werel Jt 2b pag Design of My Blast Resistant Buildings in Petrochemical Facilities Taek Cone ON BLAST RESISTANT DESIGN PETMOCHEMICAL COMMITTEE of the ENERGY DIVISION of the AMERICAN SOCIETY OF CIVIL ENGINEERS TECNIMONT / EDILE 2 4 Lut i999 TECNIMONT's property Pro. N? SHG Published by ASCE treicaseer4 i peSRaN aS a f Ths or ov gee uti in incest i stn eg of ast preene n, Seren wn mae cee sean es Eero da acl ergy leva pe eter magne et os FL tear otconrs Csgg iin tet eit ls pecan el i troonat ocean ly weedy he Tk aoe ont ea ae ofthe Exergy Divine aera Secor eh oe Feowatttonae {:Ruoknefneite-Depa desertion, 2. Perch ta {ieeprtinsefeae“Wameienreetee ett “tot Coat Sa tog ens Ts iE Ameen se fC Eager z SONSIode op ‘a Comme cn Sa Resoan Geogn dorm Ssy of Ci ngnces ASCB oa ee saat oan female, wpm, edi err ha E ‘sige saat een ys tn nang rrene a Attu cy mat FB SeSorven tame Coat) Tratcin Beret ASE Hod eh wee OB seSore ctartna toa Rein Seve, proves athebae oehieo dis Sepsens® Se 2 men Be vena SS Ree eis HEAL STBLO 9 SS gree DesSGeses PermsnesCoyi en, ASCE AMOS!" EEA prin cpg hoa an under cams ing SIRE ge ou Amen Soy of et ane rayne Coa Cua at ISBN 0-7R66.0255.5 i = ASCE Petrochemical Energy Committee ‘This publication is one of five state-of the-practce engineering ceports produced, to ite, by the ASCE Petrochemical Energy Committee, These enginceting reports are intended to be a summary of curent engineering knowledge and design practice, and Present guidelines for the design of petrochemical faclites. They representa consensus pinion of task committee members active in their development. ‘These five ASCE engineering reports are '). Design of Anchor Bolts in Petrochemical Facilities 2) Design of Blast Ressirs Buildings i Petrochemical Facies 3). Dasign of Secondary Containment in Petrochemical Faciluies 48) Guidelines for Seismic Evaluation and Design of Petrochemical Facies 5) Wind Loads on Petrochemical Facilites The ASCE Petrochemical Energy Commitee was organized by A. K, Guplain 1991 and initially chaired by Curley Tumer. Under their leadership the five tase eommittoes ‘ete formed. More recently, the Committe hasbeen chaired by Joseph A. Bohinsky and rank J, Hai, Frank J, Hse J. Marcell Hunt ‘Chevron Research and Technology Company Hudson Engineeing Corporation chairman secresary Joseph A. Behinsky Brown & Root, Ine Wiliam Bounds Fluor Danie, Inc Chay Flint Bechtel, nc John Geigel Bacon Chemical Company ‘Ajaya K Gupta ‘North Carolina State University Magdy H. Hanna Jacobs Engineering Group Steven R, Hemler Eastman Chemical Co, Gayle S. Johnson EQE Intemational, In, James Maple LA Maple & Associates Douglas J. Nyman D.J. Nyman d Associates Norman C. Rennalls BASF Corporation (Curley Turner Fluor Danie, ine. ii‘The ASCE Task Commitice on Blast Resistant Design ‘This report was prepared 10 provide guidance in the blast resistant design of ochemical facilities. Though the makeup of the committee end the writing of this ment are directed at petsochemical facilities, these yuidelins are applicable to simar design situations in other industries. Those interested in this report. should inctude Sinvciurl design engineers with dynamic design training. and experience aa well 2s ‘operating company personnel responsible to establish intemal design ard construction Practices. The task Commitee was established because of a significant interest in the Pewrochemienl inlosry in dealing with cnsily process accidents, in interpreting ‘overtnnent softy standards an inthe desire to protect employees. ‘One purpose ofthis Feport is t0 help provide some uniformity to the current mix of internal and published ‘This report is intended to be a State-oFthoPractice set of guidelines, ‘The ‘ecommendations provided are based on publshed information and actual design, A feviow of current practie, intemal company standards, published documents, and the current work of relates groups was condacted The report includes a list of refecenes to Provide addtional information. “The reference list emphasizes an emphasis on readily ‘wvalable commercial publications and government reports. Bectuse of their elevance 12 ‘his fepor, several publications deserve mention here, Two widely used documents ealng generaly with bast resistant design are 793-1300, Structures to Ress he Effects of Aecidewal Explasions from the Department of Defense and ASCE manuel 42, Design & Sirennes 10 Rest Nwiecr Weapons Effex Two publications which greatly Supplement chapters 2 and 3 are Guidelines for Evaluating the Characteristics of Vapor Clon Explosions, Flash Fives, and BLEVES, and Guidelines for Evaluating Provess Plann Buildings for Extemal Explosions and Fires, These lst two documents are fern ‘he AICHE Center for Chemical Process Safty In helping to create a consonsus set of guidelines, a number of individual and groups Provided valuable esitanee and review. These include Ted Krauthammer of Penn State, Pal R, | al Ry ~ nets I | Ley : 2) Shock Load ») Pressure Load MIGURES.3: Ideatined Shock and Pressure Londs shock/pressure front vslecty in aie ca Be approximated using the followin felationship from Newmark 19.56 = 7 U=1130(1 +0 058 Pyyt? (see) = 5.1 +0.0083 p)"* (see) 33:8 Blast Wave Length, Ly ‘The propagating bi “re any instant in time extends over timed eadial pesee Sagi shoskoresure front tavels gutweed fa oe explosion, The Pre ong ett the front and wails off 1 ambient over distance, Lay the blast 0 ihe ts fl fr bigh easy explosives eve oe, fiom 7M 5- iy.” Inthe love pressure range, ihe Tenth ofthe es be approximated y Le=Ue 33.6 Ideatized Blast Wave Parameters To simplify the blast resistant 39 Equivalent Shock Loading Blast Loading he T we FIGURE 3.4: dealized Equivalent Pressure Lond 34 DETERMINATION OF VAPOR