A variety of tube materials are encountered in
petrochemical plant heat exchangers. These materials may include 300 series stainless
steel, Admiralty brass, copper-nickel alloys, Hastelloy, titanium, Monel, nickel,
SA 789 (Alloy 2205), SA 268 (SS 439) ferritic stainless
steel, carbon steel and carbon steel fin-fan tubing. There is no single
non-destructive testing (NDT) technique that can be applied to inspect all of the tube materials. Selection of an NDT
technique depends on the tube material and also on defect types
expected. The NDT techniques available for inspection include conventional
eddy current, full saturation eddy current, remote field eddy current, magnetic flux leakage, ultrasonic IRIS and laser
optics. Each of the NDT technique has advantages and limitations. For
example, conventional eddy current is very sensitive to pits
and cracks but it is limited to non-ferromagnetic materials. IRIS is
accurate in measuring wall thickness, but it will miss small defects such as pin holes and cracks. Optical techniques
are limited to ID defects. Proper selection of the NDT techniques is
therefore a key to inspection of heat exchangers.
There are several NDT techniques available for inspection of heat exchanger tubing. A good understanding of the techniques is important for their selection. Figure 1 shows a schematic of four major techniques
Figure 1. NDT techniques for heat exchanger Tubing (a) Eddy Current testing (b) Remote Field Eddy Current Testing (c)Ultrasonic IRIS (d) Magnetic Flux Leakage
Conventional Eddy Current
Conventional Eddy Current technique (ECT) is based on measuring the impedance of a coil (1). The impedance of the coil changes as the electromagnetic field interacts with the material. Initially, the coil is placed in the tube and balanced on the defect free material. The probe pulled, and variations in coil impedance recorded. The impedance changes are related to the type and size of a defect. Conventional eddy current inspection is fast and can be performed at speeds up to 6 ft/sec.
Conventional eddy current is limited to non-ferromagnetic materials. These include stainless steel, Admiralty brass, copper-nickel alloys, titanium, Hastelloy, etc. The conventional eddy current inspection is done in two modes: differential and absolute. Differential mode detects small defects such as pitting and cracking while absolute mode detects tube wall loss. A mix channel detects defects under the support. This channel is a mix of two differential channels that are added to cancel out the tube support plate signal.
The first step in eddy current testing is detection and characterization of defects. Detection of defects is done either by the f180 or the f90 frequency. The f180frequency produces a phase spread of 180 degrees between the 0percent ID and 0 percent OD defect (see Figure 2). Calibration is first performed on a tube machined according to ASME section V, Article 8. The phase of the signal from the 100 percent through wall hole (calibration) is set to 40 degrees. Signals that fall below 40 degrees represent ID defects while signals greater than40 degrees represent OD defects.
The interpretation of defects is not trivial as other anomalies can produce signals that also look like defect signals. One such case is ferrite deposits in stainless steel tubes. Ferrite deposits in stainless steel tubes produce signals representative of ID defects. The phase of these signals is inthe 20 to 30 degree range, when using the f180frequency. This angle is representative of a 50 to 75 percent deep ID defect. The discrimination of metallic deposits from ID pits requires the use of additional frequencies. Data should betaken at 4 frequencies, and the signals should be checked for consistency. If the signal is from a defect, then all signals should correspond to either ID or an OD defect. If the signal is from a metallic deposit then there will be mismatch between the 4frequencies. Careful selection of the four frequencies is vital for this test.
Sizing of the defects is performed using a depth curve generated on an ASME calibration tube. This tube has 5 pit type defects and two grooves. These include one 100 percent hole, and4 OD flat bottom holes with depths of 80 percent, 60 percent, 40percent and 20 percent. A depth curve is generated by taking measurements on the five OD pits. Since there are no ID pits in the standard, a straight line is usually drawn between the 100percent hole @ 40 degrees and 0 percent @ 0 degrees. Inaccuracies in sizing result because of two reasons. The first being the depth curve which does not account for the effect of defect diameter on the phase. An increase in diameter reduces the phase angle of the defect. It is therefore not uncommon to obtain a phase angle in the 50 to 60 degree range from pin holes in titanium and 25 to 30 degrees from large diameter holes in Admiralty brass. Errors in can result if only the phase is used for sizing. The errors can be reduced if signal amplitude is used in addition to the phase for defect sizing. The second reason for sizing inaccuracy is connecting the 100 percent hole to 0 percent (at 0 degrees) by a straight line. This error is significantly influenced by the frequency. The error in sizing ID pits can be minimized by proper selection of the frequency.
