Embed Size (px)
Citation preview
Journal of Applied Geophysics 105 (2014) 138–146
Contents lists available at ScienceDirect
Journal of Applied Geophysics
j ourna l homepage: www.e lsev ie r .com/ locate / j appgeo
Experimental measurements of shock induced changes to themagnetization of unexploded ordnance
Stephen Billings ⁎, Laurens BeranBlack Tusk Geophysics Inc., #401, 1755 West Broadway, Vancouver, BC V6J 4S5, Canada
⁎ Corresponding author. Tel.: +1 720 306 1165.E-mail addresses: [email protected] (
[email protected] (L. Beran).
http://dx.doi.org/10.1016/j.jappgeo.2014.03.0150926-9851/© 2014 Elsevier B.V. All rights reserved.
a b s t r a c t
a r t i c l e i n f oArticle history:Received 8 August 2013Accepted 20 March 2014Available online 25 March 2014
Keywords:MagnetometryUnexploded ordnanceShock demagnetizationShock remanent magnetization
Millions of acres of land around theworld are potentially contaminated by unexploded ordnance (UXO).Magne-tometry is a technique widely used to both detect and characterize buried UXO. It has been hypothesizedthat ordnance suffer a large shock on firing and impact that erases any preexisting remanent magnetization. Ifsuch demagnetization occurs, an apparent remanence metric has been shown to be effective at distinguishinghazardous ordnance from non-hazardous metallic debris. To test the shock demagnetization hypothesis, anexperiment was conducted at a firing range to measure the magnetic remanence of sixty-five inert 81mmmor-tars before firing and after impact. As delivered, 64 of the 65 rounds had very low remanent magnetization and amagnetizer had to be used to impose various amounts of remanence on themortars. Three different categories ofinitial remanent magnetization were created (low, medium and high remanence) and these were fired at threedifferent initial velocities. Themortars that initially had low remanentmagnetization acquired amagnetization inthe direction of the Earth's inducing field after impact, with the amount of re-magnetization decreasing with anincreasing impact velocity. This effect is known as shockmagnetization. Themortars withmedium and high initialmagnetization all lost some of their magnetic remanence, with the amount of demagnetization increasing withan increasing impact velocity. However, even at the highest impact velocity, shock demagnetization of initiallyhighly magnetized mortars was insufficient to guarantee effective discrimination using apparent remanence.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
Millions of acres of land around the world are potentially contami-nated by unexploded ordnance (UXO). Themost well-established tech-niques for UXO detection are magnetics and electromagnetic induction(EMI) (Butler (2001), Nelson and McDonald (2001), Zhang et al.(2003)). These methods are very effective at not only locating UXObut also detecting many non-hazardous items such as shrapnel andcultural debris. Therefore discriminating between intact UXO and non-hazardous objects has the potential to significantly reduce clearancecosts. The richer information content and the relative immunity tomagnetic soil effects of EMI methods make them superior to passivemagnetometers for the discrimination of detected targets (Pasionet al., 2008). However, magnetometer-based methods are still widelyused, particularly in areas of difficult terrain/vegetation and for under-water applications (Salem et al., 2005). Thus, the intrinsic discrimina-tion capability of magnetic data is still of considerable interest.
Buriedmunitions composed of ferrous components (e.g. steel) causea distortion in the Earth'smagneticfield that can bemeasured by amag-netometer. However, magnetic anomalies also arise from shrapnel and
S. Billings),
other ferrous debris, as well as from geological variations in magneticminerals in the soils and rocks. To distinguish between a buried UXOand benign clutter or geology, an identifying feature of the magneticanomaly of a UXO item must be determined. Ideally, we would like tobe able to recover the shape and size of each detected anomaly's sourceand use that information for discrimination. However, there is a funda-mental ambiguity inmagnetic data whereby anymagnetic anomaly canbe represented by an equivalent layer of susceptibility (Stratton, 1941).This precludes unique recovery of a buried object's physical properties.
To proceed, we note that the response of a compact body canbe decomposed into a series of moments by a multi-pole expansion(Stratton, 1941). In most cases, measurements are made in the far-field of the object (i.e. at distances several times the object's dimensions)so that the response of the dipole component dominates the measuredanomaly. Higher order moments decay rapidly with distance and wecan therefore usually only recover a buried object's dipole momentfrom the observed magnetic data (Billings, 2004).
In simplistic terms, an object's dipole moment is a consequence ofboth remanent and induced magnetization. Remanent magnetizationis present even in the absence of an inducing field and is due to ferro-magnetic domains in the steel being locked into alignment sometimeduring the object's history. Inducedmagnetism arises becausemagneticdomains in a ferrous material tend to align with the direction of theambient field (Stratton, 1941).
