A. C. Smith and P. S. Blanton
Westinghouse Savannah River Company
Aiken, SC 29808

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

This report has been reproduced directly from the best available copy.

Available for sale to the public, in paper, from:  U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161,  phone: (800) 553-6847,  fax: (703) 605-6900,  email:  orders@ntis.fedworld.gov   online ordering:  http://www.ntis.gov/support/ordering.htm

Available electronically at  http://www.osti.gov/bridge/

Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062,  phone: (865 ) 576-8401,  fax: (865) 576-5728,  email:  reports@adonis.osti.gov

Studies of crush performance of cane fiberboard assemblies, sometimes used in overpacks for radioactive materials packages, has shown that, under low temperature or desiccated conditions, a short duration transient stress spike may occur for specimens loaded parallel to the plane of the fiberboard. This effect has been attributed to the stiffening effect of the layers of glue bonding the fiberboard sheets together. The question of the possibility of a property of the fiberboard itself being responsible for this phenomenon has been raised. As part of the study To answer this question, specimens of fiberboard were bonded with varying amounts of glue and the resulting joint evaluated. Microscopic analysis was performed of the Celotex^â and glue layer in these specimens and in specimens cut from production packages. Available test results indicate that the Celotex^â is essentially isotropic. No indication of cellulose properties which could account for the spike phenomenon was found. The study revealed that the glue layer is in the form of islands of glue, formed in pores on the fiberboard surface, connected by thin ligaments of glue.

Fiberboard overpacks which provide thermal and shock protection for drum type radioactive material (RAM) packages are frequently fabricated by bonding sheets of cellulose fiberboard together to build-up a thick-wall cylinder which encloses the containment vessel(s). Dynamic and slow strain-rate testing of the crush performance of such cellulose fiberboard assemblies, over a wide range of temperatures and degrees of humidification, has revealed a difference in response to loading parallel to and perpendicular to the plane of the fiberboard sheets, References 1 and 2. In addition, under low temperature or desiccated conditions, a short duration transient stress spike has been observed for specimens loaded parallel to the plane of the sheets, Figures 1 and 2. Celotex^â brand cellulose fiberboard sheets were used for this study. Specimens were bonded with Elmer’s^â Carpenter’s Wood Glue.

The study of the mechanism causing the stress spike included simulation of the impact tests using an ABAQUS finite element model, Reference 3, and the Celotex^â and glue structure investigation, reported here. The analytical study showed that the spike effect could be reproduced by including the effect of the glue layers in the calculational model. A parametric study was performed on the effect of stiffness of the glue layers. The stiffness of plates or columns is a function of the modulus of elasticity and the cube of the thickness of the plate (i.e., Eh³). For the parametric study, the glue affected zone, or glue layer thickness was measured for several specimens taken from a production run of 9975 packages and found to be uniformly about 1 mm thick. The parametric study of the effect of glue layer stiffness used 1 mm for the thickness and varied the modulus of elasticity of the layer. The finite element model represents the glue layer, in an idealized manner, as a uniform, discrete sheet of material.

The results of the parametric study on layer stiffness were compared with the impact test results. The value of stiffness (combination of assumed thickness and modulus of elasticity) which yielded deformation closest to that of the test specimens was used for a 3-D analysis to determine the effect of the spike phenomenon on the safety performance of the 9975. Although the finite element model confirmed that the safety of the package, questions on the mechanism of the spike remained.

Examination of the slow strain rate results for loading parallel to the cellulose fiberboard sheet shows that, in every case, the stress rises sharply, initially, before reaching a peak on the order of 200 psi, then drops off for an extended period of buckling and crushing, Reference 2 and 4. Then, as the void is closed the stress strain curve turns rapidly upward. The initial steep rise appears to be an elastic behavior (i.e., linear). In contrast, for loading perpendicular to the plane of the Celotex^â, the specimens crushed uniformly, with slowly increasing stress until the void was closed. As the void is closed, the stress strain curve turns rapidly upward, Figure 1. The steep rise associated with closure of the void space is called "lock-up" or "densification" in the literature.

