Final Report Summary - SPOTTRACK (Development of an automated spot weld inspection device for safe vehicle repair)
Executive Summary:
There are thousands of spot welds in a car and it is often necessary to remake some of these welds during repair. On the production line the spot welds are routinely destructively or semi-destructively tested but it is not possible to destructively test repair welds. Matching test samples can be made so that a destructive peel test can be carried out but there is currently no suitable method to ensure the integrity of the actual welds going into service. Repairers do not have the robotic welding equipment used by OEMs and they will be dealing with a huge variety of car models using different steel sheet thicknesses and joint configurations. They also do not have the resources to employ specialist inspection personnel to interpret inspections. However, there is a need for assured reliability of the spot welds made in car repairs. The integrity of the spot welds can be vital to the crash performance of a car and the safety of its passengers and it is the aim of repairers to restore the vehicle to its original crash-worthiness.
To address this need, the SpotTrackTM project aimed to develop an automated spot weld inspection device. The project partners (Vermon, Tecnitest, ACT, Nardoni, TWI, KTU, CAB, Tofas and Karl Vella) have worked together to firstly assess the market needs. This study identified a number of desirable attributes of the system for the consortium to aim for, such as lightweight (<2kg), low cost (<2000 Euros) and a probe that is straightforward for a non-expert to position and can be used in difficult to access regions.
Numerical modelling including finite element modelling was used to guide the design of the device. The modelling results enabled the visualisation of the sound field throughout the whole volume at any point in time during the test which resulted in a deeper understanding of the effects of the test variables on the reliability of the results. The models were developed using ABAQUS and were used to predict the effects of different probe configurations, excitation conditions and the interaction of ultrasonic waves with a range of typical spot weld defects.
The first SpotTrackTM prototype was then developed including a bespoke signal processing procedure. The device uses an array probe to allow greater inspection resolution with a multiplexer which allows the signals to be collected without the need for expensive phased array equipment. In order to ensure the SpotTrackTM device was capable of reliably determining the quality of spot welds, a set of samples was created including good welds and a range of defective welds: undersized welds; stuck welds (where the zinc coating melts and sticks the two plates together but a proper bond is not achieved); oversized (burnt) welds; and porous welds. The samples were characterised using macro sections, radiographs and conventional ultrasonic C-scan measurements. Data were gathered from the samples and assessed using the signal processing algorithms developed to provide a simple pass or fail indication and a categorisation of the type of weld.
Improvements were then made to the SpotTrackTM device and second prototype was developed. This included an increase in the frequency of the probe and improvement in signal quality, a guide for positioning the probe to help keep it centralised and parallel to the surface of the panels, and refinements to the signal processing procedure. The second prototype was then used to re-test the laboratory samples and trialled in a car under repair in a body shop. The laboratory tests showed that the false-call rate (cases where the algorithm gave an incorrect pass or fail decision) was around 1% which was well below the original target of 5%. The guide was found to beneficial in aligning the probe but in the field trials there were access problems in certain regions of the car and recommendations for improvements were identified. The results from the field trials were found to be broadly in line with visual inspection of the spot welds and the expectation of the welder.
The SpotTrackTM prototype was successfully tested in the laboratory and in the field. The cost target was exceeded but methods for reducing the cost were identified in addition to plans formed by the SMEs for selling the device to alternative markets. The prototype met its weight and accuracy targets.
Project Context and Objectives:
There are thousands of spot welds in a typical car. Automotive OEMs rely on two combined approaches to ensure the quality of these welds. Firstly, complete sub-assemblies are routinely pulled off the production line once every 10 days and torn apart weld-by-weld. Secondly, in-process monitoring is used to ensure that the weld quality remains satisfactory between one tear-down test and the next. Some OEMs perform non-destructive tests on a small number of spot welds (say 25) on 1 vehicle in 10, but this is uncommon. However, the automotive repair industry does not have these opportunities; these approaches cannot be used to ensure spot weld quality in the body repair shop. Repairers do not have the robotic welding equipment used by OEMs; they will be dealing with a huge variety of car models using different steel sheet thicknesses and joint configurations and tear-down testing is clearly not possible. Also, OEMs have the resources to employ specialist inspection personnel whereas the repairer will not be an expert in ultrasonic inspection of spot welds. Despite this, there is still a need for assured reliability of the spot welds made in car repairs. The integrity of the spot welds can be vital to the crash performance of a car and the safety of its passengers and it is the aim of repairers to restore the vehicle to its original crash-worthiness.
