Follow the force, not the drawing
A connection is not a picture on a sheet; it is a relay. Something upstream is pushing, pulling or twisting, and that force has to leave one piece of steel, travel through everything the detailer put in its way, and arrive in the next piece without stopping. Every failure a steel joint has ever had is the same failure: the load reached a place it could not cross.
So walk it. Pick the biggest number at the joint - the beam end reaction, the brace axial, the flange couple from a moment frame - and follow it hand over hand: out of the member, into the angle or plate, through the bolt or the weld, into the plies that bear against it, across the plate body, through the second set of fasteners, into the supporting flange, and finally into whatever the column stands on.
Each heading below is one crossing. Detailers, fabricators and erectors each own a different stretch of the same path, which is why a joint can be drawn correctly, made correctly and still fail - the handoff between two competent parties went unchecked. Read a connection as a chain of crossings and the checks arrange themselves.
The first handoff: member face into connecting element
Force leaves the member through whatever the connection actually touches. A shear tab welded to a column flange takes the beam reaction out of the beam web only - the flanges contribute nothing there, which is why a coped beam with its top flange cut away still works and a beam with a damaged web does not. Single-angle and double-angle connections take it out of the web as well; a seated connection takes it out of the bottom flange in bearing.
Fit-up governs this first crossing more than any calculation. A gap behind a shear tab, an angle sprung open by a bent leg, a bolt group forced into position with a bull pin instead of a drift pin - each one means the plies are not sharing the way the detail assumed. Erectors know the sound of it: bolts that will not run down by hand, plates that spring when the wrench comes off.
Gauge lines and edge distances are the physical record of that assumption. When a fitter shifts a hole line to clear a stiffener, the load path moves with it, and eccentricity the design never carried appears in the plate. Fabrication and erection tolerances - straightness, plumbness, mill variation, the permitted setting of an anchor rod - are published in AISC 303, Code of Standard Practice for Steel Buildings and Bridges, and in EN 1090-2, Execution of steel structures and aluminium structures. Field fit that lands outside them is an RFI, not a judgement call.
Through the shank: bolts, plies and the hole they sit in
Once force is in the connecting element, the shank has to carry it. A bolt in a double-angle connection sees two shear planes; the same bolt in a single plate sees one, and it moves twice the stress for the same reaction. Whether the threads fall in the shear plane changes the strength again - the X and N designations on a bolt schedule are not decoration, and a longer grip with a shorter thread run is what puts the shank where the drawing says it is.
Bearing and tearout govern the plies, not the bolt. Steel in front of a bolt hole either crushes or tears out to the free edge, and which one happens depends on ply thickness, hole type, spacing and how much material sits between the hole and the plate edge. A short edge distance is the single most common thing that turns a nominally adequate bolt group into a plate that unzips - and it is nearly always created on site, by a fitter trimming a plate to clear something.
Hole type is part of the path. Standard, oversized, short-slotted and long-slotted holes each change the bearing area and the amount of movement before load is picked up, and slotted holes have an orientation the erector must respect. Bolt grades, assemblies and their marking are covered by ASTM F3125, Standard Specification for High Strength Structural Bolts and Assemblies; installation, joint type and faying-surface condition by the RCSC Specification for Structural Joints Using High-Strength Bolts. In a slip-critical joint the force crosses by friction before it ever touches the shank, so paint overspray, faying-surface condition and a missing hardened washer are structural defects, not housekeeping.
Through fused metal: what the throat actually carries
Welded crossings carry force through fused metal whose cross-section nobody can see. For a fillet, the throat does the work, not the leg the inspector measures - which is why an undersized leg costs proportionally more capacity than it looks like it should, and why a convex bead does not make up for a short one. Length matters equally: a return that was never wrapped, a crater at the end of a run, or a stop-start that left slag are all missing throat.
