Case Study: Implementation of a New Construction Method for Integrated Management of Embankment Volume Using ICT Construction Machinery and Point Cloud Data

Embankment work at construction sites is not merely transporting and piling soil; it is an extremely important process that requires meticulous management based on precise volume calculations. Conventional surveying methods involved creating cross-sectional diagrams from multiple measurement points and calculating volumes from them, but this approach had many challenges and limitations. The more complex the site terrain, the more subjective decisions enter into the placement of measurement points, causing errors to accumulate in the final volume calculations. Moreover, even when multiple surveyors perform the same task, results can vary, undermining consistency in management. However, by leveraging 3D point cloud data, these issues can be greatly mitigated. A 3D point cloud is coordinate data of millions to tens of millions of points acquired by laser scanning or drones, and by analyzing this data with advanced processing techniques, more accurate and reliable volume calculations have become possible.

In this article, ICT construction machinery refers to heavy equipment equipped with GPS that, when linked with 3D construction designs, can automatically perform optimal construction. By combining this technology with point cloud data, construction accuracy is dramatically improved. From the three perspectives of pursuing accuracy, improving efficiency, and reducing costs, we will explain the value of volume calculation using 3D point clouds.

Conventional Embankment Volume Calculation Methods and Challenges

Traditional embankment volume calculations have mainly been carried out by combining two methods: the cross-section method and the longitudinal-section method. In the cross-section method, multiple cross-sections are set perpendicular to the construction alignment, the volume of soil in each section is calculated, and the total embankment volume is obtained by summing them. This approach is particularly effective for linear projects such as roadworks and has been a reliable method used on many sites for many years. Typically, cross-sections are set at 10 m (32.8 ft) to 20 m (65.6 ft) intervals, and the ground elevation at each location is inferred from multiple survey points. However, when the site topography is complex—for example, when valley terrain is intricate, existing structures are scattered, or rock outcrops are present—careful judgment is required to determine appropriate section locations.

At this stage of selecting cross-section positions, there is a serious problem that the surveyor's experience and judgment can greatly influence the results. For example, if cross-sections are taken at 10 m (32.8 ft) intervals at one site but at 15 m (49.2 ft) intervals at another, the calculated results may differ even for the same terrain. In the worst case, calculation errors can reach 5 percent or more, and in large-scale projects this can result in differences in earthwork volume of hundreds of cubic meters (thousands of ft^3) or more. Furthermore, fine terrain variations between cross-sections cannot be fully captured by the cross-section method. If the ground surface changes in a wavy manner, there are locally deep depressions, or fills have been constructed to accommodate complex terrain such as alluvial fans, these features are not reflected in the calculations, leading to systematic errors in the final volume.

The longitudinal-section method suffers from the same problems. In this approach, cross-sections are created along the project's centerline and volumes are calculated from terrain changes before and after construction. However, terrain changes in the direction perpendicular to the centerline are not easily captured accurately by these section methods. In particular, when the terrain changes sharply near the edges of an embankment, many measurement points are required to capture those changes accurately, which in turn greatly increases the surveying workload. It is also difficult to confirm that no unintended terrain changes occurred between the pre-construction and post-construction surveys, since the terrain can change slightly over time.

Characteristics and Advantages of 3D Point Cloud Data

3D point cloud data has the potential to fundamentally solve the problems of conventional cross-section methods. A 3D point cloud is a digital record of the coordinates of countless points in space, and each point contains three-dimensional information: X coordinate, Y coordinate, and Z coordinate (elevation). Dense point cloud data like this can be efficiently acquired by combinations of laser scanners, cameras and IMUs (inertial measurement units) mounted on drones, and GNSS receivers. Importantly, this point cloud data captures even the fine undulations of the ground surface. Terrain changes on the order of millimeters (mm; 0.04 in per mm) are also recorded and stored as data.

The greatest advantage of 3D point cloud data is its overwhelming density of measurement points. In conventional surveying, depending on site conditions, it was common to have dozens to hundreds of measurement points per hectare. In contrast, 3D point clouds yield tens of thousands to hundreds of thousands, and in some cases millions, of points for the same area. This dramatic difference in density records all the subtle variations of terrain, and as a result makes it possible to calculate volumes with much higher accuracy. Also, once 3D point cloud data are acquired, they can be analyzed from various angles and viewpoints afterward. The high reproducibility — that different surveyors analyzing the same data will obtain the same results — is another significant advantage that conventional methods could not provide. Because the objectivity of the data is guaranteed, if questions arise later about calculation results, it is easy to reuse the data and recompute.

