Meshing

Meshing
Introduction Mesh classification Structured mesh generation

Algebraic grid generation Elliptic grid generation

Unstructured mesh generation Special topics

Mesh adaptation First cell height calculation

Introduction
The partial differential equations that govern fluid flow and heat transfer are not usually amenable to analytical solutions, except for very simple cases. Therefore, in order to analyze fluid flows, flow domains are split into smaller subdomains (made up of geometric primitives like hexahedra and tetrahedra in 3D and quadrilaterals and triangles in 2D). The governing equations are then discretized and solved inside each of these subdomains. Typically, one of three methods is used to solve the approximate version of the system of equations: finite volumes, finite elements, or finite differences. Care must be taken to ensure proper continuity of solution across the common interfaces between two subdomains, so that the approximate solutions inside various portions can be put together to give a complete picture of fluid flow in the entire domain. The subdomains are often called elements or cells, and the collection of all elements or cells is called a mesh or grid. The origin of the term mesh (or grid) goes back to early days of CFD when most analyses were 2D in nature. For 2D analyses, a domain split into elements resembles a wire mesh, hence the name. An example of a 2D analysis domain (flow over a backward facing step) and its mesh are shown in pictures below.

The process of obtaining an appropriate mesh (or grid) is termed mesh generation (or grid generation), and has long been considered a bottleneck in the analysis process due to the lack of a fully automatic mesh generation procedure. Specialized software progams have been developed for the purpose of mesh and grid generation, and access to a good software package and expertise in using this software are vital to the success of a modeling effort.

Mesh classification
Meshing
Introduction Mesh classification Structured mesh generation

Algebraic grid generation Elliptic grid generation

Unstructured mesh generation Special topics

Mesh adaptation

First cell height calculation

<< Introduction | Mesh Classification | Structured Mesh Generation>>

As CFD has developed, better algorithms and more computational power has become available to CFD analysts, resulting in diverse solver techniques. One of the direct results of this development has been the expansion of available mesh elements and mesh connectivity (how cells are connected to one another). The easiest classifications of meshes are based upon the connectivity of a mesh or on the type of elements present.

Contents
[hide]

1 Connectivity-Based Classification o 1.1 Structured Meshes o 1.2 Unstructured Meshes o 1.3 Hybrid Meshes 2 Element-Based Classification

Connectivity-Based Classification
The most basic form of mesh classification is based upon the connectivity of the mesh: structured or unstructured.

Structured Meshes
A structured mesh is characterized by regular connectivity that can be expressed as a two or three dimensional array. This restricts the element choices to quadrilaterals in 2D or hexahedra in 3D. The above example mesh is a structured mesh, as we could store the mesh connectivity in a 40 by 12 array. The regularity of the connectivity allows us to conserve space since neighborhood relationships are defined by the storage arrangement. Additional classification can be made upon whether the mesh is conformal or not.

Unstructured Meshes
An unstructured mesh is characterized by irregular connectivity is not readily expressed as a two or three dimensional array in computer memory. This allows for any possible element that a solver might be able to use. Compared to structured meshes, the storage requirements for an unstructured mesh can be substantially larger since the neighborhood connectivity must be explicitly stored.

Hybrid Meshes
A hybrid mesh is a mesh that contains structured portions and unstructured portions. Note that this definition requires knowledge of how the mesh is stored (and used). There is disagreement as to the correct application of the terms "hybrid" and "mixed." The term "mixed" is usually applied to meshes that contain elements associated with structured meshes and elements associated with unstructured meshes (presumably stored in an unstructured fashion).

Element-Based Classification
Meshes can also be classified based upon the dimension and type of elements present. Depending upon the analysis type and solver requirements, meshes generated could be 2dimensional (2D) or 3-dimensional (3D). Common elements in 2D are triangles or rectangles, and common elements in 3D are tetrahedra or bricks. As noted above, some connectivity choices limit the types of element present, so there is some overlap between connectivity-based and element-based classification. For a 2D mesh, all mesh nodes lie in a given plane. In most cases, 2D mesh nodes lie in the XY plane, but can also be confined to another Cartesian or user defined plane. Most popular 2D mesh elements are quadrilaterals (also known as quads) and triangles (tris), shown below.

3D mesh nodes are not constrained to lie in a single plane. Most popular 3D mesh elements are hexahedra (also known as hexes or hex elements), tetrahedra (tets), square pyramids (pyramids) and extruded triangles (wedges or triangular prisms), shown below. It is worth noting that all these elements are bounded by faces belonging to the above mentioned 2D elements. Some of the current solvers also support polyhedral elements, which can be bounded by any number and types of faces.

Since all 3D elements are bounded by 2D elements, it is obvious that 3D meshes have exposed 2D elements at boundaries. Most of the meshing packages and solvers prefer to club exposed elements together in what is known as a surface mesh (for the purposes of applying boundary conditions, rendering meshed domains and visualizing results). A surface mesh does not have to be 2D, since volume meshes may conform to domains with non-planar boundaries. Many meshing algorithms start by meshing bounding surfaces of a domain before filling the interior with mesh nodes (such algorithms are also known as boundary to interior algorithms). For such algorithms, generation of good quality surface meshes is of prime importance, and much research has been done in the field of efficient and good quality surface mesh generation. Since surface meshes are geometrically somewhere between 2D and 3D meshes, they are also sometimes known as 2.5D meshes.

Structured mesh generation

Meshing
Introduction Mesh classification Structured mesh generation

Algebraic grid generation Elliptic grid generation

Unstructured mesh generation Special topics

Mesh adaptation First cell height calculation

<< Mesh Classification | Structured Mesh Generation | Unstructured Mesh Generation>> The simplest algorithms directly compute nodal placement from some given function. These algorithms are referred to as algebraic algorithms. Many of the algorithms for the generation of structured meshes are descendents of "numerical grid generation" algoritms, in which a differential equation is solved to determine the nodal placement of the grid. In many cases, the system solved is an elliptic system, so these methods are often referred to as elliptic methods. The best basic refence on this topic is the book of Thompson, Warsi, and Mastin[1]. There are more recent texts available, but this is the classic book on the subject, and it is available online here.

Contents
[hide]

1 Algebraic Grid Generation 2 Elliptic Grid Generation 3 Grid Marching Methods 4 Other Techniques 5 References

Algebraic Grid Generation

The simplest way to obtain a grid would be to specify the grid coordinates some vector function, or as the result of

where is the "index" vector, sometimes referred to as a computational coordinate. For our purposes here the entries of the computational coordinate will range from zero to a maximum. If such a function can be found for a given geometry, then the actual generation of gridpoints is

straightforward. The problem, however, is that the determination of the function is not neccessarily that easy. In practice, it is sometimes easier to add an intermediate parametric space, denoted by , in between the physical space representation of the grid and the computational space representation of the grid:

The entries in the computational coordinate are taken from the unit interval. This representation can help simplify matters, especially in the one dimensional case. Many mesh generation systems (both structured and unstructured) require the generation of boundary grids before interior cells can be generated. This is an area in which algebraic grid generation is ideal - typically, we want to specify boundary edge point distributions quickly, with a minimum of complexity, and a high degree of repeatability. Consider a line segment joining two points and together. The segment can be expressed as the linear form

