Terms & Conditions


Go to -> Company Details

Go to -> Licensing of Oasys software – Licensing Documentation PDF

Go to -> Help with ordering

Privacy and Cookies

Data Security

Website Terms and Conditions

Software Licensing Terms

Privacy and Cookies

We may change this privacy notice from time to time by updating this page.

What information do we collect?

When you use this website, we may collect the following information:

  • the areas of the website that you visit
  • information about your computer, such as which browser you are using, your network location, the type of connection you are using (e.g. broadband, ADSL etc) and your IP address

We do this by using cookies, which are small files that help us track how our visitors use the website and enable us to understand where we can improve your experience. If you would like to find out which cookies we use and the information they track see our Cookies Policy.

Once you submit or register information through our website we will know who you are and your activities on this website and information about you and/or your company may be recorded on our systems. For example, we may ask for personal information when you download our software including:

  • your name
  • company name
  • email address
  • postal address
  • telephone number
  • country where you are based
  • Social media ID
  • your comments/questions
  • services/markets you are interested in

We may also collect personal information from telephone calls and/or other correspondence with you.

What do we do with the information we collect?

The information we capture is used for various purposes. The main purpose is to provide you with our services (whether available via the website or offline). We also use the information for:

  • website development
  • understanding how our visitors interact on the website
  • understanding what our clients are interested in
  • understanding what potential clients are interested in
  • dealing with enquiries/concerns
  • marketing our services and people to you
  • market research
  • service development
  • internal record keeping


We would like to provide you with information about our services and other information which we think you may find interesting. We may send you such information by post, email and/or telephone, unless you have asked us not to do so.

We will not provide your personal information to other organisations for marketing purposes without your explicit consent.

If at any time you do not want your information used for direct marketing purposes, please contact us or follow the unsubscribe link in our marketing email messages.

Who do we share this information with?

We may share your personal information with companies acting on our behalf who will only use the information to provide that service. However, we will retain control of your data and any third party service provider that we use must act in accordance with our instructions. We may also share your personal information with a purchaser or potential purchaser of our business.

In some circumstances, we may have to disclose your personal information by law, because a court or the police or other law enforcement agency has asked us for it.

How to get copies of or amend the information we have collected

You may request details of the personal information that we hold about you under data protection laws. If you would like a copy of the information held about you please write to us at oasys@arup.com or at: Data Protection Officer, 13 Fitzroy Street, London, UK, W1T 4BQ. Please note that we may charge a small £10 administration fee for information requests.

If you think any information we have about you is incorrect or incomplete, please email us as soon as possible. We will correct or update any information as soon as we can.

top ]

Data Security

We are committed to ensuring that your information is secure. In order to prevent unauthorised access or disclosure we have put in place suitable physical, electronic and managerial procedures to safeguard and secure the information we collect, including locked cabinets, electronic password protection and pass card access to buildings.

If at any point you suspect or receive a suspicious communication from someone suggesting they work for Oasys or a website claiming to be affiliated with Oasys, please forward the communication to us or report the incident by email to oasys@arup.com or in writing to Oasys, 13 Fitzroy Street, London, UK, W1T 4BQ as soon as possible.

top ]

Website Terms and Conditions

The contents of this web site are protected by copyright and other intellectual property rights under international conventions. No copying of any words, images, graphic representations or other information contained in this web site is permitted without the prior written permission of the webmaster for this site.

Oasys accepts no responsibility for the content of any external site that links to or from this site.

top ]

Software Licensing Terms

Terms and Conditions of Purchase

The full conditions of purchase and maintenance for all Oasys software are set out in the Oasys Software Licence and Support Agreement. All prices are subject to TAX at the current rate.

Prices and specifications are subject to change without notice – please ask for a written quotation.

Although every care has been taken to ensure the accuracy of all information contained herein, the contents do not form or constitute a representation, warranty, or part of any contract.

Superseded Versions of Terms and Conditions

Oasys keeps copies of all superseded versions of its terms and conditions.

Maintenance & Support Services

Twelve months support and maintenance is included with most products. Thereafter maintenance is typically 20% of the current sales price.

An annual maintenance service is available for most programs after the first year.

This service includes:

  • telephone/fax/email/web based support
  • free software updates available via internet download
  • personalised output header for many products


Find the answer to frequently asked questions from existing users.

Most popular FAQs

Plane Frame
To create the above 2D plane frame in GSA do the following (note that the attached PDF has this text plus additional useful graphics).

  1. From the File menu select New… or click the New (Ctrl +N) button on the Standard tool bar
  2. In the New Model Wizard fill in the job details, then click Next>
  3. Select the structure type; in this case set to Plane. Click the Units… button to and reset the units by clicking the kN-m button, then OK, and Finish
  4. You should now have a blank graphics window and the gateway open. You will note that there are a number of shortcut buttons at the bottom of the screen. Select Nodes
  5. As the node units are set to m, ensure that the coordinates are given accordingly.
    Input the coordinates given here. You will note that the nodes appear in the graphics window as you work.
    x z
    0 0
    4 0
    8 0
    18 0
  6. The rest of the nodes could be placed in the same way, but instead we will copy up the nodes to form the next layer. Click on the Graphics window to make it active then click the Select Nodes button on the Cursor Mode toolbar. Click and drag a window around the nodes.
    Open the Sculpt Toolbar by selecting that button on the Cursor Mode toolbar. Click the Move/Copy Selection button. Shift the nodes 1, 2 & 3 by 3m in the Z direction; click the Preview button to ensure that GSA will give what you expect, then select OK.
  7. Switch on the node numbers by clicking the Label Node Numbers button on the Display Favourites tool bar.
    Repeat step 6 with nodes 3 + 4 to create the left hand haunch, but this time copy by 4m not 3m.
  8. Select node 7 and copy 5m in the X direction and 2m in the Z direction.
    Note that during these operations the Nodes table automatically updates.
  9. To set the frame supports, select the nodes 1 – 4, then right click on the graphics window to open the menu, then Modify Selected nodes. Alternately, you can click on the Modify Selected button. You will note that there are releases possible in the six degrees of freedom, but as we have set this model to a plane frame a number of them are greyed out. For each degree of freedom you can set to Restrain, Free, Free plus a support stiffness (spring support), or don’t modify that freedom. There are also a number of default restraint types; click on the Pin button, then OK.
    To display the supports click the Label Restraints button on the Display Favourites tool bar. The word Pin should now show next to the modified nodes.
  10. Now to create the analysis elements. Click the Elements button at the bottom of the screen to open the elements window. We could define the elements by stating the topology, but it is easier to draw them on. Click on the graphics window then click the Add Elements Sculpt Tool button. Click on the nodes in turn to create the layout shown above. Again note that the elements are filled in automatically.
  11. We now need to add the releases to the elements. Click the Select Elements/Members button on the Cursor Mode toolbar, then select the two braces by clicking to the right of them and dragging left (dragging to the left selects the elements that the window crosses, dragging to the right selects the elements within the window). Right click on the graphics window and Modify Selected Elements on the pop-up menu. Check the Modify type of 1D element and set the toggle box to Tie (this means that those elements are tension only) and OK. You will note that the two braces are now a different colour to the other elements.
    Select the two horizontal elements and again Modify Selected Elements; this time click the Beam Releases button, check Modify next to each yy, then set the radion button to ReleaseOK and OK again to exit the dialogs. To confirm the releases click the Label Element Releases button on the Display Favourites tool bar. The releases will now show.
  12. The elements currently have no section data, so let’s define that now. Click the Sections button at the bottom of the screen. The easiest way to define the sections is to click the Wizard button on the Data Options tool bar.
    To do the columns first, on the wizard dialog, set the Name to Column, and the he material to Steel. You will note the wide range of section definition types available; this time set the radio button to Catalogue and click next. There are four catalogues to choose from; for this example set the Catalogue to British, the Type to Universal Columns and the Section to UC254x254x89; click Next, then Finish.
    Repeat to create Brace, Beam and Rafter section definitions with suitable section profiles.
  13. Now to associate the sections with the elements. On the Graphic Display toolbar click the Section Display button. Select all the vertical elements (the cursor mode should still be set to Select Elements/Members) using the Ctrl or Alt keyboard buttons to add or remove from the selection group, and right click to Modify the Selected Elements. Check the Modify property to and set to 1.
    Repeat for the braces and set to property 2, the horizontal elements to property 3 and the sloping beams to property 4.
    To make it easier to see which element is set to what property, on the Graphic Display toolbar click the Label and Display Methods button , change to the Display Methods tab and set the Colour Elements to By Property, then OK. You should now see the different sections in different colours.
  14. Loading next. On the gateway on the left of the screen, double click on Loads, then Nodal Loading then Node Loads to open the table. Also switch on the display of loads in the graphic window buy clicking All Load Diagrams on the Display Favourites tool bar.
    To apply a wind force to the horizontal beams, type the following into the Nodal Loading table on a single line:
    Nodes: 5
    Load Case: 3
    Axis: Global
    Direction: x
    Value: 5
    A horizontal point load should now be visible on the frame. If it is not, change the Cases choice on the graphic window to L1
  15. Now open the beam loading table from the gateway. To apply UDLs to the left hand horizontal beam, select it in the graphics window, then right click and Copy Selection As List. In the Beam Loading table past the selection to the Beam list. Set the load case to 2, the type to Uniform, Axis to Global, Projected to ~, Direction to z, and the load value to -6 kN/m. Note that the gravity load is negative as it acts in the opposite direction to the z axis.
    Select and copy the right hand horizontal beam to a new beam list, set the values as above except the Type to Linear, and the Load 2 to -8 kN/m. You should now have a varying load increasing to the portal frame column. If the load slopes away then swap the Load 1 and 2 values around.
    Now select and copy the rafters to a new load line, again load case 2 but set the projected to Projected.
  16. To add self weight, open the Gravity Loading table from the Gateway and accept the defaults by just pressing Return.
  17. To analyse the frame you click the Analyse button on the GSA tool bar. You may get some warnings, but for this exercise do not worry about them unless the analysis will not run; if that is the case check the data already input.
    You can now display the results graphically using the remaining buttons on the Display Favourites tool bar and the Deformed Image on the Graphic Display tool bar.
  18. To create a load combination, enlarge Cases and Tasks on the Gateway on the left side of the screen, then double click on Analysis Tasks. Right click on the Analysis Tasks window and select New Analysis Task… from the menu. Click Next and add the three standard load combinations 1.4L1+1.6L2, 1.4(L1+L3), 1.2(L1+L2+l3)) as separate cases. Click on Next and Finish to analyse them.
    On the graphics window the case or combination to view can be selected using the Cases toggle box on the top of the window; you can also go through them in sequence by clicking on the adjacent + and – buttons.
  19. You can print out the graphics window with the results; annotate the result values by clicking the Select for Annotation button on the cursor mode tool bar and click or window the elements that you want to see the results for.
    Tabular results can be chosen by clicking the Output button on the bottom of the screen; double click on the data that you wish to view.
  20. Congratulations, you now have a fully analysed GSA frame.


