This paper presents an overview of research conducted by the authors and others both experimentally and analytically in the area of blast wave interactions with buildings and structures in an urban landscape. Empirical tools for the assessment of blast wave resultants for simple-geometry situations are first presented before the confining effects of buildings along a street are discussed. Street layouts such a T-junctions and crossroads are then introduced and the way that blast wave characteristics are affected by street width and the height of buildings along the street is assessed. The influence of the location of the explosive charge within the street layout on the blast load experienced by buildings is also reviewed. It is likely that building façades closest to the explosion will fail and the effect on the loading experienced further from the detonation is therefore assessed. A complementary study of the loads developed inside a building whose façade has failed is also discussed. Apart from the blast loads on buildings in a street in which there is an explosion, buildings in adjacent streets will also experience blast loading. Studies concerned with the phenomena of ‘channelling’ and ‘shielding’ are presented and the extent to which both processes occur in any given situation discussed. The various studies reported throughout the paper indicate that simple tools are often inappropriate and the development of empirical rules for more complex geometries may not be reliable. This view is supported by studying blast wave propagation in a realistic cityscape comprising streets of different widths with several layout features all bounded by buildings of different size, shape and height. Such studies conclude that computational fluid dynamics analyses may sometimes be the only satisfactory approach for the engineer seeking a reliable assessment of blast loads experienced by buildings in an urban landscape.
When a single isolated building is loaded by the blast wave produced by the detonation of a quantity of high explosive, calculation of the pressure–time history experienced by the building is generally relatively straightforward, particularly for the side of the building directly facing the blast. Loading information can be obtained from graphs of blast resultants plotted against scaled distance contained in manuals such as TM5-13001 or TM5-855-12 and texts such as Baker et al.3, Smith & Hetherington4 or Mays & Smith5. Software such as ConWep6 and PSADS7, which have user-friendly interfaces, automate the process of blast load assessment for simple geometries. An example of the variation of incident and reflected pressure with range from a hemispherical charge comprising 200 kg TNT derived from the use of Ref.6 is shown in Fig. 1.
Remennikov8 presents a review of methods for predicting bomb blast effects on buildings that are essentially isolated, while von Rosen et al.9 present the results of an experimental and numerical investigation of the loading developed on the rear face of a simple rectangular structure when a blast wave impinges on the face of the building.
However, if the geometry of the scenario becomes more complex (e.g. when an explosive device is detonated in an urban environment where there are many buildings), assessment of the loading experienced by a particular building becomes more difficult. Such assessment becomes even more complicated should the façades of some buildings partly or completely fail, allowing the blast to enter the building. As Remennikov10 states: ‘Civil engineers today need guidance on how to design structural systems to withstand various acts of terrorism.’
This paper presents a review of the work in the area of blast wave propagation in a complex urban environment that has been undertaken by the authors and others at Cranfield University, Defence College of Management and Technology (formerly the Royal Military College of Science), Shrivenham since 1998 and also makes reference to complementary work by other investigators.
The influence of street configuration
One of the earliest investigations was by Feng as reported by Smith et al.11. In this study an approximately 1/50th scale model of a straight city street was constructed with buildings (made from appropriately sized reinforced concrete beams) along each side. Two separate buildings (made from 6 mm thick steel plate) were located at each end of the street and a small charge (replicating a large explosive device) was detonated in the middle of the street at different distances along it as illustrated in Fig. 2 where the bomb is shown located approximately one-third of the way along the street from the left-hand building. Reflected pressure–time histories were measured on the side of Building A in Fig. 2 facing the street and compared with measurements when the buildings along the street were removed. Fig. 3 shows the peak reflected pressure, Pr, measured by the gauge as the scaled distance, Z (= distance of the charge from Building A divided by the cube root of the TNT equivalent charge mass), increases. It was found that the presence of the street buildings increased the pressure (and also the impulse) experienced by the building by a factor of up to about four, thus clearly demonstrating the blast ‘confining’ effects produced by the buildings along a street. The increase in pressure will be at a maximum here because of the non-deforming nature of the very robust model building façades used in this investigation. In reality, this enhancement would be somewhat less, because of the likelihood of glazing failures in the vicinity of the explosion and further along the street. Even allowing for some façade failure, the important point is that this enhancement could not have been accurately predicted using simple tools.
