Final - April 1998 Verification of Proposed Design Methodologies For Effective Width of Slabs In Slab-Column Frames By: Jacob S. Grossman Rosenwasser / Grossman Consulting Engineers, P.c. ~ NYC Abstract ‘This paper evaluates several design methodologies for lateral loads using recent flat siab-column frame experimental data [1]. This data considers the effects of connection and panel geometry, as well as the cracking caused by construction loads, gravity loads and lateral loads. “The design methodologies evaluated in this paper are as follows: the methodology proposed by the researchers of the experimental data [1]; an earlier design methodology [2] (which has been used since the late 1970's in the design of many structures by this author); and a permutation of this earlier design methodology. ‘A“rational’ methodology to estimate the contribution of slab-column frames to the stifiness of 3-D structures at various lateral load levels of interest (serviceability, strength design, and limit states), along with the results of intial efforts to verity this methodology via measurements of high-rise flat siab structures, is presented. Keywords: effective width; stab-column frames; fat plate; structure; serviceability; strength design; stiffness; drift limits; ductility; redundancy; dynamic properties; damping; wind; seismic; design methodologies. Background ‘The designer must know the dynamic properties of the structure (its stiffness, mass and damping) in order to evaluate the lateral loads the structure will absorb during seismic or wind storm events, as well as to determine the structure's ability to successfully dissipate such action. In concrete structures, two of the three properties mentioned above (stiffness and damping) are variables ‘dependent on time (history of events) and the load-level the structure is absorbing or has absorbed. ‘The design process is iterative ~ requiring a review of the structure's various building blocks for stifiness degradation at the load levels of concem (serviceability, strength design and limit states).Final - Apt! 1996 Verification of Proposed Design Methodologies For Effective Width of Slabs In Slab-Column Frames By: Jacob S. Grossman nyc Abstract ‘This paper evaluates several design methodologies for lateral loads using recent flat slab-column frame experimental data [1]. This data considers the effects of connection and panel geometry, aS well as the cracking caused by construction loads, gravity loads and lateral loads. ‘The design methodologies evaluated in this paper are as follows: the methodology proposed by the researchers of the experimental data [1]; an earlier design methodology [2] (which has been used ince the late 1970's in the design of many structures by this author); and a permutation of this earlier design methodology. ‘Arational” methodology to estimate the contribution of slab-column frames to the stiffness of 3D structures at various lateral load levels of interest (serviceabilty, strength design, and limit states), along with the resuits of intial efforts to verity this methodology via measurements of high-rise fat stab structures, is presented. Keywords: effective width; slab-column frames; flat plate; structure; serviceability; strength design; stiffness; drift imits; ductility; redundancy; dynamic properties; damping; wind: seismic; design methodologies. Background The designer must know the dynamic properties of the structure (ts stiffness, mass and damping) in order to evaluate the lateral loads the structure wil absorb during seismic or wind storm events, as ‘well as to determine the structure's abilty to successfull dissipate such action. In concrete structures, two of the three properties mentioned above (stifiness and damping) are variables dependent on time (history of events) and the load-level the structure is absorbing or has absorbed. “The design process is iterative -- requiting a review ofthe structure's various building blocks for stiffness degradation atthe load levels of concem (serviceabilty, strength design and limit states).‘Among the more commen building blocks (columns, beams, shear-walls) flat slabs and plates are used extensively, most notably on the East Coast, in the design of high-rise apartment buildings. In these buildings, architectural constraints did not allow for the incorporation of many beams and shear walls in the design. Engineers, responding to the demand for this economical type of structure, have attempted to estimate the stiffness of the two-way slab and its contribution toward resisting lateral loads. Inthe late 1970's, ACI Committee 318 made a concentrated effort to provide Code directions that ‘would incorporate fiat slabs and plates as part of the lateral load resisting system. Certain factions within this committee recommended the extension of the use of the Equivalent Frame Method (EFM) {AGH-13.7) [3], which was developed for gravity loads, thus creating a unified gravity and lateral load design approach. Other members of the committee recognized the complex nature of this approach. ‘They pointed out that the EFM was specitically developed for gravity loads via a limited research study of square panels. These members concluded that it would be imprudent to use the EFM in the design for lateral loads without additional research. ‘The disagreement within the Code committee prompted a recommendation to the Reinforced Conerete Research Council (RCRC) to assign the late Professor Vanderbilt the task of ‘accumulating the available research and theoretical studies on this subject. Part of his charge was to verify i the EFM could safely analyze lateral loads. Vanderbilt accumulated most of the information available prior to 1981 on this subject [4] and described the different approaches a design team might select. A considerable part of his report focused on the use of the EFM (a.k.a. the “transverse-torsional” method), concluding that this method would be suitable, but only with some adjustments. Prof. Vanderbilt determined adjustments ‘were