Experimental and Numerical Study of Polymer-Retroﬁtted Masonry Walls under Gas Explosions

: Unreinforced masonry walls are extensively used in the petrochemical industry and they are one of the most vulnerable components to blast loads. To investigate the failure modes and improve the blast resistances of masonry walls, four full-scale ﬁeld tests were conducted using unreinforced and spray-on polyurea-reinforced masonry walls subjected to gas explosions. The results suggested that the primary damage of the unreinforced masonry wall was ﬂexural deformation and the wall collapsed at the latter stage of gas explosion. The presence of polyurea coatings could effectively improve the anti-explosion abilities of masonry walls, prevent wall collapses, and retain the ﬂying fragments, which would reduce the casualties and economic losses caused by petrochemical explosion accidents. The bond between the polymer and masonry wall was critical, and premature debonding resulted in a failure of the coating to exert the maximum energy absorption effect. A numerical model for masonry walls was developed in ANSYS/LS-Dyna and validated with the test data. Parametric studies were conducted to explore the inﬂuences of the polyurea-coating thickness and spray pattern on the performances of masonry walls. The polyurea-coating thickness and spray pattern affected the resistance capacities of masonry walls signiﬁcantly.


Introduction
Most materials involved in petrochemical plants are flammable and explosive hydrocarbon liquids and gases. Leakage can lead to vapor cloud or gas explosions, which creates a great hazard to existing structures and factory staff. Meanwhile, for historical reasons, unreinforced masonry walls have been extensively used in the petrochemical industry as non-load-bearing components in reinforced-concrete frame buildings, and they are one of the most vulnerable components to blast loads. Most casualties and injuries sustained during external explosions are not caused by the direct effect of the explosion, but rather by the disintegration and fragmentation of a wall, which can be propelled at high velocities by the blast [1]. To reduce the potential hazards to employee safety, it is necessary to study the performances of masonry walls under explosions and find a cost-effective strengthening method.
Due to the potential threats of terrorist attacks and accidental explosions, many researchers have focused on the behaviors and dynamic responses of masonry structures under blast loads. Shi et al. [2] conducted a series of field experiments to investigate the failure mechanisms and fragment characteristics of unreinforced masonry walls. The fragment size distribution was also analyzed, and  [13,14] 2005 CMU walls Sprayed-on polymer Explosive DYNA3D Alsayed et al. [8] 2016 CMU walls GFRP Explosive AUTODYN Elsanadedy et al. [9] 2016 CMU walls GFRP Static YES Wang et al. [15] 2016 Clay brick, AAC Sprayed-on polymer Explosive chen et al. [10] 2019 AAC panels CFRP/sprayed-on polymer Explosive Processes 2019, 7, 863 3 of 18 In this study, a series of full-scale field tests on masonry walls subjected to gas explosions were conducted. A simplified micro-model was also developed to predict the dynamic responses of polymer-retrofitted masonry walls using ANSYS/LS-DYNA. The predictions from the numerical simulations were compared with the test data to validate the numerical model. Parametric studies were carried out to evaluate the effects of the polyurea layer thickness and the spraying pattern on the responses of strengthened masonry walls subjected to gas explosions.

Experiments
A specially designed testing system was prepared for the gas explosion tests. The test system consisted of a pipeline, wall specimens, data-recording instruments, and gas distribution instruments, as shown in Figure 1. [8] 2016 CMU walls GFRP Explosive AUTODYN Elsanadedy et al. [9] 2016 CMU walls GFRP Static YES Wang et al. [15] 2016 Clay brick, AAC Sprayed-on polymer Explosive chen et al. [10] 2019 AAC panels CFRP/sprayedon polymer Explosive In this study, a series of full-scale field tests on masonry walls subjected to gas explosions were conducted. A simplified micro-model was also developed to predict the dynamic responses of polymer-retrofitted masonry walls using ANSYS/LS-DYNA. The predictions from the numerical simulations were compared with the test data to validate the numerical model. Parametric studies were carried out to evaluate the effects of the polyurea layer thickness and the spraying pattern on the responses of strengthened masonry walls subjected to gas explosions.

Experiments
A specially designed testing system was prepared for the gas explosion tests. The test system consisted of a pipeline, wall specimens, data-recording instruments, and gas distribution instruments, as shown in Figure 1.