Table 1Mortar test-matrix showing parameters that were varied. The numbers in brackets are theactual numbers achieved in each category.
Remanence Low Medium High Total
Projectile Trans. Axial Trans. Axial
81 mm (minimum charge) 4 (5) 4 (6) 4 (4) 4 (0) 4 (6) 20 (21)81 mm (medium-charge) 4 (9) 4 (5) 4 (4) 4 (0) 4 (5) 20 (23)81 mm (maximum charge) 4 (5) 4 (8) 4 (4) 4 (0) 4 (3) 20 (20)
139S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
Artillery projectiles andmortars will suffer at least two large shocks:one during the initial firing of the projectile or mortar and the other onimpact with the ground. Altshuler (1996) postulated that the shocksexperienced during firing and impact partially erase the remanentmag-netization of a UXO item. He further noted that the direction of inducedmagnetization in typical munition items is constrained to lie withinabout 60° of the Earth's field. Billings (2004) developed the apparentremanent magnetic discrimination method to exploit these shock in-duced changes to magnetization. Magnetic data from each anomalyare inverted to produce an initial dipole model. This recovered dipolemodel m is then compared to a family of induced magnetic responsesof all munition items suspected at the site. An apparent remanence γis computed as
γ ¼ 100 Δmk kmk k ð1Þ
where Δm is the minimum deviation of the recovered dipole momentfrom the induced dipole for an ordnance item. The deviation is normal-ized by the magnitude of the estimated dipole (‖m‖) to produce apercentage. This comparison process defines an apparent remanentmagnetization for each target in the ordnance library and allows a dis-crimination ranking criteria to be established. Due to shock demagneti-zation, UXO items in the munition library that are encountered at a siteare hypothesized to have low values of apparent remanence. Geologicalanomalies, shrapnel and other metallic debris will, in general, notmatch the induced model very well and hence will have large valuesof apparent remanence. The premise of correlating the likelihood oftargets being UXO with low apparent remanence was tested at severalMontana sites and was shown to be a reliable process for classification(Billings, 2004; Billings and Youmans, 2007). However, later measure-ments collected at two live sites by Billings (2009) raised some doubtsabout the general applicability of apparent remanence for ordnancediscrimination. The present work therefore seeks to develop a morecomprehensive understanding of the physical phenomena behind theapparent remanence method.
We describe an experiment conducted at the Aberdeen Test Center(ATC) in 2009 where the remanent magnetization of sixty-five inert81 mm mortars were measured both before and after-firing in a seriesof controlled tests. The mortars were filled with a wax simulant to en-sure that their mass was the same as live rounds. The ATC firing testswere intended to determine:
• Whether shock demagnetization reduces the amount of remanentmagnetization remaining after impact;
• Whether ordnance are re-magnetized in the Earth's field after shockdemagnetization;
• How impact velocity affects the amount of shock demagnetizationthat occurs; and
• Whether rounds with extremely large remanent magnetizationundergo sufficient shock demagnetization to be correctly classifiedusing the apparent remanence method.
Shock and stress induced changes to magnetization of metals androcks have been the focus of a large volume of literature in a numberof fields including paleomagnetization (e.g. Louzada et al., 2010) andnon-destructive testing (e.g. Staples et al., 2013). In 1949, Brown inves-tigated changes to magnetization caused by mechanical disturbancesand related them to the hysteresis loop (Brown, 1949). Brown changedthe independent variable in Rayleigh's Law from magnetizing forceto stress. He postulated that the magnetizing force or stress causesdisplacements of the walls separating adjacent domains and more fa-vorably magnetized domains grow at the expense of their neighbors.Nagata (1971) showed that isothermal remanence carried by a rockcan be substantially reduced, and a remanence proportional to the am-bient field can be acquired at pressures comparable to those suffered bya rock when struck by a geologic hammer. Gattacceca et al. (2010)
conducted experiments to better understand shock demagnetizationand shock remanent magnetization (SRM) caused by high velocity im-pacts of planetary or asteroidal surfaces. They found that the amountof SRM was independent of the initial remanence and dependent onlyon the strength and direction of the field at the time of impact.
2. Materials and methods
Billings (2009) describes the “Magnetic Remanence InterrogationPlatform” (MRIP), a device for measuring the contributions of the in-duced and remanent magnetizations of a steel item. The MRIP com-prises six three-component fluxgate magnetometers symmetricallydistributed around a rotating sample holder. Samples are placed onthe holder and are slowly spun through two complete rotations. Themeasurement is repeated after the sample is physically rotated by 90°,so that the initial vertical axis becomes horizontal.
