FEA 26: Isoparametric Elements
Summary
TLDRThis video explains isoparametric elements, a type of finite element used in structural analysis, where transformations occur between local (natural) and global coordinate systems. Isoparametric elements employ the same shape functions for both displacement interpolation and coordinate mapping. The video covers the benefits of mapping four-sided and 3D elements from natural coordinates to global coordinates, and explores the process of creating stiffness matrices through this mapping. Additionally, it explains the importance of the Jacobian matrix in ensuring accurate integration and evaluating element quality.
Takeaways
- 🔄 Isoparametric elements use the same shape functions to interpolate displacement and map between local and global coordinate systems.
- 🌍 Elements can be defined either directly in the global system or transformed from a local system, with isoparametric elements opting for the latter.
- 🟥 Defining four-sided elements directly in global coordinates can be complex due to shape variations, whereas using local coordinates simplifies element definitions.
- 📐 The local coordinate system for isoparametric elements is called 'natural coordinates,' often using variables like 's' and 't'.
- 🔗 A key aspect of isoparametric elements is mapping from the natural coordinate system to the global coordinate system.
- 📊 The mapping between natural and global coordinates uses shape functions that are typically employed to interpolate displacements.
- 🧮 Calculating element stiffness matrices becomes easier using this mapping, reducing complex global integrations into simple operations in the natural coordinate system.
- 🔍 The Jacobian matrix plays a crucial role in the transformation, linking global and natural coordinates and helping with derivative calculations.
- 💡 The determinant of the Jacobian matrix represents the ratio of the element's area in global and natural coordinates, providing insight into element quality.
- ⚠️ A well-formed element has a nearly constant Jacobian, and a good rule of thumb is that the minimum determinant should be at least 70% of the maximum.
Q & A
What are isoparametric elements?
-Isoparametric elements are a type of finite element where the same shape functions are used to interpolate both displacements and map from a local coordinate system (natural coordinates) to a global coordinate system.
What is the advantage of using a local coordinate system in isoparametric elements?
-Using a local coordinate system simplifies element definitions because a single mathematical mapping can be applied to all elements, regardless of their shape or orientation in the global coordinate system. This makes the process of defining elements more consistent and easier.
Why is transformation necessary for four-sided elements in a mesh?
-Four-sided elements in a mesh can take on a variety of shapes and orientations. Defining each one directly in the global coordinate system would be complex and inefficient, so a local-to-global transformation simplifies this process and ensures consistency across the mesh.
What are natural coordinates in the context of isoparametric elements?
-Natural coordinates refer to a fixed coordinate system (often labeled s and t) used to define the element within a standard shape, such as a square. These coordinates are used in the local system for transformation to the global coordinate system.
How does mapping from natural coordinates to global coordinates work in isoparametric elements?
-Mapping is achieved by using the shape functions that define the displacement in the local coordinate system to relate the positions in the natural (local) coordinate system to the global coordinates. This ensures a consistent mapping for all elements.
What role do shape functions play in isoparametric elements?
-Shape functions in isoparametric elements are used for two purposes: they interpolate displacement within the element, and they are used to map the positions from the natural (local) coordinate system to the global coordinate system.
How is the stiffness matrix computed in isoparametric elements?
-The stiffness matrix is computed by integrating over the element, but rather than integrating in global coordinates (which would be complex), the integration is performed in the natural coordinate system (with limits from -1 to 1), making the process easier and more consistent.
What is the Jacobian matrix, and why is it important in the mapping process?
-The Jacobian matrix represents the derivatives of global coordinates with respect to natural coordinates. It captures the transformation information needed to map between the two coordinate systems. The inverse of the Jacobian is used to calculate derivatives needed for strain and stiffness matrix calculations.
Why is the determinant of the Jacobian matrix important?
-The determinant of the Jacobian matrix represents a scale factor that is the ratio of the element's area in the global coordinates to the area in the natural coordinates. This scale factor is critical for accurate integration and stiffness matrix calculation.
What does it mean if the determinant of the Jacobian matrix is not constant?
-If the determinant of the Jacobian matrix is not constant, it indicates that the element is not well-shaped or uniform. This could lead to an approximation error during numerical integration, affecting the accuracy of the stiffness matrix and the overall element quality.
Outlines
🔍 Introduction to Isoparametric Elements
This paragraph introduces isoparametric elements, which involve defining an element in a local coordinate system and transforming it to a global one. Unlike regular elements, isoparametric elements use the same shape functions for displacement interpolation and for mapping between local and global systems. The discussion begins with comparing element transformation options, emphasizing the challenges of defining elements directly in a global coordinate system, especially for non-square four-sided figures. Using a local coordinate system simplifies this, despite the complexity of the transformation.
