Magnetic sensor for nondestructive evaluation of deteriorated prestressing strand

A Thesis entitled Magnetic Sensor for Nondestructive Evaluation of Deteriorated Prestressing Strand by James D Wade Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Master of Science Degree in Civil Engineering Dr Douglas Nims, Committee Chair Dr Vijay Devabhaktuni, Committee Member Dr Brian Randolph, Committee Member Dr Patricia R Komuniecki, Dean College of Graduate Studies The University of Toledo May 2010 Copyright 2010, James David Wade This document is copyrighted material Under copyright law, no parts of this document may be reproduced without the expressed permission of the author An abstract of Magnetic Sensor for Nondestructive Evaluation of Deteriorated Prestressing Strand by James D Wade Submitted to the Graduate Faculty in partial fulfillment of the requirements for the Master of Science Degree in Civil Engineering The University of Toledo May 2010 The objective of this thesis was to develop a non-destructive in-situ magnetic technique to investigate the remaining cross-sectional area of prestressing strands Corrosion is the predominate failure-mechanism in box-beam bridges The current method, visual inspection, is not sufficient as the strands may not be exposed to the investigator An inaccurate estimate of the remaining area of strands can lead to an overestimated strength of the bridge A new technique involving an electromagnet and magnetic theories was researched for this thesis The experiments conducted have shown that it is possible to distinguish between different cross-sectional areas using an electromagnet and Hall sensors Experiments with an air-gap were first used to simulate concrete cover and provide viability of the technique These experiments showed that as the cross-sectional area increased so did the induced magnetic field To further the research, concrete blocks were used in place of the air-gap to better simulate field conditions Again, these experiments showed an increase in the induced magnetic field as the cross-sectional area was increased Using the data from the air-gap iii and concrete block experiments, an approach to determine the cross-sectional area of a corroded strand under concrete cover was investigated iv Acknowlegements I would like to thank Dr Douglas Nims for his guidance and engineering expertise during the production of this research and paper, Dr Vijay Devabhaktuni and Bertrand Fernandes for their contributions in establishing the electrical aspect of this research, and Dr Brian Randolph for being on the thesis committee and providing valuable feedback on the research problem and solution In addition, I would like to thank the US Department of Transportation for funding this research through the University of Toledo-University Transportation Research Center and Dr Gottfried Sawade from the University of Stuttgart, Germany for sharing information about the electromagnet used in his research In addition, the design engineers from Ohio Magnetics, Sophie Zaslavsky and Paul Sheridan were extremely helpful and patient during the design and procurement process of our final electromagnet I would also like to thank MMIStrandCo for their donation of prestressing strand and the Ohio Department of Transportation for allowing me to hold my presentation at their facility in Bowling Green, Ohio Finally, I would like to thank my family for their support and encouragement throughout the entire process v Contents Abstract iii Acknowledgements v Contents vi List of Tables ix List of Figures x Introduction 1.1 Overview 1.2 Problem Statement 1.3 Research Objectives 1.4 Research Theory and Approach Literature Review Preliminary Experiments 13 3.1 Background 13 3.2 Procedure 14 3.2.1 Trial 15 3.2.2 Trial 16 3.2.3 Trial 16 3.2.4 Trial 18 3.2.5 Trial 19 3.2.6 Trial 20 3.2.7 Trial 21 vii Ohio Magnetics 24 Secondary Experiments 34 5.1 Solid Steel Round Experiments 34 5.2 Prestressing Strand Experiments 38 5.3 Estimating Cross-Sectional Area Loss 40 Conclusion 48 6.1 Results 48 6.2 Future Research 50 References 51 Appendices Appendix A vii 52 List of Tables 3-1 Trial experimental data 15 3-2 Trial experimental data 16 3-3 Trial experimental data 17 3-4 Trial experimental data 18 3-5 Trial experimental data 19 3-6 Trial experimental data 20 3-7 Trial experimental data 22 3-8 Brief summary of all trial experiments 22 4-1 0.125 inch thick specimen results 27 4-2 0.1875 inch thick specimen results 28 4-3 0.375 