The scope of this work was extended by Whalen, also reported in Ref.11. In this study, model street configurations—a crossroads, a T-intersection, a 90° bend and a cul-de-sac as well as a straight street—were built from steel plate. Small charges designed to replicate a large explosive device were detonated at the centre of the crossroads and the intersection, at the place where the street turned through 90°, at the end of the cul-de-sac and halfway along the straight street (marked ‘+’ in Fig. 4). Along every street the steel plates, representative of the buildings, were sufficiently tall such that no effects at street level were produced as a result of the blast reaching the top of the building. Pressure–time histories were recorded at a number of points on the ‘façade’ of one of the buildings as shown in Fig. 4 where a line adjacent to one of the façades in each configuration shows that it was equipped with an array of pressure transducers. The experimental waveforms were analysed using a ray-tracing technique to determine the origin of the peaks—sometimes there were several—on the record. Also, from the experimental records, impulses were calculated and compared with ‘free field’ calculations performed using ConWep. The results of this latter exercise clearly demonstrated the blast enhancing effect of different street configurations with the cul-de-sac producing the greatest enhancement and the crossroads configuration the least as demonstrated in Fig. 5 which shows the ratio of the reflected impulse ir measured in a confined street location (designated ir street configuration ) to that which would result with no confinement (designated ir UHAB where UHAB stands for Unconfined Hemispherical Air Burst) plotted against scaled range from the charge. The difference between the various geometries can be explained in terms of the volume available for the blast wave to expand into—the smaller the volume, the greater the density of blast energy and the larger the resulting blast load.
This study was complemented by a more detailed experimental and numerical investigation of straight streets using the blast simulation code Air3d12 by Smith & Rose13. In the relatively limited experimental study, small-scale straight streets of variable width w were bounded by buildings of variable height h. The set-up is shown schematically in Fig. 6, which indicates that the model streets were configured using a combination of concrete beams and blocks. The figure also shows a typical charge location and the eight places where pressure transducers could be installed flush with the model building façades. The streets were of sufficient length such that any ‘end effects’ associated with the blast reaching the end of the street were negligible. Experimental and numerical results were in good agreement (as evidenced by Fig. 7 where Air3d and experimental pressure–time histories at transducer location 8 in the wide street configuration are shown) and allowed further configurations to be analysed using Air3d. The main conclusions drawn from the investigation were that street width and building height influence the magnitude of the positive phase impulse. A scaled street width w/W1/3 > 4.8 m/kg1/3 (where W is TNT equivalent charge mass) is sufficiently wide such that the loading experienced by buildings on one side of the street is not affected by reflections from buildings on the other side. Fig. 8 shows scaled impulse vs scaled distance along streets bordered by infinitely high buildings for increasing scaled street width compared with a single infinitely large reflecting surface (labelled ‘true’) and when no confining buildings are present (labelled ‘side-on’). Figs. 8a–d show the results for scaled widths of 1.6, 2.4, 3.2 and 4.8 m/kg1/3, respectively. As street width increases, reflections from façades become less important and when w/W1/3 = 4.8 m/kg1/3 street and ‘true’ impulses coincide until a scaled distance of about 7 m/kg1/3. This implies that reflections are too weak or too late arriving to have a significant effect on the impulse. Figs. 9a–d show scaled positive phase impulse vs scaled distance along the street for the same values of scaled street width as Fig. 8. These graphs show that a scaled building height h/W1/3 >3.2 m/kg1/3 means that the buildings are effectively infinitely tall and there is little significant enhancement at street level of the positive phase impulse that is apparent at h/W1/3 = 3.2 m/kg1/3 when the scaled height is increased further. It was also shown in the same study that the negative phase impulse is enhanced because of the expansion of the blast wave when it reaches the top of the building. For practical purposes, the study showed that a value of h/W1/3 =3.2 m/kg1/3 can be used to define the situation when the effect is maximized as little change occurs for greater scaled heights. Finally, beyond a scaled distance from the charge of Z = 2.0 kg/m1/3, negative phase impulse exceeds the positive phase impulse for all street widths and all scaled building heights <12.8 m/kg1/3. This observation may go some way to explaining the anecdotal evidence that much of the glazing in city streets is drawn into the street by the passage of a blast wave following an explosion. Clearly, negative phase impulse cannot be ignored, since the region beyond Z=2.0 kg/m1/3 is likely to encompass most of the street.