Setting up a staged analysis is very simple in GSA.

  1. Ensure that Analysis Stages are enabled in the Preferences (Tools | Preferences | Advanced Features)
  2. Define the stages (Gateway | Analysis Stages | Stage Definition). This is specified as an element list e.g. “pb1 to pb3 not(21 to 30)”
  3. If necessary, define properties (Gateway | Analysis Stages | Analysis Stage Properties) and constraints (Gateway | Constraints) for different stages
  4. Define the Analysis Task to work with that Analysis Stage

You can display the stages in the graphical window by changing the Display combo box. You will note that elements not in the current stage are shown in a dashed grey line. You can also colour the elements according to the first stage in which the element is included by setting ‘colour by stage’ in the Labels and Display Methods dialog.

There are several uses of Analysis Stages.

Building a multi-span beam or bridge

If you have a multi-span beam, the usual thing is to analyse the whole of it at once, but if it is not built all at once (say for a bridge??) then this will not give realistic results. To illustrate this, let’s look at a simple two-span concrete beam.

Take a two element beam with symmetrical spans. Apply a load to each span, a load case to each beam, one span in stage 1 and both spans in stage 2.

Analysing the first span by itself, we get a standard pin-ended result:

Multi-span Beam Stage 1

Stage 1

The second analysis case includes both spans, but only loads the second span:

Multi-span Beam Stage 2

Stage 2

Adding these both together in a combination case we get the moments in the permanent condition:

Multi-span Beam Stage 1 + Stage 2

Stage 1 + Stage 2

To take into account the construction sequence (and assuming linear analysis) we can envelope this combination with the first Analysis Stage, which had a higher span moment:

Envelope of (Stage 1) and (Stage 1 + Stage 2)

Envelope of (Stage 1) and (Stage 1 + Stage 2)

Compare this with the moments from the traditional analysis of both spans loaded together. We can see that the hogging moment is double and the span moments reduced and even halved; figures way beyond normal limits of redistribution.

Both spans loaded together > </div> <p align=

Both spans loaded together on the whole model

This analysis, of course, makes many assumptions about the sequence of formwork striking and casting, but it does indicate how useful it is to consider construction sequencing when designing structures.

The GSA model used in this example, <Multi-span beam.gwb>, is attached.

Changing restraints and properties

Another advantage of GSA’s Analysis Stages is the ability to change the properties of restraints and sections through the Analysis Stage Properties.

In this example, we have a crane lifted beam

Crane Lifted Beam

that is then encased in structural concrete.

Beam encased in structural concrete

We can achieve this by changing the beam’s section properties in the Analysis Stage Properties table (Gateway | Analysis Stages), where we change all elements with property 1 to property 3 for that stage.

Analysis Stage Properties table

In this example we changed the beam properties, but we can also do that for 2D elements, springs, masses, links and cables.

We can also adjust the restraints in the Generalised Restraints table (Gateway | Constraints) – note that in this model there are no nodal restraints set in the nodes table; all the restraints are set here. Other constraints, such as Joints, Rigid Constraints, etc. can also be specified to apply only to specific stages. WATCH IT: if you remove a restraint you need to apply forces in the stage to represent the effect of the removal of the restraint.

Generalised Restraints table

The GSA models used in these examples are attached below.

Attached files: Multi-span beam.gwbStaged analysis.gwb

Ribbed slabs as shown below are common in engineering practice. To model the ribbed slab in details in finite element analysis when doing whole model analysis may create a huge model that may take too long to analyze or the model size is just too large to be analyzed using available computers. To simplify the modeling of the ribbed slab, orthotropic material 2D elements may be used if the global behaviors of the ribbed slab are interested. The parameters for the equivalent orthotropic material 2D elements may be derived as follows; Ribbed floor

Definition of isotropic material ribbed slab:

E – Young’s modulus

G – Shear modulus

n – Poisson ratio

Ho – The equivalent bending thickness about x axis

Other geometric dimensions are shown on the diagram above

Definition of orthotropic material 2D element:

Ex – Young’s modulus in x direction

Ey – Young’s modulus in y direction

nxy – xy Poisson ratio (y direction strain generated by unit strain in x direction

nyx – yx Poisson ratio (x direction strain generated by unit strain in y direction

Gxy – xy plane shear modulus

Gyz – yz plane shear modulus

Gzx – zx plane shear modulus

H- The equivalent thickness of the orthotropic material 2D elements

Out-of-plane bending with negligible shear and in-plane deformation

The equivalent bending thickness Ho about x axis of the ribbed slab can be obtained assuming the average bending stiffness of the ribbed slab is the same as the bending stiffness of the uniform thickness slab, this gives:

Eqn 1(1)

If (T – t) is larger than b, T should take as t+b to prevent overestimating Ho

To enable the orthotropic material 2D element to have the same bending stiffness about x axis as the ribbed slab, we have:

Ey = E(2)

H = Ho

The unit width bending stiffness of the ribbed slab about y axis is:

Eqn 3(3)


Eqn 4(4)

To enable the orthotropic material 2D element to have the same bending stiffness about y axis as the ribbed slab, we have:

Eqn 5(5)

It is reasonable to assume:



Eqn 6(6)


Eqn 7(7)

As assumed, shear deformation is negligible, we can assume:

Eqn 8(8)

To summarize, the parameters for the orthotropic material 2D elements to simulate the ribbed slab are:

  • Thickness H             Eqs (2)
  • Ex                                     Eq(6)
  • Ey                                     Eq(2), the same as isotropic material Young’s modulus
  • nxy & nyx                  Eq(7)
  • Gxy, Gyz, Gzx            Eq(8), the same as isotropic material Shear modulus

The above parameters are obtained assuming that the slab is subjected only to bending and shear deformation is negligible, so they can be used if these conditions are satisfied. The use of orthotropic material 2D elements is for simulating the global behavior of the ribbed slab, so the stresses obtained from the orthotropic material model do not represent the stresses in the ribbed slab model. If stresses of ribbed slab can be calculated using the moments from the orthotropic material model and the real geometry of the ribbed slab.

If the slab is mainly subjected to in-plane forces, the parameters for the orthotropic material 2D elements to simulate the ribbed slab can also be obtained in the similar way.

Alternatively or the layout of the ribbed floor (e.g. steel decks with ribs in one direction) is complicated but still behaves orthotropically, the overall properties of the ribbed floor can be obtained by analysing a small piece of the ribbed floor that is modelled in details. Once the overall properties of the ribbed floor are known, the equivalent orthotropic material properties can be worked out in a similar way as shown above, then orthotropic material 2D elements can be used to simulate the overall behaviour of the whole ribbed floor.

As a general guide:

The quad 8 (and tri 6) elements provide a quadratic approximation of displacement over their element domain and so provide the best accuracy when compared with the linear approximation of quad 4 (and tri 3) elements.  The 8 nodes required to accommodate quadratic behaviour makes such elements the most numerically expensive in terms of processing time, memory requirements and stored file size.  Where such constraints are not an issue, quad 8 elements offer the best performance.

