Final Report Summary - VORTEXSCAN (Vortex Robot for Rapid Low Cost Scanning and Improved Non-Destructive Testing of Large Concrete Structures)
Executive Summary:
Large public infrastructure facilities around the world like dams, cooling towers and bridges use cement as the main building material. Although made from a durable material many of these large assets have begun to age and are in need of periodic inspection to ensure their integrity. Current inspection routine involves setting up scaffolds that have to be moved around for personnel to reach to the whole of the surface.
The VORTEXSCAN project has developed a novel NDT automated system incorporating advanced Ground Penetrating Radar (GPR) and low frequency Ultrasonic Testing (UT) techniques and a robotic platform able to operate at heights. It will enhance inspection of in-situ structures so that degradation due to fatigue or natural incidents can be determined on time, before a breakdown or catastrophic failure occurs.
It is expected that the end users will benefit substantially, as inspections will be made faster without imposing trained personnel in hazardous conditions, with repeatability and lower costs, thus extending the lifetime of critical structures.
The VORTEXSCAN project has achieved the following results:
• An optimally designed, through simulations and lab testing, impeller able to generate adhesion to carry a payload of more than 2kg while overcoming external disturbances
• A compact and lightweight vortex robot with payload capability of up to 2kg, able to operate vertically or horizontally at heights on curved concrete surfaces and controlled remotely
• A visual tracking aid system for the robot’s navigation by the operator
• A GPR antenna, through simulations and lab testing, with increased sensitivity and penetration depth capability of ~40cm on concrete structures
• A portable GPR system integrated on a compact and ruggedized chassis with a full weight of 1kg
• A pitch-catch dry couplant low frequency UT system able to penetrate at depths of 50cm on concrete structures with an increased accuracy and defect detection capability of as small as 2 times more than the average size of concrete grains
• A custom pulser-receiver unit able to optimally drive the UT couple with the required requirements and a weight of less than 1kg
• Post processing algorithms that enabled increased repeatability effectiveness
• Laboratory trials were performed on various concrete samples demonstrating the capability of the system
The project has achieved the main objectives with a final trial carried out on a big concrete sample showing that the full VORTEXSCAN system is operating satisfactory. The NDT systems can be both applied on other applications and materials under the necessary fine tuning, such as central sewage piping, railway concrete sleepers, asphalt motorways. The robotic platform has a potential to deploy other NDT techniques for concrete inspection and could be used as well for maintenance of other structures as glass windows in tall buildings.
Project Context and Objectives:
The Need
NDT of large concrete structures is currently carried out utilising access techniques such as abseiling and scaffolding which is time consuming and is very costly to carry out. For example, it takes a week to erect scaffolding to access a 7m high double storey building.
With inspection required to be carried out at least once a year by law*(1), utilising the current NDT techniques would necessitate that, operations shut down to facilitate inspection preventing use of the facility. For Nuclear Power Plants (NPPs), the average loss of generating capacity is 15 days per year leading to a €165m revenue loss across Europe. This has contributed to the high cost of inspecting structures – a cost which includes the loss of production time together with the actual inspection costs. These factors present an opportunity for an innovative technology to rapidly and effectively inspect concrete structures thereby reducing the cost of inspection and down time caused by shutting down operations.
While concrete is strong in compression, it has very low tensile strength. The latter is improved by adding steel reinforcement bars (rebars) to the concrete. The inspection of concrete structures is required to test for defects such as potential corrosion of the rebars, cracks and voids in the concrete as well as delamination of the different concrete layers. These flaws weaken the concrete structure causing various hazards to individuals that utilise such facilities.
The inspection process poses a risk to the safety of the workers. Scaffolding and rope access techniques currently in use attract very high insurance charges and, moreover, EC directives on health and safety at work*(2) strengthen the need for an automatic inspection technology.
The vortex robots that have been developed cannot be used in this application as they have a very limited carrying capacity, typically less than 200gm. There is therefore a need for larger robots which are capable of carrying the NDT equipment as they navigate the vertical walls. Furthermore, the current state-of-the-art in UT and GPR equipment is bulky and impossible to carry using a lightweight robot, so making these portable is very important.
The SME partners in the VORTEXSCAN project have identified the potential commercial use of innovative vortex robot technology for use in safety inspections for concrete structures in power plants such as nuclear reactor containments and cooling towers.
The Solution
The VORTEXSCAN project sought to develop an automated vortex robot which is capable of carrying UT and GPR test equipment to inspect concrete structures and the accompanying software architecture. A small vortex robot measuring with a carrying capacity of up to 2kg was developed to easily navigate vertical curved walls. Vortex robots adhere to a surface by creating negative pressure created by rotating an impeller at high speeds*(3). The advantage of creating negative pressure by generating a vortex inside a chamber is that an increasing pressure gradient develops outwards from the centre of the chamber to its periphery which will be at atmospheric pressure. This produces a force that attaches the robot to any surface enabling it to move vertically*(4).
The robot and the GPR/UT system were powered by a pack of Lithium-ion polymer (Lipo) batteries with robot control and data acquisition performed wirelessly. In order to conserve power and ensure batteries last the expected 1 hour, magnetic adhesion technology which requires no energy would be used to provide most of the adhesion force for the climbing robot. Using this technology a permanent magnetic flux is focussed and connected to concrete rebars – attaching the robot to rebars magnetically. The robot was tele-operated through a radio controller from the ground. The tele-operator used real-time information from the GPR system to direct the robot along vertical rebars in order to get optimal magnetic adhesion.
We also developed novel portable sensors which combine GPR and UT to non-destructively inspect concrete walls for size of small cracks, air voids, concrete delamination as well as locating rebars. The sensors, with a combined weight of around 2kg and providing sufficiently high spatial resolution and depth penetration, were mechanically attached to the robot. The GPR system of a frequency between 1600 – 2400 MHz was designed to provide a high resolution with a depth range of 40–50cm. This GPR system together with low frequency UT provided measurement to small defects size of 2 times more than the average size of concrete rains in concrete. During inspection, the robot stops every 30cm to lower the probes to be in contact with the concrete and gather data before lifting them again. Innovative software architecture provided the interface for communication between the NDT apparatus and the vortex robot. The robot also incorporated a communications hub for the control systems and data transfer from the NDT system to the inspector’s computer via wireless radio link.
Data from the two NDT techniques were displayed together to diminish the limitations of each and create an accurate representation of the underlying material and its defects. The data were saved for offline visualisation so that an inspector can view A-scans and be in a position to decide where to repeat inspections.
The display of UT and GPR data would give accurate and timely technical information on structural health. This allows lifespan legislation on concrete structures to be verified and ensure appropriate decision making on maintenance programmes, structural integrity and lifespan.
The inclusion of robotics in the field of concrete structure inspection through the VORTEXSCAN project enabled faster scanning. The robotic system is expected to eliminate the need of manual inspection with associated access equipment. In many instances loss of plant generation capacity will be averted as the inspection can occur when the plant is operating, resulting in financial savings. VORTEXSCAN will significantly reduce the cost of inspecting structures.
Robotic inspections on tall concrete structures will support safer working conditions by reducing the time required for people to access and stay on the walls under high risk working conditions. Once the automatic system has finished the inspection, the skilled personnel would only need to work on specific points where a defect has been identified. The additional benefit of this automated system is the reduction in both labour and insurance costs.
Competitive advantage
The VORTEXSCAN project has developed an automated robotic system equipped with two NDT techniques (UT and GPR). Currently there is no alternative technology which combines inspection data from two techniques. Our system rapidly inspects as well as provide a great amount of technical data, a prerequisite for accurate assessment of the structural health of concrete walls.
The development of a vortex robot which maximises payload while being operated wirelessly requires the creation of new knowledge. The SME partners own all the IP generated in the project. Exploitation of this project’s results will place the SMEs in an ideal position to increase their market share and exploit this and adjacent markets.
In undertaking this project we aimed to achieve the following objectives:
Scientific Objectives
• To determine the optimal impeller size and speed to generate adhesion to carry a payload of up to 2kg while overcoming air resistance and impeller vibration
• To develop novel sensor-centric data fusion models appropriate to fusion of disparate range image data
Technical Objectives
• To develop a post process algorithm that will enable increased repeatability effectiveness
• To develop a vortex robot with payload capability of up to 2kg that is operated wirelessly
• Image fusion and processing to categorize and quantify defects detected by the approach used
• To develop portable GPR and UT sensors with a combined weight of less than 2kg
• To design and develop a dry couplant low frequency UT sensor to provide sufficient measurement accuracy to size defects as small as 2 times more than the average size of concrete grains
• To design and develop GPR sensor with increased antenna sensitivity and penetration depth range of 40-50cm
----------------------------------------------------------------------------------------------------------------
*(1) International Committee on Industrial Chimneys Model Code - ATLAS Guide to the inspection and maintenance of reinforced concrete chimneys and natural draught cooling towers
*(2) Official Journal of the European Communities, “European Parliament and Council Directive 89/391/EEC
*(3) http://www.wired.com/gadgetlab/2007/06/vortex-robot-do/
*(4) http://www.vxsim.com/documents/Vortex-SDK-Brochure.pdf
Project Results:
The main scientific and technological results arising from the project are as follows:
System specification and sample preparation was carried out, which included sample procurement, a user requirement survey and production of the functional specifications of the VORTEXSCAN system. Firstly, information concerning the end-user requirements (working conditions for concrete structures inspections, types of defects found inside, etc.) was collected and analysed.
