> For the complete documentation index, see [llms.txt](https://docs.openbrim.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.openbrim.org/templates/openbrimfea/geometry-fea/surfaces-fea.md). # Surfaces \[FEA] An FE Surface is a two-dimensional shell element spanning three or four nodes. It is used for decks, slabs, webs, walls, diaphragm plates and any component whose behaviour is governed by plate action rather than beam action. A shell carries both **membrane** action (in-plane forces) and **bending** action (out-of-plane moments), plus a drilling degree of freedom that lets it connect properly to beams and to neighbouring shells. It contributes to the **stiffness matrix**, carries **mass** through its thickness and material density, and **generates load** through self-weight. ## Geometry **Node 1 / Node 2 / Node 3 / Node 4:** The corner joints. Supply three for a triangle, four for a quadrilateral. *Engineering influence.* Node **order** defines the element's local axes and the sign of its reported forces, and it must go consistently around the perimeter — a quadrilateral whose nodes are listed in a crossed order produces a self-intersecting "bow-tie" element with a meaningless Jacobian. Consistent ordering across a mesh also matters for interpreting results: shells whose local axes alternate direction report Mxx of opposite sign on adjacent elements, which looks like a discontinuity in the results but is only a sign convention. Element **shape** drives accuracy. Quadrilaterals close to rectangular are the most accurate; accuracy degrades as elements become skewed, warped out of plane, or extreme in aspect ratio. Triangles are stiffer than quads for the same mesh density and are best used to transition between regions rather than to fill a whole panel. **Thickness:** The shell thickness. *Engineering influence.* Thickness is the dominant stiffness parameter and it does **not** act linearly across the two behaviours: membrane stiffness scales with `t`, but bending stiffness scales with `t³`. Doubling the thickness therefore doubles in-plane stiffness while increasing out-of-plane stiffness eightfold. It also scales self-weight linearly. A small thickness error has a large effect on deflection and on how load distributes between plate action and any supporting beams. Thickness also determines whether transverse shear deformation matters, which interacts with the Plate Type setting below. **Material:** The material supplying modulus, Poisson's ratio and density. A shell with no material has no stiffness and no self-weight. **Area (readonly) / Weight (readonly):** Derived values. Compare the total weight of a deck against a hand calculation as a quick check that thickness, material density and mesh extent are all right. {% hint style="warning" %} These two columns are derived and cannot be edited, but they are not flagged read-only, so typing into them appears to work and the value is discarded on redraw. {% endhint %} ## Global Offset **Node 1–4 Offset X / Y / Z:** Rigid offsets in **global** directions between each node and the shell's actual corner. *Engineering influence.* The usual purpose is eccentricity: connecting a deck shell at its mid-surface to girder nodes at the girder centroid, so that composite action develops through the lever arm between them. That lever arm is what makes the deck act compositely with the girders, so an incorrect offset changes the composite section properties the model effectively has, and therefore the share of moment carried by deck versus girder. Zero (the default) puts the shell corner at the node. ## Local Offset **Node 1–4 Offset X / Y / Z:** The same offsets expressed in the element's **local** axes. *Engineering influence.* Local offsets follow the shell's orientation, which is the better choice for an offset defined relative to the shell itself — a mid-surface-to-soffit offset on a deck that is superelevated or curved. Global offsets suit genuinely global quantities. Do not enter the same physical offset in both tabs; they add. ## End Releases **Node 1–4 Tx / Ty / Tz / Rx / Ry / Rz:** The connection condition at each corner, in global axes. {% hint style="danger" %} **Sign convention — the same as node supports and line end releases:** Entering `-1` intending to release a corner produces a **fixed** connection. Use `0` to release. Right-click for *Set to Free*, *Set to Fixed* and *Edit Stiffness…* rather than typing the number. {% endhint %} | Value | Meaning | | ------------ | ----------------------------------------- | | **−1** | **Rigid / fully connected** (the default) | | **0** | **Released** — free in that direction | | **positive** | **Semi-rigid** spring of that stiffness | *Engineering influence.