2.1. Selection of test-rounds
Personnel at Aberdeen Test Center conducted a study to identifyprojectiles that could be used for pre and post-fire/impact measure-ments of induced and remanent magnetic fields. A moderate size pro-jectile was sought that had, or could easily be converted to, an inertcounterpart. The size and configuration of the projectile had to besuch that magnetic field readings could easily be taken with the MRIP.Another considerationwas that the rounds have trajectories and impactvelocities that would enhance recoverability and minimize test setupcosts. By analyzing assembly, geometry, mass, trajectory characteristicsand expected penetration depths an acceptable candidate was identi-fied. The M889A1 mortar cartridge is a common round with an inerttarget practice counterpart: the M879 cartridge. The empty mass ofthe M879 is 3.3 kilograms (kg) and when wax-filled weighs the sameas a M889A1 HE round (4.14 kg). Both mortar types have a diameterof 81 mm. The abundant use of the M879 round makes it easily attain-able for testing. Further, mortar firing systems are easily set up, theirrounds typically have small dispersions, and associated penetrationdepths are small. The M879 round would therefore facilitate a low testcost with good recoverability potential.
Charge increments and firing elevations were selected for theM879to provide a variety of launch and impact shocks. Expected remanenceand projectile flight characteristics are shown in Tables 1 and 2.
2.2. Test matrix
An axi-symmetric ordnance item has a much larger induced mag-netization in the axial direction, ma, than in the transverse directionmt. From prior measurements of similar sized rounds we expectedmt = 0.07 amperes-meter2 (Am2) and ma = 0.24 Am2 for the 81 mmmortar (Billings et al., 2006). After measuring the rounds with theMRIP we were able to more precisely define these numbers asthe mean values of sixty-one different MRIP measurements: mt =0.0826 Am2 and ma = 0.375 Am2 with standard deviations of 0.0015and 0.0041 Am2 respectively.
The magnetic remanence discrimination method of Billings (2004)considers a round to be a UXO if the apparent remanence is b50%,although typically a larger cutoff is used to account for uncertainty in
Table 2Flight characteristics for the M879 mortars.
Charge Muzzle velocity(m/s)
Time of flight(sec)
Angle of fall(from horiz) (°)
Impact velocity(m/s)
0 66 11.3 56.2 641 152 23.9 56.8 1312 250 36.5 78.6 210
Fig. 1. (Top) Photograph of 81mmmortar on the MRIP turn-table before firing. (Bottom)The entry hole and tail of one of the 81 mm mortars fired with 0 charge increment.
140 S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
the recovered dipole moment. As shown in Table 1, we planned to usethree categories of remanence which we defined as follows:
• Low remanence, which we define as mr b mt/3: Roughly speaking,we would consider a remanence, mr, of less than mt/3 to be small.Even if no shock demagnetization occurred, the apparent rema-nence would never be larger than 50% so that the item should becorrectly identified as a potential UXO (calculated assuming aworst case with the item perpendicular to the Earth's field and theremanence in the opposite sense to the induced magnetization, sothat γ = mt/3 / (mt − mt/3) = 50%);
• Medium remanence, which we define as mt/3 bmr b ma: In this case,the apparent remanence could vary from0% to (theoretically) infinity.At worst the remanent magnetization would exactly cancel the in-duced magnetization producing a moment of 0 (and an apparentremanence of infinity). Depending on the orientation of the roundand the remanence, shock demagnetization may or may not be re-quired for these rounds to be classified as potential UXO.
• High remanence: mr N 4 ∗ ma: In this case, the remanence wouldnever be smaller than 75% (Calculated assuming a best case of rema-nence along the transverse axis with the remanence and inducedmagnetization aligned in the same direction). Therefore shock de-magnetization would have to occur for the round to be classified asa potential UXO.
We measured all rounds soon after arrival at ATC and found that allexcept one were in the low-remanence category. Consequently, roundswere magnetized using the methodology described in Section 2.4. Forthe two higher remanence categories, we had two classes: one withthe remanent magnetization predominantly along the traverse axis,and the other with the remanence predominantly along the axial axis(see Table 1). It was very difficult to magnetize items in the transversedirection and any attempts to generate items in the “high transverse”remanence category resulted in additional samples in the “mediumtransverse” or “high axial” categories. One of the high-axial remanencecategory items fired at maximum charge could not be recovered, thusthere were only three items in that category instead of the intendedfour items.
2.3. Measurement procedure
The following measurement procedure was followed:
1. MRIP set-up: The MRIP platform was set-up close to the ATC de-magnetization facility but at least 50 m from any power-linesand roads with passing traffic. Candidate areas were swept withSchonstedt magnetometers to ensure that the immediate area ofthe MRIP platform was free of large ferrous items.
