Taking Off a Steel Package for Tender: the Tonnage and the Painted Area

Half past three, and the frame is still four PDFs

What came down with the enquiry is a general arrangement at 1:100, two elevations, one cross section, a sheet of general notes and a specification clause saying the protective treatment is to ISO 12944 for corrosivity category C3 at medium durability. There is no mark schedule. There is no bill. Fabrication drawings happen after award, which is the whole difficulty: the return is due at noon and it wants a fabricated tonnage and a quantity of shop primer, and both of them have to be produced from a drawing that was never drawn to be measured.

The reflex at this hour is to get the tonnage settled, because tonnage is what the rate hangs off, and then to reach for last job's paint allowance as so many litres per tonne. That second move is the one that quietly loses money, and it loses it in whichever direction the section mix has moved. Coating is bought by the square metre and steel is sold by the tonne, and the number connecting them is not a constant of steelwork — it is a property of this particular package, set by how heavy the average section in it happens to be. A frame of stocky columns and a frame of light bracing and rails can differ by a factor of three or four on painted area for the same weight, and nothing on the drawing announces which one you are looking at.

So there are two takeoffs to run before noon and only one of them is about weight. What follows is the order they go in: lengths off the arrangement, then the steel that is in the package and not on the plan, then the tally closed into a tonne, then the same member list turned sideways to measure its skin. How a per-metre mass is derived for a section whose designation does not carry one — an imported angle, a cold-formed rail, a cut-down beam — is a separate exercise with its own guide on this site, and this page assumes that column is already filled in.

One steel, two geometries

The reason the ratio moves is a piece of geometry rather than a piece of estimating. Mass per metre follows the cross-sectional area of the shape. Coating per metre follows the perimeter of the same outline. Grow a section and its area grows roughly as the square of the linear dimensions while its perimeter grows as the first power, so surface area per tonne falls as the steel gets heavier, and it falls steeply. That single relationship is behind every mismatch between a weight takeoff and a paint takeoff, and once it is in front of you the rest of the afternoon organises itself.

It bites hardest at the ends of the range, which is where real packages sit rather than in the comfortable middle: a shed frame is mostly light purlin, rail and bracing and its skin is enormous for its weight, while a short-span transfer structure is a few heavy columns and deep girders and its skin is small. Price both with one litres-per-tonne allowance and you are wrong twice, in opposite directions, by more than the margin on the paint line. The fix is not a better rate but a second quantity, measured from the member list the tonnage already needed.

Gridlines are not member lengths, and on a tender that is fine

Beams get measured grid to grid because that is what the drawing dimensions, and grid to grid is not the length that gets cut. A beam framing between two column centres is short by half the depth of the column at each end, and short again by the erection clearance the detailer leaves between the beam end and the column face. On a 6 m bay into 254 mm columns the two half-depths alone come to 254 mm, and a ten millimetre erection gap at each end takes it to 274 — better than four percent of the bay, measured and never fabricated. Take it anyway. A tender takeoff that runs a few percent over on member length is on the safe side of the only error that matters at this stage, and the alternative is deducting connection allowances from every beam on the plan with two hours left, for an accuracy nobody is going to check against a drawing that does not yet exist.

Columns are the ones to slow down on, because their length is the dimension the plan does not carry at all and the one people most often take from the wrong place. A column runs from the underside of its baseplate to the top of the section, which is neither floor to floor nor level to level: it starts at the top of the foundation or the top of the pocket, picks up the grout space beneath the plate, and finishes at whatever the roof or the next splice actually requires, which on a portal frame with a haunched eaves is well above the eaves level everyone quotes. Read those off the section and the elevations, not off the level schedule, and read every column rather than assuming the row is identical, because the one on the grid line where the ground falls away is longer than its neighbours and it is longer by more than you would guess.

Anything on a slope wants its true length rather than its plan projection, which sounds too obvious to write down until the third hour of a takeoff. A rafter across a 12 m span at a modest pitch is measurably longer than 12 m along its own line, and a diagonal in a braced bay is the hypotenuse of the bay rather than either dimension anybody quoted. Both get measured flat off a plan by tired people. Set the rule at the start of the sweep and hold to it, because it is not a judgement you want to be making member by member at five o'clock.

Keep the count as you go, in a column beside the lengths, because the piece count is a quantity in its own right and it is not recoverable from the tonnage afterwards. Fabrication hours, handling, painting setups, transport and the erection sequence are all driven by how many pieces there are rather than by what they weigh, and the tonnage divided by the count gives an average piece weight that tells you more about the character of the job than the tonnage does. A package at half a tonne a piece and a package at eighty kilograms a piece can carry the same total and are not the same contract.