🔄 Mapping Between Coordinate Systems
The focus here is on the transformation process between the natural coordinate system (local) and the global coordinate system for isoparametric elements. A square element in the natural system is mapped onto a more complex shape in the global system, with nodes being transformed accordingly. This mapping uses shape functions to relate coordinates in both systems. The main challenge is ensuring that the transformation remains consistent across all elements. The same shape functions used for displacement are now employed to map positions between these two coordinate systems.
📐 Understanding the Matrix Mapping
This section explains how the position within the element (in terms of x and y coordinates) is derived using shape functions and a vector of nodal positions in the global system. The shape function matrix used for displacement is now applied to nodal positions, creating a mapping system. This transformation results in a one-to-one correspondence between points in the local and global systems, simplifying complex calculations like stiffness matrix integration by transforming operations into the natural coordinate system.
📊 Chain Rule and Jacobian Matrix Derivation
The discussion moves into how the Jacobian matrix is derived through the chain rule to calculate the necessary derivatives of shape functions with respect to global coordinates. The Jacobian matrix, which consists of partial derivatives of global coordinates with respect to natural coordinates, plays a critical role in transforming these derivatives. The inverse of this matrix is used to find derivatives of shape functions, aiding in mapping and integration in finite element analysis (FEA). This matrix allows the transformation of operations from global to local systems in a manageable way.
Mindmap
Keywords
💡Isoparametric Elements
💡Local Coordinate System
💡Global Coordinate System
💡Shape Functions
💡Natural Coordinates
💡Mapping
💡Jacobian Matrix
💡Stiffness Matrix
💡Bilinear Quadrilateral Element
💡Chain Rule
Highlights
Introduction to isoparametric elements and their importance in transforming between local and global coordinate systems.
Isoparametric elements use the same shape functions to interpolate displacements and to map from local to global coordinates.
Defining elements directly in the global system can be complex, especially for four-sided figures, making local system transformation preferable.
Natural coordinates are used in the local system for isoparametric elements, typically labeled as S and T, to simplify element definitions.
A key advantage of the local-to-global transformation is reducing the complexity of handling different element shapes.
The transformation challenge is to relate the local system's basic square element to the global system’s irregularly shaped element.
Shape functions, used to define displacement interpolation, are also employed to map positions from natural to global coordinates.
The mapping matrix for position interpolation reuses the same shape function matrix used for displacements, allowing consistency across the element.
The Jacobian matrix is used to define the transformation between natural and global coordinates, critical for calculating element stiffness.
The inverse of the Jacobian matrix allows calculating the partial derivatives of shape functions with respect to global coordinates.
Chain rule derivatives are used to connect shape function derivatives in the natural coordinate system to global coordinates.
The determinant of the Jacobian matrix serves as a scale factor, representing the ratio of the element area in global versus natural coordinates.
A constant Jacobian determinant indicates high-quality elements, while variations imply numerical integration approximations during stiffness calculations.
When the Jacobian determinant is not constant, the element quality can be assessed by comparing the minimum to maximum determinant ratios.
A rule of thumb is that the minimum determinant should be at least 70% of the maximum to ensure good element quality.
Transcripts
this video introduces isoparametric
elements isoparametric elements are an
element type where we define the element
in a local coordinate system and then we
transform it to a global coordinate
system we've done that before with bar
and beam elements but now with
isoparametric it's a particular type of
transformation specifically if you look
at the term there ISO means same and
parametric is related to the parameters
that define our element so specifically
an isoparametric element is going to use
the same shape functions that are used
to interpolate displacement and we're
going to use those shape functions to do
a mapping from our local coordinate
system to our global coordinate system
before we get into isoparametric
elements let's quickly review what our
options are for element transformation
what we have is elements that we need in
2d space and we have a global coordinate
system so one option we have is that we
define every single element directly in
the global system the benefit of that is
we don't need to worry about
transformation at all everything's done
with a global coordinate system however
every four-sided figure has a different
shape in the global coordinate system
well not every every square is Silla
square but we have a lot of non square
four-sided figures that show up in a
typical mesh and every single one of
those will need its own definition in
the global system we even have to get
new definitions for rotated squares
diamond shapes for example it can be
done this process of defining the
element properties directly the global
system can be done for three noted or
triangular elements and it works very
successfully for those but four-sided
has dramatic shape changes that require
a different definition for every new
type of shape that the four-sided figure
can form into so the alternative is to
define the elements in some sort of
local system and then transform them to
the global system in order to do the
calculations that is by far easier even
though the transformation as you're
going to see gets a little bit messy
it's far easier because we have one
definition for every four-sided figure
then
so we get that simple single simple
