Effects of close range blasts on steel frames. Experimental testing and numerical validation

Abstract

During last decades, there was an increased interest from research and design professionals to provide effective strategies in protecting buildings and other assets from the direct effects of blasts or other incidents. Experimental tests, conducted over a large range of distances and charge weights, helped at developing analytical approaches and charts which can be used to calculate blast parameters. Due to the lack of test data and inapplicability of common scaling rules, in the last years special attention was devoted to close-in blasts, located in the proximity of the structural elements. Such explosive charges may cause extreme lo-cal damage of the elements or even complete loss of load bearing capacity. In the study presented in the paper, two types of beam-column assemblies have been tested under explosive charges detonated close to the specimens. Numerical models, developed using Extreme Loading for Structures software, were validated using the test data collected in the experimental program.

Full text

EFFECTS OF CLOSE RANGE BLASTS ON STEEL FRAMES.
EXPERIMENTAL TESTING AND NUMERICAL VALIDATION
Florea Dinua,b, Ioan Mărgineana, Andreea Sigauana, Attila Kovacsc,
Emilian Ghicioic and Dragos Vasilescuc
a Department of Steel Structures and Structural Mechanics, Politehnica University Timisoara, Romania
b Laboratory of Steel Structures, Romanian Academy, Timisoara Branch, Romania
c INCD INSEMEX Petroșani, Romania
Abstract: During last decades, there was an increased interest from research and design pro-
fessionals to provide effective strategies in protecting buildings and other assets from the di-
rect effects of blasts or other incidents. Experimental tests, conducted over a large range of
distances and charge weights, helped at developing analytical approaches and charts which
can be used to calculate blast parameters. Due to the lack of test data and inapplicability of
common scaling rules, in the last years special attention was devoted to close-in blasts, locat-
ed in the proximity of the structural elements. Such explosive charges may cause extreme lo-
cal damage of the elements or even complete loss of load bearing capacity. In the study pre-
sented in the paper, two types of beam-column assemblies have been tested under explosive
charges detonated close to the specimens. Numerical models, developed using Extreme Load-
ing for Structures software, were validated using the test data collected in the experimental
program.
1. Introduction
Over their lifespan, the buildings may be subjected to a variety of actions. Design and execu-
tion of buildings should take into account that some of these actions may reach extreme val-
ues, much higher than those currently used at initial design. Therefore, the buildings must be
able to resist such events without being damaged disproportionately compared to the original
cause. One example is the gas explosions that are produced in residential buildings due to
failure or improper use of the gas facilities. In such situations, the structure can be severely
damaged, leading to partial or complete failure, e.g. the Ronan Point building, in London, in
1968. Explosions caused by gas accumulations have great significance, given the large num-