2. Baseline survey of all rounds: All of the 81 mm mortars were mea-sured on the MRIP platform (see Section 2.3.1 for specific details)and sorted into low, medium and high remanence categories. Ini-tially, all but one of the rounds were in a low-remanence category.During this process each round was assigned a unique label and areference pointwas engraved on one side of the round. Allmagneticmeasurements were subsequently oriented with the z-axis alignedwith the long (axial) axis of the ordnance with positive pointingtowards the ordnance nose, the y-axis transverse to the ordnance
symmetry axis and pointing towards the reference marker, andthe x-axis orthogonal to the y- and z-axes to form a right-handedcoordinate system.
3. Magnetization: Because all but one of the rounds were in a low-remanent state, the medium and high remanence rounds had tobe magnetized in non-destructive evaluation (NDE) equipment(see Section 2.4 for specific details).
4. After magnetization in the chamber, the magnetic remanence ofeach round was measured on the MRIP, and additional magnetiza-tion was imposed if necessary.
5. Transportation of rounds and attachment of ignition cartridges: Therounds were then transported to a bunker close to the firing rangeand the ignition cartridges corresponding to charges 0, 1, and 2were attached.
6. Relocation ofMRIP and set-up adjacent to thefiring range: TheMRIPplatform was moved closer to the firing range, in a location free ofsignificant magnetic variation and at least 50 m from the nearestroad and power-line. The selected location was close to the firingpoint so that rounds could be measured immediately before firing.
7. Pre-firing measurement: A few hours before firing, the remanentmagnetization of each round was again measured on the MRIP.The values measured just before firing agreed to within a fewpercent of the values measured before attachment of the ignitioncartridges.
Fig. 2. Example hysteresis curve.
141S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
8. Firing of projectiles: Projectileswere then fired into the impact area.A number of spotting rounds were first fired to ensure that the testrounds would land within a specific area of the test-field.
9. Impact location: Spotters in a control tower observed the ordnanceimpact and estimated a range and bearing to each target.
10. Relocation of rounds: The recovery team navigated to the approxi-mate location of each round and placed a pin-flag at the entrance
0 0.25 0.5 0 0−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Transverse Rem. (Am2) Transverse
Axi
al R
eman
ence
(A
m2 )
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Axi
al R
eman
ence
(A
m2 )
(a) Charge 0 (b) C
Fig. 3. Change in remanence afterfiring and impact for (a) charge 0; (b) charge 1; and (c) chargefor the 81 mm mortar. A gray line joins the before firing and after impact values for a given ite
hole which was typically clearly visible. Some of the 0 charge incre-ment rounds had their tails sticking out of the ground and wereeasily recovered (Fig. 1).
11. Excavation of rounds: For the shallower rounds (charges 0 and 1),shovels were used for excavation, while for the deeper rounds acombination of an excavator and a shovel was used. Every effortwas made to ensure that each round did not suffer a significantshock during the excavation process. Because an excavator wasused for the deeper charge 2 rounds, there is no guarantee thatthe rounds were extracted without further shocks. All roundswere recovered within 6 h after they were fired.
12. Measurement of depth and orientation: The orientation and depthof each round that was excavated with a shovel were measuredbefore removal. The measurements included the orientation ofthe y- and z-axes of the ordnance (the y-axis was identified bythe reference mark engraved on each item). Each round was alsophotographed.
13. Holding location: Each roundwasmoved to a holding location adja-cent to the MRIP.
14. Post impact measurement on MRIP: Each round was measured intheMRIP device at a location close to the impact locations (b500m).
2.3.1. MRIP measurement procedureThe following is a summary of themeasurement procedurewe used:
1. The MRIP was installed in a fixed location away from magneticnoise sources and buried ferrous objects or geology: (a) The sampletable was oriented towards Magnetic North; (b) Each flux-gatemagnetometer was rotated so that it read 0, or near zero in the
.25 0.5 0 0.25 0.5
Rem. (Am2) Transverse Rem. (Am2)
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
Axi
al R
eman
ence
(A
m2 )
harge 1
1
(c) Charge 2
Before firingAfter impact
2. The two dashed black circles delineate the transverse and axial inducedmagnetizationsm.
142 S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
x-direction, and so that the y- and z-components are related byBz = −By tan I (where I is the inclination of the Earth's field).
2. Calibration data were collected using a coil with known magneticmoment that was successively oriented in three different directions.
3. Background measurements were collected before and after mea-suring each round so that diurnal variations could be removedfrom the sensor data.
4. Each item was carefully placed in the sample holder (Fig. 1) so thatits reference frame was aligned accurately with the sensor's frameof reference (e.g., the y-axis of the sample is parallel to the y-axis ofthe sensor coordinate system).
5. Flux-gate magnetometer data were collected at a 12 Hz (Hz) ratewhile the sample turntable was rotated through 360° (at a rate ofabout 1 rotation every 30 s).
