You are on page 1of 10

Focus on Insulation

Metal-deck corrosion:
Three case studies
A combination of factors was found to have contributed to major,
unpredictable patterns of deck corrosion

by Richard P. Canon, RRC, PE

The roofing industry has exhibited Project No. 1 over an area used to charge batteries
a growing concern in recent months for lift trucks. The rusting was ex-
regarding corrosion of metal decks. The discovery of corrosion of the plained as being related to battery acid
Several recent articles in Professional metal deck was initially made on the fumes, a theory later invalidated. Origi-
Roofing ¹ and NRCA Bulletin 15-91 ² roof of Project No. 1 in 1985, about nally, both occurrences were rational-
have discussed a growing concern in three years after construction, when ized as localized anomalies and not a
the roofing community regarding corro- several new penetrations were cut in source of major concern.
sion of metal decks. During the last two the asphalt built-up roof (BUR). This
years the author's roof consulting firm exposed moderate-to-severe corrosion In 1989, six years after construction,
has investigated three projects that of the deck. This corrosion was initially pieces of decking began falling to the
have experienced severe, premature thought to be dormant and isolated to production floor in various areas of the
corrosion of metal roof decking. Two of active leaks. plant. There were numerous leaks doc-
the decks were primed but not painted, Two years later a small area of cor- umented in the plant with no logical
and one was galvanized (ASTM A 446 roded deck was replaced after rust was pattern. Maintenance personnel also
steel with hot dipped galvanized coat- observed on the bottom of the deck discovered several rather odd "sink
ing conforming to ASTM A 525, G-60).

The roof assembly for each project


is shown in Figures 1, 2, and 3. Table
A summarizes the history and data
regarding each project. All three used
phenolic insulation as part of the sys-
tem. This article will discuss the evi-
dence that a combination of factors
involving the insulation, moisture in
the system and inadequate protection
by the deck treatments may have con-
tributed to deck corrosion in these
three projects. Figure 1: Project No. 1 Roof Cross Section

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 1 of 10
TABLE A
Parameter Project No. 1 Project No. 2 Project No. 3

Project location Northwest South Carolina Coastal South Carolina Southeast Michigan

Year constructed 1982 1987 1987

Age at investigation 7 years 2 Years 4 Years

Roof age at removal 8 Years 3 Years Pending 5 Years

Generic membrane Aggregate surfaced BUR Ballasted EPDM Mechanically Fastened


CSPE

Building Use Plastic Molding Retail Store Office Building

Insulation Type One layer phenolic One layer phenolic 1 layer wood fiber
2 layers phenolic

Insulation facer Foil over corrugated paper Fiberglass Fiberglass

Number of layers 1 1 3

½ inch wood fiber


Thicknesses 1.4 inch 2.3 inch 1 inch phenolic
1 inch phenolic

Unusual Interior None None None


Conditions

Deck Type Metal Metal Metal

Deck Gauge 22 22 20

Deck configuration Intermediate (B) Intermediate (B) Intermediate (B)

holes" on the roof. The owner was nearly every board joint appeared as a nolic insulation, which has been deter-
concerned and requested an investiga- white line indicating voids at insulation mined to be 23.4 percent (at 90 percent
tion by the author to determine what board joints. The same lines were visi- relative humidity and 68E F)³.
was happening to the roof system, ble on the roof on a frosty morning.
particularly the decking. Dry insulation extracted fromsome
Test cuts and cores confirmed the of the areas we had interpreted to be
The author's first task was to deter- observations of the IR survey. Areas of dry were found to contain as little as
mine the extent and severity of the phenolic insulation interpreted to be 1.6 percent MCBW.
corrosion. An infrared (IR) roof mois- wet were very wet. Moisture Content
ture survey indicated at least 6,000 By Weight (MCBW) was measured up The roof membrane had numerous
square feet 2, or 11 percent of the roof, to 1,025 percent. Most wet material was deficiencies, which had resulted in
had wet insulation with varying de- found to contain about 250 percent more than 100 documented leaks since
grees of moisture. The author also ob- MCBW. Compare these values with the occupancy. The author also found
served on an IR camera screen that equilibrium moisture content for phe- evidence from the test cuts that the