definition but we have a transformation
challenge so we're going to have to
define some sort of mathematical mapping
between our local system and our global
system that is how most four-sided
figures are our maps are transformed
we're also going to see that for most
three-dimensional elements as well
you
before we get into how we map from a
local coordinate system to a global
let's focus in on that local coordinate
system for isoparametric elements
instead of talking about a local
coordinate system we call it the natural
coordinates for an element and here what
I've shown is the bilinear quadrilateral
element in its natural coordinate system
I'm going to call that coordinate system
SMT some other authors will use a
different set of variables here but it's
basically just not the global XY system
just like for the bilinear quadrilateral
we did previously the axes are centered
in the middle of the element but now the
element has a fixed size it is two units
tall and two units wide and the node
numbering is as shown so previously when
we looked at the bilinear quadrilateral
it had a width of two B and a height of
two H and here were the shape functions
for that element now what we're going to
do is convert that by setting B and H
equal to one so we have a standard size
and then we're going to take our s and
turn our X in turn into an S and our Y
in terms of T that gives us our new
shape functions defined in terms of the
natural coordinates s and T so now let's
talk about how we map from that natural
coordinate system over to our global XY
coordinate system so we start out in the
natural system with this basic square
element two by two dimensions and what
we want to do is map to a global system
where we have an element say that looks
like this where we have some nodes and
sines of corners but it's no longer a
square shape what we're trying to do is
relate these two shapes to each other so
that we can do things in the natural
coordinate system that's consistent
between every elements but the
properties will be relevant in the
global so here in the global system each
one of these nodes has nodes numbers x1
y1 x2 y2 and so on that each node there
we need to have some correspondence from
node 1 and global with coordinates of x1
y1
to node 1 in the natural with
coordinates of negative 1 negative 1 we
need to define a transformational
mapping that gives us the x coordinate
for every corresponding s and T
coordinate and similarly gives us the y
coordinate for every s and T coordinate
that means we can basically map on the
grid corresponding to the natural
coordinate system within this single
element that's fundamentally what we're
trying to do here put the S&T coordinate
system somewhat warped inside this
element when we do this mapping with
something called an isoparametric
element what we're going to use is the
elements shape functions previously
defined for displacements we're going to
use them to map the position from
natural to global so let's get started
on that mapping what I've got shown here
is the interpolation within the the
natural coordinate system of the
displacements so I have the N 1 n 2 & 3
and n 4 defined in terms of the natural
coordinates s and T and then I've got
you defined there as well so this we
know this definition that's how we find
displacement from degrees of freedom
it's the interpolation function so
displacement inside the element given by
the relationship between the nodal
displacements and the shape functions
what we want to do is now find the
positions inside an element in other
words x and y in terms of the nodal
positions so we're going to write a
relationship written in matrix form as
the vector X positions in the element is
equal to the shape function matrix
multiplied by a vector of nodal
positions rather than a vector of nodal
displacements so in other words this is
what we are writing as our mapping
system we're reusing those same shape
functions but now we're multiplying them
by the positions of each of the nodes in
the global system instead of multiplying
by them the
displacements of each of the notes so
that gives us the the full matrix form
is we have a vector of the nodal
positions x1 y1 into the position of
node 1 in global coordinates x2 y2 its
position of no.2 in global coordinates
and so on and then we have our shape
function matrix the exact same one we're
going to use for displacements but
remember it is now defined in terms of
the natural coordinate system s and T so
let's explore this mapping just a little
bit more we have what I'm calling a
little X vector that is the positions x
and y it's going to be equal to a shape
function matrix that's defined in terms
of the position in my natural coordinate
system and then I'm multiplying it by a
vector which is all of the nodal
positions in the global coordinates so
I'm mapping from the global over to the
local or vice versa it gives us a
one-to-one correspondence so there's one
point in the natural coordinate system
it will have a corresponding point in
the global system and the reason that
this helps us out is that we can define
our shape function matrix which we
already have in natural coordinates but
we can also define our B matrix that way
and we need to do some mapping we need
to take advantage of the mapping in
order to do that we'll see that in a
moment
secondly once we have the B matrix we
then need to calculate the stiffness
matrix to calculate the stiffness matrix
means integrating over the element every
element in global coordinates has a
different shape integrating over each of
those would be very complicated to try
to code but instead if we can use the
mapping to do our integration in the
natural coordinate system well that's
very easy we're going negative 1 to 1 in
X and y and then we're done so those are
the two real benefits we get out of
doing this mapping ok so let's work on
finding that V matrix it starts to get a
little bit hairy here so hang on tight
we know that V is going to be the
partial derivative matrix operator
multiplied by the shape function matrix
or in long form that looks like this but
we need to know the derivatives with
respect to X and Y because that's how
strain is related to displacement that's
where this came from the strain
