Magnetic Resonance Imaging of concrete Dr Chris Burgoyne Department of Engineering University of Cambridge Assessment of Concrete Structures How can we tell what is going on inside concrete? We would like to know:- Has the concrete hardened? Is there corrosion? Is there cracking? Where are the cracks? Coring Systems to monitor concrete modulus Drilling into concrete to extract sample With drill to extract concrete dust for chemical analysis With core to extract intact sample To determine stress-state To examine crack structure In both cases procedure is destructive Ultra-sonic pulse velocity Relates Elastic modulus to speed of sound Assumes Concrete heterogeneous Can be affected by steel Modulus related to speed of sound Strength of concrete related to modulus Impact Echo Location of flaws Ultrasound Detects flaws by listening to echo If flaw is large, multiple echoes Difficult to distinguish one big flaw from many small flaws (www.uml.edu) If we can do it for babies, why not for concrete? (www.babyscanning.co.uk) 1
Can we use different techniques on a laboratory scale? Non-destructive-testing Methods for Concrete Structures. Irie et al Sound is reflected at interfaces Concrete has too many interfaces so it becomes very difficult to extract image Two possible techniques Computed Tomography (CT-scan) X-rays Magnetic Resonance Imaging (MRI) Magnetic field CT scanning CT examples S S An X-ray is a shadow mask due to density of the material A series of X-rays is taken from different directions The images can be processed numerically to infer density at volume cells (voxels) within the sample Works well where sample shows large density variations through the sample Ideal for medical applications where bone structure is being sampled Bone has density different from water Less good for studying soft-tissue Density close to that of water and fairly uniform CT applied to concrete Structure difficult to see because density of sand-cement very similar to that of aggregate Fractures can be seen if filled with air and if they occupy a full voxel Otherwise the density difference is too small Thus CT scanning does not appear to be an ideal tool for studying concrete Fracture of Concrete - Macro scale Features visible to the eye (~mm) Meso scale Features visible under optical microscope (μm) Micro scale Features visible under electron microscope (~nm) Variety of scales 2
At the meso-scale We would like to know:- Do cracks pass through or around aggregate? effects of size, shape, strength? What is the exact mechanism of bond breakdown? effect of bar geometry? Effects of load sequence? At the meso-scale Finite elements allow us to model the structure numerically But how can we check results experimentally? Need to be able to look inside concrete without destroying the sample Meso-scale section Specimen cut, polished and examined under microscope Limited number of sections can be obtained and only after unloading Destructive http://ciks.cbt.nist.gov/~garbocz/nistir6399/node22.htm Internal Structure observed using liquid metal Woods Metal injected while specimen under load, and hot Specimen cooled which freezes the metal Specimen cut and polished Observed under SEM Surface can be etched with HCl to reveal more detail Resolution limited by SEM Destructive Wood's metal melts at about 75 C 42.5% Bi, 37.7% Pb, 11.3% Sn, 8.5% Cd Nemati and Monteiro: A New Method to Observe Three- Dimensional Fractures in Concrete Using Liquid Metal Porosimetry Technique. (C & CR, 27/9, 1333-1341, 1997). Magnetic Resonance Imaging MRI detects protons in water Can be used to detect Structure of concrete by detecting water while cement paste is still wet Can be used to detect Fractures in concrete if they are filled with water MRI principles Protons spin axis normally random In the presence of a magnetic field the axis of spin is aligned Field must be very strong (2 Tesla 20,000 times background) B 0 3
