Four Degrees on the Metal, Seventy in the Specification
Half past seven on a February morning, north-east elevation, second lift. The frames landed last week and have been sitting on the deck under a tarpaulin. The setting-out sheet in the folder says the stack joint is six millimetres and gives no other information, which is the same as giving none at all, because nobody wrote down what temperature six millimetres was meant to be six millimetres at. The infrared thermometer in the bag reads four degrees on the extrusion. The specification the frame was designed against assumes that same extrusion will reach seventy in full sun in August.
That difference is the whole job. A stack joint is not a dimension you transfer from a drawing; it is a position in a travel. The travel is fixed — it is decided by how long the member is, what alloy it is, and how wide a temperature band the designer says the metal will swing through — and the only thing today's reading decides is where in that travel you start. Set the joint at the wrong place in the travel and you have not made an error of a millimetre or two. You have made an error of an entire season.
It fails quietly and it fails late. A joint set on a cold morning to a figure meant for a mild afternoon closes up over the spring, reaches metal on metal somewhere in July, and from that point every further degree of expansion is being resisted by two extrusion ends bearing on each other instead of sliding past. The load has to go somewhere: into the splice sleeve, into the bracket, into the anchor, into the glass. Nothing about that shows up on a snagging walk in March.
Shoot the Metal, Not the Forecast
Air temperature is the wrong number and it is the number everyone reaches for, because it is written on a phone. An extrusion responds to what it is, not to what the shade air around it is. A dark anodised or dark powder-coated mullion on a south or west elevation in spring sunshine will sit fifteen to twenty degrees above the air beside it; a mill-finish or light-coated one runs closer to ambient. Go the other way and the same physics applies in reverse: on a clear night a north elevation radiates to a cold sky and the metal drops below the air temperature the weather station recorded, which is why the coldest design case for a facade is rarely the coldest air on record.
So the reading has to be taken on the elevation you are working, on the member you are about to splice, at the moment you set the joint. Shoot it with an infrared thermometer aimed at the extrusion rather than at the glass, and be aware that a bright anodised finish is a poor emitter and will read low — a strip of matt tape on a spare offcut, or a contact probe taped to the member, settles that argument. Take it again after lunch if the sun has come round the corner, and again on the next elevation, because two faces of the same building at the same hour are not the same temperature. Frames stacked in sun on the deck are not the temperature of frames hanging on a shaded wall either, and a bay set straight off the stillage carries the stillage's temperature for a while yet.
The design range is a different kind of number and it does not come from site. It comes from the facade designer, built up from climatic design data and a solar allowance chosen for the finish, and it is the input that sets the total travel every joint on the elevation has to accommodate. If the range in the specification looks too narrow for a dark finish on a sun-facing elevation, that is a design question and it belongs with the designer before the first joint is set, not with the installer holding a feeler gauge afterwards.
| Temperature | What it actually describes | Where it comes from |
|---|---|---|
| Shade air temperature today | What the forecast and the site thermometer agree on | Useful for deciding whether a sealant will cure and whether the gasket is workable, and for very little else on this page |
| Metal temperature at the moment of setting | What the extrusion in your hands is, on the elevation and the finish you are working | An infrared or contact reading on the member being spliced, taken bay by bay and written down |
| Hottest design metal temperature | The top of the band the joint must survive, with solar gain and the finish already inside it | The facade designer, from climatic design information such as the ASHRAE Handbook — Fundamentals or BS EN 1991-1-5 Eurocode 1: Actions on Structures — Thermal Actions, plus a solar allowance |
| Coldest design metal temperature | The bottom of the same band, including radiant cooling below air temperature on a clear night | The same source. A shaded elevation under a clear sky, not the record low air temperature, is usually the governing case |
Setting the Stack Joint
The stack joint is the splice between two lengths of mullion, one floor above the other, and it is deliberately the loosest connection in the wall. The extrusions are joined by an internal sleeve that locates them and lets them slide, the joint is covered rather than closed, and the entire vertical growth of a stick or unitised system is supposed to be taken up there and nowhere else. A wall with anchors at every floor and a rigid splice between them is a wall with no route for its own expansion.