6. The sample was rotated by 90° so that the top of the sample pointedEast, andmeasurements were collected through another 360-degreerotation.
7. Using a non-linear least squares optimization algorithm in Matlab(Coleman and Li, 1996), a model was fit to the magnetometerdata acquired with the MRIP for each item. The model comprised(a) three vector components of remanent magnetization along theprincipal axes of the target; (b) three vector components of inducedmagnetization along the principal axes; (c) the three Euler anglesof the principal axes relative to the sensor's frame of reference; and(d) the two-components of the lever arm of the center of magnetiza-tion (apparent location of dipole moment) of the item relative to theaxis of the turntable, aswell as the height of the center of magnetiza-tion above the turntable. Quantities c) and d)were typically differentfor the two sets of rotation data due to differences in positioning andorientation of the samples. See Billings (2009) formore details on themodel fitting process.
0 0.05 0.1−0.2
−0.15
−0.1
−0.05
0
0.05
0.1
0.15
0.2
Transverse Rem. (Am2)
Axi
al R
eman
ence
(A
m2 )
0 0.−0.2
−0.15
−0.1
−0.05
0
0.05
0.1
0.15
0.2
Transverse
Axi
al R
eman
ence
(A
m2 )
(a) Charge 0 (b) Ch
Fig. 4. Zoomed in view of the change in remanence after firing and impact for (a) charge 0; (b)given item. Also shown on this plot are contours that delineate the regimes where apparent re
2.4. Magnetization of rounds prior to firing
The behavior of ferromagnetic materials below the Curie tempera-ture under the influence of a magnetic field is described by a hysteresisloop (Fig. 2). The application of amagnetic field to amulti-domain graincauses the preferential growth of domains with magnetization parallelto the field. If the applied field (H) is strong enough, domain walls aredestroyed and magnetization (B) reaches saturation (Bs). On removalof the magnetizing field, domains reform and move back towardstheir initial positions. Because of lattice imperfections and internalstrains, domain walls settle in a different (local) energy minimum anda remanent magnetization results (Br).
To magnetize the 81 mm mortars, an H-6472 model direct current(DC)-magnetizer was used. The active coil in the H-6482 demagnetizeris 60 cm (cm) in diameter and 15 cm wide, with the useable field ex-tending approximately 23 cm on either side of it. Thus an item upto 60 cm in length can be accommodated either axial or transverse tothe coil axis. A DC-magnetic field of up to 8 kA/m was applied to the81 mmmortars, with the amount of magnetization confirmed by mea-suring the sample in the MRIP. The M879 mortar is made from HF-1steel (HF is hot-forged), which is a high carbon content steel. A searchof the literature did not reveal the magnetic properties of HF-1 steel.From Woolman and Mottram (1964), a typical high carbon steel willhave a magnetic saturation value on the order of 1.81 T. An appliedfield of 8 kA/m is sufficient to produce amagnetic flux density of around1.67 T, or approximately 92% of the magnetic saturation value of high-carbon steel. Thus, the H-6472 is, in theory, capable of taking an 81mm mortar through most of the range of the hysteresis loop in Fig. 2.As discussed in Section 2.2, using this process we were able to imposea large axial remanence but not a large transverse remanence on the81 mm mortars. This is presumably due to shape demagnetization
05 0.1
Rem. (Am2)
0 0.05 0.1−0.2
−0.15
−0.1
−0.05
0
0.05
0.1
0.15
0.2
Transverse Rem. (Am2)
Axi
al R
eman
ence
(A
m2 )
arge 1 (c) Charge 2
Before firingAfter impact
charge 1; and (c) charge 2. A gray line joins the before firing and after impact values for amanence is always less than 50% (solid line) and 75% (dot-dashed line).
0 0.2 0.4
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Perpendicular (Am2)
Par
alle
l (A
m2 )
0 0.2 0.4
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Perpendicular (Am2)
Par
alle
l (A
m2 )
(a) Charge 0
1
(b) Charge 1
Before firingAfter impact
Fig. 5.Change in remanence afterfiring and impact for (a) charge 0; and (b) charge 1,withremanent moments split into components parallel and perpendicular to the induced fieldat the time of impact. A gray line joins the before firing and after impactmeasurements fora given item.
143S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
which significantly reduces the size of the internal magnetic fieldwhichcan be imposed in the transverse direction.
3. Results
In Fig. 3 we plot the axial and transverse remanent magnetizationbefore and after firing for charges 0, 1 and 2. A gray line joins the mea-surements for a given round. Also shown on the plot are the transverseand axial induced magnetizations for the 81mm rounds. These providean indication of the magnitude of the remanent magnetization com-pared to the possible range of inducedmagnetizationwhich can assumeany value between theminimal transverse and maximal axial magneti-zations (depending on the orientation of the mortar relative to the am-bient magnetic field). Several observations can be made from the plots:
1. As the number of charge increments increases a larger proportion ofthe initial remanent magnetization is lost after impact. This confirmsthat shockdemagnetization occurs and demonstrates that the degreeof demagnetization is dependent on the shock of impact.