necessary in the computation of the torsional link K,. He suggested that 4 < ¢, be used in this computation. Disagreements persisted within the ACI Code Committee 318 (for the 1983 Code) and the proposal was tabled. At this time, the author of this paper developed an “effective width” design mode! [2] that was ‘extracted from the various papers and meager research information reviewed in [4] and was modified based on engineering judgment, This code proposal was tabled due to a lack of experimental verification. Due to the committee's lack of consensus regarding this issue, the AC! 318-83 Code [5] is vague (see Commentary [6] Section 13.3.1.2). It became obvious that more experiments were necessary. Meanwhile, established practices continued. In flat slabs, a lack of proper detailing in the joint between the column and the slab can cause a considerable loss of stiffness. ACI 318 (section 13.3) specifies the portion of the unbalanced‘moment (at square suppors it s equal to 60%) to be transferred by flexure, via concentration of reinforcement within a narrow band, “column head,” which equals a width of (C, + 8h), where C, is the transverse width of the support and h isthe slab thickness. The remainder of the unbalanced moment is transferred by the eccentricity of shear about the centrold of the critical section (ACI 318, section 11.12). This empirical method does not accurately describe the transfer of unbalanced moments, butis assumed to generally provide conservative results. Unfortunately tis to rig for Use in workable design. A description of the evolution of this method is in reference (7), along with recommendations that allow for the rore flexible provisions necessary for workable application, ‘The review of the state-of-the-art assembled by Prof, Vanderbilt indicated to the critical analyst that no explicit rational directions could be provided without additional pertinent research. Most attempts to provide direction for both the “transverse-torsional” and the “effective width’ design models were based on the elastic plate theory ~ modified and “ironed out” to agree with the meager test results available, In the ettempt to correlate theory and tests, an important parameter, the ability of the connection between the floor member and the support to develop the predicted unbalanced ‘moment and shears, was generally not considered. Another often neglected parameter was the availabilty of redundancy, which allows for the redistribution of and better utilization of available capacity elsewhere in the structure. On the flip side of the coin, construction procedures and loads could reduce available stiffness more than anticipated by the presence of the service loads. Tests of three-dimensional models that would also review the slab-column connection (considering concentration of reinforcement within the “column head,” the presence of capitals, drops, etc.) and also vatied panel lt,, CC, aspect ratios were necessary before any explicit rational directions could be provided. Meanwhile, the lack of testing strained the credibility of the practitioner who had to continue to design flat slab structures. ‘The stalemate within the 1983 Code Committee eventually initiated a “shopping list" of needed research [8] which was forwarded to RCRC. Two requests from this “shopping lst" pertinent to the subject of this paper are quoted below: A. “Re-investigate transfer of unbalanced gravity and lateral loads in flat slabs. Code provisions (section 11.12.2.3) [5] are too rigid. Ad Hoc Committee investigation (chaired by Professor James MacGregor) could use a deliberate research effort to reevaluate equation (11-40) especially with an eye to allowing for some flexibility in the portions assigned to flexure and eccentricity of shear.” B, “Study 1/8 scale 3-D models with varied aspect ratios of spans 4y¢, column sizes C,/C, etc. to help formulate simple Code provisions for flat slab participation in lateral load resistance. Such a study should consider the slab-column joint. Emphasis should be onstructures designed for moderate wind or seismic forces and which have a deflection index of about 500. The total reinforcing provided should be based on Code requirements to satisfy the test load. Concentration of reinforcing over support should be done by varying the spacing and size and not by increasing the total amount, Also re- examine the practicality of elastic analysis for gravity loads (such as EFM of chapter 13) in Conorete structures.” ‘Several researchers have reviewed and contributed to Item A. The Technical Committee's (ACI- ‘ASCE 952) recent report [9] incorporated substantial improvements, allowing for a large measure of flexibility at the exterior supports. Independent studies, by this author, of available research indicated that a moderate measure of flexibility is also available at the interior supports. The flexibility in the portions assigned to flexure and eccentricity of shear is made possible when direct shear, due to gravity loads and the quantities of flexure reinforcing within the “column-head," are within prescribed lower limits. This improves ductile behavior of the slab-column joint, ACI Code Committee 318 reviewed and implemented above recommendations in the recently published 1995 Code. ‘The investigation of item B was assigned by RRC to Prof. Jack P. Moehle. His research [1] is possibly the first effort on this subject that allows a “rational” approach to resolving the effectiveness of fiat plates in resisting lateral loads. Several papers {10, 11], describing the resuits ofthis research land a proposed methodology for the jateral load analysis of flat-plates, have been reviewed by RCRC (Verification of the proposed design methodology for lateral loads that resulted from this research is ‘provided later in this paper). The research also reviewed gravity loads Code procedures and ‘concluded {10] that, “Neither the Direct Design Method (DDM) or the Equivalent Frame Method (EFM) of ACI 318-83 accurately reproduce the moment fields of the