The arithmetic is the plain linear one — change in length equals the coefficient of expansion times the length times the change in temperature — and the useful trick is to run it twice. Take a 3.9 m aluminium mullion between splices, a design band from minus twenty to seventy, and a reading of four degrees on the metal this morning. Across the whole band the member travels about 8.3 mm, so that is the total the joint has to hold. From four degrees up to seventy there are sixty-six degrees still to come, which is about 6.1 mm of expansion the joint has not yet absorbed. Add the residual clearance that must survive at the hottest condition — call it 1.5 mm, and the system supplier's detail governs the figure — and the gap to set this morning is a little over 7.5 mm, not six.
Set six anyway and the consequence is arithmetic, not opinion: the 6.1 mm still to come uses the whole six and runs about eight hundredths of a millimetre past it, so the two ends meet at around sixty-nine degrees — a degree short of the design maximum, and earlier still on a real extrusion carrying a burr on the cut end. The 1.5 mm that was meant to survive the hottest hour of the year has gone, and with it the room the sleeve, the gasket and the joint cover were relying on. The reverse error is less dramatic and still wrong: a joint over-opened on a warm afternoon is still visibly gaping in January and, on some systems, has lost the sleeve engagement the supplier requires.
Length is measured between splices, not between slabs. A mullion spliced twice through a double-height lobby is two shorter members with two smaller gaps, not one long member with one big one, and a run left unspliced through three floors because it arrived in one piece is a run that has to move three floors' worth of growth at whatever joint is above it. Where a steel back-frame carries aluminium pressure plates, size the joint on the member that is actually spliced, since aluminium travels roughly twice as far as carbon steel for the same swing.
Then make sure the joint is free to do the job you sized it for. A splice bridged by wet paint, a gasket run continuously through and sealed, a self-tapping screw driven through the sleeve to stop a rattle, a site weld added to fix an alignment problem — each one pins the joint, and a pinned joint does not stop the movement, it redirects the force into something never designed to take it. Where the joint must be sealed for water or air, it is sealed with something that stretches over the travel, on a bond breaker, not with a bead gripping both faces and the back of the gap at once.
- Confirm the member length between splices, not between slab soffits, from what is actually hanging on the wall.
- Read the metal temperature on that member, on that elevation, at that moment — not the air, not the forecast, not this morning's reading reused after lunch.
- Take the design band and the required residual clearance from the system supplier's detail and the facade specification, and query the band before setting anything if it looks thin for the finish.
- Work the gap, set it with a feeler gauge or a purpose-made setting block rather than by eye, and check it at both flanges of the extrusion.
- Check the sleeve engagement at the joint's open limit as well as at today's gap, so a cold night does not pull the splice short.
- Write the bay reference, the length, the reading and the gap on the sheet before moving the platform.
| Member material | Coefficient (× 10⁻⁶ per °C) | Movement in 4 m over 90 °C | Where the figure comes from |
|---|---|---|---|
| Aluminium extrusion, 6000 series | 23.6 | about 8.5 mm | Aluminum Design Manual, The Aluminum Association |
| Structural carbon steel | 11.7 | about 4.2 mm | AISC Steel Construction Manual |
| Austenitic stainless steel | 17.3 | about 6.2 mm | Published data for the austenitic grades; the producer's data sheet governs |
| Soda-lime float glass | about 9 | about 3.2 mm | Varies with composition — take it from the glass processor's data sheet rather than from a single published number |
| PVC-U window profile | roughly 60 to 80 | 22 to 29 mm | Varies widely by compound and by colour; the profile system's own literature governs, and a dark profile is the worst case |
| Normal-weight concrete | about 10 | about 3.6 mm | Aggregate type dominates and the spread is close to a factor of two, so take it from the mix design, not from a table |
Give it the length between splices, the alloy, the design band and the reading you have just taken off the metal, and it returns the gap for today along with the open and closed limits the joint has to stay between — which are the two numbers a snagging inspection in a different season needs.
The length of one mullion between stack joints — usually the floor-to-floor dimension.
The material of the mullion itself, which sets how far it moves per degree.
The lowest temperature the mullion metal itself is expected to reach in service.
The highest temperature the mullion metal itself is expected to reach in service.
What the mullion is reading right now, on the face being installed.
How much clearance must survive when the mullion is at its hottest and the joint is at its tightest.
Setting gap at today's temperature
0.246 in
- Total movement across the design range
- 0.31 in
- Gap at the coldest design temperature
- 0.36 in
- Gap at the hottest design temperature
- 0.06 in
- Expansion still to come from today's reading
- 0.19 in
- Contraction still to come from today's reading
- 0.12 in
They open the calculator with your figures already in it
Mullion Stack Joint Setting Gap Calculator: 0.246 in — 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 — 0.246 in — 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
- Covers axial thermal movement of the mullion only. Frame shortening, creep in a concrete structure and differential slab movement all land on the same joint and none of them is temperature-driven.
