|NDT.net - March 2001, Vol. 6 No. 03|
12th International Symposium
on Nondestructive Testing of Wood
The Process of NDE Research for Wood and Wood CompositesJozsef Bodig
Emeritus Professor, Wood Science and Civil Engineering
Colorado State University
Principal Scientist, EDM International, Inc., Fort Collins
The keynote address for NDT 2000 provides an overview of the various nondestructive evaluation (NDE) methods and their applicability to wood products. It discusses the end goal requirements of a typical NDE research and identifies the components of the project that need to be emphasized. The presentation focuses on the sonic stress wave spectral analysis approach. Based on the author’s personal experience, examples are given on five previous research projects.
It is a great privilege for me to address the NDE 2000 conference which is attended by delegates from various countries all over the world. It is also an honor and great recognition for delivering my keynote address at the country where I was born and raised and at the excellent university where I have started my college education.
The focus of my presentation is not to report on any particular research project. Rather, I would like to highlight some of the overall issues involved in the research and development (R&D). I prefer the term nondestructive evaluation (NDE) over nondestructive testing (NDT) since the former encompasses a much broader range of activities.
There are a number of issues involved in the R&D of wood and wood composites NDE. Questions need to be raised, such as, what are the goals of the research; what resources are available for the investigation; what is the current state of the art; what level of approach is feasible, and what to do with the results if the set goals are only partially accomplished. Other questions, like what is the purpose of the research: curiosity, fundamental development or applied technology have to be answered before one can formulate the scientific approach and seek funding to carry out the R&D.
The field of NDE in general and that of wood and wood composites specifically are still in the infancy of development. We have to gain much more knowledge before we will be able to predict successful outcomes in advance. Consider, for example, the widely used relationship between strength and modulus of elasticity (E) of wood products. We can expect only partial correlation between these two variables as Hearmon of UK (Hearmon, 1966) have stated nearly half a century ago that there is no theoretical relationship between strength and stiffness. This is because a localized flaw controls strength while stiffness is an expression of the integrated effect of the overall characteristics of the material.
The NDE research and development may take different approaches depending on the nature and geometry of the object or product to be studied. While the categories provided below are arbitrary, they represent major differences in an NDE approach.
We have a strong interest in determining the behavior and characteristics of a standing tree. Properties of a tree, like its growth rate, strength, internal decay, defect distribution, density variation, etc. are driving NDE research activities. While some of these properties can be estimated by taking so-called small intrusive measurements, such as cutting off branches, taking increment core samples, resistograph drilling, etc., the overall goal always should be to predict tree properties through non-intrusive NDE.
A tree is a complicated structure for NDE consideration. Its shape varies by species, age, climatic conditions and silvicurtural practices. Some of its properties fluctuate with the season. Enclosure of wood by an outside bark, presence of branches and underground roots create very difficult geometries and inhomogeneities for an NDE.
Is this tree strong enough to stand up to the next wind storm? What product would provide optimum use? Is the tree healthy enough to keep it growing or should it be harvested because of its deteriorating condition? These types of questions drive our desire to provide new NDE methods for a tree, be it in the forest, in a park or at a street edge.
While a solid wood product is somewhat less complicated geometrically and material composition-wise than a tree, it is complicated enough to give us challenges for its NDE. Solid wood may be grouped into two subcategories:
Poles and piles retain the typical circular and tapered geometry of a tree, but the bark has been removed as well as branches and roots cut. Variation of growth characteristics (growth rate, density, knots, spiral grain, checks and splits) along the length and in the radial direction creates complexities in their evaluation.
Further complexities arrive with poles and piles because they are usually dried and preservatively treated. In addition, the applications of poles and piles are quite different. Poles are used typically for utility and construction applications where the butt section is oriented in the same way as in the tree. On the other hand, piles used typically for foundation base and marine application are driven in the ground by tip first. Thus, assessing the properties of poles via NDE requires a different approach than that of piles.