inch thick specimen results 28 5-1 Gauss reading for 1018 steel rounds tested at various air-gaps 35 5-2 Gauss reading for 1018 steel rounds tested at various concrete gaps 37 5-3 Gauss reading for prestressing strand tested at various air-gaps 39 5-4 Gauss reading for prestressing strands tested at various concrete gaps 39 5-5 Results from corroded strand experiments 43 5-6 Estimates of remaining cross-sectional area of samples and at 15 16 " concrete gap viii 45 List of Figures 1-1 I-70 collapse 1-2 I-70 collapse 1-3 Corroded tendons 1-4 Hall Effect principle 1-5 Basic block diagram of Hall Effect sensor 3-1 Test equipment set-up 13 3-2 Specimen position for trial 14 3-3 Trial area vs field strength with linear fit equation 15 3-4 Trial area vs field strength with linear fit equation 16 3-5 Trial area vs field strength with linear fit equation 17 3-6 Trial area vs field strength 19 3-7 Trial area vs field strength 19 3-8 Trial area vs field strength 20 3-9 Trial set-up 21 3-10 Trial set-up 21 4-1 Ohio Magnetics test set-up 25 4-2 Air-gap vs flux density 29 4-3 Area vs flux density 29 ix 4-4 Flux lines through longitudinal axis of specimen 31 4-5 Magnetic domain walls 31 4-6 Domain growth curve 32 4-7 Ohio Magnetics electromagnet 33 5-1 Test set-up for 1018 steel rounds with air-gap 35 5-2 Gauss reading vs cross-sectional area for 1018 steel rounds with air gaps 36 5-3 Test set-up for 1018 steel rounds with concrete gap 37 5-4 Gauss reading vs cross-sectional area for 1018 steel rounds with concrete gap 38 5-5 Typical prestressing strand section 38 5-6 Gauss reading vs cross-sectional area for prestressing strands with air-gap 5-7 39 Gauss reading vs cross-sectional area for prestressing strands with concrete gap 40 5-8 Test set-up for corroded strand 41 5-9 Separation of wires in corroded strand 42 5-10 Example calculation of estimating remaining cross-sectional area 44 5-11 Magnetization curve for iron 47 A-1 Drawing of electromagnet design by Ohio Magnetics 54 x Gauss Reading (G) 1018 Steel Rounds with Concrete Gap 2400 2200 7/8" Gap 2000 1800 1/8" Gap 1600 1/4" Gap 1400 5/16" Gap 0.0000 0.1000 0.2000 0.3000 0.4000 0.5000 15/16" Gap Area (sq in.) Figure 5-4: Gauss reading vs cross-sectional area for 1018 steel rounds with concrete gap 5.2 Prestressing Strand Experiments The next experiments were conducted in the same manner as those described in 5.1, with the exception that prestressing strand was used in place of the 1018 steel rounds A typical section of prestressing strand can be seen in figure 5-5 The results from the air-gap and concrete block experiments can be seen in tables 5-3 and 5-4 and figures 5-6 and 5-7, respectively Once again, the results showed that the magnetization increased as the cross-sectional area of the specimens increased Figure 5-5: Typical prestressing strand section (www.ivysteel.com) 38 Table 5-3: Gauss reading for prestressing strand tested at various air-gaps Air Gap (in.) Diameter (in.) Area (in ) 3/8 0.0850 7/16 0.1150 1/2 0.1530 3/5 0.2170 1/8 3670 4250 4490 5310 1/4 2850 3290 3560 4070 1/2 2060 2200 2360 2650 3/4 1740 1830 1940 2090 1630 1690 1760 1880 1/4 1530 1580 1620 1711 1/2 1495 1533 1572 1638 3/4 1453 1478 1512 1566 1415 1436 1464 1503 Table 5-4: Gauss reading for prestressing strands tested at various concrete gaps Concrete Gap (in.) Diameter (in.) Area (in ) 3/8 0.0850 7/16 0.1150 1/2 0.1530 3/5 0.2170 7/8 1733 1820 1910 2060 1/8 1595 1645 1700 1790 1/4 1560 1605 1648 1732 5/16 1540 1590 1637 1712 15/16 1462 1480 1505 1546 Prestressing Stand with Air Gap 5900 5400 Gauss Reading (G) 4900 1/8" Air Gap 4400 1/4" Air Gap 3900 1/2" Air Gap 3/4" Air Gap 3400 1" Air Gap 2900 1/4" Air Gap 2400 1/2" Air Gap 1900 3/4" Air Gap 2" Air Gap 1400 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 Area (sq in.) Figure 5-6: Gauss reading vs cross-sectional area for prestressing strands with air-gap 39 Gauss Reading (G) Prestressing Strand with Concrete Gap 2250 2050 7/8" Gap 1850 1/8" Gap 1650 1/4" Gap 1450 5/16" Gap 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 15/16" Gap Area (sq in.) Figure 5-7: Gauss reading versus cross-sectional area for prestressing strands with concrete gap 5.3 Estimating Cross-Sectional Area Loss As discussed in chapter 4, it was believed that with more data and a better magnet an equation to estimate the remaining cross-sectional area of a strand could be developed However, due to many variables, such as magnetic permeability and metallurgy, an