- Assumes the splice is free to slide. A stack joint pinned by a sealed gasket, a paint bridge or a fastener through the sleeve does not move, and the load goes somewhere else.
Everything in a Bay That Is Allowed to Move
Read a glazed bay from the weather face inward and the movement allowances are separate components with separate budgets, bought separately, installed by different operations and capable of failing on their own. The cap bead over the joint has a movement capability printed on a data sheet. The glass has an edge clearance around it. The frame has its splice. The whole assembly has a joint onto the structure behind it. None of those budgets can spend another one's, and the common site error is to assume that because there is a gap somewhere, the movement has somewhere to go.
What is worth noticing about the drawing below is that the components doing the work are mostly empty space. The clearance around a lite, the gap at a splice, the width of a perimeter joint — these are not gaps left over from the parts, they are the parts. Which is why the most reliable way to build a wall that cannot move is to be tidy: fill the head channel with mortar because it looked unfinished, run the gasket through the splice because it looked neater, pack the perimeter with foam because it was draughty. Every one of those is somebody improving a component out of existence.
One bay of a glazed wall, taken apart
- Exterior cap seal and joint cover — the weather line over the joint, which has to stretch across the same travel the joint underneath it is making Caulk & Sealant Calculator
- Insulating glass unit — grows less than half as far as the aluminium holding it, and cannot yield at all when it is pinched Architectural Glass Weight Calculator
- Edge clearance, gaskets and setting blocks — the empty space that keeps the lite off its frame, sized for the frame's movement rather than for the glass's own Window Setting Block Count Calculator
- Mullion and its stack joint — two lengths of extrusion on a sliding sleeve, which is where the wall's vertical growth is supposed to go Mullion Stack Joint Setting Gap Calculator
- Perimeter movement joint — takes thermal movement, interstory drift and slab-edge tolerance at the same time, in the same width Curtain Wall Perimeter Joint Width Calculator
- Slab edge and structural opening — moves on its own account through deflection, shrinkage and creep, none of which is temperature-driven
The Joint Between the Wall and the Building
The perimeter joint is asked to take three unrelated things at once and it is normally sized for one of them. Thermal movement of the wall is the one everybody remembers. Interstory drift is the second — the frame racking sideways between floors under wind or seismic load, which arrives at the head joint as shear rather than as extension. The third is the structure's own behaviour: a slab edge that deflects under live load, shortens under prestress, and creeps for years afterwards. ASTM C1472 Standard Guide for Calculating Movement and Other Effects When Establishing Sealant Joint Width is the document that sets out how those are established and combined, and its point is the one that gets missed on site: they are added, not compared, because they can all be asking for room on the same afternoon.
Width then follows the sealant's rated capability rather than the gap that happens to be there. Under ASTM C920 Standard Specification for Elastomeric Joint Sealants the movement class is a percentage of the installed joint width, so six millimetres of anticipated movement into a Class 25 polyurethane needs twenty-four millimetres of joint before any tolerance is added at all. A silicone qualified at Class 50 halves that. This is why the class printed on the data sheet is a design decision and not a procurement one — it frequently decides whether the joint fits behind the cover or whether the cover has to be redrawn.
Construction tolerance is then added on top of the movement width, not shared with it. It is the erection tolerance of the primary structure plus the setting tolerance of the wall, and a cast-in-place slab edge poured against a bowed form will use all of a generous allowance in one bay and none in the next — which is exactly why the joint is surveyed before it is specified as a single number for the elevation. ASTM C1193 Standard Guide for Use of Joint Sealants covers the joint's build, and the part of it that gets skipped is the bond breaker: a bead adhered to three faces cannot extend, so the backer rod is not packing, it is the component that decides whether the sealant is a two-sided joint or a rigid strip.
Feed it the three movements added together rather than the largest of them, the class off the sealant data sheet, and a tolerance taken from the slab edge you actually surveyed — and read the backer rod diameter and bead depth it returns as part of the answer, not as an afterthought.
Thermal movement, interstory drift and structural deflection added together as one number.
The ASTM C920 class printed on the sealant's data sheet.
How far the gap as built may differ from the gap as drawn.