2. The rounds with low initial remanent magnetization increase theirremanence after impact. This confirms that shock remanent magne-tization occurs: the round acquires a component of remanent mag-netization in the direction of the ambient field at the time of impact.
The second observation regarding remagnetization is more readilyapparent in the zoomed-in views shown in Fig. 4. Close-inspection ofthese plots reveals that the amount of magnetization decreases withan increasing charge size. Also shown on the plots is a contour whichdelineates the regime where the apparent remanence would alwaysbe less than 50% regardless of the final orientation of the projectile.If all projectiles were within this regime, discrimination using the ap-parent remanence metric would be a viable and potentially reliablemethod for UXO discrimination (for these rounds and shock regime).For each of the charge increments, there are a number of cases wherethe after impact remanence places the item outside this regime andhence discrimination using remanence would not be reliable. A secondcontour delineates the regime where all possible orientations wouldproduce an apparent remanence of less than 75%. There are still a num-ber of after impact remanence values that lie outside this regime, withthe number decreasing with increase in charge increment. Thus appar-ent remanence discrimination becomes more reliable the larger theshock of impact, but even at high shock rounds with large preexistingremanence may not be sufficiently demagnetized for reliable discrimi-nation with the apparent remanence metric.
Close inspection of the axial and transverse remanent magnetiza-tions of the mortars fired with charge 0 in Fig. 3a reveals a tendency ofsome of the high remanence rounds to acquire more transverse rema-nence after impact than they possessed before firing. Fig. 5 providesan alternative visualization of the change in remanence between theinitial and after impact scenarios and provides an explanation for thisobservation. The remanence is split into components parallel and per-pendicular to the inducing field at the time of impact. We assume thatthe final orientation of the round is the same as the orientation at thetime of impact and use theMRIPmeasured transverse and axial inducedmagnetizations to estimate the direction of the inducing fieldwithin themortar. We can only show results for mortars fired with 0 and 1 chargeincrements as orientationmeasurementswere notmadeon themortarsfired with two charge increments. Almost all the mortars fired withone charge increment appear to be moving toward a final state with amodest amount of magnetization in the ambient field direction andwith zero magnetization in the direction perpendicular to the ambientfield. This progression results in the increase in the remanentmagnetiza-tion evident in the transverse component of magnetization of several ofthe measurements shown in Fig. 3a. These observations are a conse-quence of shock remanent magnetization of the ordnance during impact.
Next, we take a closer look at the distribution of the apparent rema-nence values for the low, medium and high initial remanence samples
both before firing and after impact (Fig. 6). To calculate these cumula-tive distributions we generated 5000 random orientations of eachitem, and computed the apparent remanence that resultedwhen the in-duced and remanent magnetizations were added together. We thencombined the apparent remanence distributions for each item withina given class (e.g low initial remanence, before firing, charge increment0 constitutes one class) to come up with one cumulative distributionper class. Considering the low initial remanence category first (firstrow of Fig. 6), we see that the apparent remanence is larger after impactthan it was before firing. In addition, the change in apparent remanenceis greatest for the mortars fired with 0 charge increment. This resultreflects our earlier observation that shock remanent magnetization de-creases with increasing charge increment. Turning now to the mediumand high initial remanence categories (second and third rows of Fig. 6)we see that the apparent remanence tends to decrease after impact,with a greater change with increasing charge increment. For mediuminitial remanence, charge increments 1 and 2 are sufficient to reducethe apparent remanence to less than 50% for virtually all possible orien-tations. For high initial remanence, there are still significant proportions(40% and 25% for charge increments 1 and 2) of orientations with ap-parent remanence greater than 50%.
The results presented above concern the distribution of apparentremanence for any arbitrary final orientation of the mortar. As suchthey could be overly pessimistic as the final remanent magnetization
(a) (b) (c)
(g) (h) (i)
(d) (e) (f)
Fig. 6. Cumulative histograms of apparent remanence for rounds before and after firing. The results are presented separately for the different charge increments (as columns) and fordifference amounts of initial remanence (low, medium and high are top, middle and bottom rows respectively).