siab under service gravity loads; however, either could have been used in design to produce proportions and reinforcement similar to those used in the slab, Henee, either would have produced acceptable service load performance. ‘The research concluded that flat plates have a large capacity to redistribute both gravity and lateral moments. There has been observation of this redistribution in many cases of actual construction. This should point the way toward the simplification of existing Code provisions (EFM) for gravity loads. Research / Design Significance Rational methodologies to estimate the contribution of slabs to the stiiness of the 8-D structure are presented. These methodologies consider the stitiness degradation of flat plates caused by construction and the various lateral load levels usually of interest to the designer of the structure (serviceability, strength design, limit state). The process of correlating the most accurate methodology to actual construction is underway ~- with early testing based on actual weatherconditions at low and moderate load levels indicating satistactory results. A more accurate evaluation, by the designer, of the dynamic properties of structures in which flat slabs/plates are integral parts is now possible. This will result in better estimations of: the lateral loads induced on the structure; the distribution of such loads to the individual members; and the structure's sway at the various load levels considered. Rational Methodologies for “Effective Width” The concept of “effective width” representing the stiffness of an “equivalent beam” has been ‘thoroughly reviewed in the literature and is summarized in 4]. The effective wicth factor, ot, is ‘obtained from the requirement that the stiffness of a beam of equivalent width, ot, will equal the stifness of the full width, &, of the flat plate panel. The effective width accounts for the behavior of the slab which is not fully effective across its transverse width. Factors affecting the effective width are numerous. Elastic plate analysis can identity those {factors which are more dominant then others. However, for a non-homogenous material, such as concrete, a review of actual field conditions is necessary to supplement theory. Therefore, construction loads and procedures must be considered, along with the concentration of reinforcing ‘over the supports: C,/C, and 4, aspect ratios; the development of cracks caused by shrinkage, by restraint of stiff supports or by loads; the capacity ofthe joints to develop unbalanced moments and to redistribute excessive demands; etc. Above all, prudent methodologies must describe the degradation of the stiffness of the stabs due to the level of the lateral loads. Each of the three methodologies described in this paper is put to a sensitivity review to match the UCB test results [1]. The UCB test encompasses a variety of parameters such as aspect ratios CC, and 4/t, gravity loads and construction procedure influences. While other parameters have yet to be explored (such as the effects of penetrations, unequal column spans in a unilateral direction, plan offset of columns, ete.) the UCB tests do provide an initial means to develop a “rational” approach to verifying the contribution of flat plates to lateral stifiness in actual construction. The three design methodologies reviewed in this paper are designated as follows: + Methodology “TWR® (Extracted from the various papers and meager research information reviewed by Vanderbilt in Reference [4]. This method was modified based on engineering judgment.) + Methodology “JSG” (One of several sensitivity studies of Methodology “TWR" in which a few parameters have been altered.) + Methodology “HWNG" (Developed by UCB researchers [11}.)Methodology “TWR” Description of Proposed “Effective Width” Rules Inthe late 1970's (as part of the effort for Code input described earlier in this paper -~- see “Background’), this author assembled a design methodology to obtain the effective slab width which correlates to the acceptable ati limit (at design load levels) of about h,/400 to h/500. An abstract of this design methodology was published [2] with a footnote disclaimer that the results must be verified by tests in progress at the time [1]. This methodology is duplicated below, with some minor adjustments, ‘The effective width at the center line of an interior slab-column joint, 04, , is computed as follows: (Of, = 0.8, +0, (/ 4) + (Cz-C,)/2 Eg. (1) Eq, (1) allows the recognition of non-square columns and panels and the transition from two-way to one-way slab action as support width C, approaches panel width &. In Eq. (1) he terms are defined as in Chapter 13 of the ACI Code, except that interior supports is taken as the average of the two spans parallel to the direction of the lateral load ~- one in front of and one in back of the slab-column Joint. The effective width of the slab supported by two adjacent columns is then taken to equal the average of the values at the supports, and has the limits: 02 4 SO S054 Eq. (2) To evaluate a, at edge colurnns, with the edge of the slab parallel to the direction of the lateral load, the following procedure was used: is assumed to equal the transverse centerline distance to the column in the adjacent interior panel. ta is computed as if the edge column is an interior column having é thus obtained, Adjustments are then made by multiplying by Eq. (1) by (f+ (L/2))/% where ¢, equals the distance between the column centerline and the parallel edge of the slab. Reference [2] indicated that Eq. (1) appears to provide “rational” estimates of stiffness for two-way slabs to be used in a second-order analysis in structures restricted by the drift and stability boundaries established in the reference. ‘Rational’ stiffness is defined here as the stiffness of the structural member at, or close to, the design (service) load level. The drit and stability boundaries recommended [2] were developed to provide serviceable structures for lateral wind loads and to minimize adverse P- A effects, should the initial estimate of member stitiness by the designer be significantly erroneous. These requirements dictated that the structure being designed have:|. Final (second order) story deflection index h, 4, 2 400 and total structure deflection index Bh,/24,2 500 (with loads magnified due to P- effects). 