Key Takeaways
A credible finite element analysis submission is more than a colourful model or a long output file. It should allow the design and checking engineers to follow the assumptions, test the model’s behaviour, and connect the results to code-compliant construction details.
- Define the structural system, restraints, materials, and load paths clearly.
- Verify loads, combinations, design criteria, and relevant Singapore requirements.
- Demonstrate mesh sensitivity and investigate numerical warnings rather than hiding them.
- Compare important results with hand checks, equilibrium checks, and simplified models.
- Present an auditable report that connects analysis, member design, drawings, and revisions.
1. Understand the role of FEA in BCA submissions
Finite element analysis can provide useful insight into structures with irregular geometry, complex load paths, or interactions that are difficult to capture with simple models. For a BCA submission, however, the software output is supporting evidence rather than a substitute for engineering judgement. The checking engineer needs to understand what was modelled, why it was modelled that way, and whether the reported behaviour is credible.
When finite element analysis is appropriate
FEA is most appropriate when the structure contains features such as transfer zones, unusual openings, shell action, complex connections, or significant interaction between elements. It can also help investigate local force effects that are not visible in a simple beam-and-column idealisation. The method should be proportionate to the engineering question: a detailed model is not automatically a better model if its inputs and boundary conditions are uncertain.
The design engineer should state the purpose of the analysis at the outset. That purpose might be to determine global actions, assess local stresses, study deformation, or examine stability. A clear purpose helps the checker distinguish governing results from incidental peaks in a large data set.
How FEA supports structural plan submissions
A structural plan submission must demonstrate that the proposed system is safe, serviceable, and consistent with the drawings and specifications. FEA can support that demonstration by documenting global responses, reactions, member actions, deformation patterns, and the behaviour of critical regions. It is especially useful when the load distribution cannot be explained adequately through a conventional manual calculation alone.
The report should explain how analysis results were carried into the design. A contour plot without units, load-case identification, or a stated design interpretation is difficult to check. By contrast, a concise result summary gives the Accredited Checker a route from the model to the structural plans and calculations.
The responsibilities of the design engineer and checking engineer
The design engineer remains responsible for selecting the structural concept, establishing design criteria, building the model, interpreting results, and preparing the design. The checking engineer independently reviews those decisions, including loads, combinations, boundary conditions, analysis methods, stability, member design, and detailing. Independence does not mean repeating every keystroke; it means applying an appropriately independent route to the important engineering conclusions.
The review should also consider whether the submission is complete. Drawings, calculations, analysis files or extracts, material specifications, geotechnical information, and relevant design assumptions should tell one consistent story. Where a result appears unusual, the designer should explain it before the checker has to infer the reason from an unexplained image.
Verification versus validation in structural engineering
Verification asks whether the numerical model and implementation solve the stated mathematical problem correctly. Typical activities include checking units, element formulation, convergence, equilibrium, and sensitivity to modelling choices. Validation asks whether the mathematical model is a reasonable representation of the real structure, including its supports, materials, connections, construction state, and expected physical behaviour.
Both are needed for FEA Verification , BCA Submissions. A model may converge neatly and still be wrong because a restraint was idealised incorrectly. Conversely, a physically sensible model may produce unstable results because its mesh, contact definition, or numerical settings are unsuitable.
2. Establish a reliable analysis model
Model reliability begins before the solver runs. The geometry, material properties, section definitions, element connectivity, supports, and load introduction points should correspond to the intended structure and to the information shown on the submitted drawings. A checker will usually focus first on the model’s assumptions because errors at this stage can make later numerical precision misleading.
Defining geometry, materials, sections, and structural systems
The model description should identify the structural framing system, member dimensions, slab and wall thicknesses, material grades, stiffness assumptions, and relevant constitutive properties. It should also distinguish between analytical dimensions and construction dimensions where offsets, centreline modelling, or effective widths are used. These choices affect stiffness, load distribution, and design actions.