144 S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
is influenced by the ambient magnetic field at the time of impact. Thatis, there is some relationship between the induced magnetization andthe after impact remanentmagnetization. Fig. 7 plots the apparent rem-anence in the as-found orientation of each round for the before firingand after impact remanent magnetizations. We can only show the re-sults for the mortars fired with 0 and 1 charge increment as no orienta-tion measurements were made on the mortars fired with two charge
Fig. 7. Apparent remanence in the ‘as found’ orientation for before firing and after impact.
increments. Similar to the results in the previous paragraph there is anincrease in apparent remanence for the mortars that initially had lowremanent magnetization, and a decrease in apparent remanence formortars with medium and higher initial remanent magnetization.
We next present a simple model that approximates the change inmagnetization observed in themortars fired at the three different chargeincrements. The model is linear and comprises a percentage changein magnetization and an offset value. The offset value accounts forthe observation that the mortars are not progressing to a state of zeroremanence, but instead appear to be acquiring a shock remanentmagne-tization in the direction of the ambient magnetic field. Linear modelswere generated for axial and transverse remanence and for remanenceparallel and perpendicular to the ambient field at the time of impact(no results are provided for these last two cases for the mortars firedwith two charge increments). The modeling results are shown in Fig. 8with model coefficients listed in Table 3. For the mortars fired withzero charge increment, just over 50% of the initial magnetization in theaxial or ambient field directions remains after impact, compared to 23%left for one charge increment. The percentage change in axial remanenceof themortars fired with two charge increments is similar, but the offsetvalue is smaller (that is, the amount of shock remanent magnetizationacquired in the direction of the ambient field is decreased). Larger de-creases inmagnetization in the transverse and perpendicular to ambientfield directions occur for all charge increments.
4. Discussion
The MRIP device was used to make measurements of the magneticremanence of sixty-five 81 mm mortars before firing and after impact.
(a) (c)
(b) (d)
Fig. 8. Linear model of the change in remanence for the mortars fired at different change increments: (a) compares axial remanence; (b) compares transverse remanence, while (c) com-pares themagnetization in the direction of the inducedmagnetization at the time of impact (charge 2 results not shown as no orientationmeasurements weremade), with (d) the resultsfor the remanence perpendicular to the inducing field.
145S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
As delivered, 64 of the 65 rounds had very low remanentmagnetizationand a magnetizer had to be used to impose various amounts of rema-nence on the mortars. Three different categories of initial remanentmagnetization were created (low, medium and high remanence) andthese were fired at three different charge increments (0, 1 and 2 chargeincrements). After impact, the mortars that initially had low remanentmagnetization acquired amagnetization in the direction of the inducingfield, with the amount of shock remanent magnetization decreasingwith increasing impact velocity. Themortarswithmediumand high ini-tial magnetization all lost some of their magnetic remanence with theamount increasing with an increasing impact velocity (from ~50% at 0charge increment to ~70% at two charge increments). Even at thehighest impact velocity, shock demagnetization of initially highly mag-netized mortars was insufficient to guarantee effective discriminationusing apparent remanence.
The firing tests were intended to determine:
• Whether shock demagnetization reduces the amount of remanentmagnetization remaining after impact: Shock demagnetizationoccurred and resulted in projectiles with significant initial rema-nent magnetization losing between 50 and 70% of their initialmagnetization.
• Whether ordnance acquire a shock remanent magnetization whenthey impact the ground: Projectiles do acquire a shock remanent
Table 3Parameters of the linear models in Fig. 8.
Charge increment Charge 0 Charge 1 Charge 2
Coefficients Offset %Change Offset %Change Offset %Change
Axial remanence 0.0401 51 0.0466 23 0.0281 22Transverse remanence 0.0249 31 0.0104 22 0.009 10Parallel to induced 0.0471 52 0.0421 24 NA NAPerpendicular to induced 0.0256 31 0.0088 15 NA NA
magnetization in the direction of the ambient field at the time of im-pact. The amount of this acquired remanence appears to decreasewith increasing impact shock.
• How impact velocity affects the amount of shock demagnetizationthat occurs: At the lowest impact velocity considered (64 m/s),remanent magnetization decreased by about 50% compared to a70% decrease at the highest impact velocity (210 m/s). The lowerimpact velocity rounds also acquired a larger shock remanentmagne-tization in the direction of the inducing field at the time of impact(~0.047 Am2 compared to 0.028 Am2).
• Whether rounds with extremely large remanent magnetizationundergo sufficient shock demagnetization to be correctly classifiedusing the apparent remanence method: For this type of soil and pro-jectile, shock demagnetization is insufficient to erase the remanentmagnetization of the highly magnetized rounds even at the highestimpact velocity achieved (210 m/s).
5. Conclusions
The main conclusion of this experiment is that discrimination usingapparent remanence is not reliable enough to guarantee the excavationof all detected UXO under all impact scenarios. The study also providedconvincing the evidence of both shock demagnetization and shockremanent magnetization:
• Shock demagnetization: The controlled firing tests described heredemonstrate that Altshuler's (1996) hypothesis that shock demagne-tization occurs in ordnance was correct. As intuitively expected theamount of demagnetization increases with increasing impact shock.However, the maximum impact velocity achieved at ATC (210 m/s)was insufficient to completely erase the preexisting remanence ofthe 81 mmmortars that had a high initial magnetic remanence.