Laminates are layered products with the laminae composed of wood (glued-laminated timber,plywood, etc.) or in combination with other materials (overlaid wood products, reinforced glulam, etc.).
Typically, an NDE researcher faces the following major issues with laminates:
Typically, sheet type laminates require the evaluation of different properties (e.g.rolling shear) that elongated laminates (bending, compression and tensile strength). Thus, it is not expected that a generalized NDE procedure would be applicable for all laminates.
Wood composites are generally defined as products in which wood is used in particle or fiber form and these components are bound together by an adhesive. Generally, the adhesive does not coat the wood particles entirely, hence they are only spot glued. In case of paper, while the fibers are held together by Hydrogen bonds, they are usually supplemented by wet-strength adhesives.
Among the most important factors influencing NDE of composites are:
Since wood composites are multiphase products, a researcher needs to be careful to select those NDE variables that more closely respond to the controlling variables of interest (such as the glue bond strength) while minimizing the influence of the other components. Not an easy task to accomplish.
The experimental design and the overall goals of the R&D are directly affected by the type of application envisioned. While applications can be categorized a variety of ways, I chose four categories.
Basic Research and Data Collection
This type of activity is usually conducted when there is little or no fundamental NDE knowledge. For example, an investigator wishes to know the NDE characteristics of a decay pocket in a timber pile. He or she may start by selecting a healthy round section of wood, collect a set of NDE data on this section. Then, additional NDE data are collected as a cavity is created and progressively enlarged in the wood section. Finding a relationship between the size of cavity and NDE variables can help to define the measurement needs for decayed wood.
Another approach one may employ in a pre-research experimentation is to use a variety of NDE techniques on a few selected test specimens. These test specimens usually represent extreme cases, like the smallest and largest specimens, healthy and most degraded material, etc. The effectiveness of the various NDE techniques are sorted according to their accuracy of predicting the destructive property. Then one or two of the most effective NDE methods is chosen to conduct the full investigation.
Fundamental research is needed on a variety of materials, composites and structural components. The level of evaluation can range from the quantification of gross overall characteristics to measurements related to the anatomical level. We are especially in need of defining failure mechanisms with the identifiable controlling factors. Such knowledge would narrow our search by focusing on the most promising NDE variables.
Unfortunately, quite often our NDE research may be characterized by a random search where we collect a large number of NDE variables and let the statistical level of significance dictate our choices. This is especially true with multiple regression analysis where we choose a half dozen or more variables. This approach makes future use of the system complicated and repeatability questionable.
Product Quality Evaluation
Material or product quality evaluation represents an important NDE application. The NDE method often is the quality control (QC) tool. As with most QC activities, the NDE method has to be reliable and fast to meet the production speed.
Some of the more important QC requirements of a reliable NDE method include:
In some cases the NDE of the final product is not feasible such as for a glulam beam. In these cases the component, such as the laminae, are graded by NDE prior to lamination and the lay-up organized in such a way that the expected behavior of the finished product can be established.
In-Situ Structural Components
A major category of NDE application is the evaluation of the degree of degradation of built-in structural components. Such evaluation is needed to quantify the remaining strength of the members and to make statements about the reliability, safety and life expectancy of the structure. There are several difficulties encountered with both researching and application of NDE on in-situ members. Some of the more important factors include:
Almost all NDE methods used for various materials and products can be applied to wood and wood composites. The choice depends on the specific application. In this section I am providing some observations about the applicability of the various NDE methods, i.e., my preference and bias toward them.
The simpler and oldest NDE method is the visual observation. Many evaluations today in practice still rely entirely on visual observations and the subjective interpretation of these observations.
In many situations visual observations all are needed to make decisions about the criticality of the condition of the product or component. Some of the key observations that serve this purpose include:
The visual-observation-based decisions are typically “go, no-go” types, meaning the product is rejected or judged to be suitable for continuous service. In some cases repair or reinforcement may be recommended.
