Ella bridge discretisation for load application

SOFiSTiK
1

Hi all,
I am trying to understand how Ella works. Does it use influence lines to determine the envelopes?

Does it discretise the structure and apply the load every say 2m to determine the most onerous location for the calculation of the envelopes? If yes can I control the distances between two consecutive load positions?

Thanks in advance

2

Hello

ELLA evaluates the influence lines of load trains along a specific lane. The theoretical background is described within the ELLA manual.

SSD / TEDDY > Help > User Manuals > All Manuals… > ELLA

The unfavourable load position is calculated with a three-stage method (see ELLA Manual chapter 2.6.2 Unfavourable Load Position)

  1. typical load patterns are positioned at different locations perpendicular to the traffic axis
  2. load train is placed at selected positions
  3. most unfavourable value of all analysed positions is examined with a local extreme value calculation

So the influence line is create for all elements. But you can reduce the amount of elements with the command “ELLA > CALC FROM … TO … INC”

Best regards
Frederik Höller
Your SOFiSTiK Support Team

3

Thanks for your answer.
I have one more question regarding reducing the computational load with ella. I have a very long structure and I was wondering if there is a way to tell Ella to apply the load patterns in certain locations only? Ella is very powerfull in generating load patterns and applying the loads to quads and beams in the model.

In my model I want to apply a brute force approach so I want to limit the load application points to say 0, L/4, L/2, L. This will reduce my problem to only 4 loadcases x N load alignments per span per span. I tried to use the SMIN SMAX commands in Teddy but I think this command only limits the points for which the influence lines will be calculated.

Thanks in advance

4

With the commands “CALC” and “APPL” the computing effort can be reduced, otherwise I don’t know of any method to implement a brute force algorithm. Try to
Please have a look a the chapter 3.1 Input Language and the the stated commands within the ELLA manual.

image
(source: SOFiSTiK ELLA manual, page 3-1)

Furthermore you can reduce the runtime of ELLA with the following tips.
Switch off the output generation (ECHO FULL NO) or generate a reduced output. Do not use ECHO LPOS only if it is absolutely necessary. Reduce the number of elements in the evaluation (options FROM TO INC in command CALC). In case of a system where the bridge deck is made from area elements (QUADs) use INT 3 instead of 5-9 in command LSEL.

I hope these tips help you.

1 Like
5

If you want to minimise the load placements you shouldn’t be using ELLA.
ELLA performs one calculation per element and load line (several placements)
ASE performs one calculation for all elements and one single load placement.

In general you could say that you should use:

  • ELLA, i.e. influence lines if you have
    Many load placements
    Few elements
    E.g. a very long continuos beam

  • Ase with discrete load placements if you have:
    Few load placements
    Many elements
    E.g. a tunnel with several levels made up of quads and traffic only on the top level

6

Thanks for the answers. In the end i assigned all quads within two consecutive spans to another group and used rsel and qgrp to limit ella to 2 spans and it reduced the amount of computation by a lot (i have 5 spans in total).
I find ella very powerfull because I think it does two things

  • generate influence lines which is the resource intensive part i am trying to bypass
  • calculate envelopes and set up load cases and load patterns which might take too long if done by sofiplus and ase. This part is the one i want to use without the influence lines by limiting the number of positions. For a repetitive problem it seems to work really well with qgrp.

Thanks for all the help.

1 Like