CLOUD DESIGN OVERPRESSURES actual design overpressres iy be stated to the detign engineer in «wo ways: * Tepelinnlst is a set blanket statement such as; “All buildings shall be detiened for a peak reflected overpressure of X pai (&Pa), 4 peak eee overpressure of Y psi (kPa), and a duration of 2 milisesonds* & f Liter refinement is to specity overpressures and durations based on Gieance between the structure and 2 potenil source. The Utara se wre a tspped blocks or » continous fiction The bulaing designer ‘would then determine design lands based onthe appropriate distance, ie CIES Fr the sbove design criteria may bave been developed from « sie ‘specific study, from commonly used criteria or from hisoneat ng dis aaa dy i the most comprehensive approach. Site specifi stds to Sealy and quantify exposion hazards are usually conducted by hese process ELLY Seciaistox by specialty constants. Ther are several exepe hier yo be wir ctthof which may be done ina vary of anys. The stem ne rele rato ofthe avaiable methods. More dtaied information's wansslon, ceo, uilding Guidelines and API RP-782. 310|, Define the release: “This step may be based on a worst possible case based on the maximum anount of material within & process loop, of a worst probable (ccedible) case selected from a hazards review. Blast overpressures are specific to companies, processes and sites and it is therefore impractical to quantify a uniform minimurn ot maximunt blast overpressure ‘A survey of the blast resistant design practices of some oyerating companies and contractors within the industry shows that blast resistant design is considered for buildings 50 to 1,200 feet (15 to 365 moters) from vapor cloud explosion hazards, However, most industry standards cover buildings in the 100 to 400 foot (30 10 120 meter) range. The blast loading specified varies considerably depending on plant type, spacing and model used to quantify the explosion. Overal, che speciied bast, Yoads used for design have side-on overpressures ranging from 1.5 10 15 psi (10 W 103 KPa) with positive phase duration ranging from 20 10 200 ms. These loads re for buildings spaced from 100 to 200 feet (30 to 60 meters) from an explosion 2. Formation of an explosive cloud: ‘This step is often done using two computer models. ‘The first isa source emissions madel which ealeulates what happens atthe interface between the contained material and the atmosphere into which is being released. The second is a dispersion model which calculates how the released material disperses and mixes with the air. ‘Amount of energy contributing t0 the explosion: This may be based on a = fraction of the (otal amount of material available or by determining the mass ofthe cloud that is within the flammable limits, It may be further refined by looking atthe level of confinement within the area ofthe cloud. 4, Calculation of blast overpressure parameters: There ae three major methods in use today. One is the TNT Equivalency Method which gives inaccurate results for vapor cloud explosions, “The other two methods are the Strehlow Curves from Baker 1983 and the Multi-Enetgy Method from TNO 1985 Both provide a family of curves based on flame speed or explosion strength, ‘These curves are used to select dimensionless parameters which ate then unsealed to determine the actual overpressures. Overpressures may be determined at the point of the structure closest to the source and then applied to the entre structure, If the structure is large, the average source. Generally, the greater the spacing, the lesser the overpressure and impulse, bout the longer the duration ofthe blast loding, Historical deta from industrial explosions are hard to accurately quantify as these can only be approximated by beck calculating from observed deformations of structures. Blast overpressures from vapor cloud explosions ate expecially dificult to ‘quantify because they tend to be directional, come from multiple sources, and vary with site conditions. Additionally, there is les information available than for high ‘explosives. In one company’s review of five recent vapar cloud explosion incidents, as measured at 8 range of 200 to 1,000 feet (60 to 300 meters), peak reflected pressures inthe range from 2 psi (14 kPa) with a 38 ms duration to 12 psi (#3 KPa) ‘with 2 33-ms duration have occurred, ‘These pressures correspond to sie-on ‘overpressures ranging ftom 1 psi (7 kPa) to 5.5 psi (38 KPa), An extensive lst of this 'ype of explosion data is included in Lenoir 1993, ‘overpressure on the surface or the overpressure at the centroid of the surface may be | Used. Norally a building should be designed considering the potential bast wave from any horizontal direction, but nt all diections simultaneously 3.8. BUILDING BLAST LOADING To design a blast resistant building, the design enginecr first has to determine foads on the building as @ whole and on each individual structural component such ag ‘yall, roof, frame, et. from the free field blast overpressure usually provided by the facility owner. To establish these loads, the design engineer should understand the interaction ofthe propagating blast wave with the building = Commonly used criteria inciudes Sti-22, and CJA 1992, Both documents specify 4 least two biast overpressures for buildings spaced 100 feet (30 meters) from 2 vapor cloud explosion hazard as follows: = 4. High pressure, shor duration, triangular shock loading: Side-on overpressure of 10 psi (69 KPa) with a duration of 20 miliseconds When a blast wave strikes a building, the building is loaded by the overpressure and drag forces of the blast wave. The interaction between the blast wave and a Structure is quite complex as shown schematically in Figure 3.5, For the purpose of