Examination of the impact testing results revealed a high, short duration stress peak (or spike) in the initial phase of crushing of several of the Celotex^â specimens, Figure 2, when the loading was parallel to the plane of the Celotex^â sheets. The test conditions under which this behavior occurred were at very low temperature (-40°C, -40°F) and, in one case, for a desiccated specimen at ambient temperature. This behavior was not observed for loading perpendicular to the sheets

Literature on the behavior of cellular materials under dynamic crushing provides insight into the mechanism leading to the spike in the present results, References 5 and 6.

Open cell behavior is dominated by the structural response of the cell walls. Crushing in such materials depends upon the geometry of the individual elements and inertia. The crushing takes place by local instabilities in deformation of the cell walls, and typically begins at the face and propagates inward. The inertia of the cell walls can modify the local quasi-static mechanism within the structure, leading to less compliant modes of failure and requiring higher loads to cause crushing. Initial peaks are common for (anisotropic) cellular materials with load applied parallel to cell walls. Examples of this are metal honeycombs and woods.

In particular, results for crushing of honeycomb structures, both in the literature and in SRS tests, show high initial peaks, followed by a uniform crushing plateau region termed "progressive buckling". Once buckling extends throughout the structure (lock-up), the stress rises very rapidly.

In summary, the initial response to loading of cellular structures is elastic, with the stress being carried (elastically) by the cell walls. As the load increases above the point where the walls become structurally unstable, they begin to buckle, with consequent reduction in load carrying ability. As the deformation continues, the buckling propagates through the material, at a fairly consistent stress level. When essentially all of the cell volume has buckled, so that all the void is closed (i.e., lock-up), the stress increases rapidly.

Under dynamic conditions, the loads in all of these regimes increase as a function of strain rate.

The results of the Celotex^â impact tests are fully consistent with the behavior described in the literature. For loading perpendicular to the plane of the sheet, the Celotex^â displays no elastic behavior above a few ten’s of psi and crushes consistently, with stress rising as lock-up is approached. The glue layers carry no load in this orientation.

The behavior of specimens loaded parallel to the planes of the Celotex^â is determined by the glue reinforced layers, which divide the material into 2-D cells. The initial crushing is determined by the glue layers which act as plates, stabilized by the weaker Celotex^â interstitial material. The buckling of these glue-reinforced-cellulose-fiber layers causes the entire specimen to buckle, if unconstrained. The buckling dominates the crushing behavior until further buckling cannot occur. Beyond this point, the Celotex^â crushes in a manner like that for perpendicular loading, rapidly attaining lock-up.

Buckling is determined by the structural characteristics of the glue-layer cells (filled with Celotex^â). The important characteristics are the thickness and spacing of the glue layers and the structural properties of the glue layers. The structural properties of the glue layers depend upon whether the glue is acting as a relatively pliable material (moderate temperatures and humidity) or as a relatively stiff, elastic material (very dry or cold conditions).

The evaluation of the response of the bonded Celotex^â assemblies from stress-strain results and examination of bulk deformation provided a consistent postulate for the presence of the stress spike found in testing. However, these results alone are unable to eliminate other effects which may be postulated. These can be summarized by the basic question: Can the spike phenomenon be the result of basic Celotex^â material properties, rather than being caused by the glue layer?

On a mechanistic level, this question may be restated as: Do local effects like crushing at the contact faces (in the manner of honey-comb structures), micro-inertia and densification of Celotex^â cells actually account for the stress spike, rather than the behavior of the larger cells established by the glue layers? Related questions are: If the effects were the result of properties of the cellulose matrix, would the behavior be consistent with observed static and dynamic crush test results? And, during the impact event, does localized crushing take place in the cellulose matrix in a crush zone that is small compared to the test cube? Finally, If the glue layer does determine behavior, what is the effect of glue layer fabrication parameters on crush strength of the assemblies?