• The core end user requirements are to provide an automated system that will be able:
o To provide in service NDT without the need to stop the facility’s production
o To detect defects (air voids and cracks) at potentially critical areas with a rough given specific minimum resolution at a depth of 0.5m
o To be transported easily by not specialized personnel as well
o To provide wide area coverage (autonomy) without the need for continuous setups
o To provide consistent results over time and be in general highly reliable
o To provide rapid and accurate inspection
o To provide lower costs of inspection
• For proof of concept needs the project focused on detecting defects at a maximum depth of ~0.5m at structures with a dense mesh (high strength structures) of reinforcing rebars.
• Regarding defects, we focused on concrete delamination, air voids and cracks. An extensive literature review was performed in order to identify technical details about them, the reasons behind their creation
• The minimum size to detect was set to 2 times more than the average size of aggregate present in the mixture
The inspection procedure included two technologies, low frequency ultrasonic testing (UT) and high frequency ground penetration radars (GPR). For both of them an extensive literature review was performed in order to identify state of the art systems, technological advances, similar applications approach, potential weaknesses / limitations and upgrade opportunities and ideas.
The developed GPR system complied with the following core specifications and requirements:
• Overall weight of 1kg
• High frequency system in the range of 1600-2400MHz
• Increased sensitivity and resolution
• Penetration depth up to 0.5m in concrete
• Optimal design of the GPR antenna after simulations under diverse but specific scenarios
• System optimization after testing on custom made samples
• Full system integration on-board the robotic vehicle
• NDT data dispatch wirelessly to the operator’s control PC
• Easy presentation and interpretation of data
With regard to the UT system, the following specifications and requirements were set:
• Overall weight of the full prototype of max. 1kg
• Operating frequency in the range 40-500KHz
• Defects detection sizing up to 2 times the average size of aggregates
• Increased sensitivity and resolution
• Penetration depth up to 0.5m in concrete
• Optimal design of the UT parameters technique after simulations under diverse but specific scenarios
• System optimization after testing on custom made samples
• Full system integration onboard the robotic vehicle
• NDT data dispatch wirelessly to the operator’s control PC
• Easy presentation and interpretation of data
With regard to the robotic vortex vehicle that will carry all the NDT and other assistive equipment the following core specifications were set and followed:
• Maximum payload ~2kg, including all NDT components and subsystems
• Adhesion technology: Vacuum via an onboard vacuum generating system
• Operation on curved surfaces with big radii
• Provision of electromechanical interfaces for the NDT integration
• Wirelessly controlled
• Autonomous navigation features
• Power supply via an umbilical cable
• Onboard controller
• Easy to use and control user interface
Moreover, with regard to samples provision a few of them have been developed for the validation of the GPR and UT NDT systems. They should incorporate rebars and special features, as tendon holes, drilled holes, etc.
Finally, an implementation plan was produced, including a detailed technical risk analysis, describing the management structures and procedures for the implementation of the project, as well as the risk management to be taken into account and the potential mitigation measures suggested.
More specifically a great effort was put to solidly address core risks around two the NDT systems’ performance.
GPR SYSTEM DESIGN AND DEVELOPMENT
The GPR system development included three main phases: design, development and performance validation.
> Design
The overall objective was to develop a portable, advanced and high operating frequency GPR prototype subsystem having an overall weight of 1 kg, that will be able to provide an increased sensitivity and resolution in terms of defects detectability at a penetration depth range up to 50 cm in concrete, enabling also easy interpretation of obtained data.
The developed GPR sensor would be able to fit in the dimensions of the vortex robot based on the system specifications defined.
The development of GPR antennas can be undertaken in two different ways. One way was to use empirical methods and process to a series of trials to achieve the desired goals. The other way was to use simulations tools to predict the performance of the antenna under investigation. Obviously, the second way was the preferred one offering key advantages as it could provide some insight to the designer regarding the electromagnetic field behaviour, allowing at the same time the investigation of many antenna configurations without the need of any hardware.
In this research work, numerical electromagnetic models for the propagation of radar waves within concrete were used to predict the fundamental properties of the radar waveform and the potential bases for creating images from these waveforms. The detailed objective was to theoretically assess the performance of multiple GPR antenna frequency approaches using electromagnetic wave simulation tools for the propagation of radar waves within concrete, aiming to predict the required antenna frequency and characteristics that would be the most effective in detecting concrete defects of interest found in realistic structures. The definition of the parameters of the optimum features led to the selection of the GPR sensor that was used in the VORTEXSCAN system.
In order to construct a GPR model in two dimensions, using commercially available software, the following assumptions were made:
• All media were considered to be linear and isotropic.
• The transmitting antenna was modelled as a line source. The use of a line source was a consequence of the assumption of the invariance of the problem in one direction.
• The constitutive parameters were, in most cases, assumed not to vary with frequency.
The GPR numerical simulation work was based on the Finite-Difference Time-Domain (FDTD ) method and had very useful features for antenna modeling. A series of numerical parameter studies were carried out in order to assess the potential performance of high frequency GPR approaches for detecting features and deterioration in concrete structures. The simulation tool was used to represent different defect types (i.e. voids, delaminations) around reinforcing steel in concrete and to model multiple antenna frequency approaches aiming to determine the required antenna characteristics that were most effective in revealing the damage and concrete properties of interest.
Modelling materials and deterioration
The deterioration of concrete often changes the electromagnetic properties. These changes may occur in the fine structure of the material, such as the formation of additional interfaces and dielectric discontinuities created by delaminations, voids, etc, or they may occur in the bulk properties of the material, such as an increase in the dielectric constant (decreased velocity) or an increased in the attenuation due to higher moisture content, chloride content, and porosity.
One key factor in using waveform or imaging models is the determination of the electromagnetic representation of the materials being modeled. Concrete defects such as cracking and delaminations, chloride contamination, voids, etc and other ordinary concrete properties were represented by material dielectric properties and by geometric discontinuities. Different analytical models have been developed and used by other researchers aiming to evaluate the dielectric properties of concrete (i.e. variations in permittivity and conductivity) with respect to frequency, moisture content or other agents. These physical properties derived from published experimental results were used to develop input into the GPR numerical modeling carried out under this research work, aiming to assess changes in the GPR signal response as they are related to concrete material property changes.
Different types of defects with varying sizes and thicknesses, and placed at different locations and depths in concrete were simulated and numerically tested. The investigated concrete deterioration included voids and delaminations, where the latter are usually represented by small air-filled or water-filled cracks that usually occur close to the rebars. The dielectric properties assigned to the above layers depended upon the type of delamination and void that was being simulated
In addition to the above deterioration layers, rebars with varying diameters placed at different positions and depths in concrete were modelled as well. Since the reinforcing bar reflections are usually superimposed on the reflections from other irregular subsurface anomalies (i.e. delaminations), it is very important to incorporate the effect of rebars in any analytic model. This is because the smaller reflections from the delaminations are usually masked by larger rebar reflections.
Simulations
A series of numerical modeling scenarios were carried out in order to predict the required antenna frequency and characteristics that were most effective in detecting internal features such as reinforcing and concrete defects (i.e. air-filled and water-filled delaminations) of interest found in realistic structures.
The numerical simulation results obtained from the GPR modeling showed that a monostatic geometry antenna could provide accurate inspection of subsurface structural elements and concrete deterioration, offering enhanced detectability of very small defects (on the order of 1mm in thickness) and closely-spaced targets with superior resolution and with sufficient penetration for concrete to be adequately resolved in depths up to 40 cm. GPR antenna characteristics (i.e. transmitter-receiver separation distance, transmitter/receiver height above the ground level) close to commercially available and widely used high-frequency GPR systems from major manufacturers seem to provide sufficient resolution of the reflected signal produced from the internal features.
> Development and preliminary testing
After the implementation of the numerical simulation studies, the next stage was the delivery of a suitable GPR sensor satisfying the technical objectives of the VORTEXSCAN project. Based on the modelling results and literature sources, the appropriate geometry, characteristics and material properties of the fundamental components were determined and integrated into a portable, advanced GPR prototype subsystem having an overall weight of 1.3 Kg. The GPR subsystem operated in the vicinity of the ground using a high-frequency shielded dipole antenna of monostatic geometry. The antenna consisted of planar dipoles which were enclosed in rectangular shields in order to increase the required resolution at a penetration depth range up to 40 cm.
