SMALL-SCALE MODELING OF

EXPLOSIVE BLASTS IN URBAN SCENARIOS

R. C. Ripley
1
, B. von Rosen
2
, D. V. Ritzel
3
and D. R. Whitehouse
1

1
Martec Ltd. Combustion Dynamics Group
1888 Brunswick St., Suite 400, Halifax, NS, B3J 3J8, CANADA
2
Canadian Explosives Research Lab (CERL), Natural Resources Canada - Minerals
and Metals Sector, 555 Booth St., Ottawa, ON, K1A 0G1, CANADA
3
Dyn-FX Consulting Ltd., 19 Laird Ave. N., Amherstburg, ON, N9V 2T5, CANADA


The present investigation deals with modeling of explosive blasts in
urban scenarios using scaled geometries. Both experimental and
numerical results are considered for three scenarios that represent a
single building target, streets between structures, and a vacant
courtyard. These scenarios have been instrumented to show simple
and oblique reflection, clearing, channeling and diffraction of shock
waves caused by air blast. Results from 50 g C4 charges are presented
at 1/24 scale. The goal is to demonstrate the capability to capture blast
trends as seen in experiment using rapid coarse mesh calculations.
Good agreement between the results of experiment and numerical
simulation is shown.


INTRODUCTION

Explosions in urban scenarios are of great concern with respect to predicting the
vulnerability of structures and humans to blast. Full-scale testing of realistic urban blast
scenarios related to terrorist bombs is often prohibitively expensive and time
consuming. Practical alternatives, which can dramatically reduce the time and cost to
deliver results, include small-scale experiments and numerical simulation.
Although it is generally not feasible to use small-scale testing to assess structural
response or damage, within certain limitations this is a well-proven means to assess
blast loading. Small-scale experiments have been used in the past for validation of
hydrocodes, e.g., [1] and [2]. The purpose of the present investigation is to validate the
rapid blast modeling capability of the Chinook CFD code against small-scale test results
in anticipation of deployment for modeling of full-scale scenarios.
Results from new experiments are presented here to examine the effect of charge
position and target geometry on the accuracy of blast pressure and impulse predictions.
Of particular interest is the shock interaction behind the building, where predictions
using simple analytical models [3], are not valid. Experiments were conducted at
different stand-off distances and at different incidence angles relative to the geometric
center of the target geometry. The target is representative of a typical three-story urban
structure and the scaled charge size is representative of a typical truck-bomb threat.
Many blast parameters can be scaled for charge masses ranging from grams to
tonnes [4]. The most widely used method of blast scaling is Hopkinsons cube-root
law for scaled distance, time and impulse. Further, Ohashi [5] showed that scaling of
some blast parameters could be extended to the milligram range. Cube-root or energy
scaling is generally considered valid for blast scenarios where the explosive source is an
idealized prompt detonation and the regime of interest is well outside the fireball
expansion. Energy scaling does not apply until distances at which the charge mass can
be neglected against the mass of air engulfed by the propagated blast [6]. The present
investigation deals with far-field explosive blasts where scaling laws are applicable.


SMALL-SCALE CONFIGURATIONS

The small-scale configurations were tested on the Canadian Explosives Research
Laboratory (CERL) blast table, which is 1.2 x 2.1 m in planform. The table features an
array of mounting holes on 7.6 cm centers that facilitate the placement of sensors
(pressure transducers) and modular 38.1 x 38.1 x 7.6 cm aluminum building blocks.
The three scenarios studied are shown in Table 1. Due to the fixed block size, a
T-intersection between walls exists in Series 2 and 3, but this is not expected to impact
the results.

TABLE 1. Illustration and description of small-scale urban scenarios.
Series Schematic Description
1

The first scenario involves a single building target that is 91.4
cm by 38.1 cm and 38.7 cm tall (including a 0.6 cm thick plate
covering the walls to enclose the building).
2

The second scenario adds 38.1 cm tall walls adjacent to
three sides of the rectangular block to simulate streets
between buildings. The width of the side and rear passages is
15.2 cm. This configuration highlights the shock channeling
effects and complex shock interactions behind the building
and in the passages.
3

The third scenario removes the primary building, leaving the
surrounding walls to form an empty courtyard. This setup
demonstrates shock focusing in a re-entrant corner scenario,
and identifies the effect of removing the building.
Scaled Explosive Charge

The charge mass for all trials was 50 g of C4, formed as a rectangular cube (2.5
x 2.5 x 5.0 cm) with its longitudinal axis oriented upwards. The charges were initiated
from the bottom using electronic detonators. The center of the charge was bisected by
the table surface, directing one-half the blast energy at the target model. Three stand-off
distances were tested; only the results of 1.37 m are presented here. This gives about
1 atm overpressure at the gauge locations of interest, which provides good sensitivity in
the sensor range used in the experiment. The C4 charges were comprised of nearly-ideal
RDX explosive and the stand-off distance tested corresponds to a spherical mass
equivalent of 70 charge radii, indicating that both energy scaling and spherical charge
assumptions are valid.
On the full scale (24:1), this charge size corresponds to 345 kg of explosive
mass for a hemispherical blast, with an equivalent stand-off distance of 32.9 m. The
building height in full scale is 9.3 m.

