XANSYS Message: 240 [Go back to message list]
[bookmark on del.icio.us]
No rating yet
Rate item:

Subject: Re: Intersection of two plates
Author: Guus Janssen
Date: 1999-01-15 10:37:00

Jan,
Yes I did not give all the details of the setup and purpose of the analysis
but I thought my question was clear enough to be answered. I have given
some additional information and comments below.

> From: Jan Christian Anker

> This is too difficult for me, so I have to ask some silly
> questions:

> 1) Do I understand it correctly that there one wall
> with thickness 18 mm?
> 2) Is there only one stiffener per 300 mm?
> 3) Is the stiffener welded to the wall?

> If the answer to items 1 - 3 above is 3 X "yes", then
> you could do the modeling using shell elements
> without any constraint equations (Jan Christian's
> first rule of modeling would be "never use constraint
> equations").

The answer is No to item 1 (it is a vessel with straight walls) and Yes to
the other 2-items.

> The elements would be at the
> centerlines of wall and stiffeners, preferably with
> elements of length corresponding to half the wall
> and stiffener thicknesses at the intersection,
> very good judgements concerning the thicknesses
> of these (short) elements have to be made.

> Personally, I would use 2-D solid elements. In this
> way, you do not introduce any errors at the
> intersections. As a bouns, you would have the
> opportunity of modeling the geometry of any weldings
> present. If you need a full 3-d model, you would
> want to use brick elements (Jan Christian's second
> rule of modeling would be "there is no structure found
> in nature that has zero volume or zero extension, so
> always use solid elements").

A brick model for a plate type structure is not apropriate in my opinion.
Only in case of detail stressing near the welded zone it would be.

> What I did not ask was:
> a) What is the purpose of the analysis?
To find stresses and for some places where displacements are restricted the
deformations in a static, linear analysis.

> > -----Original Message-----
> > From: G. Janssen [SMTP:G.Janssen@s...]
> > Sent: Friday, January 15, 1999 12:49 PM
> > To: xansys list
> > Subject: [xansys] Intersection of two plates

> > From: "G. Janssen"

> > I'm modelling a plane wall with stiffener plate perpendicular to the
wall.
> > I use shell 181 for both wall and stiffener.

> > | | (wall) (T=18 mm Width between stiffeners 300 mm)

> > |Q|===== (stiffener (20x100 or 150 mm))

> > A pressure load at the left side of the wall will be applied.

> > At point Q, the intersection of the two planes I choose to define two
> > lines
> > of nodes, one at the centreline of the wall, and on at the topside of
the
> > wallelements. The connection between these two are made by means of
> > Contraint Equations.

> > First the general FE question:
> > Do I indeed need to model this intersection in this way or should I
have
> > taken the easy way of generating one line of nodes at the intersection
> > lines of the mid-planes of the two plates?

> > Second the ANSYS related question.
> > To generate the CE's for nodes on the two lines I have mede two
macro's.
> > The first one to connect nodes at the keypoint locations of the lines
and
> > a
> > second one for the inetrior nodes of the lines.
> > In this proces I now arrive in the end with CE's with non unique DOF
> > errors.
> > Therefore: Is it possible to check if a node is already on a CE, either
as
> > MAster or Slave?
> > (Or is there a better way in ANSYS to couple the nodes on two lines)

> > Thanks

> > Guus Janssen

Posts possibly associated with message #240AuthorDateScore
235Intersection of two platesGuus Janssen1999/01/15 
236Re: Intersection of two platesPavel Manas1999/01/15 
239Re: Intersection of two platesJan Christian Anker1999/01/15 
240Re: Intersection of two platesGuus Janssen1999/01/15 
245Re: Intersection of two platesChristopher Wright1970/01/01