Similar expressions are possible for other curves connecting the two points. Of particular interest is the cubic Bzier curve, which allows the specification of direction and location at both endpoints and can be written as

where the

's are the control points and again

By changing the functional expression for , we can change the grid distribution along the line segement. These functions are often referred to as stretching functions, and there are many choices available. The simplest choice is a uniform distribution, in which we set

where . For cases in which grid clustering is desired, the hyperbolic trigonometric functions such as the hyperbolic tangent are a popular choice. A simple one-parameter hyperbolic tangent stretching function is defined by

where is the stretching factor and . This function partitions the unit interval and allows the specification of a single location. This sort of distribution is good for wall-normal grid distribution in viscous flows. This distribution is due to Vinokur [2]. Vinokur's procedure for the

determination of the proper stretching factor to obtain desired spacings uses the derivatives of the stretching functions. Suppose we wish for our first grid spacing to be . This can be taken to mean that or that . Vinokur's procedure guarantees the latter. A related double-sided stretching function (that gives symmetric spacings about given by ) is

This function is good for duct flows, such as turbulent channel flow. In situations in which different grid spacings are desired, a stretching function can be constructed that has specified spacings at both ends: and . Vinokur gives such a function, first defining

The stretching factor is found from the solution of the transcendental equation

The final grid distribution is then given by

Again, Vinokur's procedure ensures that the derivative conditions and , and not the grid spacings obtained in the direct evaluation of the stretching function. For the generation of interior cells, algebraic techniques are also available, most usually in the form of interpolation between boundary faces.

Elliptic Grid Generation

The oldest numerical grid generation techniques are based upon the solution of elliptic PDE's. Typically, a Poission-type equation is solved given the boundary grid distribution to generate interior nodal points. The solution domain is often topologically equivalent to a cube in 3D and a square in 2D. Consider the solution domain shown below with the indicated boundary resolution.

The simplest technique we could use here would be a solution of the Laplace equation using the standard second order finite difference stencil. This approach simplies into a form that is easily solved using Jacobi or Gauss-Seidel iterative techniques:

with Dirichlet boundary conditions is discretized as

An initial grid is shown below in (a), and the resulting grid (after iterations) is shown in (b).

Note that the grid spacing near the curved section increases and then decreases as we move left to right, and that grid lines near the left and right boundaries are not very orthogonal. These issues are reasons that production grid generation techniques are usually more complicated. The addition of control functions allows for better grid clustering properties, which will be necessary for viscous flow simulations.

Stretched grid method

From Wikipedia, the free encyclopedia Jump to: navigation, search

The stretched grid method (SGM) is a numerical technique for finding approximate solutions of various mathematical and engineering problems that can be related to an elastic grid behavior. In particular, meteorologists use the stretched grid method for weather prediction[1] and engineers use the stretched grid method to design tents and other tensile structures.

Contents

1 FEM/BEM mesh refinement 2 Minimum surface problem solution 3 Tensile fabric structures form finding 4 Unfolding problem and cutting pattern generation 5 References 6 External links

FEM/BEM mesh refinement

In recent decades the finite element and boundary element methods (FEM and BEM) have become a mainstay for industrial engineering design and analysis. Increasingly larger and more complex designs are being simulated using the FEM or BEM. However, some problems of FEM/BEM Engineering analysis are still on the cutting edge. The first problem is a reliability of engineering analysis that strongly depends upon the quality of initial data generated at the preprocessing stage. It is known that automatic element mesh generation techniques at this stage have become commonly used tools for the analysis of complex real-world models.[2] With FEM/BEM increasing popularity comes the incentive to improve automatic meshing algorithms. However, all of these algorithms can create distorted and even unusable grid elements. Fortunately, several techniques exist which can take an existing mesh and improve its quality. For instance smoothing (also referred to as mesh refinement) is one of such methods, which repositions nodal locations, so as to minimize element distortion. The Stretched Grid Method (SGM) allows the obtaining of pseudo-regular meshes very easily and quickly in a one-step solution(see [3]). Let us assume that there is an arbitrary triangle grid embeded into plane polygonal singlecoherent contour and produced by an automeshing procedure (see fig.1) It may be assumed further that the grid considered as a physical nodal system is distorted by a number of distortions. It is supposed that the total potential energy of this system is proportional to the length of some -dimensional vector with all network segments as its components.

Fig. 1 A triangle grid bounded by plane polygonal single-coherent contour

Thus, the potential energy takes the following form

where - total number of segments in the network, - The length of segment number , - an arbitrary constant. The length of segment number may be expressed by two nodal co-ordinates as

It may also be supposed that co-ordinate vector distorted network and co-ordinate vector expression for vector may be written as

of all nodes is associated with nonis associated with the distorted network. The

The vector determination is related to minimization of the quadratic form incremental vector , i.e.

by

where - the number of interior node of the area, - the number of co-ordinate After all transformations we may write the following two independent systems of linear algebraic equations

where - symmetrical matrix in the banded form similar to global stiffness matrix of FEM assemblage, and and 1, 2. - incremental vectors of co-ordinates of all nodes at axes 1, 2, - the right part vectors that are combined by co-ordinates of all nodes in axes

Fig. 2 Left: distorted 2D grid, right: corrected grid

The solution of both systems, keeping all boundary nodes conservative, obtains new interior node positions corresponding to a non-distorted mesh with pseudo-regular elements. For example, Fig. 2 presents the rectangular area covered by a triangular mesh. The initial auto mesh possesses some degenerative triangles (left mesh). The final mesh (right mesh) produced by the SGM procedure is pseudo-regular without any distorted elements.

As above systems are linear, the procedure elapses very quickly to a one-step solution. Moreover, each final interior node position meets the requirement of co-ordinate arithmetic mean

of nodes surrounding it and meets the Delaunay criteria too. Therefore, the SGM has all the positive values peculiar to Laplacian and other kinds of smoothing approaches but much easier and reliable because of integer-valued final matrices representation. Finally, the described above SGM is perfectly applicable not only to 2D meshes but to 3D meshes consisting of any uniform cells as well as to mixed or transient meshes.

Minimum surface problem solution

Mathematically the surface embedded into a non-plane closed curve is called minimum if its area is minimal amongst all the surfaces passing through this curve. The best-known minimum surface sample is a soap film bounded by wire frame. Usually to create a minimum surface, a fictitious constitutive law, which maintains a constant prestress, independent of any changes in strain, is used.[4] The alternative approximated approach to the minimum surface problem solution is based on SGM. This formulation allows one to minimize the surface embedded into non-plane and plane closed contours.

Fig 3. Catenoidal surface

The idea is to approximate a surface part embeded into 3D non-plane contour by an arbitrary triangle grid. To converge such triangle grid to grid with minimum area one should solve the same two systems described above. Increments of the third nodal co-ordinates may be determined additionally by similar system at axis 3 in the following way

Solving all three systems simultaneously one can obtain a new grid that will be the approximating minimal surface embedded into non-plane closed curve because of the minimum of the function where parameter . As an example the surface of catenoid which is calculated by the described above approach is presented in Fig 3. The radii of rings and the height of catenoid are equal to 1.0. The numerical area of catenoidal surface determined by SGM is equal to 2,9967189 (exact value is 2.992).

Tensile fabric structures form finding

Fig. 4 Hypar (hyperbolic paraboloid)

Fig. 5 Saddle-type awning

For structural analysis, the configuration of the structure is generally known priori. This is not the case for tension fabric structures. Since the membrane in a tension structure possesses no flexural stiffness, its form or configuration depends upon initial prestressing and the loads to which it is subjected. Thus, the load-bearing behaviour and the shape of the membrane cannot be separated and cannot be generally described by simple geometric models only. The membrane shape, the loads on the structure and the internal stresses interact in a non-linear manner to satisfy the equilibrium equations.