The antenna consisted of planar dipoles arranged onto Printed Circuit Board (PCB) substrate and enclosed in rectangular metal boxes which act as shielding. The shields were designed to eliminate above ground interference or external signals and prevent electromagnetic emissions from the antenna affecting surrounding electronic equipment, as well as to improve the antenna directivity. Above the antenna enclosure and on the internal upper side of the GPR housing box, the radar electronic PCB board was accommodated, also connected to the serial memory. Furthermore, in order to control the radar data collection and enable the acquisition of encoded results in the scan direction, an encoder connected to a survey wheel was used to provide distance measurement along the inspection area and initiate a triggering pulse for the radar signal at predetermined distances. Data was acquired at user defined distance intervals so that the position of each trace along a survey line was given by the position of a radar trace in the GPR data file. This simplified GPR data processing procedures and positioning of identified targets.
> Data processing
After the system development, tests were run on custom made samples. Before that the necessary data processing system was setup.
It is well known that the mixing and convolution of effects that go into the formation of GPR return signals often make them difficult to interpret without some level of signal processing to put them in a format that is amenable to human examination. Thus, the main objective was to develop a data analysis processing system that would allow the condition, display and easy interpretation of GPR signals received from the antenna in the robot. The data processing software system had been developed to carry out data condition, displaying and image processing on the recorded B-scan GPR signals. The important problem for the B-scan GPR image was to transform or migrate the unfocused B-scan image to a focused image showing the object’s correct location and dimension. This image processing technique was called migration or focusing.
> Testing and results
Finally, the evaluation of the performance of the developed GPR technology was carried out under laboratory conditions, where three concrete samples of varying dimensions and with different embedded structural features of known characteristics were tested. The validation results produced from this study indicated the high potential and efficiency of the developed GPR subsystem to accurately detect internal concrete features with superior resolution and with sufficient penetration for concrete to be adequately resolved in depths up to 40 cm.
UT SYSTEM
The UT system has been first of all designed through simulations under selected scenarios and concrete models. Then it was tested in the lab on custom made samples. The methodology selected was the pulse echo technique.
> Design through modeling
This technique was first proposed in 1940, however it has seen little application for the inspection of concrete, although much present research is being undertaken to rectify this. Emerging concrete pulse echo techniques are concerned with the evaluation of concrete from a single surface, with the aim to: detect air filled voids and cracks; locate structural elements such as reinforcement and ducts; and provide information on geometrical dimensions. The technique requires a short time length pulse (broad in frequency domain) to be transmitted into the structure by an electro-acoustic transducer and received on the same surface by the same (pulse-echo) or an additional transducer (pitch-catch).
To achieve the broadband transducer characteristics necessary for pulse echo, a suitable material is backed onto the piezoelectric material to damp oscillations. The unwanted effect of this is to reduce the amplitude of the transmitted signal and the sensitivity of the receiver, hence a reduction in received signal to noise. Attempts over the years have been made to develop pulse-echo techniques for the inspection of concrete structures, all of which suffered from impractical cumbersome transducers. In literature it has been reported that using separate transducers and advanced signal processing techniques is the key to improve effective pulse-echo tests on concrete.
Analysing the commercial sensors and systems available for concrete inspection, it was observed that the central frequency of the sensor was determined by the thickness of the sample under inspection. A compromise between accuracy and attenuation was finally found due to the fact that the higher the frequency the higher the accuracy, but also the higher the attenuation. Modelling had been carried out in order to confirm the pre-selected central frequency. Different aggregate sizes had been considered, with two different percentages of aggregates present in the mixture: 20% and 80%.
> Hardware design-development and selection
With regards to the Transducers Module, UT shear sensors had been adopted based on the modelling analysis results. Shear sensors do not need couplant to be used which makes operation more practical and convenient.
A Pulser-Receiver module had been specifically design for the VORTEXSCAN application. The module is composed of pulser, receiver, and processing control unit. The Pulser-Receiver module can generate short pulses with very high amplitude which are converted to mechanical waves when applied to the transducer. The frequency of the pulses can be varied from few kHz to 300 kHz. A GUI system control software was developed to control the pulser-receiver module, perform experiment parameter setting and data acquisition.
The shear sensors are robust to ‘structural noise’ and do not require couplant to be deployed, which is convenient for automatic inspection, taking into account that the inspection will not be is carried out manually but using an autonomous robot.
The pulser receiver of the system generates short pulses which are converted to elastic waves when applied to the transducer. The frequency of the pulses can be varied from few kHz up to 300 kHz. It includes 3 different amplification stages and a power supply.
The receiver amplifies the signal received by the transducer to adapt it to a correct range for the ADC. As the ultrasonic signal reflected from the defects are of very low amplitude, the voltage generated by the transducers are very low, usually μV or few mV. Therefore, an amplification stage between the transducer and the ADC is required. The factors to be considered are frequency bandwidth, slew rate, cost, VCC, gain, filters, noise ratio and size of the final PCB.
This signal must be suitable for processing to obtain the needed information about the condition of the concrete structure. The preamplifier must be as low-noise as possible and must not modify the shape of the waveform and include band-pass to reduce the noise outside the frequency range under analysis.
> Test results
The prototype system was tested for the inspection of the concrete sample with known internal structural features, such as reinforcement bars, plastic ducts and post- tensioning tendon. Two experimental setups had been tested: transmission mode and reflection mode. The propagation wave velocity was identified via transmission mode and used for identify the depth of the inner objects via reflection mode. A series of A-scan signals were acquired along profile line in order to construct 2D B-scan image representing the cross-sectional view. From B-scan images, the reflections at and inside the plastic duct, including the front edge, the central steel bar and the rear edge, were clearly identified. Even at low frequencies the UT sensor can successfully and clearly detect the post-tensioning tendon, which is a 12 mm diameter strand. It can be concluded that the developed UT subsystem is able to detect the interior objects (plastic, metal) and interior defects (crack, void, etc.) of the size as small as 12 mm or even smaller.
In the VORTEXSCAN integrated system, GPR and UT subsystems will be both installed on the Vortex robot. GPR employs electromagnetic rather than acoustic waves, and offers fast measurement without contact to the surface of the concrete. However, it’s difficult to use GPR to detect the interior of tendon ducts, i.e. post-tensioning tendon. As a complementary inspection tool to the GPR subsystem, UT subsystem is employed for detailed examination of the post-tensioning tendons.
ROBOTIC VEHICLE
The robotic vehicle design was divided in three core sections:
• Vortex system based on a custom designed impeller
• Robotic vehicle navigation system
• Robot’s controller
Regarding the vortex system, it was mounted on onboard the robotic vehicle so that the vehicle was lightweight, easy to navigate and with extended autonomy. The solution of having a vacuum generator near the operator, with a heavy, cumbersome and stiff vacuum tubing was considered not suitable for this application.
The heart of the vortex system is the impeller along with its chamber. The optimal design was based on an extensive literature review, results from simulations as well as feedback through realistic tests.
> State of the art
Axial impellers, as their name implies, move the airstream long the axis of rotation. The air is pressurized by the aerodynamic lift generated by the fan blades. Sometimes these fans can be interchanged with radial fans, but they are commonly used in low-pressure and high volume applications. Some advantages of this group of impellers are:
- Compactness
- Light weight
- Low cost
However, this type presents a wide variety of disadvantages for the goal pursued, such as:
- Problems in situations where the air flow must vary considerably
- Anti-stall devices or impeller over-sizing to avoid the previous problem
- Higher rotation speed to achieve the same airflow capacity as centrifugal impellers
- Need for clean air
In terms of radial or centrifugal impellers the classification is a bit wider. Nevertheless, a general definition says that these devices move air first radially outwards by centrifugal action, and then tangentially away from the blade tips. As the air moves from the impeller hub to the blade tips, it gains kinetic energy which is then converted to a static pressure increase.
Within radial impellers there is another classification regarding the geometrical arrangement of the blades, namely, radial, forward inclined and backward inclined.
• Radial. The blades extend straight out from the center and its variation “radial-tip” the blades are completely straight in the outer diameter getting inclined against the rotation direction and not reaching the inlet. This family is commonly used for high speeds, low volumes and high pressures. Paying attention to their characteristics curves, we understand that these impellers are not stable and the motor can be overloaded. The power curve is always growing despite the fan curve starts to decrease after the optimum.
• Forward inclined. The blades are curved in the direction of rotation. In general, these impellers have more blades than the previous and the next group, what means weight increase. Such impellers have a low efficiency and poor control of the airflow. Furthermore, they are mostly used for high airflow and low pressure at low speed applications.