Wall Specimens
One unstrengthened specimen and one rear-face-strengthened specimen were prepared. According to the Chinese standard (GB50003-2011), MU 15 clay bricks and M 5.0 mortar were used for the clay brick walls. The wall specimens with dimensions of 3 × 2 × 0.12 m were lumped with clay blocks with dimensions of 240 × 115 × 53 mm and 10-mm-thick mortar in a "running" pattern, as shown in Figure 2. Large wall dimensions were selected to avoid scale effects and to reflect the actual performance of a reinforced masonry unit wall under blast loading [15].

Wall Specimens
One unstrengthened specimen and one rear-face-strengthened specimen were prepared. According to the Chinese standard (GB50003-2011), MU 15 clay bricks and M 5.0 mortar were used for the clay brick walls. The wall specimens with dimensions of 3 × 2 × 0.12 m were lumped with clay blocks with dimensions of 240 × 115 × 53 mm and 10-mm-thick mortar in a "running" pattern, as shown in Figure 2. Large wall dimensions were selected to avoid scale effects and to reflect the actual performance of a reinforced masonry unit wall under blast loading [15].

Gas Distribution Instruments
To maximize the explosion overpressure, ethylene with a volume concentration of 5.5% was chosen as the combustible gas in the tests. The oxygen concentration in the mixture was measured by an oxygen content analyzer, and the ratio of fuel gas to air was adjusted over time to ensure uniform mixing of the flue gas and air.

Gas Distribution Instruments
To maximize the explosion overpressure, ethylene with a volume concentration of 5.5% was chosen as the combustible gas in the tests. The oxygen concentration in the mixture was measured by an oxygen content analyzer, and the ratio of fuel gas to air was adjusted over time to ensure uniform mixing of the flue gas and air.

Device Main Body
The device main body included a pipeline and flaring, as shown in Figures 3 and 4, respectively. The pipeline was used to store gas. Its diameter and length were 500 mm and 50 m, respectively. To create a more uniform explosion load on the wall surface, the flaring, with dimensions of 2 × 2 m, was installed at the end of the pipeline.

Gas Distribution Instruments
To maximize the explosion overpressure, ethylene with a volume concentration of 5.5% was chosen as the combustible gas in the tests. The oxygen concentration in the mixture was measured by an oxygen content analyzer, and the ratio of fuel gas to air was adjusted over time to ensure uniform mixing of the flue gas and air.

Device Main Body
The device main body included a pipeline and flaring, as shown in Figures 3 and 4, respectively. The pipeline was used to store gas. Its diameter and length were 500 mm and 50 m, respectively. To create a more uniform explosion load on the wall surface, the flaring, with dimensions of 2 × 2 m, was installed at the end of the pipeline.

Data-Recording Instruments
The data-measuring system was used to record the explosion overpressure, displacements, and wall failure modes. It was composed of a synchronous trigger instrument, data acquisition instrument, piezoelectric pressure sensors, displacement sensor, and high-speed cameras. The pressure sensors were used to measure the peak overpressure and positive pressure acting time of the explosion shockwaves, as shown in Figure 4. The measurement range was 0-10 MPa. The displacement sensor was used to measure the displacement of the central points of the test walls, as shown in Figure 5, with a measurement range of −5 to +5 cm. High-speed cameras were arranged on the left and right sides of the test walls to record the failure process of the wall during the test, and the speed of high-speed cameras were 6400 frames. The signals from the pressure sensor and displacement transducers were logged by the data acquisition instrument.

Data-Recording Instruments
The data-measuring system was used to record the explosion overpressure, displacements, and wall failure modes. It was composed of a synchronous trigger instrument, data acquisition instrument, piezoelectric pressure sensors, displacement sensor, and high-speed cameras. The pressure sensors were used to measure the peak overpressure and positive pressure acting time of the explosion shockwaves, as shown in Figure 4. The measurement range was 0-10 MPa. The displacement sensor was used to measure the displacement of the central points of the test walls, as shown in Figure 5, with a measurement range of −5 to +5 cm. High-speed cameras were arranged on the left and right sides of the test walls to record the failure process of the wall during the test, and the speed of high-speed cameras were 6400 frames. The signals from the pressure sensor and displacement transducers were logged by the data acquisition instrument. the explosion shockwaves, as shown in Figure 4. The measurement range was 0-10 MPa. The displacement sensor was used to measure the displacement of the central points of the test walls, as shown in Figure 5, with a measurement range of −5 to +5 cm. High-speed cameras were arranged on the left and right sides of the test walls to record the failure process of the wall during the test, and the speed of high-speed cameras were 6400 frames. The signals from the pressure sensor and displacement transducers were logged by the data acquisition instrument.