2. Small stabilty index Q = (EP) 4,/ Hh, < 0.15 using a rational stifness estimate. Provision for ductility and redundancy in the struoture. [Service loads are used in the above: y= final story deflection (inclusing P-A effects); {4,= inital 1 order story deflection; h = story height; H = latera load accumulated down to level [EP = total gravity loads for levels above and including evel i] Reference [2] indicates that the small stability index Q was selected in order to assure no larger than 10% increase in lateral shears and moments, even if erroneous stiffness assumptions (sufficient to increase the computed sway by about 50%) are selected by the designer. Since the estimate of stiffness was in doubt, this approach minimized the penalty erroneous assumptions might have caused. Notice that service loads are used in both the sway and stability computations. Proper detailing of reinforcing is also necessary to develop the moments and to allow redistribution where necessary. Concentration of reinforcement to develop the unbalanced moment within the “column head” (C, + 3h) should be allocated so that the ratio of top to bottom reinforcement is about 2 to 1 when reversal of moment occurs on both sides (windward and leeward) of the column. This reinforcement is sized to be developed within column dimension ©, and should not exceed 10.375p, in order to allow for ductile behavior and redistribution of excessive moments without spalling of the concrete [18]. Concentrating reinforcement within the “column head” was deemed necessary {0 improve all aspects of slab behavior and maintain the stiffness of the slabs. Eq, (1) was intended to account for a “normal” amount of accidental cracks developed during the construction process (in addition to the design loads), some yielding of the reinforcing over supports ‘and sorne bar slippage (all are usually observed in average construction). Reference [2] suggests that this design methodology is applicable for structures in mild climate zones where lateral wind or seismic forces are low so that the structure is generally behaving elastically. In moderate seismic zones, about 75% of the values given by Eqs. (1) and (2) were recommended. A footnote disclaimer alongside equation (1) in reference [2] reads: “Laboratory testing now (May 1986) underway at the University of Califomia at Berkeley to study vertical and lateral load resistance of flatplate construction should disclaim or verity the accuracy of this equation. “The results of this test at UCB [1] are used to verity the various design methodologies reviewed in this paper. At the time “TWR" was developed (the late 1970's) it was compared to a methodology described by Kahn and Sbarounis [12] and was found to produce slightly more flexible structures. It has been Used since the early 1980's in the design of over two hundred structures.‘The use of any methodology, including “TWR", requires several "judgment" calls to allow it to be sed in actual construction. Slabs in structures are penetrated by many slooves and slots. Often, more than half the supports are alongside siab edges. The designer must use “judgment” in further reducing stitiness based on the location of the slab edges or the size and proximity to supports of the mechanical penetrations. The design “judgment” necessary to account for penetrations and edge conditions cannot easily be described in Code language. Clearly, additional research on this subject is necessary. This author associated the reduction in joint confinement of an edge column (where the lateral ‘moment is parallel to the edge) with a 20% reduction to the effective width of the slab. This value was riot based on laboratory studies, but was selected as representing one side lost which contributed torsional capacity to transfer moments (for square interior supports with both transverse faces intact, 40% of the unbalanced moments are Code prescribed to be transferred by other means than by direct flexure). Exterior columns with the edge of the slab perpendicular to the direction of moment ‘also suffer from reduction in joint confinement and will elongate or shorten due to axial chord action so that a portion of the captive moment will be distributed. Therefore, a similar reduction (20%) was used for these columns as well. The effective width of slabs at comer columns (with two faces devoid of slabs) was reduced to 64% (0.8 x 0.8) of the effective width computed by Eq. (1). Methodology “TWR’, as described above, was extensively reviewed for compliance with the results of the UCB [1] test. The comparison of measured dift results to computed results (at load levels producing drifts of hy/400) indicated a need for a small adjustment to Eq. (1) when the fat plate is very thin and d/h < 0.9. An earlier parametric study by this author [15] evaluated the parameters influencing the effective ‘moment of inertia, !,, by ACI Code Equation (9-7) for slabs and beams supporting gravity loads. This parametric reviow was made forthe parabolic moment envelope (the results of gravity loads acting alone) and not for the combined moment envelope of gravity and lateral loads (a thorough research of this item is stl needed). However, section 13.3.1.2 of ACI Commentary [6] indicates that reasonable estimates of sttiness of beams (E |) for lateral loads may also be obtained by using Code equation (@-7) to compute |, Reference [15] suggests simplified equations to compute |, directly: When MMe > 1.6: L/h= 01MM), 20.351, Eq. @) Where K, = (d/0.9h) (1/10.4 + (1.4 MY/OM,) (fy/100,000)]) Eq. (4) Iln Eq, (4) the value enclosed in the bow brackets {} is equal approximately to unity ‘when fy = 60,000 psi, therefore, In this common