Minimum designed joint width
1.25 in
- Width demanded by movement alone
- 1 in
- Construction tolerance added
- 0.25 in
- Backer rod diameter
- 1.56 in
- Sealant bead depth at mid-joint
- 0.5 in
- Movement the finished joint can absorb
- 0.31 in
They open the calculator with your figures already in it
Curtain Wall Perimeter Joint Width Calculator: 1.25 in — 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 — 1.25 in — 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
- Sizes a butt joint working in extension and compression. A joint also racking in shear — a head joint at a drifting floor edge — needs the shear component resolved into the movement figure before it is entered.
- Says nothing about adhesion. Every substrate pairing in a perimeter joint (aluminium, sealed concrete, air barrier membrane) needs its own primer and its own adhesion test.
Glass Moves Less Than What Holds It
Soda-lime float glass has a coefficient of expansion under half that of the aluminium around it, so the clearance at the edge of a lite is not, mostly, there for the glass's own growth. It is there for the frame's growth, for the frame's deflection under wind and dead load, for erection tolerance, and for the plain fact that glass has no capacity to yield. Everything else in the bay can take a small abuse and recover. A lite that ends up bearing hard on a frame at one corner cannot, and it announces the fact with a crack that runs from that corner at the first sharp cold snap.
The mechanics of keeping it clear are old and well documented. Setting blocks of the correct hardness carry the weight at roughly the quarter points of the bottom edge, so the load is spread and the lite is held up off the sill with its edge clearance intact all the way round; face clearance and bite are what hold it in the pocket without gripping it. The National Glass Association's Glazing Manual and IGMA TM-3000 North American Glazing Guidelines for Sealed Insulating Glass Units are where those clearances belong, along with BS 6262 Glazing for Buildings and BS 8000-7 Workmanship on Building Sites — Code of Practice for Glazing in British practice. What none of them tolerate is a block improvised from an offcut of gasket, which compresses to nothing by the second summer and drops the lite onto its frame.
There is a second movement problem inside the glass itself, and it surprises people who have only thought about frames. The edge of a lite sits under a gasket in shade; the centre takes full sun. The differential puts the cool edge in tension, and annealed glass has a modest tolerance for it. Where the shading pattern, the coating, the interlayer or an internal blind pushes that differential up, the answer is heat-strengthened or tempered glass to ASTM C1048 Standard Specification for Heat-Strengthened and Fully Tempered Flat Glass, chosen from the processor's own thermal stress analysis rather than from a rule of thumb. The same reasoning reaches the sealed unit's edge: ASTM E2190 describes what a unit is qualified to survive, and a unit clamped by a hard bead has its edge seal carrying the frame's movement.
A Glass Block Panel Is Not Part of the Wall
A glass block panel behaves as one rigid, non-structural sheet, and every detail around it exists to keep the building's movement out. It carries its own weight and nothing else. It is not a bearing wall, it is not a shear panel, and any load the surrounding structure hands it goes straight into a material that has no way to shed it. So the panel bears at the sill and floats everywhere else: a compressible expansion strip runs the full length of both jambs and the head, panel anchors cross the mortar bed joints into the jambs to hold the panel in plane without restraining it in its own plane, and the head detail is a channel the panel sits inside rather than under.
The single most common way a panel is ruined is somebody tidying up. The expansion strip at the head is compressible and it looks unfinished, so it gets pointed solid with mortar by a bricklayer doing what a bricklayer normally does, and at that moment the panel becomes a rigid infill in a moving opening. Deflection in the lintel above then lands directly on the top course of glass. The strip is not a filler and it is not a temporary protection; it is the reason the panel survives the first heating season.
Inside the panel, horizontal wire or ladder reinforcement laid in the bed joint at the manufacturer's stated course interval ties the joints together and spreads stress across the panel instead of letting it concentrate at one block. The interval is a manufacturer's figure and it varies with unit size and panel dimensions. So do the panel's maximum area, width and height, which are capped both by the glass unit masonry provisions of the adopted building code and by the block manufacturer's own published limits, and which differ between exterior and interior panels and between standard and thin units. Take those from the edition your jurisdiction has actually adopted and from the system's literature — beyond them, the panel needs intermediate stiffeners and it is no longer a straightforward infill.
Fire-rated block is a separate case again. Where the panel is a rated opening protective, the assembly is a tested one and its detail is the listed detail: NFPA 257 Standard on Fire Test for Window and Glass Block Assemblies is the test standard behind it, with NFPA 80 Standard for Fire Doors and Other Opening Protectives governing installation. Nothing about mortar, anchors, strip or reinforcement in a rated panel is available for substitution on site, however sensible the substitution looks.