Material parameters should be traceable to the design basis. If stiffness modifiers, cracked-section properties, composite action, or orthotropic behaviour are adopted, the report should explain their source and purpose. Consistency between the analysis model and the member design assumptions is particularly important for reinforced concrete structures.
Representing supports, restraints, releases, and connections
Supports are not merely convenient points that make a model stable. They represent the way forces and movements are transferred to foundations, bearings, adjoining members, or the ground. The report should state which translations and rotations are restrained, where releases are applied, and how connection stiffness or eccentricity is represented.
Over-restraint can suppress real movement and inflate forces, while under-restraint can create mechanisms or unrealistic deformation. Connection assumptions deserve specific attention at transfers, bracing nodes, beam-column joints, base plates, and interfaces between different structural systems.
Modelling load paths and interactions between structural elements
A reliable model makes the intended load path visible. Gravity loads should move through slabs, beams, columns, walls, foundations, and supporting ground in a way that agrees with the structural concept. Lateral actions should likewise have a defensible route through diaphragms, frames, cores, bracing, foundations, and the ground system.
Where elements interact, the designer should explain whether the model captures that interaction directly or approximates it through constraints, springs, links, offsets, or equivalent loads. This is also where CORENET X may be relevant to the wider coordination process, since coordinated digital information can help teams identify conflicts before regulatory approval, but the engineering model still requires professional review.
Identifying simplifications that require engineering justification
Every model simplifies reality. The question is whether the simplification preserves the behaviour relevant to the design decision. Examples include rigid diaphragm assumptions, omitted secondary framing, smeared reinforcement, linear soil springs, or idealised rigid connections.
A useful report records the simplification, its expected effect, and the check used to confirm that it is acceptable. If an omitted component could change stiffness, force redistribution, vibration, stability, or detailing, it should be addressed explicitly rather than left implicit.
3. Verify loads, combinations, and design criteria
Load verification is a central part of independent technical checking. The reviewer should trace each action from its source and tributary area into the model, then confirm that the combinations used for strength and serviceability match the design basis. This includes the possibility that a model is numerically correct but loaded with the wrong magnitude, direction, duration, or combination factor.
Checking dead, live, wind, seismic, and imposed construction loads
Dead loads should include the structural self-weight and relevant finishes, walls, façade components, services, and permanent equipment. Live loads should reflect the actual occupancy and use rather than a convenient default. Wind and seismic actions require attention to direction, exposure, mass assumptions, torsional effects, and the applicable local criteria.
Construction loads, temporary storage, erection stages, maintenance actions, and other imposed conditions may govern particular members even when they do not control the final global design. The checker should ask whether these actions have been assigned to the right locations and whether the model’s mass and stiffness reflect the condition being assessed.
Applying Singapore Standards and relevant BCA requirements
The design basis should identify the applicable Building Control requirements, Singapore Standards, National Annexes, project specifications, and any accepted engineering assumptions. Structural plan submissions commonly rely on Singapore Standards for concrete and steel design, with local requirements affecting wind, seismic considerations, durability, and detailing.
The report should not simply list codes. It should show where they influence load factors, combinations, material parameters, limit states, member checks, and detailing. A careful reviewer also checks that cited editions and design parameters are consistent throughout the calculations and drawings.
Reviewing ultimate and serviceability load combinations
Ultimate combinations are used to assess strength and stability, while serviceability combinations address deflection, cracking, vibration, movement, and other performance criteria. The distinction should be visible in the model setup and in the result presentation. Envelope results need enough information for the checker to identify which load cases actually govern.
Particular care is required when combinations include nonlinear behaviour, staged loading, tension-only elements, contact, or second-order effects. The designer should explain whether envelopes are generated before or after nonlinear solution and how that affects the interpretation of reported actions.
Accounting for staged construction, temporary works, and existing conditions
Construction sequence can alter stiffness, support conditions, load transfer, and the age or strength of materials when actions are applied. It may therefore be necessary to assess stages such as propping removal, transfer loading, excavation, backfilling, prestress application, or connection installation.
