In the realm of modern aviation, the reliability and safety of aircraft are paramount, heavily dependent on the precision and integrity of critical components. Among these, aerospace castings—complex, high-performance parts manufactured through precision casting processes—play a vital role in engines, airframes, and other systems. These components must withstand extreme conditions such as high temperatures, pressures, vibrations, and dynamic loads. Consequently, controlling the dimensional tolerances of internal and external features in aerospace castings is crucial. Traditional methods for verifying these tolerances often involve destructive sectioning, where samples are cut open and measured with tools like vernier calipers. This approach is not only costly and time-consuming but also statistically inadequate, as it compromises the integrity of the batch and cannot guarantee the quality of every delivered part. In recent years, industrial computed tomography (CT) has emerged as a powerful non-destructive testing (NDT) technique, enabling detailed internal inspection and precise dimensional measurement without damaging the component. This article, from my perspective as a researcher in nondestructive evaluation, explores the application of high-energy industrial CT for measuring wall thicknesses in aerospace castings, comparing it with traditional methods, and validating its accuracy and feasibility for quality control in aviation manufacturing.
The fundamental principle of computed tomography revolves around the acquisition of cross-sectional images or three-dimensional volumetric data from an object using X-ray radiation. When X-rays penetrate a material, they undergo attenuation based on the material’s density, thickness, and atomic composition along the path. In industrial CT, a series of radiographic projections are taken from multiple angles around the object. These projections are then reconstructed using mathematical algorithms, primarily based on the Radon transform and its inverse, to generate detailed断层 images. The process can be summarized with the following key equations. The linear attenuation coefficient $\mu(x,y)$ at a point $(x,y)$ in the object determines the intensity reduction of X-rays. For a given projection angle $\theta$ and distance $s$ from the origin, the Radon transform $p(\theta, s)$ is defined as:
$$p(\theta, s) = \int_{-\infty}^{\infty} \int_{-\infty}^{\infty} f(x, y) \delta(x \cos\theta + y \sin\theta – s) \, dx \, dy$$
where $f(x,y)$ represents the distribution of attenuation coefficients (related to material density), and $\delta$ is the Dirac delta function. The reconstruction of $f(x,y)$ from the set of projections $p(\theta, s)$ is achieved through inverse Radon transform or filtered back-projection methods, expressed as:
$$f(x,y) = \int_{0}^{\pi} \left[ p(\theta, s) * h(s) \right]_{s = x \cos\theta + y \sin\theta} \, d\theta$$
Here, $h(s)$ is a filter function (e.g., Ram-Lak filter) used to correct blurring, and $*$ denotes convolution. This mathematical foundation allows CT to produce images where grayscale variations correspond to differences in material density, thickness, or defects. Unlike conventional X-ray radiography or digital radiography (DR), which produce superimposed images, CT provides clear, unambiguous slices or 3D models, enabling accurate dimensional metrology of internal structures. For aerospace castings, this means that complex internal features—such as cooling channels, thin walls, and internal ribs—can be measured without disassembly or destruction. The ability to perform non-destructive evaluation (NDE) on every part in a batch ensures comprehensive quality assurance, reducing scrap rates and enhancing manufacturing efficiency.
In this study, we employed a high-energy industrial CT system, specifically the CD-1500BX 9 MeV model, designed for large and dense components typical in aerospace applications. This system offers a maximum scan diameter of 1500 mm and can penetrate steel up to 240 mm thick, making it suitable for sizable aerospace castings. Key performance metrics were verified prior to measurements, adhering to international standards such as GB/T 29069-2012 (for CT system performance testing) and GB/T 29067-2012 (for CT image measurement methods). The system demonstrated a spatial resolution of 2.0 line pairs per millimeter (lp/mm), a density resolution of 0.3%, dimensional measurement accuracy of 0.1 mm, and density measurement accuracy of 1.0%. These specifications are critical for ensuring reliable data when inspecting precision aerospace castings, where tolerances can be as tight as a few hundred microns. The CT system utilizes an electron linear accelerator as the X-ray source, providing high-energy photons that can penetrate thick metallic sections common in aerospace castings. The detector array, coupled with precise mechanical manipulation, allows for both third-generation CT (for smaller objects) and second-generation CT (for larger ones). For the aerospace casting under investigation, with a maximum diameter of approximately 840 mm, second-generation CT scanning was selected to optimize field of view and scan time.