Pseudo-NDE methods are often used in conjunction with visual observations. These include such methods as:
While the pseudo-NDE methods can help to expand the information obtained with visual examination, they, for the most part, are also subjective in nature. For example, an increment core or a resistograph drilling record may be used to establish the extent of a decay cavity. However, it is a much more difficult task to determine the boundary of incipient decay.
Sonic Stress Wave
Sonic stress wave is one of the most popular NDE methods used for wood and wood composites. Stress waves are generated either through an impact or by a forced vibration. Because of the difficulty of coupling a forced vibration generator to a piece to be evaluated, the majority of investigators favor the impacting method (Thomson, 1981).
While it is true that a specimen can be vibrated without physical contact by a sound source, there is a large energy loss due to this transfer method. Hence, the non-contact technique so far received only limited investigation.
The sonic stress wave method itself is usually employed in two different ways using either the speed of sound or the vibration spectrum.
Speed of sound measurement which is often used to express the dynamic modulus of elasticity is very popular with wood products. The reason for its popularity is its simplicity in instrumentation and computation. Typically, speed of sound is converted to modulus of elasticity (E). In spite of its simplicity, using E as an NDE parameter has several shortcomings:
On the other hand, using the stress wave spectrum from which NDE parameters are chosen can overcome many of the shortcomings the speed of sound approach presents (Bodig 1995), including:
Ultrasonic Stress Wave
Ultrasonic stress wave NDE is similar to the sonic stress wave approach except that it is applied at higher frequencies. The ultrasonic method is very popular with homogenous, nonporous materials for detection of flaws (Mal and Bar-Cohen 1988). While this method is also used with wood and wood composites, it is less effective due to the porous and discontinuous nature of this class of products (Beall 1987).
Ultrasonics are most effective for manufacturing quality control such as detecting delaminated areas in laminates and blows in particle composites. This is due to the ability of ultrasonic waves to be concentrated in a small area. Ultrasonics are used either as a:
In the former, the ultrasonic stress wave is sent through the thickness of the material and characteristics of the wave are recorded on the output side of the panel. On the other hand, the pulse-echo system relies on the measurement of the echo which is created by internal surfaces, such as material separation. The pulse technique is used to detect the depth where the separation occurs, such as the blow in a particle board or the depth where a timber pile is decayed.
Other NDE Methods
Time does not permit to discuss all the NDE methods applied to wood and wood composites. Here, only brief descriptions are given of those techniques the author is familiar with.
Deflection Method is employed mostly with lumber and pole type products. The deflection is measured at a safe load level from which E is computed. From here on the process is the same as with the dynamic E approach.
Electrical Properties are used as NDE parameters such as the relationship between moisture content and electrical resistance of wood. Electrical resistance is also used in the detection of decay with in-situ examinations.
Gamma Radiation is a useful tool for quantifying decay. It is also employed as a trace element for quantifying the distribution of preservatives in wood. One of the limitations of this method is the regulations associated with the use of a radioactive source.
Penetrating Radar method is in its formative stage with wood products. The method promises to be able to detect and quantify degradation at inaccessible locations.
X-ray Method is mostly used in laboratory environment or in production lines due to the bulky nature of the x-ray source and the measuring equipment.
To set up an NDE research project the investigator has to justify the need for such a project. Aside from scientific curiosity, one has to establish that there is a problem requiring an NDE research. In addition, it has to be shown that the currently used approach does not provide acceptable reliable answers, or is it too costly to use it.
The researcher needs to answer positively at least the following questions:
Setting up a Research Project
The first and most critical component of setting up a research project is the establishment of its objectives. The objectives need to be stated in such a way that at the completion of the project one can determine whether the objectives were accomplished or not. Unfortunately, there are cases where the research project is declared successful, even though in reality its objectives were not accomplished. In these cases the closing conclusion often states that more research is needed. If such is the case, that conclusion should have been known before the research project started.
