Investigating and Imaging Construction Details Non-Intrusively
In order to improve our knowledge of how most effectively to use x-ray equipment to assess the interior construction of walls and to produce radiographs that display the most information, laboratory radiographs were taken with a variety of setups. These were used to identify the most appropriate approach to address many situations confronting the preservationist.
To produce radiographs that display interior details of a given wall most effectively, multiple factors must be taken into consideration. These can broadly be grouped into three categories: the geometry of the setup, using the x-ray to optimize contrast and minimize distortion, and image enhancement. The first two factors can be employed to generate the most informative radiograph possible with the equipment available and the latter can be used to improve the radiograph to see subtle features or identify the relative position and size of components of the wall system.
The geometry of the setup can dramatically influence the quality of the radiograph. To understand why, it is important to understand how components within a wall (the “object” of an x-ray, when discussed below) generate an image on a radiograph. If you think of the x-ray source as a light, then the image on a radiograph is the shadow cast by the object. All the characteristics of edge sharpness, size and distortion discussed below can be observed by standing in the bright light from a window and casting shadows on the wall opposite the light source.
The edge sharpness, or “fuzziness”, of a radiograph is partially controlled by geometry. If you have ever been in a room or outside with multiple overhead lights, you probably noticed that you cast multiple shadows. One large light source creates a shadow with fuzzy edges, in part because every portion of the light source is creating a shadow. Thus, the size of the light source can influence the sharpness of the image cast. On an x-ray source, this light source is the aperture. The smaller the aperture, the better. The XR200 has an aperture of 1/8 inch (3 mm), which minimizes the fuzziness effect. Given that the source aperture is small, but not a point, the multiple shadow problem can be improved by reducing the distance between the object and the imager and increasing the distance from the source to the object. However, as the distance from the source increases, more pulses are required to penetrate the object due to the inverse square law. In essence, an optimum distance from source to object exists for every setup. However, without knowing all of the parameters or having unlimited time in the field, it is necessary to establish guidelines that can be used as a starting point for the field work.
Minimizing the distance between the object and the imager is important for several reasons. The closer the object is to the imager, the smaller the image of the projected object is on the radiograph, as it approaches its actual size (Figure 16). Because of the shadow effect at the edges due to the aperture size and diffusion of the x-rays, the object edges appear sharpest (less fuzzy) when the imager is closest to the object. Finally, by moving the imager farther from the object, the intensity of the x-ray is diminished (the inverse square law). All three of these effects can be seen in Figures 16 and 17.
If the object is perpendicular to the line between the source and the center of the imager, its shape will show little, if any, distortion (as in the middle screw in Figure 18), but if it is at an angle (such as the screws on the right and left in Figure 18) the portion of the screw closest to the source will appear larger, creating distortion. The net effect is that the image appears to the observer as if one is looking at it from the direction of the source (i.e., the screw on the right below has the bottom coming toward the observer).
Other changes in geometry, such as having the source and imager not perpendicular to one another, can create different kinds of distortion. The three images in Figure 19 demonstrate that as well. Using the same setup as in Figure 12, a 1⁄2-inch metal grid was attached to the outer face of the 4-inch by 4-inch wood block. Then, with the source at 24 inches from the block and perpendicular to it, the center radiograph was taken. The radiographs on the left and right in Figure 19 were taken with the source moved 6.5 inches to the left and right, respectively, and angled back toward the center of the block (always pointed towards the center screw). Note that the offset angles create a different sense of whether the screws are angled (the image on the left looks almost as though the screws are simply different sizes). Further, their arrangement relative ￼￼￼￼to one another appears to be different, even though the relative position of the screws within the block is unchanged.
As the distance between the source and object increases, the energy of the pulses per unit area penetrating the object decreases as the area of the beam increases. As such, there is an optimum distance to place the source from the object to maximize resolution and coverage of the object. To address this, objects of different thickness were x-rayed at distances of 12, 24 and 36 inches between the source and the object (in all cases, the imager was placed directly against the back side of the object to maximize contrast). A six-inch thick block of wood with an embedded spike was x-rayed at these distances, using 10 pulses (pulse number is discussed below).