Figure 2. ECT Inspection. (a) Detection of defect under a tube support plate using the mix channel (b) Calibration and Depth Curve obtained on a calibration tube. (Instrument: TC 5700)
Special ECT Techniques: Rotating and multi-coil probes
Special ECT probes are used for applications where the bobbin coils can miss defects. These may include detection of circumferential cracks in finned tubes and cracking next to the tube sheet. Bobbin coils cannot detect circumferential cracks in finned tubes because the eddy currents are parallel to the crack and are not interrupted by the crack. In such a case, multi-channel pancake coil probes should be used. A typical probe would have 4 to 8 pancake coils placed around the circumference. The coils are balanced against each other in the sound material. Any imbalance in the signal is an indication of a defect.
Another application where bobbin coil cannot be used is cracking near the tube sheet. This area is also referred as the expansion zone. Simple multi-channel coils can be used but the liftoff caused by the expansion zone produces a large signal that can mask the crack signal. In such a case, there are two options: motorized rotating pancake coil (MRPC) probe or a multi-channel send-receive probe. The MRPC probe contains a single pancake coil that is rotated as the probe is pulled slowly in the inspection area. This probe is very sensitive to the cracking in the transition zone. The second option includes multi-channel send-receive probes. One such probe design is called the CECCOprobe (2). The send-receive coils are placed around the circumference of the probe. This send-receive probe design makes it lift-off insensitive. The probe is therefore more sensitive than the differential multi-channel coil probe. The inspection speed with the multi-channel send-receive coil is significantly faster than the MRPC probe.
Full Saturation Eddy Current
The principle of full saturation eddy current is the same as conventional eddy current. The technique is applicable to partially ferro-magnetic materials such as MONEL, Alloy 2205,nickel and ferritic stainless steel or thin ferro magneticmaterials such as ferritic stainless steel. The full saturation probe contains a conventional eddy current coil and a magnet. The magnet saturates the magnetic field in the material. Once saturated, the permeability of the material drops to one; and the principles of conventional eddy current are applicable.
The main problem with full saturation technique is to assure that the material has been fully saturated. This can be confirmed by running a test on a calibration tube. Once the tube is fully saturated, it should produce a normal phase spread on the OD defects in the calibration tube. It is therefore very important to machine the tube from the same material as that of the exchanger. This is especially important in MONEL whose permeability is not fixed and can vary from batch to batch. In addition, the strength of the magnets in a full saturation probe can vary from vendor to vendor. Weaker magnets will not saturate the material and will produce a noisy signal. It is imperative to check the quality of the probe before the inspection. The application of full saturation ECT depends on the permeability of the material, thickness and tube diameter. Larger diameter tubes will allow placement of larger magnets, whereby, slightly thicker tubes could be saturated.
Sizing of OD defects is done similar to conventional eddy current. Phase cannot be used for sizing ID defects because the depth of the defect does not influence the phase of the signal. The sizing of ID pits is, therefore, done purely on the basis of signal amplitude.
Partial Saturation Eddy Current Technique
This technique is applied on ferromagnetic tubes that are too thick to be fully saturated. The technique uses a conventional eddy current instrument and monitors changes in impedance caused by changes in permeability. The permeability changes with thickness. A loss of thickness increases the intensity of the magnetic field and hence reduces the permeability. The reduction in permeability changes the impedance of the coil that is measured by the eddy current instrument.
Since this technique depends on gross changes in permeability, it is limited to variations in tube wall loss. Small defects such as pits will not influence the total magnetic field and are therefore insensitive to this technique.
Remote Field Eddy Current Technique
Remote Field Eddy Current Technique is based on the transmission of an electromagnetic field through the tube material (3). The exciter coil generates eddy currents at low frequency in the circumferential direction. The electromagnetic field transmits through the thickness and travels on the outer diameter (see Fig 1). A receiver coil that is placed in the remote field zone of the exciter picks up this field. In this zone, the wall current source dominates the primary field directly from the exciter. The separation between the two coils is between 2 to 5 times the tube ID.