A need for an inspection device was therefore identified. The SpotTrackTM project aimed to develop a new device using ultrasonics that will allow the user to tell, quickly and reliably, whether a spot weld is acceptable by giving a simple and automatic pass or fail indication.
The biggest challenge with such a system is to ensure the reliability of the pass or fail indication. This is related in part to the fact that, in a repair situation, the surfaces in the region of the spot weld are often complex geometries and are difficult to access. This leads to difficulties in the accurate positioning of the NDT probe which can have a significant effect on the inspection result. It is even common for trained operators to disagree.
Currently, when working on older cars made from low strength steels, body shops create spot welds on steel sheets of the same thicknesses as those in the component to be repaired. The trial weld is then subjected to a ‘peel test’. The peel test is a widely used destructive test, in which the two panels are forced apart. The results of this test are used to verify that the welding parameters are satisfactory. Subsequent welds made using these parameters are assumed to be adequate. However, it is rarely possible for the mock-up to represent the surrounding geometry and conditions of the actual weld. Therefore the assumption that the test spot weld is identical to the one on the car can be untrue. In some cases the spot weld may not be tested at all. Furthermore, more modern cars are built from high strength steels. The peel test cannot be applied to such steels because, when spot welds in such steels are subjected to peel testing, fracture often occurs in the nugget itself. Without the peel test, the repair shop has no way of ensuring that a sound weld is being made. The proposed new system will have the advantage of testing the actual weld made within a matter of seconds and without the need for interpretation. Most importantly, it will ensure that the actual spot welds going into service are safe.
The SpotTrackTM project has developed novel modelling and signal processing techniques for inspection of spot welds. Those techniques have been used to develop novel routines and inspection procedures that enable a reliable pass or fail indication to be given in an easy to use robust device applicable to current and new materials. The main aims of the project were to:
• Develop accurate and reliable predictive models for the ultrasonic testing procedure and ensure the predicted results are in agreement with actual observations with false call rate <5%, i.e. in fewer than 5% of cases will a sound weld be reported as a fail or a defective weld be reported as a pass.
• Use modelling to design new inspection parameters and equipment and provide valuable information for signal processing development.
• Develop signal processing routines that allow a reliable pass or fail decision to be given both for conventional automotive steels and new lightweight materials (e.g. high and ultra-high strength steels).
• Develop a low cost device (<€2k)
• Develop a hand-held and lightweight device (<2kg)
• Develop a novel ultrasonic probe that will cope with the access requirements of a repair body shop
• Produce a prototype that is user-friendly with robust tool positioning.
Project Results:
The finite element software ABAQUS was used to study the capabilities of a range of ultrasonic array options with respect to their ability to distinguish between different spot weld conditions and the numerical modelling software, CIVA was used to study the ultrasonic beam profile for a range of probe configurations. The investigation determined that an array probe combined with the employment of full matrix capture is the best way to get the enhanced capability required to deal with the varied situations experienced in a car repair centre.
An assessment of the current state-of-the-art in signal processing of ultrasonic spot weld inspections was reviewed and assessed. Building on this knowledge, algorithms for identification of the backwall and intermediate reflections, and their parameters (amplitude, delay time) estimation as well as for estimation of weld size were developed. Initial tests were carried out on simulated data from the finite element models. An additional idea of estimation of the nugget diameter was also tested and it was found that this was possible with accuracy of twice the width of the array elements.