A further study of the influence of street junctions on the characteristics of the blast that propagates into the streets beyond the junction was undertaken by Rose & Smith14 that used both a more comprehensive experimental programme than Whalen's11 as well as Air3d. The study concentrated on crossroads, T-intersections and 90° bends with the physical and numerical models configured in such a way that the streets could be considered infinitely long and relatively narrow (scaled width w/W1/3 = 1.6 m/kg1/3). Fig. 10 shows schematic views of the various configurations. In each of Figs. 10a–g, transducer locations are shown numbered 1–6 or 1–8 depending on the layout. Typical charge locations are marked by ‘X’ together with a letter (‘l’ for ‘straight-on’ and ‘r’ for ‘right angle’ configurations) and a number (‘0’ for ‘centred’ locations and 1, 2 or 3 for successively increasing equal increments from the ‘centred’ location).
Scaled building height h/W1/3 was kept constant at 3.2 m/kg1/3, a figure identified above as suitable for maximizing both positive and negative impulse at street level. Firstly, the results of this study confirmed the conclusions about the effect of confinement discussed above. Secondly, it was determined that the distance of the charge from the junction influences the extent to which the blast diffracts at the junction (and enters the other streets leading off the junction). The larger the distance of the charge from the junction, then the greater the degree of diffraction that occurs at the junction, as opposed to reflection and transmission back down the street in which the charge is located. Finally, it was observed that the intensity of the blast propagating round a 90° bend was similar to that which propagates along a straight street. It was also evident that, even in these relatively simple geometries the formulation of simple rules to predict blast resultants at particular locations could not be done with any confidence for either positive or negative phase impulses as indicated. Figs. 11a and b show the results for T- junctions. Fig. 11a shows the variation of scaled positive phase impulse vs scaled distance along the street for charges locations r0, r1 and r2 compared with the results for a straight street. Fig. 11b shows the variation of the scaled negative phase impulse with scaled distance.
These studies were complemented by those of Remennikov15 who investigated a scenario involving a simplified street geometry, producing results in support of those reported above. Additionally, Dörr et al.16 conducted a small-scale experimental study in which charges representing a terrorist explosive device were detonated in a street near T-junction and crossroads configurations. Their findings tended to support the studies discussed above in that they commented that blast reflections associated with these geometries increase the potential hazard of an explosive device detonated in city streets.
Effects of façade failure
The façade of a building will be the first element to experience a blast load and it is tempting to consider preventing severe building damage by provision of a very robust façade. As Sakula17 states: ‘The challenge for building designers in general and façade engineers in particular is the need to provide clients with informed and balanced advice on the appropriate level of protection without creating a world of fortress buildings’. This issue has been addressed by, for example, Ettouney et al.18 who comment: ‘As the building's exterior is its first real defence against the effects of a bomb, how the façade responds to the (blast) loading will significantly affect the behaviour of the building.’
The damaging effects of a blast wave propagating along a city street were mentioned above in connection with the influence of street width and building height. One question that might be asked is: how is blast propagation affected if, instead of presenting a solid surface to the incident blast wave, a building façade has some ‘holes’ in it? This situation could arise if an explosion caused some façade failure to a building at close range and blast entered the building through the failed region; less energy would then be available to propagate along the street. A study was conducted by Smith et al.19 using a 1/50th scale model of a long straight city street bounded by tall buildings. Both sides of the street were constructed in such a way as to allow different amounts of façade failure to be represented. Regular gaps in the façades (termed ‘porosity’) were created ranging from about 24–77% of the façade area and were chosen following a survey of real buildings to assess the proportion of the façade that could be damaged by the impact of a blast. Fig. 12 shows one side of a 48% porosity street façade constructed from brick, steel and plywood with pressure transducer ports at a number of locations. Twelve possible such locations are available as can be seen in the lower left-hand side of the model, though only four were ever used at one time. Experiments and simulations were also conducted with both sides of the street ‘solid’ and compared with results for the ‘porous’ facades. It was found that, at a particular location on the porous side of the street, there was an approximately linear decay in impulse delivered as the degree of porosity increased as shown in Fig. 13. This means that energy from the blast was removed as a consequence of it entering the building through the failed façade.