case, K, = d/0.9h]{tis thus recognized that |, / lis, in part, dependent on the ratio o/0.9h, When this value is reduced, as in thin slab members, the stifiness of the slab is also reduced. Therefore, itis necessary to make adjustments to ot/,of thin slabs by multiplying OZ, in Eq. (1) by d/0.9h. This small adjustment brings measured and computed drifts in close conformity to each other. Ez, (1) at interior supports could now be expanded to also include edge, exterior and comer supports along side adjustments for thin stabs as follows: th, = [0.3 b+ Cr (yl) + (Ca CiN2] (A0.9h) (Kes) Eq. where Ker .0 at interior supports = 0.8 at exterior and edge supports .6 at comer supports ‘The effective width of edge supports (with the edge parallel to the direction of the lateral load) requires a final step to adjust the results of Eq. (5) by the factor (t+ & /2)/ 4 described eartier. Eq, (6) describes the effective width of stabs which have degraded in stiffness by lateral loads, ‘causing a critical story sway of about h/400. ‘The results comparing predicted behavior of the UCB test by the above described procedure for a range of story drift indexes are displayed later in this paper, under subheading “TWR" (Tables D, E, F, end G). Methodology “JSG” Description of Proposed “Effective Width” Rules Methodology “JSG" is an offshoot of Methodology “TWR” and is one of many sensitivity reviews ‘whose purpose is to evaluate how the various parameters influence the effective width of flat plates and to ascertain if improvements in the accuracy of predicting behavior can be realized. In this ‘methodology all computational procedures are identical to the procedures used in Methodology “TWR,”" except Eq. (6) replaces Eq. (6). In Eq, (6) the average of the clear spans 4, (rather than cic spans) is used and 4 is set to be smaller than or equal to 4; (the latter adjustment is incorporated in order to examine if Vanderbilt's recommended adjustment for EFM [4] has validity here). Smaller effective width values were the result of these adjustments which, in turn, produced somewhat larger drift estimates. Oly = 10.3 by + Cr (x) + (Co- Ci/2] VO.8h (Key) Eq. (6) where x =4/t, $1.0‘The results found by comparing the predicted behavior of this methodology to the measured resulls of UCB tests [1 ] are displayed later in this paper, under subheading "JSG" (see Tables D, E, F, and @). Methodology “HWNG” Description of Proposed “Effective Width” Rules ‘This methodology was developed by the researchers of the UCB tests, Itis described in depth in Reference [11] and is summarized below. For interior supports and edge connections with bending perpendicular to edge. Of = (20, + 4/9)B Eq. (7) For edge supports with bending parallel to the edge. Ot, = (Cy + 4/6)8 Eq. (8) where B = 5Cy/- 0.2 (LL40-1) > "y Eq. (9) ‘or, approximately, 8 = 4 (Cyt) 2 "fy Eg. (10) (LL= ive load or construction loads and ff accounts for loss of stifress under loads). In the following sensitivity review the approximation for f given in Eq. (10) is used. No other adjustments (similar to Kp for “TWR") are made at edge and comer columns, as none were recommended by the researchers and the results of inital studies indicated that larger discrepancies between measured and computed deflections will occur when such adjustments are employed. ‘The results found by comparing the predicted behavior of this methodology to the measured results of the UCB tests [1] are displayed later in this paper, under subheading "HWNG" (see Tables D, E, Fand G). It was observed that Methodology "HWNG" more closely described the effective width Of the slab at a load level, causing a critical story sway of about h/200. The UCB Test Description and Adjustments Needed Figures 1, 2a and 2b indicate the variety in geometry. Tables 1 and 2 indicate the chronology of the Flat Plate tests and their results at UGB [1]. This information is duplicated from the UCB report with the generous permission of Professor Moehle. 10The purposes ofthis test, to review the Code's gravity load requirements and to estimate the lateral stiffness of flat plates, were further extended to see if reduced Code requirements would provide satisfactory behavior. The placement of reinforcing at the south haif of the test model followed ACI Code minimum requirements, while liberties were taken to cause a large redistribution of ‘moments in the north half by reducing the amount of negative steel and increasing, where needed, the positive steel reinforcement. This test indeed verified that large redistribution is possible for both gravity and lateral loads. However, the overall reduction in total reinforcement, especially n top steel, (Gee Fig, 2a) did reduoe the stiffness of the north half of the test slab. In order to develop design criteria for the effective width of a flat slab, the effects of the reduced sfifiness of the north half had to be considered. This was accomplished by dividing the measured slab-column connection stiffness in the north half by the measured stiffness of a similar connection in ‘the south half of the test model, The average of two tests in each direction were used to estimate this slifiness ratio, R1 (NS400 with NS200 in the N-S direction and EW400 with EW200 in the E-W direction). The results are isted in Table A. The drift index of 400 is in close proximity to the recommended drift index of 500 in reference [2] for wind design. The dift index of 200 is ‘approximately the recommended index for seismic in several national Building Codes. Thus, these tests were selected (over others listed in Table 1) to estimate the reduction in stiffness in the north halt. Gravity ioads of the single flat plate level tested were quite small and could not provide the stiffening effects larger gravity loads provide to columns in high rise structures. The test model columns were heavily reinforced and the cracked moment of inertia was used in the analysis [1]. Table B indicates the ratio of R2 = |,/, of the supporting columns forthe test model. The stiffness