A first SpotTrackTM prototype was put together and initial signals using the device were successfully gathered and analysed. The difference between the case of a good weld and loose weld was clear from the initial test results. The initial prototype was close to the desired weight limit but when the device is contained in one box, it is expected that the weight will be well within the 2kg limit. The prototype exceeded the desired cost. However, routes for reducing the cost were suggested such as identifying a supplier capable of making a large number of systems that combine the pulser-receiver unit with the multiplexer thus removing any redundancy in the current configuration.
In order to assess the SpotTrackTM device, a set of around 30 spot weld samples were created containing both good welds and welds with a broad range of defects. The samples were characterised using macro sections, conventional C-scans and radiography. This ensured that there was adequate information available on the samples for assessment and validation of the prototype. The samples were tested using the first prototype. Figure 1 of the attachment shows an example of one of macro sections of the samples.
It was found that in most cases the sample was categorised correctly by the algorithm. There were some ‘unknown’ or ‘burnt’ results mostly relating to pores. However, since the acceptance criteria for pores vary with manufacturer and the known pores were close in size to the limits, this was not necessarily an undesirable outcome. Excluding the pores cases taking ‘unknown’ to be a ‘pass’, the total number of cases tested (including repeat scans) was 97 and out of these one ‘good’ case was incorrectly identified as ‘burnt’. This translates to a false call rate of around 1% which is well within the target of 5%. Figure 2 of the attachment shows some examples of the algorithm results.
In addition to the pass or fail indications, methods for measuring the size of the nugget were included in the algorithm and tested. It was not yet confirmed how accurate the values reported by the algorithm were as it is not possible to determine the nugget size without destructive examination of the sample. Some potential improvements to the sizing algorithm were identified.
Two probes with different array configurations were manufactured and tested on the samples. The performance of the two probes was found to be similar but some problems were identified in the signal quality for both cases and improvements such as damping and central frequency to reduce the pulse width were suggested.
Further modelling work was carried out to help identify the limits of the system. The algorithm was found to classify the pores cases as either ‘unknown’ or ‘burnt’. A parametric modelling study on pore size showed that there was a potential new technique that could be used to detect pores with a diameter of 0.5mm or greater. This is close to typical car manufacturers’ acceptance criteria of pore size no greater than 10% of the weld diameter.
Further modelling work was also carried out to study the effect of surface profile on the inspection outcome. It was shown that the surface profile of the spot weld can significantly affect the signal quality. However, the profile is closely linked to the weld parameters and quality and this is therefore thought not to present a problem for the algorithm.
The use of a rubber delay line was identified as a practical solution for the new prototype with a shape designed to couple to the spot weld profile. Its effects on the signals were investigated using modelling and experimentation and it was found that there was reduced signal quality using the bullet shape. The shape and form of the delay should therefore be improved in the new prototype. Either an enclosed water delay or a different shape/hardness of rubber delay line were identified as potential solutions.
In addition, experimental work on probe angle and position relative to the centre of the weld was carried out and it was shown that this can have a significant effect on the signal quality. It was therefore recommended that the prototype be modified to include a positioning guide that helps the user to place the probe correctly.
Improvements were made and a second SpotTrackTM prototype was developed. The improvements included an increase in the centre frequency, sensitivity and pulse length of the probe. This was shown to improve the performance on a flat stainless steel target and has therefore enhanced the performance of the prototype. Data were also gathered using two different multiplexing systems and a comparison of the algorithm output showed that the multiplexer selected for the SpotTrackTM device was fit for purpose.
A probe positioning guide was designed and manufactured. The guide performs two functions which are: ensuring the probe is kept perpendicular to the surface of the spot weld; and allowing the operator to align the probe to the centre of the spot weld.
Finite element analysis and theoretical calculations were carried out to help guide the design of a captive water column probe holder. Figure 3 in the attachment shows the predicted pore pressure from one of the simulations used. Improvements to the signal processing algorithm were implemented and it was adapted to work with the captive water column data collection method.