The International Colloquium on Stability and Ductility of Steel Structures, Timisoara, Romania
ber of residential buildings (and other types of spaces) where it is currently used. Hazards can
be caused both by external distribution networks and the devices that use the gas inside. For
example, in the United States, the average probability of producing a gas explosion is about
1.8 × 10-5/ year [1]. Studies conducted in the UK showed similar values of the order 2.3 ×
10-5/ year to 1.86 × 10-5/ year. In other countries, e.g. Romania, there are no public data about
such events, and this makes it difficult to assess the risk of such events based on the number
of dwellings. The second important issue is the maximum pressure in the rooms affected by
the explosion. Studies conducted after the accident in the Ronan Point building showed that
these pressures rarely exceed the value of 17 kPa. Even if this value is much higher than "or-
dinary" loads, it is less than the amount recommended in EN 1991-1-7 [2], namely 34 kPa.
The pressure resulting from gas explosions is therefore very high compared to permanent or
live loads (occupancy, wind, snow), which rarely exceed 4-5 kPa. The presence of voids lim-
its first pulse intensity and pressure fluctuations of the second phase presents a reduced signif-
icance for the structural response. Therefore, it can be considered that the effect on the struc-
ture is quasi-static (Fig. 1.a).
Unlike internal gas explosions, external explosions are usually produced from intentional
causes, the source being both classic explosives (TNT, C-4, Semtex) but also improvised
charges, like those based on ammonium nitrates, a chemical fertilizer widely used in agricul-
ture. Because of the intentional nature, these events cannot be modeled using a Poisson distri-
bution model, which is widely used to model the random occurrence of rare events [1]. Explo-
sions caused by explosives, known as detonations, create shock waves that propagate at high
speeds from the point of detonation. The initial phase, of positive pressure, is followed by a
longer and less intense negative pressure phase, Fig. 1.b. Unlike gas explosions, detonations
induce significant dynamic effects, and therefore they must be accounted for in design.
a) gas explosion ([3]) b) blast wave ([4])
Fig. 1: Typical pressure-time records
In terms of the effects on the structure, distance between the source of the explosion and
structure plays a very important role. When the explosive charge is placed at greater distance
from the structure, failure usually occurs due to excessive bending. The pressure released by
the shock wave can be considered as uniformly distributed on the surface of the element. An
explosion at a very short distance from the building causes a localized failure, like punching
or shear failure. This is caused by the rigidity of the structural element, which produces iner-
tial resistance to blast. Punching or shear-type failure occurs before structural element is able
to respond by bending.
The study presented in the paper investigated the response of beam-column steel elements
subjected to close-in blasts. The research focused on the evaluation of typical steel frames re-
sponse when subjected to direct effects of a blast and the direct calculation of blast loads for
further numerical simulations. Two 3D frame specimens were tested inside a bunker using
different charge weights, located at different distances. A numerical model was also validated

Florea Dinu, Ioan Mărginean, Andreea Sigauan, Attila Kovacs, Emilian Ghicioi and Dragos Vasilescu
using the advanced nonlinear dynamic analysis software Extreme Loading for Structures [5].
The study is part of a research program devoted to the design of structures to sustain extreme
loading events without collapse [6].
2. Specimens, test setup and blast loading
2.1 Description of specimens
Two 3D specimens were designed and constructed for testing under blast effects inside a bun-
ker. Specimens were extracted from a typical moment resisting steel frame structure, see high-
lighted areas in Fig. 2.a. Due to space limitation in the bunker, the specimens were scaled
down from 8.0 m span to 3.0 m span. First specimen (see Fig. 2.a, Fig. 3.a) includes a column
(blast induces deformations in the major axis direction), two half-span longitudinal beams rig-
idly connected to the flanges of the column using extended end plate bolted connections and
one half-span transversal beam, connected to the column web using a simple clip angle con-
nection.
a) plan layout b) specimen I with the bunker details c) specimen II
Fig. 2: Plan layout of the reference multi-story building with the position of the specimens extracted
for testing and 3D viws of specimens
* flange of HEB260 column reduced to 160 mm width
a) Specimen I b) Specimen II
Fig. 3: Details of the specimens
8888
8
8
8
E
D
C
A
1 2 3 4 5
8
B
II
I

