Local: (800) 806.6051
International: (253) 473.5000
Request a quote


By C.L. Williamson, Z.L. Iams

Download PDF


Rigid polyurethane foam used as Over-pack in transport containers, utilizes a variety of mechanisms to mitigate the thermal assault of a “regulatory burn.”   Polymer specific-heat and foam k-factor are of limited usefulness in predicting payload protection. Properly formulated rigid polyurethane foam, as a consequence of its pyrolysis can provide additional safeguards. These ablative mechanisms are effective even when the foam has been crushed or fractured as a result of trauma.  The dissociative transitions from polymer to gas and char, and the resulting gas- transport  of  heat  from  inside  the  package  out  into  the  environment,  are  efficient  thermal mitigators.  Also important is the in-situ production of an intumescent, insulative, carbonaceous char that confers thermal protection even when a package’s outer steel skin has been breached.

In our test program, 19 liter (Five gallon) steel pails were filled with protective foams and then exposed on one face to the flame of an “Oil Burner” as described in the US Federal Aviation Administration (FAA) “Aircraft Materials Fire Test Handbook”.   When burning #2 diesel at a nominal rate of 18.5 pounds (8.39 kg)/hr, the burner generates a dirty, high emissivity flame that impinges the pail face with the thermal intensity of a full scale pool-fire environment.   Results of these tests, along with Thermo Gravimetric Analysis (TGA) and Differential Scanning Calorimetry (DSC) analysis on the subject foams are reported, and their relevance to full size packages and pool-fires are discussed.


Packaging design for the safe transport of nuclear materials would be far simpler if thermal protection was the only requirement.  There are many insulative inorganic materials adequate for this purpose.  However, transportation packages are also called upon to protect their “payloads” from kinetic accidents1  (and in today’s world even explosive blasts2) before thermal exposure. Hence the package must first preclude inadvertent release resulting from loss of containment and preserve package insulative integrity for subsequent fire blocking.

General Plastics has installed rigid polyurethane foam in transportation over-packs and impact limiters for this dual purpose since 1971.  Foam Impact energy absorption is reasonably well understood,  but  deep  insight  into  its  thermal  protective  mechanisms  has  remained  elusive. Whether or not a package survives an IAEA fire3 depends upon many variables, including those of both package design and materials.  Our research objective was to isolate and quantify foam properties in order to provide enhanced protective characteristics.


To begin, we hypothesized likely mechanisms for the thermal protection imparted by LAST-A- FOAM® FR-3700 series rigid polyurethane foams.4 These mechanisms are both physical and chemical and result from foam characteristics including specific heat, thermal conductivity (k- factor) and ablative characteristics such as pyrolysis gas heat transfer, evolved combustion products, enthalpy of anoxic pyrolysis, conditions for char formation and char quality.


Though fires are notoriously difficult to scale, we sought to design a medium-scale test that would yield engineering data relevant to full size fires and packages.  Regulatory fires impinge on all sides of a package, and of course the area/volume ratios and the payload thermal mass of designs vary greatly.

To better understand foam decomposition, we reviewed Thermogravimetric Analysis (TGA) on the foam both tested in air and Nitrogen, as well as Differential Scanning Calorimetry (DSC) tested in Nitrogen. For our medium scale testing, we elected to utilize a two dimension test methodology by subjecting only one surface of our test article to thermal assault.  For pool fire representation, we chose the “Oil Burner”, as called out in commercial aircraft “Joint Airworthiness Regulations” FAR/JAR 25.853, which is used for testing the flammability of seating and the fire hardening of cargo wall liners.  This burner is completely described in the US FAA’s “Fire Test Handbook”5.

The 280 mm wide x 152 mm high burner exit cone, when adjusted to produce a ~1050°C temperature at a distance of 100 mm from the burner exit cone, results in a radiant heat flux of 19.6 W/cm2 on our calorimeter6. This is somewhat higher than that for pool fire heat flux referenced7 for Diesel, JP4 or Gasoline (Petrol) at 13.0 W/cm2, though less than that for Butane at 22.5, Propane at 25.0 or LNG at 26.5 W/cm2.


Differential Scanning Calorimetry (DSC) was performed on FR-3710 foam (10 lbs/ft3 density before crushing) in an open pan, using a 50 cc/min N2 purge.  The high purge rate helped separate initial endotherm from the anoxic exotherm resulting from the later recombination of molecular fragments into lower energy oligomers.  These oligomers “are transported into the gas phase, and sometimes referred to as tar.”8    If vented (not condensing in cooler parts of an overpack) these.

In practice, the anoxic exotherm is not self sustaining, and the FR-3710 foam self-extinguishes when the external heat flux ceases, and the unit cools. TGA’s, figures- 1 and 2, were conducted on FR-3706 foam in both air and Nitrogen, disclosing the first (main) decomposition temperatures of 338 and 354 °C respectively.  Most interesting is that throughout the broad temperature range of approximately 340 through 650 °C, the weight remaining in air is greater than in N2- an indication that Oxygen significantly enhances char formation!