Moreover, 3D point cloud data is very easy to compare over time. By acquiring data at multiple points in time—such as the initial, middle, and completion stages of embankment construction—and overlaying them, it becomes possible to grasp in detail which parts of the embankment have progressed and by how much, or whether uneven construction is occurring. Such time-series analysis enables improvements in construction quality and early detection of unexpected problems. Furthermore, if a problem with construction quality is found, tracing its cause is also made easy. For example, if it is discovered that the fill height at a certain location falls short of the plan, data analysis can help infer whether the reason is equipment failure, operator error, or ground settlement.

On-site Methods for Acquiring 3D Point Cloud Data

There are several options for efficiently acquiring 3D point cloud data at embankment sites. The first is the use of drones equipped with laser scanners. This approach has the major advantage of being able to survey large areas in a short time. Because drones can be flown without extensive manual intervention, measurements can be conducted safely even on hazardous slopes or in areas that are difficult to access. Generally, they are flown to an altitude of about 100 m (328.1 ft) above ground and passed over the construction area multiple times to collect measurement data. With drone-based surveying, it is possible to measure several hectares of area in a single day, which is especially time-efficient for large construction zones. However, drone surveying has the drawback of being susceptible to weather: it is difficult to fly on rainy or very windy days, and because the sun’s position differs significantly between morning and afternoon, the resulting change in shadow patterns can affect the quality of the point cloud data.

The second option is to use a terrestrial laser scanner. In this method, the scanner is mounted on a tripod and measurements are taken sequentially from multiple positions on-site. A terrestrial scanner has the advantage of higher accuracy than a drone. Typically, errors are kept to less than a few centimeters (less than a few in), and measurement accuracy at close range is particularly good. It can also acquire relatively accurate data in complex terrain and around existing structures. However, because the area covered by a single scan is limited, large construction areas require many scan positions, which results in increased time required for surveying. Furthermore, it is necessary to perform registration to align multiple scans obtained from different positions into a unified coordinate system, and this process requires technical effort.

The third option is to use photogrammetry. By processing high-resolution images taken from multiple angles, it generates a 3D point cloud. This approach has the significant advantage of being low-cost to implement, and a drone equipped with a high-performance camera can achieve sufficient accuracy. However, on soil surfaces with a uniform texture, extracting feature points can be difficult, risking a reduction in measurement accuracy. In particular, freshly deposited soil may have a uniform surface that prevents photogrammetry algorithms from finding feature points, so caution is required. The optimal method should be decided by comprehensively assessing multiple factors such as the project scale, on-site terrain, accuracy requirements, budget allocation, and weather conditions.

Process for calculating volume from 3D point cloud data

When 3D point cloud data is acquired, the subsequent data processing stage becomes extremely important. First, the acquired point cloud data contains noise, and this noise must be removed. Noise refers to points generated by measurement errors rather than the actual ground surface. For example, points acquired while the drone was vibrating, points reflected from low-reflectivity objects, points capturing dust in the atmosphere, or points representing parts of existing structures that were mistakenly recorded all constitute noise. Properly removing these greatly improves the accuracy of subsequent processing. Statistical methods that automatically detect outliers are often used for noise removal.

Next, it is necessary to distinguish between embankment areas and non-embankment areas. This is called classification, and the computer can either automatically extract embankment regions from terrain features or an operator can manually designate the areas. Automatic classification is fast, but its accuracy is not always perfect, so verification work is indispensable. Manual designation takes time, but yields more reliable results. In practice, a hybrid approach—checking the results of automatic processing and adjusting them as needed—is the most efficient. For example, operators visually confirm the boundary lines of embankment areas and correct any discrepancies with the automatic classification results.

Once the embankment area is determined, compare the two point cloud datasets of the pre-construction ground surface and the post-construction ground surface, and calculate the volume difference. This calculation is performed by aligning the two point clouds in a unified coordinate system, determining the elevation difference at each point, and aggregating those differences into a grid. The calculation results yield not only the total volume of the embankment area but also the volume for each subarea. This makes it possible to grasp construction progress in detail. For example, by dividing the construction area into 5 m (16.4 ft) square grids and calculating the volume within each grid, it becomes immediately clear which parts of the work are behind schedule.

Steps for Practical Use in the Workplace

When actually using 3D point cloud data at an embankment site, a phased approach is effective. As the first stage, record the pre-construction conditions in detail. Because the measurement accuracy at this stage will serve as the baseline for all subsequent comparative calculations, you must pay careful attention. It is also important to carry out multiple measurements and verify the consistency of the data. Recording all conditions—measurement time, weather conditions, drone flight parameters, and so on—will be useful later when performing quality assessments.

In the second stage, after construction begins, data are collected periodically. Generally, measurements are taken when the embankment fill reaches a predetermined reference volume, or when it reaches quantities that serve as indicators of construction progress (for example, 30 percent, 60 percent, 90 percent of the plan). These periodic measurements allow checking whether construction is proceeding as planned. If deviations from the plan are detected, their causes can be identified and the construction plan adjusted. For example, if it is found that embankment progress in a particular area is falling well below the plan, an investigation can determine whether the cause is an issue with the placement of heavy equipment or an unexpected change in soil properties, and appropriate countermeasures can be implemented.