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 2 of 10
joints in the phenolic foam had been
butted, or up to ¼ inches wide, at the
time of construction. They were now
up to 1½ inches wide with an average
width of 1 inch. *[W E ESTIMATED THE
TOTAL CUMULATIVE AREA WHERE THERE
IS NO INSULATION DUE TO OPEN BOARD
JOINTS WAS 2,465 FT 2. THIS MEANT
THAT ABOUT 5% OF THE INSULATED
ROOF AREA WAS NOT INSULATED!]*

This was concluded, by observa-


tion, to be the result of shrinkage (i.e.
due to dimensional instability) of the
insulation, not improper placement.
The author believes the resulting gap
could have contributed to the corro-
sion in two possible ways:

! First, as the insulation shrank it


placed the membrane under ten-
sion in the region above the joints.
This may have resulted in water
infiltration through the membrane
in this region.

! Second, at the wide joints, water


Figure 2: From Project No. 1, severe corrosion of deck, plate, and fastener.
vapor would have been allowed to
rise freely. It thus could have con-
densed in those joints on the un- the deck. In the most heavily corroded tion interface. When it occurred close
derside of the membrane or near areas, all surfaces were corroded: enough to the deck, this phenomenon
the top of the insulation's flange, web, and flute. Numerous gal- could result in a leaching effect.
sidewalls. This would have led to vanizedinsulation fastener plates were
wetting of the edges of the insula- heavily pitted and corroded. After extensive study, the author
tion in the wide joints as moisture concluded that water had entered the
dripped from the top of the roof The author determined that the insulation from four sources:
assembly in those areas. interior conditions were such that a ! Leakage through the membrane.
vapor retarder would not typically be ! Condensation within the roof in-
At the test-cut locations, the author deemed necessary. Relative humidity sulation.
observed the aluminum foil facer was in during the heating season is generally ! Condensation in the areas of wide
varying states of decomposition. How- about 40 percent. The bottom of deck board joints.
ever, metallurgical and chemical analy- temperature is about 75E F. But since
sis concluded the foil was not a factor there was no vapor retarder, condensa- It is the author's opinion that the
in the corrosion. The foam was typi- tion was determined to be a significant presence of the moisture seems to
cally crushed, broken, and fractured. factor in the corrosion analysis. The have resulted in serious, accelerated
Many of the insulation fasteners in the reason for this is that during the winter corrosion of the metal deck, possibly
test-cut locations had corroded from months, the dew point would fall within because it caused an acid to leach from
0.172 inches diameter down to 0.05" the insulation (which is where it should the insulation (which will be discussed
diameter. Most of these fasteners had be), and most critically could have later in this article).
some scaly rust. Corrosion was typi- occurred, under certain conditions, in
cally most severe on the top flange of close proximity to the deck-to-insula- From the IR survey, visual inspec-

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 3 of 10
tion of the underside of the deck and Project No. 2
test-cut data, the author projected that Phenolic insulation with a fiber-
13 percent of the roof deck would have On September 22, 1989, South glass facer was in contact with the
to be removed and replaced. Further, Carolina was visited by a very unpleas- deck. In many locations, the facer had
the author estimated that about 30 ant Hurricane Hugo. The roof on this become firmly adhered to the corroded
percent of the deck would be lightly to project, which is located near the coast, deck. When the insulation was
moderately corroded and would require was exposed to some of Hugo's higher removed, the deck came up with it in
field coating. winds. The ballasted EPDM roof bil- strips the width of the flange. The as-
lowed and was ripped in several places. sociation of the fiberglass as a reactant
The physical condition of the insu- The owner arranged for emergency in the corrosion process has not been
lation and the extent of corrosion, how- repairs soon after the hurricane. During fully investigated, but is currently not
ever, led the author to recommend a final repairs three months later, exten- felt by the author to be significant.
total removal of the roof system to the sive corrosion of the primed deck was
deck. It would then be necessary to observed. After 1,710 square feet of The author examined the EPDM
replace seriously deteriorated deck and steel deck had been replaced, with more membrane for evidence of factory defi-
to coat moderately corroded deck with replacement anticipated, the author was ciencies or deficient field seams or
a high-solids epoxy coating system. requested to review the situation and flashings. No serious problems were
After this preparation, a new roof sys- provide recommendations. found that could explain the extent of
tem would be installed. observed corrosion.
The author extracted numerous test
During the demolition the author cuts from the roof to observe the deck- The owner reported that prior to
found the IR and interior survey tech- ing: several were taken in storm-dam- Hugo, he had only one persistent leak
niques assisted in predicting areas of aged wet areas; most were in areas not in the store. There was nothing in the
corrosion in some areas but were of known to have been damaged during history of the roof to indicate that the
little value in most others. Although Hugo. deck was rusting except for some or-
the process gave the author an indica- ange stains observed on the end walls
tion of the problem and assisted in The results were of great concern. immediately after Hugo. Despite the
developing a basic magnitude of the Out of nine test cuts, eight had disappointing results on Project No. 1
extent of the corrosion, it was not as moderate-to-full penetration rusting of with non-destructive testing to locate
dependable as he would have hoped.