Transverse fillets are stronger than longitudinal ones, and several codes permit taking credit for the direction of loading. That credit belongs to the engineer, not the welder, and a field weld sized on the assumption of transverse loading in a joint actually loaded along its length has no reserve left. Where a bracket carries an eccentric load, the weld group behaves like a bolt group in the same geometry - the far end of the run does most of the work.
Procedure controls whether the metal is sound at all. Preheat, interpass temperature, filler classification and prequalified joint geometries live in AWS D1.1, Structural Welding Code - Steel; in Canada in CSA W59, Welded Steel Construction; in Australia and New Zealand in AS/NZS 1554.1, Structural steel welding. Which of those governs is set by the jurisdiction on the permit, not by the fabricator's habit. Weld access hole geometry, backing bar treatment and run-off tab removal are the details most often skipped in the field, and each of them removes throat or leaves a notch exactly where the stress concentrates.
Argument about a fillet weld happens with a tape and a gauge in hand at this exact crossing, so the capacity the leg and length are worth belongs on the same page rather than back in the office.
The nominal leg size of the equal-leg fillet weld.
The electrode classification, which sets the weld metal's classification strength FEXX.
The total effective length of the weld seam.
Nominal weld shear capacity
71.4 kips
This is the NOMINAL weld strength Rn from AISC 360 §J2.4. It is not a usable design capacity: apply φ = 0.75 for LRFD, or divide by Ω = 2.00 for ASD, and check the connected base metal separately — the base metal, not the weld, governs many joints.
- Effective throat thickness
- 0.22 in
- Capacity per unit length of weld
- 9.21 kips/in
They open the calculator with your figures already in it
Fillet Weld Shear Capacity Calculator: 71.39 kips — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 71.4 kips — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- 0.60 FEXX is the base case, for a weld loaded parallel to its own axis. AISC J2.4 permits a directional increase for welds loaded at an angle to their length — up to 1.5 times for a fillet loaded transverse to it — so this reads conservative for an end weld and exact for a side weld, and a group containing both has to be handled by the code's combination rule rather than by adding the two lengths together.
- Length is taken as fully effective however long it runs. For end-loaded fillet welds longer than about 100 times the leg size AISC applies a reduction factor, because stress along a long lap is not uniform and the far end does less work than the near end; below four times the leg size the effective length is cut back instead. A long lap weld does not deliver capacity in proportion to its length.
- The leg size is accepted without any of the rules that bound it. AISC Table J2.4 sets a minimum leg by the thickness of the thicker part joined, so the weld cools slowly enough not to crack, and a weld run along a plate edge is capped at the plate thickness — less 1/16 in (1.6 mm) on material 1/4 in (6 mm) and over — so the edge is not melted away. A capacity computed from a leg outside those bounds describes a weld that will not be accepted.
- The result is a group total, and it assumes the load passes through the group's centroid. Where it does not — a bracket, a shear tab, a bracing gusset — the weld carries torsion on top of direct shear, the far corner reaches its limit while the rest is barely stressed, and the capacity from an instantaneous-centre analysis falls well below this length × strength product.
Through the plate body: net section, block shear and the Whitmore width
Having crossed the fasteners, the load now has to travel through the plate itself, and a plate has three separate ways to refuse. Net section rupture takes the line of holes across the width. Block shear tears a chunk out - shear down one leg of the bolt pattern, tension across the other - and it punishes tight edge distances and short bolt patterns hardest. In compression, a gusset buckles over the effective width spreading from the connected length, the conventional thirty-degree spread that every brace connection is checked on.
Coped beams add a fourth. Cut the top flange away and the remaining web behaves like a plate cantilevering off the connection, prone to local buckling and to cracking from the re-entrant corner if the cope was flame-cut without a radius and left unground. A cope extended on site to clear a duct has changed the strength of that crossing, and nobody upstream knows it happened.