Fig. 6 The dancefloor cover grid model

Fig. 7 Render of the dancefloor cover

Fig. 8 Real dancefloor cover

The preliminary design of tension structures involves the determination of an initial configuration referred to as form finding. In addition to satisfying the equilibrium conditions, the initial configuration must accommodate both architectural (aesthetics) and structural (strength and stability) requirements. Further, the requirements of space and clearance should be met, the membrane principal stresses must be tensile to avoid wrinkling, and the radii of the doublecurved surface should be small enough to resist out-of-plane loads and to insure structural stability (work [5]). Several variations on form finding approaches based on FEM have been developed to assist engineers in the design of tension fabric structures. All of them are based on the same assumption as that used for analysing the behaviour of tension structures under various loads. However, as it is noted by some researchers it might sometimes be preferable to use the so called minimal surfaces in the design of tension structures. The physical meaning of SGM consists in convergence of the energy of an arbitrary grid structure embeded into rigid (or elastic) 3D contour to minimum that is equivalent to minimum

sum distances between arbitrary pairs of grid nodes. It allows the minimum surface energy problem solution substituting for finding grid structure sum energy minimum finding that provides much more plain final algebraic equation system than the usual FEM formulation. The generalized formulation of SGM presupposes a possibility to apply a set of outer forces and rigid or elastic constrains to grid structure nodes that allows the modelling of various outer effects. We may obtain the following expression for such SGM formulation

where
- total number of grid segments, - total number of nodes, - length of segment number , - stiffness of segment number , - coordinate increment of node at axis , - stiffness of an elastic constrain in node at axis , - outer force in node at axis .

Unfolding problem and cutting pattern generation

Once a satisfactory shape has been found, a cutting pattern may be generated. Tension structures are highly varied in their size, curvature and material stiffness. Cutting pattern approximation is strongly related to each of these factors. It is essential for a cutting pattern generation method to minimize possible approximation and to produce reliable plane cloth data. The objective is to develop the shapes described by these data, as close as possible to the ideal doubly-curved strips. In general, cutting pattern generation involves two steps. First, the global surface of a tension structure is divided into individual cloths. The corresponding cutting pattern at the second step can be found by simply taking each cloth strip and unfolding it on a planar area. In the case of the ideal doubly-curved membrane surface the subsurface cannot be simply unfolded and they must be flattened. For example in,[6][7] SGM has been used for the flattening problem solution. The cutting pattern generation problem is actually subdivided into two independent formulations. These are the generation of a distortion-free plane form unfolding each cloth strip and flattening double-curved surfaces that cannot be simply unfolded. Studying the problem carefully one can notice that from the position of differential geometry both formulations are the same. We may consider it as an isometric mapping of a surface onto the plane area that will be conformal mapping and equiareal mapping simultaneously because of

invariant angles between any curves and invariance of any pieces of area. In the case of singlecurved surface that can be unfolded precisely equi-areal mapping allows one to obtain a cutting pattern for fabric structure without any distortions. The second type of surfaces can be equi-areal mapped only approximately with some distortions of linear surface elements limited by the fabric properties. Let us assume that two surfaces are parameterized so that their first quadratic forms may be written as follows

The condition of conformity for two surfaces as is formulated in differential geometry requires that

where :

- the ratio of the surface distortion due to conformal mapping.

It is known that the first quadratic form reflects the distance between two surface points and . When -ratio is close to 1 the above eqn converges to condition of isometric mapping and to equi-areal mapping respectively because of invariant angles between any curves and invariance of any pieces of area. Remembering that the first stage of form finding is based on triangular mesh of a surface and using the method of weighted residuals for the description of isometric and equi-areal mapping of the minimum surface onto a plane area we may write the following function which is defined by the sum of integrals along segments of curved triangles

where
- total number of grid cells, - weight ratios, - the total mapping residual, - the constant that does not influence the final result and may be used as a scale ratio.

Considering further weight ratios we may transform eqn. into approximate finite sum that is a combination of linear distances between nodes of the surface grid and write the basic condition of equi-areal surface mapping as a minimum of following non-linear function

where
- initial length of linear segment number , - final length of segment number , - distortion ratio close to 1 and may be different for each segment.

The initial and final lengths of segment number may be expressed as usual by two nodal coordinates as

where
- co-ordinates of nodes of the initial segment, - co-ordinates of nodes of the final segment.

According to the initial assumption we can write mapping. The expression for vectors and be written as

for the plane surface with co-ordinate increments term use may

Fig. 9 Cut out of the twin-peaks awning

Fig. 10 Initial form of the patch

Fig. 11 Plane patch pattern

The vector

definition is made as previously

After transformations we may write the following two independent systems of non-linear algebraic equations

where all the parts of the system can be expressed as previously and vectors of pseudo-stresses at axes 1, 2 that has the following form

and

are

where
- total number of nodes that surround node number , - the number of global axes.

The above approach is another form of SGM and allows the obtaining of two independent systems of non-linear algebraic equations that can be solved by any standard iteration procedure. The less Gaussian curvature of the surface is the higher the accuracy of the plane mapping is. As a rule, the plane mapping allows to obtain a pattern with linear dimensions 1-2% less than corresponding spatial lines of a final surface. That is why it is necessary to provide the appropriate margins while patterning. The typical sample of cut out is presented in Figs. 9, 10, 11.

Mesh adaptation
Meshing
Introduction Mesh classification Structured mesh generation

Algebraic grid generation Elliptic grid generation

Unstructured mesh generation Special topics

Mesh adaptation First cell height calculation

<< Unstructured Mesh Generation | Mesh Adaptation | First cell height calculation>>

Mesh adaptation, often referred to as Adaptive Mesh Refinement (AMR), refers to the modification of an existing mesh so as to accurately capture flow features. Generally, the goal of these modifications is to improve resolution of flow features without excessive increase in computational effort. We shall discuss in brief on some of the concepts important in mesh adaptation. Mesh adaptation strategies can usually be classified as one of three general types: r-refinement, h-refinement, or p-refinement. Combinations of these are also possible, for example hprefinement and hr-refinement. We summarise these types of refinement below.

r-Refinement
r-refinement is the modification of mesh resolution without changing the number of nodes or cells present in a mesh or the connectivity of a mesh. The increase in resolution is made by moving the grid points into regions of activity, which results in a greater clustering of points in those regions. The movement of the nodes can be controlled in various ways. On common technique is to treat the mesh as if it is an elastic solid and solve a system equations (suject to some forcing) that deforms the original mesh (see for instance Stretched grid method). Care must be taken, however, that no problems due to excessive grid skewness arise.

h-Refinement
h-refinement is the modification of mesh resolution by changing the mesh connectivity. Depending upon the technique used, this may not result in a change in the overall number of grid cells or grid points. The simplest strategy for this type of refinement subdivides cells, while more complex procedures may insert or remove nodes (or cells) to change the overall mesh topology. In the subdivision case, every "parent cell" is divided into "child cells". The choice of which cells are to be divided is addressed below. For every parent cell, a new point is added on each face. For 2-D quadrilaterals, a new point is added at the cell centroid also. On joining these points, we get 4 new "child cells". Thus, every quad parent gives rise to four new offsprings. The

advantage of such a procedure is that the overall mesh topology remains the same (with the child cells taking the place of the parent cell in the connectivity arrangement). The subdivision process is similar for a triangular parent cell, as shown below. It is easy to see that the subdivision process increases both the number of points and the number of cells.