The quad 4 and tri 3 elements in contrast provide only a linear approximation of displacement over their domain and so while less accurate, their requirement for only 4 nodes per element make such elements an attractive practical alternative.  This alternative becomes more relevant from GSA 8.5 with an improvement to the out-of-plane performance of linear elements by the introduction of the MITC formulation.  This new formulation removes the previous problems of hourglassing associated with the original ‘Mindlin’ formulation and so, while still linear in approximation, offers a competitive alternative to the quadratic elements.

There are two locations where you can set the usage of the MITC elements:

  1. for all new analyses – Tools | Preferences | Miscellaneous | GSS Solution | Plate formulation default
  2. for existing or particular analyses – Analysis task | Advanced | 2D Element Analysis | Plate formulation

In the GSA preferences you can switch the release and restraint symbols on and off.

The default is off and you get the following displays of releases

And restraints

With the option checked you get restraint symbols (note that the rotations are in red and the translations in green)

And restraints

We have noted that some machines have had graphical problems with GSA

For example when selecting entities in the graphics window:

  • Dragging a selection leaves a trail of rectangles on the screen
  • Selection marks are incorrectly displayed

Other problems have included

  • The contents of the graphical window not matching the windows dimensions
  • Graphical updates not occurring automatically
  • Images not displayed in perspective views

These problems are related to the graphics card and its drivers.

In most cases one of the following solutions resolves the problems, in order of recommendation:

  1. In GSA, toggle ‘Tools | Preferences | Miscellaneous | Advanced | View Options | GDI overlay in Graphic Views’
  2. Install the latest build of GSA
  3. Update the graphics card driver (note that Windows Update and Device manager may incorrectly indicate that the driver is up to date – check for a more recent version on the graphics card manufacture’s website)
  4. Switch to a non-Aero desktop theme (if running Windows 7)

You can determine the graphics driver manufacturer using the following steps

  • Click the Start menu button
  • Click “Run”
  • In the window that opens type “dxdiag”
  • Click the Display tab (there may be more than one) and see the Manufacturer listing

When downloading the latest Graphics Drivers, make sure that you obtain them from the website of the graphics card vendor, not that of the laptop/desktop. So, if you use a particular laptop, but you determine from the steps above that your graphics card is by a certain manufacturer, then go to the graphics card manufacturer’s website, and download the latest drivers from there.

Python has become increasingly popular for engineering computing in the recent years. Apart from being free and open source, it provides several excellent library add-ons and packages such as SciPy that can be useful for engineering analysis and design workflows.

Apart from a standard Python environment/interpreter, you will need the PythonCom and win32com libraries, which you can be obtained from http://sourceforge.net/projects/pywin32/files/pywin32/.


Here is an example Python script, which opens one model, deletes the results and re-analyses it, then saves as a new file:

# GSA Python script

import win32com.client

import pythoncom

gsaobj = win32com.client.Dispatch("Gsa_8_7.ComAuto")


gsaobj.Open("C:\\Program Files\\Oasys\\GSA 8.7\\Samples\\Stair.gwb")





gsaobj = None

All FAQs

Occasionally you can get error messages related to the Section database, whether “Corrupted section database found” or “Sectlib.db3 is missing”.

There are a number of solutions:

  1. when the warning message prompts to find the section database, click Yes and browse to C:\Program Files\Oasys\GSA 8.7\ and select sectlib.db3
  2. If this file is not there or is corrupt then you might copy sectlib.db3 from a colleague
  3. or download and install the latest build of GSA, as it is possible that the downloaded file or installation process were corrupted

Axes are essential to the geometric description of a GSA model. When the structure is orthogonal, the default axes are all that are needed. But when beams and grids start veering off or curving round, getting clever with the axes can save you a lot of problems.

Global axes

Standard axes are provided by GSA. Of these, the most fundamental is the global axis set. All other axes, whether standard, implicit, or explicitly defined axes are defined with respect to the global axis set. GSA adopts the convention that axes labelled with uppercase X,Y,Z are in global axes; otherwise lowercase axis labels are used.

Use of axes

Nodal coordinates are defined with respect to an axis set. Usually this is the global axis set, though this can be changed by setting the ‘current grid’.  (See below.)

Nodal constraints act in constraint axis directions. This defaults to global, but can be set to any standard or user defined axis set.

Element axes are often defined in terms of the element topology, though several element types can have their element axes defined by reference to a particular axis set.

Results are usually reported in output axes directions. These default either to global directions or to element axis directions, depending on the result type.  Many results can be transformed from these default directions to specified output axis directions by overriding the default output axes.

Axis set definition

Axes table

Axis sets are specified as being one of these three types:

Cartesian axes

Cylindrical axes

  1. Cartesian, in which coordinates are defined in terms of lengths along each orthogonal axis. (Global axes are Cartesian.) Cartesian axes are good for structures in which the members are typically at right angles to each other.
  2. Cylindrical, in which coordinates are defined in terms of the plan distance from the origin (r), the angle about the z axis (θ), and the length along the z axis. Cylindrical axes are good for when working with spiral staircases and round buildings.
  3. Spherical, in which coordinates are defined in terms of the radius (r), an angle about z (φ), and an angle of inclination from z (θ).

For cylindrical axes, the x vector is the direction of zero rotation (θ=0) and the xy vector defines the z=0 plane of the cylinder.

Axis sets have an origin, specified in global directions.

Spherical axesCartesian axis sets are oriented by means of an x vector that defines the direction of the x axis from the origin, and an xy vector that, together with the x vector, defines the xy plane. Both of these vectors are defined in global directions. (N.b the xy vector does not need to point exactly in the intended y axis direction.) The z axis is orthogonal to the xy plane.


For sperical axes, the x vector is the direction of zero rotation (φ=0) and the xy vectordefines the plane of 90° rotation (θ=90).

Standard axes

In addition to global axes, GSA provides several standard axis sets, including cartesian axes in the YZ and ZX planes, and cylindrical and spherical axis in the XY plane.

User defined axes

You can define user axes as Cartesian, cylindrical or spherical. Unlike the global axes and other standard axes, these might have their origin somewhere other than 0,0,0 and might be rotated from the global directions. For example, if you have an area of structural grid that is skew to the rest of the building, then you will need a user defined axis, with the origin set at the ‘hinge’, and oriented to align with that part of the structure.Spiral stair

Grid Planes

Grid planes are related to axes, in that you refer to an axis set and z elevation to define a grid plane. A grid plane may be thought of as a virtual surface in the GSA model. As with axes, GSA provides standard grid planes, and you may define others.

The ‘Current Grid’

The current grid is the grid plane in which nodal coordinates are currently reported and defined. In Graphic Views the grid is displayed on the Current Grid. The name of the current grid is displayed in the status bar, and can be specified via ‘Data | Define Current Grid”.

In the old days engineers had to inspect stresses in local element axis directions as these were certain to be in an axis set that were aligned with the element (i.e. element in the xy plane). So as to give engineers scope to inspect results in a direction of their choice, but still with respect to an axis set with z normal to the element, GSA offers forces and stresses reported in projected axis directions:

  • the x axis of the specified output axis is projected onto the element to produce a ‘projected x axis’
  • the z axis is normal to the element, and
  • the projected y axis is orthogonal to the projected x and normal.

(See ‘projected axes | fundamentals’ in the GSA help.)

The default output axis is ‘global’, but you can change this. No results are produced for elements that lie in a plane normal to the output axis xy plane.

In the old days engineers had to inspect stresses in local element axis directions as these were certain to be in an axis set that were aligned with the element (i.e. element in the xy plane). So as to give engineers scope to inspect results in a direction of their choice, but still with respect to an axis set with z normal to the element, GSA offers forces and stresses reported in projected axis directions:

  • the x axis of the specified output axis is projected onto the element to produce a ‘projected x axis’
  • the z axis is normal to the element, and
  • the projected y axis is orthogonal to the projected x and normal.

(See ‘projected axes | fundamentals’ in the GSA help.)

The default output axis is ‘global’, but you can change this. No results are produced for elements that lie in a plane normal to the output axis xy plane.

Note the following when interpreting 2D element results.