The ‘porosity’ study was complemented by an experimental and Air3d-based investigation of the loading developed inside a building in a straight street once a façade failure had allowed the blast to enter19. A 1/50th scale building, 10 storeys high, was constructed with a frangible façade and incorporated in the street. The other side of the street was ‘solid’. Pressure transducers were fitted flush with the surface of the material used to model the building ‘core’ (representing, say, a lift shaft in a real building) facing the street. The building (with porous façade removed) and transducer locations are shown pictorially in Fig. 14. The results of both the experimental and numerical investigations indicated that, if the blast has been allowed to enter the building, the resultant internal loading could be of sufficient magnitude to cause damage both to the building fabric and the building's occupants.
It is possible to express the results of these investigations in another way: all buildings that are potential targets should have façades that are as robust as possible. Some, close to the explosion, will undoubtedly fail but, beyond what is likely to be a relatively small distance from the explosion, if façade integrity can be maintained (and the building frame is sufficiently robust to accept the resulting increase in transmitted load), the resulting level of individual building damage will be reduced; this must be a desirable outcome.
Shielding and channelling
Prediction of blast wave characteristics becomes more complicated as the number of buildings in the urban (or suburban) environment interacting with the propagating blast increases. Intuitively, it might be expected that, when an explosive device is detonated in the heart of an urban environment, buildings close to the explosion would shield buildings in adjacent streets running parallel to the one in which the detonation occurred. Conversely, the presence of many streets in the vicinity of the explosion could, depending on their orientation, provide channels for the blast to run along and become enhanced as a result.
It has been acknowledged that to address these scenarios, there is a requirement for tools more sophisticated than ConWep. Tools that are available include AT Planner20 which, although it addresses the problem of complex geometries and provides an assessment of building damage, does not comprehensively account for the shielding effect of buildings between the explosive source and the ‘target’ structure. The program EBlast21 is an expert software system that provides damage and injury distances for a wide variety of explosive devices in urban environments. The program uses ‘enhancement factors’ (obtained by numerical modelling) to account for the effect of blast wave channelling in urban environments.
In order to gain a better insight into the phenomena of ‘shielding’ and ‘channelling’, Smith et al.22, 23 conducted investigations into blast wave propagation through arrays of buildings (replicating houses rather than large city buildings though, suitably scaled, the investigation could represent a cityscape) by means of small-scale experiments and Air3d simulations.
In the first of these two studies, several parallel lines of terraced houses and regular arrays of detached houses were studied. Streets ran between the lines of the terraced houses and the rows of the detached houses. Fig. 15a shows the experimental arrangement for terraced houses and Fig. 15b shows the corresponding set-up for an array of detached houses realized with a combination of concrete beams and blocks. Experimental and numerical measurements of blast wave resultants were made at a number of locations in the different arrays. In both studies, it was observed that the presence of a building between the explosion and the ‘target’ building produced a shielding effect and the blast resultants were reduced compared to the situation where there was no intervening building. As an example of the results obtained, the histograms of Fig. 16 illustrate the situation for terraced houses. The numbers below the histograms are the percentage of the blast pressure loading buildings of different height h (= 4, 8 and 16 m, representing single, double and four storey buildings) in different rows (Row 2 being nearer the explosion than Row 4; Row 1 was adjacent to the explosion) compared to pressures that would have been produced if no buildings were located between the charge and the ‘target’ building. Similar results were obtained for impulse reduction. Most effective shielding was evident when the charge was small and the buildings were tall and close together. Shielding was least when the charge was large and the buildings small and widely spaced. It was consistently observed that the interaction of the blast with the first row of buildings was the dominant factor defining the extent of shielding to subsequent rows of buildings.