adjustment, RS (see Table C), for the effective panel width in the North half of the test model, is obtained by averaging Rt values (slab-column connection stifiness rato from Table A) at each end of the panel. Comparison of “Effective Width" Rules to Test Data First order analysis (P-A is negligible) by SAP9O software [16] was ultlized to verify the accuracy of the design methodologies tested. Rigid joints were assumed in the analysis. The actual constructed sizes (such as 3.3" average slab thickness) and measured modulus of elasticity as presented in (1] were used, The corrections to stitiness based on the softer north haif were applied by multiplying the modulus of elasticity, E, of the members in the north half by R1, R2 and FR. ll‘The stitiness ofthe test model in the E-W direction was influenced similarly by the softer north- half, regardless of the direction of the load. However, in the N-S direction, when the load action was toward the north, the influence was greater because the limited top stee! at the north building edge (in tension for this loading direction) provided minimal stifmess. For that reason evaluating the results in the N-S direction should lean more heavily on tests with the load acting south. In the E-W direction either test (load east or west) can be used. ‘The results of the three design methodologies described in this paper are compared to the test results at UCB [1] (see tables D, E and F). The measured column shears were considered to provide the best means to gauge the accuracy of the design methodologies. In Table D, column shears from S400 (total shear of 9100 Ib in the southerty direction resulting in a drift of 0.12"), and from EW400 (total shear of 11,470 Ib in the westerly direction, resulting in a similar drift) are compared to the predicted computer anaiysis of the three design methodologies and are listed in reverse order of compliance with the UCB test results. ‘The computed columns shears for each of the above desoribed methodologies are tabulated alongside the test results. The non-square supports (Tables D-2 and D-4) were separated from the square supports (Tables D-1 and D-3) to better determine the effect of member geometry. Standard deviations (assuming test shear results as a base 100%) are also provided. Finally, dit predictions by the three methods and standard deviation for shear distribution for the total structure are shown in Table D5. ‘Table-£ for drift index of 200 and Table-F for drift index of 800 are similarly provided withot further adjustments to member stiffness (Le. stiffness of members remain unchanged in each of the Tables D, E, and F). Discussion of Results ‘The purpose of this review is to establish a "rational" design methodology to estimate the stiffness contribution of flat plates. Many designers will provide a structure with sulticient stiffness so that for the common service design load level (50 to 100 year wind force or mild to moderate seismic forces) a sway index of 400 (or better) for the critical story is realized, This is the load level the design methodologies reviewed should be tailored to satisfy. The results of the comparison of these three ‘methodologies to the UCB test at a sway index of 400 are tabulated in Table D. Table E tabulates the comparison between the methodologies at larger drifts (h,/200), which are more likely to occur at moderate to more severe seismic design loads. Table F provides the comparison at lower load levels 12(6 to 10 year return) for which serviceability (comfort, perception to motion) are being reviewed and which may cause sways not to exceed approximately (800). ‘The Code's Commentary [6] (Section 18.3.1.2) suggests that a conservative assumption of drift is appropriate for structures of unbraced frames. However, in structures having dual-systems (shearwalis and frames) a more accurate prediction of slab stitress is desirable so that the frame members are property proportioned to resist the lateral forces and moments their relative stiffness will attract. Since the dual-system is more dominant in structures where drift limits are of concern, this review will attempt to develop a methodology which accurately predicts the sway. ‘Tables D through F have been prepared using Eqs. (2) and (5) for methodology “TWR", Eqs. (2) and (6) for “JSG”; Eqs. (7), (8) and (10) for“HWNG." These equations have not been adjusted, as yet, for the different levels of stiffness anticipated at each load level of design concern indicated above. The ratio of computed to measured difts in Tables D through F allows us to determine a "stiffmess degradation’ factor, Ke, 80 that a single unified methodology can provide a measure of the flat plate contribution to stiffness at any load level of design concem (see Table G). DRIFT PREDICTIONS: ‘The ability for methodology “TWR" to predict drifts is sufficiently accurate for design purposes as it “straddles” the measured test results. The computed/measured ratio (see Table D) is 92% and 108% in the N-S and E-W directions respectively for dit index hy/400. In most residential structures (in which fat slabs are paramoun)) rectangulaity of panels is random and is shared in both directions. ‘Therefore, itis likely that a small over-estimation or under-estimation, based on the panels rectangulatity, wil be balanced in the average structure. This methodology is, therefore, verified to provide “rationat” drift values at load levels appropriate for wind induced sways (hy/400), without additional adjustments to stifness (Ky = 1.0). ‘Adjustments for stiffness degradation, K, = 0.8, straddles the computedimeasured ditt (see Table E) for Methodology “TWR’ at a dit index of 200 and will provide sutficiently accurate drift results at this design level. For a drift index of 800 itis necessary to increase the effective width to 1.1 (02,) (therefore, K,= 1.1) in order to more accurately estimate the available stiffness. Similar corrections are also adequate for Methodology “JSG". Methodology “HWNG" provides overly conservative drit