No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Hydrodynamics in a randomly packed bed of cylindrical particles: A comparison between PR-CFD simulations and MRI experiments
ResearchGate has not been able to resolve any citations for this publication.
Multitubular reactors are commonly used in industry for processes involving highly exothermic chemical reactions. This reactor type consists of individual tubes, with a small diameter compared to the particle size. These slender beds facilitate heat management, but also give rise to flow maldistribution, which decreases the reactor efficiency. The aim of this article is to validate particle‐resolved simulations using Magnetic Resonance Imaging experiments while focusing on the flow maldistribution. The packing structure used in the simulations is reconstructed from the experimental images to facilitate a one‐to‐one comparison. A good match between experiments and simulations is found for the averaged flow profile, probability density function of the velocity in axial direction and even the local velocity distributions. However, a correction of the experimental results for magnetic susceptibility artifacts is necessary to obtain a similar match in the probability density functions and the local profiles.
Single-phase fluid flow velocity maps in Ketton and Estaillades carbonate rock core plugs are computed at a pore scale, using the lattice Boltzmann method (LBM) simulations performed directly on three-dimensional (3D) X-ray micro-computed tomography (µCT) images (≤ 7 µm spatial resolution) of the core plugs. The simulations are then benchmarked on a voxel-by-voxel and pore-by-pore basis to quantitative, 3D spatially resolved magnetic resonance imaging (MRI) flow velocity maps, acquired at 35 µm isotropic spatial resolution for flow of water through the same rock samples. Co-registration of the 3D experimental and simulated velocity maps and coarse-graining of the simulation to the same resolution as the experimental data allowed the data to be directly compared. First, the results are demonstrated for Ketton limestone rock, for which good qualitative and quantitative agreement was found between the simulated and experimental velocity maps. The flow-carrying microstructural features in Ketton rock are mostly larger than the spatial resolution of the µCT images, so that the segmented images are an adequate representation of the pore space. Second, the flow data are presented for Estaillades limestone, which presents a more heterogeneous case with microstructural features below the spatial resolution of the µCT images. Still, many of the complex flow patterns were qualitatively reproduced by the LBM simulation in this rock, although in some pores, noticeable differences between the LBM and MRI velocity maps were observed. It was shown that 80% of the flow (fractional summed z -velocities within pores) in the Estaillades rock sample is carried by just 10% of the number of macropores, which is an indication of the high structural heterogeneity of the rock; in the more homogeneous Ketton rock, 50% of the flow is carried by 10% of the macropores. By analysing the 3D MRI velocity map, it was found that approximately one-third of the total flow rate through the Estaillades rock is carried by microporosity—a porosity that is not captured at the spatial resolution of the µCT image.
Additive manufacturing of catalyst and sorbent materials promises to unlock large design freedom in the structuring of these materials, and could be used to locally tune porosity, shape and resulting parameters throughout the reactor along both the axial and transverse coordinates. This contrasts catalyst structuring by conventional methods, which yields either very dense randomly packed beds or very open cellular structures. Different 3D-printing processes for catalytic and sorbent materials exist, and the selection of an appropriate process, taking into account compatible materials, porosity and resolution, may indeed enable unbounded options for geometries. In this review, recent efforts in the field of 3D-printing of catalyst and sorbent materials are discussed. It will be argued that these efforts, whilst promising, do not yet exploit the full potential of the technology, since most studies considered small structures that are very similar to structures that can be produced through conventional methods. In addition, these studies are mostly motivated by chemical and material considerations within the printing process, without explicitly striving for process intensification. To enable value-added application of 3D-printing in the chemical process industries, three crucial requirements for increased process intensification potential will be set out: i) the production of mechanically stable structures without binders; ii) the introduction of local variations throughout the structure; and iii) the use of multiple materials within one printed structure.
Packed bed reactors are widely used to perform solid‐catalyzed gas‐phase reactions and local turbulence is known to influence heat and mass transfer characteristics. We have investigated turbulence characteristics in a packed bed of 113 spherical particles by performing particle‐resolved Reynolds‐averaged Navier–Stokes (RANS) simulations, Large Eddy Simulation (LES), and Direct Numerical Simulation (DNS). The predictions of the RANS and LES simulations are validated with the lattice Boltzmann method (LBM)–based DNS at particle Reynolds number (Rep) of 600. The RANS and LES simulations can predict the velocity, strain rate, and vorticity with a reasonable accuracy. Due to the dominance of enhanced wall‐function treatment, the turbulence characteristics predicted by the ε‐based models are found to be in a good agreement with the DNS. The ω‐based models under‐predicted the turbulence quantities by several orders of magnitude due to their inadequacy in handling strongly wall‐dominated flows at low Rep. Using the DNS performed at different Rep, we also show that the onset of turbulence occurs between 200≤Rep≤250.