Look at the plate the way the force does. Draw the failure line by hand across the bolt pattern and ask what area remains on it; that thirty-second sketch on a scrap of plate catches more real problems than re-reading the connection schedule. Where the remaining area looks thin, the answer is usually another bolt row or a longer plate, not a bigger bolt.
The turn: when the path has to carry moment
Turn the path through ninety degrees and it becomes a couple. A moment connection does not carry moment as moment - it splits it into a tension force at one flange and a compression force at the other, and those are the numbers that actually cross the joint. In a bolted end-plate or flange-plate detail, the tension side is the whole problem: bolts in tension, a plate bending between them, and prying action that adds force the bolts never see on the drawing.
Bolt count on the tension flange sets the geometry of everything else - plate thickness, bolt gauge, whether stiffeners are needed, whether the flange plate will fit between the column flanges at all. Settling that number late means a fabricator re-detailing after material is already cut.
Seismic frames tighten every one of these crossings. Prequalified geometries, demand-critical welds, protected zones where nothing may be attached, and backing and tab removal requirements come from AISC 341, Seismic Provisions for Structural Steel Buildings, AISC 358, Prequalified Connections for Special and Intermediate Steel Moment Frames for Seismic Applications, and AWS D1.8, Structural Welding Code - Seismic Supplement. Drilling a hanger through a protected zone severs a load path that was qualified by full-scale testing.
The moment has just become a flange force landing on a bolt group, and the bolt count that force demands is the one number that has to be settled before the fabricator cuts plate, so it is worth running here.
The factored moment the connection must resist.
The distance between the extreme tension and compression bolt rows (or bolt row and compression flange).
The rated shear or tension capacity of a single bolt for this connection.
Bolts needed
8 bolts
A simplified equivalent-force method — a full moment connection design also checks bolt group eccentricity, prying action, and plate bending per AISC 360 Chapter J and should be verified by a structural engineer.
- Equivalent force at bolt group
- 82.98 kips
They open the calculator with your figures already in it
Steel Moment Connection Bolt Count Calculator: 8 bolts — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Estimated cost — your price
This site holds no price list for this material — local prices vary too much to publish honestly. Enter your supplier's price and the result is costed with it.
What this calculation does not cover
- The count assumes the lever arm entered survives the layout, and it has no idea whether the bolts fit. Spacing of about three diameters and the edge distances in AISC Table J3.4 have to be found within the end-plate width and either side of the beam flange, so a count of twelve on a 400 mm (16 in) arm can force the rows apart — changing the arm — or add a row inside the couple that does not carry the same force. Where the number will not fit the geometry, the connection changes, not the bolt count.
- A moment connection carries the beam's vertical reaction as well as its moment, and this divides the moment alone. Shear normally goes through a separate web plate and its own bolt group; where the same bolts take both, tension and shear interact and each bolt's available capacity drops below its tabulated single-action value, so a group sized here for the couple can be short once the reaction is added.
- Everything counted is on the beam side. The couple delivers its force into the column, whose flange bends, whose web yields, crumples or buckles opposite the compression flange, and whose panel zone shears — the checks that decide whether continuity plates, a doubler plate or a heavier column are needed. Those are what usually add the cost to a moment connection, and none of them changes the number of bolts returned here.
Into the supporting member: flange, web and panel zone
Arriving is not the same as crossing. The supporting member has to accept the force locally: a column flange bends between its own web and the bolt lines, a girder web can crush or crumple under a concentrated force, and a moment-frame panel zone shears as a small independent element inside the column. Continuity plates and doubler plates exist for exactly these crossings, and they are the plates most often value-engineered out and most often missed in the field.
Web crippling and local yielding show up wherever a beam sits on a bearing plate or another beam frames in without a stiffener. On existing structures, look for the tell-tale wrinkle in a web under an old crane rail or a stacked mezzanine beam.