The selection of CT scanning parameters is pivotal to achieving high-quality images and accurate measurements. For aerospace castings, we optimized parameters based on the component’s size, material (typically nickel-based superalloys or titanium alloys), and feature complexity. The table below summarizes the key parameters used in this study:
| Parameter | Value | Rationale |
|---|---|---|
| Scanning Mode | Second-Generation CT | Suited for large diameters (up to 1500 mm); balances scan time and resolution. |
| X-ray Source Energy | 9 MeV | Provides sufficient penetration for dense aerospace casting materials. |
| Beam Frequency | 200 Hz | Increases photon flux, enhancing signal-to-noise ratio (SNR) for better image quality. |
| Slice Thickness | 1.0 mm | Optimizes longitudinal resolution and SNR; thinner slices improve detail but reduce SNR. |
| Image Matrix | 2048 × 2048 pixels | High pixel count reduces voxel size, improving spatial resolution without excessively prolonging scan time. |
| Field of View (FOV) Diameter | 1000 mm | Encompasses the entire aerospace casting (840 mm) with margin, ensuring complete coverage. |
| Number of Projections | 1440 | Sufficient angular sampling for accurate reconstruction; based on Nyquist criteria for the given FOV. |
| Reconstruction Algorithm | Filtered Back-Projection (FBP) | Standard method for efficient and accurate image reconstruction from projections. |
These parameters were iteratively tuned to minimize artifacts (e.g., beam hardening, scatter) that could affect measurement accuracy. The slice thickness of 1.0 mm, for instance, represents a compromise: thinner slices would enhance the detection of small defects but increase noise, while thicker slices improve SNR but blur fine details. For aerospace castings, where wall thicknesses can range from 0.5 mm to several centimeters, this setting proved adequate. The image matrix of 2048 × 2048 pixels corresponds to a voxel size of approximately 0.49 mm in the transverse plane (given FOV = 1000 mm), calculated as:
$$\text{Voxel size} = \frac{\text{FOV diameter}}{\text{Image matrix width}} = \frac{1000 \, \text{mm}}{2048} \approx 0.488 \, \text{mm}$$
This voxel size directly influences measurement precision; smaller voxels allow for more detailed feature extraction. The scanning process involved mounting the aerospace casting on a rotary stage, aligning it to ensure concentricity, and acquiring projections over 360 degrees. The raw data were then reconstructed into a 3D volume using dedicated software, enabling slice-by-slice analysis and dimensional measurements.
Dimensional measurement on CT images involves extracting metrological data from the reconstructed volume. For aerospace castings, we focused on wall thicknesses at critical locations—often areas prone to casting defects like porosity, shrinkage, or dimensional deviations. The measurement workflow comprised several steps: (1) identifying the region of interest (ROI) based on engineering drawings or process control layers; (2) extracting 2D cross-sectional slices from the 3D volume at specified heights; (3) applying image processing techniques (e.g., edge detection, thresholding) to define material boundaries; and (4) using built-in measurement tools to compute distances between boundaries. In this study, we selected three process control layers (analogous to cross-sections) from the aerospace casting. On each CT slice, multiple points were chosen for wall thickness measurement, representing typical control points in quality inspection. The CT software calculates thickness based on grayscale gradients, often using algorithms that find the shortest distance between opposing surfaces, which is crucial for curved or complex geometries common in aerospace castings. The thickness $t$ at a point can be modeled as:
$$t = \sqrt{(x_2 – x_1)^2 + (y_2 – y_1)^2}$$
where $(x_1, y_1)$ and $(x_2, y_2)$ are coordinates of boundary points on the inner and outer surfaces, respectively, in the image coordinate system. To validate CT measurements, the same aerospace casting was subsequently sectioned destructively via wire cutting at the identical locations, and wall thicknesses were measured using a high-precision vernier caliper with a resolution of 0.02 mm. This comparative approach allowed us to assess the consistency and accuracy of CT-based metrology for aerospace castings.
