ARRB: Let’s Get Cracking
Posted under: The pavement is a road agencies most valuable asset and cracking is one of several pavement condition parameters which, when monitored, can assist road managers to maintain their network. Traditionally, crack measurement has been a manual process but with recent developments in automated data collection, ARRB Group has developed an innovative 3D crack measurement system which can automatically identify and classify cracks. 3D laser sensors were first introduced as a means of measuring the transverse profile. However, with further advancements, this technology is now being used to assist asset managers to identify cracks and other defects in the pavement surface.
In this era of reduced funding and competing priorities road agencies usually attempt to maintain the condition of their most valuable asset, the road network, by implementing a pavement management system. This tool can help them make the most of their roads budget. However, for a pavement management system to be of value, it must be populated with fit-for-purpose data which accurately represents the pavement condition.
Out of all the pavement condition parameters measured, surface distress is often the primary parameter upon which decisions are made. Surface distress usually comprises a number of individual defects such as delamination, potholes, ravelling, patching etc. Of these distresses, many view cracking to be the most important. A cracked surface allows the ingress of water into the sub-layers which can lead to the rapid deterioration of the pavement, especially if the pavement is subject to high traffic flows or heavy loading.
Cracking is also one of the hardest distresses to identify, particularly using automated or semi-automated methods. Road agencies have been keenly awaiting an accurate, repeatable and economically viable means of automatic crack detection, particularly fine cracking, as the earlier it is identified, the quicker the cracks can be sealed and the damage to the pavement minimised.
The recent development of 3D crack measurement systems has gone a long way to fulfilling this need. Cracking was traditionally measured by foot or from the window of a slow moving vehicle. This type of manual data collection is slow, expensive and relatively unsafe as the rating team is exposed to the prevailing traffic conditions. Later on systems were developed that took video images of the pavement surface from a survey vehicle. This radically improved the speed and safety of the data collection. However, the images were still required to be rated manually, a subjective process that does not always lead to accurate and/or repeatable results. However, it did allow for 100% coverage of the road network.
In time, the introduction of external lighting and the improved performance of the pavement cameras resulted in higher resolution images and the ability to see finer cracks. However, it was still not possible to identify the cracks automatically.
The ARRB Automatic Crack Detection (ACD) system is based on the Laser Crack Measurement System (LCMS) developed by INO and marketed by Pavemetrics, a Canadian company. The system has been fully integrated into the ARRB’s Hawkeye survey platform and operates in tandem with Hawkeye’s other data collection modules.
The ACD comprises two high-performance 3D laser units that are fitted to the rear of the survey vehicle, 2.2 m above the pavement.
Each unit consists of two main components: a high-power spread line laser and a high-speed 3D camera mounted off-axis to the laser light source. When combined, the two 3D laser units project a 4 m wide laser line consisting of over 4,000 measurement points onto the pavement. Half of the image is captured by each camera which interprets the distortions to the straight laser line as variations in the vertical surface profile. Because of the high pixel resolution, measurement accuracies of 0.5 mm are claimed.
A picture of the road surface can be built up by combining sequential transverse profiles which, at 90 km/h, are only 5 mm apart (less at lower speeds). The LCMS sensors produce both range and intensity profiles which are merged to produce a 3D image. The information contained in this image allows the ACD system to automatically identify cracks and a variety of defects.
The data processing software divides the road surface into small sections 5 m long by 4 m wide which are automatically checked for cracks and other surface defects.
The variation in the colour gives an indication of the average width of each crack. The pink lines indicate the area in which the cracking is analysed and are automatically determined based on the presence of the more reflective line markings.
The 3D system is capable of producing a high-resolution picture of the road surface in which cracks and other defects are readily identifiable. However, like most automated systems, the LCMS isn’t perfect and some of the fine cracks can be missed especially if ‘pumping of fines’ is evident. This can be improved by modifying the analysis algorithms.
Similarly, for road agencies it is important that the 3D system produces results that are compatible with the crack reporting requirements of their pavement management system (PMS) or road management system. In most cases this means being able to replicate visual crack measurements that have been made either on foot or through the window of a slow moving van. The ACD can do this as the resolution of the images is sufficient to identify the crack type and, because the image is calibrated, the extent of cracking can be accurately measured in both the longitudinal and transverse directions. The ACD system can also measure crack width which allows the severity of the cracking to be determined.
Whilst the crack map and crack widths are determined automatically by the analysis software, ARRB has developed its own software to automatically classify the type and extent of cracking in accord with the requirements of several asset management systems, thus eliminating the need for human input. By doing this it produces an objective, repeatable measurement.
The ACD has been extensively trialled by ARRB. Its repeatability and reproducibility was tested over numerous test sites with asphalt, concrete and spray seal surfaces and various road classifications.
One of the major advantages of 3D imaging over conventional cameras is that it does not require additional lighting to overcome the effect of shadows. This significantly reduces power consumption. The ACD system can even operate at night without compromising the quality of the images. Additionally, the 3D images are unaffected by lens distortion but more importantly in terms of asset management, 3D systems measure an extra dimension – depth – which can be used to identify various pavement surface condition parameters in addition to cracks. These include the following:
- Rutting – rutting is determined from the transverse profile which comprises more than 2,000 measurement points in each 2 m wheel path. This equates to a 1 mm transverse resolution which is significantly greater than the current point laser systems which typically have up to 15 lasers. The system can also limit the analysis to the trafficable width of the pavement as the 3D systems have the ability to identify line marking and edge drop off.
- Potholes and delamination – a 3D system can locate potholes and delamination and by measuring their depth and diameter calculate their area and volume. This enables the accurate calculation of fill material required for pavement rehabilitation.
- Ravelling – one definition states that ravelling is the ‘progressive disintegration of the pavement surface by loss of both binder and aggregates’. The ACD identifies ravelled surfaces by calculating the volume of aggregate loss over the area.
- Surface texture – a laser profiler usually measures macrotexture longitudinally over the pavement. However, the ACD does this in the transverse direction. Its outputs have been compared against other texture measuring devices with good results.
Another feature of the 3D system is its ability to measure the reflectivity of the pavement surface. This enables the automatic detection, classification and assessment of lane markings, directional markings etc.
In conclusion, the ARRB ACD system is capable of accurately measuring cracking, a critical pavement condition parameter, plus a host of other parameters which are important to most asset managers and pavement/asset management systems. As the technology and analysis software continues to improve, it is more than likely the ACD system will measure an even wider range of pavement elements. The small size and low power requirements of modern 3D systems means that they can be integrated into a standard data collection vehicle, making this technology ideal for local road networks in both urban and rural environments.
One of the most important features of the ACD system is its ability to produce high-resolution pavement images, independent of and unaffected by ambient lighting conditions. These images can be processed by powerful software to automatically detect cracking and provide an assessment of
crack type, extent and width. This data can then be collated and formatted, emulating the visual crack assessment, to produce results that are compatible with the information contained in a road agency’s PMS or historical database.
As with any automated system, 3D systems will not always be 100% correct but they do produce an objective and very repeatable result over 100% of any road network in a safe and economical manner.
Visual assessment of the pavement surface has come a long way from the days of walking the pavement or viewing it through the window of a slow-moving van. Australian state road agencies have already begun using the ARRB ACD system to measure cracking ,thus showing their willingness to adopt new technologies for managing road pavements. 3D technology certainly has the potential to improve on the accuracy and extent of pavement condition data currently collected by state and local government agencies so let’s get cracking.