equation was not obtained during the time frame of this thesis Nevertheless, another, more empirical approach for estimating the remaining viable cross-sectional area was developed This approach was developed during and after experiments with corroded prestressing strands Corroded strands of a ½” diameter from a box-beam bridge were 15 tested with a " and 16 " concrete block gaps Only ½” diameter strand was tested 40 because it was all that was available in a pre-corroded condition Two samples were tested and readings were recorded every inches along their length at the right pole Wires in sample were removed to create regions of increased cross-sectional area loss The test set-up for these experiments can be seen in figure 5-8 Observations of the strands along their length were recorded as follows: Figure 5-8: Test set-up for corroded strand 1) Sample i) From 4"-12" wires were visibly present ii) At 16" wires were visibly present with separation between the wires (see figure 5-9) iii) At 20" wires were visibly present iv) At 24" 4-5 wires were visibly present v) From 28"-32" 4wires were visibly present vi) At 36" there is a transition from to wires being visibly present vii) At 40" wires were visibly present 41 2) Sample i) At 4" wires were visibly present ii) From 8"-16" wires were visibly present iii) At 20" wires were visibly present iv) At 24" wires were visibly present v) From 32"-36" wire was visibly present Figure 5-9: Separation of wires in corroded strand In addition to the observations previously listed, the strands were severely corroded as can be seen in figures 5-8 and 5-9 The results from these experiments can be seen in table 5-5 42 Table 5-5: Results from corroded strand experiments (measured in Gauss) Distance Along 1/4" Concrete Gap Longitudinal Axis (in.) Sample Sample 1670 1540 1690 1570 12 1670 1590 16 1620 1540 20 1590 1510 24 1550 1500 28 1590 1490 32 1530 1490 36 1520 1450 40 1470 15/16" Concrete Gap Sample Sample 1430 1390 1440 1400 1430 1400 1420 1400 1410 1380 1410 1360 1410 1350 1390 1340 1390 1320 1360 The first instinct when estimating the remaining viable cross-sectional area was to measure the diameter of the strand at the points of interest with calipers This, however, was impractical because the strands had expanded during the corrosion process, and in the field, the strand will be embedded in concrete, thus, making it impossible to measure the diameter with calipers To overcome these issues, a healthy (non-corroded) strand was dissected and separated into its seven individual wires From previous results, a healthy ½” diameter 15 strand produced a 1505 Gauss (G) reading at a 16 " concrete gap Again, first instinct was to assume that this would be distributed evenly over the wires (215 G per wire), 15 however, when one wire was placed on the magnet at a 16 " concrete gap, a reading of 1350 G, nearly 90 percent of the total, was observed As each wire was added to the magnet, an increase of 20-25 Gauss was observed for each wire added In addition, it was observed that wires were not always attracted to each other, but in most instances repelled one another This is believed to be due to how the magnetic domains in each 43 wire align themselves in the field The orientation of one wire to another produces a give and take affect that results in a net magnetization of 1505 G when they are all together With the data and observations from the experiments, an approach to estimate the remaining viable cross-sectional area was proposed The proposed approach used the known readings of a healthy strand and that of one wire taken from a healthy strand at a given gap An example of calculations to determine the remaining viable cross-sectional area of a strand can be seen in figure 5-10 The example was calculated in MathCAD Bstrand := 1505G Magnetization for a healthy 1/2" strand at a 15/16" gap Bwire := 1350G Magnetization for a healthy wire taken from a 1/2" strand at a 15/16" gap B := Bstrand - Bwire = 155  G B M per.wire := = 25.833  G Net magnetization per wire for the remaining wires Magnetization at a point on a corroded strand M corroded := 1390G As := 0.153in Area of steel in 1/2" diameter strand As Awire := = 0.022 in Area per wire in a 1/2" strand wire := (Mcorroded - Bwire) + = 2.548 M per.wire Estimate of total viable wire left The +1 wire accounts for the wire that was not used to determine the net magnetization per wire Acorroded := Estimate of remaining viable cross-sectional area