• Backward inclined. The main feature of this group is that the blades are inclined towards the opposite direction of rotation. These impellers are:
o lighter than the forward inclined ones because in this case a smaller number of blades is required,
o optimum for low flow and medium-high speed applications, and
o energetically stable. Non-overloading motor characteristic.
> Assumptions:
• According to the requirements, vacuum must be created to adhere the robot against the wall. Because of that, flow conditions can vary widely depending on the amount of suction desired by means of mainly rotation speed.
• Small dimensions of blade either in length or in height mean that the rotation speed must be high.
• The robot must be as flat as possible to avoid moments on wheels and therefore to make the system more stable.
• Many different types of equipment need to be installed on board which implies that the larger space available on the platform the better.
> Choices:
• It can be said that centrifugal impellers have a more uniform and smoother working zone, avoiding stall at any speed. Also they can work at high speeds of rotation and their design is more convenient for low airflow.
• In terms of vacuum chamber and frame, axial impellers need to expel air perpendicularly to its ration plane. It means that immediately after the impeller an axial channel is required to drive the air out. In case of radial impellers, the air is expelled radially and whether the flow is driven in that direction or through a vertical duct, more mechanical solutions can be considered.
For all these reasons, it was decided to use a radial backward inclined impeller. Both, curved and airfoils were taken into account in this study
Its final tuning was made by means of theoretical calculations, simulations and in lab tests. This design flow was the pilot to define and determinate the optimal working point and to examine its deviation from the original project specifications.
> Simulation and results
Several 3D models were designed including the blades, hub, air inlet and outlet. The main factors under consideration and design were:
• Number of blades
• Width of blade (constant or varying)
• Height of blade
• Angles of blades related to the relative velocity at the inlet and outlet.
• Inlet diameter
Based on the project specifications, it is immediate to get the pressure required. Since the maximum weight of the robot is 2kg and the payload must be of up to 2kg, the suction generated must be enough to lift up an overall mass of 4kg. With regard to rotational speed preliminary calculations revealed that the speed should be between 15,000 and 20,000rpm to get the suction levels required.
The process to determine the inlet diameter was done in parallel along with the speed one, reaching the conclusion that for working within the speeds showed before, the most suitable diameters would be between 30 and 40mm.
Having chosen a backward inclined impeller limits the number of these elements. Typically, the design of that group of impeller goes from a few blades, around 4, to hardly 14. With regard to the blade’s height two restricting factors were set:
• The robot should be as flat as possible what means that if a too high impeller is included it can contribute to make the robot falling down from the wall.
• For the same diameter, higher blades mean larger surfaces what entails more air moved and hence, more aerodynamic drag. The direct result of the previous reasoning is obvious: more power. This situation would increase the weight of the motor and hence, the weight of the overall structure.
Simulations were not done only with constant blades but with variable profiles of height as well. Finally, with respect to the blade’s width and profile it was decided to model thin ones. Radial impellers are usually designed as a decreasing profile being thicker in the inlet region and thinner in the outlet. The reason is because according to aerodynamic theory, this type of geometry favors the airflow around a certain profile, therefore the more efficient is the airflow the better is the profile and hence, the suction force.
Several different shapes of blades were designed, including airfoils, and simulations undertaken for all of them using commercially available software. The simulations resulted in 4 valid designs that were manufactured and tested. The chamber’s design included the inlet and outlet diameter and height. Several configurations were made and simulated extensively and a full prototype was manufactured for extended testing in the lab. The experimental jig included a high speed AC motor selected off the shelf, a typical plastic tapered chamber and 4 3D printed impellers. The impeller selected could offer around ~8 Kg at 2000RPMs which was considered highly promising providing as well a good safety factor for the whole VORTEXSCAN system. The final used impellers were machined from hard engineering plastics and the driving motor used is a high speed outrunner DC motor.
> Robotic vehicle design
The vehicle (platform) design was very simple and straightforward. It consists of 4 wheels, all driven for high toque capacity. The wheels are made of elastic and durable material for high coefficients of friction and endurance.
The main chassis is composed by two carbon fibre plates which host the vacuum chamber, impeller, motor and electronic components. These plates are fixed and assembled by means of the motors brackets and four steel spacers. The motor brackets have been specifically designed and manufactured in white nylon to fix the motors to the lower plate.
The last structural components are the wheels and the vacuum ring. The wheels were manufactured in nylon and they have a rubber layer to ensure good traction and avoid skidding, See previous picture.
The vacuum sealing system is shown in the image below. It is composed of a plastic interface and a rubber or foam ring which is in contact with the surface. Both tyres and ring can be changed and customized depending on the type of surface where the robot has to move along.
The driving motors are DC ones with an integrated gearbox. Moreover, they include an encoder for speed control and odometry. Its control was done remotely via an RF transmitter receiver. In addition, the main controller was based on raspberry and controllers and a power supply sized and selected off the shelf.
After the system’s full integration, it was tested extensively in lab conditions, on dry concrete walls. The results were highly promising, presenting reliability in order to incorporate all NDT equipment onboard for final tests.
> Tracking system
Depending on the inspecting surface/structure an inspection map may be needed. This is the main reason why a tracking system was required. The implemented solution was a visual tracking system performed by a simple webcam and a custom software developed. The algorithm is able to identify the mobile target and provide its relative position to a fixed reference spot, which can be the initial point or any other relevant spot of the structure. Once the fixed position has been defined, the software can track the robot position in three different ways:
• Following the object with no tracking.
• Following the object with a temporary tracking record that can be erased after a defined after a period defined by the user.
• Following the object with a constant tracking, so by the end of the inspection it is possible to check what areas have been swept.
A mimic of the software with real images regarding the tracking method was developed for testing and validation purposes. The evolution of the software is noticeable, which goes from simple forms (magnet pieces) on a white board to the real robot image.
> Integration testing and conclusions (see the attached document "VORTEXSCAN_Final Report_Figures")
All components were installed onboard the robotic vehicle achieving a payload of 2 Kg. Powering of the full system was done via an umbilical and data dispatch towards to the master PC for further processing and examination.
The GPR could perform continuous inspections with no stops except for those where the operator wanted to change parameters, whereas the UT inspections needed a spot by spot procedure expending around 40 seconds to acquire the result of every measurement.
Thanks to the GPR controlling software and, especially, the encoder, the inspection could be interrupted at any moment for the UT measuring and then resume it, since the GPR keeps records of the last position when the movement was paused. The output data were acquired and analysed afterwards to verify the status of the concrete reinforcing bars.
Potential Impact:
The VORTEXSCAN system is destined to improve the industrial competiveness across the EU. Currently, the most widely used procedures for inspecting concrete structures use scaffolding and rope access techniques, such methods are time consuming and labour intensive. The VORTEXSCAN project does not require scaffolding and will enable rapid inspection of the complete concrete structure resulting in time and cost savings. An average Nuclear Power Plant (NPP) shuts down approximately 15 days per annum for inspection and maintenance resulting in loss of revenue amounting to €165m for all NPPs in Europe. This money could be saved with our automatic inspection system as inspection can be performed without disrupting the operation of the plant. It will also give reliable assessment of concrete characteristics thereby improving the cost-effectiveness of maintenance programmes for concrete structures by locating structures in need of priority repair and, equally importantly, structures that do not need repairing. Hence, our technology promotes methods of prolonging the working life of existing concrete structures through better inspection.
This increase in operational life will lead to a reduction in the capital costs for the replacement of assets. Specifically, 30% of commercial NPPs in Europe have more than 30 years of operation and 60% operate between 20 and 30 years of service life. The design life of existing NPPs is set to be 30 – 40 years, so without the advantage that VORTEXSCAN technology can bring, a significant number of these will shutdown. Currently, nuclear energy provides 30% of the electricity supply in Europe. With a number of NPPs reaching the end of their lifespan, VORTEXSCAN technology will potentially provide an extension to their operational lives and enable a phasing of shutdowns to eliminate the predicted power shortages. This will reduce economic loss due to the sustenance of electricity supply which is crucial to the manufacturing and service delivery sectors.
Use of the VORTEXSCAN product reduces the time required for inspectors to access and stay at the walls of large structures under high risk working conditions. Once the automatic system has finished the inspection, the skilled personnel will only be required to work on specific points where a defect has been identified. The dramatic reduction of the inherent risks due to the automatic machine inspection conforms to the requirements of the current Prevention of Risks at work Act34 (European Directive 89/391/CEE).
The use of this technology also reduces the risk of premature failure of concrete structures. Failure of structures result in financial losses, environmental damage, and emission of toxic radiation in NNPs. Prevention of such disasters can be greatly aided by better timely and thorough inspections through the VORTEXSCAN technology.