If the magnetic field is disturbed the axis of spin rotates When the disturbance is removed the axis returns in a characteristic way and a radio-frequency signal is produced This signal can be detected and analysed MRI principles Main magnetic field direction Magnetisation of the water protons MRI signal The emitted waves are at Radio Frequencies (~100 MHz) The sample to be scanned sits inside a Radio-frequency aerial (RF Probe). The probe both excites the water protons and listens to the response To form an image we must know where in the sample the MRI signal comes from. Sample The main magnetic field strength is varied across the sample by a set of gradient coils There are three sets of coils which produce a gradient in each of the three spatial dimensions. B gradient Magnetic field strength varies over sample RF response from different places has a different frequency All signals received at same time Signal strength Time Z frequency changes with position Signal must be processed using Fourier transfoms to extract spatial information MRI of water Water is H 2 O Hydrogen is proton plus electron Returned signal depends on how free the protons are to vibrate This depends on the mobility of water Bulk liquid water very mobile strong signal Water adsorbed onto a surface weaker signal decays more quickly Solid water (ice) low mobility almost impossible to detect MRI Material Characteristics The MRI response is characteristic of the material - two time constants measured T 1 time constant of process of return to thermal equilibrium after disturbance by radio-frequency pulse T 2 time constant of the process of loss of phase coherence in the transverse plane 4
Signal variation The different characteristics of the signal can be used in medical imaging to distinguish between different soft tissues (internal organs, tumours, blood, etc) Medical MRI complements medical CT scanning http://sprojects.mmi.mcgill.ca/braintumor /section2/subsection2/default.htm Typical values Pure bulk water with 2 Tesla magnet T 1 and T 2 both about 3 seconds MRI relatively easy In porous media T 1 in range 0.03 0.3 seconds T 2 in range 30 3000 microseconds MRI much harder Values in large pores in concrete will differ from values in narrow fractures Spin Echo Protons disturbed by two pulses of RF The signal produced as the protons return to their equilibrium state is recorded after a period T E (Echo time) After signal has decayed the process can be repeated after time T R (Repetition time) The process can be repeated to improve the signal-tonoise (S/N) ratio Most images here taken with:- T E = 4.5 ms T R = 100-800 ms 4 repetitions MRI Equipment 2 Tesla superconducting solenoid magnet with 310 mm bore Magnet Cooled to 269 C with liquid helium MRI Equipment 2 Tesla superconducting solenoid magnet with 310 mm bore Contains gradient coils which generate 1.14 Gauss/mm with bore 108 mm This contains radio-frequency probe with a cylindrical space 54 mm in diameter; the fieldof-view is located in the central 100 mm. Gradient coils RF probe Magnet Sample Field of view 40 mm * 40 mm transversely divided into digital matrix 256 * 256 pixels gives resolution 156 * 156 μm 40 mm longitudinally divided into 32 slices each 1250 μm long 5
Aggregate in water Placing aggregate in water allows material suitability to be assessed Note that the water returns a signal the aggregate appears as a shadow Both limestone and quartz show clear images with high contrast Magnetic materials Ferromagnetic materials locally affect the imposed magnetic field strength Resulting image is distorted Both these images should be circular but are distorted by presence of iron Effect of cement type Hardened cement samples with circular holes containing water Plaster shows signal from matrix and no distortion Ordinary Portland Cement shows distortion of circular holes due to iron in cement and no signal once hardened White Portland Cement WPC shows no distortion but also no signal Concrete type Hardened samples with limestone aggregate:- WPC shows no signal Plaster shows signal WPC and plaster shows signal As concrete hardens Signal intensity Structure can be determined up to 8 hours after casting Structure of normal concrete Can be observed if After one hour (more clear) After 8 hours (less clear) 1 10 100 1000 Time after casting (hours) 1. use White Portland Cement 2. use Limestone Aggregate 3. the sample is scanned within first 8 hours after casting 6