An additional point to be noted is that this type of mesh adaptation can lead to what are called "hanging nodes." In 2D, this happens when one of the cells sharing a face is divided and the other is not, as shown below. For two quad cells, one cell is divided into four quads and other remains as it is. The highlighted node is the hanging node.

This leads to a node on the face between the two cells which does not belong (properly) to both of the cells. The node "hangs" on the face, and one of the cells becomes an arbitrary polyhedron. In the above case, the topology seemingly remains same, but the right (undivided) cell actually has five faces. Further, there are two ways of h-refinement: Isotropic and Anisotropic. Isotropic refinement, is what was described a little back. In isotropic refinement, new points are added in both the directions, say x and y. In anisotropic refinement, the division takes place in one predominant direction. Thus, in short, an isotropic refinement for a quad would produce four new offsprings,

while anisotropic refinement would only generate two. Anisotropic refinement is made use of, when the user knows that the flow feature is predominantly to be resolved in one direction, for e.g. Boundary Layers. However, there are situations where an anisotropic refinement alone may not be satisfactory, such as a shock-boundary layer interaction. The simplest reifnement anyone can think of is to divide all cells in the domain. This is referred to as "Uniform Refinement". Although it does improve the solution vastly, it is easy to realise that we are going for a huge unwanted effort in doing so. For e.g., in the farfield region of an airfoil, cell division is not bringing in any improvement because the flow there does not experience any activity. Therefore, to achieve the goal of mesh adaptation, the refinement is done at "selected" regions alone based on certain criterion. This is referred to popularly as AMR or Adaptive Mesh Refinement. It is to be remarked that AMR does not only encompass division of cells into smaller ones ( Refinement ), but also the agglomeration of smaller cells into a larger one (De-refinement or coarsening), when the need arises.

p-Refinement
A very popular tool in Finite Element Modelling (FEM) rather than in Finite Volume Modelling (FVM), it achieves increased resolution by increasing the order of accuracy of the polynomial in each element (or cell). In AMR, the selction of "parent cells" to be divided is made on the basis of regions where there is appreciable flow activity. It is well known that in compressible flows, the major features would include Shocks, Boundary Layers and Shear Layers, Vortex flows, Mach Stem , Expansion fans and the like. It can also be seen that each feature has some "physical signature" that can be numerically exploited. For eg. shocks always involve a density/pressure jump and can be detected by their gradients, whereas boundary layers are always associated with rotationality and hence can be detected using curl of velocity. In compressible flows, the velocity divergence, which is a measure of compressiblity is also a good choice for shocks and expansions. These sensing paramters which can indicate regions of flow where there are activity are referred to as ERROR INDICATORS and are very popular in AMR for CFD. Just as refinement is possible by ERROR INDICATORS as mentioned above, certain other issues also assume relevance. Error Indicators do detect regions for refinement, they do not actually tell if the resolution is good enough at any given time. In fact the issue is very severe for shocks, the smaller the cell, the higher the gradient and the indicator would keep on picking the region, unless a threshold value is provided. Further, many users make use of conservative values while refining a domain and generally end up in refining more than the essential portion of the grid, though not the complete domain. These refined regions are unneccesary and are in strictest sense, contribute to unneccesary computational effort. It is at this juncture, that reliable and resonable measure of cell error become necessary to do the process of "coarsening", which would reduce the above-said unnecessary refinement, with a view towards generatin an "optimal mesh". The measures are given by sensors referred to as ERROR ESTIMATORS, literature on which is in abandunce in FEM, though these are very rare in FVM.

Control of the refinement and/or coarsening via the error indicators is often undertaken by using either the 'solution gradient' or 'soultion curvature'. Hence the refinement variable coupled with the refinement method and its limits all need to be considered when applying mesh adaptation. Journal bearing operation

The figure demonstrates a hydrodynamic journal bearing and a journal rotating in a clockwise direction. Journal rotation causes pumping of the lubricant (oil) flowing around the bearing in the rotation direction. If there is no force applied to the journal its position will remain concentric to the bearing position. However a loaded journal displaces from the concentric position and forms a converging gap between the bearing and journal surfaces. The pumping action of the journal forces the oil to squeeze through the wedge shaped gap generating a pressure. The pressure falls to the cavitation pressure (close to the atmospheric pressure) in the diverging gap zone where cavitation forms. According to Michael M. Khonsari and E. Richard Booser two types of cavitation may form in journal bearing:

Gaseous cavitation associated with air and other gases dissolved in oil. If the oil pressure falls below the atmospheric pressure the gases tend to come out of the oil forming gaseous cavitation voids. The cavities are carried by the circulating oil to the pressurized converging gap where they redissolve in the oil and disappear without any damaging effect. Vapor cavitation forms when the load applied to the bearing fluctuates at high frequency (e.g. bearings in high RPM internal combustion engines). The oil pressure instantly falls causing formation of cavities due to fast evaporation (boiling). When the pressure rises the vapor cavities (cavitation bubbles) contract at high velocity. Such collapse results in impact pressure, which may erode the bearing material.

The oil pressure creates a supporting force separating the journal from the bearing surface. The force of oil pressure and the hydrodynamic friction force counterbalance the external load F. The final position of the journal is determined by the equilibrium between the three forces. In the hydrodynamic regime the journal climbs in the rotation direction (left side of the bearing). If the journal works in boundary and mixed lubrication the hydrodynamic pressure force disappears (the other two forces remain). Thus, the climbing direction is opposite to the rotation direction and the journal rolls up the right side of the bearing. to top Reynolds Equation Lubricant pressure distribution as a function of journal speed, bearing geometry, oil clearance and lubricant viscosity is described by Reynolds equation:

Where: h local oil film thickness, dynamic viscosity of oil, p local oil film pressure, U linear velocity of journal, x - circumferential direction. z - longitudinal direction. Close form solution of Reynolds equation can not be obtained therefore finite elements method is used to solve it. Analytical solutions of Reynolds equation exist only for certain assumptions:

Sommerfeld Solution

The equation is solved with the assumption that there is no lubricant flow in the axial direction (infinitely long bearing assumption).

Ocvirk Solution

Ocvirk solution for infinitely short bearing assumption neglects circumferential pressure gradients (first term of Reynolds equation).

Where: Cr radial clearance Cr = (D-Dj)/2, r bearing radius, Dj - journal diameter, - eccentricity ratio = e/Cr e - absolute bearing eccentricity, B bearing length, p0 - cavitation pressure, So - Sommerfeld number (see below). Sommerfeld and Ocvirk solutions are applicable only in the region of positive pressure. to top Engineering calculations of journal bearing performance Engineering analysis of a journal bearing performance may be made by using Raimondi-Boyd charts containing numerical solutions of Reynolds equation for bearings with different length-todiameter ratio. These engineering calculations are based on the dimensionless bearing parameter called Sommerfeld number: So = F /(DB) Where: F bearing load, relative bearing clearance = (D-Dj)/D, B bearing length,

D bearing diameter, Dj - journal diameter, dynamic viscosity of oil, angular speed of journal. Raimondi-Boyd analysis is based on the assumption that the oil viscosity does not change when it passes through the bearing. Since the real oil heats up in the bearing and its viscosity drops it is assumed that oil viscosity is determined at the effective temperature Teff = Ten + 0.5Tas (Tas is assumed temperature rise). The calculation procedure is as follows:
1. 2. 3. 4. 5. 6. 7. 8. Temperature rise Tas is assumed. Effective temperature Teff = Ten + 0.5Tas Oil viscosity at the temperature Teff is determined (viscosity-temperature chart). Relative bearing clearance is calculated = (D-Dj)/D Sommerfeld number is calculated: So = F /(DB) Values of Sommerfeld number and geometry parameter B/D determine the bearing eccentricity , which is obtained from the eccentricity chart (So, B/D). Minimum oil film thickness hmin = 0.5D(1 ) Eccentricity value and the geometry ratio B/D are used for determination specific coefficient of friction f(, B/D), angular position of the minimum oil film thickness (, B/D) and oil flow parameter Q(, B/D). Coefficient of friction f = f Oil flow rate Q = D3Q Power loss due to friction P = fFD/2 Calculated oil temperature rise Tca = P/(cQ) where c is specific heat capacity of oil Calculated oil temperature rise Tca is compared to the assumed value Tas and if there is a significant difference between them a new assumption of the temperature rise is made.