1. Print or open the general arrangement at a fixed scale and mark up in colour as you go, so a member that has been measured is visibly different from one that has not.
2. Take beams grid to grid without deducting for column depth or connection clearance, and write on the sheet that you have done so, so the number is a decision rather than an oversight.
3. Measure every column individually from underside of baseplate to top of steel, off the section and elevations, and never off the floor-to-floor levels.
4. Take every sloping member — rafters, braces, valley and hip steel, stair strings — at true length from an elevation or by computing it, never by scaling the plan.
5. Keep a piece count in its own column next to the length column; you will need it for handling, painting setups and loads out, and it cannot be recovered later.
6. Group by mark or by section size rather than by grid line, so the per-metre mass column can be filled in once per group instead of once per member.
7. Close the loop before you total anything: every column has to appear on two elevations and one plan, and every beam has to land on something at both ends.

The steel that is in the package and not on the plan

The framing plan shows the frame, and the frame is not the package. What is missing is rarely a beam anyone would notice missing; it is a category — a whole class of member that lives on a different drawing series, in an architect's detail, in the cladding subcontract, or nowhere at all until the fabricator asks about it. Those categories are where a tonnage goes wrong by a margin no amount of careful measuring on the visible steel can recover, because the error is not in the measurement, it is in the scope.

Work the categories deliberately rather than hoping to notice them. Read the general notes, then the specification's scope clause, then the drawing register itself, and ask of each line whether steel appears on it. A register with an S-series and an SK-series where you have only opened the S-series is the commonest way half a secondary package goes unmeasured. Where the answer is genuinely unresolvable in the time available, that is a qualification rather than a guess.

The perimeter the drawing never dimensions

Painted area is the same member list read sideways: perimeter of the cross-section, multiplied by length, multiplied by count. The perimeter is the one term nobody has handed you. Some section tables publish it directly — the SCI publication P363, the Blue Book, carries surface area per metre and surface area per tonne alongside the section properties for the UK series — and where the table you are using does not, the outline can be assembled from three dimensions you already have.

For an I section, an H section or a channel, the painted perimeter comes out as four times the flange width, plus twice the overall depth, less twice the web thickness. It is worth walking the outline once to see why, because the result is more useful than it looks: the two flange outer faces contribute two widths, the two flange inner faces contribute two more widths less the web that interrupts them, and the four flange edge thicknesses cancel exactly against the two flange thicknesses each of the two web faces loses at its top and bottom. Flange thickness drops out of the answer altogether. Two sections of the same serial size, one nearly twice the weight of the other, have very nearly the same skin.

Put two real sections through it and the spread stops being theoretical. Take a light beam of the 305 by 102 serial size at about 25 kg/m: with a depth near 305 mm, a flange near 102 mm and a web near 6 mm, the outline gives roughly 1.005 m of perimeter per metre run. A tonne of it is 40 m of steel, so a tonne of it carries about 40 square metres of skin. Now take a heavy column of the 305 by 305 serial size at about 198 kg/m: depth near 340 mm, flange near 315 mm, web near 19 mm, and the outline gives about 1.90 m per metre run. A tonne of that is a little over five metres of steel and carries under 10 square metres. Same family, same rolling mill, four times the paint per tonne.

The outline arithmetic ignores root fillets and toe radii, and it errs generously rather than meanly: a concave fillet is a shorter path than the square corner it replaces, so the true wetted perimeter runs a percent or two below the outline figure. On a tender that is the right direction to be wrong in. Two further distinctions are worth holding onto. Section factor tables written for fire protection quote a boxed perimeter as well as a profile one, and only the profile figure describes what a spray gun has to cover — take the box value into a paint quantity and you underbuy on every open section in the package. And connection material moves the two takeoffs in opposite directions: a gusset or a cleat is two broad faces of area for very little weight, so plate inflates painted area proportionally more than it inflates tonnage.

Read the coating clause before you price a litre

The specification decides the quantity long before any calculator does, and the clause is usually four facts in one sentence. ISO 12944-2 classifies the environment — C1 through C5 for atmospheric exposure, CX for extreme, Im1 to Im4 where the steel is immersed or buried — and ISO 12944-1 sets the durability bands the specifier then chooses from, low up to seven years, medium from seven to fifteen, high from fifteen to twenty-five and very high beyond that. Those two together point at a paint system in the ISO 12944-5 tables, and it is the system, not the environment, that tells you how many coats and what total dry film thickness the steel is going to receive. Reading C3 and stopping there leaves the most important half of the clause unread.