The final area of demolition was


12,250 square feet or 23 percent of the
project, compared to an initial estimate
of 13 percent. The area coated with
epoxy was 6,250 square feet or 12 per-
cent.

The corrosion patterns were


unpredictable. The author found the
severity of corrosion could change
from board to board. In this case, total
removal of the roof to the deck expo- Figure 3: Project No. 2 Cross Section
sure was essential in order to accu-
rately assess the extent of corrosion
and the need for replacement or field the decking. The flange was typically corroded decking, procedures for infra-
coating. There does not appear to be a much more severely corroded than the red, nuclear and capacitance testing
viable non-destructive technique that webs or flute. In the most severe areas were reviewed.
would have differentiated accurately all surfaces of the deck were seriously
the "good" deck from the "bad" in corroded. Because the roof system was bal-
such a situation. lasted EPDM, the author ruled out the

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 4 of 10
IR camera and capacitance meter. The observe the installation of the new The author concluded there were
moisture readings from a nuclear gauge roof. A total of 14,900 square feet, or four "sources" of the water that en-
were questionable, because the author 17 percent of the roof deck, was re- tered the system and initiated the
got low readings on all test-cut loca- placed and 52,094 square feet or 60 corrosion. In order of significance they
tions except one where the insulation percent, was coated with epoxy. Prior are:
was found to contain 1,980 percent to coating, the deck was carefully pre-
MCBW. Here the nuclear count was 19, pared. The use of epoxy requires appli- ! Condensation on the bottom of the
which is not a high reading for 2 inches cators who are trained in the proper membrane.
of insulation that was found to be techniques for this type of coating.
soaking wet. ! Condensation near the deck-to-insu-
The demolition daily exposed ran- lation interface.
In other test cuts where rust was dom, and extensive, corrosion of the
observed, the foam was found to con- decking that the author would not have ! Condensation at insulation joints.
tain only 3.8 percent to 11.4 percent known about if the roof system had not
MCBW and nuclear counts were all remo ved. To have followed a course ! Leaks prior to Hugo.
only four. Therefore, for ballasted other than total removal of the roof
EPDM, non-destructive testing was system would have been a very seri- ! Leaks
~ from Hugo damage.
concluded to be of little to no value in ous mistake. Total removal is required
correlating wet insulation to areas of to assure all corroded deck is removed. It was observed by the author that
rusted deck. the corrosion was more severe at damp
During the course of the retrofit, areas compared to very wet areas. The
Visually surveying the bottom side the author made several observations. acidity of the insulation was tested in
of the deck was misleading too, be- different locations using a pH meter,
cause only about 5 percent of the deck The foam was heavily damaged and was found to range from 4.3 pH in
exhibited visible pitting. In these areas, from compression failure. The author the very wet insulation to a highly
the moisture contents were found to be believes most of this occurred during acidic 1.6 pH in the damp insulation.
about 1,980 percent. the placement of the ballast. Obvi- (The pH scale is from 0-14, with 0 being
ously, there was also some damage acidic, 7 being neutral and 14 being
An orange stain was visible at nu- done during the repairs. The insulation caustic.)
merous flute lines at end laps in the did not display any instability prob-
deck and the exterior end walls. This lems; i.e., there was no shrinkage. This finding initially seems contrary
orange stain was found to be rust that There were some areas where perlite to logic, until you consider that the
had been carried to the flutes of the was in contact with the deck adjacent concentration of an acid is diluted as
deck and flowed in the building after to phenolic foam in these locations. In more water is added to it, and thus its
Hugo. The staining was not found to these locations there was no rust under pH is closer to neutral. Therefore,
be a reliable indicator of rusted deck- the perlite but there was severe rust acidity can be affected by the quantity
ing. only inches away under the phenolic. of water present, but perhaps not in
*[ W E ALSO NOTICE LESS CORROSION ways you might expect.
The author saw no option but to UNDER LOCATIONS WHERE TAPERED EDGE
recommend again complete removal of STRIPS WERE INSTALLED OVER THE PHE- PROJECT NO. 3
the roof system to the deck, for two NOLIC FOAM. IN THESE CASES , CORRO-
reasons:We had to visually inspect the SION STARTED AT THE TOE OF THE TA- The author was recently requested
top surface of the deck for rust, and we PERED EDGE STRIP . to consult on a third project that exhib-
were concerned that the continued ited corrosion of the metal deck. This
presence of wet insulation could pose IN ASSESSING THE SOURCE OF THE roof was about four years old and the
potential future corrosion problems. EXTENSIVE CORROSION ON THIS PROJECT deck had a G-60 galvanized coating.
WE AGAIN CONCLUDED THE PHENOLIC The insulation consisted of two layers
During the demolition, there was a FOAM WAS THE CULPRIT . THE BASIC of 1-inch phenolic foam covered by
full-time observer on site to collect EQUATION FOR CORROSION WAS : half-inch wood-fiber board. As shown
samples, to log and map the demol- PHENOLIC FOAM + WATER = in Table C, the membrane was a CSPE
ished areas, to take photographs and to CORROSION ]* mechanically attached system.