The strip belongs at the head and both jambs — the sill is where the panel bears, so it takes mortar rather than compressible material — while this returns the whole panel perimeter, which leaves the sill run in the answer as surplus unless you take it off. The wire reinforcement quantity falls out of the course count, which is why the block size changes the order as much as the panel size does.
The width of the glass block panel opening.
The height of the glass block panel.
The nominal size of the square glass blocks used.
How many courses apart horizontal wire reinforcement is placed.
Expansion strip needed
16 ft expansion strip
Glass block panels are non-structural (self-weight only) and rely on the perimeter channel/expansion strip and internal reinforcement to accommodate building movement — always follow the specific glass block system manufacturer's installation requirements for panel size limits.
- Number of courses
- 6 courses
- Total wire reinforcement length
- 12 ft
They open the calculator with your figures already in it
Glass Block Panel Expansion Strip & Wire Calculator: 16 ft expansion strip — 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 strip length returned is the full rectangle perimeter, 2 x (width + height). The calculation draws no distinction between jambs, head and sill, and adds nothing for corner laps, for the thickness of the strip itself, or for the perimeter channel, frame or anchor track the strip sits in. Many glass block systems bed the panel in mortar at the sill and call for the compressible strip at the jambs and head only, so check your system's perimeter detail before ordering to this number.
- This is a materials take-off, not a structural or wind-load check. Nothing here tests the panel against the manufacturer's published limits for maximum panel area, maximum unsupported span or aspect ratio, and nothing flags the point at which an intermediate stiffener, a structural surround or a heavier reinforcement schedule becomes mandatory. The calculator will return a figure for a panel far larger than any glass block system permits.
- Courses are simply the panel height divided by the block's nominal module. No allowance is made for the sill mortar bed, for the head expansion strip, or for joints laid wider or narrower than the nominal module, and the panel width is never divided by the block module at all, so nothing checks that your width lands on whole blocks. Glass block cannot be cut, so the opening has to be designed around whole units in both directions.
- The wire figure is nothing more than reinforced courses multiplied by panel width. It excludes lap and splice allowance at joints and at the jambs, extra reinforcement at panel anchors, at the course below the head or around any opening in the panel, and any vertical reinforcement. Where the spacing you select is larger than the panel's course count, the figure comes back as zero.
- Only expansion strip and horizontal wire are quantified. Blocks, mortar or mortar-substitute silicone, spacers, panel anchors, jamb and head channel, sill sealant and backer rod, and finished joint sealant are all outside this calculation.
The One Clearance You Do Not Get to Size
Everything on this page so far has been arithmetic you are entitled to do. Fire-rated glazing is the exception, and it is worth knowing precisely where the exception starts. In a rated assembly the edge clearance, the bite, the frame, the glazing tape, the beads and the fasteners are all part of what was tested, and the assembly holds its rating only in the configuration that passed. The tests are UL 9 Fire Tests of Window Assemblies and UL 10B and UL 10C for door assemblies; NFPA 80 governs the installation of opening protectives; NFPA 257 covers window and glass block assemblies. A clearance opened up by two millimetres because the opening was tight is not a small deviation, it is an untested assembly.
So the label and the installation instructions are the governing documents on that opening, and the thermal reasoning behind the rest of the elevation stops at its frame. Confirm the listing before the unit goes into the opening, keep the instructions with the glass rather than in the site office, and if the opening as built will not take the listed clearance, that is a report, not an adjustment.
Use it to know roughly what to expect before you open the box — it returns typical published figures for common glazing and frame pairings, and it says plainly that the listing for your actual product, not this reference, is what you install to.
The fire-rated glazing product category.
The frame material the glazing is set into.
Typical listed edge clearance
0.37 in
This is a TYPICAL reference figure only, not a calculated or code-compliant value — fire-rated glazing edge clearance is set entirely by the specific tested and listed assembly (UL 9/UL 10B/UL 10C, NFPA 80/257). Always verify the exact bite/clearance against your specific product's UL listing and the manufacturer's installation instructions before installation; deviating from the tested clearance voids the fire rating.
