TECHNOLOGY TOPICS > quality control

FEA studies of impact loads

on NNPB refillable bottles
Dr Wenke Hu, William G Slusser and Gary Smay discuss impact load
considerations when using NNPB forming technology for refillable beer bottles.
In a previous study, the internal analysis (FEA), utilising an Autodesk the carbonation level of typical beers, as established by
pressure and vertical load mechanical simulation programme(2, 3). worldwide specifications.
performance of refillable bottles In these studies, the physical A 3D symmetrical model was created using
was evaluated through computer dimensions of the bottles were Solidworks for each of the four glass container designs.
stress analysis. The evaluation maintained constant throughout the The outer surface profile was created from information
compared the same bottle designs analyses. This was done to avoid that was provided on technical drawings of these four
made by both the narrow neck dimensional changes that would add specific bottles. The glass weights shown in table 2
press and blow (NNPB) process complexity to the stress analysis. were calculated on a theoretical basis starting from the
and the blow and blow process It is understood that keeping the minimum thicknesses, while simultaneously considering
using two different approaches(1): dimensions constant will affect the the max-to-min thickness ratios for each of the two
(a) The minimum thicknesses were overflow capacities. For the current forming processes, along with the two approaches being
maintained constant, while the bottle designs, these were found to evaluated in this study.
maximum and average thicknesses vary by about 3%. While this volume Impact analysis: For any given impact, three stress
were allowed to fluctuate based on variation would have to be taken modes are generated as shown in figure 2: (a) The contact
typical maximum to minimum (max- into account in actual commercial mode on the outside glass surface at the impact site; (b)
to-min) thickness ratios for the NNPB practice, it did not significantly alter the flexure mode on the inside surface, directly opposite of
and the B/B processes; and (b) the the results of the stress analyses. the impact site; and (c) the hinge mode that occurs on the
average thicknesses were maintained outside surface at a distance away from the impact site(4).
constant, while the minimum and CONTAINER FINITE The prediction of these stresses in a container
maximum thicknesses were allowed ELEMENT ANALYSIS subjected to an impact load is difficult due to the
to fluctuate, based on typical max- Thickness distributions and dynamic nature of impact. Fully dynamic analysis requires
to-min thickness ratios for the NNPB computer modeling: The max-to- substantial computational modeling and time. Alternatively,
and the B/B processes. min thickness ratios that were used for a low speed impact of just metres per second
In the present study, the in this study for the NNPB and B/B (velocities typically encountered by glass containers), a
discussion is extended to include processes are shown in table 1. quasi-static approach can be accurately used to evaluate
impact forces. The same approach These values are based on numerous the impact stresses, since the duration of an impact is
from the previous study was used for measurements of bottles made by long in comparison to the time period of the fundamental
four different beer bottle sizes and the B/B and NNPB processes in frequencies(2).
designs (330ml, 500ml, 650ml and unrelated studies. The minimum A methodology has been developed that permits the
750ml), as shown in figure 1. The thickness values for refillable beer prediction of impact stresses by combining finite element
impact stress indices of each design bottles were chosen based on the analysis with an impact index concept(2). The impact index
were obtained through finite element body diameter of the container and can be written as

Figure 1: 3D Solidworks model for four different bottle sizes. The green coloured region represents the Figure 2: Tensile stresses developed during impact.
thickness distribution.

96 GLASS WORLDWIDE > issue sixty six 2016 Subscribe online at www.glassworldwide.co.uk
 TECHNOLOGY TOPICS > quality control

(1)

where σ is the stress, V is the impact velocity and F is the

peak force generated during the impact. While σ/F can be
obtained directly from finite element analysis, the force Table 1: Typical maximum to minimum thickness ratios for the NNPB and the B/B processes.
index F/V can be written as follows(2,5):

(2)

where m is the effective mass and k is the contact

stiffness. These values are dependent on the centre of
gravity, radius of gyration, bottle dimensions and bottle
weight and must be determined for each unique bottle
design being considered(5).
Combining equations (1) and (2) results in:
Table 2: Calculated glass weights.
(3)

surface, directly opposite of the occurred in the bearing surface
The use of equation (3) results in an impact stress per impact point and the in-plane hinge region. The stresses in each of these
unit velocity index that can be used to evaluate container stress occurred on the outside key regions were obtained for each
impact performance in a manner similar to the way in surface at the same height as the of the four designs and for both
which an internal pressure stress index is currently used(2). impact point. The maximum hinge manufacturing processes.
In this study, the unit impulse force in the FEA stress usually occurred in the upper
analyses was applied directly to the minimum thickness shoulder region for the shoulder RESULTS AND DISCUSSION
location at the shoulder and heel contact of the subject impact, while the maximum hinge The results from the evaluations
bottles. The stresses generated in the model for key stress usually occurred in the lower of the four different bottles
regions (such as the contact point, flexure region, in-plane most heel region, below the contact produced the same general trends.
hinge region and shoulder and heel hinge regions) were point for the heel impact. Moreover, Consequently, for simplicity of the
calculated. The flexure stress occurred on the inside bearing surface hinge stresses discussion, only the results from the
 TECHNOLOGY TOPICS > quality control

