Figure 1-3
Table 1-2

Introduction
The US Army Research Laboratory Weapons and Materials Research Directorate (ARL-WMRD) undertook the development of a screening test program designed to alleviate the subjectivity of testing various types of chemical cleaning agents used prior to liquid penetrant (PT) testing. This paper presents the test methodology as well as the benefits and drawbacks of utilizing the method. Data were acquired from eight prospective alternatives to a control cleaner, methyl ethyl ketone (MEK). The results demonstrated the test method's unique ability to separate the potential alternatives by quantitatively assessing their performance in the PT process.

---

The false positives of the equipment were easily recognizable and therefore interpretable.

---

Background
Liquid penetrant testing is widely accepted as a simple, quick, low cost method for nondestructively testing components for the presence of anomalies, either inherent or service induced. Due to its intrinsic low tech simplicity, operator complacency can arise from utilizing long standing specified procedures. Ever increasing environmental restrictions and regulations continue to force change throughout PT processing but mainly on the front end, in precleaning and in the most common application, fluorescent penetrant testing. To the NDT operator, a change in the pretesting cleaning process is virtually undetectable in terms of the inspected components' visual appearance. However, it has been widely documented that some cleaning solutions have detrimental effects on the detectability of some anomalies. The newly implemented cleaning solutions are typically not even in the same chemical family as those they are replacing. As such, they have varying effectiveness on the contaminants being removed in addition to special requirements for their processing procedures. More specifically, silicated aqueous cleaners have been widely scrutinized (Gooding and Whitehouse, 1996; Rummel, 1998; Grendahl and Champagne, 1998). The end user is faced with the hardship of being forced into change either financially by penalties or by complete environmental restriction in the case of 1,1,1 trichloroethane (TCA). The last ten years have seen a bombardment of new and improved cleaning solutions claiming to be the best ever, but these are usually no improvement at all and at times are harmful to the substrates. The materials compatibility data are most often lacking due to the investment expense.

Industry components, contaminants and processes vary, but they all have a common objective with NDT. They are all looking for anomalies. This objective places many requirements on the cleaning solutions that have been widely ignored. The solutions must clean the individual user's contaminants from the surface, they must have the ability to remove the contaminants from cracks and they must not mask or obscure a crack from being detected. It would simply be impossible to test every new environmentally friendly cleaning solution by utilizing the probability of detection (POD) method each time a salesman came to the door. Make no mistake, there is nothing wrong with the method itself, it is merely time consuming, requiring many repetitions, different operators and different users to draw correct conclusions. While this method is quite good at establishing confidence in one set of procedures, it is less efficient at quantitatively screening the respective PT performance of many cleaning solutions. This inefficiency prompted the US Army to develop a quantitative performance based test procedure for screening cleaning solutions while comparing them to the accepted method. Material compatibility and mechanical cleaning techniques were obviously an integral part of the screening process but they will not be discussed in this work.

 

Method
Establishing basic parameters and variable minimization were the first steps toward testing the feasibility of an alternative technique for screening liquid penetrant precleaning processes. A certified inspector typically interprets indications. Since there can be considerable variation in interpretation from inspector to inspector, a calibrated, precise light photometer was utilized to detect the brightness of the known indications achieved from each prospective cleaning solution or process, thereby making the fluorescent penetrant testing results quantitative rather than qualitative and removing human subjectivity. The most difficult indications to detect are the smaller cracks and discontinuities. They are also the most demanding with which to achieve reliable and consistent liquid penetrant results. Therefore, testing was performed on fatigue crack induced bars fabricated from the Ti-6Al-4V and Inconel 718, two materials known to develop tight anomalies. Cracks were generated between 0.5 and 1.5 mm (0.02 and 0.06 in.) in length. The cracked test specimens were prepared by machining a specimen geometry that could be fitted into a fatigue testing machine and cycled until a specific crack size was obtained. Calculations were made to determine the proper loading configuration prior to fatigue. In order to quantify the exact length and depth of the crack, a number of cracks had to be opened and measured. Measurements of crack depth and length were determined by fractographic analysis and scanning electron microscopy (SEM). The actual fatigue cracked region would contain transgranular morphology that was easily distinguished from that of the laboratory induced overload, which would yield ductile dimpled topography. Once a particular crack size was attained repeatedly, as confirmed by opening the crack and examining the fracture surface, the exact parameters of the fatigue test were noted and repeated. These test specimens and a set of commonly employed NDT specimens referred to as TAM panels (testing and monitoring panels - thin chrome plated panels indented from the backside to produce small starburst crack patterns, which vary in length, on the front) were utilized. Other standard NDT panels have not been shown to be discriminating enough to resolve the differences in the solutions. In other words, it doesn't take precision cleaning to reveal large open cracks. All test specimens were subjected to an application of a baked on standard contaminant typical for the military rotary aircraft industry. The contaminated specimens were then cleaned with varying test solutions and processes and passed through a liquid penetrant testing line, the last stage of which contained the quantitative data acquisition equipment. Geometric variables were specified and held constant during data acquisition. Results were obtained from several prospective cleaning solutions as well as a control group cleaner, which employed cleaning with methyl ethyl ketone (MEK). Figure 1 depicts a schematic of the data acquisition system. The light photometer records light in the units of foot-lamberts that are equivalent to approximately 10.8 lm/m² (9.1 lm/yd²).