As seen in Figure 20, the radiograph taken at 12 inches illuminates only the central portion of the imager. The details at the edges of the object (and imager) are not visible. The radiograph taken at 24 inches illuminates the entire imager surface (although the central portion is most exposed), while at 36 inches, the radiograph becomes grainy and has lower contrast. It is also much darker before post-processing due to the reduced energy of the x-ray beam with increased distance from the object). For most building inspection applications using these systems, the optimum source-to-object distance ranges from 18 inches to 24 inches.
From the above discussion, it is clear that the optimum conditions for creating clear radiographs are when the source, object, and imager are in parallel planes and positioned perpendicular along a single line of site. The imager should be placed directly on the back side of the object, thereby minimizing the distance between object and imager and maximizing contrast on the radiograph. Contrast is discussed below. The optimum spacing between source and object, when the object is a wall, is typically between 18 and 24 inches. This spacing allows the x- ray beam to fill the imager completely without losing detail along the sides and minimizing the number of pulses necessary to penetrate the wall.
After geometric concerns are addressed, generating high-quality images requires an understanding of the final contrast produced in a radiograph. This is a function of the composition of the materials to be examined, primarily the density and thickness. Contrast is also affected by the type of radiation used to penetrate the object and scattering of the beam. With the equipment used in this research, the investigation was limited to studying the appropriate number of pulses to use for a variety of situations. The limitation was because the 150 kV source produced x-rays over a fixed range of wavelengths.
A series of tests were run using individual blocks and test walls with different numbers of pulses, placing the source 24 inches from the object. Figure 21 shows the results of over and under exposure. At 10 pulses, little of the wood grain or the boundaries between boards are visible, while at 50 pulses, only the rod has not been “washed out” in the center of the radiograph due to over exposure.
It could be argued that with post-processing, under exposure is not a problem. However, under-exposed radiographs can display loss of detail after post- processing. The six-inch block with an embedded spike, x-rayed using 2, 10, and 24 pulses is shown in Figure 22. With additional pulses comes clarity and definition of the difference between the wood block and the spike. Further, the wood features at the bottom of the radiograph are enhanced. All these x-rays were under-exposed and then post-processed.
To determine a reasonable number of pulses for solid timber of varying thicknesses, a 10-inch-thick timber was notched into a stepped block with 3⁄4, 1⁄2 and 1⁄4 of the original thickness (7.5, 5 and 2.5 inches thick, respectively). The block was x-rayed using between 16 and 100 pulses. Figure 23 presents three of these radiographs, showing the relative brightness and contrast for the various thicknesses of the stepped block.
To determine a reasonable number of pulses for a typical wall constructed of 2- inch x 6-inch lumber and containing a variety of fasteners, a test wall was built. The addition of exterior and interior sheathing and siding were added to determine the number of pulses were required to penetrate the added features.
Figure 24 shows examples of some of the radiographs taken at 24 inches from the source to the wall surface, using a range of pulses. This wall had sheetrock on the interior and cedar siding over oriented strandboard on the exterior. Although not traditional historic materials, these materials were used because of the ease of changing the layup of the wall for different tests.
Using 16 to 24 pulses, the wall components are clear in the radiograph – all the screws, nails, metal clips and wires are visible; the wood grain in the 2 x 6 studs is clear, and even the overlap in the cedar siding is visible. At 8 pulses, some of these features are difficult to see, while at 40 pulses, features in the center of the radiograph are lost due to over exposure. Figures 25 through 29 show additional configurations of this same wall.
In summary, the optimum number of pulses for most wood-frame wall constructions ranges from 6 to 30, depending on depth and wall components, unless masonry finishes are present. With a typical wood-frame wall, ranging in depth from 6 to 12 inches, initial radiographs can be taken using 10 pulses. An examination of the resulting radiograph will allow the operator to appropriately select the optimum number of pulses for the situation.
To more accurately refine the number of pulses needed, an examination of the histogram will help. The histogram is a graph showing the distribution of frequencies in an image. If the histogram is compressed to left, more pulses are required to produce optimum contrast. If it is compressed to right, use fewer.