Despite the substantial simplicities inherent in pseudo-continuum models of fixed bed reactors, there is a continued interest in the use of such models for predicting fluid flow and transport scalars. In this paper, we aim to quantitatively address the inadequacy of 2D pseudo-continuum models for narrow-tube fixed beds. We show this by comparing with spatially resolved 3D results obtained by a robust and integrated numerical workflow, consisting of a sequential Rigid Body Dynamics and Computational Fluid Dynamics (RBD-CFD) approach. The RBD is founded on a physics-based hard-body packing algorithm, recently proposed by the authors (Moghaddam, E.M., Foumeny, E.A., Stankiewicz, A.I., Padding, J.T., 2018. A Rigid Body Dynamics Algorithm for Modelling Random Packing Structures of Non-Spherical and Non-Convex Pellets. Ind. Eng. Chem. Res. 57, 14988–15007), which offers a rigorous method to handle resting contacts between particles. The methodology is benchmarked for simulations of flow fields in all flow regimes, for 5 <Rep <3,000, in random packings of spheres and cylinders with tube-to-pellet diameter ratios, N, between 2.29 and 6.1. The CFD results reveal a remarkable influence of local structure on the velocity distribution at the pellet scale, particularly in low-N packings, where the spatial heterogeneity of the structure is very strong along the bed axis. It is also demonstrated that azimuthal averaging of the 3D velocity field over the bed volume, which has been considered as an advancement over plug flow idealization in classical pseudo-continuum models, cannot reflect the role of vortex regions emerging in the wake of the pellets, and leads to underestimation of the local velocity values by more than 400% of the inlet velocity.
In this work, gas maldistribution in random packings of cylinders and Raschig rings is studied with particle-resolved direct numerical simulation (PRDNS) employing a finite volume solver for Cartesian grids and the immersed boundary method for the treatment of complex geometries. The geometry of the random packed bed has been obtained in a separate simulation based on a sequential deposition of particles in a cylindrical column and a simplified mechanics with additional radial force allowing for higher packing densities. The first part of the work focuses on comparison of the packings of cylinders and rings with respect to their ability to distribute the flow uniformly over the cross-section of the column when the gas in the form of a narrow jet enters along the column axis and impinges on the bed directly. The second part is devoted to analysis of three different inlet configurations and their influence on gas maldistribution in the packing of Raschig rings. It has been found that the random packing of rings has comparable properties of flow uniformisation to the same arrangement of full cylinders with significantly lower pressure drop and much less pronounced effect of formation of localised jets at the bed outlet. Among the considered inlet configurations the most beneficial is the one in which the gas duct is tangential to the walls of the distribution chamber.
A sharp interface implicit Immersed Boundary Method (IBM) is developed and used for direct numerical simulations of the flow through open-cell solid foams with a cellular structure. The complex solid structure of the foam is resolved on a non-boundary fitted Cartesian computational-grid. A single representative unit cell of the foam is considered in a periodic domain, and its geometry is approximated based on the structural packing of a tetrakaidecahedron. Simulations are performed for a wide range of porosities (0.638-0.962) and Reynolds numbers (0-500). Flow is enforced by applying a constant body force (momentum source) for three different flow directions along the {1 0 0}, {1 1 0} and {1 1 1} lattice-vectors. The drag force on the foam is calculated and a non-dimensional drag/pressure drop correlation is proposed that fits the entire data set with an average deviation of 5.6%. Moreover, the accurate numerical simulations have helped to elucidate the detailed fluid-solid interaction in complex porous media.
Packed bed reactors are widely used in the chemical and process
industry amongst others for highly exothermic or endothermic catalytic
surface reactions. Such reactors are characterized by a small
tube to particle diameter ratio D/d to ensure a safe thermal management.
For the design of such apparatuses the well-known correlations
for packed bed reactors cannot be used, because these reactors
are dominated by the influence of the confining wall, which affects
the porosity and the velocity field and as a result also the species
and temperature distribution within the bed.