Another important issue with the research project is its scope. Often, to gain funding, the investigator promises to solve all the issues associated with a given topic. In reality such a lofty goal is never achieved as is physically seldom possible. Instead, the scope of the work has to be stated in clear specific statements.
The scope of research should state not only what will be accomplished, but just as importantly, what will not be considered. There should be a clear understanding of what factors will be evaluated and what limits will be placed on these factors. For example, the scope of work may state that the effect of temperature and moisture content will be quantified (not just studied or evaluated). However, the temperature range that will be quantified is limited to above freezing and the moisture content will be below fiber saturation point. Such specific scope limitations will not create undeliverable expectations on the part of the funding agency and it will be much easier to show that the scope of the work is accomplished.
It is important that the investigator has a clear vision of the entire flow of the planned research. The specific research steps need to be established and conclusions drawn at the completion of each step to determine if the project is proceeding according to plan. These periodic progress evaluations should be done in such a way that, if needed, the remaining research steps can be modified to accomplish the objectives.
Logic would dictate that if it becomes evident that the original objective of the research cannot be accomplished, the project be stopped and the remaining budget returned to the funding agency. However, in reality such a drastic action is not known to be practiced and the investigator begins modifying the original objectives to justify continuing the research. This is the case even though funding may not have been provided if the modified objectives were known prior to the establishment of the project.
Experimental design can make or break a research project. The nature of the experimental design itself is dictated by the objectives and scope of the project. The experimental design serves two fundamental purposes:
The extent of the data collection for quantifying NDE parameters and their influencing factors require a much larger effort than that needed for proof-of-concept. Key elements of the experimental design for quantification purposes should include as a minimum:
Proof-of-concept experimental design itself may have two different goals. One of the goals would be to determine whether a new concept is applicable for the material and variables of interest. Another goal may be to determine if the concept already known to be applicable to a given material and range of variables is applicable to a wider range of variables.
In either case, the data collection may be limited to the extreme values of each variable considered. However, it is important that data also be collected at the mid-range of each variable to establish nonlinearity of the relationship. Once the data collected, typically the measured destructive property values, are compared to the corresponding predicted (theoretical) values. Then conclusions are drawn about the applicability and the errors associated with the NDE method evaluated.
The final step of a research project is to draw conclusions about its success, failure, usefulness and applicability. It is important that the conclusions precisely state what has been accomplished. Some of the more important components of the conclusion should include:
It is just as important to identify which of the objectives were not accomplished and which of the variables are not significant, than to list those that are. It is much more desirable to clearly define the limitations of the results than state broad claims that cannot be substantiated by later independent verifications.
A researcher may be satisfied to produce a report on the conducted NDE research. However, eventually a question is raised whether the results could be used to produce a tool that can be put to practical application. While the researcher may not be the one who produces the NDE tool, he or she should be familiar with the questions raised by the one who will consider the feasibility of such an undertaking. Some of the more important elements of the feasibility evaluation are discussed in the next sections.
While a researcher usually claims great success and wide applicability for the developed NDE method, an impartial evaluation is needed to realistically assess the researcher’s claims. The evaluator will compare the new information to previous knowledge and assess the extent the new method is superior to those already in use and may ask questions like:
A negative answer to any one of the above questions may terminate the feasibility study before the economic feasibility is considered.
If the technical feasibility is judged favorable or acceptable, the entity interested in the commercialization of the NDE equipment and/or the method will proceed to the economic feasibility study. After defining the technical capabilities and potential applications, the interested commercialization entity will likely conduct a market survey. Some of the questions of the market survey are typical of such an undertaking, others are specific to the NDE itself. The process usually includes the following steps:
Producing a Commercial NDE Product
Once technical and economic feasibility’s established, the next step is to produce a commercial NDE device. Assuming availability of finances, the process usually takes the following major steps:
Typically, the product will likely undergo additional modifications as users will raise questions and report problems with the device or the NDE method. Eventually the device becomes outdated or competing products appear on the market. At that point the device may be withdrawn or major modifications made to it.