design, the resulting blast loading can be simplified, as illustrated in Figure 2.6, based ‘on the idealized shock wave discussed in Section 33.6, The bast wave in Figure 3.6 is shown traveling horizontally let to right. However, depending on the location of potential explosion hazards relative to the building site, the blast could strike the ‘building from any direction and may, in the case of an elevated explosion source, slant downward towards the building, ». Low pressure, long duration, triangular loading’ Side-on overpressure of 3 si (21 kPa) with a duration of 109 milliseconds. ‘These blast loadings have been widely used in the past for blast resistant design ‘hroughout the industry. However, many owners have developed specific blast loading criteria more in line with their specific circumstances, With advances in the ‘modeling of vepor cloud explosions (Baker 1983, CCPS Explosion Guidelines), the trend is toward the use of VCE based blast loads eae. es = ou 312a E =a : a - qe a ree tarefacton > a = SET [> tt ton ; cm ee eee 4a " cra wave \ reflected ‘$= Clearing Distance } shock — Blast a (Lesser of B/2 or 1) seas se os te nen (eo reese ; tn et Poets Dre front 7 = Pa * Averaged Overpressure f FIGURE 3.6: Blast Loading General Arrangement for n Rectangular Building ' Sock, fee (om Fortes 195) lit since bing td amperes viva cones a tne | ry sou (ete sverprnte,sieon eveyone depose an ncpae is ton tras) aed psu, Raed oe sont ese sicea ee 6) Dita cot ase ate dee J Fast tn antag |The was ie he exptin suc il epee feel oeprese 4 ae | As discussed previously, the reflected overpressure amplifeation Of the blacy sue | depends on the angle of incidence, a, and on the rise-tima, tof the side-on ~ 1 t ‘overpressure pulse. For design purposes, the normal shock reflection conditions (a. = 1 | CUE Sable anmelden enon ce Oca J i es vortex: otherwise. However, in some cases oblique reflection (about 30° to 60°) may be 4) Ditaetion is complete Shock front passes heyond structure ‘ore critical to the overall building because the fll reflected overpressure could load too adjacent sides of the building The reflected overpressure decays to the FIGURE 3.5: Schematic of Blast Wave Interaction Stagnation pressure, P, in the clearing time, t as defined below and illustrated in ~ with a Rectangular Building (from TNO Green Book) Figure 37, 7 Pr=Pot Cote an oP 3142 = te a ty t GURE 3.7: Front Wall Loading fe3S/U< ea) where, $= clearing distance, the smaller H, or B/2 Hm building height wiling width ‘quntion 3.8 and Section 3.2, the duration of the reflected overpressure effec, should not exceed that ofthe fee Feld postive overpressure, _, li order to se the dynamic response charts based on a triangular shaped load, the bilinear pressure-tine curve shown in Figure 3.7 ean be simplified to an equivalent Iwiangle.” This equivalent load is compoted by equating the impulse for each loed shape and using the same peak pressure, P, The impulse, lor under the bilinear pressure-time curve is 1-051, -P) +05, G9) ‘The duration, ta ‘of the equivalent triangle is determined from the following equation DP met) PIPL TL 6.10) 382 Walls sidewalls are defined relative tothe explosi shown in Fi © explosion source as shown in Figure 3.6, These walls will experience less blast loading than the front wall, due to lack of essere reflccion and to atenaton of the Bast wave wit dizance fem the explosion source, In certain cases, the actual side wall loeding is combined with ther blast induced forees (such as in-plane forces for exterior shear ‘wals). ‘The seneral form of side wall blast loading is shown in Figure 3 8, FIGURE 3.8: Roof and Side Wall Londing As a blast wave travels along the length ofa structural element, the peak side-on overpressure wil not be applied uniformly, It varies with both time and distance. For example, ithe length of the side wal equals the length of the blast wave, when the peak side-on overpressure reaches the far end of the wall the overpressure a the near end has returned to ambient, reduction factor, Cz, is weed to account for this effect in design. Values for Cy, see Figure 3.9, are dependent on the length of the structural element, Ly, in the direction ofthe traveling bast wave. Ifthe blast wave i= traveling perpendiculr to the span, then Ly should be equal to a nominal unit with of the element, ‘The equation for side walls is as follows: PL= CPt Coae ou where, = effective side-on overpressure ‘The side wall load has a rise time equal to the time it takes forthe blast wave 10 travel across the element being considered, ‘The overall duration is equal to this rise ‘ime plas the duration of the fee-feld side-on overpressure, 10 Ce ou ol 10 10.9 Ly FIGURE 3.9; Effective Overpressure Values (from TM 5-1300), 3.16vo we hemes es hanes pee sof weal md amd 3.53 Roof Londi For 2 building with a fat roof (pitch less then 10°) itis normlly assumed that ‘ellection does not occur when the blast wave travels horizontally. Consequently the roof will experience the side-on overpressure combined with the dynamic vied Bressure, the same as the side walls. The dynamie wind force on the root acts in the pposteciection to the overpressure (upward). Also, consideration should be given fo variation of the blast wave with distance and time as it travels across avoot Coment The cesulting roof loading, as shown in Figure 3.8, depends on the ratio of [yast wave lengts to the span of the oof element and on its orientation relative to the iestion of the Bast wave. The effective peak overpressure for the roof elemments are salelated using Equation 3.11 similar to the side wal, 35.4 Rear Wall Loading Rear wall loading is normally used only to determine the net overall fame loading, Because the rear