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 5 of 10
Leaks were reported soon after half of the board and 33 percent the top flange and was attacking the
occupancy. There were numerous defi- MCBW in the lower half. The top layer base metal. No full-penetration rust
ciencies in the installation of the mem- of the phenolic contained 9.4 percent was observed. *[THE CORROSION OF
brane. The persistence of the leakage MCBW and the bottom layer 6.9 per- THE GALVANIZED COATED DECKING WAS
led to initiating a partial replacement cent MCBW. The author believes this PREDICTED FROM RESEARCH GATHERED
program with the boundaries for re- stratification of moisture is significant, ON PROJECT NO . 1, BUT NOW IT WAS
placement based in large part on the as it indicates moisture accumulation CONFIRMED.]* Although the corrosion
results of an IR survey. As the replace- related to the drive of moisture from an was not as severe as on Project 2 (both
were the same vintage foam) there was
a basis for serious concern.

*[A NOTHER INTERESTING OBSERVA-


TION WAS MADE WHERE A DECK WELD
HAD BEEN PAINTED. THIS PROTECTED
THE METAL DECK FROM CORROSION
DESPITE ABUTTING CORROSION ON THE
BARE DECK. IT APPEARED THAT A GOOD
COAT OF PAINT OUT PERFORMED THE
GALVANIZE .]*