of corroded strand  100  = 36.406 Percent cross-sectional area loss due to corrosion wire  Awire = 0.056 in  As - Acorroded %lost := 100 -   As   Figure 5-10: Example calculation of estimating remaining cross-sectional area 44 In this example, the observed reading of 1390 G was taken from sample at 36” along its length where it was observed that to wires were present If to healthy wires were present there would be a cross-sectional area of 0.066 in2 to 0.087428 in2 The estimate of 0.056 in2 is reasonable as the strand is severely corroded Estimates of the remaining area for samples and can be seen in table 5-6 Table 5-6: Estimates of remaining cross-sectional areas of samples and at a 𝟏 Distance Along Longitudinal Axis (in.) 12 16 20 24 28 32 36 40 𝟏𝟓 𝟏𝟔 " concrete gap Remaing Area (in ) Sample Sample 0.089544 0.055701 0.098005 0.064161 0.089544 0.064161 0.081083 0.064161 0.072622 0.04724 0.072622 0.030318 0.072622 0.021857 0.055701 0.013396 0.055701 0.003526 0.030318 For this approach to be used the gap as well as the magnetization of a healthy strand and wire at that gap would need to be known The purposed estimating technique does contain shortfalls, such as validation, range restrictions, and contradicting estimates Although the estimates provided are reasonable when compared to the visual evidence of the strand, there is still no validation of the actual amount of remaining steel To rectify this, rust could be removed from the lab specimens and the area calculated, but investigation into the healthy diameter and 45 area of a ½” nominal strand has hindered this attempt The area of steel listed with a ½” strand is 0.153in2 which when distributed over seven wires yields 0.021857in2 per wire This in turn yields a calculated diameter of 0.1668in., which does not match the measured diameter of 0.130in The area of 0.153in2 may be nominalized and be the reason for this discrepancy Neither the ASTM nor other publications have an explanation of how the area is calculated or give an approximate diameter size for the wires that are to be used in the strand Investigation into this issue will need to be continued so this or any other estimation approach can be properly calibrated Another shortcoming of this approach may be attributed to inadequate field strength As previously mentioned, the magnetization of one wire is 1350G, which is within the range of the estimating approach, but once the magnetization drops below approximately 1320G the estimate increases from the expected One explanation of this discrepancy is due to the field strength of the magnet being used 15 The field strength of the magnet alone on the right pole is 1460G At 16 " concrete gap a ½” diameter strand has a magnetization of 1505G From other specimens tested at smaller gaps, it is known that the magnetization of a ½” diameter specimen is capable of being higher As with the first magnet at this distance, there is less magnetization In short, the magnetic capacity of the strand is not being filled However, 15 at this same distance (1 16 ") the wire’s capacity may be filled or at least have a higher a higher degree of saturation to area than the strand This cause’s a few problems when trying to estimate the area The estimation is more or less based on a linear approach, but the data show that the magnetization although nearly linear, is curved Nevertheless, it 46 appears that this causes the estimates to err on the side of caution, i.e there is more area remaining than estimated The other problem is that there could be two estimates for a magnetization of 1460G; 0in2 or 0.115in2 for this region Since the magnetization curve of a strand or any other ferromagnetic material is non-linear (see figure 5-11), there will be errors when estimating the area Nonetheless, these errors could be minimized by optimizing the electromagnet design for size of specimens and gap being tested for In other words, if the field strength were increased to ensure the specimens being magnetized would be at or near complete saturation, the error in estimating would be reduced as the data would become more linearized Until such an electromagnet becomes available, the estimating approach proposed here gives reasonable, conservative estimates and, with more testing and scrutinization, could become an effective estimating tool Figure 5-11: Magnetization curve for iron (http://info.ee.surrey.ac.uk/Workshop/advice/coils/mu/BH_iron.png) 47 Chapter Conclusion 6.1 Results The results obtained