wall io is opposite in direction to the ftant wall oad, te “nciusion tends t reduce the overall neal blast force. Rear wall effects are many ‘imes conservatively neglected The shape of the reat wall loading is similar to that for side and roof louds, however the rise time and duration are influenced by a not well understood pattern cf srilover Fzom the roof and side walls snd from ground reflection effec The esr ‘wall blast load lags that for the front wall by L/U, the time forthe blast wave to travel the length, L, of the building, The effective peak overpressure is similar to that for Side walls andi calculated using Equation 3.11 (P, is normally used to designate the ‘ear wal peak overpressure instead of P), Available references indicate two distinet Values forthe rise time and positive phase duration, ENO Green Book and ASCE Manual 42 vse esiteria that appears to be based on longer duration hlast lads, The positive phase has rise time of 4S/U and a total ration of 1, (Figure 3.108). Note that for blast loads of @ moderate to short ‘ration, the rise Gime may approach or exceed 1, Information is not provided on. Ww ht WU+SIU, yt, +t b) Based on TM 5-130 FIGURE 3.10: Rear Wall Loading 8} Based on TNO Green Book 317 ealing with this situation TM 5-1300 provides eriteria computing the reat wall load as though it were an extension of the roof, Though graphs are provided to determine the rise time and duration, for most typical control bulidngs, the positive phase will have a rise time of approximately SIU followed by a duration oft (Figure 3.106) 355 Prame Loading {In adkltion to the roof loading, the framing system forthe building will experience the diffaction loading which is the net loading on the front and rear wall taking into ‘ceount the time phasing, During the time, L/U, that it takes the Blast wave to travel fiom the front to the back ofthe building the structural framing willbe subjected to the targe horizontal unbalanced pressure on the front wall. After that time the front ‘wall loading is partially offset by the rear wall loading. Figute 3.11 shows the general fori for te lateral frame loading, 3.5.6 Negative Press re And Rebound Loading ‘The components ofa building wil also experience blast load effects, opnosite in Airection tothe primary blast load effets, due to the negative plase (suction) of the bast wave as discussed in sections 3.3.1 and 3.3.2 above, together with the rebound Df the structural components from the inertial effects of the overpressure loading "As noted above, the negative pressure forces are generally ignored since they are relatively small or are unquantfied for vapor cloud explosions. However, the structural cormponents of the building should be adequately detailed 0 perform satisfactorily for the rebound effects. These effects can he quantified from the time history dynamic analysis of the structural components as discussed in Chapter 6, ot approximated by use of design charts such as provided in 7M $-/300 or A. Mama 42, uy wht FIGURE 3.11: Net Lateral Load on a Rectangular Building (Grom TNO Green Book) 318r di APPENDIX BLAST LOAD EXAMPLE This example ilustrates the ealeulation of blast loading on the components of a Luang subjected to « shock wave traveling horizontally. ‘The building dimersiong areas follows: Width, B93 (28-4 m9 gn length = 67 @OAtH) tg CO > Tight 18. 64S a Isa Blast Londing ore ‘A blast wave fas boen given and will be applied normal to the long side of the tuilding: 2s futher determined that the distance tothe explosion and the lengih of the bulding are such thatthe overpressure and duration do not change igrifeanty ‘over the length ofthe buliing, The blast (shock) wave paremeters are 1s follows peak side-on overpressure, P= 6 psi (41 kPa) Figure 3.3) duration, & = 0.05 see shock fon velocity, U #11300 + 0.058 Py 1130 (1 + 0.088 psy 312 MVsee (400 nvsee) ‘(Equation3.5) length of pressure wave, b= Ua) = (1.312 Mee) (0.05 s02)= 66 (20.1 mp (Equation 3.6) Peak dynamic wind pressure, (Equation 3.4) y= 0022 (Pu) = 0.622 (6 psi)®=0.8 psi (6 KPa) ron Wall Londlig: 2 Font wal asumed to span venically from foundation to roof, The design willbe fora typical wal segment one foot wide rellected overpressure, (Equations 3.2 and 3.3) P= [2 +005 (Pal Py = (2 +.0.05 (6 pe) (6 psi)= 13.8 psi (OS Pa) clearing distance, (Section 3.5.1) ofl orB2=15R (45m) tellected overpressure clearing time, (Equation 3.8) ("3 (S/U) 2.5 in (64 mm)| eae be HIMI AH fall Ferdi ogi oi 10 10 100 Strain Rate, invin/seg 5.7: Erfect of Strain Rate an Dynamic Material Strength (from TM 5.1300) tis possible to determine the actual stcan rate of a material during calculation of {dynamic response using an iterative procedure, A rate must be assumed and a DIF Selected, The dynamic strength is determined by multiplying the statie strength (increased by the strength increase factor) by the DIF. The time required to reach [aximum response can be used to determine a cevised strain rate and a revised DIF, This process is repented until the computed strain rate matches the assumed value, There are uncerainies in many of the variables used to calculate this response and Aetermination of strain rates with great accuracy is not warranted 7M 5-1300 and other references suggest selecting DIF values based on pressure ‘ange or scaled distance to the explosion source, ‘This method groups blast loads of less than afew hundred ps into the low pressure category witha single DIF value for tach sess ype Tor petrochemical faites, he vas majority of structures wil fll in this low pressure category DIF valses vary for different stress types in both conerete and steel for several feasons. Fexural response is ductile and DIF values are permitted whieh reflect actual