Selection of Gauge and Charge Locations

The data acquisition system employed at CERL permits instrumentation of up to
16 pressure transducers. To allow commonality between configurations, all 16 gauges
were positioned on the table surface. The sensor locations were selected to show
reflection, clearing, channeling and diffraction of shock based on the Series 2 scenario.
Fig 1 shows the placement of 16 pressure transducers, which have been
concentrated in the corners and have taken advantage of geometric symmetry. This
investigation focuses on the results from locations with interesting physics, i.e., behind
the building. Fig 1 also identifies the location of the charges for the three scenarios.


FIGURE 1: Layout of gauge and charge locations [numbered SERIES-TRIAL] for small-scale test (left)
and experimental configuration (right) illustrating apparatus including CERL blast table, modular walls,
sensor mounting locations, and the explosive charge positioned with a stand.
NUMERICAL MODEL

Balloon Analogue Model

It was presumed that the blast from the rectangular charge developed into a
uniform quasi-spherical blast prior to its interaction with the target. The balloon model
approach of Brode [6] was employed in the numerical calculations. The explosive itself
was represented as a spherical volume of high-pressure, high-temperature ideal gas (the
explosive products) containing the equivalent energy of the explosive charge.
Thermodynamic properties of carbon-dioxide, with ratio of specific heats of 1.27 and
molecular weight of 44.01 g/mol, were chosen for the explosive products as this best
provides behaviour similar to that of C4 products [7]. To be consistent with the nominal
density of solid C4, the initial density of the gas in the balloon model was 1600 kg/m
3
.

Calibration of Explosive Model

The explosive energy of the C4 charge is applied to the numerical model based
on its mass and TNT equivalency. The TNT equivalency of C4 is not unique and
typically ranges from 1.15 to 1.60 depending on how it is measured. As a reference,
CONWEP uses a C4 equivalency of 1.37 for pressure and 1.19 for impulse [3].
A 1-D spherical model was used to calibrate the numerical model by matching
the experimentally observed pressure at Gauge 1. Several explosive equivalence values
were tested empirically to calibrate the model. The pressure equivalence of 1.37 was
found to provide the best match for peak pressure and arrival time, as compared to the
experimental results, and impulse, as predicted by CONWEP (Fig 2). The selected
equivalence was then applied to a full 3-D model for additional validation (Fig 2).


FIGURE 2: Calibration of C4 detonation using 1-D spherical model (left) and full 3-D model (right).
Spherical Blast Mapping

Preliminary calculations with cubical charges in 3-D showed that the blast wave
approached spherical within a 1.0 m radius. The high-resolution 1-D blast profile was
mapped onto the 3-D mesh as a spherically symmetric source. This enabled a coarser
mesh to be used in 3-D to reduce run time and deliver results more rapidly. The blast
profile was selected at an instant when the blast wave reached 1.0 m in radius, prior to
its interaction with the targets. The extent of the fireball can be seen in the blast profile
(Fig 3) as the interface between detonation products and air behind the blast wave. Fig 3
illustrates the blast profile and pressure mapping on the surface mesh only. The full 3-D
mesh contained 713,490 to 815,541 tetrahedral cells depending on the scenario.


FIGURE 3: Detonation profile at t = 1.12 ms (left); explosive blast pressure profile applied to 3-D
unstructured computational mesh (right).


COMPARISON OF RESULTS

Effect of Diffraction Angle

The effect of diffraction angle was investigated using Series 1, where the
charges were positioned in a radial array through 90 (Fig 4). Gauge 10 located at a rear
corner of the building was selected for comparison. Overall, good qualitative agreement
of the P-t histories was achieved, including the double peak in Test 1-3 (60 incidence)
that is captured by both the experimental and numerical models. Peak pressures
typically differed by 19% and impulse by 15%. This is excellent agreement considering
that the calculation was performed without any effort to refine the mesh. Further, the
uncertainty in the experimental results has not been quantified.
Blast Focusing in a Re-entrant Corner

Fig 4 illustrates the effect of charge location on the reflected pressure observed at Gauge
16 located in a corner. The peak reflected pressure and time of arrival increase as the
distance of the charge from the corner is increased, indicating a shock focusing effect.
In all cases, the peak pressure in the numerical results is less than that of the experiment.
The numerical results are somewhat diffuse in differentiating between the incident and
reflected shock, but accurately capture the reflected blast decay. This is reasonable
considering that a coarse mesh was used to generate results rapidly. Impulses predicted
by the numerical results are about 20% greater, indicating that the numerical approach is
capable of providing loading estimates and hence damage predictions.