Microscopic examination of the glue layers was undertaken to resolve these questions.

The amount of glue present in typical production 9975 packages was determined by comparing assembly weight with the weight of Celotex^â used in the assembly. The results show that around 5.4 Kg of (12 lbm) glue is required for assembly. Comparison of the thickness of separate Celotex^â sheets with bonded assemblies shows that the average added thickness due to bonding is 0.010 mm (0.004 in).

In order to evaluate the effects of the amount of glue used for bonding the Celotex^â sheets on the resulting glue layer thickness, a comparison test was performed. Three 51 mm (2 in) square specimens (each made of two squares) were prepared using varying amounts of glue.

Specimen 1 used just sufficient glue to coat the mating surfaces.

The bottom half of Specimen 2 was coated on one surface with a uniform, heavy layer of glue, completely covering the mating surface. The top piece of Celotex^â was placed in position on the glue.

For Specimen 3, both mating surfaces were coated with a uniform, heavy layer of glue, completely covering the mating surfaces. The top piece was then placed in position, glue layer to glue layer, on the bottom piece.

To insure consistent interface load, 300 grams weights were placed on each specimen for 24 hours (until the glue was dry). The weights provided uniform loading on the specimens.

The specimens were weighed before and after application of the glue and then after the glue dried. The solids fraction of the glue is 52%, so its volume decreases significantly as it dries, Reference 7.

Excess glue flowed from between the sheets for both specimens 2 and 3 when the weight was applied. After the glue dried, the puddles of excess glue were cut away and the specimens were weighed. The weight of dry glue retained by the specimens was:

The specimens were cut in two using a band saw, and the apparent thickness of the glue affected zone was determined at several locations, using a vernier caliper, for each. The average apparent or nominal thicknesses of the glue layers were:

The glue affected zone appear to be of consistent thickness across the specimens, Figures 3, 4 and 5.

Finally, the overall thickness of each of the specimens was measured to determine the increase in assembly thickness caused by the presence of the glue layer. The thickness measurements, taken across the cut faces of the specimens, were:

The increasing change in specimen thickness with quantity of glue applied is attributed to two processes. First, as the Celotex^â absorbs moisture from the glue it becomes locally soft. Second, as the glue drys, it shrinks slightly. It is worthy of note that the average thickness for 9975 production assemblies lies between the values for Specimens 1 and 2.

The glue layers were also examined under a microscope, Figures 6, 7, and 8. The glue layer was readily apparent in most regions along the mating line for Specimen 1, although there were few locations where a layer of pure glue was present. There was little indication of permeation into the surface, due to capillary action. Specimen 2 was similar, but the glue layer was more obvious and displayed fewer regions where it was not readily apparent. Specimen 3 was similar in appearance, but the glue layer was distinctly visible and continuous across the specimen. For comparison, the glue layer in a typical specimen taken from a production 9975 is shown in Figure 9, for reference.

Examination of the surface of the Celotex^â sheets (i.e, surfaces representative of those to which the glue was applied) revealed that the surfaces were highly porous, Figures 10 and 11. (The porosity of the Celotex^â is 0.67.) There was also a lot of surface "fuzz" apparent. The glue clearly engulfs the surface fuzz and flows into the surface voids. Beyond surface porosity, there is no indication of glue being drawn by capillary action into the Celotex^â structure. It appears that the thickness of the apparent, or nominal glue layer is established primarily by the contact of the asperities on the mating surfaces. This is supported by the consistency of overall thickness measurements, given above. The glue fills the space between the "high points" and fills (at least partially) the voids in the porous surfaces. Lesser amounts of glue applied to the mating surfaces fill the voids relatively less consistently, whereas the heavy coatings of glue filled the available voids more consistently and to a greater depth. Although the cellulose fiber between the glue filled voids act in parallel with the glue layer, the strength of the cellulose structure is small compared to that of the glue, Figure 12.