The second investigation involved the detonation of small charges in both regular and random arrays of buildings with different areal densities Ad. Areal density was defined as the ratio of the total ‘footprint’ area of an array to the total area occupied by the array. Blast resultants were measured experimentally and calculated using Air3d along a straight continuous reflecting surface just beyond the array of buildings. In some cases there was a ‘direct line of sight’ from the charge location to the reflecting surface and in others a ‘direct line of sight’ did not exist. Fig. 17 shows arrays of model buildings on the explosives range at Shrivenham: Fig. 17a shows a regular array while Fig. 17b shows a random array. In both illustrations, pressure–time histories were measured along the solid reflecting surface outside the arrays. Fig. 18 shows schematically the random array with an areal density of 28.6%. Firstly, the findings discussed in Ref.22 were confirmed; no matter what the configuration of the buildings, their presence between a detonation and a measuring location will produce a reduction in blast loading. Secondly, it was observed that the differences in the total loading produced on the whole of the reflecting surface for different values of Ad and between ‘regular’ and ‘random’ and ‘direct line of sight’ and ‘no direct line of sight’ arrays are small. On average, a reduction of just over 10% in the load developed on the measuring surface was observed compared with when no buildings were present. This result suggests that two processes are occurring in any given array and the overall effect at the measuring surface is as a result of a combination of shielding and channelling. This process is illustrated schematically in Fig. 19. In other words, in a complex array of identical buildings, the blast reducing effects of shielding are offset by the blast enhancing effects of channelling.
The results reported above relate to relatively simple geometries in terms of both building size and shape and street layouts. For more detailed assessment in cases where the buildings are not small, identical units and the street layout is complex, it may be necessary to carry out a full numerical simulation of a specific scenario. As discussed above, investigators have sought to provide guidance by generating empirical and semi-empirical data based on dimensions such as street width and building height, which are significant in describing the blast environment in urban and suburban settings. However, these simple rules seem to be of diminishing usefulness if the actual building geometry and street layout differs significantly from the simple arrangements which provide the source data for the rules.
In order to address this perceived difficulty, Smith et al.24 present the results of an experimental and numerical investigation of a generic cityscape containing buildings of different size and shape and streets of different width and orientation and with charges detonated at a number of locations. Fig. 20 shows the geometry investigated rendered using ARCon+ software25, with some of the buildings ‘cut away’ for clarity and with the charge locations indicated. A schematic view of the scenario with a charge at location 1 is shown in Fig. 21. Three complex pressure–time histories (recorded in three separate experiments at the Gauge 21) are presented in Fig. 22 where the repeatability of results should be noted. The experimental investigation was carried out using buildings at 1/50th scale and the simulation was conducted with the most recent version of the Air3d code, Air3d which includes adaptive mesh refinement as discussed by Rose & Smith26. Fig. 23a shows Air3d-generated pressure (in kPa) and Fig. 23b impulse (in kPa-ms) contours on the façade of Building 6 where it will be seen that the loading is, as expected, non-uniform. However, it should also be noted that regions of high and low loading do not necessarily occur where they might intuitively be expected; ‘hot spots’ occur where a building surface might be expected to be shielded and relatively low loads are evident where a direct line from charge to ‘target’ might be expected to produce a higher load. The results of the study concluded that, to provide the good level of detail that would be required by the hazard assessor in complex-geometry urban environments, the use of computational fluid dynamics (CFD) techniques is likely to offer the best way ahead.
Such an approach is endorsed by a number of investigators including Löhner & Baum27 who concluded that ‘survivability assessments for buildings can nowadays be carried out on PCs in a reasonable amount of time.’ Such assessments ‘should provide substantially better predictions than ‘line of sight’ calculations, in particular for the more complex geometries that may entail shock focusing and reflections.’ Fairlie et al.28 also advocate the use of a three-dimensional CFD code—in this case AUTODYN3D—for blast loading on buildings ‘in more complex geometries typical of a modern urban environment’. In the paper noted above, Remennikov15 indicated that the use of both analytical techniques and sophisticated numerical simulations can provide an effective approach to determining blast loads in an urban environment.
It is reasonable to suggest that well-resolved, three-dimensional CFD analyses of blast wave propagation in complex urban environments will be commonplace among the structural engineering community within a few years.
The overview presented above leads to the following conclusions.
For simple geometries, simple tools could be acceptable.
For blast propagation along relatively simple-geometry city streets, rules can be formulated to predict blast resultants on building façades.
For more complex city street layouts, such rules become difficult to develop.
When buildings bounding streets respond and façades fail, any such rules must be altered.
When blast propagates through arrays of simple buildings, both shielding and channelling effects occur.
For ‘real’ cityscapes, reasonably accurate prediction of blast resultants may require numerical simulation.
Finally, the reader's attention is drawn to Smith & Rose29 that contains a more fundamental review of interaction of a blast wave with a building structure together with an introduction to ways of developing building robustness and other techniques to improve their chances of surviving the blast loading generated by the detonation of a high explosive charge.