estimates at a drift index of 400 and is more suitable to estimate drifts at a drift index. of 200 (for which K, = 1.0). Corrective ratios (not provided by the researchers) of K, = 1.33 and Ky= 1.45 are appropriate to increase stifiness at drift indexes of 400 and 800 respectively. ‘The increased stiffness can be realized by multiplying f obtained from Eq, (9) oF (10) by Ky. 13‘When the various corrective measutes (K, ) indicated above are applied to the computed eftective widths, the ratios of computed to measured drifts indicated in Tables D through F will be adjusted (neglecting the small column contribution to drift) approximately as shown in Table G. TABLE G Computed Drift / Measured Drift - Using Adjustment K, Mathodalogy Methodology Methodology Diift Index "HIG" “yse" wr NS EW Ag NS EW Avg. NS EW Ag. hy400 (94% 106% 100% 108% 111% 107% 92% 108% 100% Kyet.s3 Ke10 Ke 10 ny200 105% 110% 108% 106% 109% 108% 96% 106% 101% Ke10 Ky=08 K=08 800 98% 107% 100% 100% 111% 106% 91% 107% 99% Kea 145 Ket Katt With these K, adjustments, more accurate estimates of sways at various load levels may be realized from any of the three methodologies reviewed. Methodology “TWR", however, also provides {as will be shown below) the greatest accuracy In distribution of the lateral forces to the individual members. The adjusted stiffness of Methodology “TWR" are superimposed on Figs. 3 and 4 (Figs. 6.9 and 6.10 in Ret.1 - these figures compare the accuracy of the various methodologies described in Ret.1). ‘SHEAR DISTRIBUTION PREDICTIONS: Tables D through F indicate that design methodology “TWR" provides the most accurate prodiction of shear distribution in square and non-square columns at all drift indexes. For example: ‘At drift index of 400, standard deviation for shear distribution into all the columns is 6.6% in the North- South direction and 7.1 % in the East-West direction (see Table D-5). Results for methodology “JSG” indicate somewhat less accuracy in predicting the distribution of lateral shears to the supports. Vanderbilt's recommendation to limit 4 < ¢, for the torsional link of the “EFM” Model did not improve the accuracy of shear distribution for the “effective width" Model and is, therefore, not warranted here. For example: At drift index of 400, standard deviation of 6.6% in the N-S and 7.7% in the E-W (see Table D-5). Methodology “HWNG* totally ignores parameter ¢, and the rectangularity of support in computations for 0.¢;. It is shown to have reduced accuracy in distributing the lateral shears to the vatious members. For example: At drift index of 400, standard deviation of 10.8% in the N-S and 10.1 % in the E-W (see Table D-5). 14“Rational” Design Recommendations For Structures With Flat Slabs Structures must be investigated at diferent stages for different design parameters. In the design for wind, serviceability for non-structural elements (such as partitions and cladding) must be reviewed at design level forces. ‘The threshold of non-structural damage to partitions is estimated to occur at a (service load level) drit index of about 400. If not exceeded, ths critical story drift index will also minimize an adverse increase in P-A effects, should the initial estimate of stifiness by the designer be significantly erroneous. This will influence the design for the whole structure (for design load levels) 10 be at about 500 or more to keep the more critical levels at 400. In addition, for certain lightweight and slender structures [17] serviceability to minimize perception of motion must be also reviewed. In this case, stiffness dependence is only part of the issue - perception of motion is a function of mass, damping, a host of other parameters, and, toa lesser extent, of stiffness. This review for serviceability is made for a more frequent occurrence of (even though lower) wind loads. At this load level, the structure's period is shorter and its damping is lower. One may anticipate a drift index of the order between 2000 and 800 for this review. For seismic design a more inelastic behavior is anticipated. However, Code dit limits are more iberal (200 2). ‘The dynamic properties of the structure, its damping and periods, will influence the design loads {or both wind and seismic design. Wind tunne! testing will predict larger wind loading for larger period buildings because wind gust energy is concentrated at high periods. Seismic loads will generally be reduced for tall buildings having periods 2 seconds or larger, once the structure Is forced to enter a post-yield state. Additional cracking at the post-yield state will increase the damping and elongate the periods of the structure. Both these functions will reduce the lateral loads on the structure. ‘Therefore, itis proper to obtain base shear for seismic loads in tal structures using the partial secant stiffness at a dit index of 400 which will provide a threshold of intial yielding in the structure, Measured damping, at ambient load levels, of recently constructed concrete structures (with most of the non-structural elements not yet in place), varied between 1% and 1'/,%. Referenoe [17] proposed that for generally elastic behavior, at design wind forces, 2'/,% minimum damping could be anticipated. For serviceability at intermediate wind forces, to estimate perception to motion, 1'/, - 2% ‘damping range should be assumed. For seismic events, anticipating larger excursions into the Inelastic range, 5% damping has usually been assumed for design purposes by the profession with much larger 10 - 20% and more anticipated just prior to collapse. Setting design criteria for h/400, h/200 and h,/800 for non-homogenous material such as concrete is a complex proposition. The stiffness of the structure and its damping at any of these dit 15levels are also dependent on previous events. A review with pre-assigned anticipated degradation levels, to establish periods and range of damping for each design parameter, should encompass any eventual probability of design requirements. For elastic (wind) design, a