The RFECT is quite effective in the inspection of carbon steel tubes. The technique however is limited to wall loss. While many attempts have been made to detect pits with RFECT using differential receivers and multiple receiver coils, the sensitivity for pit detection is limited and usually unacceptable. RFECT pit detection can be demonstrated in cleancalibration tubes, but the noise produced by rust in the heatexchanger tubes masks the signals from the pits. The inspection speed with RFECT is also significantly lower than conventional ECT. While conventional ECT can easily be performed at a speed up to 6 ft/sec, RFECT is limited to about 10 inches/sec.
Flaw sizing with RFECT is done using the Voltage-Plane curves (Figure 3). These curves are used to size tube wall loss but not pits. The curves relate flaw depth, flaw length, and the flaw circumference to the phase of the remote field signal. Inaccuracies result because the geometry of the actual flaw is not defined as in the calibration defects. Ultrasonic IRIS is therefore used to verify the RFECT measurements.
Figure 3. Voltage Plane curves used for sizing wall loss in carbon steel tubes (Instrument: TC 5700).
Magnetic Flux Leakage
This technique is based on the influence of defects on a magnetic field. The method is limited to ferromagnetic materials. The MFL probe consists of a magnet with two types of magnetic pickups: coil type and Hall element. The coil type sensor picks up the rate of change of flux while the Hall type picks up absolute flux. The coil detects small defects that cause perturbations in the flux (see Fig 1). The rate of change of flux induces an output voltage (Faradays Law) which is read bythe MFL instrument. Since the output voltage is directly proportional to the rate of change of flux, a constant pull speed should be maintained. Sudden changes in speed will induce an electromagnetic voltage, which can be misinterpreted as defects. Since the coil output is proportional to the rate of change of flux, sharp flaws produce larger signals and are more sensitive to MFL coils. In fact, the coils can totally miss long areas of wall loss if the changes in wall thickness are gradual. The Hall element sensor is used to detect gradual wall loss.
The output of the MFL coils is related to change of flux caused by the defect, but not the defect size. This technique therefore cannot be used to size flaws. A small diameter, 25percent deep pit will produce a larger signal output than a large diameter 75 percent deep pit with a gradual change in depth. In addition, rust at the ID surface will also produce noise type signals that can overshadow the defect signals. Hall Element measures the absolute flux and can be used for sizing wall loss type flaws (not pits). But the output of the Hall element depends on the orientation of the sensor in the probe and the location of the defect: ID or OD. ID defects will produce larger signals than OD defects because the field strength on the ID is higher than OD.
Ultrasonic Internal Rotary Inspection System (IRIS) is based on the principle of measuring thickness using ultrasonic waves. The IRIS probe consists of an ultrasonic transducer that is lined up in the centerline of the tube and incident on a rotating mirror. The mirror reflects the beam in the radial direction as it rotates in the tube. The IRIS probe scans the entire circumference of the tube as it is pulled out of the tube. The IRIS display includes the cross-section of the tube and a C-scan of the tube (see Figure 4).
The IRIS method is mostly used for inspection of carbon steel tubes and is sometimes used in non-ferromagnetic tubes for defect verification. The method is very accurate for thickness measurement as well as detecting ID and OD pits. IRIS will, however, miss pinholes and cracks. The method is also slow with inspection speeds limited to 3 inches/second. Because of the inability of maintaining water coupling during the entire tube length, the technique does not result in 100 percent coverage. Some areas can be missed. The inspection also requires good cleaning prior to inspection. Improper cleaning will result in un inspected areas.
One of the limitation of IRIS is the minimum measurable thickness. As the tube gets thinner, the time difference between the ID and OD signals gets smaller. This time difference reaches a limit so that the ID and OD signals cannot be resolved. The minimum level of thickness measurement depends on the tube material (ultrasonic velocity) and surface roughness of tube. In general, for in-service carbon steel tubes, thickness below 0.035inches cannot be measured. The thickness limit for new (smooth) tubes can be as low as 0.025 inches.
Figure 4. IRIS display includes the C-scan and the tube cross-section. (a) ID damage in Nickel Tube (b) OD wall loss in carbon steel tubing. (Instrument: TC 5700).