The impact face of a specimen loaded parallel to the plane of the Celotex^â sheets was examined under the microscope to determine if any crushing of the cellulose fibers had occurred. It was found that while the specimen experienced gross crushing, the cellulose fibers themselves were not crushed, Figure 13.

The microscopic examination of the glue layers enabled resolution of the questions posed above.

Celotex^â is composed of cellulose fibers arranged essentially randomly. The cellulose fibers are cellular themselves, but the fiberboard formed from the fibers is not cellular. Under the microscope, the material resembles a close-up view of a straw bale. The fibers are sufficiently random in orientation that little difference would be expected for loading perpendicular to the plane of the sheet compared to loading parallel to the sheet.

The expectation that the material is essentially isotropic is supported by the laterally constrained slow strain-rate test results given in Reference 2. In the laterally constrained test, specimens were placed in a fixture, so that gross buckling could not occur, and loaded parallel to the plane of the sheets. With bulk buckling inhibited, the Celotex^â stress-strain performance was comparable to that for loading perpendicular to the sheets.

Because of the low modulus of elasticity of Celotex^â fiberboard, the stiffness (Eh³) of specimens loaded parallel to the plane of the sheets is very low and buckling of the specimen would be expected, rather than direct crushing. This is confirmed by the behavior of the glued assemblies which, inspite of the stiffening provided by the glue layers, consistently fail by buckling unless artificially constrained, as in the laterally constrained test.

Crushing of the structure does not result in crushing of the cellulose fibers. In addition it produces no local regions of crushing like that seen in crushing honey-comb structures, Figure 13. In the absence of the larger scale cells formed by the glue layers, the porous fiber structure crushes consistently and uniformly, like specimens loaded perpendicular to the plane of the sheets, References 1 and 2.

In the crush process, the porous matrix exhibits micro-inertial effects, causing the impact test specimens to respond more stiffly than the corresponding slow strain rate results. In the absence of cell walls oriented parallel to the direction of the load, and carrying the load as elastic members, the micro-inertial effects do not cause a stress spike, like that reported in the literature for cellular materials with cell walls oriented parallel to the direction of the load.

As noted earlier, when the void is closed by the crush process, the stress rises steeply. This is the phenomenon called lock-up or densification. For cellular structures, this occurs when buckling of the full height of the cell walls is complete (hence the void is reduced to minimal values). For the porous Celotex^â material, this occurs when the void is closed to the extent that the cellulose fibers themselves begin to be crushed. In the present study, this behavior is displayed in both impact and slow strain rate results for loading parallel and perpendicular to the plane of the sheets when the strain exceeds 50%.

The evaluation of the effects of fabrication parameters on the glue layer indicate that the glue layer thickness (and consequently its stiffness) is not highly sensitive to the fabrication process. The comparison of weight gain shows that there is more glue retained in the case of heaviest glue application, Specimen 3. This is not reflected in the measurements of apparent (or nominal) thickness of glue layer, however. The microscopic examination of the specimens shows that the interfacial layer is more continuous and slightly thicker, for Specimen 3. It appears that more of the surface pores are filled with glue for this case.

The glue layer appears to take the form of a film of fiber and glue that varies in thickness from location to location. That is, the thin membrane connects a random array of closely spaced "islands" of thicker glue, corresponding to the larger voids in the Celotex^â surface. The ligament minimum thickness does not appear to be strongly affected by the amount of glue employed, (compare Figures 6, 7 and 8) although more of the voids must be filled when more glue is applied, so the "islands" are closer together and layer of more consistent thickness. Consequently, the affect of variation in manufacturing method, with corresponding variations in amount of glue, does not greatly affect the h³ term in the stiffness parameter. Rather, having a more consistent layer would have the same effect as increasing the modulus of elasticity of the glue. In this case, the stiffness (Eh³) would be directly proportional to the amount present.

The glue layer does not lie in a single plane, so that it is wavy. This further reduces its stiffness for loading parallel to the layer. A cross section of the glue layer looks much like a string of beads. Like a string of beads, the stiffness of the assembly is determined by the stiffness at the minimum cross section (i.e., by the string), rather than the stiffness of the beads.