Over forty years of professional activities as a wood scientist, I had the opportunity to be involved in several NDE projects. My experience has been limited to the field of mechanical properties. In an attempt to support some of the points I made earlier, I am now providing a few examples from these projects.
In-situ Strength Evaluation of Utility Pole
Research conducted on the strength evaluation of utility poles in service resulted in a commercial device called “PoleTestTM”. Figure 1 shows the setup of this device on a utility pole for in-situ NDE.
|Fig 1: “PoleTestTM” NDE Device Used to Estimate the Remaining Bending Strength of Wood Utility Poles in Service.|
The principle of operation involves the impacting of a nail head, embedded in the wood by an impactor to create a transverse stress wave. Sensors next to the nail and on the opposite face record the stress waves as a time function. These time records are converted to frequency spectrums and specific characteristics of these spectrums are used through multiple correlation’s to provide the best estimate of the remaining bending strength of the pole.
The relationship between strength and NDE parameters are based on a large number of full-scale test data. Several thousands of poles were evaluated first by NDE then by destructive testing. Figure 2 shows the flow diagram which explains the operation principle of PoleTestTM (EDM 1995).
|Fig 2: Flow Diagram of PoleTestTM Operation Principle|
After turning the instrument on, the keyboard is used to characterize the test pole by species and location of the NDE. Through each impact the NDE signal is converted to specific NDE variables which in turn are entered into the mathematical model used to predict the modulus of rupture at the break point (MORBP) which is given the same location as the NDE. After adjustments for pole diameter and height fraction (location of NDE along the length) an MORBP value is obtained. The result of three impacts are averaged and compared to two other sets of three impacts. If the three sets of average MORBP values are less than five percent apart their overall average is displayed as the predicted results in a larger than five percent variation an error message is displayed on the screen.
Strength of Cooling Tower Columns
The interior of industrial cooling towers is usually constructed of timber. One of the major components of these structures is wood columns usually of 89 by 89 mm in cross section and connected end to end up to ten story heights. Because the cooling tower environment can easily degrade any material, periodic evaluations are conducted for these columns to determine their remaining strengths and life expectancies.
To assist in the determination of the remaining strength we have developed an NDE method which can be used for in-situ measurements (Bodig and Pandey 1996). As shown in Figure 3, an nstrumented hammer is attached to the segment of column of interest that impacts a pin embedded in the wood. About one meter away an accelerometer is attached to the opposite face of the column. A signal conditioner provides the amplification of the signals obtained through the instrumented hammer and the accelerometer. The time records are stored in a computer file.
|Fig 3: Schematics of the In-situ Column NDE Test Setup.||Fig 4: Relationship Between Actual and NDE-Predicted Crushing Strength of Redwood Columns.|
As shown in Figure 4, a very good correlation exists between actual and NDE-predicted crushing strength of columns. The prediction has a smaller error at the lower strength, a desirable feature where the primary purpose of the NDE is to identify the weakest columns. The NDE also provides data enabling the computation of modulus of elasticity and hence, the remaining buckling strength of the column.
Strength of Finger-jointed Lumber
An in-line NDE methodology is needed to accept or reject each finger joint as it is manufactured. We attempted to develop such a method (Anthony 1992). Because the location of the evaluation needs to focus on the finger joint itself, the ultrasonic range of stress wave was judged to be the most suitable method for the NDE method.