For alteration and strengthening work, existing conditions should be based on available records and site information, with uncertainty stated clearly. Temporary works should not be treated as outside the structural design discussion when they influence the permanent structure or create critical intermediate states.
4. Demonstrate mesh and numerical convergence
Mesh quality affects what an FEA model can resolve and how confidently local results can be interpreted. A checking engineer will look beyond the number of elements and consider element formulation, aspect ratios, integration, connectivity, refinement, and the relationship between mesh size and the structural feature under review. Numerical convergence is evidence of stability in the result, not proof that the idealisation is physically correct.
Selecting suitable element types and mesh density
Element types should reflect the behaviour being studied. Beam, shell, solid, interface, contact, and link elements each carry assumptions about degrees of freedom, thickness, stiffness, and force recovery. The model description should explain why the selected formulation is appropriate for the geometry and design question.
Mesh density should be sufficient to capture curvature, gradients, load introduction, openings, and local action effects without creating unnecessary numerical noise. A coarse global mesh may be reasonable for overall behaviour if critical regions are treated separately and the limitations are stated.
Refining the mesh around openings, supports, connections, and discontinuities
Local refinement is often needed where geometry or load transfer changes abruptly. Openings, re-entrant corners, concentrated supports, connection zones, wall boundaries, transfer regions, and abrupt changes in thickness deserve particular attention. Refinement should follow the physical feature, not simply produce a denser picture.
The refined results should be interpreted with care. A local peak may indicate a real design demand, a recoverable nodal effect, or a mathematical singularity. The report should state how the value was extracted and how it was used in member or connection design.
Checking sensitivity to mesh changes
A practical convergence study compares selected response quantities across at least two or more mesh densities. Useful quantities may include reactions, global displacement, section forces, averaged stresses, or the demand in a defined design region. The comparison should use the same load case, boundary conditions, extraction method, and result location.
A small change in mesh should not cause a large unexplained change in a governing design conclusion. If it does, the designer should refine the investigation, adjust the result extraction method, or acknowledge the uncertainty and design the region conservatively.
Recognizing singularities, distorted elements, and numerical instability
Singularities can occur at point loads, sharp corners, fully rigid constraints, idealised supports, and abrupt changes in stiffness. Their stress values may increase without converging to a finite local number. They should be identified rather than presented as ordinary material stresses.
Distorted elements, excessive aspect ratios, negative Jacobians, ill-conditioning, warnings, and non-convergent iterations also require investigation. A stable-looking contour plot cannot override solver diagnostics or a deformation pattern that conflicts with the structural system.
5. Validate model behavior and analysis results
Validation is where the numerical model is tested against engineering expectations. The review should consider whether reactions balance applied actions, whether deflected shapes make sense, and whether force distribution follows the intended load path. Results should be read as behaviour, not just as isolated maximum values.
Comparing FEA results with hand calculations and simplified models
Independent calculations provide a valuable scale check. A beam strip, tributary-area calculation, frame idealisation, diaphragm model, or simple stability estimate can reveal an incorrect stiffness, load assignment, unit conversion, or support condition. The simplified method does not need to reproduce every local stress; it should test the response that matters to the design conclusion.
For complex structures, alternative modelling approaches may be appropriate for critical elements or overall behaviour. Differences should be explained through modelling assumptions rather than dismissed because two methods do not produce identical numbers.
Reviewing reactions, equilibrium, deformation, and load distribution
The first review of output should include total applied load, support reactions, residual forces, rigid-body motion, and global deformation. These checks often find errors faster than a page-by-page review of contour plots. Reaction patterns should also be compatible with the foundation arrangement and the expected structural action.
Deformation shapes provide a useful visual test. A slab that bends in the wrong direction, a frame that sways unexpectedly, or a wall that attracts no lateral force may indicate a connectivity or restraint problem rather than an interesting engineering discovery.
Checking deflection, vibration, cracking, buckling, and stability behavior
Serviceability checks should match the intended use and the design criteria. Deflection limits, crack control, vibration, long-term effects, buckling, P-delta response, and global stability may require different combinations and modelling assumptions. The design team should state which behaviours have been assessed and which are outside the model’s scope.