For spatially resolved simulations of packed bed reactors a randomly
packed bed has to be generated and meshed. Special attention
has to be paid on the mesh at the particle-particle and particlewall
contact points. We developed a method, which flattens the
particles locally in the vicinity of the contact points and we could
show, that this method does not affect significantly the bed properties
and the fluid dynamics in terms of bed porosity, radial porosity
distribution and pressure drop.
Based on this published work we developed a workflow and a tool
which allows an automated simulation: generation of a packed bed
with DEM (discrete element method), meshing and solving the
transport equations with a finite volume code. The whole workflow
is done within the software package STAR-CCM+ by CD-adapco.
The simulation time could be reduced significantly and depends on
the number of particles (typically 1-2 days for 1500 particles).
Further we used the built-in DEM capability to generate random
packings of non-spherical particles like cylinders and Raschig rings,
which are more often used in the chemical industry, as a composition
of spherical particles. For the meshing and the CFD calculation
these approximated shapes are replaced by their original exact
shape.
With the described workflow we have investigated spherical as well
as non-spherical packings with D/d between 2 and 10 and packing
heights between 10d and 40d. The results are validated in terms of
bed porosity, radial porosity and velocity distribution and temperature
profiles with experimental results from literature.
Based on these results the interplay between the flow field, the
temperature, the species distribution and the chemical reaction in
packed bed reactors can be investigated in detail.
Cylindrical particles are widely employed in chemical engineering when a high surface area is necessary for process intensification. However, reliable information about the orientation of non-spherical particles in random packed beds is very scarce. Moreover, a potential effect of the orientation of cylindrical particles on fluid flow has not been examined yet in a real packing geometry. In this work, a new experimental approach has been applied to perform systematic studies of global and local orientation of cylindrical particles in random packed beds. Five different column-to-particle diameter Dc/Dp ratios were examined, ranging from 4 to 15.1. Orientation distribution was measured in the entire bed, with the bottom region excluded and in the core zone. Next, the experimental cases were reconstructed numerically with different packing densities. Finally, the effect of different orientation distributions of cylindrical particles, including full cylinders (FC) and Raschig rings (RR), on pressure drop (for identical bulk void fraction) in a selected packed bed configuration was examined numerically through computational fluid dynamics simulations. Reproducibility of particles’ orientation distributions was found to be dependent on the number of particles per layer. A substantial effect of the column wall and its bottom on particles orientation was observed. The effect of orientation of FC on pressure drop was found to be moderate, however, in the case of RR, the maximum difference in pressure drop among investigated orientation distributions was as high as 400%.
Particle-resolved computational fluid dynamics (PRCFD) is the most detailed modeling approach for fixed bed reactors. A major issue of PRCFD is to ensure a high quality of the underlying 3D calculation mesh around the particle-particle and particle-wall contacts, which otherwise results in convergence problems. A still unresolved challenge, especially for packed beds of non-spherical particles, is to avoid this poor-quality mesh at contacts by modification of the packed bed geometry, while providing a reasonable number of mesh cells and physical accuracy. Based on the so called local caps method, a novel mesh-type independent routine was developed to modify all occurring contacts of a packed bed of rings and to generate a computationally efficient high-quality mesh. Thereby, one of the two particles involved in a point or line contact was inflated and used as cap geometry, whereas area contacts were combined via the contact surface. The impact of the cap size on mesh quality, packed bed structure, fluid flow and heat transport was investigated for a technically relevant slender packed bed under various conditions. For a small ratio between solid and fluid thermal conductivity (λPλF≤23.1), and ReP=50−600, a cap size smaller than 0.8% of the outer particle diameter did not have significant impact on the simulation results. To apply the approach to other packing structures and particle shapes, it is recommended to investigate at least three different cap sizes to determine at which cap size the independence from contact modification is achieved.
Flow, heat, and mass transfer in fixed beds of catalyst particles are complex phenomena and, when combined with catalytic reactions, are multiscale in both time and space; therefore, advanced computational techniques are being applied to fixed bed modeling to an ever-greater extent. The fast-growing literature on the use of computational fluid dynamics (CFD) in fixed-bed design reflects the rapid development of this subfield of reactor modeling. We identify recent trends and research directions in which successful methodology has been established, for example, in computer generation of packings of complex particles, and where more work is needed, for example, in the meshing of nonsphere packings and the simulation of industrial-size packed tubes. Development of fixed-bed reactor models, by either using CFD directly or obtaining insight, closures, and parameters for engineering models from simulations, will increase confidence in using these methods for design along with, or instead of, expensive pilot-scale experiments.