The extent of the corrosion on this


Figure 4: Project No. 3 Roof Cross Section project is currently not known pending
possible demolition of the system
ment progressed, it became apparent area of high vapor pressure to an area (which the author has recommended).
that there was significantly more wet of low vapor pressure. Condensation The author does not anticipate there
insulation than the IR had indicated. on the bottom of the CSPE membrane will be much if any deck replacement,
Also during the replacement, a more was observed by the author at the two but does expect a significant area will
disturbing fact surfaced: The galva- cuts made in wet areas. The galvanized have to be field-coated.
nized decking was corroding under the fastener plates were slightly corroded
areas of wet insulation. but all of the screws appeared free of On this project there had been no
rust. The author found a saturated area firm determination at this time of the
around the fasteners, which may have source of the water that triggered the
The author visited the site on a been an indication of a thermal bridge corrosion, but the following are our
cold, late winter morning and extracted (short circuit) at the screws. prime candidates:
three exploratory test cuts. Two were in
known wet areas (from the IR) and one Upon exposing the galvanized ! Condensation on the bottom of the
was in an area interpreted to be dry. deck, the author's worst expectations billowed membrane during positive-
The moisture profile in the wet area is were confirmed. There was aggressive pressure operation of the HVAC sys-
interesting: corrosion on the deck, primarily at the tem.
! Wood fiber = 356 percent MCBW. joints in the lowest insulation board
! Top layer phenolic foam = 1,734 and on the deck at the fastener penetra- ! Condensation within the roof as-
percent MCBW. tion point. There was also a precipitate sembly with a fluctuating dew-point
! Bottom layer of phenolic foam = on the webs of the deck. The flange of location.
551 percent MCBW. the deck was corroded, but there was
The equilibrium moisture content of little to no corrosion on the web or the ! Top-side leakage through membrane
the wood fiber is 15 percent and of bottom of the flute. This has been a deficiencies.
phenolic is 23.4 percent, so these common phenomenon observed on the
insulation samples were very wet. three projects. ! Damage to the membrane from other
trades.
On the dry test cut, the wood fiber The corrosion had progressed all
was 51 percent MCBW in the upper the way through the G-60 coating on The author has recommended the

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 6 of 10
Metal Decks: More than a place to hang your roof
On each of the projects described in the accompanying designed to transfer wind loads (shear loads) through a
article, the metal deck was designed as more than just a diaphragm, the wind load is transferred from the windward
platform for the roof. The deck in such a design serves wall to the roof deck and out to the walls parallel to the
three primary functions: wind direction.

! Supporting vertical dead and live loads. This transfer is accomplished by attaching the deck to the
structural supports (joists or beams) with welds or screws.
! Providing a wind uplift resistance component. If the decking has either rusted away from the fasteners or
has corroded to the point it is no longer a monolithic plate
! Providing resistance to lateral (sideways) wind loads capable of distributing loads (a diaphragm), the structure
striking the building. This is referred to as diaphragm could receive extensive damage or possibly rack or
action. collapse during a wind that is within the structure's design
load, or even less.
When a steel roof deck is seriously corroded, its capacity
to resist these loads is compromised. The deck may no Note: The assessment of loss of structural integrity
longer be capable of supporting snow loads, the dead should be made by a structural engineer who is knowl-
load of the roof system, foot traffic, etc. The wind resis- edgeable regarding deck corrosion and its effects.
tance for an adhered system and a mechanically attached
system are provide by screwing the insulation or mem- As a critical component of such structures, metal roof
brane to the metal deck. decks are anticipated in the design to serve the life of the
building - which in most cases is 50, 75, or even 100 years.
If, for example, 50 percent of the deck has turned to A building owner may expect to have to replace localized
*[CRUSTED]* rust, the uplift resistance of its attachment to areas of rusted deck when he replaces his roof, but should
joists or beams - and the pull-out resistance of insulation not expect to replace his entire roof deck, or large portions
fasteners - is reduced proportionately to the reduction of of it when doing so.
metal thickness. When wind strikes a building that was

inclusion of an air retarder or a change of INVETECH, Houston, Texas in ana-


to a fully adhered system in the re- lyzing the chemical reactions related to "It is considered axiomatic that
placement to reduce or eliminate any the corrosion process. Product litera- water will find its way into a roofing
problems with the HVAC positive pres- ture and his research indicated that the system, either due to leaks or to con-
sure. phenolic foam contained an aromatic densation. Water absorption by the
sulfonic acid as a catalyst used to cre- phenolic foam can be extremely high.
Chemical Analysis ate the cells of the foam. The water dissolves the sulfonic acid,
forming an acidic environment which
What is going on in the roof sys- Slater found that the deck primer contacts the roof decking. The organic
tem to propagate such aggressive was a purely organic coating. Chloride materials used to coat the deck (primer)
corrosion in such a short time? The was present in the corrosion product and protect from atmospheric corrosion
common denominators in these pro- with 0.001 percent to 0.01 percent by are unable to withstand these acidic
jects were phenolic foam insulation in weight. Sulfur content, on the other environments and quickly break down.
contact with a metal deck and the pres- hand, was found to be in the range of
ence of water *[MOISTURE]*. 0.01 percent to 0.24 percent by weight Accelerated corrosion of the bare
(i.e., from 10 to 200 times more sulfur steel ensues. While it is acknowledged
During the author's investigation of than chloride). that the available quantity of acidity in
Projects 1 and 2 he worked with a a given volume of the foam is limited,
metallurgist/chemist, Dr. John E. Slater Slater's report stated: and thus cannot by itself sustain accel-