strain rates, Shear stresses in concrete produce bite failures and thus require 8 degree of conservatism to be applied to the selection of a DIF. Additionally test data for dynamic shear response of concrete materials is not as well established 25 ‘compressive strength, Strain rates for tension and compression in steel and concrete ‘members are lower than for flexure and thus DIF values are necessarily lower, 533 Values for dynamic increase factors are presented in a variety of references although most are based on the same data source. Additional data has been provhiced in various test programs but has not been assembled into a central source. | Much of the data that has been published is based on high strain rate tests and many of the recommended values are arbitrarily chosen, Table $.A.1 provides eecommendled DIFs for reinforeed concrete and masonry and Table $.A.2 contains values for structural steel, cold-formed steel and aluminum. SSS Dynamie Design Stress Strsin hardening effects in steel members and concrete reinforcing are modeled in SDOF analysis by using a design stress which is greater then yield. During dynamic response, the stress level at critical sections in 2 member vary with strain of the section. In the elastic region, the strain across the section varies with location fiom the neutral axis of the member, Beyond this region, the member experiences plastic ‘response in which the fiber stress of the entire section exceeds the elastic limit. AL ‘is point, the stress is constant over the cross section but is still changing with total ‘member strain, Stee! members experience an increas in siess in the strain hardening ‘region until the ultimate dynamic material stress it reached. After ths point, the fiber steess decreases with increasing strain until rupture occurs. Concrete exhibits an increasing stress until the maximum compressive stess is reached afer which the stress level decreases with additional deformation. Because of ts britle nature, strain hardening does not occur in concrete; however, reinforcing steel will exhibit this coffee To predict tie dynamic response, it would be necessary to continuously vary the saterial stess with deformation. This variation is difficult to model using SDOI analysis methods because it requires tracking a complex resstance-deflestion curve at cach time step. It is desirable to represent the design material sess as a bilinear stress-strain curve in which sires increases linearly with strain to yield and a constant value afer yield (refer to Section 7.2.5). This produces a simple, bilinear resistance eflection curve as shown in Figure $.8 which includes strain hardening effects ard is «elatvely easy to incorporate into the SDOF analysis, To achieve this simplification, while accurately modeling the dynamic response, itis necessary to select a design siress equal to the average stress occurring in the actual response. This can be done by estimating a maximum response renge and using recommendations in Tables 5.A.4 and 5.A°5 for stee! members and reinforcing At low response ranges, the maximum design stress is equal to the dynamic yield stress. At higher response ranges, the design stess is increased to account for strain hardening. In theviiial portion of the response, this increased design siress will ‘esult in an overprediction of resistance. As greater deformations oceur, the stress level, and thus resistance, wll be underpredicted by the design stress selg Ra 3 é Ke Denton FIGURE 5.8: Dilinear Resistance-Deflection Curve Fine omen methods (FEM) ar capable of incorporating complex vations in ail ses in the tine vaning tipo Wile tee sek nace vale, they ae Quite cmple snd, n many css, thers ek aan seins in et ond prediction The yams watel poner eee this soon can bused FEM eleltons, however is soped en i rx scion ay to bo ule oat FEM Sods emtan aoe as odes which are betes indcators of capa eons, See ager ee Asayss Metts, for acon internation, Tn S® Chater 8 Dy 56 DEFORMATION LIMITS Response deformation limits are used to ensure that adequate response to blest teas if provided. ‘These limits are based on the type of structure or component, onstruction materials used, location of the structure and desired protection level, The primary method for determining adequacy of & structure for conventional design is evaluation of the stress level achieved compared with the maximum ste Hermited_ Delecons are also checked fr certain members although hss pitas Gane for serviceability or architectural reasons rather than structural reqitemente Bisst loaded members however, reach or exceed yield stresses to achions ay 5 § ]8 8 Ky = mass transformation factor = “xo * & tac aay canbe efooed ang these equivalent prtameters in ae ae (2 2 fe Of the corresponding actual valves hernate form of the bilinear dynamic EE |B co 1B © 3 i “eauaton 6 3) then Becomes. oe 8 g R Matron ae le ES leg ale 3 Becomes \iflgile ala als % loadsmasstrancforma eajéi|2 2/2 2 |e = |g i KuMarky=n (66) a 7 E he See ee ees Kiune= KuK ee;Feis 813 818 & 1s gelee]e sys s 7 er eee equation above are gF ls i need! xccording to the stress range of the member. These changes aze illustrated a = 068 fr sna supa bam wh union eee ee EP | z 1 foading. The ra ian fetors are also shown a2 |2e a See Pet ise falbe]! 6/8 2/8 8 |e hey 22/82 g é qi cE i ae ae y ga |e gle 5 gS gelgelf gjes jae [72 u Po oo as 3 bs seji ald 2 |! 