Expected final online publication date for the Annual Review of Chemical and Biomolecular Engineering, Volume 11 is June 8, 2020. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
In 2006, Dixon et al. published the comprehensive review article entitled “Packed tubular reactor modeling and catalyst design using computational fluid dynamics.” More than one decade later, many researchers have contributed to novel insights, as well as a deeper understanding of the topic. Likewise, complexity has grown and new issues have arisen, for example, by coupling microkinetics with computational fluid dynamics (CFD). In this review article, the latest advances are summarized in the field of modeling fixed-bed reactors with particle-resolved CFD, i.e. a geometric resolution of every pellet in the bed. The current challenges of the detailed modeling are described, i.e. packing generation, meshing, and solving with an emphasis on coupling microkinetics with CFD. Applications of this detailed approach are discussed, i.e. fluid dynamics and pressure drop, dispersion, heat and mass transfer, as well as heterogeneous catalytic systems. Finally, conclusions and future prospects are presented.
Particle resolved direct numerical simulation (PR-DNS) has been used extensively to obtain closures for heat transfer from static particle arrays. However, most of the currently available closure models are valid for packings of spherical particles only. We present closure models for momentum and heat transfer in densely packed cylindrical particle assemblies of different aspect ratios (2, 4 and 6). Our packings are generated using the Discrete Element Method (DEM). Subsequently, the void space is meshed with a high quality computational grid, and steady-state DNS simulations are completed to provide insight into the local heat transfer and pressure drop characteristics. The variation observed in the values for the local heat transfer rates from our PR-DNS study implies the necessity of specifying confidence intervals when reporting a correlation for the corresponding Nusselt number. Our newly developed correlations are applicable to densely packed beds of cylindrical particles in the porosity range (0.405 < ε < 0.539), and allow the estimation of the variability of the Nusselt number.
In the article, a model is proposed for a single-phase flow in a random packed bed of arbitrary particles – Raschig rings in particular. The Navier-Stokes flow equations are solved using a projection method on a structured grid and the complex geometry of the bed is captured with the immersed boundary method and direct forcing approach. The model allows for analysis of flows in complex geometries with classical and widely available finite volume or finite difference solvers for Cartesian grids. The generation of unstructured computational grids fitted to the bed particles is not necessary. The realistic bed geometry has been obtained in the simulation of the filling process in which the particles are sequentially introduced into a cylindrical container and move until mechanical equilibrium is attained. The pressure drop in the bed as the function of gas velocity has been found to be in good agreement with empirical formulas for beds of Raschig rings and sample results on detailed flow characteristics are presented.
One of the major issues for particle-resolved CFD simulations of fixed beds are particle-particle and particle-wall contacts. A fixed bed of cylinders is studied with different local contact modifications, i.e., caps and bridges method for line and area contacts, and caps and united method for overlaps resulting from composite DEM-particles. Effects are analyzed regarding pressure drop, local velocity, and local heat transfer for 191 < Rep < 763. Modeling particle-wall contacts is becoming more important with increasing Rep. Contact modifications inside the bed are becoming less important, since convective heat transfer is dominant. The stable caps method shows promising results in comparison with heat-transfer predictions. However, it overestimates convective heat transfer in the contact regions for high flow rates. The bridges method shows good results for pressure drop and heat-transfer predictions. A difficulty remains the choice of the thermal conductivity of these bridges.
Cited By (since 1996):8, Export Date: 5 November 2013, Source: Scopus, CODEN: CESCA, doi: 10.1016/j.ces.2012.06.055, Language of Original Document: English, Correspondence Address: Deen, N.G.; Multiphase Reactors Group, Department of Chemical Engineering and Chemistry, Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, Netherlands; email: N.G.Deen@TUe.nl, References: Centrella, J., Wilson, J., Planar numerical cosmology. II. The difference equations and numerical tests (1984) Astrophys. J. Suppl. Ser., 54, p. 229;
In the development of meshes for computational fluid dynamics (CFD) simulations of transport in fixed beds of spheres, particle–particle and wall–particle contact points often present difficulties. We give results for drag coefficient (CD) and heat flow (Q) for flow past sphere–sphere and wall–sphere contact points, focusing on higher flow rates typical of industrial steam reformers (500 < Re < 10,000). Global methods, in which all particles in a bed are either shrunk or enlarged uniformly, change bed voidage giving erroneous results for CD. Local methods, in which bridges are inserted or spherical caps are removed only at the points of contact, give much better results for CD. The bridges approach is preferable for heat transfer, as fluid gaps reduce heat transfer too much, and particle overlaps increase it. A set of graphs is presented to allow estimation of the error introduced by the various methods of dealing with contact points.