This technique uses a rotating Laser beam that scans the
ID surface as the probe is pulled out of the tube. The reflected laser beam is picked up by a lateral detector that
measures changes in proximity caused by variations on the ID surface. The
information received by the detector is processed to create an image of the ID surface. The technique is limited to ID
surface inspection with a speed up to 3 inches/sec. The technique also
requires the tube to be cleaned to avoid any unnecessary optical scattering. Because of the slow speed and cleaning
requirements, the technique is used to compliment other methods such as eddy
Selection of Techniques
The selection of a particular technique depends on the material and type of defect. Table 1. lists the tube materialsand the recommended techniques.
Conventional Eddy Current is the most suitable technique for inspection of non-ferromagnetic tubes. These include stainless steel, Admiralty brass, copper-nickel alloys, Hastelloy, titanium, etc. The inspection can be done with regular bobbin probes for the tube length and under the supports. Special probes are required for detection of circumferential cracks in finned tubes and cracks next to the tube sheet.
Partially Ferromagnetic and Thin Ferromagnetic Tubes
The partially ferro-magnetic materials include MONEL and SA 789 (Alloy2205). Thin ferromagnetic materials include the ferritic grade of stainless steel SA 268 (SS 439) also known as seacure. Full Saturation ECT is the most sensitive technique for these tubes. This technique can be applied if full saturation can be demonstrated on the calibration tube. There a maximum thickness limit for application of this technique. The thickness limit depends on the tube diameter. Larger tube OD will allow placement of stronger magnets and can therefore saturate thicker tubes. The maximum thickness limit for partially ferro-magnetic tubes is about 0.085 inches for tube OD of 0.75. The thickness limit for ferritic stainless steel tubes is about 0.030 inches for tube OD of 1.0inches. Ferritic stainless requires a stronger magnetic field to saturate the material.
Carbon steel tubes and carbon steel tubes with aluminum fins fall into this category. There are three electromagnetic techniques that can be applied for these tubes. These include remote field ECT, Partial saturation ECT and magnetic flux leakage. The first two techniques are limited to detecting large areas of wall loss that are longer than 0.5 inch. Magnetic flux leakage is sensitive to both pitting and wall loss, but cannot size the defect depth. The sensitivity of the flux leakage technique is also impaired by noise signals produced from oxides (rust) in the tube ID. In addition, MFL is more sensitive to ID defects than OD defects.
The most reliable technique for ferromagnetic tubes
is ultrasonic IRIS. Table 2 shows that the flaw detection reliability of IRIS is 83 percent compared to 77 percent
Reliability of NDT Techniques
Electric Power Research Institute (EPRI) conducted a study in1998 to measure the flaw detection performance of NDT techniques(4). The study was done on mockup samples with both service induced defects and man-made defects. The defects were representative of corrosion and mechanical wear type damage forms initiating on ID and OD sides of tubes. Seven NDT vendors participated in the study.
Table 2 provides the highest flaw detection measured by any technique. The table shows very clearly the high reliability of ECT on non-ferromagnetic tubes and the high reliability of IRIS on carbon steel tubes. However, the IRIS reliability drops to 28% on stainless steel because IRIS is not sensitive to small pits and cracks.
Table 2. Flaw Detection Performance by tube material and NDT technique.
The study also showed variation in reliability caused by differences in operator. When inspecting the same mockups using two different operators, a significant change in the reliability was measured. Table 3 shows the conventional ECT tests performed by two different operators. The results clearly show that poor operator training degrades the results. In case of stainless steel, the flaw detection dropped from 91% to 58%. These numbers clearly show the importance of training for eddy current inspectors. The training should be done on samples with service induced defects and the operators should be qualified on mockup samples.
Table 3. Flaw detection performance by ECT. Effect of operator on flaw detection
*Highest ECT flaw detection by NDE Associates,Inc. using TC 5700
There is no single NDT technique that can be applied
for inspection of all the heat exchanger tubing materials. A
multi-technology approach is, therefore, required. Using the
multi-technology approach, the most appropriate technique should be selected for inspection. Improper selection of the
technique will result in missed defects and inaccurate sizing. In addition,
technicians performing the test should be properly trained and tested on mockups. The technicians should be aware of
the limitations of each technique. Training should also include education on resolving false calls from defects and knowledge
of factors that can affect defect sizing.
NDE Associates, Inc.