9. 10. 11. 12. 13.

The above calculation procedure is repeated for several iterations un

Hydrodynamic lubrication theory

Dr. Dmitri Kopeliovich A theoretical analysis of hydrodynamic lubrication was carried out by Osborne Reynolds. The equations resulted from the analysis has served a basis for designing hydrodynamically lubricated bearings. The following assumptions were made by Reynolds in the analysis:

The lubricating fluid is Newtonian - the flow is laminar and the shear stress between the flow layers is proportional to the velocity gradient in the direction perpendicular to the flow (Newtons law of viscosity):

Where: dynamic viscosity of oil, v linear velocity of the laminar layer, y - the axis perpendicular to the flow direction.

The inertia forces resulted from from the accelerated movement of the flowing lubricant are neglected. The lubricating fluid is incompressible. The pressure of the fluid p is constant in the direction perpendicular to the laminar flow: dp/dy=0 (assumption of thin lubrication film). The viscosity of the fluid is constant throughout the lubrication film.

Consider the equilibrium of a unit volume in the lubricant film.

The pressure forces act on the right and the left faces of the unit volume. The shear forces resulting from the relative motion of the laminar layers act along the upper and the lower faces of the unit volume. Assuming that there is no flow in z direction the equation of the equilibrium of the forces in the direction of the flow is as follows:

Substitution of from Newtons law of viscosity (1) results in:

The velocity function is obtained by integrating the equation (2) with respect to y:

The constants of integration C1 and C2 may be determined from the boundary conditions: v=U when y=0, then C2=U v=0 when y=h, then:

Substituting C1 and C2 in (3) we obtain:

The total flow of the lubricant is:

Substituting v from (4) we get:

According to the assumption about incompressibility of the lubricant the flow Q does not change in x direction:

Differentiating the equation (5) with respect x results in:

This is Reynolds equation for one dimensional flow. It can be used with the assumption of no flow in z direction (bearings with infinite length). If the flow in z direction is taken into account (bearings with side leakage of the lubricating fluid) then the analysis results in Reynolds equation for two dimensional flow:

Where: h local oil film thickness, dynamic viscosity of oil, p local oil film pressure, U linear velocity of journal, x - circumferential direction.

z - longitudinal direction. Close form solution of Reynolds equation can not be obtained therefore finite elements method is used to solve it. Analytical solutions of Reynolds equation exist only for certain assumptions:

Sommerfeld Solution

The equation is solved with the assumption that there is no lubricant flow in the axial direction (infinitely long bearing assumption).

Ocvirk Solution

Ocvirk solution for infinitely short bearing assumption neglects circumferential pressure gradients (first term of Reynolds equation).

Where:

Cr radial clearance Cr = (D-Dj)/2, r bearing radius, Dj - journal diameter, - eccentricity ratio = e/Cr e - absolute bearing eccentricity, B bearing length, p0 - cavitation pressure,

So - Sommerfeld number (see below). So = F /(DB) Where: F bearing load, relative bearing clearance = (D-Dj)/D, D bearing diameter, angular speed of journal. Sommerfeld and Ocvirk solutions are applicable only in the region of positive pressure. Mechanisms of wear
Dr. Dmitri Kopeliovich Wear is the removal of the material from the surface of a solid body as a result of mechanical action of the counterbody. Wear of Engine bearings is the removal of the bearing material from its surface caused by sliding friction between the bearing and the crankshaft. Wear may combine effects of various physical and chemical processes proceeding during the friction between two counteracting materials: micro-cutting, micro-ploughing, Plastic deformation, cracking, fracture, welding, melting, chemical interaction. The mechanisms of wear:

Abrasive wear

Adhesive wear Fatigue wear Corrosive wear Erosive wear

Abrasive wear

Abrasive wear occurs when a harder material is rubbing against a softer material.

If there are only two rubbing parts involved in the friction process the wear is called two body wear.

In this case the wear of the softer material is caused by the asperities on the harder surface.

If the wear is caused by a hard particle (grit) trapped between the rubbing surfaces it is called three body wear. The particle may be either free or partially embedded into one of the mating materials.

In the micro-level abrasive action results in one of the following wear modes:

Ploughing. The material is shifted to the sides of the wear groove. The material is not removed from the surface. Cutting. A chip forms in front of the cutting asperity/grit. The material is removed (lost) from the surface in the volume equal to the volume of the wear track (groove). Cracking (brittle fracture). The material cracks in the subsurface regions surrounding the wear groove. The volume of the lost material is higher than the volume of the wear track.

to top

Adhesive wear

Adhesion wear is a result of micro-junctions caused by welding between the opposing asperities on the rubbing surfaces of the counterbodies. The load applied to the contacting asperities is so high that they deform and adhere to each other forming micro-joints. The motion of the rubbing counterbodies result in rupture of the micro-joints. The welded asperity ruptures in the non-deformed (non-cold worked) regions. Thus some of the material is transferred by its counterbody. This effect is called scuffing or galling. When a considerable areas of the rubbing surfaces are joined during the friction a Seizure resistance (compatibility) seizure of one of the bodies by the counterbody may occur. The factors decreasing adhesive wear:

Lower load. Harder rubbing materials. Contaminated rubbing surfaces. Presence of solid lubricants. Presence of a lubrication oil. Anti-wear additives in oil.

to top Fatigue wear

Fatigue wear of a material is caused by a cycling loading during friction. Fatigue occurs if the applied load is higher than the fatigue strength of the material. Fatigue cracks start at the material surface and spread to the subsurface regions. The cracks may connect to each other resulting in separation and delamination of the material pieces. One of the types of fatigue wear is fretting wear caused by cycling sliding of two surfaces across each other with a small amplitude (oscillating). The friction force produces alternating compression-tension stresses, which result in surface fatigue. Fatigue of overlay of an engine bearing may result in the propagation of the cracks up to the intermediate layer and total removal of the overlay. to top Corrosive wear Wear may be accelerated by corrosion (oxidation) of the rubbing surfaces. Increased temperature and removal of the protecting oxide films from the surface during the friction promote the oxidation process. Friction provides continuous removal of the oxide film followed by continuous formation of new oxide film. Hard oxide particles removed from the surface and trapped between the sliding/rolling surfaces additionally increase the wear rate by three body abrasive wear mechanism. to top Erosive wear