A significant body of current research is aimed at developing methods for numerical simulation of flow and transport in porous media that explicitly resolve complex pore and solid geometries, and at utilizing such models to study the relationships between fundamental pore-scale processes and macroscopic manifestations at larger (i.e., Darcy) scales. A number of different numerical methods for pore-scale simulation have been developed, and have been extensively tested and validated for simplified geometries. However, validation of pore-scale simulations of fluid velocity for complex, three-dimensional (3D) pore geometries that are representative of natural porous media is challenging due to our limited ability to measure pore-scale velocity in such systems. Recent advances in magnetic resonance imaging (MRI) offer the opportunity to measure not only the pore geometry, but also local fluid velocities under steady-state flow conditions in 3D and with high spatial resolution. In this paper, we present a 3D velocity field measured at sub-pore resolution (tens of micrometers) over a centimeter-scale 3D domain using MRI methods. We have utilized the measured pore geometry to perform 3D simulations of Navier-Stokes flow over the same domain using direct numerical simulation techniques. We present a comparison of the numerical simulation results with the measured velocity field. It is shown that the numerical results match the observed velocity patterns well overall except for a variance and small systematic scaling which can be attributed to the known experimental uncertainty in the MRI measurements. The comparisons presented here provide strong validation of the pore-scale simulation methods and new insights for interpretation of uncertainty in MRI measurements of pore-scale velocity. This study also provides a potential benchmark for future comparison of other pore-scale simulation methods.
In this review paper we introduce currently available methods based on the Navier–Stokes equations for simulation of particulate flows which fully resolve the particles. The methods can be classified under two general categories based on the treatment of the underlying mesh namely fixed mesh methods and body-conformal mesh methods. We first consider body-conformal mesh methods and their properties. We then examine different steps of such algorithms and their application to many particle flow problems and argue why such implementations may not be feasible. Alternatively we discuss fixed mesh methods and categorize them into two subcategories namely immersed boundary methods and fictitious domain methods. A critical review of each method and their variations is provided bearing in mind the application to the particulate flow systems. The algorithms are covered in detail providing suggestions and guidelines for a successful implementation. Fundamental concepts such as discrete delta functions, body forces and calculation of surface integrals in fixed mesh methods are introduced in a simple and coherent way with simple examples and many illustrations. Major variations which are successfully applied to particulate flows are identified and the possibility of addition of heat transfer phenomena to the methods are discussed.
For liquid flow, superficial velocity profiles inside a packed bed were obtained for monodisperse packings of spheres, deformed spheres, cylinders and Raschig-ring by averaging over several thousand local measurements of the axial flow components within a cross-section. At fully developed flow, the profiles are constant along the packing except for short inlet and outlet zones of about three particles layer length with either a buildup of the profile at the inlet or a degeneration at the exit of the packed bed. Besides velocity profiles, porosity distribution functions were also evaluated experimentally. These were introduced in the extended Brinkman equation which simulates the fluctuating profiles well if the average porosity of the packing is replaced by the radial porosity distribution. The effective viscosity is adjusted to obtain best agreement between measurements and solutions of the Brinkman equation. This effective property, however, influences the flow profile in the very near wall range only, up to approximately a distance of about half the particle diameter. The effective viscosity is found to depend on the porosity data close to the wall, on particle shape, on the Reynolds number, and on the pressure loss relation.
Three-dimensional structural magnetic resonance imaging (MRI) and MRI velocimetry have been used to fully characterise the structure of the interparticle pore space and the single-phase flow field in a packed bed of alumina catalyst particles. Three orthogonal components of the velocity (Vx, Vy and Vz) are acquired such that the fluid velocity vector is determined at a pore-scale resolution of . The pore space has been analysed by unambiguously partitioning the pore space into individual pores. Characteristics of the individual pores are combined with the MRI velocity data to determine quantitative statistical information concerning flow through these pores. The ability of the lattice-Boltzmann simulation technique to predict the flow field visualised by MRI is also demonstrated by performing the simulation on a lattice derived directly from the MRI experimental three-dimensional image of the structure of the packed bed. A direct comparison of the MRI and lattice-Boltzmann results shows there is good agreement between the two methods. Using the pore analysis in conjunction with the velocity information, the flow field through the pore space is shown to be highly heterogeneous with 40% of the fluid flowing through only 10% of the pores. We also show that the lattice-Boltzmann data may be used to calculate average molecular displacement propagators similar to those acquired experimentally for such systems. The effect of the wall on the fluid velocity and porosity is calculated as a function of distance from the wall. Some difference between the MRI and lattice-Boltzmann results are observed close to the wall because of inertial effects in the high velocity channels which are not simulated by the lattice-Boltzmann method. Finally, we present initial results from the extension of this work to two-phase flow in packed beds. A case study of the visualisation of the extent of wetting of the packing as a function of time following start-up is presented.