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 7 of 10
erated corrosion indefinitely, the salts OF THE ROOF ON PROJECT NO. 1 WHICH COMPOSITION OF PHENOLIC FOAM (NOT
produced during the initial corrosion WAS EIGHT YEARS OLD AT REPLACEMENT THE FACERS OR AN INCREASE IN THE
process will remain within the corro- AND ON PROJECT NO. 2 WHICH WAS TWO COMPRESSIVE STRENGTH), WE HAVE NO
sion product (rust), rendering its con- YEARS OLD, THE ACTUAL CORROSION ASSURANCE THAT THE PRESENCE OF
ductivity high. Thus, continued high RATE WAS 3.5 TO 14 MILS PER YEAR PHENOLIC FOAM ON A PRIMED OR GAL-
corrosion rates occur in the presence of (0.0295" ÷ 8 YEARS = 3.6 MILS PER YEAR VANIZED DECK WILL NOT CORRODE THE
moisture trapped in the foam and atmo- AND 0.0295" ÷ 2 YEARS = 15 MILS PER DECK.]*
spheric oxygen. The sulfonic acids YEAR ). THIS IS HARDLY AN ACCEPTABLE
released by moisture from the phenolic CONDITION.]* The author believes that his inves-
foam therefore act as both an initiator tigation on these projects has raised
of corrosion and as a catalyst for fur- Recommendations certain points that require further
ther corrosion." study and discussion within the roof-
What can be done to avoid or mini- ing industry.
Regarding the use of galvanized mize deck corrosion under the condi-
decking, Slater's report states: tions described in these case studies? First, more designers and contrac-
The author suggests these as possible tors should be aware that phenolic
"It has been suggested that galva- preventative steps: foam insulation is vapor-permeable
nized decking is more resistant to at- and therefore capable of absorbing
mospheric corrosion and hence more *[! ISOLATE PHENOLIC FOAM FROM water. This has ramifications relative
resistant to the type of attack found on METAL DECK WITH A SEPARATOR to: material storage (that is, stored in a
the (investigated) roof. The corrosion BOARD place with low humidity); possible in-
rate test data clearly indicate that the clusion of a vapor retarder in situa-
galvanizing cannot be relied upon to ! INSTALL A " PERFECT " ROOF THAT tions not normally requiring one; ap-
protect against the sulfonic acid attack. HAS NOT NOR WILL LEAK DURING ITS plication of this insulation during hot,
Furthermore, the sulfonates can be LIFE (NOT A PRACTICAL humid weather, and design consider-
anticipated to catalyze atmospheric EXPECTATION)]* ations due to the additional weight of
corrosion of the galvanized layer which absorbed moisture.
remains after the initial acid attack, in ! Use epoxy coating or equivalent
the same way as for the ungalvanized applied to prime-coated metal The industry also needs a better
steel. *[THUS GALVANIZING IS NOT CON- deck. method to determine the absorption
SIDERED TO BE THE PROTECTIVE COAT - potential of phenolic insulation.
ING OF CHOICE IN THIS APPLICATION.]*" ! Use G-90 galvanize coating, not G- ASTM C1126 is the "Standard Specifi-
*[TESTS WHICH WERE CONDUCTED 60, for longer resistance to corro- cation for Faced or Unfaced Rigid Cel-
AND REPORTED BY A PHENOLIC FOAM sion. lular Phenolic Thermal Insulation."
MANUFACTURER STATED THE FOLLOW - This standard refers to ASTM C209,
ING: *[! H A V E THE INSULATION which includes a moisture absorption
MANUFACTURER WARRANT DAM- test. This test does not adequately
"UNDER THE MOST SEVERE POSSIBLE AGE TO DECK FROM CORROSION model the conditions to which a roof
TEST EXPOSURES , I.E. UNPROTECTED FOR "X" YEARS insulation will be exposed. The test
STEEL, ACCELERATED TEST CONDITIONS , measures the percentage water ab-
AND CONTINUOUSLY WET ( PHENOLIC ! USE A ROOF SYSTEM THAT IS EASY sorption increase by volume with a
FOAM) INSULATION, MEASURED CORRO- TO VISUALLY INSPECT FOR PHYSI- two-hour immersion. For one manu-
SION RATES WERE BELOW 2 (1.8) MILS CAL DAMAGE facturer, this was 2.0 percent by
PER YEAR AFTER 200 DAYS ......" volume. With a density of the foam of
! USE STAINLESS STEEL SCREWS 2.7 pounds per cubic foot that is 46
THE THICKNESS OF 22 GAUGE PRIMED percent MCBW. Further, immersion
METAL DECKING IS ABOUT 0.0295" ! CONSIDER USING ANOTHER INSULA- does not model a vapor drive to which
THICK. IF WE CAN EXPECT CORROSION TO TION in-service insulation is exposed. *[THE
REMOVE 2 MILS (0.002") PER YEAR , THE INDUSTRY NEEDS A BETTER TEST.]*
ANTICIPATED DECK LIFE WILL BE ONLY 15 UNLESS THERE HAS BEEN OR WILL
YEARS ! BASED ON OUR OBSERVATIONS BE A MAJOR CHANGE IN THE CHEMICAL *[This leads us then to a second