Test Procedure
The supporting structure was designed to fix the test walls on the ground. All the specimens were mounted 1 m away from the flare. The pressure sensors were installed at the end of the pipeline, and the displacement sensor was arranged at the center of the rear surface of the wall. The spark igniter was installed on the pipe inlet flange, which meant that the igniter was at the edge of the gas cloud. The ignition voltage of spark igniter was 1000 V. The ethylene was pumped from the cylinder gas tank to the pipeline and flaring until the target concentration was reached. Finally, the gas mixture was ignited by a high-voltage power source. The data acquisition system and high-speed cameras were triggered to record data simultaneously.

Test Procedure
The supporting structure was designed to fix the test walls on the ground. All the specimens were mounted 1 m away from the flare. The pressure sensors were installed at the end of the pipeline, and the displacement sensor was arranged at the center of the rear surface of the wall. The spark igniter was installed on the pipe inlet flange, which meant that the igniter was at the edge of the gas cloud. The ignition voltage of spark igniter was 1000 V. The ethylene was pumped from the cylinder gas tank to the pipeline and flaring until the target concentration was reached. Finally, the gas mixture was ignited by a high-voltage power source. The data acquisition system and high-speed cameras were triggered to record data simultaneously.

Unreinforced Wall
Explosion tests of the unreinforced masonry wall were conducted to analyze the failure mode and anti-explosion capacity of the wall. The test results of the masonry wall without an explosion-resistant coating under the first explosion (test 1) are shown in Figure 6. As shown in Figure 6a, large-area concave deformations appeared on the explosion face of the unreinforced wall, and the deformation was almost symmetric. The compression failure of the unstrengthened masonry wall was caused by the high reflected peak pressure. A ⊥-shaped penetrating crack appeared in the upper half of the wall, as shown in Figure 6b. The fracture lengths in the horizontal and vertical directions were approximately 1.8 and 1.4 m, respectively, and the widths of cracks were about 25 mm. Bricks almost fell from the central area of the back blasting surface, and cracks developed radially from the center of the wall to the surrounding area. The wall was severely damaged, which indicated that the explosive load was close to the ultimate load-bearing capacity of the unreinforced wall. The blast load time histories were also recorded during the test, and Figure 7 shows that the peak overpressure of the explosive shockwave was about 5.5 MPa. However, no displacement data were obtained due to the mechanical failure.
almost fell from the central area of the back blasting surface, and cracks developed radially from the center of the wall to the surrounding area. The wall was severely damaged, which indicated that the explosive load was close to the ultimate load-bearing capacity of the unreinforced wall. The blast load time histories were also recorded during the test, and Figure 7 shows that the peak overpressure of the explosive shockwave was about 5.5 MPa. However, no displacement data were obtained due to the mechanical failure.  To study the distribution of wall debris under explosive loading, subsequent tests were carried out on the wall. The peak overpressure of the explosive shockwave was about 5.2 MPa, as shown in Figure 8. The failure process of the unstrengthened wall specimen during the second explosion (test 2) is shown in Figure 9. Compared with the tests of high explosives in Table 1, the peak pressures from gas explosions were lower but the duration of pressure waves were longer, which lead to different structural responses. Local damages such as spalling or punching damage were more likely to occur when the masonry walls were under close-in TNT explosions as compared to the overall flexural deformations in gas explosions. The test results were similar to those carried out by Li et al. almost fell from the central area of the back blasting surface, and cracks developed radially from the center of the wall to the surrounding area. The wall was severely damaged, which indicated that the explosive load was close to the ultimate load-bearing capacity of the unreinforced wall. The blast load time histories were also recorded during the test, and Figure 7 shows that the peak overpressure of the explosive shockwave was about 5.5 MPa. However, no displacement data were obtained due to the mechanical failure.  To study the distribution of wall debris under explosive loading, subsequent tests were carried out on the wall. The peak overpressure of the explosive shockwave was about 5.2 MPa, as shown in Figure 8. The failure process of the unstrengthened wall specimen during the second explosion (test 2) is shown in Figure 9. Compared with the tests of high explosives in Table 1, the peak pressures from gas explosions were lower but the duration of pressure waves were longer, which lead to different structural responses. Local damages such as spalling or punching damage were more likely to occur when the masonry walls were under close-in TNT explosions as compared to the overall flexural deformations in gas explosions. The test results were similar to those carried out by Li et al. To study the distribution of wall debris under explosive loading, subsequent tests were carried out on the wall. The peak overpressure of the explosive shockwave was about 5.2 MPa, as shown in Figure 8. The failure process of the unstrengthened wall specimen during the second explosion (test 2) is shown in Figure 9. Compared with the tests of high explosives in Table 1, the peak pressures from gas explosions were lower but the duration of pressure waves were longer, which lead to different structural responses. Local damages such as spalling or punching damage were more likely to occur when the masonry walls were under close-in TNT explosions as compared to the overall flexural deformations in gas explosions. The test results were similar to those carried out by Li et al. [17], wall specimens experienced damage modes of flexural deformation and a through crack occurred at the middle of the walls. The crack developed with the rise of pressure and the wall specimens collapsed at the latter stage of gas explosion.
Processes 2019, 7, 863 7 of 19 [17], wall specimens experienced damage modes of flexural deformation and a through crack occurred at the middle of the walls. The crack developed with the rise of pressure and the wall specimens collapsed at the latter stage of gas explosion.    The fragment distributions of the wall specimen after the gas explosions are shown in Figure 10. The wall collapsed and broke along the bottom, top, and left joints of the frame, with only a 0.4-mwide wall remaining on the right side. The debris was dispersed in an area of approximately 7.6 m in length and 5 m in width. Masonry walls are often used as filling walls of frame structures. The debris produced by a masonry wall under a blast wave can cause severe injuries to indoor personnel.

