Building a model that answers the question

Mass: what moves with the building

The seismic mass is the mass that participates in the inertia. That is not the same as the mass used in a gravity analysis, and it is not the same as the factored design load.

Include: self weight of the structure; permanent finishes, screeds, ceilings and services; cladding; permanent equipment; the appropriate quasi-permanent fraction of the imposed load; the mass of any water or stored material that is always present.

Exclude: the transient part of the imposed load that will not be there; snow, except the fraction codes require; the mass below the level at which the structure is considered fixed.

Be careful with: partitions, which are often mass and stiffness both; plant, which may be isolated and therefore not fully participating; tanks, where sloshing means part of the water does not move with the structure.

The single most common mass error is using a load combination rather than a mass. A factored load is a force with a safety factor in it. Seismic mass is a physical quantity and has no factor.

The period goes as √m, so a 20% mass error moves the period by about 10%, and the base shear moves roughly in proportion to the mass. Neither error is subtle, but both are invisible in the output unless the mass is checked.

Stiffness: the state the structure will be in

For a steel frame the elastic stiffness is a reasonable model, with the qualification that connection flexibility and panel-zone deformation are real and are often omitted.

For reinforced concrete it is not. A concrete member under seismic demand is cracked, and cracked stiffness is a fraction of gross:

• beams roughly 0.3 to 0.5 of gross;
• columns 0.5 to 0.7, depending on axial load;
• walls 0.5 or lower.

Using gross section properties gives a structure that is too stiff, a period that is too short, and — since the short-period end of the spectrum is where accelerations are highest — a base shear that may be too high while the DRIFT is much too low. One error, two consequences in opposite directions, and the unconservative one is the drift.

Infill panels are the other large stiffness question. A masonry infill is very stiff until it fails, and a frame with infill on the upper floors and open ground floor has a soft storey whether or not the model says so.

Damping: consistent with the method

Module 12 covered this and the practical instruction is short.

In a modal analysis, specify modal damping ratios directly. You get exactly what you asked for in every mode.

In a direct integration, a damping matrix is needed. Rayleigh damping delivers the target at exactly two frequencies and something else everywhere else. Choose the two anchor frequencies to bracket the modes that matter — commonly the first mode and the highest mode with significant mass — and then check what the intermediate modes actually received.

A Rayleigh damping model anchored at the first and second modes gives very heavy damping to the higher modes, which quietly suppresses the floor accelerations an equipment engineer may be asking for.

Constraints: the ones you need and the ones you must not have

Rigid diaphragm where the floor genuinely acts as one — a concrete slab in plane. It removes the in-plane floor degrees of freedom and reduces the model enormously. Do not apply it to a floor with a large opening, a long narrow plan, precast units without a topping, or a transfer level.

Base fixity. A model fixed at the ground floor assumes the basement and foundation are rigid. Where the structure continues below with basement walls, the fixity level is a decision, and it moves the period.

Accidental restraints. The commonest modelling fault of all: a node inadvertently restrained, a member connected to a support it should not touch, a rigid link left over from an earlier version. These make the structure stiffer than it is and are hard to see in a plot.

Accidental mechanisms. The opposite: a member with releases at both ends and no other restraint, a node connected to nothing, a slab element with no in-plane stiffness. These produce near-zero-frequency modes that carry almost no mass, and they are the reason a first mode should always be inspected before it is used.