For reinforced concrete, member verification includes flexure, shear, punching shear, axial-flexural interaction, torsion where significant, deflection, and crack control. For steelwork, it may include section classification, local and lateral-torsional buckling, column stability, P-delta effects, plate stability, fatigue where relevant, and connection behaviour.
Investigating unexpected stress concentrations and nonlinear responses
Unexpected peaks should trigger a question, not an automatic design decision. The designer should determine whether the result arises from geometry, load introduction, contact, mesh, material behaviour, restraint, or a real force-transfer mechanism. Nonlinear responses should be described with their solution controls, convergence criteria, material laws, and loading path.
Where a local concentration is physically meaningful, it should be carried into an appropriate design check. Where it is a numerical artefact, the report should explain the averaging, integration, or idealisation used to obtain a meaningful engineering quantity.
6. Connect FEA results to member design and detailing
FEA becomes useful for approval only when its results lead to defensible member sizes, reinforcement, steel sections, connections, and details. The design calculations should identify how actions were recovered, combined, enveloped, or averaged before being checked. They should also make clear where engineering judgement was used to convert a continuum result into a practical design region.
Translating forces and moments into reinforcement or steel design
Shell forces, beam actions, solid stresses, and nodal reactions must be translated into design quantities that match the selected code checks. For reinforced concrete, this may involve flexural reinforcement, shear reinforcement, development length, cover, spacing, confinement, and crack control. For steel, it may involve section capacity, stability, plate behaviour, welds, bolts, and force transfer through connections.
The extraction method matters. A peak at one node should not automatically dictate reinforcement across an entire slab, while averaging over too large a region may conceal a critical local demand. The report should state the design strip, section cut, averaging width, or connection interface used.
Verifying beams, columns, slabs, walls, foundations, and connections
Member design should cover the critical structural elements and all relevant limit states. The following compact view helps organise the relationship between analysis output and checking focus.
| Design area | Typical actions or checks | Detail requiring attention |
|---|---|---|
| Beams and slabs | Flexure, shear, torsion, deflection, crack control | Development length, spacing, congestion |
| Columns and walls | Axial-flexural interaction, slenderness, stability | Confinement, splices, wall boundaries |
| Foundations | Bearing, settlement, structural capacity, punching | Differential settlement, pile caps, load distribution |
| Steel members | Section capacity, local and lateral-torsional buckling | Bracing, effective properties, fatigue where relevant |
| Connections | Force transfer, stiffness, strength, ductility | Eccentricity, edge distances, weld access |
The table is a planning aid, not a replacement for project-specific calculations. The governing analysis actions still need to be traceable to the relevant member, connection, foundation, and detailing checks.
Reviewing punching shear, anchorage, buckling, and second-order effects
Local failure modes often control even when global stresses appear moderate. Slab-column regions should be checked for punching shear, while anchorage and development lengths should suit the actual force transfer and available geometry. Steel members require appropriate buckling lengths, bracing assumptions, and second-order treatment.
Columns and walls may require biaxial interaction, confinement, slenderness, and P-delta checks. Foundations may require attention to group effects, settlement, eccentricity, and the compatibility between geotechnical assumptions and structural reactions.
Confirming that drawings and details match the analytical model
The final review should compare model geometry, member sizes, levels, openings, supports, reinforcement zones, connection assumptions, and construction joints with the drawings. A change made during coordination can invalidate an earlier analysis assumption even when the model file remains unchanged.
Details should therefore be checked as part of the analytical evidence, not after it. Reinforcement layouts, cover, bar spacing, splices, anchorage, steel stiffeners, bolts, welds, and base plates must be buildable and consistent with the forces the model transfers.
7. Prepare FEA evidence for checking and BCA approval
A good calculation package lets an independent reviewer reproduce the design logic without searching through disconnected files. It should begin with the design basis and structural concept, then move through model assumptions, loads, combinations, analysis checks, member design, drawings, and revisions. For Singapore authority work, a professionally prepared submission also needs to sit within the responsibilities of the QP and the applicable checking process.
