Analysis of Roof Components at Condominiums

Consultant Site Investigation:
On 1 and 2 May 2002 this consultant visited a Condominium project to inspect alleged corrosion problems on a section of roof, labeled CU-33 and CU-34, on the roof plan diagram, dated 22 May 2002. This consultant was informed that the roof was originally constructed in 1965, and that a phenolic-containing roof system was installed in 1987-1988. As indicated in the report, the roof insulation was 1.2" thick, mopped, with a four ply coal tar membrane with organic felts, with a top pour and gravel. The metal form was hot dipped galvanized carbon steel. A layer of light weight concrete (LWC) was placed directly on the metal form, with phenolic foam placed on top of the LWC. A bituminous material separated the phenolic foam from the LWC.

The roof was initially inspected from within the living spaces in the ceiling of a ballroom, in a closet adjacent to the ballroom, in a corridor between the ballroom area and the no longer used kitchen area, and in the kitchen area. Severe corrosion, including full penetration by corrosion reactions, was identified in each of the areas inspected. The corrosion product was primarily red in color (hematite), although there were several indications of black corrosion product (magnetite) where the metal was in direct contact with the LWC. Three samples of corrosion product were collected from inside the 1 building. These were labeled Samples 1, 2 and 3. Sample 1 was taken from an area in the ceiling of the closet next to the ballroom. This area is under a partially enclosed patio, and no phenolic foam had been applied over the LWC. Sample 2 was taken from the ceiling of the ballroom in an area near a roof drain. This sample was taken from an area of the roof where the phenolic foam had been removed in 2001. It was also identified as an area where a workman had partially fallen through the roof. Sample 3 was collected from an area of roof near the parapet. These areas are identified on the roof plan in the report of 22 May 2002.

The external roof sections were also examined, and two sections of roof were removed. These were "sandwich" sections wherein the composite roof consisting of the membrane, the foam, the LWC and the metal sections were connected with roofing screws and a section, approximately 2' by 2' were cut, bagged and sent to this consultant's laboratory. Sample 4 was collected from an area adjacent to the enclosed living space, near the flashing joining the building to the roof. Sample 5 was a section of metal form and LWC collected from approximately the middle of the roof area where the phenolic foam had been removed in 2001. This sample was also forwarded to this consultant's laboratory. This consultant's samples 1, 2 and 3 were hand-carried back to the laboratory.

Consultant Laboratory Investigation:
Specimens from each of the components of the roofing system were immersed in water for a period of 24 hours, after which the water samples were measured for leachable chloride content and pH. The protocol for these tests and the resultant pH and chloride concentrations are shown in the following table:

Upon opening the bagged samples of 4 and 5, it was noted that a significant amount of moisture had been released from the samples and the insides of the plastic bags were coated with water. Additionally the LWC samples were exceptionally heavy and friable for a product of this type. Accordingly, an LWC sample from each of the "sandwiches" was weighed and exposed to a drying furnace at 150 degrees F for 24 hours. After the exposure the samples were re-weighed with the following results:

Samples of each of the corroded area and associated roof components were sent to Quantitative Technologies, Inc. (QTI) for a determination of formate and sulfonate concentrations. The samples that were provided to QTI included: (a) corrosion product samples from areas 1, 2, 3, 4 and 5, (b) a sample of LWC from area 1, which was under the patio (inside the closet), where no phenolic had been installed, (c) samples of the tops of the LWC at areas 4 and 5, and (c) samples from the bottoms of the LWC at areas 4 and 5. The results of those analyses are appended to this report. To summarize those results, no detectable formate or sulfonate was identified in any sample except for the corrosion sample at area 2, where a workman had partially fallen through the roof. Even in this area the formate level was at only 202 ppm, judged to be a trace amount.

Consultant Conclusions:
It is obvious that the corrosion problems observed in the roof of the Condominium project are not related to the chemical composition of the phenolic foam that had been used for insulating purposes. Having examined dozens of roofs where phenolic foam did play a role in accelerating corrosion reactions, the damage to the roof bears none of the similarities observed in previous studies. When phenolic foams are associated with corrosion processes, formate is always observed in the corrosion product. Sulfonate is observed when the roofing system has been exposed to significant amounts of liquid phase water, and the phenolic foam is in direct contact with the metal form. With the exception of one sample, none of the corrosion product samples, or the light weight concrete samples showed measurable indications of formate, and none of the samples showed any measurable indications of sulfonate (note that the detectable limit is less than 50 ppm). The one sample of corrosion product that showed trace amounts of formate was in an area where a workman had penetrated the roof, and where the phenolic foam had been removed prior to this inspection. The disturbance of the roofing system, both by damage caused by the roof penetration, and by the high probability that removal of the phenolic foam contaminated the underlying roof and corrosion product with trace amounts of formate, explains the presence of the formate. It should be noted that, in prior investigations, corrosion induced by phenolic foams resulted in thousands of parts per million formate in the corrosion product, in contrast to the 202 ppm reported here. Even 3 the upper surfaces of LWC, closest to the phenolic foam, were devoid of either formate or sulfonate, in each of the areas examined, including where the roof had been damaged.

The observation that the LWC was virtually saturated with water explains the corrosion damage to the metal form. It is likely that the LWC has been wet for a considerable period of time, and that conventional corrosion by aerated water is the cause of penetration of the metal form. Once the LWC was saturated, the construction of the roof inhibited drying out of the materials, causing it to act as a wet poultice against the steel metal form, in much the same manner as a sponge retains water. It is significant that the pH of the LWC was well below that expected for cementitious products (11 and 7 respectively). Cementitious products based on portland cement usually have pH values higher than 12-13. These results indicate that the soluble hydroxides usually found in these products have either oxidized by the aerated water, or have leached out of the LWC, causing the product to approach a neutral pH. Either of these possibilities takes a considerable amount of time. The levels of chloride observed in the LWC are nominal, and to be expected for a seacoast environment. Note that no chloride was detected in the corrosion product, indicating that chloride is not a factor in the corrosion of this metal form. While galvanizing can protect steel from corrosion, by acting as a sacrificial material, its does so by preferentially corroding the zinc. When the thin galvanizing layer is compromised, corrosion of the steel metal form occurs.

To conclude, there is no evidence whatever that the phenolic foam played any role in the corrosion damage observed in the metal form.

To see the resume of the expert associated with this case study, see the link below.