Fracture of concrete Can we observe fractures in hardened concrete? Limestone in hardened plaster Fracture of concrete Plaster not representative of normal concrete Uncracked Cracks filled with air (Crack black) Cracks filled with water (Crack white) Difficult to determine location of crack Aggregate leaves shadows which alter level of grey Use normal concrete Unconfined Hardened samples Bond specimen Mould Structure image obtained soon after casting Fracture image shows no structure Confined by aramid spirals Partially bonded bar Potato marker Compression samples Confined and unconfined samples tested in axial compression Final crack pattern Unconfined Confined by aramid spirals 7
Image is multiple slices Final crack pattern Sliced images can be manipulated 1. Remove concrete 2. Combine slices 3. Make movie or 4. Visualise in VR software Combined images (bond specimen) Fractures of concrete AFRP bar AFRP bar (debonded) Structure Fracture Combined Can be observed after loading 1. By filling voids with water 2. Scanning sample 3. Combining images with structure scans Can we observe fractures when loaded? 3F 125 To observe fractures under load Load beam in flexure 25 25 F F 550 F F F F 163 RF-probe 3F Test beam inside scanner Test frame must be non-magnetic Reinforcement must be aramid fibres Use medical syringes as hydraulic jacks Beam position Field of view RF Probe 70 14 30 163 548 740 Gradient coil Magnet Gradient coil RF Probe Field of view Beam position Deflection 54 108 176 Space is very confined Reaction frame must be within the scanner Specimen and frame must not touch the scanner when the beam deflects The scanner must not be damaged if the beam breaks 70 8
Reinforcing cage made from twisted aramid fibres in resin RF probe Loading jacks Beam Water tank Reaction frame Forming composite structure image Test procedure Water tank filled Beam loaded by external hydraulic pressure through jacks deflection controlled Water tank drained Specimen scanned Process repeated until failure Structure Before loading Fracture Different stages of the loading process Note that these images are 3D views showing the internal crack geometry of the same specimen Processed Structure Processed fracture Composite image 9
Can we infer more? Look at T 1 and T 2 maps T 1 T 2 Value of T 1 or T 2 Signal intensity due to T 1 or T 2 Look at this area in detail Both T 1 and T 2 affected by nearness to the concrete surface Can we use this to tell us more about the fractured state? Effect depends on materials A acrylic/acrylic B acrylic/marble C acrylic/cement Fracture widths change with load Signal intensity changes Can we obtain more information? 10
Use NMR to get bulk values for T2 Water in cement has three components, each with different T 2 values Chemically combined (T 2 < 10-3 s) Gel Water (T 2 ~.01 s ) Capillary and fractures (T 2 >.05 s) These can be detected in bulk using NMR and the longer ones in detail by MRI Detection of onset of cracking Bulk NMR measurements are quick (minutes), so specimens can be tested many times Detailed MRI measurements are slow (8 hours) so would only be carried out when NMR tells us that significant cracking has occurred. Do we see what we think we see? Seek a Gold Standard comparison Inject specimen with low-viscosity resin; then either Grind down surface Take physical slices Take photographs to compare images MRI Image MRI image superimposed on digital photo Conclude that the MRI images are in the correct place Microscope ~5μm pixels Digital Camera ~ 19μm MRI ~156 μm 11
Fracture 1A 1B 2A 2B 3A 3B 3C 4A Comparison of Fracture Sizes (microns) Filteredlight photograph 95 80 80-95 100 250-285 540-600 355-395 100 Penetrationlight photograph 110 90 70-90 105 325-360 565-610 390-410 100 Microscopic measurements 46.5 65.1 72.3 101-123 220.4 545-630 300-440 106.7 MRI black/white x 156 μm 3 0-1-2 2 1-3 3 7-8 3-4 2-3 MRI volumetric 26 56 (M) 85 103 207 520 (R/F) 260-320 95-175 Conclusions Magnetic Resonance Imaging can be carried out to determine the structure of concrete and the fracture state of concrete Crack widths can be calculated from the MR image and from the properties of the MR signal (T 2 ) Onset of cracking can be determined by NMR measurements Future work Linking structure scans to finite element analyses compare predicted crack patterns with observed Use of metallic non-ferrous reinforcement to generate bond failures observe crack patterns Speed up scanning Improve resolution Long-term goal Look outwards? Samples outside scanner Real structures Allow for iron? Use distortion caused by iron to predict where iron is located Negate effects of iron from scanned images 12