Four, 19 liter (5 gal.) steel pails, ~30 cm in diameter (tapering to 28 cm diameter at the rear) x~33 cm deep, were filled with different densities of FR-3700 rigid urethane foam. Nominal foam densities were 0.108, 0.174, 0.305 and 0.413 g/cm3 (6.74, 10.86, 19.03, and 25.77 lbs/ft3).   3.18 mm thick, stainless steel lids were then welded onto the pails.  Each lid was vented with a 23.8 mm hole in its center.  Test pails were then positioned as shown in figure-3, with the burner cone 100 mm from the lid.

Figure -3 “Oil Burner” and test pail set-up

When testing, the burner is turned on and the test pail lid exposed to the ~1080 °C burner flame. Tests begin when the pail lid/hot-face temperature reaches 801°C and ends 30 minutes later (though temperature recording continues until all thermocouples pass their peak temperatures). Hotface temperatures during tests average around 950 °C as determined using 1.57mm, ungrounded, stainless-steel sheathed, type-k thermocouples in metal-to-metal contact with the rear of the hotface.

After the burner is turned off, the test pail is allowed to continue burning and then cool, while still positioned on the stand. The burn distance from the hotface is a readily discernable as a sharp transition between foam and char; this is as might be expected from the TGA curves (figures 1 and 2).   We know from the TGA that this transition is centered at 354 °C (in N2).   At the end of the test, after all components have cooled, the pails are weighed, lids removed, char weighed and examined. The recession distance (burn distance) from the hot-face to the un-degraded foam is measured and recorded.


Table  I  shows  the  foam  recession  distance-  the  depth  of  foam  consumed  in  a  30  minute regulatory exposure vs. the foam density.   In all cases, varying amounts of carbonaceous char were  found  in  the  space  where  foam  had  been consumed.  The  regression  line  that  can  be calculated from the data in table 1 indicates that foam effectiveness increases with increasing foam densitythough at a decreasing rate.  A 0.10 g/cm3 density foam recesses about 10.7 cm. Doubling the foam density to 0.20 g/cm3 does not cut the foam recession distance in half (to 5 cm).  Rather, it reduces it by only 3.5 cm (a 7.2 cm recession distance).   This relationship should be useful for package designers performing their initial calculations.

Table I Effects of foam Density on foam thermal effectiveness

Density, g/cm3










Initial Weight, g 5116 6576 9543 11228
Final Weight,  g 4690 5940 8799 10563
Wt. Loss g 426 636 744 665
Extinguish Time, minutes 5:25 8:15 9:24 9:11
Recession Distance, cm 10.2 8.00 4.7 3.8


Table II illustrates the effect of exposure time on a single foam density.  Six tests were conducted using six identical pails foamed with the same 0.181 g/cm3  FR-3700 series foam.    Exposure times were 5, 10, 15, 20, 25, and 31 minutes.  Results calculated from Table II show that foam recession distance increases at a decreasing rate with respect to time.

Table II Recession Distance as a function of exposure time


Exposure Time, minutes 5 10 15 20 25 31
Density g/cm3














Initial Weight, g 6655 6761 6778 6695 6649 6680
Final Weight,  g 6380 6382 6259 6082 5952 5945
Wt. Loss, g 275 379 519 613 698 735
Extinguish Time, minutes 5:31 6:32 4:34 4:20 6:02 6:53
Recession Distance, cm 2.54 4.13 5.59 6.16 7.62 8.13

There is likely more than one mechanism at work.  Firstly, as the foam recesses away from the hotface, heat flux to the un-degraded foam is reduced as the inverse-square of the recession distance, a minor effect at these distances.  What we believe the regression illustrates most clearly is the insulative effect of carbonaceous char.  As the un-degraded foam retreats, thicker char layer is able to mitigate more of the heat flux.


A fire impinging on a damaged container is a difficult scenario to model.  Drop damage can penetrate or rip the outer metal skin.  Additionally, impacted foam is likely to be cracked or crushed. To reproducibly simulate severely damaged foam (or other materials) we elected to test pails filled with randomly packed, 2.54 cm test cubes as shown in Figure 4.  The packing fraction for randomly packed cubes with little orientational ordering, or registration between cubes, is about 0.5829.  We discovered, that by just dropping foam cubes into an empty pail, and then spreading them out for lid installation, we achieved packing fractions in the range of 0.56 to 0.61, and with practice, the range could likely be reduced to between 0.58 and 0.60.

Figure-4 Random packed foam cubes

By cutting three 46mm holes in the lower pail hot-face, and a similar vent hole at the upper rear of the pail (as oriented for testing) we create a chimney effect that pulls hot combustion gasses and outside air through the interstices between the foam cubes.  These gas pathways simulate a damaged unit, where foam has been punctured and/or fractured.  When tested in this manner, combustible organic materials lacking intumescent properties may exhibit difficulty self- extinguishing.  Four test pails of randomly packed cubes were prepared:

Tests-400519-1, 040521-1, and 04729-1 were conducted on standard and modified FR-3709 foams; Test-040728-1 with FR-3718 at a density of 0.296 g/cm3 was conducted as a direct comparison with Test 040630-1

Test 040630-1 was conducted on GP’s FR-10112, a rigid, Isocyanurate foam, Isocyanurate foam possess better high temperature performance than polyurethane, but lack intumescent properties. Test-040723-1  was  conducted  on  uncoated,  high  density,  ASTM  C  208-95,  Type-2  cane fiberboard.

