The system makes noise with an average loudness of 47.7 dB over a time period of 0.6 s (±1). This is a large time frame that will likely distort the measurements of the condition of the concrete.

(29)

Chapter 5

Conclusion

In this thesis, an impactor design for an in-pipe impact-echo inspection method was designed according to the following requirements:

• The system must have an impact time of 40 to 80 µs. With this impact time, a frequency range from 12,5 kHz (80 µs) up to 25 kHz (40 µs) can be released into the concrete. This is higher than the thickness frequency of concrete in the sewer pipelines, which is needed to assess its condition (Pleijsier 2019).

• The energy upon impact must be between 0.7 - 2.2 J to be able to measure the signal over the noise and not to destroy the concrete pipeline (Kovler, Wang, and Muravin 2018).

• To measure a clear signal, the impactor should be retracted directly after the first impact and before a second impact is made.

• The system itself should be as silent as possible in order to measure the condition of the sewer pipeline clearly.

• The system could be to be able to make an impact throughout the whole cross-section of the sewer pipeline. If this requirement is satisfied, the condition of the whole cross-section of the sewer pipeline can be assessed.

A morphological analysis was performed to obtain new design ideas in a structured way (section 2.1). Out of the 8 generated design ideas, the inverted pendulum - spring design was further analyzed, designed, and tested. A Lego prototype was built to observe any favorable outcomes of the design early on in the thesis. From this Lego prototype it was observed that a 5 times stronger front spring compared to the rear spring could achieve the requirement of making a single impact.

This relationship was used to determine the desired value of the rear spring.

Afterwards, an expression for the angular acceleration as a function of the front and rear springs and the gravitational force was found (section 2.3. From this equation, it was clear that the acceleration caused by the front spring is approximately 100 times larger than the acceleration caused by the gravity. This means that the requirement of making an impact throughout the whole cross- section of the pipeline can be met using this design. When simulating the system, only the angular acceleration caused by the front spring and the gravitational force was simulated, and the loss in impact velocity due to the rear spring was accounted for by desiring a high impact velocity.

From these simulations, the desired impact velocity of 15 m/s was achieved using a spring with a constant of 2500 N/m. A steel ball with a diameter of 13 mm was used to reach the required impact time.

After the system was built, its impact time and impact energy and noise over time were measured.

The impact time was directly determined from a force-time half-sine pulse and was 29.5 µs on average. The impact energy was approximated by finding a linear relationship between the potential energy before and after impact with the concrete surface. It turned out 70% of the energy before impact is transitioned into the material. This linear relation was used to find an average approximated impact energy of 0.41 J . The average approximated impact energy without the use of the rear spring was 0.53 J , which means that the rear spring reduces approximately one-fifth of

(30)

CHAPTER 5. CONCLUSION

the impact energy, like the ratio of the front and rear spring. Down below, an overview is given showing the desired and realized system properties. Finally, the system itself makes some noise in a time frame of 0.6 s (±1), which likely distorts the measurements of the condition of the concrete.

Table 5.1: Desired and realised system properties.

Property Desired values Realised value

Impact time 40 - 80 µs 29.5 µs

Impact energy 0.7 - 2.2 J 0.41 J

Single impact Yes Yes

Silent system Yes No

Impact throughout the cross-section Yes Yes of the sewer pipeline

(31)

Chapter 6

Recommendations

The first recommendation for future research on this impactor design is to start by using the linear relationship between energy before and after impact with a concrete surface. From this, a desired kinetic energy before impact can be derived, and from this, the desired spring constant for the front spring.

The second recommendation is to increase the mass of the system and mainly the thickness of the rod. Using a thicker rod reduces the deformation in the beam, which enables the use of a smaller initial angle. This smaller initial angle means the system has a larger distance to accelerate over.

The increase in the inertia of the rod is acceptable since this will increase the impact time which is too low in the current design which can be seen from table 5.1.

The final recommendation is to design different mounting pieces, which could be connected to the spring instead of using a key ring. The key rings add a lot of noise to the system.

(32)

Bibliography

Boven, E.J van (2019). Localization of structural flaws in concrete sewer pipes by physical interaction inspection with a robotic arm.

Carino, N.J. (2001). The impact echo method: An overview. url: https://tsapps.nist.gov/

publication/get_pdf.cfm?pub_id=860355.

Choo, J.H (2019). Impactor design and development for in-pipe sewer inspection robot.

Kovler, K, F. Wang, and B. Muravin (2018). Testing of concrete by rebound method: Leeb versus Schmidt hammers.

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

Sansalone, M. and N.J. Carino (1986). Impact Echo: A method for Flaw Detection in Concrete Using Transient Stress Waves. url: https://nvlpubs.nist.gov/nistpubs/Legacy/IR/

nbsir86-3452.pdf.

(33)

Appendix A

Appendix

Down below, the code used in the simulations in Simulink is given.

1 f u n c t i o n y = f c n ( u )

2

3 k = 2 5 0 0 ; % S p r i n g c o n s t a n t

4 d2 = 5 e −2; % D i s t a n c e from p i v o t

t o where t h e s p r i n g i s a t t a c h e d

5 x i n i t = 5 . 7 e −2; % I n i t i a l l e n g t h o f t h e

s p r i n g

6 x = 2∗ d2 ∗s i n( (p i−u ) / 2 ) − x i n i t ; % D e f l e c t i o n i n t h e s p r i n g

7 mrod = 5 . 8 e −3; % Mass rod

8 m b a l l = 7 . 7 e −3; % Mass beam

9 m = mrod + m b a l l ; % T o t a l mass o f t h e

syst em

10 r b a l l = 0 . 0 0 6 5 ; % Radius o f t h e b a l l

11 g = 9 . 8 1 ; % G r a v i t a t i o n a l

c o n s t a n t

12 l = 20 e −2; % Length o f t h e beam

13 lcom = ( ( ( l / 2 ) ∗mrod ) +( l ∗ m b a l l ) ) / ( mrod+m b a l l ) ; % Length form t h e p i v o t t o t h e c e n t r e o f mass

14 I r o d = ( 1 / 3 ) ∗mrod ∗ ( l ˆ 2 ) ; % I n e r t i a rod

15 I b a l l = ( 2 / 5 ) ∗ m b a l l ∗ ( r b a l l ˆ 2 ) + m b a l l ∗ ( l ˆ 2 ) ; % I n e r t i a b a l l w i t h p a r a l l e l a x i s theorem

16 I = I r o d + I b a l l ; % I n e r t i a o f t h e mass +

beam

17

18 y = ( ( k∗ d2 ∗x ) / I ) ∗s i n( u / 2 ) +((m∗ g ∗ lcom ) / I ) ∗s i n( u ) ;

Impactor Design for In-pipe Impact-echo Inspection Method: A Morphological Analysis