displacement relationship however the
shape functions are now defined in terms
of our natural coordinate system so in
other words what we're looking for is
the derivative of say N 1 with respect
to X but N 1 is defined with respect to
s and T so what we're looking for are
the derivatives of the shape functions
with respect to x and y let's try the
chain rule for example DN I let's say I
is 1 2 3 or 4 plus DN I DX is going to
be equal to DN IDs times DSD X plus DN I
DT times DT DX that would be the chain
rule expansion of the derivative DN I DX
similarly for DN n IV Y we have a
similar relationship so that's the chain
rule expansion how does that help us
well see in just a moment let's write
this in matrix form because hey it's Fe
a and that's what we do so DN I DX DN i
dy becomes a vector and it's going to be
equal to a matrix which has the four
coefficients in there so d s DX DT DX D
s dy and DT dy notice that this is a
each one of these terms is a derivative
of the mapping the x2 T or y2 T of s DT
coordinate system that of course is all
going to be multiplied by my shape
function derivatives which I can now
solve for because shape functions are
defined directly in terms of s and T so
if we can figure out what's in that
matrix we can calculate the derivatives
of the shape functions with respect to x
and y which means we can then go and
find B but we know the vector with the
shape functions we don't know the
derivatives of the natural coordinates
with respect to the global remember our
mapping is X defined in terms of s and T
so we don't have the inverse let's
temporarily call that matrix that we
don't know
J and minus one the inverse of J so this
is the inverse of J why the inverse well
let me show you so J is something we
call the Jacobian matrix J inverse is
something we've just defined it's the
partial derivatives of s and T with
respect to x and y something that we
don't know on the other hand J is
something that we do know and I'll show
you how we get there so first off let's
take that expression and we'll multiply
both sides by the Jacobian J in other
words the inverse of the one we just saw
and when I do that I get DN d SD n DT is
equal to the Jacobian times DN DX and DN
D Y so now I have the thing that I don't
know on the right hand side multiplied
by something else that at the moment I
don't know and it's equal to the thing
that I do know not normally the
relationship that we'd like to use but
it's going to be helpful to us let's go
ahead and write out the chain rule
expansion of the two terms in the left
hand vector so DN DF is equal to DN DX
times DX DF plus DN D Y times dy DF
similarly we have a relationship like
that for DN DT a chain rule expansion
when I take that and I write it in
matrix form the left there vector looks
the same as what's above the matrix ends
up being each of those coefficients
again but now these are things that we
know D X D s and D Y D s are defined can
be directly calculated from our mapping
and then that's multiplied by again the
vector that we don't know but by
comparison of these two right-hand
equations we can see I've just defined
what the Jacobian is equal to it is four
terms which are the derivatives of my
global coordinates with respect to my
natural coordinates so you head hurting
yep
this actually works it makes sense it's
a little bit too a little bit difficult
to grasp so let's take it a step further
kind of bring it all home how do we get
to be with this information one of the
first things that we want to observe is
that if we can calculate J it's a 2 by 2
matrix we can find the inverse of it
once we calculate it so that's how we're
going to end up going back to that upper
left equation here which is going to
allow us to get those derivatives with
respect to x and y that we need to have
in the b matrix so now we have a
definition of a Jacobian matrix it is
the partial derivatives of global with
respect to local coordinates so the
matrix that we then need to find ni
derivative with respect to X and ni
derivative with respect to Y is the
inverse of that Jacobian and because
it's a 2 by 2 it has a simple definition
for the inverse so J inverse is defined
here we can use that to find each of
these partial derivative components of B
specifically what we're going to do is
for each of our shape functions 1 2 3 &
4 we're going to take the derivative
with respect to X or the derivative with
respect to Y that's what we're looking
for we don't know those they're going to
be equal to the Jacobian inverse
multiplied by the derivative of the
shape function with respect to s and T
the natural coordinate systems which
they are defined in terms of so there's
a lot more here that we're going to
explore in the next video just to wrap
this up I'm going to talk about a couple
of properties of this Jacobian matrix we
just created so first off note that its
terms are the derivatives of the global
coordinates with respect to the natural
coordinate system so it effectively is
that transformation matrix it's
capturing all the pertinent information
about our mathematical mapping another
interesting property is that when we
take the determinant of J that's a scale
factor it is actually equal to the ratio
of the elements area
global coordinates to the element area
in natural coordinates so ratio as in a
fraction we will use this property it'll
be critical when we find the stiffness
matrix for integration
another interesting point is that when
we have a high quality element that
determinant J is a constant term so when
we have an element which is a rectangle
J is constant
however if J is not constant in other
words it depends on position then the
numerical integration that we're going
to use to find the stiffness matrix is
not going to be exact it will be an
approximation that's going to be
important in the next video also if the
determinant is not constant we can use
the ratio of the maximum determinant
divided by the minim I'm sorry the
minimum divided by the maximum to assess
the degree of element quality and
specifically a good rule of thumb is
that the minimum should be at least 70%
of the maximum in other words it's close
to constant and then we can say we have
a good quality or a good shaped element
5.0 / 5 (0 votes)