* Corner releases are used far less often on shells than on lines, and they are easy to misuse: releasing a corner disconnects the shell from the mesh in that direction, which can leave a hole in the load path or make the element unstable. The common legitimate case is releasing a rotational degree of freedom where a shell meets a member that cannot carry that moment. Releasing the same degree of freedom at every corner leaves the element unrestrained in that direction. ## Overrides **Section Thickness:** An absolute override of the shell thickness, applying to stiffness, self-weight, mass and the drawn extrusion alike, and leaving the Thickness parameter itself intact. One lane still reads the Thickness parameter directly and so does not see it: export to SAP2000 or Midas. **Stiffness Thickness:** An absolute override of the thickness used for **stiffness only**. Membrane, bending and transverse-shear stiffness, the geometric stiffness behind buckling and P-Delta, the thermal load's rigidity and the recovered forces and stresses all follow it. Self-weight and mass do not, and neither does anything geometric: the drawn extrusion, tendon face offsets and the depth a through-thickness temperature gradient is measured across all stay on the section thickness. Blank or 0 means unset, and every quantity then uses one thickness exactly as before. *Engineering influence.* Use it where the stiffness a deck or panel should contribute is not the stiffness its own thickness implies **and** changing the thickness would falsify the weight: a bending thickness distinct from the membrane thickness, an effective thickness for a stiffened or corrugated panel smeared into a flat shell, or a haunched region modelled with uniform elements. It carries the `t` and `t³` scaling together, so it moves membrane and bending stiffness at once; to change one action alone the Membrane / Bending modifiers below are the sharper tool. {% hint style="warning" %} Weight is the reason this parameter exists, so check it. With a Stiffness Thickness set, the Weight column and every self-weight reaction stay on the *section* thickness. If you meant the shell to weigh what the override says, change Thickness or Section Thickness instead. **Composite results follow it.** The composite lane places its neutral axis and builds the section its stresses are read on from the same analysed thickness, so the reported centroid and the reported stresses describe the section the solve actually used. A Stiffness Thickness applied through a staged or case-scoped Override reaches composite output too. **Export:** OpenSees carries the override exactly, because its shell section is per element and takes a thickness and a density that together reproduce both the stiffness and the true mass. SAP2000 and Midas cannot represent it and write the base Thickness parameter, so an exported model has the real thickness for stiffness too. Those two exporters also drop every other per-element override (Section Thickness, the material overrides, the stiffness modifiers). {% endhint %} {% hint style="info" %} Two consequences worth knowing before you rely on a result: **Stresses are reported on the stiffness section.** Forces per unit length are recovered from the overridden rigidity, and the extreme-fibre bending stress is taken at half the *stiffness* thickness, not at the shell's real faces. That keeps stress consistent with the strain the model computed; it is not the stress in the real material at the real face. **A Stiffness Thickness alone is not a shell.** If you set it and leave Thickness at 0, you get an element that stiffens the model normally but weighs nothing, contributes nothing to any reaction or quantity, and draws with zero height. Set a real Thickness as well. {% endhint %} **Material Modulus of Elasticity / Poisson's Ratio:** Absolute overrides of the material properties. **Membrane Fxx / Fyy / Fxy Modifier (local):** Multipliers on the in-plane stiffness terms. **Bending Mxx / Myy / Mxy Modifier (local):** Multipliers on the out-of-plane stiffness terms. *Engineering influence.* These modifiers are how code-prescribed stiffness reductions are applied — cracked-section properties for a concrete deck or wall in a seismic analysis, or a deliberate softening in one direction to represent one-way behaviour or a jointed slab. They act in the shell's **local** axes, so their meaning depends on the element's orientation; applying a directional modifier to a mesh whose local axes are inconsistent produces a stiffness field that varies element by element. Reducing a modifier softens the shell in that action, increases its deflection, and **sheds** force into stiffer adjacent elements — so softening a deck increases the demand on the girders. A modifier of 1 (the default) leaves the term unchanged; 0 removes that stiffness entirely and will usually make the model unstable. Separating membrane from bending matters: a deck may be cracked in bending while retaining most of its in-plane stiffness, and modifying both together overstates the reduction. **Plate Type:** The bending formulation — **Thin** (Kirchhoff, no transverse shear) or **Thick** (Mindlin, shear-deformable), or Default to inherit the global setting. *Engineering influence.* Thin plate theory ignores transverse shear deformation, which is accurate while the element is slender relative to its span and increasingly wrong as it thickens. Thick (Mindlin) includes shear deformation and is the right choice for a deep or stubby panel; it is also what matches a SAP2000 thick-plate model. For a thin deck the two give nearly identical answers, so the setting matters most for thick slabs, pier caps modelled as plates, and deep webs. Choosing Thin for a genuinely thick element **overstates** stiffness and understates deflection. **Triangle Formulation:** The membrane/bending formulation used for three-noded shells — **DKGT** (the default, quad-compatible drilling) or **ANDeS** (high-performance membrane). *Engineering influence.