Seismic weight is the mass your building brings to an earthquake, and it is built almost entirely out of the numbers you already computed in the gravity takedown. Codes define an effective seismic weight, W in ASCE 7’s notation, as the full dead load plus specific extras: a slice of storage live load, the partition allowance, the operating weight of permanent equipment, and a share of heavy roof snow. Ordinary occupancy live load stays out, because people and furniture are not reliably present or attached when the ground moves. That definition makes seismic weight a bookkeeping exercise on top of a good takedown, and a trap on top of a sloppy one: every storey’s W, and where it sits in plan, comes straight from your dead load model, which is why StructLoads is the right starting point, its per-storey, per-element dead loads give you W as a traceable sum rather than an area guess, and the same model hands you the mass distribution the analysis needs.
Why earthquakes load mass, not area
Wind pushes on surface; earthquakes shake mass. Ground acceleration forces every kilogram of the building to accelerate with it, and by Newton’s second law the inertial force is mass times acceleration, which is why seismic demands scale with weight and why the equivalent static approach that most low- and mid-rise design uses computes a base shear as a seismic coefficient times W. Halve the weight and you halve the seismic force; add a heavy roof garden and you have added seismic demand in the same proportion, a trade gravity design alone never surfaces.
That physics explains the composition of W. Dead load is in because it is always there, rigidly attached, and it accelerates with the frame. Ordinary live load is out because the code judges its reliably-present, attached fraction to be small: the crowd, the desks, the stock on shelves in a shop mostly are not bolted to the structure, and their average presence during a rare event is low. The exceptions are exactly the cases where that judgment fails: storage occupancies, where racks stay loaded as a way of life, contribute a defined share (ASCE 7 takes 25 percent of the storage live load); partitions, which are real dead weight wherever they stand, enter through the partition allowance when the floor design used one; permanent equipment enters at its operating weight, fluids included; and where the flat roof snow load is heavy (above 1.44 kPa or 30 psf in ASCE terms), a 20 percent share of it joins W, since deep winter snow is effectively resident mass.
Building W from the takedown, storey by storey
The workflow is a disciplined re-read of your gravity model. For each storey, W collects: the slab and framing self-weight tributary to that level, the superimposed dead load (finishes, services, ceilings) as built into the takedown, the partition allowance where it was applied, cladding and facade weight for the storey height it hangs on, the storey’s share of walls and columns (conventionally half the wall and column weight above and below the floor line), plus the storage and equipment and snow shares where they apply. Sum per storey, then sum the storeys for the total W the base shear formula wants.
| Component | In seismic weight? | Notes |
|---|---|---|
| Structural self-weight | Yes, fully | Slabs, beams, columns, walls from the takedown |
| Superimposed dead load | Yes, fully | Finishes, services, ceilings as modeled |
| Partition allowance | Yes, where used | Real distributed wall weight, not a live load |
| Ordinary occupancy live load | No | Not reliably present or attached |
| Storage live load | Partially | 25 percent share in ASCE 7 practice |
| Permanent equipment | Yes | Operating weight, contents included |
| Heavy flat-roof snow | Partially | 20 percent where pf exceeds 30 psf / 1.44 kPa |
The storey-by-storey structure matters beyond the total, because the analysis distributes base shear up the height in proportion to each level’s weight (and height), and it locates each storey’s inertial force at that storey’s centre of mass. A takedown that lumps three storeys together, or smears cladding weight evenly when it actually concentrates on two facades, degrades both the vertical distribution and the torsion picture. This is where the multi-storey takedown discipline pays a second dividend: the same per-level, per-element bookkeeping that makes column loads traceable makes seismic mass traceable, and in StructLoads that bookkeeping already exists as data rather than as a spreadsheet you rebuild.
A worked storey shows the arithmetic’s shape. Take a 30 m by 20 m office floor: a 200 mm concrete slab at roughly 4.8 kPa self-weight gives 2880 kN; superimposed dead load at 1.5 kPa (screed, ceiling, services) adds 900 kN; the partition allowance at 1.0 kPa adds 600 kN; perimeter cladding at, say, 1.2 kN per metre of facade over a 3.5 m storey and a 100 m perimeter hangs about 420 kN on this level; and the storey’s half-shares of columns and core walls contribute perhaps 350 kN in a concrete building of this size. Storey W: roughly 5150 kN, with the office live load of 2.4 kPa contributing exactly nothing. Now make it an archive floor instead: a 7.2 kPa storage live load contributes 25 percent, 1080 kN, and this single reclassification adds more seismic mass than the cladding and partitions combined. The numbers are illustrative, yours come from your own takedown, but the proportions are the lesson: dead-side bookkeeping dominates, and occupancy classification can swing a storey’s mass by twenty percent at a stroke.
