Michael Oberrauch, July 2017
There are many reasons to use non-destructive testing (NDT) methods for the inspection of a product. To do so, however, a product should follow a number of design-constraints or guidelines, depending on the favoured NDT method: Design to NDT. Designing a product with higher testability in mind reduces the test effort which leads to direct cost reduction. It is therefore essential to regard the needs of NDT during the design phase of a product [1].
This article mainly refers to the topic of mechanical engineering, but there are many parallels to other fields, including civil engineering and environmental engineering.
The product development contains all activities necessary to produce a product. The process starts with an idea and ends when the product is launched [2][3]. Figure 1 displays a comparison of the production process based on VDI 2221 [2] and the product development process based on Pahl/Besitz [3]. The phases which refer to the design and use of NDT are also marked and will be discussed in the subsequent chapters.
Product development starts with the planning phase. In this phase, market analysis (the needs of the customer) and benchmark analysis (business and design goals) are conducted. Subsequently, the concept phase begins. It includes the planning of the product and the definition of emergent tasks. After this phase, the product exists as a concept. The concept summarises the functions, specifications, requirements and solutions which were previously formulated. In the design phase, the concept with all its properties is transferred into a design model (mostly CAD). Thereby the model is created by starting from an outline and progressing into greater levels of detail, resulting in a detailed design. This detailed CAD-model is further elaborated until it fulfils the requirements of a series-production. It may be useful to manufacture some prototypes to ascertain whether it is necessary to produce auxiliaries such as special tools or fixtures to enable a problem-free start into series-production. Normally, the market launch commences with prototypes for select customers to be tested under real conditions. With the market launch, the lifecycle of the product has begun.
Figure 1: Left: the production process with examination at the end, based on VDI 2221 [2] Right: The product development process based on Pahl/Besitz [3] with design and use of NDT marked |
NDT is an early topic in the development process (see figure 1). During the planning and concept phase, among the first considerations are quality assurance, monitoring during production- and service-time (online-process- and health-monitoring), testing and testability of the product. These issues refer to the topic of NDT and are the basis for the decision-making process concerning the optimal NDT method for the product. It is not always easy to answer this question as there are a lot of affecting parameters.
In the ideal case, the optimal NDT method arises from the requirements of the NDT method itself and the properties of the product. Relevant properties of the product are possible defects and imperfections, material, geometry and testability. The requirements of the NDT method are detectable defects and imperfections, applicable material and geometry, testability (e.g. applicable from one side) and test effort. As aforementioned, in the ideal case the optimal NDT method is arrived at from these properties and requirements. Otherwise it may be necessary to run through an iterative cycle in which the product may be optimized in its design and/or material to be testable and/or another NDT method is chosen. A common reason necessitating this iterative cycle is an infeasible test effort, for instance the non-availability of the optimal NDT method in the institute. This circumstance implies that mostly the product has to be optimized in its design (changing the material is rarely feasible) to be testable with a certain NDT method.
This is the point at which the real topic of design to NDT becomes pertinent. All parameters resulting from the requirements of the chosen NDT method should be used as an additional degree of freedom in the (refining) design process [1] (e.g. material choice in the concept phase, design and geometry of the product, etc.).
One of the biggest advantages of a product designed for good testability is its usefulness long before the product has started its lifecycle. Depending on the NDT method and the scope of testing, NDT methods can even be used when the product only exists in slug-state during production (see figure 1). However, NDT can also be used when the product is already in service during its lifetime. This indicates that the benefit of NDT on a product can range from the concept phase (testing exemplarily some slugs – online process monitoring) over to the design phase (testing first prototypes) till the end of lifetime (health monitoring), where the product is not needed anymore.
The topic of online process monitoring plays an important role since the degree of automation in the manufacturing process is increasing. Especially in the case of relatively young processes such as the manufacturing of composites. These processes have many parameters which interfere with each other, and quality assurance remains a key priority. Furthermore, these processes are divided into more distinct steps compared to the manufacturing of metal products. Here, NDT is often the only possible method which is industrially applicable [4]. It is highly advantageous to be able to test the product after every step in the manufacturing process. When doing so, imperfections and defects can be detected early and corrected when possible. NDT thus helps to reduce material costs and rejection rates.
There is a wide variety of NDT methods. This wide variety leads to a wide range of parameters impacting the testability of a product. For example, the impact echo method has a geometry-restriction for testability of a thickness to aspect ratio of 1:5 [5][6]. Other possible constraints of a NDT method are the size of detectable anomalies in a product. The ultrasonic echo method for instance can only detect imperfections in the product as long the detected imperfection is not deeper in than the start-up length of the measurement signal. Otherwise there is the risk of interferences with the echo of the start-up length. The start-up length of the measurement signal in turn depends on the thickness of the tested product.
A generic model of a testing situation can be described with three main parameters [1]:
These three entities interact with each other, enabling the distinction of different testing situations. These situations depend largely on the faculty responsible for the testing. For instance, the design and stress departments may be interested in the presence of a certain anomaly in a given geometry, whereas for NDT, the available method is of importance. Figure 2 schematically shows the interaction between these three main parameters. The circle can be passed clockwise or non-clockwise. Depending on the direction, the parameters change from an active to a static position or vice versa. For instance, either a geometry can be examined by a testing method or a testing method can examine a geometry.