FIGURE 4: Effect of charge incidence angle on pressure diffraction at corner of building (left) and shock
focusing observed in a corner (right). [solid lines experiment; broken lines numerical]

Effect of Scenario Layout

Fig 5 compares the effect of the three scenario configurations. The results
behind the building demonstrate that the single building alone (Series 1) had the lowest
pressure, while the building and surrounding walls (Series 2) had the greatest pressure.
Most significantly, the removal of the building (Series 3) resulted in a lower peak
pressure than with the building present, even though this case is essentially an
unobstructed incident blast that is normally considered the worst case. This is a result of
the superposition of waves that are channeled through the passages surrounding the
building and waves that are diffracted over the top of the building, ultimately colliding
behind the building. The building alone has a strong sheltering effect illustrating simple
blast diffraction. Comparison between experiment and numerical simulations show
broad agreement and clearly distinguish the effect between configurations.

FIGURE 5: Effect of urban scenario on blast observed at Gauge 14 (behind building).
[solid lines experiment; broken lines numerical]

CONCLUSIONS

Overall, the Chinook code has been shown to accurately capture the effects of
pressure clearing, shock diffraction, blast channeling and complex wave reflections. As
compared to experiment, the numerical results tended to underpredict peak pressure and
overpredict impulse, which can be attributed to the coarse mesh resolution (1.26 cm)
employed in 3-D. However, the numerical results clearly distinguish effects of charge
location and scenario layout. The goal was not to obtain the best comparison for a single
test, rather to demonstrate the capability for many tests using a generic coarse mesh.
Pressure and impulse levels observed in this study are insufficient to cause
human fatality near the urban structures; however, the present results are applicable to
human injury and window damage predictions. The results have shown that the
combination of charge position and scenario layout can result in dramatically different
outcomes depending on target location. This indicates that simplified numerical
modeling approaches may have difficulty in predicting actual outcomes in complex
scenarios, and that experiment and numerical simulation using CFD are more suitable
for these predictions.
On the scale that the experimental tests were performed, quick turnover between
tests allowed as many as 20 tests per day to be completed for a given configuration.
Similarly, the numerical simulations required about 2 hours of CPU time per test on a
1.5 GHz desktop, allowing blast predictions to be determined rapidly.
The combination of scale modeling and numerical simulation can be an
important tool in supporting and complimenting large-scale experiments due to cost and
effort benefits. Rather than being a disadvantage, the absence of non-scalable effects
provides the following:
1) Test environment is clean, simple and close to ideal. Complications encountered in
full-scale tests can be avoided, such as fragmentation, debris and field data
acquisition issues.
2) Blast calculations can be validated using relevant models and geometries.
3) Changes in target scenario can be investigated quickly with additional small-scale
tests and numerical simulations.
4) Results of small-scale investigations can be used to assist in planning an effective
full-scale test matrix. This will allow full-scale investigations to focus on the non-
scalable phenomena found in thermobaric and fuel-air devices.


RECOMMENDATIONS

The experimental results contain a significant amount of noise, which may be a
result of table ringing and shock interaction with the table supporting structure. It is
recommended that efforts be made to reduce the noise. Some wave reflections from the
far-field boundaries in the numerical model were observed. Future studies should place
model extents farther from the target geometry. Mesh refinement should also be pursued
to improve resolution of the numerical results.
All measurements of overpressure and impulse were obtained at ground level;
future investigations should focus on the results measured at various elevations. The
effect of charge shape should be investigated further using either new experiments or
additional numerical simulations. The modeling of detonation products could also be
improved by using a more realistic equation of state such as JWL.


REFERENCES

1. G. E. Fairlie, Efficient Analysis of High Explosive Air Blast in Complex Urban Geometries
using the AUTODYN-2D & 3D Hydrocodes, MABS 15, Sept. 14-19, 1997.
2. C. E. Joachim, B. J. Armstrong, and D. D. Rickman, Free-Field Airblast on Structures:
Computational Model, MABS 17, June 10-14 2002, Las Vegas, NV.
3. D. W. Hyde, CONWEP Conventional Weapons Effects Programme, U.S. Army Engineer
Waterways Experiment Station, Vicksburg, MS, 1992.
4. W. E. Baker, Explosions in Air, University of Texas Press, Austin, 1973.
5. K. Ohashi, H. Kleine, and K. Takayama, Characteristic of Blast Waves Generated by Milligram
Charges, Proceedings JSSW (2001), pp. 441-481. (in Japanese)
6. H. L. Brode, The Blast Wave Resulting from a High Temperature, High Pressure Sphere of
Air, RM-1825-AEC, The Rand Corp. [Unpublished].
7. D. V. Ritzel and K. Matthews, An Adjustable Explosion-Source Model for CFD Blast
Calculations, ISSW 21, Great Keppel Island, Australia, July 20-25, 1997.