Microscopic examination of the Celotex^â structure has shown no characteristic which supports attributing the initial stress spike to any property or characteristic of the cellulose fiberboard material or matrix. This conclusion is supported by the laterally constrained crush test results.

The microscopic examination of the glue layer indicates that it has the form of a thin membrane pebbled with islands of glue, similar to a string of beads in cross section. The stiffness of this structure is determined by the stiffness of the minimum thickness regions. Since the minimum thickness is determined by contact of the asperities of the Celotex^â surface, the minimum thickness is not strongly affected by the amount of glue applied. Consequently, close specification of glue application in the assembly process is not required.

1. Miller, K.W., Khan, J., Khandkar, M. Z. H., A. C. Smith, Evaluation of Celotex for Shock Protection of Packages, To Be Presented at 2001 ASME PVP Conference.
2. Smith, A. C., Vormelker, P. R., Chapman, G., Creech, G., Khan, J., Miller, K. W., Khandkar, M. Z. H., Effect of Orientation and Strain Rate on Crush Strength of Cellulose Fiberboard Assemblies, To Be Presented at 2001 ASME PVP Conference.
3. Gong, C.,Wu, T. and Smith, A. C., Computational Parametric Analysis of Mechanical Behaviors of Celotex Implanted with Glue Plates, To Be Presented at 2001 ASME PVP Conference.
4. Walker, M. S., Packaging Materials Properties Data, Oak Ridge report Y/EN-4120, Jan. 1991
5. Jones, N. and Wierzbicki, T., Editors, Structural Crashworthiness and Failure, Chap. 8, Dynamic Compression of Cellular Structures and Materials, Elsevier Applied Science, London, 1993.
6. Easterling, K. E., Harryson, R. Gibson, L. J., and Ashby, M. F., On the Mechanics of Balsa and Other Woods, Proc. R. Soc, London, A383, pp 31-41, 1982.
7. Product Data Sheet: Elmer’s Professional Carpenter’s Wood Glue, Elmer’s Products,Inc. Columbus,OH.
8. Safety Analysis Report – Packages, 9965, 9968, 9972-9975 Packages, Westinghouse Savannah River Co. Report WSRC-SA-7, Rev. 10, 2000.

The authors wish to recognize G. Chapman and G. Creech, who performed the testing at SRS, and K. Miller of the University of South Carolina, for their conscientious support. Mr. Miller performed the impact testing at USC, under the direction of Dr Jamil Khan and Dr. Curtis Rhodes of the Mechanical Engineering Department of the University of South Carolina.

Figure 3. Glue Layer evaluation test Specimen 1. Average width of glue layer was
0.035 in. 0.68 gm of glue were retained by this specimen.

Figure 4. Glue Layer evaluation test Specimen 2. Average width of glue layer was
0.045 in. 1.3 gm of glue were retained by this specimen.

Figure 5. Glue Layer evaluation test Specimen 3. Average width of glue layer was
0.041 in. 2.46 gm of glue were retained by this specimen.

Figure 6. Microscopic view of glue layer in Specimen 1.

Figure 7. Microscopic view of glue layer in Specimen 2.

Figure 8. Microscopic view of glue layer in Specimen 3.

Figure 9. Glue affected zonde of Celotex assembly. The ink dot on the left is on glue layer and approximately
the width of visible layer. The lines bound the glue affected zone

Figure 10. View normal to the surface of a Celotex^q sheet in the un-damaged condition.

Figure 11. Edge view (cross section) of a Celotex^q sheet in the un-damaged condition.

Figure 12. Cellulose fibers between the glue filled voids (vertically lined region)
act in parallel with the glue in the glue affected zone.

Figure 13. View of impact face of Celotex for crushed specimen loaded parallel to the plane of the
Cellotex sheet. The view shows that the cellulose fibers are not crushed.