Skewed and sloping members complicate the arrival. Force arriving off-axis puts a component into the supporting member's weak direction, and bent plates, sloped shear tabs and skewed end plates all need the geometry checked in three dimensions rather than on the single elevation the shop drawing shows. Where a brace lands at a beam-to-column joint, three load paths meet in one gusset and the resultant has to be resolved before anyone argues about weld size.
Out of the column and into the concrete
Down the column the same force keeps travelling, and the last steel-to-concrete crossing has its own vocabulary. Axial load spreads through the base plate into bearing on the grout and concrete below, and the plate has to be thick enough to distribute it without dishing at the edges. Grout that was never fully packed under the plate - a common outcome when the shim stacks are tall and the mix is stiff - leaves the load riding on the shims alone.
Horizontal force is a separate question. Friction under the plate, a shear lug, or bearing of the anchor rods against the plate holes each carry it differently, and anchor holes in base plates are deliberately much larger than the rods, so plate washers must be welded to the plate before shear is assumed to cross that way.
Tension in an anchor rod pulls a cone of concrete with it. Breakout, pullout, side-face blowout and edge distance are governed by ACI 318, Building Code Requirements for Structural Concrete, in jurisdictions that adopt it, and by EN 1992-4, Design of concrete structures - Design of fastenings for use in concrete, elsewhere; the concrete design code named on the permit decides which. Anchor rods set out of position and then bent back into the plate holes with a torch have lost both their embedment geometry and their material properties.
A column base, exploded
- Column shaft — arrives priced by weight per foot of section, and its footprint is what the plate below has to spread across the concrete Steel I-Beam Weight-per-Length Calculator
- Base plate — sized so the pressure it hands the concrete stays inside what that concrete may take, and thick enough not to dish at its own edges Steel Column Baseplate Concrete Bearing Pressure Calculator
- Grout bed — has to be packed solid rather than bridging the shim stacks, and its pocket volume is a small pour somebody still has to order Baseplate Anchor Bolt Count & Grout Pocket Volume Calculator
- Anchor rods — carry tension by dragging a cone of concrete with them, so embedment depth and edge distance matter more than rod diameter Concrete Anchor Bolt Breakout Capacity Calculator (ACI 318)
- Concrete pedestal — where the force finally stops, and the one item on this list poured rather than delivered, ordered by volume weeks ahead of any steel Concrete Calculator
This crossing reduces to one division - the load over the plate area it is spread across - checked against what the concrete underneath is permitted to take.
The total factored axial (compressive) load the column transfers to the baseplate.
The baseplate's footprint area in contact with the concrete.
The specified compressive strength of the supporting concrete.
The ratio of the supporting concrete area (A2) to the loaded baseplate area (A1).
Actual bearing pressure
453 psi
The bearing pressure under this plate is below the design bearing strength, with phi = 0.65 applied to the ACI nominal shown with it — ACI 318 gives it. The load compared here has to be a factored one, not a service load. Being under one limit is not a design. Nothing else is checked here — not the other limit states, not the connections, not the member the load arrives from.
- Nominal bearing strength Bn/A1
- 6,164.1 psi
- Design bearing strength phi*Bn/A1 (phi = 0.65)
- 4,006.67 psi
They open the calculator with your figures already in it
Steel Column Baseplate Concrete Bearing Pressure Calculator: 453 psi — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 453 psi — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- The pressure reported is the axial load spread evenly over the whole plate, which is only true for a concentric load with no base moment. Once a column moment gives the resultant an eccentricity greater than one sixth of the plate length, the plate bears on part of its area only and the peak pressure at the compression edge rises well above the uniform figure shown here.
- The A2/A1 ratio is taken as typed and is never checked against real geometry. ACI permits the square-root term only where A2 is geometrically similar to and concentric with the loaded area, lies wholly within the support, and can be reached on a load spread no flatter than two horizontal to one vertical, so a shallow pier or a column set near a pier edge has a smaller usable A2 than its plan dimensions suggest.