The mistakes that actually move the answer
Seismic weight errors cluster into five patterns worth checking by name. Forgetting the partition allowance: floors designed with a partition allowance carry that weight in reality, and dropping it from W understates mass on every storey at once. Ignoring cladding: heavy facades, masonry, precast, stone, add substantial perimeter mass, and hanging it on the wrong levels distorts the storey distribution even when the total survives. Treating storage as ordinary occupancy: a warehouse or archive floor at 25 percent of its storage live load can add more to W than its entire dead load allowance for finishes, and classifying it as office-like live load silently deletes that mass. Using service-level snow judgment: the snow share is a defined fraction triggered by the mapped flat-roof value, not a personal estimate of winter. And double-counting through combinations: W is an input to the seismic force, and the seismic load case then combines with gravity per the code’s combinations; mass shares of live and snow inside W do not mean those loads also appear at full value in the same seismic combination, a bookkeeping distinction the load combination rules handle once you keep W’s role separate.
A sixth pattern is subtler: vertical irregularity by accounting error. Mass irregularity provisions compare adjacent storey weights, and a bookkeeping mistake, equipment assigned to the wrong level, a green roof’s saturated weight omitted, can either mask a real irregularity or invent a fake one, changing the required analysis method. The remedy is the same traceability that protects the gravity side: every kilogram in W attributable to a component you can point at, per storey, which is exactly the form a StructLoads takedown already holds.
Where seismic weight meets the rest of the design
W is the demand side’s seed, and it echoes through the design in ways worth anticipating. Base shear scales with it linearly in the equivalent static method, so mass-saving decisions, lighter floors, lighter cladding, are seismic-force reductions with the same directness that they are foundation savings, the same lesson fast column axial load thinking teaches about gravity efficiency. Overturning and drift checks inherit the storey distribution, so honest per-level mass placement is a stability input, not a nicety. And foundations see the combination: gravity loads from the takedown plus seismic actions computed from a W that came from the same takedown, which is the practical argument for one consistent load model feeding both, rather than a gravity spreadsheet and a separate seismic guess that drift apart over revisions.
For regions designing to other codes the shape survives translation: Eurocode 8 builds its seismic mass as dead load plus combination factors on variable actions, category-dependent shares of live load and snow, the same reliably-present logic with different symbols, and the general structural load taxonomy underneath is identical. The universal is the discipline: seismic weight is your dead load model, extended by the code’s short list of resident extras, and it is only ever as good as the takedown it stands on.
Key takeaways: seismic weight
Effective seismic weight is the takedown wearing a second hat: all dead load, the partition allowance, permanent equipment at operating weight, a quarter of storage live load, and a fifth of heavy flat-roof snow, summed per storey with honest placement in plan and height. Ordinary live load stays out. The dangerous errors are omissions, partitions, cladding, storage mass, and lumping that blurs the storey distribution. Build W from a traceable gravity model, StructLoads holds exactly that, and the seismic side inherits numbers you can defend instead of area-based guesses that surface in review.
Quick answers
How do you calculate seismic weight from dead loads?
Sum, per storey, the full dead load from your takedown, structure, superimposed dead load, cladding, plus the partition allowance where used, permanent equipment at operating weight, 25 percent of storage live load, and 20 percent of flat-roof snow where the mapped value exceeds 30 psf (1.44 kPa). Ordinary occupancy live load is excluded. The per-storey sums feed the vertical force distribution, and their plan placement sets each level’s centre of mass, so build W from the takedown’s element-level numbers rather than a blended area rate.
Why is live load mostly excluded from seismic weight?
Because seismic weight models mass that is reliably present and moves with the structure, and ordinary occupancy live load, people, furniture, loose contents, is neither dependably there during a rare event nor rigidly attached. The code keeps the exceptions where that reasoning fails: storage floors stay loaded as a way of life (hence the 25 percent share), partitions are real standing weight, permanent equipment is bolted down, and deep roof snow in heavy-snow regions is resident for whole seasons.
Does the partition allowance count in seismic weight?
Yes. Where the floor design carries a partition allowance, that weight represents real walls standing on the floor, and it belongs in W even though live-load-style reasoning might tempt you to drop it. Forgetting it understates mass on every storey simultaneously, which biases base shear low. The allowance’s dual character, designed like a distributed load, weighing like dead load, is exactly why it deserves a deliberate line in the seismic weight sum rather than an assumption either way.
What is the biggest mistake in seismic weight calculations?
Omission by misclassification: storage floors treated as ordinary occupancy (deleting the 25 percent share that can rival the finishes allowance), heavy cladding smeared or forgotten, equipment assigned to the wrong storey. These errors bias W low or distort the storey distribution, and they are invisible in a lumped total. The defense is per-storey, per-component traceability, every kilogram attributable, which is a property of a good takedown, and the reason W should be derived from one rather than estimated beside one.
Can StructLoads give me the seismic weight?
StructLoads builds the model W comes from: per-storey, per-element dead loads, superimposed dead load, partition allowances, and wall and cladding weights, traceable to components. Summing W per storey from that model is a re-read rather than a new calculation, and the same data locates storey mass for distribution and torsion checks. The seismic coefficients and analysis belong to your seismic code and tools; the mass side, which is most of the error surface in practice, is exactly what the takedown already knows.