thus far are promising that this will develop into a new magnetic NDE technique A consistent reliable relationship between cross-sectional area and magnetization was found for a variety of steel specimens, including corroded prestressing strand at varying distances from the magnet face It was observed that as the cross-sectional area of the specimens increased so did the magnetization An approach for estimating the remaining area of corroded strands was developed The estimate results seemed reasonable when compared with visual inspection of the strand This research has demonstrated that the proposed electromagnetic system for detection of corrosion in deteriorated prestressing strand has significant potential The primary objective, proof of concept, was achieved 6.2 Future Research Thus far, the research and results have demonstrated that the technique proposed in this thesis has potential To further validate this technique more laboratory 48 experiments, field tests, and theoretical work must be completed Laboratory testing should consist of embedding strands in concrete at various distances These experiments should be used to validate and calibrate, or invalidate the estimating method proposed in Chapter Also, water content is believed to affect the magnetization of the strands, so testing should be done at various time intervals as the concrete hardens and the water content decreases In addition, the influence of neighboring prestressing strands and/or mild reinforcement needs to be investigated These experiments should be similar in geometric configuration as what can be expected in a bridge beam Field tests would involve the scanning of a beam on an existing bridge A bridge scheduled for demolition by ODOT has been identified as a potential candidate for field testing However, before field testing could be done, a carriage system for the magnet needs to be developed Currently, this system is being developed at The University of Toledo Data acquisition and analysis would also need to become quicker and computerized Currently, a data acquisition system has been used to re-complete the preliminary experiments with the 1018 steel rounds However, it was discovered that the Hall sensors being used did not have the capacity to produce accurate results at certain gaps and specimen diameters New sensors for this system are being ordered, as well as transitioning the system from a desktop computer to a laptop The complexity of this technique also requires that an expert in metallurgy be consulted It is known that the magnetic properties are changed with the introduction of defects, such as dislocations, grain boundaries, and boundaries between phases in the atomic structure Also, the stress state of the strand will change its magnetic properties 49 The unknown is how much these defects and the stress state will affect the overall estimate of remaining cross-sectional area and how they can be accounted for With new experimental data and metallurgical data, the estimating approach proposed can be calibrated or a new approach can be developed to estimate the remaining cross-sectional area In addition, the metallurgical data may be helpful in adjusting known theorectical equations for net magnetization to produce more accurate estimates As with any research endeavor, there is a chance of failure, but nothing thus far has shown that this is impossible 50 References Ali, M, Maddock, A “Evaluation of Corrosion of Prestressing in Concrete Using Nondestructive Techniques.” GHD Pty Ltd., Sydney 2003 Askeland, D The Science and Engineering of Materials Massachusetts PWS-KENT Publishing Company 1984 FHWA “Corrosion Costs And Preventive Strategies in the United States.” http://www.corrosioncost.com/pdf/techbreif.pdf March 2002 Hillemeier B, Scheel H “Magnetic Detection of Prestressing Steel Fractures in Prestressed Concrete.” Materials and Corrosion 1998 Indacochea, J, Rumiche, F, Wang, M “Assessment of the Effect of Microstructure on the Magnetic Behavior of Structural Carbon Steels Using an Electromagnetic Sensor.” Journal of Materials Engineering and Performance September 2007 Nims, D “Nondestructive Inspection of Deteriorated Prestressing Tendons in Concrete Bridges.” The University of Toledo March 2008 Devabhaktuni, V, Nims, D “Magnetic Sensor for Nondestructive Evaluation of Deteriorated Prestressing Strand.” The University of Toledo December 2008 51 Appendix A Figure A-1: Drawing of electromagnet design by Ohio Magnetics 52