Figure 2: The basic components of the generic model to describe a testing situation [1] |
The following provides a brief example of how one can use the generic model to identify the optimal NDT method for the product. Table 1 presents the seven most common NDT methods with different attributes and yet also displays the relationship between the basic components of the generic model. Table 2 displays the most common (simplified) geometries and anomalies with different attributes. Table 1 is, in respect to the generic model, expanded by the attributes: based on technology, applicable to materials, applicable from one side and test effort.
It should be noted that both tables display only a section of the whole and they only represent one possibility of how the entities and attributes can be displayed. There are much more attributes which can describe the testing methods more accurate and a product can also contain much more different geometries and anomalies than displayed in table 2. However, table 1 and 2 display a decent variety of testing methods, geometries and anomalies.
Example: One can have the question for the ideal way to inspect products
The answer is the following (table 1):
Another query could be about the best method for detecting foreign substances in metal products. As only the methods 2, 3, 4 and 5 can detect foreign substances, but only method 3 is applicable to metal products with acceptable testing effort, method 3 is the one to choose (Table 1).
Table 1: Testing methods with different attributes [1][6]
Name | Based on Technology | Applicable to Materials | Applicable to Geometry | Applicable from one Side | Can detect Anomaly Type | Test Effort |
---|---|---|---|---|---|---|
Method 1 | Ultrasound, Impact-Echo | all | A, B, D (partially for air coupled) | NO (for air coupled), YES (others) | I, II, IV, V, VI | + |
Method 2 | X-Ray and Computer Tomography (CT) | all | A, B, C | NO | I, II, III | - |
Method 3 | Thermography | all | A, B, C, D, X (active IR-TG) | YES | I, II, III, IV, V, VI | 0 |
Method 4 | RADAR and Microwaves | CFRP, Concrete, the detected material has to be electroconductive | A, B, D, X (especially with high resolution) | YES | III (for sufficient impe-dance contrast), V | 0 |
Method 5 | Eddy Current (Induction and Electrical Field) | Electroconductive, CFRP | A, B, C, X | YES | I + IV (if perpendicular to surface), II, III (if not electroconductive), V | + (fast, no preparation) - (a lot of training needed) |
Method 6 | Dye penetration, Magnetic particle | all | A, B, C, X | YES | I, II, IV, VI (Dye: if open to surface, Magnetic: if near to surface) | + |
Method 7 | Tap testing | all | A, B, D | YES | I, II, IV, VI | + |
Table 2: Geometries and Anomalies with different attributes [1][6]
Geometry Name | Shape | Anomaly Name | Type | |
---|---|---|---|---|
Geometry A | Flat | Anomaly I | Delamination | |
Geometry B | Curved | Anomaly II | Porosity | |
Geometry C | Connection | Anomaly III | Foreign Substance | |
Geometry D | Rough surface | Anomaly IV | Disbonds | |
Geometry X | Just near surface | Anomaly V | Hollow | |
Anomaly VI | Cracks |
These brief examples convey how the design to NDT can be used during the development process. However, as the previous chapter demonstrates a lot of departments work on the same product, each with its own perspective. It is not remarkable that good communication between the individual departments is one of the biggest issues when designing a product in respect to the needs of NDT.
As already mentioned, table 2 displays the most common geometries (see table 2). These geometries are very simple and can be specified to be more accurate. One possible way to do this is to add specifications on the exact radius of curvature. The curvature can be calculated as \kappa=\frac{\Delta \phi}{\Delta s}, where \phi is the corresponding angle to the arc s. The circle of curvature itself can be chosen at random because if \phi={constant} the arc s=r*\phi is proportional to the radius r (see figure 3) [7].
Mosch M. and Grosse C.U. conducted research on a model for an automated analysis and valuation of testing situations [4]. The model utilizes the exact radius of the curvature of the geometry to calculate if a given geometry can or cannot be tested with given NDT methods. Thereby the part gets divided into a sufficiently detailed network of triangular elements before each individual element is analysed on the curvature to its neighbour. All values of the curvatures of an element are compared to one another, but only the biggest one is stored for the element. In this way, the whole part is analysed and results in a coloured picture (see figure 4) [4]. To see if an analysis of the whole part with a certain NDT method is feasible, one has to compare the maximum permissible curvature of the NDT method with the maximum curvature of the analysed part.
If the inspection of the whole part is not feasible with this NDT method, it is at least possible to display which areas of the part are testable (see figure 5) [4].
Figure 3: Calculation of the curvature, with circle of curvature [7] | Figure 4: Analysis of the curvature of a real component geometry. The colour indicates the curvatures to the neighbour elements [4] | Figure 5: Testable (blue) and non-testable (red) areas of the component geometry for a fictitious NDT method with a maximum permissible curvature of \kappa_{max} ≤ 30 [4] |