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 8 of 10
point. The industry must acknowledge try needs to know what the deck MCBW.) NEXT REMOVE THE FOAM
that phenolic foam is not "closed-cell" manufactures can and will provide to FROM THE CONTAINER. PLACE A FEW
relative t o moisture. When the prod- the end user. *[THE STEEL DECK INSTI- SMALL STRIPS OF METAL DECKING IN
uct can absorb 46% in two hours and TUTE RECOMMENDS FIELD PAINTING THE THE SOAK WATER AND MONITOR FOR A
ultimately nearly 2,000% MCBW, it DECK BECAUSE THEY SAY THE FACTORY WEEK OR SO. WE OBSERVED RUST ON
ain't closed-cell! This is important dur- PRIMER IS " AN IMPERMANENT AND THE EXPOSED CUT EDGES OF THE STRIPS
ing the design stage. Structural engi- PROVISIONAL COATING." W OULD IT BE IN ABOUT 48 HOURS. W HEN WE
neers base their designs and size POSSIBLE OR PRACTICAL FOR THE MET - CHECKED THE PH AFTER THE METAL
structural members on the assumption AL DECK SUPPLIER TO FURNISH A PRE- HAD BEEN IN THE SOLUTION FOR A
that the insulation is dry and will re- PAINTED PRODUCT WITH MORE THAN A WEEK, WE FOUND THE PH HAD RISEN
main dry, not that someday itweigh
may 20 " PROVISIONAL COATING"? IF NOT , THEN DUE TO THE CHEMICAL REACTION TAK-
times more than they were told it THE CHOICES ARE: A G-90 GALVANIZED ING PACE IN THE CORROSION PROCESS.
would. Mechanical engineers base COATING; AN ALUMINUM ZINC ALLOY
their heating and cooling loads on dry COATING; A FIELD APPLIED PAINT OR M ORE SOPHISTICATED TESTS CAN
material with a stated "R" value, not a COATING; OR CHANGING TO ANOTHER BE RUN BUT WE HAVE FOUND THESE
depreciated, lower value as a result of DECKING SYSTEM. SIMPLE TESTS ARE QUITE INFORMATIVE.
moisture absorption. These are impor- YOU MAY WANT TO RUN SIMILAR TEST
tant liabilities that must considered.]* WHAT CAN YOU DO? ON OTHER INSULATIONS FOR YOUR
INFORMATION.
In addition, the matter of positive DURING OUR INVESTIGATIONS WE
pressure from an HVAC system billow- HAVE CONDUCTED A SIMPLE TEST TO FOR THOSE WHO HAVE IN -PLACE
ing a single-ply membrane is a situa- MONITOR PHENOLIC FOAM CORROSION PHENOLIC FOAM, IT MAY BE PRUDENT TO
tion that is not receiving adequate PROPERTIES AND MOISTURE ABSORP- DETERMINE IF A PROBLEM CURRENTLY
attention from manufactures, design- TION. YOU MAY WANT TO PERFORM EXISTS AND THEN RECHECK OCCASION-
ers, or contractors. The requirement YOUR OWN TESTS TO SEE WHAT HAP - ALLY. TAKE A FEW EXPLORATORY TEST
for roof pressure relief valves, air re- PENS WHEN A 2" CUBE OF DRY PHENOLIC CUTS AND SEE IF THERE IS A PROBLEM.
tarders or fully adhered systems needs FOAM IS PLACED IN ABOUT 150 ML OF ON OUR PROJECT NO. 1, 35% OF THE
further study and consideration in the DISTILLED WATER . EVERY 24 HOUR PE- DECK HAD TO BE EITHER REPLACED OR
design process. With the exception of RIOD MEASURE THE PH WITH INDICATOR COATED AFTER ONLY EIGHT YEARS. ON
only a few manufacturers, air infiltra- PAPER OR A P H METER . A LSO PROJECT NO. 2, 77% HAD TO BE RE-
tion remains a non-issue. ACCURATELY WEIGH THE SAMPLES . W E PLACED OR COATED AFTER ONLY TWO
FOUND THE PH DROPPED FROM 6.6 TO YEAR S . TIME MAY BE YOUR ENEMY IF
Regardless of the type of insula- 3.3 IN ONLY 48 HOURS. THE MCBW IN- THE DECK IS CORRODING. DAMAGE MAY
tion to be used, there needs to be a CREASED TO OVER 200% IN THE SAME BE MINIMIZED IF CAUGHT EARLY
thorough, industry wide review of TIME PERIOD. (THAT IS FIVE TIMES THE ENOUGH.]*
metal decks and corrosion. The indus- TWO-HOUR IMMERSION VALUE OF 47%

