Most seismic models that fail review do not fail because the engineering was wrong. They fail because a procedural step was skipped: insufficient modes, unscaled base shear, a mass source that did not include what it should. This guide works through the three analysis methods, when each is required, and the specific checks that decide whether a model stands up to an authority or peer reviewer.
The three methods
| Method | What it does | Typical use |
|---|---|---|
| Equivalent lateral force | Applies a static lateral load pattern derived from a single fundamental period | Regular, low to mid-rise structures within code height and irregularity limits |
| Modal response spectrum | Combines the responses of multiple vibration modes against a design spectrum | The default for most buildings of any consequence, and mandatory where ELF is not permitted |
| Time history | Integrates the equation of motion against ground motion records, step by step | Irregular or critical structures, base isolation, energy dissipation devices, performance-based design |
The method is not a free choice. ASCE 7 sets out permitted analysis procedures against seismic design category, structural system, height and irregularity. Eurocode 8 does the equivalent through its own applicability criteria for the lateral force method. If your structure fails the regularity or height tests, the simpler method is unavailable regardless of preference.
Equivalent lateral force, and where it stops
ELF reduces the seismic demand to a base shear derived from the seismic weight, the design spectral acceleration at the fundamental period, the response modification factor and the importance factor. That base shear is then distributed up the building with an exponent that accounts for higher mode effects in taller structures.
It works well for what it was built for: a regular building where one mode dominates. It becomes unreliable when the mass or stiffness distribution means no single mode dominates, when torsion is significant, or when a setback or soft storey concentrates demand at one level. That is precisely why the code restricts it.
Modal analysis comes first, and mass participation is the gate
Before any response spectrum results mean anything, the modal analysis has to capture enough of the structure's mass. ASCE 7-16 requires enough modes to capture at least 90 percent of the actual mass in each orthogonal direction. Eurocode 8 sets a comparable requirement.
A model that reports 70 percent participation is not a conservative model, it is an incomplete one: the missing 30 percent of mass is not being shaken at all. The usual causes are too few modes requested, a mass source that excludes some of the seismic weight, or diaphragms modelled in a way that locks mass out of the lateral system.
Check mass participation before you look at a single drift or force result. It is the cheapest check in the sequence and the one that invalidates everything downstream if it fails.
Modal response spectrum analysis
Each mode is treated as a single degree of freedom oscillator, its peak response read from the design spectrum at that mode's period, and the modal responses then combined. Because the modal peaks do not occur simultaneously, they cannot simply be added.
SRSS versus CQC
The square root of the sum of the squares is the long-standing combination rule, and it is appropriate when the modes are well separated in period. Where modes are closely spaced, which happens routinely in buildings with similar stiffness in two directions or with significant torsional response, SRSS can misestimate the combined response because it assumes the modal responses are statistically independent.
Complete quadratic combination accounts for correlation between closely spaced modes and is the safer default for buildings with torsional modes near the translational ones. Most modern software offers both. The choice should be stated in the calculation report rather than left as a software default nobody examined.
Base shear scaling, the step most often missed
A response spectrum analysis frequently produces a base shear lower than the ELF value, because modal combination recovers some of the conservatism built into the static method. Codes do not allow the full benefit of that reduction.
ASCE 7-16 requires that where the modal base shear falls below 85 percent of the ELF base shear, the modal results are scaled up so that it does not. Skipping this produces member forces that are uniformly and invisibly low throughout the structure, which is the most dangerous category of modelling error because nothing in the output looks wrong.
Show the scaling calculation in the report. A reviewer who cannot see whether scaling was applied will assume it was not, and will be right often enough to keep asking.
Time history analysis
Time history analysis integrates the structure's response to an actual or synthetic ground motion record through time. Linear time history keeps the structure elastic. Non-linear time history allows plastic hinging, and is the basis of performance-based design.
It is the most informative method and the most demanding. Ground motions have to be selected and scaled to match the design spectrum over the period range of interest, usually with a suite of records rather than one, and the results interpreted as a statistical set. Non-linear work additionally requires defensible hinge properties and a view on degradation, which is modelling judgement rather than code compliance.