They open the calculator with your figures already in it
Fire-Rated Glazing Typical Edge Clearance Reference: 0.374 in — 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 — 0.37 in — 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
- A fire rating is not an impact rating. Glazing in a door, a sidelite or any other hazardous location has to satisfy safety-glazing impact requirements as well, and traditional wired glass is where the two collide — it is weaker under impact than plain annealed glass of the same thickness, and its use in those locations has been restricted for years. Confirm the lite is listed for both before the edge clearance matters at all.
- One figure is returned for every lite size, and listings do not always work that way. A larger lite expands more, so a listing can give a different clearance band for a big lite than for a small one, and this page never asks how big the opening is. Read the size ranges in the listing before treating the figure as fixed.
How a Bound Joint Reports Itself
A wall that cannot move makes noise, and the noise has a timetable. Ticking or a sharp crack along an elevation as the sun leaves it in the late afternoon, and again as it arrives in the morning, is a joint sticking and releasing under load — stick-slip in a splice or under a cover plate. Occupants report it as pipework, as birds, as the lift. It is worth taking seriously on a new wall because it is the earliest honest evidence that something in the movement chain is bound, and it arrives long before anything visible does.
The visible symptoms come at the extremes and therefore come late. A mullion bowed between anchors, a joint cover that has telescoped out of its neighbour or popped its clip, a gasket walked out of its race and hanging in a loop, a fastener sheared at a bracket and found on a canopy roof below. On the glass side the signature crack for a pinched lite starts at a corner; the signature crack for thermal stress starts perpendicular to a shaded edge, in the middle of a run rather than at a corner, and is often mistaken for impact damage. Fogging inside a sealed unit on a wall that is otherwise sound points at an edge seal that has been asked to carry frame movement.
Sealant tells its own story and it is the easiest to read. A bead torn through its own thickness generally means the joint was narrower than the movement demanded. A bead released cleanly off one face is a preparation or primer problem instead. A bead thinned into a taut membrane across a wide joint went in too deep, with no backer rod behind it or with the rod pushed too far in.
The awkward part is the timing. Every one of these symptoms belongs to a temperature the building may not reach for months, so a wall handed over in February has been proved against roughly nothing. That is not an argument for waiting. It is an argument for the joint gaps having been set correctly in the first place, and for there being a record of what they were set to, because the alternative diagnosis in year two involves stripping covers off an occupied elevation to find out.
The Number Nobody Wrote Down
The logbook is a ten-second-per-joint habit and it is the only thing that answers a question raised eighteen months later. Bay reference, member length between splices, metal temperature, the gap actually set, the gauge used, the time of day and which elevation. That record turns a disputed joint into a checkable one: anybody can take today's reading, work the expected gap for that bay at today's temperature, and compare it against what the sheet says was set and what a feeler gauge finds now. Without it, a joint that has closed up is indistinguishable from a joint that was never opened, and the two have completely different remedies.
It also does something for the next job. Fifty logged readings across a winter and a summer are real evidence about whether the design band in the specification matched what that finish on that orientation actually does — the single input everything else here multiplies. Hand the sheet over with the operation and maintenance information rather than binning it at practical completion, and the consultant who arrives in year five to chase a noise has somewhere to start other than a cover plate and a crowbar.
What to have settled before the first joint is set
None of this is a material order in the usual sense — most of it is empty space with a dimension on it. Settle these six and the wall's movement has somewhere to go; miss one and the movement goes through whatever is stiffest.
- Design temperature band, with the finish and orientation in it — The band drives the total travel and therefore the size of every joint and cover on the elevation. If it looks thin for a dark finish on a sun-facing wall, raise it before setting anything.
- Member length between splices, bay by bay — Not floor-to-floor off the drawing. A run that arrived unspliced through two floors has to move two floors' worth at the next joint up.
- Residual clearance the joint must keep when fully closed — From the system supplier's detail. It is what the sleeve, the gasket and the cover live in at the hottest hour of the year, and it is never zero.
- Perimeter joint width, built from three movements added together — Thermal movement plus interstory drift plus slab-edge tolerance and deflection. The sealant's ASTM C920 class then decides how wide that has to be.
- Expansion strip, anchors and reinforcement for any glass block panel — Strip at the head and both jambs, anchors across the bed joints into the jambs, wire at the manufacturer's course interval — and a head detail nobody is allowed to point solid.
- A temperature log sheet in the folder, not in somebody's head — Bay, length, metal reading, gap set, gauge, time. It costs nothing now and it is the only defensible record when the elevation is investigated in a different season.
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.