PEB optimisation explained: how tapered frames get lighter

Why most PEB frames carry steel they don’t need

Ask any PEB designer how a main frame gets sized and you will hear the same story. You pick a starting section from experience, run the analysis in STAAD.Pro, look at the utilisation ratios, bump the members that fail, trim the ones that look lazy, and run it again. Five, eight, sometimes ten rounds later every member passes, the deadline is close, and the frame goes to detailing.

Nothing about that process is wrong. It is how most of us were trained. But it has one built-in weakness: it stops at the first section that passes, not the lightest section that passes. Those are almost never the same section.

A tapered built-up I-section has at least five variables you can move: flange width, flange thickness, web depth at each end (which sets the taper), and web thickness. Top and bottom flanges can differ. Each variable has a practical range and a set of standard plate sizes. Multiply those together across a column and two rafter segments and you have thousands of valid combinations for one frame. By hand, we look at maybe ten or twenty of them.

So the frame that ships is a good frame. It is just not the lightest good frame, and on a building with twenty identical frames that gap is paid for twenty times over.

What PEB optimisation actually means

In the plainest terms, PEB optimisation (or PEB optimization, depending on which side of the Atlantic your code book came from) is a structured search. For each tapered member we ask: of all the sections that satisfy every check in IS 800:2007 or AISC 360, which one weighs the least?

Three things have to be true for that search to be worth anything:

• The checks must be the real checks. Axial, bending about both axes, shear, combined interaction, lateral-torsional buckling and the width-to-thickness (slenderness) limits of the section, clause by clause, for every load combination in the model.
• The candidates must be buildable. Plate thicknesses from the sizes your fabricator stocks, flange widths in sensible increments, web depths and tapers your welding line can handle.
• The forces must be honest. A lighter frame is a more flexible frame, so the member forces change as sections change. The optimisation has to work from the analysis of the frame as it is being designed, not from forces frozen at the start.

Get those three right and what comes out is an optimised frame: one where most members run at a utilisation close to, but not above, the code limit, and none carries steel just because it was quicker to round up.

Where the steel hides in a tapered frame

After optimising a fair number of frames, the same few places keep giving steel back. If you are reviewing a frame by eye, these are worth a second look.

Flanges that are thicker than they need to be

Flange area does most of the work in bending, so it is tempting to add thickness when a member fails. Often a slightly wider, thinner flange carries the same moment with less steel, as long as the outstand stays inside the compact or semi-compact limit. That trade-off between width and thickness is exactly what is tedious to explore by hand.

Webs that are too thick for their depth

Web plates are a big share of the weight of a deep tapered member. Many hand-sized frames use one web thickness everywhere for convenience. A deeper, thinner web can carry more moment for less steel, up to the point where shear buckling or the web slenderness limit starts to govern. Finding that point member by member is where a search earns its keep.

Taper that doesn’t follow the moment

The whole point of a tapered section is to match depth to the bending moment diagram. In practice, tapers often get fixed early from a rule of thumb and never revisited. Letting the depth at each end move independently, inside fabrication limits, is usually worth a few percent on its own.

Rafters sized for the worst segment

A rafter split into segments at the splices can use different plates in each segment. When one plate size is carried through the whole rafter, the lightly loaded middle segments pay for the heavily loaded haunch.

Manual iteration vs a systematic search

It helps to be clear about what changes when the search is done by software rather than by hand.

The last row matters most. Optimisation is not about taking the engineer out of the loop; it is about moving the engineer’s time from the boring part (trying section after section) to the valuable part (checking that the answer makes sense, detailing the connections, signing off the drawings).

What optimisation leaves to the engineer

Honesty about scope is important, so let’s be explicit. A section optimiser like ours works on the tapered built-up I-sections of the main frame. It does not:

• design the knee, ridge and splice connections, base plates or foundations;
• check deflections and drift, which you still review in STAAD.Pro after the sections are written back;
• size hot-rolled sections, or cold-formed purlins and girts (those follow their own rules, such as AISI S100 or IS 801);
• replace a senior engineer’s review of the final design.

A lighter frame is also a more flexible frame, so serviceability deserves real attention after optimisation. In our experience the strength-governed members get lighter and the deflection-governed ones stay where they are, which is exactly what should happen.

How we run an optimisation, step by step

Here is the workflow we follow on live projects with the PEB Optimization Tool, which runs alongside STAAD.Pro:

1. Model the frame as usual in STAAD.Pro CONNECT Edition (V22 Update 10 or later, including STAAD 2023, 2024 and 2025): geometry, supports, loads, load combinations and the steel design parameters for IS 800 or AISC 360.
2. Start from any reasonable section. It does not need to pass; it only needs to be in the right family so the analysis is sensible.
3. Run the optimiser. It reads the model through OpenSTAAD, searches flange width, web depth and taper together for every column and rafter, and checks each candidate against the code, utilisation and slenderness both.
4. Review the result member by member: the chosen plates, the governing check and the utilisation.
5. Write it back to the STAAD model in one click and run your final analysis and design as you always would. Check deflections and drift, then move on to connections.

If you want to see the idea before installing anything, the in-browser optimizer bench on our home page sweeps candidate sections on a sample portal frame. The tutorials walk through a full project in STAAD.Pro.

Is it worth it on your projects?

A quick way to decide: take the frame weight of your last typical building, multiply by 3–10%, multiply by the steel rate, and compare that with the time your team spent iterating on it. For most PEB manufacturers and design offices the steel alone settles the question; the hours saved are a bonus.

If you design to IS 800 and AISC side by side, our comparison of IS 800:2007 and AISC 360 for PEB frames covers where the two codes push the sections in different directions. And if your frames are a mix of clear-span and multi-span buildings, portal frame vs A-type frame looks at what makes each of them economical.