- The comparison is made in LRFD terms: a factored load against a design strength with phi = 0.65. An allowable stress design check instead compares an unfactored service load against the nominal strength divided by AISC 360 J8's omega of 2.31, which this page does not calculate, so entering a service load against the phi-reduced strength shown here overstates the margin by roughly the load factor.
- The stress is checked against the concrete only, but where the plate is set on grout or shims it is the material at the plate face that governs first. Grout weaker than f'c, an incompletely filled pocket, or setting shims carrying load before the grout cures all reduce the contact area that the uniform pressure assumes.
- Bearing is a local crushing check on the concrete directly beneath the plate. It says nothing about the soil pressure under the footing, punching or one-way shear through the pier or footing, or the bursting reinforcement needed where the concentrated load spreads out, and a pier can satisfy this check while failing any of those.
Proving the crossing: pretension, procedure and inspection
Nothing above is real until somebody proves it. Snug-tight, pretensioned and slip-critical are three different joints, and only the drawing says which one sits in front of you. Pretension can be installed by turn-of-nut, calibrated wrench, twist-off-type tension-control assemblies or direct-tension indicators, each with its own pre-installation verification and its own field evidence - a matchmark, a snapped spline, a squashed washer bump.
Inspection of welds runs from visual, which catches undercut, undersize, overlap and craters, up through magnetic particle and ultrasonic testing on complete-joint-penetration welds. Acceptance criteria and the extent of testing are set by the welding code in force and by the contract documents together; a fabricator's internal standard overrides neither.
Bolts have field rules worth knowing cold: high-strength assemblies are supplied and stored as a lot with their lubricant intact, galvanised A490 assemblies are not permitted, and reuse is restricted in ways that depend on grade and coating. Anything heated, welded on, or driven with a hammer has left the specification. Documentation matters as much as the act - an inspection record that cannot say which joints were verified is a record a future engineer will treat as no record at all.
Calibrated-wrench work needs a target set before the first bolt is run down, and the nut factor is the term that makes two identical diameters read differently on the same gun.
The empirical nut factor relating torque to pretension for the hardware's surface finish and lubrication.
The nominal bolt diameter.
The required minimum bolt pretension from the RCSC table for the bolt's grade and size.
Estimated installation torque
390 lbf·ft
Calculated torque is approximate only — friction (and thus K) varies with surface finish and lubrication by up to ±30%. RCSC does not permit torque-wrench-only control; pretension must be verified by turn-of-nut, calibrated wrench, twist-off bolts, or direct tension indicators.
They open the calculator with your figures already in it
High-Strength Structural Bolt Torque Calculator (ASTM F3125): 393 lbf·ft — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 390 lbf·ft — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- K × D × P assumes every bit of torque goes into stretching the bolt. Anything that resists turning without producing tension comes off the top first: a direct tension indicator's protrusions being flattened, a prevailing-torque or locking nut, a galling thread, grit or paint in the threads, or a washer that turns with the nut instead of staying put. None of that has an input here, and all of it lets the wrench reach the target while the bolt is still loose.
- The figure is for one continuous pull on a bolt being tightened from snug — it is not a check torque. Breaking an already-tightened bolt loose reads static friction, which is higher than the running friction K stands for, so a correctly tensioned bolt commonly moves only above this number while a badly under-tensioned one can turn at close to it. Applying this torque to an installed bolt tells you very little about the tension in it.
Where the path gets cut after erection
Load paths get severed after the structure is standing, usually by trades who have no reason to know they are doing it. Holes burned through a beam web for conduit, a hanger rod welded to a bottom flange, a cope extended for ductwork, a stiffener ground back for pipe clearance - each removes area from a crossing that was checked once and never re-checked.
Temporary work counts too. Erection lugs, tag-welded seats and lifting eyes left in place are notches on a member that may see fatigue or cyclic seismic demand, and the specification usually requires them removed and the surface ground flush. Field welding onto a member without knowing its grade and carbon equivalent risks cracking that will not show for months.