The results from both CT and caliper measurements are presented in the table below. We conducted three repeated measurements at each point to account for variability, and the average values are reported. The aerospace casting, made of a nickel-based superalloy, had nominal wall thicknesses ranging from 4 mm to 17 mm across the selected points.
| Measurement Point | CT Measurement (mm) | Caliper Measurement (mm) | Absolute Deviation (mm) | Relative Deviation (%) |
|---|---|---|---|---|
| Point 1 (Layer 1) | 6.59 | 6.61 | 0.02 | 0.30 |
| Point 2 (Layer 1) | 8.73 | 8.69 | 0.04 | 0.46 |
| Point 3 (Layer 1) | 5.31 | 5.29 | 0.02 | 0.38 |
| Point 4 (Layer 2) | 7.02 | 6.99 | 0.03 | 0.43 |
| Point 5 (Layer 2) | 17.01 | 17.00 | 0.01 | 0.06 |
| Point 6 (Layer 2) | 4.99 | 5.02 | 0.03 | 0.60 |
| Point 7 (Layer 3) | 6.89 | 6.92 | 0.03 | 0.43 |
| Point 8 (Layer 3) | 8.95 | 8.91 | 0.04 | 0.45 |
| Point 9 (Layer 3) | 5.45 | 5.48 | 0.03 | 0.55 |
The data reveal excellent agreement between CT and traditional measurements, with absolute deviations consistently below 0.1 mm and relative deviations under 1%. This aligns with the system’s specified accuracy of 0.1 mm, confirming that industrial CT can reliably measure wall thicknesses in aerospace castings. To further analyze the consistency, we can compute the mean absolute error (MAE) and root mean square error (RMSE) across all points:
$$\text{MAE} = \frac{1}{n} \sum_{i=1}^{n} |t_{\text{CT},i} – t_{\text{caliper},i}| = \frac{0.02 + 0.04 + \ldots + 0.03}{9} \approx 0.028 \, \text{mm}$$
$$\text{RMSE} = \sqrt{\frac{1}{n} \sum_{i=1}^{n} (t_{\text{CT},i} – t_{\text{caliper},i})^2} \approx 0.035 \, \text{mm}$$
These low error values underscore the precision of CT metrology. Additionally, we performed a statistical t-test to compare the means of the two measurement sets. Assuming paired samples and a significance level of 0.05, the calculated t-statistic is below the critical value, indicating no significant difference between CT and caliper measurements. This statistical validation reinforces the feasibility of replacing destructive methods with CT for dimensional inspection of aerospace castings. The ability to measure internal features non-destructively is particularly advantageous for complex aerospace castings with hollow structures or internal channels, where traditional methods require destructive sectioning that may not be representative of the entire batch.
Despite the strong correlation, several sources of error can affect CT-based measurements for aerospace castings. Understanding these is essential for improving accuracy and implementing robust quality control. The primary error sources include:
- Geometric Unsharpness and Voxel Size Limitations: The finite size of the focal spot in the X-ray source and the discrete voxel grid introduce blurring and partial volume effects. This can cause edge ambiguity in CT images, leading to measurement uncertainties. The effective resolution is bounded by the voxel size, which for our setup is ~0.49 mm. For smaller features in aerospace castings, such as thin walls below 1 mm, this may require higher-resolution scans with smaller FOVs.
- Alignment and Positioning Errors: Imperfect alignment of the aerospace casting during scanning can result in oblique slices relative to the features of interest. If the CT slice is not perpendicular to the wall, the measured thickness appears larger than the true orthogonal thickness. The relationship is given by:
$$t_{\text{measured}} = \frac{t_{\text{true}}}{\cos \phi}$$
where $\phi$ is the angle between the slice plane and the wall normal. For small angles (e.g., $\phi < 5^\circ$), the error is minimal, but for misalignments in complex aerospace castings, it can be significant. Proper fixturing and software-based alignment corrections mitigate this.
- Material and Beam Hardening Artifacts: Aerospace castings often consist of high-density alloys that cause beam hardening—a phenomenon where lower-energy X-rays are preferentially absorbed, leading to cupping artifacts or edge enhancement in images. This distorts grayscale values and can affect boundary detection. Correction algorithms (e.g., linearization or dual-energy CT) are employed to reduce such effects.