• Shockmagnetization: The controlled tests described here conclusivelydemonstrated that shockmagnetization of rounds occurs. Rounds that
146 S. Billings, L. Beran / Journal of Applied Geophysics 105 (2014) 138–146
initially started with a very small remanence acquired a magneticremanence in the direction of the inducing field at the time of impact.In addition, rounds with high initial remanence lost some of thatremanence and appeared to be progressing towards a similar finalstate of remanence as the rounds with low initial remanence. How-ever, unlike in the rock experiments reported by Gattacceca et al.(2010), for the 81 mm mortars measured here, the amount ofshock remanent magnetization varied with the size of the shock.
Acknowledgments
This research was sponsored by the Strategic Environmental Re-search and Development Program under grant SERDP MR-1380.
References
Altshuler, T.W., 1996. Shape and Orientation Effects on Magnetic Signature Prediction forUnexploded Ordnance. Proceedings, 1996 UXO Forum (Williamsburg, VA).
Billings, S.D., 2004. Discrimination and classification of buried unexploded ordnance usingmagnetometry. IEEE Trans. Geosci. Remote Sens. 42, 1241–1251.
Billings, S.D., 2009. Fieldmeasurements of induced and remanentmoments of unexplodedordnance and shrapnel. IEEE Trans. Geosci. Remote Sens. 47, 815–827.
Billings, S.D., Youmans, C., 2007. Experiences with unexploded ordnance discriminationusing magnetometry at a live-site in Montana. J. Appl. Geophys. 61, 195–205.
Billings, S.D., Pasion, C., Walker, S., Beran, L., 2006. Magnetic models of unexplodedordnance. IEEE Trans. Geosci. Remote Sens. 44, 2115–2124.
Brown, W.F., 1949. Irreversible magnetic effects of stress. Phys. Rev. 75, 147–154.Butler, D.K., 2001. Potential fields methods for location of unexploded ordnance. Lead.
Edge 20, 890–895.Coleman, T.F., Li, Y., 1996. An interior, trust region approach for nonlinear minimization
subject to bounds. SIAM J. Optim. 6, 418–445.Gattacceca, J., Boustie, M., Lima, E., Weiss, B.P., de Resseguier, T., Cuq-Lelandais,
2010. Unraveling the simultaneous shock magnetization and demagnetizationof rocks. Phys. Earth Planet. Inter. 182, 42–49. http://dx.doi.org/10.1016/j.pepi.2010.06.009.
Louzada, K.L., Stewart, S.T., Weiss, B.P., Gattacceca, J., Bezaeva, N.S., 2010. Shock and staticpressure demagnetization of pyrrhotite and implications for the Martian crust. Earthand Planetary Science Letters 290, 90–101. http://dx.doi.org/10.1016/j.epsl.2009.12.006.
Nagata, T., 1971. Introductory notes on shock remanent magnetization and shock demag-netization of igneous rocks. Pure Appl. Geophys. 89, 159–177.
Nelson, H.H., McDonald, J.R., 2001. Multisensor towed array detection system for UXOdetection. IEEE Trans. Geosci. Remote Sens. 39, 1139–1145.
Pasion, L.R., Billings, S.D., Kingdon, K.A., Oldenburg, D.W., Lhomme, N., Jacobson, J.,2008. Cooperative inversion of time domain electromagnetic and magnetometerdata for the discrimination of unexploded ordnance. J. Environ. Eng. Geophys. 13,193–210.
Salem, A., Hamada, T., Asahina, J.K., Ushijima, K., 2005. Detection of unexploded ordnance(UXO) using marine magnetic gradiometer data. Explor. Geophys. 36, 97–103.
Staples, S.G.H., Vo, C., Cowell, D.M.J., Freear, S., Ives, C., Varcoe, B.T.H., 2013. Solving theinverse problem of magnetisation–stress resolution. J. Appl. Phys. 113 (133905),133905–133906. http://dx.doi.org/10.1063/1.4799049.
Stratton, J.A., 1941. Electromagnetic Theory (McGraw Hill).Woolman, J., Mottram, R.A., 1964. The Mechanical and Physical Properties of British
Standard En Steels: Volume 1. Pergamon, Oxford.Zhang, Y., Collins, L., Yu, H., Baum, C.E., Carin, L., 2003. Sensing of unexploded ordnance
with magnetometer and induction data: theory and signal processing. IEEE Trans.Geosci. Remote Sens. 41, 1005–1015.