With an ultraviolet meter oriented as shown in Figure 2 and the center of the blacklight beam focused on the blacklight meter panel, the ultraviolet light intensity was 1.54 lm/m² ± 0.007. While the parameter for the UV intensity is considerably lower for standard fluorescent penetrant testing, it was found to yield data discriminating enough to reveal the tightest, smallest cracks and still separate the cleaners utilized. One must remember that a typical NDT inspector will not be looking for known cracks on test panels. The optimum cleaning solution is the one that provides the clearest, brightest indication with the lowest possible background fluorescence.

The fluorescent penetrant testing procedures utilized closely followed those outlined in ASTM E-165, 1209 and 1210. Dwell, rinse, dry and development times, as well as temperatures and pressures were all specified and closely monitored. Standardizing the processing stages of fluorescent penetrant testing further eliminated variables within the test method. While great pains were taken to eliminate variability from the processing parameters, day to day deviations still occurred. These unintentional changes, that were largely environmental factors, forced the processing of the control group solution every day. In that light, each test solution or process would have a quantitative comparison to the control group with the same set of processing parameters. Two common fluorescent penetrant testing processing methods, Method D hydrophilic post emulsifiable, and Method A water washable, were employed for this study.

 

Results
Table 1 and Table 2 outline the fluorescent brightness data obtained from the various cleaning processes' performance under fluorescent penetrant testing. NR represents "no reading" due to the indication not being revealed during fluorescent penetrant testing. An asterisk after a value denotes that the actual brightness values were questionable. Some of the aqueous cleaners left a diffuse or speckled pattern rather than clearly defined crack pattern on the TAM panels. This obscuring of the indications is thought to be directly related to the silicated compounds within the solutions. Figure 3 shows a schematic of the phenomenon. While the photometer projects this as a false high indication of brightness for these phenomena, the cracks themselves were dimmer and it is clearly detrimental to observing an indication.

 

Data Interpretation
While the absolute values of the data given are relatively unimportant, as subjectivity will again arise from inspector to inspector whether or not an indication will be seen or missed, the differences from the control cleaner demonstrate much more importance. If a cleaning solution provides clear bright indications with a quantitative light value comparable to or better than a control cleaning solution that is inherently environmentally unfriendly, the case to make the change can easily be made. It can clearly be seen from the data that cleaner 4, a normalized propyl bromide cleaner, demonstrated clear bright indications comparable to or better than the control group in almost all cases. Most of the other test solutions and processes failed to do so.

While removing human subjectivity from the testing process has benefits, the strict adherence to measuring brightness by acquisition equipment was revealed to be somewhat misleading. Certain cleaning solutions left difficult to remove residues that led to diffuse or speckled patches of fluorescence around cracks or other indications that would normally be passed over by an inspector. While this result should be construed as a failure of the cleaning solution, the light photometer actually records the increased brightness of the area that in some cases is comparable to the brightness of a crisp, clear indication resulting from proper cleaning. In other words, it was demonstrated that some of the residues left behind which mask indications actually trap and hold penetrant. This should not be mistaken as a good result. While masked indications may sometimes fool the machine, they would not be picked up by a typical inspector. In a sense, this was beneficial revelation. The equipment sometimes yields a false positive, however, we still have the human mind to interpret the data and recognize the false positive result. Therefore, the cleaning solutions that performed poorly in this manner would still be screened out. Additionally, there would exist a quantitative explanation for why one solution performs better or worse than another or versus a specific control cleaner or process. For the cleaners that did performed well there exists direct quantitative data for brightness. With the assumption that brighter is better, as long as an indication is clear and sharp, it was relatively simple to sort out the top performers in a timely and cost efficient manner.

The TAM panels were affected to a greater degree by the residues left behind from some cleaning solutions. It was assumed that this was linked to the panels' chrome plated finish that was much smoother than the typical surface finish specified for actual components or the fatigue crack induced specimens.

 

Conclusions
The Army Research Laboratory developed the test method described above with the intention of using it as a screening tool for cleaning solutions used prior to fluorescent penetrant testing. The results demonstrated the effectiveness of the method for removing the subjectivity experienced when typically testing new cleaning solutions. Although some fundamental flaws with the absolute value of the data acquired were observed, the false positives of the equipment were easily recognizable and therefore interpretable. The results also demonstrated that actual tangible data could be derived from the screening tests to make an argument for or against the replacement of an environmentally unfriendly solution with a prospective alternate. The Army Research Laboratory believes this test method to be of considerable use to the end users of chemical cleaning solutions who are constantly faced with the testing of new and improved products.

 

References
Gooding, Chris and Kim Whitehouse, "Effects of Pre-Cleaner Contamination on Penetrant Inspection Capability," Engine Titanium Consortium Open Forum, May 8-10 1996, Oakland, California.

Grendahl, Scott and Victor Champagne, "Alternatives to 1-1-1 Trichloroethane Prior to Adhesive Bonding and Non-Destructive Inspection (NDI)," Army Research Laboratory - Special Report, February 1998.

Rummel, Ward, "Cautions on the Use of Commercial Aqueous Precleaners for Penetrant Inspection," Materials Evaluation, Vol. 56, No. 8, August 1998, pp. 950-952.

 

 

* US Army Research Laboratory, Weapons and Materials Research Directorate, Aberdeen Proving Ground, MD 21005-5069; (410) 306-0819; fax: (410) 306-0806; e-mail <sgrenda@arl.army.mil>.

Copyright © 2001 by the American Society for Nondestructive Testing, Inc. All rights reserved.

[ Materials Evaluation ]