For most commercial buildings it is not required and not proportionate. For base-isolated structures, buildings with dampers, structures with severe irregularity, or any project where a performance objective has been contracted rather than a code minimum, it is the only method that answers the question.
Story drift
Drift governs more mid-rise designs than strength does. ASCE 7 sets allowable story drift in a table indexed by structural system and risk category, and the drift to be checked is the amplified inelastic drift, not the elastic drift the analysis reports directly. Forgetting the amplification is a common and significant error.
Drift also controls damage to non-structural elements, which is where occupants actually experience an earthquake. Cladding, glazing, partitions and lift guide rails all have their own tolerances, and a building that meets the code drift limit can still fail its facade if the facade was specified against a tighter one.
Irregularities that force a higher method
- Torsional irregularity. Diaphragm drift at one end substantially exceeding the average, usually from an eccentric core or an asymmetric shear wall layout.
- Soft or weak storey. A level with markedly lower stiffness or strength than the one above, classically an open ground floor.
- Mass irregularity. A level significantly heavier than its neighbours, often a plant floor or a transfer level.
- Vertical geometric irregularity. Setbacks changing the lateral system footprint up the building.
- In-plane discontinuity. A vertical element offset or terminated, transferring force through a diaphragm or beam never intended for it.
Irregularities have to be classified explicitly, because the classification drives both the permitted analysis method and additional detailing requirements. An unclassified irregularity is a review query waiting to happen.
A model that survives review
- Site seismic parameters, site class and risk category stated with their source.
- Response modification factor consistent with the system actually detailed.
- Mass source defined and reconciled against the computed seismic weight.
- Mass participation at or above the code threshold in both directions.
- Modal combination rule stated, with CQC used where modes are closely spaced.
- Base shear scaling calculated and shown, not assumed.
- Drift amplified and compared to the governing limit level by level.
- Irregularities assessed, classified and their consequences carried through.
- Stability and second order effects verified.
None of this is exotic. It is a sequence, and the failures come from skipping steps under programme pressure rather than from any difficulty in the engineering. Our ETABS seismic and lateral analysis service runs that sequence on every model, with the analysis basis agreed in writing before modelling begins and the native model handed over so you can verify every assumption. Loading inputs come from the same basis as our wind and seismic load analysis, and the applicable edition of ASCE 7 is covered in our guide to ASCE 7-16 and 7-22.
Frequently asked questions
What is the difference between response spectrum and time history analysis?
Modal response spectrum analysis combines the peak responses of multiple vibration modes read from a design spectrum, giving a statistical estimate of peak demand. Time history analysis integrates the equation of motion against a ground motion record through time, giving the actual response sequence. Response spectrum is the practical default for most buildings. Time history is required for base isolation, damped structures, severe irregularity and performance-based design.
How many modes should a seismic model include?
Enough to capture at least 90 percent of the actual mass in each orthogonal direction, which is the ASCE 7-16 requirement. Eurocode 8 sets a comparable threshold. A model reporting lower participation is incomplete rather than conservative, because the missing mass is not being excited at all. Check participation before looking at any force or drift result.
When should CQC be used instead of SRSS?
Use SRSS when the vibration modes are well separated in period. Use complete quadratic combination where modes are closely spaced, which is common in buildings with similar stiffness in both directions or with torsional modes near the translational ones. SRSS assumes modal responses are statistically independent, so it can misestimate the combined response when they are not.
What is base shear scaling and why does it matter?
A response spectrum analysis often produces a base shear lower than the equivalent lateral force value. ASCE 7-16 requires that where the modal base shear falls below 85 percent of the ELF base shear, the modal results are scaled up. Skipping it produces member forces that are uniformly low throughout the structure with nothing in the output appearing wrong, which makes it a particularly dangerous error.
Which story drift should be checked against the code limit?
The amplified inelastic drift, not the elastic drift the analysis reports directly. ASCE 7 gives allowable story drift in a table indexed by structural system and risk category. Note also that non-structural elements such as cladding, glazing and lift guide rails have their own drift tolerances, which can be tighter than the code limit for the structure.