Polymer-Retrofitted Wall
Test 3 was carried out to investigate the effectiveness of the polyurea layer and the performance of a polymer-retrofitted wall subjected to gas explosions. The thickness of the polyurea layer on the rear face was 6 mm. The dynamic response of the rear-face-strengthened wall specimen is shown in Figure 11.

Polymer-Retrofitted Wall
Test 3 was carried out to investigate the effectiveness of the polyurea layer and the performance of a polymer-retrofitted wall subjected to gas explosions. The thickness of the polyurea layer on the rear face was 6 mm. The dynamic response of the rear-face-strengthened wall specimen is shown in Figure 11.

Polymer-Retrofitted Wall
Test 3 was carried out to investigate the effectiveness of the polyurea layer and the performance of a polymer-retrofitted wall subjected to gas explosions. The thickness of the polyurea layer on the rear face was 6 mm. The dynamic response of the rear-face-strengthened wall specimen is shown in Figure 11.  Figure 12 shows photos of the damage of the rear-face-strengthened wall after the explosion. The wall was concave and deformed slightly, with X-shaped cracks appearing in the central area of the front face. The maximum plastic deformation at the bottom was about 0.06 m. The polyurea coating on the back blasting surface was intact, and no cracks formed. This suggested that the presence of the polyurea layer could effectively improve the blast resistance of the unreinforced wall and reduce the wall failure to local bending deformation.  As shown in Figures 13 and 14, the peak overpressure in tests 3 and 4 was 6.0 and 5.5 MPa, respectively. The peak pressures in all the tests were basically the same, which indicated that the gas mixture was relatively uniform during each test. Figure 15 shows photographs of the damage to the masonry wall after test 4. The reinforced brick wall collapsed. However, compared to the unreinforced wall, the traveling distance of the debris was significantly reduced, and the debris was almost completely distributed in the framework of the supporting body. The polyurea coating peeled off along the left, right, and lower edges of the frame. The coatings sprayed on the upper part of the frame were partially torn, whereas the remaining coatings were basically intact. The polyurea layer could absorb the explosive energy, thereby reducing the kinetic energy of the debris and preventing high-speed wall fragments from harming indoor personnel. However, in this test, premature debonding of the coating occurred, and the wall failed to produce the maximum effect of the coating. The temperature on the day of coating spraying was 9-12 °C, the wind force level was 4-5, and the humidity was about 52%. This may have been due to the low temperature of the outdoor spraying operation, causing the coating to not adhere to the frame well. The spraying operation should be carried out under suitable temperature and humidity conditions to improve the adhesion of the As shown in Figures 13 and 14, the peak overpressure in tests 3 and 4 was 6.0 and 5.5 MPa, respectively. The peak pressures in all the tests were basically the same, which indicated that the gas mixture was relatively uniform during each test. Figure 15 shows photographs of the damage to the masonry wall after test 4. The reinforced brick wall collapsed. However, compared to the unreinforced wall, the traveling distance of the debris was significantly reduced, and the debris was almost completely distributed in the framework of the supporting body. The polyurea coating peeled off along the left, right, and lower edges of the frame. The coatings sprayed on the upper part of the frame were partially torn, whereas the remaining coatings were basically intact. The polyurea layer could absorb the explosive energy, thereby reducing the kinetic energy of the debris and preventing high-speed wall fragments from harming indoor personnel. However, in this test, Processes 2019, 7, 863 9 of 18 premature debonding of the coating occurred, and the wall failed to produce the maximum effect of the coating. The temperature on the day of coating spraying was 9-12 • C, the wind force level was 4-5, and the humidity was about 52%. This may have been due to the low temperature of the outdoor spraying operation, causing the coating to not adhere to the frame well. The spraying operation should be carried out under suitable temperature and humidity conditions to improve the adhesion of the coating. Furthermore, steel angle can also be used to reinforce the coating around the frame to maximize the energy absorption effect of the coating. In order to avoid the loss of displacement data during violent explosion, non-contact digital image correlation (DIC) technology can be used to carry out the real-time monitoring of the surface displacement of the wall in future experiments. masonry wall after test 4. The reinforced brick wall collapsed. However, compared to the unreinforced wall, the traveling distance of the debris was significantly reduced, and the debris was almost completely distributed in the framework of the supporting body. The polyurea coating peeled off along the left, right, and lower edges of the frame. The coatings sprayed on the upper part of the frame were partially torn, whereas the remaining coatings were basically intact. The polyurea layer could absorb the explosive energy, thereby reducing the kinetic energy of the debris and preventing high-speed wall fragments from harming indoor personnel. However, in this test, premature debonding of the coating occurred, and the wall failed to produce the maximum effect of the coating. The temperature on the day of coating spraying was 9-12 °C, the wind force level was 4-5, and the humidity was about 52%. This may have been due to the low temperature of the outdoor spraying operation, causing the coating to not adhere to the frame well. The spraying operation should be carried out under suitable temperature and humidity conditions to improve the adhesion of the coating. Furthermore, steel angle can also be used to reinforce the coating around the frame to maximize the energy absorption effect of the coating. In order to avoid the loss of displacement data during violent explosion, non-contact digital image correlation (DIC) technology can be used to carry out the real-time monitoring of the surface displacement of the wall in future experiments.