In the third stage, after construction is completed the final volume is measured and compared with the design values. This comparison allows evaluation of construction accuracy and provides the basis for decisions on additional work or corrections if necessary. Typically, embankment works require accuracy within ±5 percent of the design values, but by utilizing 3D point cloud data this accuracy requirement can be reliably met. In addition, this data can be used to improve accuracy in future similar projects. By analyzing accumulated data to understand the difficulty of construction under specific terrain conditions, staffing and scheduling for future projects can be set more rationally.

Practical Tips for Improving Accuracy

There are several practical tips for improving the processing accuracy of 3D point cloud data. The first is the placement of ground control points (GCPs). This involves installing multiple points with known coordinates on site and linking them to the measurement data to enhance overall coordinate accuracy. In particular, in large construction areas the number and arrangement of these ground control points have a major impact on data accuracy. Generally, it is recommended to place them at the four corners and the center of the construction area. Ground control points should be made of highly reflective materials so they can be clearly identified during measurement. In aerial surveys using drones, especially over large areas, it has been reported that using multiple ground control points can improve accuracy by more than fivefold.

Second, consideration of environmental conditions during measurements. Measuring at the same time of day standardizes the direction of sunlight and minimizes the effects of shadows. Measurements should also be avoided on rainy days or days with high humidity. Under these conditions, the operation of lasers and cameras can become unstable, increasing the risk of degraded data quality. Furthermore, the ambient temperature at the time of measurement also has an effect. In extremely cold environments, a drone’s battery performance deteriorates and flight time is shortened. When planning measurements, it is important to fully take meteorological conditions into account and select the optimal measurement day.

The third is the selection of data processing software and its proper configuration. Point cloud processing software comes in a variety of products, each adopting different algorithms. By selecting the software best suited to your site’s conditions and appropriately adjusting the parameters, you can achieve higher accuracy. Comparing and evaluating results processed by multiple software packages is also an effective method for ensuring quality. For example, by processing the same point cloud data with multiple software packages and checking the differences in the calculation results, you can discover that a particular software is not well suited to the characteristics of the site.

Path to Streamlining Embankment Volume Management

By leveraging 3D point cloud data, the entire process of managing embankment volumes can be greatly streamlined. With traditional methods, surveying through volume calculation could take anywhere from several days to several weeks. In contrast, using 3D point clouds can significantly shorten the time from measurement to report preparation. Furthermore, because the data are managed digitally, multiple stakeholders can share the same information, greatly improving transparency in construction management. Owners, designers, contractors, and supervisors can all access the same data and monitor construction progress in real time.

The benefits of this kind of digitalization go beyond mere time savings. Because it enables decision-making based on accurate, detailed data, it leads to optimized construction planning and early detection of issues. In addition, the accumulated data can be used as benchmark information for future similar projects, contributing to an overall improvement in operational quality. For example, if data on embankment settlement and compaction characteristics under specific terrain conditions are accumulated, predictive accuracy for new projects will improve.

Currently, many construction companies are advancing the implementation of embankment management systems that leverage 3D point cloud data. These companies are achieving substantial cost reductions and quality improvements through automation of measurements, streamlining of data processing, and standardization of reporting processes. By introducing such cutting-edge technology at your site as well, you can expect significant benefits.

The Potential for High-Precision Measurement Using Smartphones

Recently, devices that add high-precision positioning measurement capabilities to smartphones have emerged, and they are poised to revolutionize embankment measurement. Conventional drones and scanners were very expensive and complicated to maintain, but smartphone-based solutions greatly lower the barriers to adoption. By utilizing equipment such as iPhone-mounted GNSS high-precision positioning devices, site workers can carry units small enough to be portable while obtaining position information with centimeter-level accuracy (half-inch accuracy). These mobile devices require relatively low initial investment, making it economically feasible to deploy multiple units.

Using such mobile devices at multiple measurement points to collect coordinate information can achieve an information density equivalent to that of point cloud data. Furthermore, a workflow in which data are sent to the cloud in real time, allowing processing results to be checked while on site, is also feasible. By leveraging smartphones, field staff who have not received special training can participate in measurement tasks, resulting in a significant increase in on-site flexibility. In addition to improved work efficiency, worker safety is also enhanced. Conventional measurements using surveying equipment required carrying heavy instruments and working on hazardous slopes, but by utilizing mobile devices, lightweight and safer measurements can be realized. Going forward, such mobile-based high-precision measurement systems are expected to become mainstream in embankment volume management.