  • Derived moments/forces/stresses include the principal values, in whichever direction they occur. Diagrams of principal values are drawn in the principal directions. Von Mises and Average stresses are also included in the derived results.
  • Projected moments/forces/stresses are the values reported in the x and y directions of the ‘output axes’ as projected onto the element xy plane.
  • Check which output axes you are using by clicking the ‘Axes’ button on the Contour or Diagram Settings dialog box. Diagrams of projected values are drawn in the projected output directions.
  • Moment terms follow the Timoshenko convention in which, for example, the moment Mx is based on the stress in the x direction; Mx is NOT the moment about the x axis. With this convention if a slab is in compression on the top face in both the x and y directions then the moments are both negative.
  • If you aren’t sure about the orientation of the local axes, turn on ‘element axes’ in the Labels and Display Methods dialog. The axes are coloured Red : Green : Blue for x : y : z.
  • 2D element axes are defined by projecting the 2D Element Property ‘Axis’ onto the element or, if this is set to ‘Local’, by the first edge of the element defining the element x and the first and last edges defining the element xy plane.
  • ‘Max’ and ‘Min’ forces/moments are signed, so, for example, a negative force refers to principal compression. You need to look at both results to understand the principal compression/tension regime in an element.
  • Forces and moments are given per metre run in the relevant direction. These are essentially the appropriate stresses multiplied by the element thickness. For a 0.5×0.5m square element, therefore, you would need to multiply the kN/m value by 0.5 to get the total kN force in the element.
  • The Mx+Mxy and My+Mxy results may be used to allow for in-plane twisting.

Note that the RC Slab design feature may be used for designing reinforcement requirement in RC slabs and walls.

Refer to the ‘Output Options | Results | 2D Element Results’ section of the GSA manual for more details.

Grid loading is expanded prior to analysis, either explicitly, using the ‘Tools | Expand Grid Loading’ command, or implicitly at the time the analysis is requested. An understanding of the internal process of grid load expansion will help in diagnosing grid loading problems. There are two stages to expanding grid loads. In the first stage GSA searches for panels and in the second stage GSA attributes loading to the panels. This note discusses some of the problems that arise in the first stage, the panel search.

A panel is a closed loop of elements; a typical grid plane will contain a number of panels. There are four common problems when identifying panels in grid planes, most of which, when understood, are relatively easy to deal with.

  1. Elements included in the grid plane – there are two criteria for including elements in the grid plane: the elements must be within the grid plane tolerance and the elements must be included in the element list. The first of these may need to be selected carefully, particularly when loading is applied to complex roof geometries, where the elements do not form a flat plane. In these cases it may necessary to relax (increase) the grid plane tolerance, but this may have to be done in conjunction with the element list to ensure that spurious elements do not get included.
  2. Disjointed elements – for the panel determination to work GSA has to track topologically around the elements. An element which is not connected topologically to the adjacent element (for example where a join exists) will break the topological connection leading to a failure in the panel expansion.
  3. Coincident elements – this is a similar problem to the disjointed elements but in this case the topological check is confused by doubling back on itself. Removing coincident elements, or removing the elements from the grid plane, avoids this problem.
  4. Crossed elements – where elements cross, such as bracing elements in a bay, the topological check results in a ‘twisted panel’ which cannot be resolved. Generally the bracing should not be loaded, so excluding bracing elements from the grid plane is usually the best option in this case.

In raft analysis it is not necessary to specify support stiffnesses for the nodes that interact with the soil since support stiffnesses are automatically generated during the analysis and deleted after the analysis. However, where support stiffnesses are specified these will be used as the initial soil stiffness. This can speed up the analysis if the specified stiffness is closer to the actual soil stiffness than the default initial stiffness of 1.0e10 N/m.

You can apply a pressure normal to a curved ‘surface’ of Beam elements using Grid Area Loads, as follows:

  • Define a Grid Plane onto which the whole surface can be projected. Often a grid plane parallel to the global XY will do.
  • Set the Grid Plane Tolerance such that all elements in the surface are picked up by the Grid Plane. (i.e. a large tolerance.) You may need to specify the Element List as other than ‘all’ if you want some elements not to attract load.
  • Define a Grid Area Load applied to the whole Plane of your Grid Plane.
  • Set the Axis of the Grid Area Load to ‘Local’ and the direction to z. The Local axis for the Grid Plane is such that the local z is normal to the ‘surface’ of elements (and the local x is the grid plane x projected onto the surface).

You can then expand the Grid Area Load into equivalent Beam Loads using the ‘Tools | Expand Grid Loading?’ option. (This may take a while.)

There are two ways of specifying a spring support at a node:

  1. You can assign support stiffnesses for the node on the Supports tab of the Nodes table. The stiffnesses act in the constraint axis directions for the node.
  2. You can attach a Ground Spring element to the node. The stiffnesses of the Ground Spring and the axis directions in which these stiffnesses act are specified in the spring property for the Ground Spring.

In order to apply loads to a grid plane GSA has to identify panels (enclosed regions) in the grid plane. The panel identification will fail where there are strings of element that do not form closed or form twisted panels. (Twisted panels are caused by elements that cross over each other.) In cases such as this, where these elements are a correct part of the model, the element list in the grid plane can be used to exclude these elements from the panel identification.

Fabric material is an orthotropic material and five parameters are needed to describe its elastic property. They are Ex (warp direction Young?s modulus), Ey (weft direction Young?s modulus), ?yx (warp direction Poisson’s ratio – warp direction strain generated by unit strain in weft direction), Vxy (weft direction Poisson’s ratio – weft direction strain generated by unit strain in warp direction) and G (shear modulus). As there is dependency between the two Young’s moduli and two Poisson’s ratios, only one Poisson’s ratio need be specified in GSA, Vyx. The other Poisson’s ratio is calculated by GSA using the following relationship:

Ex / Vxy = Ey / Vyx

You can compare the effects of applying a Rigid Constraint to a model with the unconstrained condition by putting the Rigid Constraint in an Analysis Stage that includes all elements in the model.

Tied Interfaces constrain slave element nodes to act with master element nodes;  they do not constrain master element nodes to act with slaves.

So, in the extreme situation where mesh A is being tied to mesh B and mesh A is half the density of mesh B, and every other node in mesh B is already connected to mesh A, when specifying the tied interface it is important that the elements in mesh A are the masters in the Tied Interface, and mesh B, the slaves.

Tied Interface properly specified

Otherwise all the slave nodes will be considered to be adequately constrained to the masters, to result in no nodes being constained by the Tied Interface.

More masters than slaves in Tied Interface

In this circumstance the Tied Interface has no effect.

There should be many slaves to serve few masters, …as ever.

Axes are key to the geometric description of an analysis model. When the structure is nice and rectangular, with gridlines parallel to the compass, the default axes are all that’s needed. But when beams start veering off or curving round, getting clever with the axes can save you a lot of problems.

Element axis definition for non-vertical beams

Once you have defined where the beam ends are in space, you then need to ensure that they are the right way around. After all, I beams are much stronger in one direction so you need to get it right. The local axes of a beam are also defined in terms of x, y and z (but lower case). The local x axis comes first, as the vector from the start to the end of the beam. The local z is then as close to the global Z as possible while keeping it perpendicular to the local x – this means that a beam web will be vertical, unless you add a rotation onto the member. Finally, the local y is the perpendicular to the local xz plane.

Element axis definition for vertical beams

For a column, that is vertical beams, the rules have to change as the local z axis is now horizontal. Now, the beam is orientated by making the local y match the global Y.

While beams may usually be aligned so that the major axis is vertical, columns typically need aligning. With GSA there are two ways of doing this: one is to set a rotation and one is to align the rotation to a separate node. Orientating to a node is ideal if your grid is at an angle and you are not quite sure what it is, or if you have a circular building or façade on an arc. It also comes in useful for aligning beams on.

You can of course use a combination of them both to get the orientation that you want.


Beam element orientation

The best way to model tapered sections in GSA is to subdivide the element into a number of shorter elements, each referring to a different section property. Please note that averaging the section properties will give a more accurate result rather than averaging section dimension.

Also, it might be useful to define the taper sections in Excel using the concatenate command to give the section text, then copy/paste that into the sections table in GSA.

This will be because you have parabolic (Quad8 and Tri6) elements in your model. These have only 5 degrees of freedom: they lack the local ZZ (drilling) stiffness. This means that nodes in the mesh that are not connected to anything else would have a zero stiffness in this direction, which would cause serious problems to the analysis. GSA gets round this by restraining the affected nodes in this direction. It shouldn’t make any difference to the overall analysis. If you look in the Advanced part of the analysis wizard, 2D Element Analysis tab, you will note a few other restraint options that you can use.

Linear (Quad4 and Tri3) elements do not normally generate this message as they default to 6 degrees of freedom. If you do get this message, usually from old models, then you can change how the analysis treats the elements. In the Analysis task’s Advanced options, go to the 2D Element Analysis tab and set the In-plane formulation to “Allman-Cook (includes zz dof)”

Load Cases

In the model, loads are assigned to Load Cases. These are numbered and can have the prefix L applied, giving them the reference L1, L2, etc.

They can also be given names in the Load Case Titles table. If Load Case names have been set then they can be selected in the loading table views.

Analysis Cases

If a default linear elastic analysis is run, then each load case will be analysed in a separate Analysis Case with the same name and number, but with an A prefix. Hence L1 -> A1, L2 -> A2, etc.

If an analysis tasks are explicitly setup by the user then the Analysis Cases may take the loads from one or more Load Case, which can also be factored at that time. For example, A1 can be derived from “1.35L1 + 1.5L2”.