List of Websites:
http://www.vortexscan.eu/
Large public infrastructure facilities around the world like dams, cooling towers and bridges use cement as the main building material. Although made from a durable material many of these large assets have begun to age and are in need of periodic inspection to ensure their integrity. Current inspection routine involves setting up scaffolds that have to be moved around for personnel to reach to the whole of the surface.
The VORTEXSCAN project has developed a novel NDT automated system incorporating advanced Ground Penetrating Radar (GPR) and low frequency Ultrasonic Testing (UT) techniques and a robotic platform able to operate at heights. It will enhance inspection of in-situ structures so that degradation due to fatigue or natural incidents can be determined on time, before a breakdown or catastrophic failure occurs.
It is expected that the end users will benefit substantially, as inspections will be made faster without imposing trained personnel in hazardous conditions, with repeatability and lower costs, thus extending the lifetime of critical structures.
The VORTEXSCAN project has achieved the following results:
• An optimally designed, through simulations and lab testing, impeller able to generate adhesion to carry a payload of more than 2kg while overcoming external disturbances
• A compact and lightweight vortex robot with payload capability of up to 2kg, able to operate vertically or horizontally at heights on curved concrete surfaces and controlled remotely
• A visual tracking aid system for the robot’s navigation by the operator
• A GPR antenna, through simulations and lab testing, with increased sensitivity and penetration depth capability of ~40cm on concrete structures
• A portable GPR system integrated on a compact and ruggedized chassis with a full weight of 1kg
• A pitch-catch dry couplant low frequency UT system able to penetrate at depths of 50cm on concrete structures with an increased accuracy and defect detection capability of as small as 2 times more than the average size of concrete grains
• A custom pulser-receiver unit able to optimally drive the UT couple with the required requirements and a weight of less than 1kg
• Post processing algorithms that enabled increased repeatability effectiveness
• Laboratory trials were performed on various concrete samples demonstrating the capability of the system
The project has achieved the main objectives with a final trial carried out on a big concrete sample showing that the full VORTEXSCAN system is operating satisfactory. The NDT systems can be both applied on other applications and materials under the necessary fine tuning, such as central sewage piping, railway concrete sleepers, asphalt motorways. The robotic platform has a potential to deploy other NDT techniques for concrete inspection and could be used as well for maintenance of other structures as glass windows in tall buildings.
Project Context and Objectives:
The Need
NDT of large concrete structures is currently carried out utilising access techniques such as abseiling and scaffolding which is time consuming and is very costly to carry out. For example, it takes a week to erect scaffolding to access a 7m high double storey building.
With inspection required to be carried out at least once a year by law*(1), utilising the current NDT techniques would necessitate that, operations shut down to facilitate inspection preventing use of the facility. For Nuclear Power Plants (NPPs), the average loss of generating capacity is 15 days per year leading to a €165m revenue loss across Europe. This has contributed to the high cost of inspecting structures – a cost which includes the loss of production time together with the actual inspection costs. These factors present an opportunity for an innovative technology to rapidly and effectively inspect concrete structures thereby reducing the cost of inspection and down time caused by shutting down operations.
While concrete is strong in compression, it has very low tensile strength. The latter is improved by adding steel reinforcement bars (rebars) to the concrete. The inspection of concrete structures is required to test for defects such as potential corrosion of the rebars, cracks and voids in the concrete as well as delamination of the different concrete layers. These flaws weaken the concrete structure causing various hazards to individuals that utilise such facilities.
The inspection process poses a risk to the safety of the workers. Scaffolding and rope access techniques currently in use attract very high insurance charges and, moreover, EC directives on health and safety at work*(2) strengthen the need for an automatic inspection technology.
The vortex robots that have been developed cannot be used in this application as they have a very limited carrying capacity, typically less than 200gm. There is therefore a need for larger robots which are capable of carrying the NDT equipment as they navigate the vertical walls. Furthermore, the current state-of-the-art in UT and GPR equipment is bulky and impossible to carry using a lightweight robot, so making these portable is very important.
The SME partners in the VORTEXSCAN project have identified the potential commercial use of innovative vortex robot technology for use in safety inspections for concrete structures in power plants such as nuclear reactor containments and cooling towers.
The Solution
The VORTEXSCAN project sought to develop an automated vortex robot which is capable of carrying UT and GPR test equipment to inspect concrete structures and the accompanying software architecture. A small vortex robot measuring with a carrying capacity of up to 2kg was developed to easily navigate vertical curved walls. Vortex robots adhere to a surface by creating negative pressure created by rotating an impeller at high speeds*(3). The advantage of creating negative pressure by generating a vortex inside a chamber is that an increasing pressure gradient develops outwards from the centre of the chamber to its periphery which will be at atmospheric pressure. This produces a force that attaches the robot to any surface enabling it to move vertically*(4).
The robot and the GPR/UT system were powered by a pack of Lithium-ion polymer (Lipo) batteries with robot control and data acquisition performed wirelessly. In order to conserve power and ensure batteries last the expected 1 hour, magnetic adhesion technology which requires no energy would be used to provide most of the adhesion force for the climbing robot. Using this technology a permanent magnetic flux is focussed and connected to concrete rebars – attaching the robot to rebars magnetically. The robot was tele-operated through a radio controller from the ground. The tele-operator used real-time information from the GPR system to direct the robot along vertical rebars in order to get optimal magnetic adhesion.
We also developed novel portable sensors which combine GPR and UT to non-destructively inspect concrete walls for size of small cracks, air voids, concrete delamination as well as locating rebars. The sensors, with a combined weight of around 2kg and providing sufficiently high spatial resolution and depth penetration, were mechanically attached to the robot. The GPR system of a frequency between 1600 – 2400 MHz was designed to provide a high resolution with a depth range of 40–50cm. This GPR system together with low frequency UT provided measurement to small defects size of 2 times more than the average size of concrete rains in concrete. During inspection, the robot stops every 30cm to lower the probes to be in contact with the concrete and gather data before lifting them again. Innovative software architecture provided the interface for communication between the NDT apparatus and the vortex robot. The robot also incorporated a communications hub for the control systems and data transfer from the NDT system to the inspector’s computer via wireless radio link.
Data from the two NDT techniques were displayed together to diminish the limitations of each and create an accurate representation of the underlying material and its defects. The data were saved for offline visualisation so that an inspector can view A-scans and be in a position to decide where to repeat inspections.
The display of UT and GPR data would give accurate and timely technical information on structural health. This allows lifespan legislation on concrete structures to be verified and ensure appropriate decision making on maintenance programmes, structural integrity and lifespan.
The inclusion of robotics in the field of concrete structure inspection through the VORTEXSCAN project enabled faster scanning. The robotic system is expected to eliminate the need of manual inspection with associated access equipment. In many instances loss of plant generation capacity will be averted as the inspection can occur when the plant is operating, resulting in financial savings. VORTEXSCAN will significantly reduce the cost of inspecting structures.
Robotic inspections on tall concrete structures will support safer working conditions by reducing the time required for people to access and stay on the walls under high risk working conditions. Once the automatic system has finished the inspection, the skilled personnel would only need to work on specific points where a defect has been identified. The additional benefit of this automated system is the reduction in both labour and insurance costs.
Competitive advantage
The VORTEXSCAN project has developed an automated robotic system equipped with two NDT techniques (UT and GPR). Currently there is no alternative technology which combines inspection data from two techniques. Our system rapidly inspects as well as provide a great amount of technical data, a prerequisite for accurate assessment of the structural health of concrete walls.
The development of a vortex robot which maximises payload while being operated wirelessly requires the creation of new knowledge. The SME partners own all the IP generated in the project. Exploitation of this project’s results will place the SMEs in an ideal position to increase their market share and exploit this and adjacent markets.
In undertaking this project we aimed to achieve the following objectives:
Scientific Objectives
• To determine the optimal impeller size and speed to generate adhesion to carry a payload of up to 2kg while overcoming air resistance and impeller vibration
• To develop novel sensor-centric data fusion models appropriate to fusion of disparate range image data
Technical Objectives
• To develop a post process algorithm that will enable increased repeatability effectiveness
• To develop a vortex robot with payload capability of up to 2kg that is operated wirelessly
• Image fusion and processing to categorize and quantify defects detected by the approach used
• To develop portable GPR and UT sensors with a combined weight of less than 2kg
• To design and develop a dry couplant low frequency UT sensor to provide sufficient measurement accuracy to size defects as small as 2 times more than the average size of concrete grains
• To design and develop GPR sensor with increased antenna sensitivity and penetration depth range of 40-50cm
----------------------------------------------------------------------------------------------------------------
*(1) International Committee on Industrial Chimneys Model Code - ATLAS Guide to the inspection and maintenance of reinforced concrete chimneys and natural draught cooling towers
*(2) Official Journal of the European Communities, “European Parliament and Council Directive 89/391/EEC
*(3) http://www.wired.com/gadgetlab/2007/06/vortex-robot-do/
*(4) http://www.vxsim.com/documents/Vortex-SDK-Brochure.pdf
Project Results:
The main scientific and technological results arising from the project are as follows:
System specification and sample preparation was carried out, which included sample procurement, a user requirement survey and production of the functional specifications of the VORTEXSCAN system. Firstly, information concerning the end-user requirements (working conditions for concrete structures inspections, types of defects found inside, etc.) was collected and analysed.