The International Colloquium on Stability and Ductility of Steel Structures, Timisoara, Romania
Second specimen (see Fig. 2.b, Fig. 3.b) includes a column (blast induces deformations in the
minor axis direction), two half-span longitudinal beams connected to the column web using
simple clip angle connections and one half-span transversal beam, connected to the column
flange using extended end plate bolted connections. For both specimens, lateral restraints
made from tubular profiles were used at the ends of longitudinal beams. IPE 220 section was
used for beams, while columns were made from HEB 260, but with flanges reduced to 160
mm width. Steel material in plates and profiles was S275J0 and bolts were grade 10.9. Table
1 summarizes the measured material properties of the specimens.
Table 1: Average characteristic values for materials in steel profiles, plates and bolts
Element
fy (N/mm²)
fu (N/mm²)
ε
y (%)
Agt (%)
yield
strength
ultimate
strength
yield strain
Total elongation
at maximum stress
Beam flange IPE220, t = 9.2 mm
345
464
0.16
28.0
Beam web IPE220, t = 5.9 mm
353
463
0.17
30.4
Column web HEB 260, t = 10 mm
407
539
0.19
27.0
Column flange HEB 260, t = 17.5 mm
420
529
0.20
27.0
End plate, t = 16 mm
305
417
0.15
17.1
Bolt, M16 class 10.9
965*
1080
1.76*
12.0
Note: * 0.2% offset yield point
2.2 Blast effects
The main hazard components of an explosion are blast (overpressure), fragmentation, and
thermal effect. In our study, only first issue has been of interest. As seen in Fig. 1.b, the peak
pressure value depends very much on the distance of the detonation point from the structure
of interest. Thus, the peak pressure value, and also the velocity of the blast wave, decrease
rapidly by increasing the distance between the blast source and the target surface. The effect
of distance on the blast characteristics can be taken into account by the introduction of scaling
laws [7]. These laws have the ability to scale parameters, which were defined through exper-
iments, in order to be used for varying values of distance and charge energy release [8]. The
experimental results are, in this way, generalized to include cases that are different from the
initial experimental setup. The most common blast scaling law is the one introduced by Hop-
kinson [9] and Cranz [10]. According to Hopkinson-Cranz law, a dimensional scaled distance
is introduced as described by Eq. (1):
( )
31
W Z R=
(1)
where: Z is the scaled distance, in m/kg1/3, R is the distance from the detonation source to the
point of interest [m] and W is the weight of the explosive [kg TNT or equivalent TNT].
As the scaled distance reduces, the peak overpressure increases. For example, in case of
Murrah Federal Building in Oklahoma City, which collapsed in 1995 as a result of a blast
produced by a large truck bomb (1814 kg equivalent TNT), located at a distance R of 1.5 m,
the scaled distance was Z = 0.12 m/kg1/3, which resulted in a very large peak overpressure,
estimated at 20 000 de kPa. The scaling law from Eq. (1) should be corrected when blast test
is done inside a bunker. A more general expression to evaluate the peak pressure (or peak
overpressure) and the variation with the distance (Eq. (2)), is the one proposed by Richards
and Moore [11]:
( )
( )
a
b
WRA×=P
(2)
where: P is the peak overpressure (kPa), A is the site constant (evaluated experimentally), a is
site exponent (evaluated experimentally, is always negative), b is the site exponent for the
charge weight (evaluated experimentally), R is the distance from the detonation source to the

Florea Dinu, Ioan Mărginean, Andreea Sigauan, Attila Kovacs, Emilian Ghicioi and Dragos Vasilescu
point of interest [m] and W is the weight of the explosive [kg TNT or equivalent TNT]. Using
experimental calibration tests, Eq. (2) can be used in case of bunker tests. In the study, cali-
bration blast tests were performed first, in order to evaluate the site exponents A, b and a.
Then, the specimens were subjected to blast of increasing intensities, obtained by increasing
the charge weight and/or reducing the distance from the blast to the specimen.
3. Experimental tests
Fig. 4 shows the first specimen tested in the bunker. In order to evaluate the site exponents A,
b and a, and then the pressure inside the bunker, two Kiestler pressure sensors were mounted
on a special frame, at 4 meter distance from the specimen, in front of the bunker venting. Ex-
plosive material used in the testing has a TNT equivalence of 1. To note that the effects of
gravity loads on the columns and beams were not considered in the test.
First four blasts used charges of m1,1 = 125 g, m1,2 = 250 g, m1,3 = 500 g and m1,4 = 1000 g,
located at 0.27 m from the beams mid-length, freely suspended (see Fig. 4). Last blast used a
charge weight m1,5 = 1000g, located at the external face of the column, at mid-height, see Fig.
4. The pressure measured after each explosion were p1,max = 20 kPa, p2,max = 40 kPa, p3,max =
80 kPa and p4,max = 175 kPa. The peak pressure during last explosion attained the maximum
value, p5,max = 1400 kPa.
Fig. 4: Plan layout and test set-up for specimen I
Fig. 5: Specimen I after the test
While for first four explosions there were no visible deformations at the level of the speci-
men, the last charge, m5, caused severe local deformations of the column near the source, and
the web complete fractured at the contact with the flange, in the fillet zone, for a length of 300
mm.
bunker
specimen
Pressure
sensors
Sand bags for protecting
connecting wires
Web fractured
in the fillet zone