Test 041123-1 was conducted on nuclear grade Balsa wood. Test 041124-1 was conducted on nuclear grade Redwood The results of these tests are shown in Table III

Table III Results from random cube experiment

Test Number 040519-1 040521-1 040729-1 040728-1 040630-1 040723-1 041123-1 041124-1
Material FR-37091 FR-37092 FR-37093 FR-3718 FR-10112 Fiberboard Balsa Redwood
Density, g/cm3




0.147 (9.17)


0.134 (8.36)


0.159 (9.92)


0.296 (18.47)


0.187 (11.67)


0.295 (18.41)


0.146 (8.97)


0.347 (21.69)



0.567 0.608 0.573 0.611 0.600 0.582 0.593 0.585
Void, % 43.3 39.2 42.7 38.9 40.0 41.8 46.8 41.5

Density g/cm3 (lb/ft3)


0.083 (5.18)


0.082 (5.12)


0.091 (5.68)


0.181 (11.29)


0.112 (6.99)


0.172 (10.73)


0.078 (4.77)


0.203 (12.69)

Initial Weight,


1808 1633 1970 3914 2421 3703 1730 4123
Final Weight, g 976 1077 1316 3253 1006 752 189 608
Weight Loss,


832 556 654 661 1415 2951 1541 3515
Weight Loss,


46.2 34.0 33.2 16.9 58.4 79.7 89.1 85.2

Time, minutes

21:20 10:50 9:34 3:40 3 Hrs+ * ~5 Hrs 3:28 Hrs 7:34 Hrs

* test terminated by closing off all vents at 3 Hrs.

The values of merit in this test series were (1) weight loss and (2) time to extinguishment. As expected, the modified FR-37093, formulated for increased intumescences performed best (for its density), followed by standard FR-37092.  FR-37091 was formulated for reduced intumescence. The isocyanurate foam (FR-10112) would likely have been completely consumed, except that the test was extinguished blocking the pail vents after three hours.

The Isocyanurate, fiberboard, balsa and Redwood tests all ended when the fire was quenched or the available fuel was depleted.  None of these overpack materials were self-extinguishing; they all require outside intervention or fuel consumption to extinguish the flame. Interestingly, ash cubes  from  the  non-intumescent  materials  tended  to  remain  “cubical”,  though  they  become smaller as they are consumed where the intumescing cubes swell and close off ventilation paths.


Thermal protection using organic materials is certainly complex as these tests have illustrated. The “Oil Burner” methodology we describe, requires minimal instrumentation, is economical (inexpensive pails are readily available), can be used for materials characterization, package design, and for comparing production foam batches with their original qualification tests.

Foam density, thickness, and the production of an intumescent carbonaceous char are important variables.  Our tests indicate that increasing foam density (increasing the mass loading) is always protective- even though foam thermal conductivity is greater at higher densities.    Of course, impact energy absorption, cost, weight, and in today’s world, even blast-wave mitigation2 are major foam density drivers.

In these tests, we found the anoxic, near step-function weight loss occurring at ~354 ºC (for FR-3700 foam) to be an ideal temperature indicator, yielding a sharp, degraded/non-degraded foam boundary, that preserves an accurate, maximum temperature record at the recession surface. Simulating damaged foam by using randomly packed cubes, proved to be a rather severe method, but was remarkably discriminatory, and we believe relevant to damaged containers.

Finally, the more we learned from the tests conducted the more questions we raised.  We did not investigate the effect of polymer heat of combustion, rate of heat release or oxygen index on foam performance.  Nor did we investigate the quantity of heat transferred to the “payload”- though it should not be difficult to add a heat sink or calorimeter to the foam at any particular foam depth.   Also left uninvestigated, was the use of refractory sheet materials arranged in parallel with foam, or the affect of burner (pool fire) temperature on intumescent char formation. These questions and others may be investigated at a later date


We wish to thank Kenneth Erickson of Sandia National Laboratories for his many helpful comments… and his DSC.  To Robert Sevasin for preparing the test articles, instrumentation, conducting the tests and recording the data.  And to Jamie Guenthoer, molecular biologist, and full time post-grad student at the University of Washington who was a great help with graphics, style, and formatting.

Related Resources


Core Distinctions: Choosing the Right Foam for Critical Composite Core Applications

Read More
CHT Travellers

Case Applications

Built to Last: 15 Years of Proven Success with Radioactive Materials Packaging with Columbiana Hi Tech

Read More
CHT Travellers


General Plastics’ Rick J. Brown, Ph.D. to Present Paper on Crash and Fire Protection for Nuclear Transportation Containers September 20 at PATRAM 2016

Read More
Nuclear Plant


General Plastics to Spotlight Premier Polyurethane Foams for Nuclear Container Liners at WM2016 Conference

Read More