• The core end user requirements are to provide an automated system that will be able:
o To provide in service NDT without the need to stop the facility’s production
o To detect defects (air voids and cracks) at potentially critical areas with a rough given specific minimum resolution at a depth of 0.5m
o To be transported easily by not specialized personnel as well
o To provide wide area coverage (autonomy) without the need for continuous setups
o To provide consistent results over time and be in general highly reliable
o To provide rapid and accurate inspection
o To provide lower costs of inspection
• For proof of concept needs the project focused on detecting defects at a maximum depth of ~0.5m at structures with a dense mesh (high strength structures) of reinforcing rebars.
• Regarding defects, we focused on concrete delamination, air voids and cracks. An extensive literature review was performed in order to identify technical details about them, the reasons behind their creation
• The minimum size to detect was set to 2 times more than the average size of aggregate present in the mixture
The inspection procedure included two technologies, low frequency ultrasonic testing (UT) and high frequency ground penetration radars (GPR). For both of them an extensive literature review was performed in order to identify state of the art systems, technological advances, similar applications approach, potential weaknesses / limitations and upgrade opportunities and ideas.
The developed GPR system complied with the following core specifications and requirements:
• Overall weight of 1kg
• High frequency system in the range of 1600-2400MHz
• Increased sensitivity and resolution
• Penetration depth up to 0.5m in concrete
• Optimal design of the GPR antenna after simulations under diverse but specific scenarios
• System optimization after testing on custom made samples
• Full system integration on-board the robotic vehicle
• NDT data dispatch wirelessly to the operator’s control PC
• Easy presentation and interpretation of data
With regard to the UT system, the following specifications and requirements were set:
• Overall weight of the full prototype of max. 1kg
• Operating frequency in the range 40-500KHz
• Defects detection sizing up to 2 times the average size of aggregates
• Increased sensitivity and resolution
• Penetration depth up to 0.5m in concrete
• Optimal design of the UT parameters technique after simulations under diverse but specific scenarios
• System optimization after testing on custom made samples
• Full system integration onboard the robotic vehicle
• NDT data dispatch wirelessly to the operator’s control PC
• Easy presentation and interpretation of data
With regard to the robotic vortex vehicle that will carry all the NDT and other assistive equipment the following core specifications were set and followed:
• Maximum payload ~2kg, including all NDT components and subsystems
• Adhesion technology: Vacuum via an onboard vacuum generating system
• Operation on curved surfaces with big radii
• Provision of electromechanical interfaces for the NDT integration
• Wirelessly controlled
• Autonomous navigation features
• Power supply via an umbilical cable
• Onboard controller
• Easy to use and control user interface
Moreover, with regard to samples provision a few of them have been developed for the validation of the GPR and UT NDT systems. They should incorporate rebars and special features, as tendon holes, drilled holes, etc.
Finally, an implementation plan was produced, including a detailed technical risk analysis, describing the management structures and procedures for the implementation of the project, as well as the risk management to be taken into account and the potential mitigation measures suggested.
More specifically a great effort was put to solidly address core risks around two the NDT systems’ performance.
GPR SYSTEM DESIGN AND DEVELOPMENT
The GPR system development included three main phases: design, development and performance validation.
> Design
The overall objective was to develop a portable, advanced and high operating frequency GPR prototype subsystem having an overall weight of 1 kg, that will be able to provide an increased sensitivity and resolution in terms of defects detectability at a penetration depth range up to 50 cm in concrete, enabling also easy interpretation of obtained data.
The developed GPR sensor would be able to fit in the dimensions of the vortex robot based on the system specifications defined.
The development of GPR antennas can be undertaken in two different ways. One way was to use empirical methods and process to a series of trials to achieve the desired goals. The other way was to use simulations tools to predict the performance of the antenna under investigation. Obviously, the second way was the preferred one offering key advantages as it could provide some insight to the designer regarding the electromagnetic field behaviour, allowing at the same time the investigation of many antenna configurations without the need of any hardware.
In this research work, numerical electromagnetic models for the propagation of radar waves within concrete were used to predict the fundamental properties of the radar waveform and the potential bases for creating images from these waveforms. The detailed objective was to theoretically assess the performance of multiple GPR antenna frequency approaches using electromagnetic wave simulation tools for the propagation of radar waves within concrete, aiming to predict the required antenna frequency and characteristics that would be the most effective in detecting concrete defects of interest found in realistic structures. The definition of the parameters of the optimum features led to the selection of the GPR sensor that was used in the VORTEXSCAN system.
In order to construct a GPR model in two dimensions, using commercially available software, the following assumptions were made:
• All media were considered to be linear and isotropic.
• The transmitting antenna was modelled as a line source. The use of a line source was a consequence of the assumption of the invariance of the problem in one direction.
• The constitutive parameters were, in most cases, assumed not to vary with frequency.
The GPR numerical simulation work was based on the Finite-Difference Time-Domain (FDTD ) method and had very useful features for antenna modeling. A series of numerical parameter studies were carried out in order to assess the potential performance of high frequency GPR approaches for detecting features and deterioration in concrete structures. The simulation tool was used to represent different defect types (i.e. voids, delaminations) around reinforcing steel in concrete and to model multiple antenna frequency approaches aiming to determine the required antenna characteristics that were most effective in revealing the damage and concrete properties of interest.
Modelling materials and deterioration
The deterioration of concrete often changes the electromagnetic properties. These changes may occur in the fine structure of the material, such as the formation of additional interfaces and dielectric discontinuities created by delaminations, voids, etc, or they may occur in the bulk properties of the material, such as an increase in the dielectric constant (decreased velocity) or an increased in the attenuation due to higher moisture content, chloride content, and porosity.
One key factor in using waveform or imaging models is the determination of the electromagnetic representation of the materials being modeled. Concrete defects such as cracking and delaminations, chloride contamination, voids, etc and other ordinary concrete properties were represented by material dielectric properties and by geometric discontinuities. Different analytical models have been developed and used by other researchers aiming to evaluate the dielectric properties of concrete (i.e. variations in permittivity and conductivity) with respect to frequency, moisture content or other agents. These physical properties derived from published experimental results were used to develop input into the GPR numerical modeling carried out under this research work, aiming to assess changes in the GPR signal response as they are related to concrete material property changes.
Different types of defects with varying sizes and thicknesses, and placed at different locations and depths in concrete were simulated and numerically tested. The investigated concrete deterioration included voids and delaminations, where the latter are usually represented by small air-filled or water-filled cracks that usually occur close to the rebars. The dielectric properties assigned to the above layers depended upon the type of delamination and void that was being simulated
In addition to the above deterioration layers, rebars with varying diameters placed at different positions and depths in concrete were modelled as well. Since the reinforcing bar reflections are usually superimposed on the reflections from other irregular subsurface anomalies (i.e. delaminations), it is very important to incorporate the effect of rebars in any analytic model. This is because the smaller reflections from the delaminations are usually masked by larger rebar reflections.
Simulations
A series of numerical modeling scenarios were carried out in order to predict the required antenna frequency and characteristics that were most effective in detecting internal features such as reinforcing and concrete defects (i.e. air-filled and water-filled delaminations) of interest found in realistic structures.
The numerical simulation results obtained from the GPR modeling showed that a monostatic geometry antenna could provide accurate inspection of subsurface structural elements and concrete deterioration, offering enhanced detectability of very small defects (on the order of 1mm in thickness) and closely-spaced targets with superior resolution and with sufficient penetration for concrete to be adequately resolved in depths up to 40 cm. GPR antenna characteristics (i.e. transmitter-receiver separation distance, transmitter/receiver height above the ground level) close to commercially available and widely used high-frequency GPR systems from major manufacturers seem to provide sufficient resolution of the reflected signal produced from the internal features.
> Development and preliminary testing
After the implementation of the numerical simulation studies, the next stage was the delivery of a suitable GPR sensor satisfying the technical objectives of the VORTEXSCAN project. Based on the modelling results and literature sources, the appropriate geometry, characteristics and material properties of the fundamental components were determined and integrated into a portable, advanced GPR prototype subsystem having an overall weight of 1.3 Kg. The GPR subsystem operated in the vicinity of the ground using a high-frequency shielded dipole antenna of monostatic geometry. The antenna consisted of planar dipoles which were enclosed in rectangular shields in order to increase the required resolution at a penetration depth range up to 40 cm.