Material Parameters
The anisotropic material model *Mat _96 (MAT BRITTLE DAM-AGE) in LS_DYNA has been used and validated in many studies to model a variety of brittle materials under explosions. Therefore, it was adopted for the clay bricks and mortar in this study. The parameters of the blocks

Material Parameters
The anisotropic material model *Mat _96 (MAT BRITTLE DAM-AGE) in LS_DYNA has been used and validated in many studies to model a variety of brittle materials under explosions. Therefore, it was adopted for the clay bricks and mortar in this study. The parameters of the blocks

Material Parameters
The anisotropic material model *Mat _96 (MAT BRITTLE DAM-AGE) in LS_DYNA has been used and validated in many studies to model a variety of brittle materials under explosions. Therefore, it was adopted for the clay bricks and mortar in this study. The parameters of the blocks and mortar used in the numerical simulations are listed in Table 2 [15,[17][18][19][20]. The MAT PIECEWISE LINEAR PLASTICITY material model was used for the polyurea coating. As provided by the supplier, the composite material had a tensile strength of 15.4 MPa, breaking elongation ratio of 200%, and tear resistance of 105 N/m. The mechanical properties of the composite material are summarized in Table 3 [21].

Geometric Model
The simplified micro-models for the masonry walls were developed in LS-DYNA, and the mortar, blocks, and polyurea layer were modeled separately. The nodes on their interfaces were merged. The wall model dimensions were a thickness of 120 mm, height of 2 m, and width of 3 m to match the tests described above, as shown in Figure 2. The sizes of blocks were 240 × 115 × 53 mm, and the thicknesses of the mortar and polyurea layer were 10 and 6 mm, respectively, identical to the thicknesses in the test specimens. The three-dimensional solid element SOLID164 was used to discretize the geometrical model. Considering the computational accuracy and efficiency, the combination of a 20-mm mesh size for the blocks and one layer for the mortar joint and polyurea coating was adopted for the numerical model.