Fixed beds of cylindrical particles are important in chemical engineering applications, but their packing structures are not as well understood or as well characterized as sphere packings. In this work, X-ray microtomography is used to obtain 3D images of 1.8 mm diameter equilateral cylinders in a 23 mm cylindrical container over a range of bulk porosities. A novel algorithm is used to computationally reconstruct the packings, resulting in data sets that give the location and orientation of each cylinder in the imaged packings. Extensive analysis has been performed, including bulk and local porosities, radial distribution functions, and parameters describing local and global ordering. The major factors affecting packing structure are the overall packing density and the proximity to the wall. At the highest overall packing densities, near-wall porosity becomes nearly equal to interior porosity, and significant global ordering occurs near the wall. For a vertical container, global ordering is characterized by the alignment of the particles with an orthogonal coordinate system that has one axis coincident with r (as defined by the container) and the other two axes in the z–θ plane, but rotated 45∘ with the horizontal. The observed structures are relevant in the context of flow maldistribution and heat transfer in fixed beds.
We present an improved method for computing incompressible viscous flow around suspended rigid particles using a fixed and uniform computational grid. The main idea is to incorporate Peskin’s regularized delta function approach [Acta Numerica 11 (2002) 1] into a direct formulation of the fluid–solid interaction force in order to allow for a smooth transfer between Eulerian and Lagrangian representations while at the same time avoiding strong restrictions of the time step. This technique was implemented in a finite-difference and fractional-step context. A variety of two- and three-dimensional simulations are presented, ranging from the flow around a single cylinder to the sedimentation of 1000 spherical particles. The accuracy and efficiency of the current method are clearly demonstrated.
CFD is a valuable tool for understanding the flow and pressure drop in packed beds. However, determining the geometry can be complex. One possible method is to use a non-invasive imaging method such as MRI, however, problems occur in processing complex geometries when using traditional commercial meshing software. This work focuses on the use of image based meshing software originally developed for the field of computational biomechanics, to create geometries from 3d MRI scans of packed beds for use with computational dynamics. For this work we focus on disordered packed beds of cylinders at low aspect ratios and Reynolds numbers of Re=1431–5074 (based on particle diameter and superficial velocity). We compare CFD studies with experimental data performed on the actual scanned beds and compare these with the correlation proposed by Eisfeld and Schnitzlein (2001). Computational data is shown to correlate well with experimental and theoretical results.
The concept of magnetic susceptibility is central to many current research and development activities in magnetic resonance imaging (MRI); for example, the development of MR-guided surgery has created a need for surgical instruments and other devices with susceptibility tailored to the MR environment; susceptibility effects can lead to position errors of up to several millimeters in MR-guided stereotactic surgery; and the variation of magnetic susceptibility on a microscopic scale within tissues contributes to MR contrast and is the basis of functional MRI. The magnetic aspects of MR compatibility are discussed in terms of two levels of acceptability: Materials with the first kind of magnetic field compatibility are such that magnetic forces and torques do not interfere significantly when the materials are used within the magnetic field of the scanner; materials with the second kind of magnetic field compatibility meet the more demanding requirement that they produce only negligible artifacts within the MR image and their effect on the positional accuracy of features within the image is negligible or can readily be corrected. Several materials exhibiting magnetic field compatibility of the second kind have been studied and a group of materials that produce essentially no image distortion, even when located directly within the imaging field of view, is identified. Because of demagnetizing effects, the shape and orientation, as well as the susceptibility, of objects within and adjacent to the imaging region is important in MRI. The quantitative use of susceptibility data is important to MRI, but the use of literature values for the susceptibility of materials is often difficult because of inconsistent traditions in the definitions and units used for magnetic parameters-particularly susceptibility. The uniform use of SI units for magnetic susceptibility and related quantities would help to achieve consistency and avoid confusion in MRI.