Combination Cases

Combination Cases add or subtract multiples of one or more Analysis Cases (note that this is only appropriate for linear analyses) and/or calculate the envelope values of a number of Analysis Cases. These are numbered and have a C prefix, they can also be named. For example, C1, with the name “ULS” might have the description “1.35A1 + 1.5A2”.

Envelopes take the results from one or more Analysis and Combination Cases can take the simple form (A1 to A3) or can include multipliers and other modifiers such as min, max, abs, and signabs. A combination case including syntax such as “(1.4 or 1.0)A1 will envelope 1.4A1 and 1.0A1. An envelope case (C1 to C3) will take the extreme values of combination cases C1, C2 and C3.

For a full list of the advanced options for combination cases refer to the GSA help file: Program Fundamentals > Cases and Tasks > Syntax of combination case descriptions.

For a quick reminder of the options type something incorrect into the combination case description to launch the reminder dialogue. The Combination Case wizard is also very useful.

The GSS Conjugate Gradient solver allows much larger models to be analysed in GSA than is possible using the default Skyline Active Column solver. The Conjugate Gradient solver may also speed up the analysis of large models. Note that the Conjugate Gradient solver can only be used for static analysis.

Invoke the Conjugate Gradient solver option by clicking ‘Advanced’ on the last page of the Analysis Wizard and selecting ‘GSS Advanced Settings dialog | Stiffness Solution | Conjugate Gradient’.

This note explains why modal analyses occasionally produce the warning: ‘missing eigenvalues’ and discusses whether this warning is cause for concern.

In an eigenvalue analysis you are solving for the roots of a high order polynomial. The eigensolver is set up to favour the lower eigenvalues, which contain the most significant modes. However it is possible for the eigensolver to miss modes, especially where another mode of similar frequency has been identified. Typically modes are more likely to be missed where several modes are clustered around a frequency, whether due to structural symmetry or some other reason. Also, it is usually the less significant modes (i.e. modes with lower effective mass) that are missed.

The automatic check for missing eigenvalues can be confused in some situations. Consider the case of a circular cantilever. This will have pairs of modes in orthogonal directions, so, if an odd number of modes is requested the ‘missing eigenvalues’ message may be given because only one of the last pair of modes is found. More generally if there is a cluster of modes near the highest requested mode this can cause the check to report missing eigenvalues.

When the ‘missing eigenvalues’ warning is given it is important to check that all the significant modes have been picked up by the modal analysis. The Dynamic Summary output is useful for checking this: if all the dynamic behaviour is accounted for the ‘accumulated effective mass / total unrestrained mass’ would be 100% in each of the X, Y and Z directions. Most codes will require that a percentage of this (typically around 90%) should be included in any dynamic response calculation, so provided the totals meet this requirement no significant modes should be missing. (Note that tall building structures almost always tend to be stiffer in the Z direction so the frequencies in the vertical direction tend to be higher; in this case it may be difficult to recover sufficient mass in the vertical direction).

In short you should not ignore the ‘missing eigenvalues’ warning, but if you are capturing the dynamic behaviour then it is not a cause for concern.

(Even when there is no report of ‘missing eigenvalues’, you should still check that all the significant modes have been picked up by the analysis.)

When you do a modal analysis in GSA the displacement results are mode shapes and the magnitude is arbitrary. In previous versions of GSA these are scaled to give a maximum displacement of 1m. In GSA 8.1 these are scaled to give a maximum displacement of one in the current ‘Length – small’ units. So now if the small length unit is specified as inches then the maximum displacement in the mode shape will be 1 in. In such circumstances some of the modal quantities e.g. modal mass will differ from those calculated in older version of GSA.

The inclusion of a rigid constraint in a model significantly increases the amount of memory required during GSS analysis. This is because all nodes in a rigid link must be resolved before any of the nodes in the rigid link can be released from memory. Manually adjusting the front vector such that it points almost normal to the plane of the rigid constraint may help reduce the size of the analysis, as may setting the ordering method to ‘Reverse Cuthill-McKee’.

Comparing reactions at a node with element forces in P-delta results will usually reveal an apparent lack of equilibrium. The reason for the difference is that the element forces are calculated using the element stiffness [K] while the reactions are calculated using the element stiffness including the geometrical effects ( [K] + [Kg] ). The forces in the element are not affected by the geometrical stiffness – they are calculated directly from the displacement and rotations. The reactions do take the geometrical effects into account, allowing for 2nd order effects.

Models containing short stiff elements can cause ill-conditioning in GSS analysis and a failure to converge in GsRelax analysis. In this circumstance Link elements should be used instead of stiff Beam or Bar elements.

Nodal forces and moments from non-linear analysis are balanced at the deformed position of the nodes. Where the displacements are large the nodal forces may appear not to be in equilibrium at the undeformed nodal positions.

Where the yield stress in a user material is given a non-zero value the material is treated as ‘elastic – perfectly plastic’ in GsRelax non-linear static analysis. A yield stress of zero specifies an elastic material; all standard materials are elastic. (Note that ultimate stress, hardening modulus and hardening parameter are not used in any GSA analysis at present.)

The time taken to do a GSS solution depends on the way the equations are set up. GSA tries to set these up in a well ordered manner, working along the longest dimension of the structure. This normally results in a stiffness matrix clustered close to the diagonal. However, when a structure contains rigid constraints these can provide topological connection which introduces many more off-diagonal terms. This results in a much larger matrix to be solved.

An example of where this problem might arise is where a structure is longer than it is tall and rigid constraints have been used to model rigid floor plates. In this situation the automatic solution order is along the length of the structure whereas the most efficient order is in the vertical direction.

If rigid constraints seem to cause a significant increase in the analysis time it may be worth telling the solver the direction in which to build the solution matrix. This can be done by going to the Analysis Specification and selecting ‘Advanced’. Then set the solution order vector to ‘User defined’ and set the vector components such that the direction is normal to the rigid constraints.

The reinforcement is orientated with respect to the element’s axes, and in particular to the X axis. The element’s X axis is angle zero, and all angles are measured anticlockwise from there.

For example, a reinforcement layer at angle 90 will be aligned with the element’s Y axis.

The completeness of RC Slab data can be established by choosing ‘Design | Check RC Slab Design Data…’ from the program menu.

If the concrete design code is ‘undefined’, then GSA assumes that the concrete and reinforcement material properties have been specified explicitly by the user in the RC Slab Design Properties Wizard. Mistakenly unrealistic values may lead to unreinforceable areas being shown in the output.

If no RC Slab Design Properties have been specified, then RC Slab Reinforcement will not be offered as an output option.

If any RC Slab Design Properties have been specified, then RC Slab will offer RC Slab Reinforcement as an output. However, all 2D elements must refer to 2D Element Properties which have corresponding RC Slab Properties, otherwise the selection of the RC Slab Reinforcement output option will result in a warning message and no reinforcement design.

RC Slab Reinforcement will not be offered as an output option unless analysis results are available.

Generally diagrams and contouring are annotated by selecting the nodes or elements for which the diagrams are to be annotated; selecting all the entities results in all being annotated. Another way of annotating all entities is by opening the Further Options dialog from the Diagram Settings dialog or Contour Settings dialog and setting ‘Annotation’ to ‘All’ rather than ‘By Selection’.

Grid loading is applied to grid planes, not to nodes or elements, so the ‘By Selection’ method of annotation is not available. Annotation of grid loading may be switched on by setting ‘Annotation’ to ‘All’ in the Further Options dialog.

Sometimes GSA crashes on HP laptop. Ensuring that GSA is using the dedicated graphics card (NVidia Quadro) will solve this problem. To do this:

  1. Open the NVidia Control Panel application, which should be installed on your machine
  2. Click on “Manage 3D Settings”
  3. Select GSA from the application list. If it’s not on the menu, you could simply browse to it
  4. Set the “Preferred Graphics Option” to the NVidia card.
  5. Close and reopen GSA

When working with odd shaped structures that have surfaces that are not parallel to global planes it is useful to have the option to produce standard views: plan, X elevation, etc. with respect to non-global axes. This can be achieved by setting the current grid to a grid plane that lies in the non-global axes and then giving the ‘Graphics | Orientation | Orient About Grid Axes’ command (also on the Sculpt toolbar). While this is set, all standard views and orientation settings are with respect to the current grid.