Numerical Model Calibration
In the tests, the pressure sensor was located at the end of the pipeline. To obtain the explosion overpressure value at the wall, after the tests, a pressure sensor with a range of 0-1 MPa was also arranged at a distance of 1 m from the flaring. The sensor was destroyed in the first measurement and reached the maximum range in the second measurement, but it was not destroyed. According to the information provided by the supplier, the damage pressure of the sensor was 2 MPa, and the positive-pressure-action time measured by the sensor at the end of the pipeline was about 2 ms. These were used as the explosion loads in the numerical simulations. Figure 16 shows the comparison between the numerical simulation and the experimental results of the unreinforced masonry wall in test 1. A large depression area and deformation appeared on the face of the explosion. The numerical simulation predicted depressions caused by the blast wave, as shown in Figure 16a. The destruction area of the front surface of the wall was basically symmetric, and the edge of the area was arc shaped. A ⊥-shaped crack penetrated the wall in the test. The calculation results showed that the same main crack penetrated to the back of the masonry wall, as illustrated in Figure 16b, and the brick in the center of the wall almost fell out. The fracture lengths in the horizontal and vertical directions were 1.78 and 1.5 m, respectively, and the widths of cracks were about 23 mm. The values of the three parameters in the test were 1.8 m, 1.4 m, and 25 mm respectively. The results of the simulation are basically consistent with the experimental data.
of the unreinforced masonry wall in test 1. A large depression area and deformation appeared on the face of the explosion. The numerical simulation predicted depressions caused by the blast wave, as shown in Figure 16a. The destruction area of the front surface of the wall was basically symmetric, and the edge of the area was arc shaped. A ⊥-shaped crack penetrated the wall in the test. The calculation results showed that the same main crack penetrated to the back of the masonry wall, as illustrated in Figure 16b, and the brick in the center of the wall almost fell out. The fracture lengths in the horizontal and vertical directions were 1.78 and 1.5 m, respectively, and the widths of cracks were about 23 mm. The values of the three parameters in the test were 1.8 m, 1.4 m, and 25 mm respectively. The results of the simulation are basically consistent with the experimental data.  Figure 17 shows the numerical simulation results of the polymer-retrofitted masonry wall in test 3. In Figure 17a, the deformation area of the central area of the wall was approximately 0.12 m 2 , and the numerical simulation predicted a value of approximately 0.14 m 2 . In Figure 17b, the coating on the back burst surface remained intact, and the calculated results showed that the coating was also intact, with slight arching in the center of the coating. Therefore, the numerical model developed in this study yielded acceptable predictions and could be used to predict the performances of polymerretrofitted masonry walls against gas explosions.  Figure 17 shows the numerical simulation results of the polymer-retrofitted masonry wall in test 3. In Figure 17a, the deformation area of the central area of the wall was approximately 0.12 m 2 , and the numerical simulation predicted a value of approximately 0.14 m 2 . In Figure 17b, the coating on the back burst surface remained intact, and the calculated results showed that the coating was also intact, with slight arching in the center of the coating. Therefore, the numerical model developed in this study yielded acceptable predictions and could be used to predict the performances of polymer-retrofitted masonry walls against gas explosions.

Numerical Simulations
The masonry walls of petrochemical buildings are larger than the specimens used in this study, and most are 240 mm thick. Therefore, based on the calibrated models, numerical models with dimensions of 4 m × 3 m × 240 mm were established according to the pattern shown in Figure 18. In the simulation calculations, the peak overpressure was 69 kPa, and the positive-pressure time was 20 ms [22,23].