The International Colloquium on Stability and Ductility of Steel Structures, Timisoara, Romania
For calibration of experimental set-up, a charge with weight m = 125 g was detonated at a
distance d = 0.27m from the position of the pressure sensors. The value of the peak pressure
was Pcalibration =1400 kPa. With the values of the pressure measured during each detonation,
the following specific coefficients of the bunker were determined: A = 3850, a = -0.73, b =
3.87/3 = 1.29. The peak pressure value of the blast wave, decrease rapidly with the distance
between the blast source and the target surface, as seen in Fig. 6. Apart from the pressure
measurements, also the permanent deflection of the specimen was measured in several points.
Second specimen was tested for three charges. First charge had a weight m2,1 = 500 g, lo-
cated at distance D1 = 0.5 m from the specimen and H = 1.0 m height from the base of the
column. No visible effects were recorded, nor in the beams or in the column. The test contin-
ued with detonation of a charge of m2,2 = 1815 g, located at the same distance and height (D2
= 0.5 m, H = 1.0 m). Web of the column has been deformed plastically out of plane for ap-
proximately 22 mm, but without any visible cracks (Fig. 7.a). The third detonation was pro-
duced by a charge blast of m2,3 = 1815 g, located at same height but closer distance, D3 =
0.20m. The web of the column has been completely removed within a length of almost 600
mm around the point of detonation, see Fig. 7.b.
Fig. 6: Peak pressure vs. distance between the blast source and the target surface
a) Plastic deformations in column web after blast m2,2
b) complete fracture of the web after blast m2,3
Fig. 7: Results of blasts for Specimen II
100
1000
10000
100000
1000000
0 1 2 3 4 5
Pressure, kPa
Distance, m

Florea Dinu, Ioan Mărginean, Andreea Sigauan, Attila Kovacs, Emilian Ghicioi and Dragos Vasilescu
To note that, because the effects of gravity loads were not considered in the test, the col-
umn did not failed. However, in real conditions, the gravity loads are always present in the
structure, and therefore such damage would be similar to a complete loss of bearing capacity
(removal of the column). The permanent deflections of the specimen were measured in sever-
al points after the test and were used for numerical model validation.
4. Calibration of the numerical model
The performance of the steel specimens undergoing close range detonations was also predict-
ed using ELS [5] (see Fig. 8). ELS utilizes a nonlinear solver based on the applied element
method [12] which is a derivative of the finite element method and the discrete element meth-
od. In ELS, the structure is modeled as an assembly of small elements, which are assumed to
be connected by one normal and two shear springs located at contact points distributed around
the element edges. The average normal strain is calculated by taking the average of the abso-
lute values of strains on each face. When the average strain value at the element face reaches
the separation strain, all springs at this face are removed and elements are not connected any
more (until they collide). Columns, beams, and plates were modeled as solid elements and
could undergo deformations at the interface between the discretized elements. The constraints,
made of tubular sections, were also modeled as solid elements. The bolts were modeled using
individual springs: one for normal stresses and two for shear stresses. The column bases and
transversal beam end connection to the bunker wall were considered pinned, and all dis-
placements were prevented.
a) Specimen I
b) Specimen II
Fig. 8: Numerical models of specimens
Blast load acts
on this area
Blast load acts
on this area

