Before peeking below to determine time and place of the paragraph in italics below, look at the issues presented
A roof deck can move excessively from changes in temperature (and moisture content).
The best place for a roof membrane is in a protected configuration where it’s not exposed to abuse, impact, or UV (or temperature change).
Vegetation on the roof isn’t necessarily harmful.
Concrete slabs can serve as ballast and keep the sand (or thermal insulation) in place.
So here’s the paragraph:
There are certainly reservations about flat roofs. The cleverest method seems (to me) to be the one Le Corbusier used in Stuttgart. He says that too much movement of the reinforced concrete due to heat and cold can cause cracks, (which) he covers with waterproof membrane (and) with a rain-dampened layer of sand, which he covers with concrete slabs. Grass grows in the layer of sand. So it always has the same humidity.
The above quotation was written in February 1928, extracted from The Bauhaus Group: Six Masters of Modernism.
With 82 years of experience since the reference to Le Corbusier, we now know that, even with the best workmanship, dead level roof systems are a design error.
Again, from The Bauhaus Group: Six Masters of Modernism:
Since roofing membranes are supposed to keep a building dry, why do we put them on top of the building where they’re exposed to all sorts of degrading influences? Examples of degradants would include ultraviolet energy, ozone, oxidation, moisture, thermal cycling, ice cracking, hail impact, physical abuse, bird droppings, chemical spills, hot steam, algae, mildew, wind-induced flexure, vibration, newly installed photoelectric panels, or intentionally vegetated roofings.
Le Corbusier was right about the virtues of protected roof membranes. Roof membranes ought to be protected, and the concept of protected membrane roofs (PMRs) has been well validated, especially in cold climates. Perhaps cold climates put two extra demands on the structure. One is design load. Because they’ve been designed to withstand heavy design loads from snow, ice, and wind, these structures are more likely to accept additional weight, such as from a ballasted roof system. In any event, the thermal insulation, which is above the roof membrane, has to be kept in place somehow.
Second, since the prevailing vapor drive is from the interior toward a colder exterior, superimposing the thermal insulation over the roof membrane allows this single membrane to serve as a vapor and air barrier.
Now jump to 2010. We have mostly exposed membranes, generally of single-ply materials that are very thin (45 to 90 mils) and vulnerable to vandalism, hail impact, and foot traffic. We’re now coating them to increase albedo and emissivity, but the coatings are of durability that isn’t yet established. We’ve complicated the roof surface not only with HVAC, cell towers, and the like, but with photocells placed close together, or with vegetated areas with unknown consequences with regard to root growth and blocked drainage.
Perhaps we have it all wrong. As early as the 1950, the concept of a modern PMR was introduced to the roofing industry. According to the January 1978 Canadian Building Digest, the roof deck was a sloped cast-in-place concrete, but it was considered that metal or wood might serve equally as well. The deck is covered with a membrane to serve as a vapor barrier and roofing membrane.
If the roof deck sloped, use of a double drain would allow any water that reached the membrane to be removed from the system. The upper drain location would address heavy rainfall and reduce the water flow between insulation or vegetation surfaces.
Baker and Hedlin pointed out that the lower, protected membrane should be the primary one and, since it’s protected from the elements, should be durable enough to protect the building even when the upper elements (PV panels, vegetated roofs, etc.) need replacement.
The unique resistance of extruded polystyrene (XEPS) to liquid water, water vapor, and freeze-thaw action was a breakthrough for the low-slope roofing industry. The inverted roof membrane assembly (IRMA) received a patent in 1968. By 1995, there were more than 50,000 PMRs in service in the United States and Canada, with thousands more in Europe, Asia, and the Middle East, and have progressed rapidly since about 1964.
PMRs were widely studied and implemented by the U.S. Corps of Engineers for use in cold regions as well. Even though protected, we now recognize some things that Le Corbusier did not:
The membrane must be sloped to drain, never dead level.
Filter fabric is needed to block fines from plugging drainage channels.
Space must be provided below pavers to facilitate upward drying.
While the earliest concepts of PMRs tried to adhere the styrene boards to the membrane so it wouldn’t float or blow away, this proved difficult to accomplish under field conditions. Instead, gravel ballast applied over a non-woven filter cloth succeeded in “rafting’ the boards together so that, if roof areas ponded during a heavy rainstorm, the boards would be restrained by the fabric. The fabric also held back fines and small gravel from working between insulation boards. Alternatives to loose ballast include freeze-thaw resistant concrete pavers, shimmed so that moisture could escape from beneath the pavers, and composite boards of cement topping directly bonded to the XEPS. In this case, tongue-and-groove side joints resist floatation.
In the case of installation of PV arrays on a roof, the big concern is the number of penetrations needed for pedestal support if such supports are used. With a sturdy and thick option of cement-surfaced composite at 15/16-inch thickness, supports could be directly anchored to the composite boards. As an alternative, 2-inch concrete planks (20 psf) placed over the composite boards could be used so that nothing penetrates to the primary roof membrane system. For extreme wind conditions, interlocking pavers have been joined together with stainless metal straps, useful at perimeters and corners. Heavy pavers have also been useful for anchoring handrails.
All of these elements – the composite boards, XEPS insulation boards, stone ballast, and pavers – could be reused (recycled) as needed.
Baker and Hedlin pointed out that, for vegetated roofs, heavy landscaping elements need to be placed directly over structural supporting elements.
Wayne Tobiasson, formerly of CRREL, points out that we’re using the wrong word (“waterproof”) to describe conventional roof membranes. Since we currently build roof membranes to be capable of withstanding minimal short-term (less than 48 hours) hydrostatic pressure that only occurs during drainage, we should be referring to them as water resistant, not waterproof. As we bury roof membranes under more “stuff,” all of which is very expensive to move aside when problems develop, it seems to me that we should be using pond liner “waterproof” systems, not our current crop of “water-resistant” roof membranes.