Structuring the calculation report and model description
Start with a concise project description, structural system, design standards, materials, geometry, and analysis objectives. Follow with boundary conditions, element types, connectivity, load paths, combinations, solution settings, convergence checks, and result extraction methods. A separate section should identify limitations and assumptions that require review.
The report should make the critical path easy to follow. If a transfer slab, wall boundary, connection, foundation, or temporary stage governs, lead the checker toward that evidence rather than burying it among unfiltered output pages.
Documenting software versions, assumptions, inputs, and design parameters
Record the software name and version, analysis modules, solver settings, units, model revision, input files, material data, section properties, load definitions, combinations, and design parameters. Screenshots can help, but they should supplement structured tables and explanatory text rather than replace them.
A revision-controlled record is particularly valuable when comments lead to changes in geometry, loads, restraints, mesh, reinforcement, or code parameters. The checker should be able to determine which model produced the submitted results.
Presenting governing results with clear units, scales, and load cases
Results should use legible units, sensible contour limits, visible deformation settings, and clear labels for load cases and combinations. The report should distinguish nodal values, element values, averaged results, reactions, envelopes, and design actions. Maximum and minimum values need context, including their location and whether they are physically meaningful.
A short governing-results schedule is often more useful than dozens of unfiltered plots. It should state the selected response, design location, governing combination, comparison or limit, and the resulting engineering action.
Addressing checking comments, revisions, and independent verification records
Checking comments should be answered directly, with the response linked to a revised calculation, model, drawing, or design decision. If a model is changed, identify the change, the reason for it, and the effect on governing results. A response that says only “updated” leaves too much for the reviewer to reconstruct.
A submission package is stronger when it preserves independent verification records, hand checks, alternative models, equilibrium checks, and design review notes. That discipline reflects the wider work of Aman Engineering Consultancy only if the link’s subject genuinely applies; here, the relevant principle is professional engineering review and authority-submission preparation, not a claim about a particular software product.
Conclusion
FEA evidence for a BCA submission should be clear enough to audit and grounded enough to trust. When the model, loads, numerical checks, validation exercises, member design, detailing, and revision records align, the analysis becomes a coherent part of the structural safety case rather than an isolated software output.
Frequently Asked Questions
Is FEA always required for a BCA submission?
No. The appropriate analysis method depends on the structural system, complexity, design criteria, and professional judgement. Conventional calculations may be sufficient for straightforward elements, while FEA can be appropriate for irregular geometry, complex load paths, transfers, interactions, or local behaviour.
What does a checking engineer review first in an FEA model?
The review commonly begins with the structural concept, geometry, materials, supports, connectivity, loads, combinations, and analysis objectives. These assumptions determine whether the numerical results have a meaningful engineering basis.
How can mesh convergence be demonstrated?
Compare selected response quantities using different mesh densities while keeping the load case, boundary conditions, extraction method, and result location consistent. Explain any material differences and show how the final design decision remains reliable.
Are stress singularities a design failure?
Not necessarily. Singularities can arise from idealised point loads, sharp corners, rigid constraints, or abrupt stiffness changes. They should be identified and handled using an appropriate engineering interpretation rather than treated as ordinary finite stresses.
What should be included in an FEA calculation report?
Include the design basis, structural system, model description, materials, element types, supports, loads, combinations, software information, numerical checks, governing results, member design, assumptions, limitations, and revision history. The report should connect analysis results to drawings and details.
How are FEA results used in reinforced concrete design?
Analysis actions may be converted into reinforcement demands, shear and punching checks, axial-flexural interaction, torsion where relevant, deflection, crack control, anchorage, development, and detailing checks. The recovery and averaging methods should be stated clearly.
Why are hand calculations still useful when FEA is available?
Hand calculations provide independent checks on scale, equilibrium, stiffness, reactions, load distribution, and critical member behaviour. They can reveal modelling or input errors that a converged computer solution will not identify by itself.