The antenna consisted of planar dipoles arranged onto Printed Circuit Board (PCB) substrate and enclosed in rectangular metal boxes which act as shielding. The shields were designed to eliminate above ground interference or external signals and prevent electromagnetic emissions from the antenna affecting surrounding electronic equipment, as well as to improve the antenna directivity. Above the antenna enclosure and on the internal upper side of the GPR housing box, the radar electronic PCB board was accommodated, also connected to the serial memory. Furthermore, in order to control the radar data collection and enable the acquisition of encoded results in the scan direction, an encoder connected to a survey wheel was used to provide distance measurement along the inspection area and initiate a triggering pulse for the radar signal at predetermined distances. Data was acquired at user defined distance intervals so that the position of each trace along a survey line was given by the position of a radar trace in the GPR data file. This simplified GPR data processing procedures and positioning of identified targets.
> Data processing
After the system development, tests were run on custom made samples. Before that the necessary data processing system was setup.
It is well known that the mixing and convolution of effects that go into the formation of GPR return signals often make them difficult to interpret without some level of signal processing to put them in a format that is amenable to human examination. Thus, the main objective was to develop a data analysis processing system that would allow the condition, display and easy interpretation of GPR signals received from the antenna in the robot. The data processing software system had been developed to carry out data condition, displaying and image processing on the recorded B-scan GPR signals. The important problem for the B-scan GPR image was to transform or migrate the unfocused B-scan image to a focused image showing the object’s correct location and dimension. This image processing technique was called migration or focusing.
> Testing and results
Finally, the evaluation of the performance of the developed GPR technology was carried out under laboratory conditions, where three concrete samples of varying dimensions and with different embedded structural features of known characteristics were tested. The validation results produced from this study indicated the high potential and efficiency of the developed GPR subsystem to accurately detect internal concrete features with superior resolution and with sufficient penetration for concrete to be adequately resolved in depths up to 40 cm.
UT SYSTEM
The UT system has been first of all designed through simulations under selected scenarios and concrete models. Then it was tested in the lab on custom made samples. The methodology selected was the pulse echo technique.
> Design through modeling
This technique was first proposed in 1940, however it has seen little application for the inspection of concrete, although much present research is being undertaken to rectify this. Emerging concrete pulse echo techniques are concerned with the evaluation of concrete from a single surface, with the aim to: detect air filled voids and cracks; locate structural elements such as reinforcement and ducts; and provide information on geometrical dimensions. The technique requires a short time length pulse (broad in frequency domain) to be transmitted into the structure by an electro-acoustic transducer and received on the same surface by the same (pulse-echo) or an additional transducer (pitch-catch).
To achieve the broadband transducer characteristics necessary for pulse echo, a suitable material is backed onto the piezoelectric material to damp oscillations. The unwanted effect of this is to reduce the amplitude of the transmitted signal and the sensitivity of the receiver, hence a reduction in received signal to noise. Attempts over the years have been made to develop pulse-echo techniques for the inspection of concrete structures, all of which suffered from impractical cumbersome transducers. In literature it has been reported that using separate transducers and advanced signal processing techniques is the key to improve effective pulse-echo tests on concrete.
Analysing the commercial sensors and systems available for concrete inspection, it was observed that the central frequency of the sensor was determined by the thickness of the sample under inspection. A compromise between accuracy and attenuation was finally found due to the fact that the higher the frequency the higher the accuracy, but also the higher the attenuation. Modelling had been carried out in order to confirm the pre-selected central frequency. Different aggregate sizes had been considered, with two different percentages of aggregates present in the mixture: 20% and 80%.
> Hardware design-development and selection
With regards to the Transducers Module, UT shear sensors had been adopted based on the modelling analysis results. Shear sensors do not need couplant to be used which makes operation more practical and convenient.
A Pulser-Receiver module had been specifically design for the VORTEXSCAN application. The module is composed of pulser, receiver, and processing control unit. The Pulser-Receiver module can generate short pulses with very high amplitude which are converted to mechanical waves when applied to the transducer. The frequency of the pulses can be varied from few kHz to 300 kHz. A GUI system control software was developed to control the pulser-receiver module, perform experiment parameter setting and data acquisition.
The shear sensors are robust to ‘structural noise’ and do not require couplant to be deployed, which is convenient for automatic inspection, taking into account that the inspection will not be is carried out manually but using an autonomous robot.
The pulser receiver of the system generates short pulses which are converted to elastic waves when applied to the transducer. The frequency of the pulses can be varied from few kHz up to 300 kHz. It includes 3 different amplification stages and a power supply.
The receiver amplifies the signal received by the transducer to adapt it to a correct range for the ADC. As the ultrasonic signal reflected from the defects are of very low amplitude, the voltage generated by the transducers are very low, usually μV or few mV. Therefore, an amplification stage between the transducer and the ADC is required. The factors to be considered are frequency bandwidth, slew rate, cost, VCC, gain, filters, noise ratio and size of the final PCB.
This signal must be suitable for processing to obtain the needed information about the condition of the concrete structure. The preamplifier must be as low-noise as possible and must not modify the shape of the waveform and include band-pass to reduce the noise outside the frequency range under analysis.
> Test results
The prototype system was tested for the inspection of the concrete sample with known internal structural features, such as reinforcement bars, plastic ducts and post- tensioning tendon. Two experimental setups had been tested: transmission mode and reflection mode. The propagation wave velocity was identified via transmission mode and used for identify the depth of the inner objects via reflection mode. A series of A-scan signals were acquired along profile line in order to construct 2D B-scan image representing the cross-sectional view. From B-scan images, the reflections at and inside the plastic duct, including the front edge, the central steel bar and the rear edge, were clearly identified. Even at low frequencies the UT sensor can successfully and clearly detect the post-tensioning tendon, which is a 12 mm diameter strand. It can be concluded that the developed UT subsystem is able to detect the interior objects (plastic, metal) and interior defects (crack, void, etc.) of the size as small as 12 mm or even smaller.
In the VORTEXSCAN integrated system, GPR and UT subsystems will be both installed on the Vortex robot. GPR employs electromagnetic rather than acoustic waves, and offers fast measurement without contact to the surface of the concrete. However, it’s difficult to use GPR to detect the interior of tendon ducts, i.e. post-tensioning tendon. As a complementary inspection tool to the GPR subsystem, UT subsystem is employed for detailed examination of the post-tensioning tendons.
ROBOTIC VEHICLE
The robotic vehicle design was divided in three core sections:
• Vortex system based on a custom designed impeller
• Robotic vehicle navigation system
• Robot’s controller
Regarding the vortex system, it was mounted on onboard the robotic vehicle so that the vehicle was lightweight, easy to navigate and with extended autonomy. The solution of having a vacuum generator near the operator, with a heavy, cumbersome and stiff vacuum tubing was considered not suitable for this application.
The heart of the vortex system is the impeller along with its chamber. The optimal design was based on an extensive literature review, results from simulations as well as feedback through realistic tests.
> State of the art
Axial impellers, as their name implies, move the airstream long the axis of rotation. The air is pressurized by the aerodynamic lift generated by the fan blades. Sometimes these fans can be interchanged with radial fans, but they are commonly used in low-pressure and high volume applications. Some advantages of this group of impellers are:
- Compactness
- Light weight
- Low cost
However, this type presents a wide variety of disadvantages for the goal pursued, such as:
- Problems in situations where the air flow must vary considerably
- Anti-stall devices or impeller over-sizing to avoid the previous problem
- Higher rotation speed to achieve the same airflow capacity as centrifugal impellers
- Need for clean air
In terms of radial or centrifugal impellers the classification is a bit wider. Nevertheless, a general definition says that these devices move air first radially outwards by centrifugal action, and then tangentially away from the blade tips. As the air moves from the impeller hub to the blade tips, it gains kinetic energy which is then converted to a static pressure increase.
Within radial impellers there is another classification regarding the geometrical arrangement of the blades, namely, radial, forward inclined and backward inclined.
• Radial. The blades extend straight out from the center and its variation “radial-tip” the blades are completely straight in the outer diameter getting inclined against the rotation direction and not reaching the inlet. This family is commonly used for high speeds, low volumes and high pressures. Paying attention to their characteristics curves, we understand that these impellers are not stable and the motor can be overloaded. The power curve is always growing despite the fan curve starts to decrease after the optimum.
• Forward inclined. The blades are curved in the direction of rotation. In general, these impellers have more blades than the previous and the next group, what means weight increase. Such impellers have a low efficiency and poor control of the airflow. Furthermore, they are mostly used for high airflow and low pressure at low speed applications.