The following Graphic View selection tools are available on the Edit menu:

  • Select String – When in the Select Elements cursor mode this selects a string of elements attached to and aligned with any currently selected elements. When in the Select Nodes cursor mode, nodes along the string of elements are selected. When in the Polyline cursor mode the polyline is extended forward from the last polyline segment, adding vertices at the positions of closest aligned displayed nodes or grid points. The straightness tolerance preference is used to determine whether an element, node or grid point is aligned, allowing this command to be used to select curved strings.
  • Select Highest Coincident – When in the Select Nodes or Select Elements cursor modes this deselects all but the highest numbered coincident items in the current selection. E.g. If nodes 1 to 30 are selected and nodes 1 and 11 are coincident and nodes 2, 12 and 22 are coincident then giving this command will result in only nodes 11 and 22 being selected. This command is useful for deleting coincident nodes and elements (though note that nodes may not be deleted graphically when they are referred to by elements).
  • Select Close To Vertical 1D Elements – When in the Select Elements cursor mode this selects all 1D elements that are within 1 degree of vertical but are not sufficiently close to vertical to be deemed vertical for the local element axes definition.
  • Select Out-Of-Range – Enabled only when contours are displayed, this command selects all entities for which the contoured data is out of the range of the current contour settings. E.g. If nodal Uz displacements are being contoured and the lowest contour value is set to -10mm and the highest to 0mm then ‘Select Out-Of-Range’ will select all nodes for which the Uz displacement is less than -10mm or greater than 0mm.

GSA offers options to save the Graphic View image to file using the ‘Graphics | Save Image | Save WMF’, ‘…PNG’, ‘…JPEG’ or ‘DXF’ commands. The use of these popular file formats is well established. Also in GSA is the option to save the parameters that you have set to define the image that is currently displayed in the Graphic View. The parameters saved include everything from the orientation and scale down to the numeric format of the annotation associated with each diagram that is displayed and the set of items to which this annotation is applied, – and everything in-between. Save the current view parameters by using the ‘View | Save Graphic View’ command. The view parameters are then entered as a record of data in the Saved Graphic Views module and will be saved with your model the next time you save. Saved views may be accessed from the ‘Views’ tab on the Gateway. Double click on the required saved view (or ‘right-click | Open’) to display the view. Note that opening a saved view for a model that has changed since the saved view was saved will apply the view parameters to the model in its current state; – only the parameters that define the view are stored with the saved view, not the data that is displayed.

Saving a Graphic View with the name ‘startup’ results in that view being the initial view when the model is opened.

When recording animated graphics to AVI it’s preferable to start the recording before starting the animation. By doing this the internal process that sets up the animation knows that the animation is for recording to AVI file and therefore establishes the correct number of frames to suit the requested animation period and AVI frame rate. Otherwise the resulting AVI file will not necessarily display the animation at the requested period and the file size will probably be a lot larger than it need be.

To record a graphics movie to AVI file first display the Recorder toolbar (using the Camera icon on the Graphic Display toolbar). Then click ‘Start recording’ on the Recorder toolbar. Each frame displayed in the Graphic View appears as a single frame in the AVI file. (Conversely, leaving the Graphic View open without changing the display while recording results in no additional frames in the AVI file.) ‘Pause recording’ may be used to avoid dialog boxes appearing in the recording, etc. ‘Freeze recording’ may be used to display a particular frame for longer than the default frame rate. The AVI data is saved to file when ‘Stop recording’ is clicked. The default frame rate and freeze time can be adjusted in the ‘Preferences | Graphics | Saved Image Settings’. The screen dimensions of the AVI movie are determined by the size of the recorded Graphic View.

It’s easy to animate graphics in GSA. Click the ‘Animate’ button and the current image is animated according to the settings specified in ‘Graphic Settings | Animation Settings’. By default if the image is displayed deformed then the animation will animate the deformation otherwise the animation will simply rotate the image. The period of animation can be specified in the Animation Settings. You can zoom, pan and rotate the image while animation is in progress.

‘Highlight coincident nodes’ and ‘Highlight coincident elements’ options are available in the ‘Graphics | Graphic Settings’ dialog.

Also, when in a selection cursor mode, the ‘Edit | Select Highest Coincident’ command removes from the current selection set all but the highest coincident item of those currently selected.

Diagrams are drawn for an entity (node or element) when that entity is displayed and not when the entity is excluded. Unfortunately grid loads are not associated with entities. The only way that you can exclude grid loads is by using volumes to exclude the reference point for the load. The reference points are defined as follows:

  • Grid point loads: the position of the load
  • Grid line loads: the centroid of the points defining the load
  • Grid area loads defined by polygon: the centroid of the polygon
  • Grid area loads defined by plane: the origin of the grid plane to which the load is applied

Press the <ESC> key to clear the current selection (of nodes, elements, polyline etc.) when in the corresponding selection mode. If there is no selection in the current selection mode or the cursor mode is not a selection mode then clears all selections. Therefore <ESC> <ESC> always clears all selections.

When nodes or elements are selected in a Graphic View the ‘Edit | Copy’ (Ctrl+C) command copies the selected items onto the clipboard in list format. This list can then be pasted into any field that expects the respective list type.

Note that the reverse is also true: Copy a list from a table and paste this into a Graphic View when in the corresponding selection cursor mode will result in that list of items being selected.

The scale at which the structure is to be drawn in printed output can be specified in the Graphic Settings dialog.

The ‘Volume’ cursor mode is an extremely quick and easy way of identifying the part of your model that you want to look at. When several volumes are defined, the subset of the model included in all the volumes is displayed. Volumes can be inclusive or exclusive.

The ‘Highlight edges’ option in the Graphic Settings dialog identifies all edges in 2D element models, – intended or otherwise! Section Display is useful for determining whether you are looking at the top of a 2D element (light grey) or bottom (dark grey); – make sure the elements consistently face one way or the other, especially if they are taking bending.

Copy (‘Edit | Copy’ or Ctrl+C) in Graphic Views behaves differently depending on the current cursor mode:

  • When in an entitiy selection mode the current selection is copied to the clip-board in ‘list’ format. The list may then be pasted into any field requiring a list, e.g. to specify nodes or elements in a loading record.
  • Otherwise the current image is copied to the clip-board in both WMF and bitmap formats. The image may then be pasted into another program, e.g. Word or Paint. The receiving program determines which format to take from the clip-board, e.g. Paint takes the bitmap whereas, by default, Word takes the WMF image.

In an entity selection mode there is also the option to ‘Copy selection as GWA’, which copies the selected entities on to the clip-board as GWA records.

If you select elements (or members, or lines) by dragging from the left towards the right around the elements then only those elements entirely within the rectangle are selected; dragging from the right towards the left selects any element partly within the rectangle. For this purpose the selection points of an element are the node positions and the centre.

Clicking the ‘Volume’ button when already in the Volume cursor mode sticks it down.  (Or ‘V’ ‘V’, but not double-click on the Volume button.)

Use the wheel on your mouse to zoom and pan Graphic Views. Rolling the wheel zooms, and you can pan by dragging the mouse with the wheel pressed like a button. These operations work in all cursor modes, so you don’t have to stop what you’re doing to manipulate the view.

The Data Defaults dialog allows you to specify what attributes new elements, nodes and members have when they’re sculpted in. Keeping the Data Defaults dialog open when sculpting can therefore save you time later, when you’d otherwise have to correct the properties of your new sculpted entities.

You can save time by saving a Graphic View with the model when you’ve got it looking how you want: Use the View menu to save the view and the Views tab on the Gateway to access your saved views. Saved views take account of any changes you make to your model, so you won’t get caught out with old results being retained.

GSA’s combinations and envelopes are a powerful way to quickly check for the worst load case. The number of permutations that they need to check can get out of control very quickly if you are not careful.

A simple combination, such as (A1 + A2 + A3) has one permutation, while the envelope (A1 or A2 or A3) has three permutations to be checked.

If you combine envelopes then the number of permutations multiply. E.g. ((A1 or A2 or A3) + (A4 or A5 or A6)), then the number of permutations is ((1 + 1 + 1) x (1 + 1 + 1)) = 3 x 3 = 9.

We have seen models with envelopes of envelopes of envelopes, with the current record combination giving 1.4×1036 permutations. To add to the problem many of the permutations were duplicates, meaning that the same load combination was being calculated and checked multiple times. Also some of the loads were insignificant in magnitude compared to other loads, and so could be safely excluded.

The recommendation is therefore to try to avoid envelopes of envelopes where you can, and to consider carefully which loads or load combinations are likely to be significant.

For full details of how to create combinations and envelopes, plus add filters and modifiers, see the help file Program Fundamentals > Cases and Tasks > Syntax of combination case descriptions.

While it is possible to create movement joints in 2D element meshes by adding translational releases to the relevant element nodes, it can be easier to do so using Joints:

  1. select a mesh that you require to be jointed to another
  2. move it a certain distance, so that it disconnects from the rest of the structure
  3. move it back, but deselect the “use existing nodes” option in the Move/Copy dialog. This will result in coincident nodes along the border
  4. open the Joints table and set the default line to the desired properties
  5. select all the nodes along the movement joint line
  6. run the command Sculpt > Constraint operations > Create Joints. This will detect the coincident nodes and join them with Joints as per the table defaults
  7. collapse any remaining coincident nodes left by the movement commands (Sculpt > Collapse Coincident Nodes…)

When modelling tunnels, it is normal to want the Ground Springs to only resist outward movement. As tunnels are (usually) circular, setting up a cylindrical axis down the centreline and aligning the Ground Springs to that is very useful. A common mistake though is to use compression-only springs; you actually need tension-only.