Numerical Simulations
The masonry walls of petrochemical buildings are larger than the specimens used in this study, and most are 240 mm thick. Therefore, based on the calibrated models, numerical models with dimensions of 4 m × 3 m × 240 mm were established according to the pattern shown in Figure 18. In the simulation calculations, the peak overpressure was 69 kPa, and the positive-pressure time was 20 ms [22,23].
The masonry walls of petrochemical buildings are larger than the specimens used in this study, and most are 240 mm thick. Therefore, based on the calibrated models, numerical models with dimensions of 4 m × 3 m × 240 mm were established according to the pattern shown in Figure 18. In the simulation calculations, the peak overpressure was 69 kPa, and the positive-pressure time was 20 ms [22,23]. According to the Chinese standard (GB50779-2012) [22], the peak reflected pressure can be calculated as follows: where Pr is the peak reflected pressure (kPa). The stagnation pressure can be expressed as where Ps is the stagnation pressure (kPa), and Cd is the drag coefficient. The drag coefficient depends on the shape and orientation of the obstructing surface. For a rectangular building, the drag coefficient may be taken as +1.0 for the front wall and −0.4 for the side and rear walls and the roof. The duration of the equivalent triangle was calculated as follows: According to the Chinese standard (GB50779-2012) [22], the peak reflected pressure can be calculated as follows: where P r is the peak reflected pressure (kPa). The stagnation pressure can be expressed as where P s is the stagnation pressure (kPa), and C d is the drag coefficient. The drag coefficient depends on the shape and orientation of the obstructing surface. For a rectangular building, the drag coefficient may be taken as +1.0 for the front wall and −0.4 for the side and rear walls and the roof. The duration of the equivalent triangle was calculated as follows:  (5) where Iw is the positive pressure impulse, S is the clearing distance, which is the smaller of H or B/2(m), te is the duration of the equivalent triangle (s), and tc is the duration of the reflected overpressure effect (s). The calculated front wall loading is shown in Figure 19.  Figure 20 shows the structural response of an unsprayed wall. According to the figure, the brick wall collapsed. Selecting the center point of the wall as the observation point, the displacement curve is shown in Figure 21. In the range of 0-0.2 s, the displacement tended to increase significantly, and the maximum displacement was 0.7 m, which was larger than the thickness of the clay brick wall. The velocity curve is shown in Figure 22. As shown in the figure, the wall velocity increased first, after which it remained basically constant, reaching a maximum velocity of 3.84 m/s at 0.028 s.  Figure 20 shows the structural response of an unsprayed wall. According to the figure, the brick wall collapsed. Selecting the center point of the wall as the observation point, the displacement curve is shown in Figure 21. In the range of 0-0.2 s, the displacement tended to increase significantly, and the maximum displacement was 0.7 m, which was larger than the thickness of the clay brick wall. The velocity curve is shown in Figure 22. As shown in the figure, the wall velocity increased first, after which it remained basically constant, reaching a maximum velocity of 3.84 m/s at 0.028 s. Figure 20 shows the structural response of an unsprayed wall. According to the figure, the brick wall collapsed. Selecting the center point of the wall as the observation point, the displacement curve is shown in Figure 21. In the range of 0-0.2 s, the displacement tended to increase significantly, and the maximum displacement was 0.7 m, which was larger than the thickness of the clay brick wall. The velocity curve is shown in Figure 22. As shown in the figure, the wall velocity increased first, after which it remained basically constant, reaching a maximum velocity of 3.84 m/s at 0.028 s.

Coating Thickness
Masonry walls strengthened at the rear face with layer thicknesses of 3, 6, and 8 mm were considered. Figure 23 shows the response of the strengthened wall with a 6-mm coating. The mortar and bricks were damaged by the blast loading, but the coating remained intact and effectively reduced the formation of wall debris.

Coating Thickness
Masonry walls strengthened at the rear face with layer thicknesses of 3, 6, and 8 mm were considered. Figure 23 shows the response of the strengthened wall with a 6-mm coating. The mortar and bricks were damaged by the blast loading, but the coating remained intact and effectively reduced the formation of wall debris.

Coating Thickness
Masonry walls strengthened at the rear face with layer thicknesses of 3, 6, and 8 mm were considered. Figure 23 shows the response of the strengthened wall with a 6-mm coating. The mortar and bricks were damaged by the blast loading, but the coating remained intact and effectively reduced the formation of wall debris.
The displacement histories at the centers of masonry walls are shown in Figure 24. The notation 0-3, 0-6, and 0-8 represent the polyurea coatings with corresponding thickness (mm) sprayed on the rear surface of clay brick wall and no coating on the front surface. The displacement of all the reinforced walls increased first and subsequently decreased gradually. The maximum displacements of the 0-3, 0-6, and 0-8 walls were 0.29, 0.21, and 0.19 m, respectively. With the increase in the coating thickness, the maximum displacement of wall decreased, which indicated that the anti-explosion ability was significantly improved. The kinetic and internal energies of the coatings are shown in Figures 25  and 26, respectively. The internal energy absorbed by the coating was much greater than the kinetic energy. With the increase in the coating thickness, the amount of wall deformation decreased, the internal energy of the coating decreased, and the wall absorbed more explosive energy through the destruction of the mortar or brick.

Coating Thickness
Masonry walls strengthened at the rear face with layer thicknesses of 3, 6, and 8 mm were considered. Figure 23 shows the response of the strengthened wall with a 6-mm coating. The mortar and bricks were damaged by the blast loading, but the coating remained intact and effectively reduced the formation of wall debris.

Spray Pattern
To study the effect of the spraying method on the explosion resistance of the reinforced walls, 3-mm-thick explosion resistant coatings were sprayed on the front (3-0), back (0-3), and both sides of the clay brick walls .
The coatings of the walls strengthened using three different spray patterns were not damaged under explosive loading. The side views of the reinforced walls are shown in Figure 27. When the polyurea was sprayed only on the front face of the masonry wall, debris was launched from the back of the wall.
The displacement curves of the different spraying methods are shown in Figure 28. The maximum displacements of the 3-0, 0-3, and 3-3 walls were 0.55, 0.29, and 0.11 m, respectively. This suggested that the masonry wall with the double-side reinforcement exhibited the best explosion resistance. For the 3-3 wall, the thickness of the coating on the front side was the same as that on the back side. The kinetic and internal energies of the coating on the front side were larger than the corresponding values of the back side, as illustrated in Figures 29 and 30, which showed that the energy absorption effect of the front face coating was better than that of the back face coating.