The International Colloquium on Stability and Ductility of Steel Structures, Timisoara, Romania
Materials were modelled using characteristic properties presented in Table 1. Blast pres-
sure was modeled using the experimental data presented in previous section. The variation of
pressure with distance obtained in the experimental test were introduced in the program, and
the point of detonation was the same, i.e. D = 0.2 m from the column face. The shaded areas
presented in Fig. 8 (left) indicate the surface of the elements loaded with the blast pressure.
Due to the dynamic, impulsive character of the blast load, strain rates effects in the materi-
al are very important. The rate dependency has been considered by means of the following
relationships[13]:
𝑓
𝑦𝑠𝑟
𝑓
𝑦
= 1 + 21
𝑓
𝑦
log έ
𝜀0 (3)
𝑓
𝑢𝑠𝑟
𝑓
𝑢
= 1 + 7.4
𝑓
𝑦
log έ
𝜀0 (4)
where: έ = strain rate, ε0 = 10-4
fy, fu = yield and tensile strength in quasi-static conditions, ε0 = 10-4
fysr, fusr = yield and tensile strength at strain rate έ.
Because the strain rate is not initially known, the analysis is done first using static material
properties, see Table 1. The strain rate is then calculated in the location of interest and materi-
al properties are corrected using Eq. (3) and Eq (4). For first specimen, the maximum strain
rate recorded in the column web was έ = 312.23 (1/sec), while for second specimen έ = 83.56
(1/sec).
Fig. 9 and Fig. 10 display the deformed shape of the specimens and the displacement histo-
ry (along Y axis). Comparisons between numerical and experimental results show a very good
agreement. The permanent deflection of the columns is very close to the measurements done
after the test. Also, the extension of damage in the columns are very similar, with the same
location and extension of fracture lines.
a) deformed shape b) displacement vs. time
Fig. 9: Specimen I, numerical vs. experimental, test m1,5
a) deformed shape after test m2,3 b) displacement vs. time for test m2,2
Fig. 10: Specimen II, numerical vs. experimental
0
10
20
30
40
50
60
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Displacement Y, mm
Time, sec
ELS
Experimental
0
10
20
30
40
50
60
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Displacement Y, mm
Time, sec
ELS
Experimental

Florea Dinu, Ioan Mărginean, Andreea Sigauan, Attila Kovacs, Emilian Ghicioi and Dragos Vasilescu
S15
S14
S13
S12
S11
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
The blast load pressure and therefore the blast effects decrease very rapidly with the dis-
tance from the explosive charge. For close-in detonations, the effects can concentrate in the
area of the element adjacent to the explosive charge, while the effects on the remaining struc-
ture can be neglected. Thus, Fig. 11.a and Fig. 11.b show the shear strains in springs located
in the filet zone of the column web, within a length of 50 cm for specimen I and 60 cm for
specimen II, respectively, measured from the bottom side of the beams. It can be seen that,
even for small charges (maximum weight 1815g equivalent TNT), the local damages are ex-
treme, but localized to the area close to the point of detonation.
a) model for specimen I
b) model for specimen II
Fig. 11: Evolution of strain in columns
5. Conclusions
The study presented in the paper focused on the evaluation of close-in blast effects against
steel frame structures. Two specimens were tested for charges of different weights and located
at different distances to the specimens. The specimens were extracted from a six-story mo-
ment resisting frame structure, designed to meet the seismic design requirements for special
moment frames. The variation of blast pressure with time and distance has been evaluated for
bunker test conditions. The results showed that charges located at close distance can produce
large damages in the members, with complete fracture of the section walls. Thus, for the spec-
imen loaded against the strong axis (charge normal to the column flange), the blast caused
sever local bending of the external flange, and fracture of the web at the contact with the
flange, in the fillet zone. In case of specimen loaded against the weak axis (charge normal to
0,25m
0,30m
s
1
s2
s3
s4
s5
s6
s7
s8
s9
s10
s11
s12
s13
s14
S
1
S
12
S
11
S
10
S
4
S
2
S
9
S
5
S
6
S
7
S
8
S
3
s
1
s2
s3
s4
s5
s6
s7
s8
s9
s10
s11
s12
s15
0,3m
0,4m
Springs
Explozive
charge
Springs
Time, sec
0.0
0.0
0.0004
0.0008
0.0002
0.0012
0.0
0.0004
0.10
0.40
0.0006
0.20
0.30
0.0008
Time, sec
Explozive
charge
B
0.0
-0.10
-0.40
-0.20
-0.30
Normal
strain
Explozive
charge
A
B
A
Explozive
charge
Shear strain