Why is this? Consider a ground spring supporting a building. Here a compression-only ground spring will resist downwards vertical movement, which is in the -Z direction. When aligning Ground Springs to a radial axis, you need to resist loads in the +R, hence the need to set them to tension-only.

  • Compression-only resists negative movement
  • Tension- only resists positive movement

2D edge loads in GSA are defined as pressures or line loads, so you need to both know what edge to apply the load to: the Labels and Display methods > On Elements > 2D element edges will help you with this. Additional operations such as using the Spin 2D Element sculpt command can help too.

Note that with GSA 8.7 you can now change the Edge load specification from Pressure to Force/Length in the General Specification dialog.

You can also apply line loads using Bar elements

  • Trace a bar member onto the Lines on the design layer and give the section minimal properties so that it will not add to the self weight or stiffness of the slab. Note that bars will give no bending stiffness to the slab anyway, but you want to avoid adding in-plane stiffness. A zero density material can also be useful, though it will generate a warning in the analysis report.
  • Ensure that in the Region definition the “Generate beam elements along aligned members” is checked.
  • Following meshing, apply the line loads as UDL loads to the resulting bars. For extra points apply the loads to property references rather than element numbers so that remeshing will not affect the results.
  • If the mesh already exists, set the element defaults to the bar, then select the line of nodes to receive the line load and use the “Add string of 1D elements” command to draw in the bars and load as above.

These notes are intended for users having problems running large GSA models.


Some multi-threading happens in GSA: the solvers run in one or more separate threads, as does the rendering of both graphics and output views. Even when not multi-threading within GSA, simply having a core or processor dedicated to GSA, leaving system and other processes to run on a different processor, will improve performance. So there is some benefit in having multiple cores. (Most new machines these days do have at least dual cores anyway.)

RAM 32-bit OS

But the biggest benefit is in RAM: the more RAM that is available, the less the need to page ‘memory’ to disk. However there are limits to the amount of RAM that a 32-bit application can address. For 32-bit apps running on 32-bit OS (the current standard) the limit is 2GB of RAM, so even if you have 4GB installed GSA will still only address 2GB.

RAM 64-bit OS

With a 64-bit Windows operating system the RAM available to GSA is effectively unlimited: it is down to how much physical memory you have installed.

64-bit OS is now our recommended platform for GSA. Note that we have separate 32 & 64-bit versions of GSA.

GWB File Opening and Saving via Memory File

By default GSA opens and saves GWB files via a memory file; e.g. when saving, the file is assembled in memory before being dropped onto disk. This is much quicker than the traditional method of reading and writing directly to disk. However, huge files can be too large to be manipulated in memory causing GSA to choke on the data. The ‘GWB read/write via memory file’ preference can be switched off to force GSA to read and write directly to disk. (When attempting to save a file that is found to be too large GSA displays a message that offers to switch off the ‘GWB read/write via memory file’ preference and try again.)

Planned for the Future

We have several ideas that we are putting into action that will make GSA more accommodating of large models. These include more efficient post processing and handling of results, making more use of multi-cores and multi-processors, both during analysis and pre- and post-processing.

Loads on building frames from runway beams or cranes are moving loads and thus it is crucial to find the worst case locations for these loads for the service and ultimate design. GSA Bridge is the ideal tool for finding these load conditions.

Defining the problem

First you need the building frame, including the crane rails. Note that as the moving vehicle loads are grids loads you will need to close across the ends with dummy elements to make a closed panel.

Note that the gable structure and cladding rails have been left off for clarity.

Next you require an axis and grid plane at the crane beam level. In this example I have created the axis origin mid way between the crane rails. Note that it needs to be at the gable end or the limit of the crane movement and that the X axis must be in the movement direction.  The grid plane, created using this axis is set to one-way with a 90° direction so that any loads are distributed directly onto the crane beams.

We then use this grid plane to define an alignment

And a path for the crane to move along

In this example I am taking a 5 tonne capacity with a bridge weight of 30 kN, a crab weight of 8 kN, and a wheel spacing of 2.6 m. As the bridge span is 18 m, it produces four wheel loads of 7.5 kN at 9 m offset from the centreline. The combined crab and lift load are applied at the wheel locations but offset to the closest approach distance (assumed 1 m from crane beam)

Moving vehicle loads

As the worst case may be with the crab on either side of the building, I have created two moving bridge loads, such that the crane starts half a metre from one gable and stops half a metre from the other. Note that the chainage takes the 2.6 m wheel spacing into account.

The moving load will be generated at 0.5 m centres, though you can make this spacing larger or smaller as you desire, with a corresponding change to the number of load cases

To create the individual loads use the menu command Tools > Bridge Analysis > Expand Bridge Loading…

You can then display the grid pint loads to check that they have been created correctly.

Create any additional permanent and variable loads as necessary.

Run a static linear analysis and then combine and factor as normal for the ultimate and serviceability limit states.

First, to spot the coincident nodes check the relevant option in the Graphic Settings wizard:

Graphic Wizard

Next select all the effected nodes.

Then use the <Sculpt | Collapse Coincident Nodes…> command. Where selected nodes are closer than the tolerance, the one with the higher node number is removed and all references to that node (such as in the Elements or Loads table) are replaced with that of the remaining node.

Note that this will not affect pairs of nodes in a joint.

First, to spot the coincident elements check the relevant option in the Graphic Settings wizard:

Graphical Wizard

Next select all the effected elements.

Then use the <Edit | Select highest coincident> command. This deselects all but the highest numbered coincident elements. You can then delete the remaining selection.

Repeat as necessary.

Yes you can.

First you need to set a different constraint Axis on the Restraints tab of the Nodes table:

Nodes table

or the Node Definition dialog:

Node definition dialog

Any restraints are then aligned with that axis, so restraining the x direction allows this node to translate only in the yz plane of ‘Axis1’:

GSA graphic

The resulting deflected shape indicates that the restraint has behaved as intended:

GSA deformed graphic
Note that the constraint axis also affects constraints. For example the directions specified in a Joint are in constraint axis directions.

See the attached example model

Face loads on 2D do not have to be constant: you can set the load to vary across the face.

2D Element Loading Table

Loads such as hydrostatic vary with depth, so it can be quite involved to calculate and apply them manually. In GSA you have some options that will help.

Create and Split

The essence of this method is that if you split an element with varying loads, they split as well. If we take this example of a single 2D element with a varying load set to something at the base and nothing at the top



 2D element with varying load



And then use the Sculpt | 2D Element Operations | Split Quad Elements command (having first selected it of course), and break it down into smaller pieces, then the loads adjust accordingly.

Split 2D element with varying load


A more flexible method is to apply the loads once the geometry is created using the Sculpt commands.

If we take this more complex (auto-generated) example, calculating and applying the varying loads is involved.

Auto-created mesh

If we select all the elements though and use Sculpt | Create Element Loading | Map Face Loads on to 2D Elements then GSA can calculate the forces for you. In this particular example the wall is 5m high, so we want the hydrostatic load to vary from 50 kN/m2 at the base (z = 0) to zero at the top (z = 5). We can achieve this by setting the pressure to be 10*(5-z):

Map Face Loads dialog

With the result:


Before the ‘GSA – Revit Structure’ link can be used it must first be registered with Revit.

If Revit Structure is installed at the time that GSA is installed then the registration will happen automatically; otherwise the registration may be invoked by running program GsaRegister.exe in the GSA program folder, e.g. <C:\Program Files\Oasys\GSA 8.2\GsaRegister.exe>.

A beam load may be applied to a string of elements using the ‘Sculpt | Create Element Loading | Create Beam Load’ command. In the Beam Load dialog, when Patch load type is selected, the positions may be specified as less than 0% or greater than 100% (or the equivalent actual lengths). In this circumstance patch loads are applied to elements found to form a string extending beyond the selected element.

Existing geometry can be transformed according to a mathematical expression using the ‘Sculpt | Transform Geometry’ command. The option allows nodal coordinates of selected nodes to be transformed to a position expressed as a function of the original position of the node. Normal mathematical notation is used in expressions, for example:

z + abs( sin(x) ) * pi

Mathematical expressions may also be used for generating node loads using the ‘Sculpt | Create Nodal Loading | Map Node Loads on Nodes’, and similar for element loads.

In any of the loading tables the loads can be sorted by case. This option is on the right-click menu in the load tables. This means that loads can be entered in any order, or new loads appended to the end of the table and then the records sorted so that all loads for any one case are grouped together.

The Task View is central to the control of analyses in GSA. The right-click menu offers many useful options for inspecting and editing tasks.

The ‘Show Description’ option changes the display in the Task View from analysis case names to analysis case descriptions. This can be especially useful as a quick check that the loading is as expected.

Alternatively the ‘Properties for ‘ option will display this information in the Object Viewer.

The Beam Section Summary output lists, per section property: the total number of elements, the total length of all the elements with that property, the total element mass, the total element surface area and the total cost of elements, – and similar for members.