• Backward inclined. The main feature of this group is that the blades are inclined towards the opposite direction of rotation. These impellers are:
o lighter than the forward inclined ones because in this case a smaller number of blades is required,
o optimum for low flow and medium-high speed applications, and
o energetically stable. Non-overloading motor characteristic.
> Assumptions:
• According to the requirements, vacuum must be created to adhere the robot against the wall. Because of that, flow conditions can vary widely depending on the amount of suction desired by means of mainly rotation speed.
• Small dimensions of blade either in length or in height mean that the rotation speed must be high.
• The robot must be as flat as possible to avoid moments on wheels and therefore to make the system more stable.
• Many different types of equipment need to be installed on board which implies that the larger space available on the platform the better.
> Choices:
• It can be said that centrifugal impellers have a more uniform and smoother working zone, avoiding stall at any speed. Also they can work at high speeds of rotation and their design is more convenient for low airflow.
• In terms of vacuum chamber and frame, axial impellers need to expel air perpendicularly to its ration plane. It means that immediately after the impeller an axial channel is required to drive the air out. In case of radial impellers, the air is expelled radially and whether the flow is driven in that direction or through a vertical duct, more mechanical solutions can be considered.
For all these reasons, it was decided to use a radial backward inclined impeller. Both, curved and airfoils were taken into account in this study
Its final tuning was made by means of theoretical calculations, simulations and in lab tests. This design flow was the pilot to define and determinate the optimal working point and to examine its deviation from the original project specifications.
> Simulation and results
Several 3D models were designed including the blades, hub, air inlet and outlet. The main factors under consideration and design were:
• Number of blades
• Width of blade (constant or varying)
• Height of blade
• Angles of blades related to the relative velocity at the inlet and outlet.
• Inlet diameter
Based on the project specifications, it is immediate to get the pressure required. Since the maximum weight of the robot is 2kg and the payload must be of up to 2kg, the suction generated must be enough to lift up an overall mass of 4kg. With regard to rotational speed preliminary calculations revealed that the speed should be between 15,000 and 20,000rpm to get the suction levels required.
The process to determine the inlet diameter was done in parallel along with the speed one, reaching the conclusion that for working within the speeds showed before, the most suitable diameters would be between 30 and 40mm.
Having chosen a backward inclined impeller limits the number of these elements. Typically, the design of that group of impeller goes from a few blades, around 4, to hardly 14. With regard to the blade’s height two restricting factors were set:
• The robot should be as flat as possible what means that if a too high impeller is included it can contribute to make the robot falling down from the wall.
• For the same diameter, higher blades mean larger surfaces what entails more air moved and hence, more aerodynamic drag. The direct result of the previous reasoning is obvious: more power. This situation would increase the weight of the motor and hence, the weight of the overall structure.
Simulations were not done only with constant blades but with variable profiles of height as well. Finally, with respect to the blade’s width and profile it was decided to model thin ones. Radial impellers are usually designed as a decreasing profile being thicker in the inlet region and thinner in the outlet. The reason is because according to aerodynamic theory, this type of geometry favors the airflow around a certain profile, therefore the more efficient is the airflow the better is the profile and hence, the suction force.
Several different shapes of blades were designed, including airfoils, and simulations undertaken for all of them using commercially available software. The simulations resulted in 4 valid designs that were manufactured and tested. The chamber’s design included the inlet and outlet diameter and height. Several configurations were made and simulated extensively and a full prototype was manufactured for extended testing in the lab. The experimental jig included a high speed AC motor selected off the shelf, a typical plastic tapered chamber and 4 3D printed impellers. The impeller selected could offer around ~8 Kg at 2000RPMs which was considered highly promising providing as well a good safety factor for the whole VORTEXSCAN system. The final used impellers were machined from hard engineering plastics and the driving motor used is a high speed outrunner DC motor.
> Robotic vehicle design
The vehicle (platform) design was very simple and straightforward. It consists of 4 wheels, all driven for high toque capacity. The wheels are made of elastic and durable material for high coefficients of friction and endurance.
The main chassis is composed by two carbon fibre plates which host the vacuum chamber, impeller, motor and electronic components. These plates are fixed and assembled by means of the motors brackets and four steel spacers. The motor brackets have been specifically designed and manufactured in white nylon to fix the motors to the lower plate.
The last structural components are the wheels and the vacuum ring. The wheels were manufactured in nylon and they have a rubber layer to ensure good traction and avoid skidding, See previous picture.
The vacuum sealing system is shown in the image below. It is composed of a plastic interface and a rubber or foam ring which is in contact with the surface. Both tyres and ring can be changed and customized depending on the type of surface where the robot has to move along.
The driving motors are DC ones with an integrated gearbox. Moreover, they include an encoder for speed control and odometry. Its control was done remotely via an RF transmitter receiver. In addition, the main controller was based on raspberry and controllers and a power supply sized and selected off the shelf.
After the system’s full integration, it was tested extensively in lab conditions, on dry concrete walls. The results were highly promising, presenting reliability in order to incorporate all NDT equipment onboard for final tests.
> Tracking system
Depending on the inspecting surface/structure an inspection map may be needed. This is the main reason why a tracking system was required. The implemented solution was a visual tracking system performed by a simple webcam and a custom software developed. The algorithm is able to identify the mobile target and provide its relative position to a fixed reference spot, which can be the initial point or any other relevant spot of the structure. Once the fixed position has been defined, the software can track the robot position in three different ways:
• Following the object with no tracking.
• Following the object with a temporary tracking record that can be erased after a defined after a period defined by the user.
• Following the object with a constant tracking, so by the end of the inspection it is possible to check what areas have been swept.
A mimic of the software with real images regarding the tracking method was developed for testing and validation purposes. The evolution of the software is noticeable, which goes from simple forms (magnet pieces) on a white board to the real robot image.
> Integration testing and conclusions (see the attached document "VORTEXSCAN_Final Report_Figures")
All components were installed onboard the robotic vehicle achieving a payload of 2 Kg. Powering of the full system was done via an umbilical and data dispatch towards to the master PC for further processing and examination.
The GPR could perform continuous inspections with no stops except for those where the operator wanted to change parameters, whereas the UT inspections needed a spot by spot procedure expending around 40 seconds to acquire the result of every measurement.
Thanks to the GPR controlling software and, especially, the encoder, the inspection could be interrupted at any moment for the UT measuring and then resume it, since the GPR keeps records of the last position when the movement was paused. The output data were acquired and analysed afterwards to verify the status of the concrete reinforcing bars.
Potential Impact:
The VORTEXSCAN system is destined to improve the industrial competiveness across the EU. Currently, the most widely used procedures for inspecting concrete structures use scaffolding and rope access techniques, such methods are time consuming and labour intensive. The VORTEXSCAN project does not require scaffolding and will enable rapid inspection of the complete concrete structure resulting in time and cost savings. An average Nuclear Power Plant (NPP) shuts down approximately 15 days per annum for inspection and maintenance resulting in loss of revenue amounting to €165m for all NPPs in Europe. This money could be saved with our automatic inspection system as inspection can be performed without disrupting the operation of the plant. It will also give reliable assessment of concrete characteristics thereby improving the cost-effectiveness of maintenance programmes for concrete structures by locating structures in need of priority repair and, equally importantly, structures that do not need repairing. Hence, our technology promotes methods of prolonging the working life of existing concrete structures through better inspection.
This increase in operational life will lead to a reduction in the capital costs for the replacement of assets. Specifically, 30% of commercial NPPs in Europe have more than 30 years of operation and 60% operate between 20 and 30 years of service life. The design life of existing NPPs is set to be 30 – 40 years, so without the advantage that VORTEXSCAN technology can bring, a significant number of these will shutdown. Currently, nuclear energy provides 30% of the electricity supply in Europe. With a number of NPPs reaching the end of their lifespan, VORTEXSCAN technology will potentially provide an extension to their operational lives and enable a phasing of shutdowns to eliminate the predicted power shortages. This will reduce economic loss due to the sustenance of electricity supply which is crucial to the manufacturing and service delivery sectors.
Use of the VORTEXSCAN product reduces the time required for inspectors to access and stay at the walls of large structures under high risk working conditions. Once the automatic system has finished the inspection, the skilled personnel will only be required to work on specific points where a defect has been identified. The dramatic reduction of the inherent risks due to the automatic machine inspection conforms to the requirements of the current Prevention of Risks at work Act34 (European Directive 89/391/CEE).
The use of this technology also reduces the risk of premature failure of concrete structures. Failure of structures result in financial losses, environmental damage, and emission of toxic radiation in NNPs. Prevention of such disasters can be greatly aided by better timely and thorough inspections through the VORTEXSCAN technology.
List of Websites:
http://www.vortexscan.eu/