Erosive wear is caused by impingement of particles (solid, liquid or gaseous), which remove fragments of materials from the surface due to momentum effect. Erosive wear of Engine bearings may be caused by cavitation in the lubrication oil. The cavitation voids (bubbles) may form when the oil exits from the convergent gap between the bearing and journal surfaces. The oil pressure rapidly drop providing conditions for voids formation (the pressure is lower than the oil vapor pressure). The bubbles (voids) then collapse producing a shock wave, which removes particles of the bearing material from the bearinINTRODUCTION

Most bearings can be described as belonging to one of four classes: (1) rolling element bearings (examples: ball, cylindrical roller, spherical roller, tapered roller, and needle), (2) dry bearings (examples: plastic bushings, coated metal bushings), (3) semi-lubricated (example: oilimpregnated bronze bushings) and (4) fluid film bearings (example: crankshaft bearings). Aside from an occasional tangent like the Porsche 1.5 litre flat four engine of the sixties and certain radial-configuration aircraft engines, almost all piston engines use fluid film bearings. This is true for the crankshaft and sometimes the camshaft, although often the latter runs directly in the engine structure. He we put the spotlight onto fluid film bearings. The point of the whole discussion is (a) to explain how fluid film bearings work (which is sometimes counterintuitive) and (b) to demonstrate how engine designers are reducing friction losses through bearing technology. Fluid film bearings operate by generating, as a by-product of the relative motion between the shaft and the bearing, a very thin film of lubricant at a sufficiently-high pressure to match the applied load, as long as that load is within the bearing capacity. Fluid film bearings represent a form of scientific magic, by virtue of providing very large load carrying capabilities in a compact, lightweight implementation, and unlike the other classes, in most cases can be designed for infinite life. Fluid film bearings operate in one of three modes: (a) fully-hydrodynamic, (b) boundary, and (c) mixed.

HYDRODYNAMIC MODE
In fully hydrodynamic (or "full-film") lubrication, the moving surface of the journal is completely separated from the bearing surface by a very thin film of lubricant (as little as 0.0001" with isotropic-superfinished {ISF} surfaces). The applied load causes the centreline of the journal to be displaced from the centreline of the bearing. This eccentricity creates a circular "wedge" in the clearance space, as shown in Figure 1.

Figure 1 The lubricant, by virtue of its viscosity, clings to the surface of the rotating journal, and is drawn into the wedge, creating a very high pressure (sometimes in excess of 6,000 psi), which acts to separate the journal from the bearing to support the applied load. The bearing eccentricity is expressed as the centreline displacement divided by the radial clearance. For example, if a bearing which has 0.0012" radial clearance (0.0024" diametral) is operating with a film thickness of 0.0001", then the eccentricity is (.0012 - .0001)/.0012 = 0.917. The bearing eccentricity increases with applied load and decreases with greater journal speed and viscosity. Note that the hydrodynamic pressure has no relationship at all to the engine oil pressure, except that if there is insufficient engine oil pressure to deliver the required copious volume of oil into the bearing, the hydrodynamic pressure mechanism will fail and the bearing(s) and journal(s) will be quickly destroyed. It is interesting to study the pressure distribution in the hydrodynamic region of a fluid film bearing. The hydrodynamic pressure described above increases from quite low in the large clearance zone to its maximum at the point of minimum film thickness as oil (essentially incompressible) is pulled into the converging "wedge" zone of the bearing. Figure 2 shows a representative sketch of the radial pressure distribution in the load-supporting area of the bearing.

Figure 2 However, this radial profile does not exist homogeneously across the axial length of the bearing. Figure 3 shows a sketch of the axial pressure distribution profile for fully-developed hydrodynamic lubrication with a non-grooved bearing surface (insert). As the picture shows, the pressure drops off rapidly at the edge of the bearing, because oil is leaking out of the edges under the influence of the high hydrodynamic pressure. Moving inward from the edges, the pressure rises dramatically. If the bearing has sufficient width, the profile will have a nearly flat shape across the high-pressure region.

Figure 3 Long ago, it was standard practice to use fully-grooved main bearings, the thought being that the groove would provide a better supply of oil to the conrod bearings. A quick study of the axial profile of the hydrodynamic pressure distribution for a grooved surface (insert), shown in Figure 4, demonstrates how any interruption of the smooth surface of the bearing in the load-carrying region will severely degrade the capacity of the bearing.

Figure 4

BOUNDARY MODE
The second mode of bearing operation is boundary lubrication. In boundary lubrication, the "peaks" of the sliding surfaces (journal and bearing) are touching each other, but there is also an extremely thin film of oil only a few molecules thick which is located in the surface "valleys". That thin film tends to reduce the friction from what it would be if the surfaces were completely dry.

MIXED MODE
The third mode, mixed, is a region of transition between boundary and full-film lubrication. The surface peaks on the journal and bearing surfaces partially penetrate the fluid film and some surface contact occurs, but hydrodynamic pressure is beginning to increase.

OPERATION
To further explain the three lubrication modes, let's examine the operation of a journal bearing from startup to steady state. Figure 5 shows a journal bearing at rest. The applied load causes the journal to contact the bearing surface (eccentricity ratio = 1.0).

Figure 5 When motion begins, the journal tries to climb the wall of the bearing, as illustrated in Figure 6, due to the metal-to-metal friction (boundary lubrication) between the two surfaces.

Figure 6 If there is an adequate supply of lubricant, the motion of the journal begins to drag the lubricant into the wedge area and hydrodynamic lubrication begins to occur along with the boundary lubrication (mixed lubrication). Assuming the load and viscosity remain relatively constant during this startup period, then as RPM increases, the hydrodynamic operation strengthens until it is fully-developed and it moves the journal into its steady-state orientation (Figure 7), defined by the eccentricity (e) and the orientation angle (a). Note that the direction of the eccentricity, and hence the minimum film thickness, do not occur in line with the load vector, but are angularly displaced from the load.

Figure 7 There are three parameters which determine the mode (boundary, mixed, hydrodynamic) in which a given bearing will operate: (1) the speed of the shaft, (2) the viscosity of the lubricant, and (3) the applied unit load. These three parameters can be combined in the following way to form a value we can call "Bearing Operating Condition" (BOC). BOC = Viscosity x RPM x Diameter x K / Unit Load (Equation 1) The Viscosity parameter is in units of absolute viscosity. The "K" value is a factor which converts RPM and Diameter into journal surface speed. The bearing unit load is the applied force divided by the projected area of the bearing (the insert width times the journal diameter).

ZN/P CURVE ("STRIBECK PLOT")

The BOC value will predict the operating mode of a bearing and the expected friction coefficient for that operating condition. The transitions between these different operating modes, and the

related friction properties are illustrated more fully in the Stribeck Plot shown in below in Figure 8. This plot (also known as a "ZN/P Curve") shows the bearing coefficient of friction (on a logarithmic scale) plotted as a function Bearing Operating Condition (BOC). The values plotted on the X-axis are nondimensionalized, and are shown as a percentage of full scale.

Figure 8 The two vertical lines in the plot area show the boundaries between the three operating modes. Area 1, from BOC = zero to about 15, is where boundary lubrication occurs. Area 2 (BOC = 15 to 35) is the region of mixed lubrication, in which, as BOC increases, the hydrodynamic pressure is developing and taking over from boundary lubrication. Area 3 is fully-developed hydrodynamic lubrication.