Finally, the second SpotTrackTM prototype was tested in field trials and it was found that the results were in line with visual inspection and the expectations of the welder. Figures 4 shows the probe being used in the field trials and Figures 5 and 6 show the probe holder and positioning guide respectively. The captive water column probe holder performed satisfactorily and whilst the probe guide was useful in the laboratory trials, it was found to be too big on the car body. Many of the spot welds were difficult to access using the current design and it was necessary to use the probe holder without the guide. For future developments, a redesign of the probe holder and guide to make the overall diameter smaller was recommended.
Potential Impact:
The SpotTrackTM project has made significant progress in the challenge of the development of an automated spot weld inspection device for safe vehicle repair. The SpotTrackTM prototype is the first of its kind and it has been possible to demonstrate its capabilities against a set of known spot weld samples and meet the target false call rate of <5%. The device has also been shown to work in a body shop environment.
The SpotTrackTM prototype includes a number of innovations as follows:
• A novel array probe and captive water column probe holder suitable for spot weld inspection
• A probe positioning device
• A low cost solution to simulate phased array capability synthetically
• A signal processing algorithm capable of distinguishing between a wide range of spot weld conditions without the need for a skilled operator
The SpotTrackTM device therefore has the potential to provide financial benefit to the consortium SMEs through potential sales to the repair market and larger automotive aftermarket companies. In addition, there are other parties that may benefit from the device such as car manufacturing OEMs and insurance companies and the SMEs have plans to explore these possibilities. The societal benefits are also important as the device will help to ensure that the crash-worthiness of repaired cars is restored to that of cars coming off the production line.
Promotion of the SpotTrackTM concept and prototype was tackled in a number of ways including raising awareness with potential end users, producing papers and articles and production of a website. The SMEs decided to limit the information that could be used in technical publications due to the novel nature of the device and therefore produce shorter magazine style publications rather than journal articles.
A working group from within the consortium visited Tofas in Turkey to promote the SpotTrack concept and learn about the challenges the prototype may face if it were to be marketed in the automotive production sector in addition to the automotive repair sector.
The consortium also held a meeting at one of Karl Vella Group’s body shops (in the UK) to help raise awareness of the SpotTrack concept and learn about the challenges the prototype will face in the repair industry.
To help promote the concept to a wider audience, the consortium made use of CAB’s links with a number of body shops and arranged to hold the SpotTrackTM field trials at a Ford accident repair body shop in Norwich. The prototype was met with interest and the results appeared to agree with visual inspection of the spot welds and the expectation of the welder.
ACT used their contacts with a number of relevant magazines and arranged for the development director of Body Shop Magazine to view the prototype at some point in the future. The development director agreed that quality control of spot welds was an issue and agreed there was an opportunity given an acceptable cost for the device.
Tecnitest attended the BINDT conference 2013 and used this as a forum to promote the SpotTrackTM concept. In order to help promote the project in more general terms, the RTD partners also wrote a number of articles and journal papers. In order to avoid revealing the novelty of the system (as requested by the SMEs), the journal papers were based on modelling and signal analysis of one of the candidate arrays that was researched early in the project and rejected in favour of the final design.
The future exploitation for the next four years is planned. In the first year the SMEs will enhance the existing prototype to demonstrate the technology to potential customers and ensure the system is sufficiently rugged for the environment of a typical body shop. A number of trial sites will be established where the system is provided free or at reduced cost to users in exchange for their agreement to make initial results of use publically available and act as demonstration sites. Then in the following three years the SMEs will concentrate on establishing a presence in major European markets for the SpotTrackTM system. The data collected in the first year following the project will be used to assist in promoting the system. Although there will be effort directed towards direct sales of the systems to body shops, a major target during this period will be insurance companies across Europe, as the most rapid route to achieve substantial sales of the SpotTrack system will be to persuade the insurers that they should specify use of such equipment by all of their authorised repairers.
List of Websites:
A project website was set up to promote SpotTrackTM as well as to act as a communication port between the partners in the background with the following address:
www.spottrack.eu
For general enquiries regarding the SpotTrackTM project please contact An Nguyen-Dinh at an.nguyendinh@vermon.com or Vermon SA,180 rue du Général Renault, 37038 Tours Cedex 1, France, Tel: +33 247 374 278, Website: www.vermon.com.