* DKGT shares the drilling convention used by the quadrilateral elements, so it balances correctly where triangles and quads meet. The cost is that its membrane behaviour is constant-strain grade and therefore stiff in in-plane bending. ANDeS has a far better membrane and is the better choice for a panel dominated by in-plane bending — a shear wall or a deep-beam web — but it keeps a moment-carrying drilling convention that only behaves in a narrow envelope. Use ANDeS **only** where all three of these hold: the triangle's neighbours are not quads; the surface is flat rather than faceted-curved; and loads are applied at nodes rather than as distributed surface loads. Outside that envelope the errors are large and not obvious — a mixed mesh misreads displacement by around 8% and stress by around 30%, a pressurized faceted cylinder with a free end bulges by roughly 189%, and under distributed load the error is erratic and gets *worse* under refinement. DKGT holds around 0.02% across the same cases. When in doubt, leave it at DKGT. **Formulation (in use):** Read-only, reporting the formulation actually applied after the Default settings resolve. Check it when you are not sure which formulation a given element ended up with. **Modifier Axis Global X:** Controls the reference axis used to interpret the directional modifiers, so that "local xx" can be tied to a consistent global direction across a mesh. *Engineering influence.* This is what makes directional modifiers usable on a real mesh. Without a consistent reference, "Membrane Fxx modifier" means a different physical direction on each element, and a deck intended to be softened longitudinally ends up softened in a patchwork of directions. {% hint style="warning" %} **Geometric Nonlinear** is honoured by the solver on shells, but the column does **not currently render** in the sheet — its parameter is declared on FE Line rather than on FE Surface, so the column is dropped when the table is built. It **defaults to ON**, so in a nonlinear analysis your shells are already running with geometric stiffness included; you simply cannot see or change that from this table. If you need it off for a particular shell, it has to be set outside this sheet. {% endhint %} ## Groups **Group:** The FE Group used for construction-stage activation and selection. A shell whose group is inactive in a stage does not exist in that stage — which is how a deck cast in a later pour, or a temporary panel, is represented. ## Time Dependent **Casting Day:** The day the shell is cast, used by time-dependent analysis to compute concrete creep, shrinkage and modulus development. *Engineering influence.* Age at loading drives creep: concrete loaded young creeps substantially more than mature concrete. For a deck this matters directly — a slab made composite and loaded early creeps more, increasing long-term deflection and shedding force toward the girders over time. The value is inert unless the analysis includes time-dependent effects and the material is configured for them, and it should reflect the real pour sequence rather than being left at a single default across a deck poured in stages. ## Verification * Display shell local axes across the mesh and confirm they are consistent. Inconsistent axes make directional modifiers and result signs unreliable. * Check **Area** and **Weight** totals against a hand calculation of the deck. * Confirm a released corner reports zero force in the released direction. * Run a patch test: a uniformly loaded, simply supported plate should reproduce the closed-form deflection closely. This validates thickness, material and formulation together. * Refine the mesh and confirm results converge. Shell results that move substantially under refinement indicate the mesh, not the model, is governing. * Compare Thin against Thick on a representative panel; a large difference means shear deformation matters and Thick is required. * Where modifiers are applied, compare against the unmodified model to confirm they took effect and in the intended direction. * Check for warped or highly skewed elements, which degrade accuracy silently. ## Common mistakes * **Entering −1 to release a corner**, which fixes it instead. * **Inconsistent node ordering**, producing alternating result signs across a mesh and, at worst, bow-tie elements. * **Forgetting that bending stiffness scales with `t³`**, so a thickness entered in the wrong unit is off by a factor of 1000 in bending, not 10. * **Using Thin plate theory on a thick element**, overstating stiffness. * **Choosing ANDeS outside its envelope** — next to quads, on a faceted-curved surface, or under distributed load — where its errors are large and not obvious. * **Applying directional modifiers to a mesh with inconsistent local axes**, softening a patchwork of directions rather than the one intended. * **Modifying membrane and bending together** when only bending is cracked. * **Meshing too coarsely** near supports and concentrated loads, where shell stress gradients are steepest. * **Omitting offsets between deck and girder**, losing composite action entirely. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.openbrim.org/templates/openbrimfea/geometry-fea/surfaces-fea.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.