of the NNPB process and the B/B process were identical,

as shown in table 2. This was due to the higher minimum
glass thicknesses from the NNPB process being essentially
offset by the lower maximum glass thickness for the same
process. As shown in table 4, the impact stress indices
for the NNPB bottles were consistently 4% to 14% lower,
compared to the bottles made by the B/B process.
These stress index reductions were due to the higher
minimum thicknesses achieved with the NNPB process,
which is the result of less thickness variation and improved
glass distribution. It should be noted that the contact
stiffness increases with the NNPB process due to less
bending deformation during the impact associated with
the higher minimum thickness. Thus, bottles made by the
NNPB process in these considerations of nearly equal bottle
weight, would exhibit significantly lower flexure and hinge
stresses compared to bottles made by the B/B process.
The exception would be the contact stresses, which are
associated with the stiffness of the impact site.
Table 3: Impact load stress indices for 330ml bottle with identical minimum thickness.
CONCLUSION
In this study, both identical minimum thicknesses and
analyses of the 330ml capacity bottle and heel contact impact sites, were identical average thicknesses for NNPB and B/B processes
will be presented here. consistently 1% to 6% higher for were analysed for shoulder and heel impacts through finite
Approach No 1 - Identical bottles made by the NNPB process element analysis. It was concluded that:
minimum thicknesses: With compared to the B/B process. This • When minimum thicknesses were maintained
identical minimum thicknesses, increase for the hinge stress is due to at the same value, bottle weight can be reduced
the resulting glass weights were the slightly lower glass thicknesses and approximately 14% through the use of the NNPB
approximately 14% lighter for bottles consequently, more glass compliance process. This weight reduction can be achieved with
made by the NNPB process than for in the hinge regions of the bottles. only small increases in the impact stress indices.
bottles made by the B/B process, Lack of structural reinforcement also • When the average thicknesses are held at the same
as shown in table 2. This weight allows for more flexibility at the point value, impact flexure and hinge stress indices can be
reduction was expected, since the of impact and consequently, slightly significantly decreased through the use of the NNPB
typical max-to-min thickness ratios increased flexure stress. The contact process while the bottle weight remains unchanged.
were less for the NNPB process stiffness was similar for bottles from Glass stiffness and the resulting contact stresses will
compared to the B/B process. the two processes, since the impacts be somewhat greater for bottles made by the NNPB
Consequently, bottles made by the were directed at the same minimum process and these values would have to be carefully
NNPB process have greater glass thicknesses and contact stress is very considered in making any decisions relative to the use
thickness uniformity and therefore, localised. of the NNPB process.
less overall weight when the Approach No 2 - Identical Further discussions are planned to include the strength
minimum thickness is fixed. average thickness: When the of glass and failure criteria considerations, in order to
As shown in table 3, the flexure average glass thickness was determine ultimately the effects of the NNPB process on
and various hinge impact stress maintained constant, the calculated refillable bottle performance. These discussions will be
indices, for both shoulder contact bottle weight resulting from the use reported in the third and final paper in this series.

REFERENCES
1. W Hu, W G Slusser, P de Haan and G Smay, Using NNPB forming
technology for refillable beer bottles, Glass Worldwide, 60, 54-56, 2015.
2. M W Davis, Effect of Lighweighting on the Impact Performance of
Glass Containers: Further Discussions, Glass Technology, 34 (4), 163-
165, 1993.
3. M W Davis and A R Shott, Optimum Thickness Distribution for
Lightweight Glass Containers, Glass, June 221-222, 1995.
4. R E Mould, The behavior of glass containers under impact. J Am
Ceram Soc 35 (9), 230-236, 1952.
5. W G Slusser and S W Spence, Filling line impact measurement.
Glass Worldwide, 62, 104-107, 2015.

ABOUT THE AUTHORS:

Dr Wenke Hu is Senior Scientist, William G Slusser is
Manager Research Services and Gary Smay is Senior
Scientist (retired) at American Glass Research

FURTHER INFORMATION:
American Glass Research, Butler, Pennsylvania, USA
tel: +1 724 482 2163
email: tbarr@agrintl.com
web: www.americanglassresearch.com
Table 4: Impact load stress indices for 330ml bottle with identical average thickness.

98 GLASS WORLDWIDE > issue sixty six 2016 Subscribe online at www.glassworldwide.co.uk