Hi, apologies for the length of this, I'm hoping as a group you knowledgeable established or future Engineers and Physicists can help me out with a problem. I have been having a discussion with a colleague about expansion of a 6m x 7m integrally waterproofed (Kryton KIM) concrete slab roof that is to have 4 toughened and laminated 1m square skylights fitted directly into rebates in the concrete with a sealant bed to give a flush walkable area rather than having upstands to mount the glass on. The skylights are positioned towards each corner of the roof and 1m in from each side. We need to assess the relative movement of the glass and the rebate quite accurately as the sealant we intend to use on has a movement limit of 5% so we need to alter the sealant bed thickness to accommodate the differential movement. The first most basic issue I need to confirm is will the hole in the concrete get bigger or smaller when the slab expands? The more I look at the problem the more unclear the answer becomes. 1. will the expansion of the slab as a whole move the opening uniformly in one direction and will the concrete adjacent to the sides of the hole expand to make the hole bigger? or... 2. will the concrete in between each of the openings expand more and as the distance is greater expand more and so make the opening smaller? or.. 3. will the opening get deformed so it is smaller across the center of each side and longer towards the corners forming a squashed box (like a vacuum inside would collapse a cube). For the sake of my sanity I have gone for the opening will get bigger, if this is the case then read on, if not feel free to LOL but please let me know... My argument is that only the expansion of the 1m section adjacent to the glass needs to be calculated and compared to the glass expansion as any expansion of the rest of the slab would 'push' the entire opening and the glass in a certain direction, only the difference in the expansion coefficients for each square meter would cause the dimension of the hole to change. My colleague does not agree and insists we use the expansion of the concrete over the complete area of the roof. If I'm wrong please let me know and dont waste your time reading on. I have calculated the expansion of each of the glass panels which are 1m square and also for a 1 meter section of concrete. The expansion coefficients for hardened glass and concrete are similar; glass is 5.6 x10-6 m/(m K)), concrete is 9.8 x10-6 m/(m K)). These are the best figures i can obtain without actually specifying an exact specification so may change slightly. The roof is in the UK and I have used a very unprobable +- 60`C temperature range from a 20'C installation temperature (ie. -40'C to +20'C and +20'C to +80'C). At these extremes the glass will expand or contract 0.3mm in each direction from the install temperature. 1m of concrete will expand or contract by 0.6mm each way over the same temperature range. which gives a difference of 0.3mm A given a 6mm bed thickness the sealant on each side of the glass panel could accommodate a movement of 0.3mm so 0.6mm in total. which gives an allowance for error of 50%. There is also the rate of expansion to be taken into account, after a very cold night (sticking with unlikely extremes) of -20'C, the sun comes up on a scorching day and the temperature of the glass due to solar gain reaches 60'C in a few minutes. The concrete is slow to heat and is still at -20'c which would equate to about -0.4mm concrete dimension but a +0.2mm glass size so 0.6mm differential from the neutral (install) dimensions, the 0.6mm allowable sealant would still be satisfactory but right on the limit until the concrete warmed up and the difference in sizes would reduce slowly. Another issue is the slab will not expand isotropically as it is integrally insulated (large EPS blocks 500mm x 240mm) placed in rows along the 7m length with 240mm x 120mm steel reinforced beams poured in situ between the rows of insulation and a 100mm rising to 200mm thick concrete slab continuously formed with the beams above the insulation. (a bit like castle wall revetments upside down if you know what I mean). Another factor is the supporting perimeter walls are made of insulated concrete forms or ICF as it is known (60mm polystyrene slabs either side of a steel reinforced 200mm concrete core) which are being poured in one hit with the roof concrete. One advantage of the ICF is that it keeps the concrete core of the wall at a relatively stable temperature but this will be largely negated by having a 42m square open roof through which this heat can escape albeit quite slowly, it will reduce the temperature swings of the roof slab to some extent. I can attach a link to some drawings on dropbox or similar if this will help and if anyone is interested please let me know. If anyone can confirm / correct or dismiss my logic above I would be most grateful as I've been trying to figure this out for a week and am really not very eager to get up in the mornings anymore. Thanks in advance for anyone who can help to solve this puzzle.