Checking a dynamic result

Check 1: the first period against a hand estimate

Always first, because everything else depends on it.

For a moment frame, T ≈ 0.1N seconds, N being the number of storeys. For any building, T ≈ 0.075H^0.75 with H in metres. For a stiff shear-wall building, expect something rather shorter.

What a mismatch means:

Reported period much SHORTER than expected — the model is too stiff. Gross concrete properties, an accidental restraint, missing releases, or a rigid diaphragm applied where it does not belong.

Reported period much LONGER than expected — the model is too soft, or too heavy. Missing members, an unintended mechanism, missing connectivity, or a mass error.

If check 1 fails, stop. There is nothing to be learned from the rest of the output until the model is right.

Check 2: the seismic mass

Take the software's total seismic weight, divide by the plan area and the number of floors. For an ordinary building expect 8 to 15 kN/m².

Much lower — mass is missing. Superimposed dead not applied, cladding forgotten, or the mass source set to self weight alone.

Much higher — factored loads have been used as the mass, or the full imposed load rather than the quasi-permanent fraction.

This check is independent of everything else in the model, which is what makes it valuable.

Check 3: the first-mode effective mass

For a regular building, the first mode should carry 60% to 85% of the total mass. For a base-isolated building it should carry nearly 100%.

Very low, under about 20% — the reported first mode is almost certainly a local mechanism, not a global mode. An unrestrained member vibrating on its own, a node connected to nothing, a slab element with no in-plane stiffness. Plot the mode shape: a local mechanism moves one small part of the structure and leaves the rest still.

Above about 90% for a multi-storey building — suspiciously high. Usually means the model has been over-constrained into behaving like a single mass.

Check 4: the base shear coefficient

V/W, the base shear divided by the seismic weight. For an ordinary building under a design earthquake, expect roughly 0.02 to 0.5.

Outside that range, assume a unit error until proved otherwise. The classic is a spectrum entered in g when the software expects m/s², or the reverse — a factor of 9.81, which turns a plausible 0.15 into either 1.5 or 0.015. Both are immediately visible in this check and almost invisible everywhere else.

This is the single most effective check in the list, because it tests the whole chain — mass, spectrum, units, combination — with one number.

Check 5: cumulative effective mass

Sum the effective masses of the modes retained. Codes require at least 90% in each direction, and that is a MINIMUM for base shear.

If floor accelerations are wanted, 90% is not enough. Acceleration is dominated by higher modes that carry almost no effective mass, so the criterion that establishes the base shear says nothing about whether the accelerations have converged. Add modes until the acceleration stops changing, and report by how much it changed.

Check 6: the time step, and convergence

For a time-history analysis: Δt ≤ T/20 for the highest mode retained, not the first. Then halve the step, rerun, and report the change in the quantity that governs.

A statement that the analysis converged is not a check. A statement that halving the time step changed the peak drift by 0.3% and the peak floor acceleration by 1.1% is a check, and it is one a reviewer can weigh.

Faults with distinctive signatures

Missing diaphragm constraint. First frequency far too low; the mode shape shows a floor distorting in plane rather than translating; wildly uneven distribution of shear between the vertical elements.

Accidental mechanism. A cluster of near-zero-frequency modes with negligible effective mass, all localised to one place.

Unit error in the spectrum. Base shear coefficient out by a factor near 9.81, or near 1 000, or near 1 000 000.

Mass error. Period out by roughly the square root of the mass ratio; base shear out by roughly the mass ratio.

Over-constraint. Period too short, first-mode effective mass too high, drifts too small.

Too few modes. Base shear plausible, floor accelerations far too low, and the cumulative mass table shows why.

Writing it down

A check that is not recorded did not happen, as far as anyone reviewing the work is concerned. For each one, record: what was checked, what value the model gave, what was expected and on what basis, and whether it passed.

The expectation and its basis are the parts that get left out, and they are the parts that make the record useful. 'First period 0.87 s; expected 0.8 to 0.9 s from 0.1N and 0.075H^0.75; pass' can be repeated by anyone. 'Period checked, OK' cannot be repeated by anybody, including its author six months later.