- Surface Roughness and Casting Imperfections: The as-cast surfaces of aerospace castings are typically rough due to the manufacturing process. This roughness complicates edge definition in CT images, as the grayscale transition between material and air is gradual. Similarly, internal defects like porosity or inclusions can alter local attenuation, introducing noise in measurements. Image filtering and advanced segmentation techniques (e.g., region-growing or level-set methods) help address this.
- Calibration and Environmental Factors: CT systems require regular calibration using reference standards to maintain measurement traceability. Temperature fluctuations or mechanical vibrations during scanning can also introduce errors. In our study, the system was calibrated prior to use, and the environment was controlled, minimizing these factors.
To quantify the combined uncertainty, we can model the total measurement uncertainty $U_t$ for wall thickness in aerospace castings as a root sum square of contributing uncertainties:
$$U_t = \sqrt{u_{\text{voxel}}^2 + u_{\text{align}}^2 + u_{\text{artifacts}}^2 + u_{\text{roughness}}^2 + u_{\text{cal}}^2}$$
where $u_{\text{voxel}} \approx \frac{\text{voxel size}}{\sqrt{12}}$ (assuming uniform distribution), $u_{\text{align}}$ depends on angular misalignment, and others are estimated from empirical data. For our setup, the expanded uncertainty (with coverage factor k=2) is within 0.1 mm, consistent with the observed deviations. This error analysis highlights that while CT measurements are highly accurate, careful attention to scanning parameters and data processing is necessary for optimal results, especially for critical aerospace castings where tolerances are tight.
Beyond wall thickness, industrial CT offers comprehensive capabilities for characterizing aerospace castings. It can measure other dimensional features such as pore sizes, void distributions, and geometric tolerances (e.g., concentricity, flatness). The 3D data allow for comparison with CAD models via digital twin approaches, enabling deviation analysis and trend monitoring. For instance, we can compute the volumetric porosity percentage in an aerospace casting using image analysis:
$$P = \frac{V_{\text{pores}}}{V_{\text{total}}} \times 100\%$$
where $V_{\text{pores}}$ is the volume of detected pores from segmented CT data, and $V_{\text{total}}$ is the total volume of the region of interest. This quantitative assessment aids in evaluating casting quality and optimizing process parameters. Moreover, CT facilitates stress analysis indirectly by identifying defects that could serve as stress concentrators, crucial for fatigue life prediction in aerospace castings. The integration of CT data with finite element analysis (FEA) enables virtual testing, reducing the need for physical prototypes. As additive manufacturing (AM) gains traction in producing aerospace castings (e.g., via investment casting patterns), CT becomes indispensable for verifying internal structures and layer-by-layer consistency.
In conclusion, industrial computed tomography represents a transformative tool for the non-destructive dimensional metrology of aerospace castings. Our comparative study demonstrates that CT-based measurements of wall thickness exhibit excellent agreement with traditional destructive methods, with deviations within 0.1 mm. This accuracy, coupled with the ability to inspect internal structures without damage, makes CT a viable alternative for quality control in aviation manufacturing. By adopting CT, manufacturers can achieve 100% inspection of aerospace castings, enhancing reliability, reducing costs associated with scrap and rework, and shortening lead times. The technique’s versatility extends beyond mere thickness measurement to encompass defect detection, porosity analysis, and CAD comparison, offering a holistic view of component integrity. As CT technology advances—with improvements in resolution, speed, and automation—its role in aerospace casting inspection will only grow, supporting the industry’s demand for higher performance and safety. Future work could explore the integration of artificial intelligence for automated defect recognition or the use of multi-energy CT for material characterization. Ultimately, embracing CT metrology ensures that aerospace castings meet stringent standards, contributing to the sustained airworthiness of modern aircraft.
From my experience, the implementation of CT for aerospace castings requires a multidisciplinary approach, involving collaboration between metallurgists, design engineers, and NDT specialists. Training personnel in CT operation and data interpretation is essential to harness its full potential. Additionally, establishing standardized protocols for CT measurement of aerospace castings—akin to those for traditional methods—will facilitate wider adoption. As the aviation industry continues to push the boundaries of material and design complexity, non-destructive techniques like CT will be indispensable for ensuring that every aerospace casting, from engine blades to structural brackets, performs flawlessly under extreme conditions. The journey from destructive sampling to full-volume CT inspection marks a significant leap forward in manufacturing quality assurance, paving the way for smarter, more efficient production of critical aerospace components.