2 a MG) = Tae 2 Latent) 2% xt fay lee MOd= spyU eth ade roey iP IS] |e St | Be | Ki7O64 Ky 050. Kyy@078—K}=050, Kye 0.33, Kyy 065 ge i «st | 2% eee | Ly|u i. e4 &) Poste Response Sa |r || Sy)" St | 32 FIGURE 6.4: ion and Transformation Factors for a Slmply tata Supported Beam i 69 etopiaina | Suan [Lad Facer | Lonped Mas | Gnome] Bending Spa] yan Diagram Range Ki | Feston k(t} Racor, Ky | Resistance, R Constant, K | Reaction, V Ply | eiawe | oss = os | i2Mpa | seen | o3en solar scaiee Epa) | os 050 /8(MpstMp- YL] 384 E5L3 | 0.39R + 0.11F r asic | 050 033 fapeme] 0” | o3er, +0 12F F Elastic 1.00 1.00 037 |4MpstMpe tf iszev? | o7R-o21F TR [TER] Ptasic | 1.00 10 033 [4M Mpc] 0 | 075k, -0.5r| F/2 Fi2 Elastic O87 076 0s2 OM IL S64EN2 | 053R-0.03F tat N aie 1.00 1.09 os | Ma 0 | 0828, -o0aF "Note: (1) Equal portions ofthe concentra. (2) E-Pis Elasic-Plastic, 2) Mpcis the ultimate moment capacity at midspan; M qs the ultimate momen: ted mass are lumped at each concentrated load. capacity at support ce TABLE 63: hasfomatin Fats On Way Mess Sipe Howden Coan Bias 1500 Loating | Sinn | Und acer | Cuneta] Unions] Eman] Some Deane Diagn | Range | ok, | Facer CH) | Paco tge | Resince Ry Comune | Ren Fepel> [caste | 058 - as Mit | 185 EVL3] Vi =026R+012F pommel lg | va-0aRvatE L PQ) | 06s ~ 050 ACM gt 2M, yt] 264 ELD | 039R +0.11F 4.4/1 MH mow | os fT oar fama) 0 lease, sours n oe 7s] Das Tegal [aE WTO OCe Vecosims or fetal cr | iw | aw 049 andar, ase? | oven ooseerteL asic | tao | too 03 aoa ic] “on [orane case Mn, a rT 04s | OMe) MBM] Wi =a OF Waooses ome WEN ee | oon | on 032 fantom, ot] s6eu hsrseccomre er { nase | too | tan 036 [aww] on” Josene cours neth Notes: (1) Equal portions of the concentrated mass ae lumped a each concentrated load (2) EP is Eas - Plastic (6) Mygis the climate mement capacity at midspan; My the altmate moment capacity at suppor,i a i i i i [ fi k i t k i i fi E ‘Transformation factors also change as the structural member progresses from the ciastic to plastic ranges and back to elastic response range. ‘The resistance ace ‘changes for the plastic range as shown by Equation 63, In actual practi, itis common to keep the transformation factors constant franuhout the analysis. Bnginering judgment is used to select the approprice fhctor, depending on the predominant response mode anticipated. A tial ood erg apbronch nay be used to evaluate the response mode behavior, An average of the elastic and plastic transformation fretors is sometimes used Fransformation factors for common one-way and twosway structural members ‘ze realy availabe fom several sources (Biggs 1964, TM 5-1300), Refer te Tablca 6.1, 62, and 6:3 fora summary of such factors for one way member, mass of the structore inches its self weight and the weight of permanetly attached equipment. Mass is simply weight divided by gravity. Approx Sometimes used in determining mass distributions of members analyzed ay SOO ‘ystems in order 10 be able to use reaily available tabulated transformation eters nal Reforming dynamic analyses of a series of SDOF systems representing & Sicture an estimate of the amount mags “ring along” with a supporting member fen must Bemade For example, a roof sede supports a portion ofthe navy st ag {Rot beams it supports which needs to be added to the girder's mass as Musrated Figure 6.5. Engineering judgments are often used in lieu of tigorous matheosied procedures, One recommendation for continuous reinforced conerete slab and bes {ype construction given by TRY $-1200 (Section 443.1) isto include 20% of ihe fuPported member's mass with the mass of the supporting member, This would Sorrespond to a supported member which is relatively flexible in comparison to the ‘poring tamer For the structure illustrated in Figure 6.5, 50% ofthe beams as considered to be lumped atthe midspan of each girder. In this example, the re gamle to be eyed in comparison to each girder. Each ease hedges individually {General transformation factor equations for distributed mass. eystems and Ton eebee of Freedom systems are given by Higgr 1964 (Chaper 8), and Clough £224 (Chapter 2). These general methods can be used in determining Lansformeccn factors for vonprismatic members or members which have nonunifore soe 10, the sysiem responds a though the load were suddenly ‘ied rel constant. Ain, the maximum ductity demand ean alo be exrese in 4 PRU Fried] 6.10) Enpirical fo:mulas have been developed to transition between these two extreme dynumie response cases. ASCE Manual 42 provides the following celationship over {he fil response range oF = t/t 2 = GT | Cue 100) on) = YORU , Que 100) a(t)" 2yu(e+07) oe: Comparisons with more exact solutions show that this relationship yek i ions show that this relationship yields results ‘o within 5%, which is usually accurate enough for most appicstions. This formula does nat lend itself to direct calculation of ductility demand in terms of the other Parameters. However, it can be solved for iy by tral teations, oi 645. Numerical Integration When simple graphical, closed form or empirical solution methods are not propriate wr do not provide sultiient information, the numerical time integration method ean be used. This method is also known as the time history method, Most texts on structural dynamics (Biggs 1968, Clough 1993, Paz 1991) provide extensive coverage on numeral solution methods for nonlinear, SDOF systems A briet summary will be given of the Newmark rumerial integration procedure, Which is commonly used to obtain the time history response for nonlinear SDOF systems, It ig most commonly used with either constant-average or linear acceleration approximations within the timestep. An incremental solution is obtained by solving the dynamic equilrium equation forthe displacement at each time step Results of previous time steps and the current time step are used with recurrence formulas to predict the acceleration and velocity at the current time step. In some cases, a total equilibrium approach (Paz 1991) is used to solve forthe acceleration at the current ime ste. To ensure an accurate and