Figure 5 shows the locations of the pulsers and the receivers in the finger joint are. The idea for the locations of the pulsers and sensors were that we needed to consider not only the characteristics of the finger joint but also those of the two pieces of lumber composing the joint. As shown in the figure, pairs of measurements were taken at four locations. To avoid interference only one pulser and one receiver were operated at a time.
|Fig 5: Location of Pulsers and Receivers for the NDE of Finger Joints|
The signals from the ultrasonic transducers were processed and NDE parameter ratios were established between those passed through the finger joint and those passed through the lumber. Then, these ratios were correlated to the actual tensile strengths of the corresponding finger jointed pieces. The resulting multiple regression equation was used for the prediction of the tensile strength of a particular finger-jointed lumber.
Figure 6 depicts how the developed ultrasonic NDE may be used for quality control purposes. The figure illustrates the quality control process where the strength of the finger joint decreases with increasing void size. Such decrease in strength may also be the result of a dulling knife, aging adhesive or any other processing variable. The issue is that with such an information at hand during production, the manufacturing would be stopped before the process gets out of control.
|Fig 6: NDE of Finger-Jointed Lumber as a Production Process Quality Control Tool.|
Strength Classification of Utility Poles
Currently in North America utility poles are sorted visually into pole classes. While a specific bending strength is assigned to each class, the actual strength of poles within a given class is highly variable. An NDE method was developed where an instrumented hammer impacts a screw embedded in the butt of the pole. A sensor at the tip of the pole senses the vibration transferred through the entire length of the pole (Bodig and Pandey 1997). Figure 7 shows the setup of the instrumented hammer.
|Fig 7: Pendulum Impactor Used to Deliver a Stress Wave to a Utility Pole.|
The NDE-predicted tip load can be used to assign a strength class to each pole. This is done by setting a lower five percent exclusion limit value for each class. Using this lower limit, 95 percent of the poles will have tip load capacities higher than the set limit. Figure 8 shows an excellent agreement between actual and strength-class assigned tip load for southern pine utility poles.
|Fig 8: Comparison of Actual and NDE Assigned Strength Class Tip Loads for Southern Pine Utility Poles.|
Quantification of Biodegradation
We joined other attempts to develop an NDE methodology whereby the presence and extent of biodegradation could be quantified (Bodig 1995). We added an additional constraint to the objectives since we wanted to develop the technique in such a way that it could quantify biodegradation at inaccessible locations.
Since in exterior use we cannot control the environmental variables, such as temperature and moisture content, for in-situ measurements, we needed to develop an NDE which can quantify the biodegradation while at the same time is insensitive to temperature and moisture variations.
In a general form, the vibration signal output, V, is assumed to be a function, f, of a number of major material and environmental variables. In mathematical form:
|V = f (D, B, M, T, G, C, …..)||(1)|
The vibration signal itself can be expressed by a number of parameters such as:
|V = f (DF, AE, SS, DE, …..)||(2)|
By equating the above two expressions and through the method of isolation, the biodegradation itself can be related to specific selected vibration spectral parameters, e.g.:
|B = f (AE, DE, ….)||(3)|
In the final form the biodegradation is quantified by a decay ratio, D. The decay ratio is defined as the fraction of decay occupying the cross section. Based on the above outlined criteria the final form of the mathematical prediction model chosen was:
|D = a+b (P/S) + c (G)||(4)|
The above decay ratio is computed at several orientations at a given cross section. This way the decay location can be mapped to scale. Figure 9 shows the NDE data collection system. Note that the process can be repeated five more times rotating the hammer and the sensors at 60 degrees each time.
|Fig 9: Schematic of the NDE Test Setup for Quantification of Biodegradation: I=Impact Pin, H=Instrumented Hammer, S=Vibration Sensor|
The decay ratio approach gives excellent results between actual and NDE-predicted decay ratios with a slope near unity (1.016), R2 = 0.743 and a standard error of estimate (SEE) of 0.164. This means that two out of three times the error in the predicted decay ratio is less than 16.4 percent. Mathematical prediction models were also developed for rectangular cross sections and for locations as far as 600 mm from the plane of NDE measurement.
|© NDT.net - firstname.lastname@example.org|||Top||