Note that the purpose of presenting this BOC (or ZN/P) curve is to demonstrate the interrelationship between friction coefficient and the BOC (ZN/P) parameters, not to instruct in bearing design. In the definitive 2001 reference text "Applied Tribology: Bearing Design and Lubrication" by Dr. Michael Khonsari and Dr. Richard Booser (ref-2:6:12), the Stribeck Plot is shown on Page 12 and is described as a "dimensionless uN/p curve relating lubrication regime and friction coefficient to absolute viscosity". That same DIMENSIONLESS curve ("ZN/P") is shown on page 2097 of "Machinery's Handbook, 24th ed." (ref-2:22:2097) The "BOC" entity (often known as ZN/P) does indeed have units, which depend completely on the units you choose for (a) surface speed converted to RPM and (b) unit load: psi, n/mm, mpa, etc. Various engineering texts use specific portions of the curve and use whatever ZN/P units they prefer. Others retain the dimensionless construct. The friction coefficient values shown in Figure 8 were taken from both "Machinery's Handbook, 24th ed." and from "Design of Machine Elements", by M.F. Spotts, Professor of Mechanical Engineering, Northwest University (ref-2:2:302). Both reference works agreed that the lowpoint is about 0.001, the fluid film range is from 0.001 to at least 0.005, the boundary region is greater than 0.1 to as high as 0.03, and the mixed region is between the other two, as shown on the plot. This curve illustrates that when operating in the hydrodynamic region (Area 3), if the unit load remains constant and either rpm or viscosity increase, the hydrodynamic pressure increases, the eccentricity decreases and the friction coefficient rises, increasing by a factor of 10 as eccentricity approaches zero. However, if rpm remains fixed and either viscosity decreases or unit load increases, then the BOC will decrease. Friction coefficient decreases logarithmically down to the low point at around BOC = 35. If the unit load continues to increase and/or viscosity continues to decrease, the BOC will move into the mixed lubrication region and the lubrication mode will change from fully hydrodynamic back to the mixed mode and friction will increase dramatically. If the load increases and/or viscosity decreases even further, BOC continues to decrease, and eventually the journal asperities break through the film and the system reverts back to the very-high-friction boundary lubrication mode. Note the values for friction coefficient. In the area of boundary lubrication, the friction coefficient is similar to that of a dry bearing (0.25-0.35). At the BOC value of 35, the friction coefficient is in the remarkably low region of 0.001, which is 50% less than the friction coefficient of deep-groove ball bearings. As the BOC increases (any combination of smaller load, higher rpm, higher viscosity) the curve shows that friction coefficient rises exponentially, approaching a value of 0.01, ten times greater than the ideal minimum. That fact illustrates why there is so much attention paid to optimizing the bearings for the application, trying to maintain the BOC in the 35-50 range.

In issues past, we have seen that combustion loads can apply forces in excess of 12,000 pounds to a rod journal. If the bearing were operating at a friction coefficient of 0.002, (BOC roughly 50), an applied load of 12,000 pounds would generate a friction load on the surface of one bearing of 24 pounds. If the diameter of the journal carrying the 12,000 lb. is 2.50", then the friction torque lost to that bearing will be 24 lbs x 1.25" = 30 lb-in or 2.5 lb-ft. If all 5 main journals carry the same load, then the friction torque lost to the main bearings alone is 5 x 2.5 = 12.5 lb-ft, which at 9000 rpm, absorbs 21.4 HP. If that journal diameter were reduced to 2.00", one might think that a 20% reduction in main bearing friction torque could be realized. However, for the same bearing width, reducing the journal diameter 20% reduces the projected area by 20%, which increases the unit loading, resulting in a reduced BOC for the same load, rpm and viscosity. Further, reducing the journal diameter by 20% also reduces the surface speed by 20%, which for the same RPM and viscosity, lowers BOC even further. Add in the effect of the very low viscosity lubricants some teams use, and the net effect can be a dramatic reduction of BOC. As long as the BOC stays within the hydrodynamic region, the smaller BOC will yield an even lower friction coefficient, which further reduces the bearing friction losses. Of course in practice, it's not that big a payoff, because the 12,000 lb. load is not applied for the whole 360 of rotation. But the illustration serves to point out an area that savvy engine designers have been successfully harvesting.

SQUEEZE-FILM LUBRICATION
There is another form of fluid-film lubrication, which adds to the load capacity in applications with oscillating loads (such as a piston engine), known as squeeze-film lubrication. Squeeze-film action is based on the fact that a given amount of time is required to squeeze the lubricant out of a bearing axially, thereby adding to the hydrodynamic pressure, and therefore to the load capacity. Since there is little or no significant rotating action in the wrist-pin bores, squeeze-film hydrodynamic lubrication is the prevailing mechanism which separates wrist pins from their bores in the rods and pistons.

GEOMETRY
Crankshaft bearings are not round. The main bearing journals and crankpins that run within these (conventionally) plain bearings are perfectly round but the bearing surfaces that surround them are not. For a start, the crush that locates a plain bearing in its housing causes distortion of the housing, the nature of which will reflect the material and geometry of the part forming it. On top of this, these bearings are actually designed to be out of round. If engine load and speed were constant and bearing geometry could always be maintained during operation a perfectly round bearing surface profile would work fine. Of course in the internal combustion engine load and speed do vary constantly and the varying loading imparted to the bearing housing constantly alters its geometry. In fact, the racing engine is an elastic device, to

an extent that is not always fully appreciated. Enormous loads go both up and down the con rod, lengthening and shortening it and distorting the shape of its big and small end. In view of this contemporary steel backed plain bearings are designed to be semi-flexible rather than rigid structures. In The Definitive V8 Engines, we showed that a naturally aspirated 2.4 litre, 750 bhp Formula One V8 running to 20,000 rpm (2006 regulations) is subject to a maximum crankpin load in the region of 13,300 lb while a naturally aspirated 5.86 litre, 850 bhp Cup V8 running to 9500 rpm is subject to about 12,500 lb. Such crankpin loads deform the crankshaft, which in turn transfers deformation to the crankcase through its main bearing journals. Thus in operation, both the rod bearing housing (conrod big end) and the main bearing housings deform. In practice it has been established that the appropriate static profile for a crankshaft bearing is normally oval, having its minimum diameter in line with the direction of maximum load. Generally this is taken to be at 90 degrees to the parting line. Bearings are therefore typically manufactured with a wall thickness that is greatest at 90 degrees to the parting line, tapering off from that point to the parting line each side by a specified amount. This is known as bearing ovality (sometimes called "eccentricity", but that usage can be confused with the eccentricity essential to hydrodynamic lubrication) and it is tailored to the characteristics of a specific engine. For example, a heavy piston assembly and high rate of piston acceleration will result in high inertia loading at the top of the exhaust stroke that will cause pronounced stretch of the con rod, this in turn significantly squeezing the big end - a high degree of ovality is required to stop the bearing then pinching the crankpin.

BEARING STRESS
While bearings are a source of friction (including consequent shearing of the oil film) and thus heat, they are also a route for heat to escape from the reciprocating/rotating assembly to the stationary structure of the engine and, more importantly, into the circulating oil. In terms of the stress that the bearings see, it is notable that the magnitude, and sometimes even the direction of the loading varies throughout the course of each stroke. How much stress a given bearing experiences is a function of net loading and bearing projected area, which fluctuates accordingly. Net loading varies dramatically with throttle and rpm, and throughout an engine cycle at any given throttle and rpm. For example, on the power stroke the compression / combustion loading on the con rod is compressive and this opposes the tensile inertia loading caused by piston acceleration. At low engine speed with wide open throttle there is less inertia loading balancing the piston combustion forces and, depending on the engine's torque characteristics this can impart greater net loading to the bearings than WOT operation at higher speeds. Conversely, at engine speeds above peak torque inertia forces come to dominate and the net effect on the bearings is increased loading compared to operation at peak torque rpm. However, the con rod loading that occurs in the vicinity of TDC overlap is extremely high tensile loading because there is very little cylinder pressure to oppose the piston acceleration. That load varies with the square of rpm, and can apply immense loads (and consequent deflections) to the cap-half insert.