numerically stable solution, 3 small time increment faust be selected, A rule of thumb is to use a value lest than of equal 1o 1/L0th of ‘ether the natural vibration period ofthe structure or the load duration, whichever is smaller. Refer to the appendix for an outline ofthe basic steps involved with solving the equation of motion using Newmark’s method. Computer programs using ramerical time integration methods for nonlinear analyses of SDOF systems (for cesample BIGGS, WBE 1990, PLASTIC, Paz 1986, and CBARCS) are available Refer to Chapter 1! for the implementation of aumerical integration in a blast design 646 Support Reactions Peehaps the mast commonly overlooked aspect of using SDOF approximations is the determination ofthe dynamic ceactions for the actual member. ‘The spring force in the SDOF system is not equal to the support reaction, In order to determine the \damic reactions, the distribution of the inertia force within the member must be considered (Biggs 1964, Chapter 5), The basic approach as illustrated in Figure 6.8 isto express the dynamic forces acting on the member, ora segment of the member, intesms ofthe displacement and acceleration atthe control point. This displacement, (0 is determined in the solution ofthe time history analysis of the equivalent SDOF sytem, Equations for the dynamic reactions of typical structural members are available from the same sources which provide the transformation factors. Refer to Tables 6.1, 6.2, and 63. These equations express the dynamic reaction in terms of the resistance and applied load, both of which vary with time oisElastic; ror |Plastic; Fa Ma | us Ls Mp | Us vo | 0 vo ‘Sui moments a “3 (clockwise +) ‘Sum moments at "a" (clockwise +) Oma + HO (Le 38h won Shy omntps AD (Le Ly vor Substitute My = Ry Lis Rey FOL by EV BSG pe Be ° Ro AO Ve Pes w= 2 daw + ew won a Be.) 039 R011 FW 038, +0.12 FE FIGURE 6.8: Reactions for a Flexural Member ‘with Distributed Mass & Load 6S MULTLDEGREE OF FREEDOM SYSTEMS are ruin of the dynamic anayis methods descbed shove for SbOF guidance fs week feedom (MDOF) systems is etcused below, Taso, Siaret 5 Provided to sid io the selection of appromane tacts methods, manates which may not be adequately represented ase SC system include USOT bulking frames, siabfocanigider Faming systare cog structures having Sgr nztrtons of ignicant amped masses An exenvle st ‘ypical multi- Sepree of fieedom structure is shown in Figure 69. “tae story building is Beth caer tl imple forces at the tp ofthe ist and second core ee be the cao for blest longs is y oO | Eh Eh] |n, 0 My EL en} [Hy Wie 8) Two-Story Frame with Rigid Gieders : x Fa rt x1 Ht RO a ° vane 1 »)MDOF Mode! ©) Bxteral Force FIGURE 6.9; Typical MDOF Structure ) Dynamic Response 65.1 Dynami Equilibrium Equation nthe avr configuton is complex or, sisicant dynam terion veer toned won one oe ee “ros ufos Re cope meson cea hn efter ilo epee rr tne a ay fcedom na sgl, eampraerave dyna als oe ent ser {iis ype af nang "Uf peel pause Fae sano sees owas 620i a | {ype of element, euch as plate and shell elements for continuum type structures, A Rourled analysis weed not be all encompassing. For example, a two sinensonal plane fame arly of «biking employing two or eee freedom is considere roach Sepe ae : dimensional model Refer to Section 6.62 for 2 discussion on modeling considerations for this type of structure age ones of MDOF systems are determined from the solution ofthe Following eniliorim equation. ‘This equation is the matrix form of the equitoring tion for « SDOF system (Equation 61) ay ea 1 (CH) + IR Yiy) = Fa 6.12) or Taste purposes, anal sluions of this equation can be obtained for only voaren cents fives sewrees of freedom, An example of an elastic-plastic, wo {earee ofeedom system analysis is given by Biggs 1964 (pp. 237242)" srcn ihe simple problem involves significant effort Solutions for MDOF systems are usually obiained through the use of finite sea cedures. Due to nonlinearities associated with plasticty and posstly aks, “tsplacements, the diret tne integration method should be weed, Wnneg Age IMearation methods for time integration are employed but, the Neonace Waeeadis Perhaps rhe most common, Other methods, such asthe Houboult Methec, Mitor-T Method and the Central Difference Method are commonly need Wnty clement applications. Refer to Ravhe 1993 for further details 65.2 Advanced Analysis Methods Ine strict sense, ae material effects are accounted for inthe analysis of the structure as hols words te appopate int Ste dag cade eqsene ra the analysis (White 1993, Chen 1994), = crn teative Ist of behavioral phenomena aad physical atrhutesafeting {Ie song and stability of stel ames is compiled in White 1991. Some eene iin Hod inslude iil impereetions, residual stresses, inal trains, ormmnste: Feiene®: clfects of sinwltancous axial force, shear” and moment on eoomon siotlae st of items could be compiled for reinforesd ‘materials. It is clear that a comprehensive adv 6a complex. ‘The tools needed to perform such advanced analyses are not yet generally availble. However, a number of commercially available finite element progeanie possess sophisticated noalinear analysis capabilities. These analysis codes do wot Incorporate the design code checks for Jacal member instabilities as is done i ‘vanced analyses. In spite of this obvious and significant ciference, the fixie slement analysis method is considered as an advanced analysis method! for pcposes ofthis report 653. Finite Element A finite element analysis method is recommended when one or more of the following conditions exist a, The ratio of a member's natural frequency 40 the natural Frequency of the support systém is in the range of 0.5 to 2.0, such that an uncoupled analysis