Shims and fills are the quiet one. Adding fillers between plies changes bolt behaviour, and beyond a certain thickness the connection needs extra fasteners or the fill developed by welding - the governing steel design code says which. Stacking finger shims to make a bad fit go away is a repair that hides its own evidence.
A hole burned through a web for conduit leaves behind exactly the net section a shop-cut opening does, so the shear still available above and below it can be worked out on the spot rather than argued about.
The beam steel's specified minimum yield strength.
The overall depth of the castellated/cellular beam at the opening.
The height of the web opening (hexagonal or circular) at the section checked.
The beam web's thickness.
Net-section shear capacity
76.8 kips
This is a basic net-section shear screening check ONLY. Vierendeel bending and web-post buckling at the opening — often the governing failure modes for castellated/cellular beams — are NOT evaluated here and require specialist software (AISC Design Guide 31, SCI P100, or FE analysis).
- Net web area at opening
- 2.56 in²
They open the calculator with your figures already in it
Castellated Beam Web Opening Net-Section Shear Calculator: 76.78 kips — shown in imperial, US market. The link sets both, so the result they see is the one on your screen.
Add the equipment this sizes
This result is a specification — 76.8 kips — not a quantity. Put the thing it sizes into your project: how many, what you call it, and your supplier’s price.
What this calculation does not cover
- 0.6·Fy·Aw is a nominal yield capacity with nothing applied to it. There is no resistance factor, no safety factor and no shear buckling coefficient, so setting this number against a factored load is unconservative twice over — apply φv or divide by Ωv, and check the slenderness of the web at the tee, because a deep thin web beside a large opening buckles in shear before it ever yields.
- Deflection usually decides these beams, and it is not the solid-web calculation. Shear deformation across the web posts adds materially to what a cellular or castellated beam moves under load, so a section that clears every strength check at the opening can still be too lively — and the long spans these beams get chosen for are exactly where that governs.
Reading a joint cold, in five minutes
Standing at a joint with nothing but a tape and a torch light, work in the same order the force does. Where does the load come from, and what part of the member is it actually leaving through? What is the first element it touches, and does that element sit flat against its neighbour with no gap the eye can catch?
Count fasteners and measure the two distances that matter - hole to hole, and hole to edge - then look for the shortest remaining material anywhere in the pattern. On welds, check length and returns before arguing about leg size; a full-length undersized fillet is a smaller problem than a correctly sized fillet that stops short.
Finish by looking at where the force leaves. Is there a stiffener where a concentrated force lands, is the supporting web unbuckled, does the base plate sit in solid grout with its washers welded? A joint that passes that walk-through will nearly always pass the calculation, and a joint that fails it will fail somewhere the calculation was never asked to look.
Walk the joint before you sign it off
Five checks that follow the same order the force travels, from the member face to the concrete. Take them with a tape, a fillet gauge and the connection schedule, and record what you measured rather than what the drawing said.
- Bolt assemblies, kept as a lot — Grade, length, washers and lubricant condition together; segregated by lot and protected from weather, with the pre-installation verification done on that lot.
- Hole-to-edge and hole-to-hole, measured on the steel — Not scaled off the drawing - any plate trimmed on site has a new edge distance and a new tearout check.
- Fillet length, returns and throat — Check the run end to end before arguing about leg size; craters, unwrapped returns and stop-starts are lost capacity.
- Continuity and doubler plates present — Confirm against the issued-for-construction detail wherever a concentrated force or a panel zone is involved; these are the first plates dropped in a cost exercise.
- Base plate grout and welded plate washers — Grout fully packed rather than bridging the shim stacks, washers welded before any shear is assumed to cross through the oversized anchor holes.
Opens the calculators above on one screen with the dimensions from this article already filled in. Quantities only — this site publishes no price list, because local prices vary too much to publish honestly.