Two numbers, one division
The arithmetic is short. A segment of length L moves by L times the coefficient times the temperature change. A sealant rated at plus or minus X percent can take X percent of its own installed width in each direction. So the width you need is the governing movement divided by X over 100.
Work the default through. A 40 foot run broken every 16 feet becomes 3 segments of 13.33 feet, with 2 joints. Each segment is 160 inches. At 34 millionths per degree F over a 160 degree swing, a full-swing movement of 160 x 0.000034 x 160 = 0.8704 inches. Installed at 70 degrees F, the closing half is 160 x 0.000034 x 80 = 0.4352 inches and the opening half is the same 0.4352 — even, because 70 is exactly the midpoint of minus 10 and 150. Divide the governing 0.4352 by 0.25 and the joint wants 1.7408 inches, which the page rounds to 1-3/4.
That is a startling answer, and it is the honest one for a material with a plastic-sized coefficient over a 160 degree surface swing on a 13 foot segment. It is also why claddings with high coefficients are not sealed into rigid joints at all: they are hung on slotted fastenings and lapped so the movement disappears into the laps rather than into a sealant bead. If your material genuinely wants a two inch joint, the answer is more joints, a different jointing method, or a different material — not a wider bead.
Install temperature is the free variable
Take that same segment with a plus or minus 25 percent sealant and vary only the day you cut it:
| Surface temperature at install | Governing movement | Joint width |
|---|---|---|
| -10 F, the coldest it gets | 0.8704 in, all closing | 3.4816 in (3-15/32) |
| 30 F | 0.6528 in, closing | 2.6112 in (2-5/8) |
| 70 F, the midpoint | 0.4352 in, either way | 1.7408 in (1-3/4) |
| 110 F | 0.6528 in, opening | 2.6112 in (2-5/8) |
| 150 F, the hottest it gets | 0.8704 in, all opening | 3.4816 in (3-15/32) |
Building at either extreme doubles the joint, because the entire swing has to go into one direction. Building at the midpoint halves it. Nobody schedules a facade around the thermometer, but it is worth knowing which way the error runs: a joint set on a hot wall in August is being asked to open, and a joint set on a frozen wall in January is being asked to close, and only one of those two failures is visible from the ground.
What this page does not decide
It does not tell you the coefficient, the movement capability, or how far apart the joints are allowed to be. All three are product figures. The coefficient in particular varies enough between formulations of the same nominal material that a class figure is close to useless — published values for plastics span a wide range, and a sheet for one grade does not carry to another. Maximum spacing is an installation instruction and it exists partly for movement and partly for reasons that have nothing to do with temperature at all, like allowing the material behind to dry.
It also says nothing about depth. A joint has to be the right shape as well as the right width, backer rod is what sets the depth and keeps the sealant from bonding to the back of the joint, and the depth relationship is a data-sheet number. Once width, depth and total length are settled, the caulk calculator turns them into tubes and backer rod.
Related pages
For movement in specific materials, the metal thermal expansion calculator covers metals including a shrink-fit mode, pipe expansion covers plumbing runs and their anchors, and wood movement and the flooring expansion gap calculator deal with the humidity-driven movement in timber, which follows completely different rules from temperature. For the trim itself, the exterior trim calculator works out the corner boards, casing and rake from the elevation, and the siding calculator handles the field.
Questions people ask
How wide should an expansion joint be?
Wide enough that the movement it has to absorb is within the sealant movement capability. If the segment moves 0.4 inches at worst and the sealant is rated plus or minus 25 percent, the joint has to be 0.4 divided by 0.25, which is 1.6 inches. There is no general answer because the movement depends on the material, the segment length and the temperature swing, and the capability depends on the sealant grade. Both come off data sheets rather than from a rule of thumb.
Why does the temperature when I install it matter?
Because the movement capability is measured from the installed width in both directions, and the installed width is whatever the run happened to be at that moment. A joint cut when the surface is at its hottest has already used up all its closing room, so the whole swing has to be absorbed by opening, and it needs roughly twice the width of the same joint cut at the midpoint of the range. The calculator shows the full spread across five install temperatures for exactly this reason.
Does a plus or minus 50 percent sealant let me halve the joint?
On this arithmetic, yes, and that is why higher-capability sealants exist and cost more. What it does not do is change the movement itself, and it does not remove the constraints that come from the other end: joints have a practical minimum width for filling and tooling, they have a depth relationship the sheet specifies, and a very narrow joint that has to stretch 50 percent is unforgiving of a bond that is not perfect on both faces. Take the capability from the sheet for the grade you are actually buying.
What if the calculated joint comes out absurdly wide?
Then the material is telling you it should not be jointed that way. High-coefficient claddings over long segments and big surface temperature swings produce joint widths that nobody would build, and the real answers are shorter segments, a lapped or slotted detail that lets the movement disappear into the overlap rather than into a bead, or a different material. A wide bead of sealant is the wrong solution to that answer, because the failure mode is not the sealant tearing, it is the material buckling or the fastenings shearing.
Should I use air temperature or surface temperature?
Surface temperature, and the difference is large. A dark surface in full sun runs far above the air around it, and a light one in the shade tracks it closely. Since the movement is proportional to the swing, using air temperature on a dark south wall understates the answer substantially. If you do not have a surface reading, the honest approach is to enter what you believe the extremes are and then look at how much the answer moves when you change them by twenty degrees either way.