approach may yield significant inaccurate results, b, ‘Time varying support reactions or memiber forces are desired in order to evalunte the structure or its foundation in great detail in an effort to minimize costs of siructural backfit modifications © Overall structural behavior is tobe evaluated with regard to structural stability rame buckling), gross displacements and Poets effects 4. ‘The structure has unusual features such as unsymmetrical or nonsniform mass and stiffness characterises, Many commercial finite element computer programs (for example ABAQUS, ADINA, ANSYS, DYNA, DYNA3D, LS-DYNA, NASTRAN and NONSAP) are readily available for nonlinear dynamic analysis. Other computer codes, sci as CBARCS, COSMOS/M, STABLE, ANSR-I heve been developed specifically for the Sesign of structures to resist blast loads. All these computer progeams possess nonlinear analysis capabilities to varying degrees. Cerin considerations should be given to achieve adequate results ata reasonable cost when using finite element analysis methods, One item to consider is the appropriateness and practicality ofthe element type, The most suitable element nds ‘nn the simplest to the most complex include spring elements, Tine (hear) elements, Platelshell elements and solid elements Another important item is to consider how the finite element outpat dats would ‘be used to confirm compliance with acceptance criteria. For example, using siress ‘output data from plate or shell elements to evaluate a ccinforced conercie sab is not 622{et Patel, Some computer coder employ « i een for pate and shel clones based on set restans{Torces and moma, aia ge on Conveio feral dsign purposes Another deny ees yh me Gateive relive dplacsments of a member her ange ee detection gst he alent defection ‘Along the fie danent method can provide the mest accurate meane for ans sce fort ads, he uncrny aso eh a lads grey docs ot jut suse Abe, te eon scutes fda deepen nd inerpeation of rests oten pete my en ee De sept! metods Guna ave Thy inl SOO ak eee Ia rsormended fr use exp in those canes uk order thence asc wih SBOP aproxtncion ean bea 64 APPLICATIONS ymimic analysis approaches for some ypeal aplication are described below, 6.6.1. Shear Walv/Di phragm Type Structures Certainly the most common type of blast resistant st t i Fesistant structure at petrochemical ‘Peles is tsinforced concrete or masonry, single story bling wake rectangular 01 prin Several considerations are essential wi the usual load path described above may not be m ‘note coseraon in the aay of he shear walaphagm systems she fective with of he Uaphragm Tange Some prion afte en se can be expected to ac as compression and eri fanote renee ee horn daphag sia The laine wi af he lege neal re ‘oposite stn he wal thst (Ocesho I Sige a ee ‘spore of 8 mente i afte by ite mail feueney and mean ee Ihe Hanes shuld be considered when deeming he Seah ae 63 A similar situation exists forthe side shear walls. Some portion of the connect fiont and rent walls will act as beam Manges asin a C-shaped cross section in plan Hore egtin, an effective with of six times the flange wall thickness may be std 662 Frame Structures Modeling of frame type structures generally involves use af @ MDOF approach ee 10 simultaneous application of lateral and vertical blast loads on the ftarne. A Simultaneous application of these forces generally results in combined axial and bending load conditions in the individual frame members which signiiearly fect the ‘member design, Otherwise, a conservative combination of the separate effects of ‘ach loading condition on the response ofthe frame must be used. Advantage can he taken of the feet that peak responses due to the vertical and lateral londs do not severally occur simultaneously ‘Another consideration for frame type structures is whether to use a two or three dimensional model. The appropriate choice depends on the symmetyy of the structural resistance, mass and the loading. if all theee are symmetri, 2 (wo dimensional plane frame model wil generally suice Some studies of one and two story plane frames have examined the level of modeling detail required to obtain reasoneble results which are summarized by Baker 1983 (pp. 442-453). These studies considered factors such as the aumber and ‘pacing of joints, member loads versus joint loads, girder Mexiblty, sweeping root loads and mass distribution among other factors, Due to the large number of variables Hudied, the reader is encouraged to refer to the referenced documen 10 obtain a clear understanding and appreciation ofthe resulis Selection of the material madel is another important factor to be considered. Some programs allow the user to specify plastic rmoment-rotetion curves for beam «lements. However, the more rigorous and most widely avilable method of defining sonlinear material properties is t0 specify the stress versus strain data, Plastic behavior is approximated at the section levl in the former method whereas, the later method tracks plastic behavior a the individual integration points (fibers) through the UWieaness ofthe member. Each method has its advantages and disadvantages, ‘The plastic hinge nonlinear material model is easier to use but usually can not consider axel load effects, Plastic hinge locations must usually be predetermined and ‘ce usually limited to the ends of the member, Analysis results which include splacements ain plastic hinge rotations which are directly comparable against acceptance criteria, ‘The more rigorous stress/strain nonlinear materi model, often referred to as the Plastic zone method, is theoretically capable of handling any general cross section Both isotropic and kinematic hardening rules are usvally availble, This method is 6-24