Surface preparation is the next line and it is a cost line, not a quality line, because it decides both the shop hours and how thirsty the steel will be. ISO 8501-1 defines the visual preparation grades — Sa 2.5 is the one most systems ask for — and the American equivalents are SSPC-SP 6 with NACE No. 3 for commercial blast and SSPC-SP 10 with NACE No. 2 for near-white. ISO 8501-3 covers the preparation of welds, edges and surface imperfections, which sounds like a detail and is not: it is the reason the specification also asks for stripe coats. A blast profile to ISO 8503 also consumes paint that never shows up as film thickness, because the first material applied fills the anchor pattern between the peaks before any measurable thickness accumulates over them.

Then subtract what is not being painted, because on most packages that is a real area rather than a rounding. Faying surfaces in slip-critical connections are either left bare or given a coating qualified for a slip class under the RCSC Specification for Structural Joints Using High-Strength Bolts, and either way they are masked in the shop. Steel to be encased in concrete is not coated. Steel receiving applied fire protection takes a compatible primer and nothing else from the paint scope, which is a separate quantity belonging to the fireproofing line. Site weld zones are held back and made good afterwards. AISC 303 on one side of the Atlantic, and BS EN 1090-2 with the BCSA National Structural Steelwork Specification on the other, are where those exclusions are conventionally written down; what matters this afternoon is that painted area is measured area less masked area, and on a heavily connected frame the difference is not small.

Two hours left: measure this, allow for that

Not everything on the drawing deserves the same care, and deciding that consciously is the difference between a takeoff that finishes and one that stops halfway with the columns done beautifully and the roof untouched. Columns and primary beams are most of the weight and are individually few, so measure them one at a time. Repetitive secondary is many pieces of small weight, so measure one bay properly and multiply, then check the multiplier against the bay count on the plan. Fittings, cleats, stiffeners and connection plate are neither measurable at this stage nor negligible, so they take a stated allowance — and the word stated is doing the work.

An allowance is only defensible when it is visible. A percentage buried in a rate is a guess that has been hidden; the same percentage written on the return as an allowance for connection material, with the basis named, is a commercial position that can be tested and adjusted at award. That distinction survives into the contract, because a client who accepts a stated allowance has accepted the assumption behind it, and a client who was never told is entitled to think the rate covered everything.

1. Sort the package by weight share before measuring anything, and give the columns and primary beams the time; they are the smallest count and the largest tonnage.
2. Measure one typical bay of secondary steel in full, then multiply by the bay count from the plan and verify the count independently off the grid.
3. Give fittings, cleats and connection plate an explicit stated allowance rather than a percentage folded into the rate.
4. Cross-check the total against a kilograms-per-square-metre ratio built from your own recent jobs of the same structural form, and treat a large disagreement as a missing category rather than a measuring error.
5. Take the painted area from the same member groups, deduct the masked and encased surfaces, and add the plate area separately because it is disproportionate to its weight.
6. Convert to litres per coat at a practical spreading rate, then round up to whole containers in the pack sizes the product actually ships in — a part tin is bought whole.
7. Write the qualifications page while the numbers are still fresh, not after the covering letter.

The assumptions go in the file, not in your head

By the time the return goes out you have made two dozen decisions nobody else can see: which drawing series you measured and which you never opened, whether beams were taken grid to grid, what the connection allowance rested on, which surfaces you assumed were masked, what loss factor turned the theoretical spreading rate into the practical one, and what you did about the stair steel that might be someone else's. Every one of them is defensible on the day it is made and indefensible six weeks later when you cannot remember making it.

Which of them belong on the return and which belong only in your own file is a commercial judgement, but the split is worth making deliberately. Anything that changes the scope goes on the return as a qualification, in plain language, because it protects the price. Anything that is a method rather than a scope — the grid-to-grid convention, the loss factor, the sample bay you extrapolated from — stays in the file, because it is how you rebuild the number when the addendum lands on Monday morning and moves three gridlines. How steelwork is measured for payment is set by the contract rather than by anybody's habit: RICS New Rules of Measurement 2 sets out the British measured-work convention for structural metalwork, and AISC 303 is the customary reference for scope and shop paint in American practice. Establishing which one the enquiry is written against, before the return leaves, decides whether your tonnage and the tonnage on the other side of the table are describing the same steel.