Spray Pattern
To study the effect of the spraying method on the explosion resistance of the reinforced walls, 3mm-thick explosion resistant coatings were sprayed on the front (3-0), back (0-3), and both sides of the clay brick walls .
The coatings of the walls strengthened using three different spray patterns were not damaged under explosive loading. The side views of the reinforced walls are shown in Figure 27. When the polyurea was sprayed only on the front face of the masonry wall, debris was launched from the back of the wall. The displacement curves of the different spraying methods are shown in Figure 28. The maximum displacements of the 3-0, 0-3, and 3-3 walls were 0.55, 0.29, and 0.11 m, respectively. This suggested that the masonry wall with the double-side reinforcement exhibited the best explosion resistance. For the 3-3 wall, the thickness of the coating on the front side was the same as that on the back side. The kinetic and internal energies of the coating on the front side were larger than the corresponding values of the back side, as illustrated in Figures 29 and 30, which showed that the energy absorption effect of the front face coating was better than that of the back face coating.
The displacements of the walls strengthened with the same thicknesses are shown in Figure 31. The maximum displacements of the 0-6, 3-3, and 2-4 walls were 0.21, 0.11, and 0.09 m, respectively. corresponding values of the back side, as illustrated in Figures 29 and 30, which showed that the energy absorption effect of the front face coating was better than that of the back face coating.
The displacements of the walls strengthened with the same thicknesses are shown in Figure 31. The maximum displacements of the 0-6, 3-3, and 2-4 walls were 0.21, 0.11, and 0.09 m, respectively. The explosion resistances of the masonry walls could be further improved by optimizing the combination of the coating spraying method and thickness.    The displacements of the walls strengthened with the same thicknesses are shown in Figure 31. The maximum displacements of the 0-6, 3-3, and 2-4 walls were 0.21, 0.11, and 0.09 m, respectively. The explosion resistances of the masonry walls could be further improved by optimizing the combination of the coating spraying method and thickness.

Conclusions
In this study, the dynamic impact responses of walls strengthened with polyuria layers were investigated through a series of impact tests under the blast loads generated by gas explosions instead of high explosives. The calibrated numerical model was conducted on parametric studies. The conclusions are summarized as follows:

Conclusions
In this study, the dynamic impact responses of walls strengthened with polyuria layers were investigated through a series of impact tests under the blast loads generated by gas explosions instead of high explosives. The calibrated numerical model was conducted on parametric studies. The conclusions are summarized as follows: (1) The wall damage modes under gas explosions were analyzed, the masonry walls experienced bending damage and then collapsed. The presence of the polyurea coatings could effectively improve the anti-explosion abilities of masonry walls, prevent wall collapses, and retain the flying fragments. Excellent reinforcement effect and no influence on normal production make polyurea coating an alternative method to strengthen existing petrochemical buildings. (2) The bond between the polymer and masonry wall is critical for the reinforcement effectiveness.
Premature debonding results in a failure of the coating to exert the maximum energy absorption effect. Spraying operations should be carried out under suitable temperature and humidity conditions to improve the adhesion of the coating. Furthermore, steel angle can also be used to reinforce the coating around the frame. (3) When polyurea was sprayed only on the front face of a masonry wall, debris was still launched from the back of the wall. The blast-resistant ability of the front polyurea-coated wall was not significantly enhanced, and it was weaker than the rear and double-side polyurea-coated walls. (4) Compared with the rear polyurea-coated wall, the blast-resistant ability of the double-side polyurea-coated wall improved significantly. Double-sided reinforcement should be used in building reinforcement to improve the wall explosion resistance. Rear-face reinforcement can be selected when construction conditions are limited. In addition, the explosion resistance of masonry walls can be further improved by optimizing the combination of the spray method and coating thickness Subsequent tests can be carried out to investigate characteristics of wall fragments, proposing an analytical model to accurately predict the debris distribution at low computational cost. Several other parameters can be included in future tests such as elongation, tensile strength, tear strength, and hardness of polyurea coating. The performance criteria for coatings could be optimized to better blast reinforcement.