GSA produces ‘stripey output’ by default; – output for alternate entities is displayed on a pale grey background, to improve legibility.

Unfortunately some printers print the pale grey background darker than is intended. Stripey output may be switched off by toggling the ‘Output | Stripey Output’ command. (And don’t forget that the ‘View | Save Default View Settings’ command may be used to preserve that setting for subsequent sessions.)

When you do a response spectrum analysis you can combine the results from the different directions in two basic ways:

  1. By linear combination of the basic responses (e.g. L21 + 0.3L22 + 0.3L23 – the so called ‘30% rule’)
  2. By the SRSS method ( e.g. sqrt(L21^2 + (0.3L22)^2 + (0.3L23)^2) ).

The latter option was offered in GSA 8.0 as a combined response. In GSA 8.1 an SRSS combination can be specified in the combination cases table; the syntax is SRSS(L21,0.3L22,0.3L23).

(The combined response option, available in GSA 8.0 and earlier, has been removed.)

GSA does not produce results for any combination case that refers to an analysis case for which results are not available. Use the ‘Combination Case and Envelope Details’ output to determine whether results are available for all contributing analysis cases: cases for which results are not available are shown in red. (In GSA 8.0 and earlier, such combination cases did output results ignoring the fact that some cases contributed no results.)

Diagrams of ‘abs’ values of envelope results can be misleading. In the case of a linearly varying diagram from, say -1 to +1, the ‘abs’ of this diagram should be from +1 to 0 at the centre to +1. However if values are not calculated along the element GSA will draw a linear diagram between the end values, i.e. +1 to +1. Requesting values to be calculated at intermediate positions along the element (‘Diagram Settings | 1D Element Results | 1D Element Intermediate Forces | Number of equidistant points’) will display a more correct absolute representation of the envelope.

Typically the content of both the ‘Properties’ and ‘Report’ tabs on the Object Viewer are best displayed in landscape orientation. Therefore docking the Object Viewer at the bottom of the GSA window is recommended. Docked at the bottom of the window the Object Viewer takes up a lot of screen space so a useful habit is to close the Object Viewer when it is not being used. This can be done conveniently using the ‘Object Viewer’ toolbar command (or Alt+3), though right-clicking on a node or element in a Graphic View and selecting ‘Display Properties’ opens the Object Viewer if it’s not already open.

Grid planes can be use for ‘Plane’ structure types. In the Current Grid Definition use the Grid Plane button to display the Grid Plane Definition dialog. In this dialog select ‘Y elevation’ for the Grid axis and ensure that the grid elevation is 0. You then have a grid plane in the global XZ plane.

You can use the CIMsteel file format to merge two models together. When importing a CIMsteel file into an existing model you are given the choice of retaining the existing data or the imported data where conflicts arise.

Bar elements are effectively pin ended beams so no moment is carried from a bar into the rest of the structure; similarly for ties and struts. However moments are calculated at intermediate positions along bar elements. These are the free end bending moments due to load applied to the bar (e.g. self-weight in a gravity case). These free end moments are based purely on equilibrium, regardless of the properties of the section. (Whereas intermediate displacement calculations do use the section properties, so setting Iyy and Izz to zero results in straight lines joining the deformed nodal positions.)

N.b. As of GSA 8.4, intermediate values are not calculated for bars, ties, and struts, unless requested in the 1D Element Results dialog box.

The ‘View | Save Default View Settings’ command saves all model independent settings in the current Graphic View or Output View as your preferred view settings for that type of view. The current settings are saved as preferences and are used as the initial settings for new views in the current and subsequent GSA sessions. For example: open a Graphic View, switch on ‘node dots’, ‘View | Save Default View Settings’ new Graphic Views will then be opened with node dots displayed. More details are given in the online help at ‘Working with the Program | Working with Saved Views and Preferred Views | Default View Settings’ (or type ‘default’ in the index).

The GSA programming interface (i.e. COM interface) can be used to control GSA from, say, a spreadsheet or other remote program. A typical use of this is where you want to carry out structural optimisation in a spreadsheet. For example in a VBA macro you can tell GSA to:

  • Open a data file.
  • Analyse.
  • Get results.
  • Delete results.
  • Get a section property.
  • Set a section property.
  • Analyse.
  • Save.

Refer to the ‘Programming and Command Line Interface’ chapter of the GSA help for more details. Sample XLS files that demonstrate the use of this feature are installed in the \Oasys\GSA ?.?\Samples folder.

2D elements sometimes appear to be corrupted by using the sculpt ?Modify elements? option to convert from ‘linear’ to ‘parabolic’ (e.g. Quad4 to Quad8). In that operation, when creating the new mid-side nodal position GSA first looks for an existing node in the required position before creating a new node. This is so as to provide good connectivity at the mid-side nodes. GSA uses the current ?tolerance for coincidence? when considering whether a coincident node exists. GSA can be misled into choosing an existing corner node as the mid-side node where such a node lies within the tolerance; typically this occurs when working with small elements. This can be avoided by setting the ‘Options | Preferences | Sculpting | Tolerance for coincidence’ to an appropriate (i.e. smaller) value.

GSA’s Quad4 and Tri3 elements can be created from DXF files that contain corresponding polygon or polyface meshes. When importing the DXF file into GSA (via ‘File | Import | AutoCAD (DXF file)’) select which type of 2D element is to be created from these DXF entities by specification in GSA’s DXF Import Options dialog.

Models imported from DXF files often suffer from bad element connectivity. The Connect 1D Elements option (‘Sculpt | 1D Element Operations | Connect 1D elements’) is effective at tidying up such models. This connects selected crossing elements by splitting them and connecting the split elements to a common node at the point of intersection. It can also be used for trimming and extending elements.

Data files containing large report log strings take a long time to open (in GSA 8.0). You can overcome this problem by opening the report (?View | Open Report View?) and deleting the contents (?View | Clear Report?).

Splitting elements graphically, whether using the split element, refine element or connect element commands, can be a slow process. Typically this happens when the elements that are being split have loads applied to them; – GSA takes a long time adjusting the element lists by replacing the old element reference with references to the new elements. The slow processing can be avoided by deleting the loads before splitting the elements and then re-applying the loads after the split.

There are three ways of deleting loads efficiently:

  1. To delete all of a particular load module (e.g. Beam Loads): right-click on the load module on the Tables tab of the Gateway and select ?Delete all?. (In fact this option is available for all modules.)
  2. To delete all loads associated with a particular load case: right-click on the load case in the Load Case Titles table and select ?Delete Loads?.
  3. To delete loads graphically: display the loads you want to delete in a Graphic View and select ?Sculpt | Delete Displayed Loading?.

Then of course you can always delete load records explicitly in the loading tables.

The Add Elements Sculpt Tool allows you to create any type of element graphically. First set the element type under Element Defaults on the Defaults tab of the Object Viewer then click on as many nodes as are required for the element type. E.g. click on four nodes to specify a new Quad4 element. If a node does not exist at the position you click and if the grid is displayed then a new node will be created at the clicked position on the grid (i.e. at the coordinates shown in the status bar at the time you click).

The message:

‘Error. SHARE.EXE was not loaded, or a shared region was locked’

– displayed when attempting to open a GWA file in GSA usually means that the GWA file is currently open in another program, such as Excel.

To quickly create a wall or slab made up of a number of 2D Elements:

  • create one 2D element defining the boundary of the wall or slab;
  • select this element;
  • split this bounding element into the required number of component elements via the various split commands under ‘Sculpt | 2D Element Operations’.

The sculpt ‘Flip Elements’ command is useful for reversing the direction of 1D elements and flipping upside-down 2D elements. Releases and offsets are preserved at the original position in global space.

When you open a model in GSA a ‘manual backup’ is automatically made. Using the ‘File | Restore Manual Backup’ command reinstates the model to the state at which the last manual backup was made. This is easier than closing the model and re-opening since the restored model is immediately displayed in all views that are currently open.

Despite GSA’s powerful Undo capability it is still worth using the ‘Preview’ option available in many of the sculpt dialog boxes. Previewing eases the process of experimenting with sculpt commands since the sculpt parameters do not need to be completely re-specified if, say, a ‘copy’ is carried out with respect to the wrong axis set.

Save time looking for the worst forces in your model by using an envelope combination case. A simple envelope looks like: “C1 or C2 or C3”; this would give you the envelope of results for combinations 1, 2 and 3. Output views for envelopes tell you which part (permutation) of the envelope is responsible for each result.

You can plot contours of values on nodes or elements that you’ve calculated outside of GSA. You do this by importing a “User Module” which is a simple text file that refers to the node/element numbers. You can import more than one module, and they can be saved with the model, so you won’t have to re-import them every session. Importing results by this route is much safer than editing GWA text files.

Stay up to date, sign up to our Newsletter

  • This field is for validation purposes and should be left unchanged.

Require more information?

  • This field is for validation purposes and should be left unchanged.

Newsletter Sign up

Please fill out your details below to receive the latest oasys news.
  • This field is for validation purposes and should be left unchanged.