BY THE TIME RUST IS OBSERVED ON THE BOTTOM SIDE OF THE DECK IT IS TOO LATE. THE CORROSION
HAS PROGRESSED TO A POINT OF COMPONENT FAILURE.

* READER NOTE: The text enclosed by asterisks and brackets *[ IN SMALL CAPS ]* was in the author’s original manu-
script, but omitted in the published article by the Editor.

References

1. Professional Roofing: March 1991, NRCA News, Page 63, Tech Transfer, Page 74; April 1991, Tech Transfer, Page 86.
2. "Corrosion Protection For New Steel Roof Decks," NRCA Bulletin 15-91, May 1991.
3. Cash, C. G., "Moisture and Built-Up Roofing," Proceedings of the Second International Symposium on Roofing Tech-
nology, NRCA, 1985, Pages 416-427.
About the Author: Richard P. Canon, a Registered Roof Consultant (RRC) and Professional Engineer (PE), is owner of Canon
Consulting & Engineering Co.,Inc. 1617 John B. White Sr., Blvd., P. O. Box 17007, Spartanburg, SC 29301-0101, (864) 574-6500.

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 9 of 10
ADDITIONAL PHOTOGRAPHS OF MISCELLANEOUS PROJECTS

Figure 5: Project 1, view of wet insulation. Figure 6: Project 1 interior view of corrosion

File: C:\DICK'S FILES\Phenolic\PHENOLIC ARTICLE- 1992.wpd [February 12, 2001(10:17PM)]

Professional Roofing (Reprinted with Permission. See Note to Reader at end of Article) August 1992
Compliments of Canon Consulting & Engineering Co., Inc.
Spartanburg, South Carolina
(864) 574-6500 Page 10 of 10

You might also like