The International Colloquium on Stability and Ductility of Steel Structures, Timisoara, Romania
the column web), punching or shear-type failure developed, with the web completely removed
for a length of 600 mm, before structural element to be able to respond in bending. The later
situation can be extremely dangerous, especially for buildings with perimeter steel moment
resisting frames and interior gravity frames, if the loads cannot be distributed to the adjacent
structural members. In the test, the complete failure of the column was prevented only due to
the absence of any gravity loads. Numerical models that were calibrated using test data indi-
cated a very good agreement, giving the possibility to extend the research to full scale struc-
tures, using different blast loading conditions.
Acknowledgments
Funding for this research was provided by the Executive Agency for Higher Education, Re-
search, Development and Innovation Funding, Romania, under grant PCCA 55/2012 “Struc-
tural conception and collapse control performance based design of multistory structures under
accidental actions” (2012-2016).
References
[1] NISTIR 7396. Best Practices for Reducing the Potential for Progressive Collapse in
Buildings. National Institute of Standards and Technology Administration, U.S. Dep. of
Commerce, 2007.
[2] EN1991-1-7: Eurocode 1: Actions on structures – Part 1-7 General actions - Accidental
Actions.
[3] Leyendecker EV, Ellingwood BR. “Design Methods for Reducing the Risk of Progres-
sive Collapse in Buildings”, Building Science Series, No. 98, National Bureau of Stand-
ards, Washington, DC, 1976.
[4] Mainstone RJ. The hazards of Explosion. Impact and Other Random Loadings on Tall
Buildings, Current Paper 64/74, Building Resch. Establishment, Garston, UK, 1974.
[5] ELS (2010). Extreme loading for structures (Version 3.1). Durham, NC: ASI.
[6] CODEC (2012). “Structural conception and collapse control performance based design
of multistory structures under accidental actions” (2012–2016), Executive Agency for
Higher Ed., Research, Development and Innovation Funding, Romania, PN II PCCA
55/2012.
[7] US Departments of the Army, Navy and Airforce (1990). Technical Manual, Army
TM5-1300, Navy NAVFAC P-397, Air Force AFR 88–22, Structures to resist the ef-
fects of accidental explosions. Washington, DC: US Dep. of Commerce, Nat. Techn.
Inf. Service.
[8] Karlos V, Solomos G. “Calculation of Blast Loads for Application to Structural Com-
ponents”, JRC 32253-2011 with DG-HOME Activity A5 - Blast Simulation Technology
Development, EU, Institute for the Protection and Security of the Citizen, 2013.
[9] Hopkinson B. British Ordnance Board Minutes 13565, 1915.
[10] Cranz C. Lehrbuch der Ballistik, Springer-Verlag, Berlin, 1926.
[11] Richards AB, Moore AJ. Blast vibration course measurement - assessment – control.
Terrock Pty Ltd, 2005.
[12] Tagel-Din H, Meguro K. “Applied element method for simulation of nonlinear materi-
als: Theory and application for RC structures”, Structural Engineering/Earthquake En-
gineering, International Journal of the Japan Society of Civil Engineers (JSCE), 17,
137–148, 2000.
[13] Kaneko H. “Influence of strain-rate on yield ratio, Kobe Earthquake Damage to Steel
Moment Connections and Suggested Improvement”, JSSC Technical Report No. 39, 1997.