Sustained high rpm operation is another threat to the bearings since it causes high temperature running, which in turn can cause excessive oil heating and with that a loss of viscosity. In this respect Cup oval running can be more taxing to the bearings than Formula One road racing. An article in Race Engine Technology, Issue 20, showed an example of cavitation damage on a big end bearing from the Cosworth 2.4 litre V8 engine of 2006, which was designed to run to 20,000 rpm. As the piston approached top dead center the tendency was for the upper portion of the titanium rod's big end to arch away from the steel crankshaft journal and for the steel backed bearing to distort accordingly. There was thus a cavity formed between the bearing and the journal upon which it ran, creating a low-pressure zone in the oil film, encouraging the formation of vapour bubbles. As the piston reversed direction the combustion pressure took out the cavity, collapsing the bubbles, which added to the loading on the big end. In fact, shock waves were formed that stressed the surface of the bearing, to the extent that material could even be lost from it. Following a problem of this nature at the Malaysian Grand Prix the oil viscosity was increased. This avoided any cavitation damage until design changes could be implemented to address the problem. High shear viscosity at high temperature is critical for bearing duty as this extreme example attests. Oil development through 2006 led to a reduction in the variation of viscosity with temperature ("viscosity index").

MATERIALS AND COATINGS

Ideally a bearing material should offer low friction properties, but given that in fullyhydrodynamic operation, the bearing surface is separated from the surface of the journal by a thin film of oil, it is clearly the lubricant rather than the respective surface materials that dominates the friction generated under normal running conditions. Therefore, if there is an adequate supply of lubrication and a suitable load / speed ratio, the material forming the bearing's working surface is not crucial in terms of frictional losses. Inevitably, however, metal-to-metal contact will occur, particularly on start up. The journal is invariably steel, and copper, for example (used as the sole material for some early bearings) running on steel has a kinetic coefficient of 0.36. However, any metal running on steel given proper lubrication has a kinetic coefficient of in the region of 0.06 (it will vary as shown in the Stribeck curve above). In view of the unavoidable metal-to-metal contact, low friction coatings are sometimes applied to bearings. For example, one manufacturer has developed an ultra-slippery moly/graphite blend, which is suspended in an inert PTFE substrate, which provides the adhesion necessary to attach it to the top surface of the bearing. This coating, only one thou thick, which is compatible with contemporary lubricants and lubricant additives, is sacrificial - the bearing will outlive it but in the meantime it is claimed to reduce friction and wear. If there is any contact it will prevent scuffing and even absorb debris.

PLAIN BEARING CONSTRUCTION FUNDAMENTALS

Typically the tri-metal plain bearing common to contemporary high-performance engines is formed as a laminated structure having a relatively thick steel backing layer in contact with the

housing, a harder, thin middle layer (copper-lead, lead-bronze, aluminum-tin, etc.) and a very thin upper layer of soft material (lead, zinc, cadmium, lead-indium, and a host of others), the top layer forming the actual bearing surface. The maximum applied pressure a bearing can carry is determined by the strength and hardness properties of the upper surface. The maximum relative velocity between the journal and the bearing is governed by the bearing's ability to dissipate the heat generated by the shearing of the oil film. Except for the rare instances of built-up crankshafts, the plain bearing is split into upper and lower halves, so that it can be fitted over the journal. One half fits into the main structure, the other into the cap. Each half is known as a shell hence this type of bearing can be referred to as plain or shell-type. Normally only one of the main bearings is designed as the thrust bearing necessary to minimize axial movement of the crankshaft. The multiple layers have been developed to provide the properties required for the specific application. While the backing will invariably be steel, a steel bearing running against a steel journal with no coating on either surface would cause high friction and wear in the boundary and mixed lubrication modes, and would provide little or no ability to allow foreign particles to embed in the material, but would instead capture them and turn them into cutting tools. Therefore, the upper layer is a softer metal, designed for minimum friction with sufficient embed ability. The idea is to allow abrasive particles to embed below the working surface and thereby minimize wear. Moreover, the softer upper layers will help the bearing act as a cushion in the face of severe operating forces. In addition to high mechanical strength and high resistance to temperature the composite bearing needs good conformability and good surface properties - it needs 'compatibility' to prevent pick up or even seizure if the oil film momentarily breaks down. Due to the mechanical properties of the soft bearing material, one might think it would be squeezed out of the bearing due to the forces acting upon it. However, the very thin nature of the soft layer, supported by the much stronger and thicker base layer, prevents the extrusion of the soft material. The inability of the applied load to squeeze out the soft layer is known as the principle of plastic constraint. Consider a thick layer of clay sandwiched between two plates of steel. If pressure is applied to the steel plates, the clay will deform and squeeze out the edges of the sandwich. But as the thickness of the clay gets ever smaller, it takes an ever-increasing amount of force to squeeze out more clay. Eventually, a thin layer of clay remains that cannot be extruded out without the application of an infinite amount of pressure. A bearing needs to conform to the shape of its housing; a shape that is constantly in a state of flux since the engine is an elastic device. In view of this the bearing is designed so that when the two halves of the housing are correctly bolted together its parting line surfaces adjoin and the bearing correctly conforms to the housing, leaving the required running clearance between its working surface and the journal. However, when a bearing shell is fitted into its respective housing its edges will stand slightly proud of the housing faces so that when the cap bolts bring the parting line surfaces together there will be a slight gap between the housing faces. When further tightening brings the faces into contact the gap will have gone and the resultant 'crush' means that the bearing is compressed like a spring and applies a radial load to its housing.

Although a plain bearing is thus an interference fit in its housing locating lugs can be fitted to assist positioning during assembly. More typically each bearing shell is retained by a pin projecting into it from the housing. These lugs or pins will help avoid any danger of movement relative to the housing in operation but that is not their primary purpose and the interference fit must be good enough in this respect to ensure reliable operation. In the case of the big end the interface between the plain bearing and its respective journal normally receives a supply of pressurized lubricant from a drilling in the journal. The relative movement of journal and bearing and the forces involved cause the oil to spread out and form the necessary film throughout the radial interface, before spilling into the crankcase.

INFLUENCES
Crankshaft main journals are subject to extremes of torsional vibration, and that influences their diameter. However, journal overlap and crankshaft balancing techniques are further factors, which may permit the use of smaller diameter and narrower journals. It is notable that the Cosworth DFV 3.0 litre V8 of 1967 had a main bearing journal diameter of 60 mm with a big end journal diameter of 49 mm. By contrast, a third of a century later a 3.0 litre V10 typically had a main journal diameter in the range 40-45 mm, a big end journal in the range 35-40 mm. However, there is also a very large difference between the operating speeds of those two engines. Since main journal diameter is a major factor in crankshaft torsional stiffness, perhaps the reduction in crankshaft torsional stiffness caused both by the reduced diameter and the increased length served to provide a greater separation between the crankshaft torsional resonance point and the much higher excitation frequency of today's engines. g