limit state having lower member stitiness of about 70% of the “rational" stiffness proposed at service loads is now required by the 1995 ACI Code. For seismic an additional evel of stitfness degradation must be reviewed (the tines just prior to a loss of 20% + of ultimate capacity) in order to evaluate collapse-prevention requirements which will test the structure's strength and ductility to the utmost. The UCB tests dicated (Table 2) that such a loss of strength did not occur until the structure exceeded 2.0% drift. Proper design for life-safety should consider limiting dit, for unteduced base shear loads, to about 1.0% of the structure's height. At this state, the stiffness of the structure may degrade to approximately 50% ofits stiffness at hy/400 (see Table 2), ‘Ambient measurements ofa structure's periods and damping can be reacily obtained from the completed structure, Preferably, this is done before mast of the non-structural elements are added. [At such time, the structure is 6-18 months old (depending on its size). Ithas undergone several wind storms, as well as experiencing the unkind influences of the construction procedure and construction loads. A large percentage of shrinkage and creep is also present. itis difficult to use ambient measurements to predict the dynamic properties of the structure at later dates at load levels which require design for serviceability and strength. Measurements: must be taken at these load levels, or laboratory testing (euch as provided, on a limited scale, by the test at UCB [1]) must be performed. ‘Ambient measurements do provide, however, a more reliable starting point from which to project and verify the design assumptions and, if they are found to be amiss, o provide the artical means (tor ‘example, dampers) to improve future behavior [17]. ‘Ambient measurements of “weathered” (old) structures taken at a low level of lateral ioads will ‘depend heavily on the past historical events influencing the structure, In most cases, such ‘measurements of the dynamic properties will indicate longer periods and larger damping values than in newly constructed structures. A UCB [1 ] recorded “ambient” reading is taken (see Table 2, LAT) at a sway index of 800 in the N-S direction. The intial stiffness recorded at this level is 177% of the average NS400 stiffness. In reference to this value, it might be reasonable to assume that “ambient” readings of frame structures will be, on average, between 1.5 to 2 times the frame stiffness at design load level - resulting in a sway index of 400. It can thus be anticipated that the predicted period of slab-column frames at a drift index of 400 will be about 20-40% longer than the measured period at “ambient.” It is expected that non-structural elements, if present, will shorten the ambient measurements and this must somehow be discounted. Conversely, an older building which has serviced many wind storms will have reduced stifiness (longer petiod) at ambient, closer to the period 16at the design level. The engineer's review of the available facts must be included in the process of evaluation. This summary points out the need for additional information, which can only be possible if actual. structures are monitored. There is a need to measure both newly constructed and "weathered" structures in order to verify the methodologies used in their design and to determine the affect non structural elements provide. A few (though, not enough) measurements foward this end have already been taken {20}. The process of verification of the proposed design methodology has also been started with encouraging initial results. See Appendix B for preliminary correlation between the methodology and the field measurements of the dynamic properties of newly constructed structures uliizing ambient (small wind) forces and the largest wind storm encountered to date. For Methodology “TWR’ -- tentatively, based on UCB tests (until adcitional tests involving actual structures at higher load levels, not yet encountered, can verify or refine such values), Ky, a factor estimating effective width degradation in stitfness at various load levels (assumed equal to unity at h/400), may be taken to equ At“Ambient” 1.5 SHEAR ''SHEAR” “SHEAR SHEAR ey a o oe aT 2 2m at a ae be bs Te ts 6 e we peo bt bt Tor a Tora, Ger SHEAR TOO% SHEAR 100% STDDEWATION “183% woz orm | Sr pEWATION 155% 93% co ‘Table €-2 at Rectangular Supports N-S ‘Table E-4 at Rectangular Supports E-W TEST MWNGEDOS SaamS TWRAOOS HWNGZoOW JSGaoON TwWRZOOW 1p SHEAR "SHEAR “SHEAR «SHEAR. 0 SHEAR | SHEAR. o 6, o reat Ce tra e 10% e 1697 @ 10% et 0, a -- 1% a = a ee es oo 8 - aaa e ee e - a a mt a Sot Tora be CA, a SHEAR Tom “01% SHEAR OO% STD DEWATION “ke SToDEWATION 194% 119% 2% Table £-5 Total Structure N-S Total Structure E-W SHEAR 181715151 15148 15150 Tet 17529 17323 "79 STDEV 110% 1% __ Ath Nos 105% Deruecr 020 OZ5t ope oe 0284 0208 oe RATIO (GOMPMEAS) 105% 5% Th 10% am ey 29Table - F ‘Sensitivity Review - Comparing Shes Table F-1 at Square Supports N-S tn Columns. (h/800) ‘Table F-3 at Square Supports E-W TEST HWNGeo0s igGa00s Twasoes Test HNNGeooW —JgGenOW —_TwREDOW] to staan “SHEAR” Senn "SHEAR io sean “SHEAR “SHEAR SH “wm 1 12 «eK 1 ro ee o 2 s & ry 2 wey as Pa zs be 20 fim eae Tatts iam Toa BB we ay Tora, 7 Bian “toe Tame Ga oo Setar “0% ‘a DeWaT 45% oz 5% ‘STDDEVATON 144% «106% =O Table F:2 at Rectangular Supports NS Table F-4 ot Rectangular Supports E-W TEST HANGHOOS SGaqs TwREnES] TEST HWNGeCGW sSGe0ON — TwREODW 1 __shean SHEAR ‘SHEAR © Saean "SHEAR ow 27, o How os 5 a 3 : 210% 2 Ss 2 a fa ot o a e e e@ a a Jorn 2081808 Bie Tora 390, Srean “dome ‘Som “oe SHEAR 0% STDDEVATON 160% aT ‘sro pewATION Table F-5 Total Stucture NS Total Structure EW ‘SHEAR eG ery 9 =e 2 Stooev 1% 0% ae 20% DEFLECT 0.060 RATIO (COMPAIEAS) 0950 110% TOACKNOWLEDGMENTS Funding for this correlating review was provided primarily by the firms of Rosenwasser/Grossman Consulting Engineers, P.C. and WWES, Inc. | would like to thank the generous volunteer efforts of ‘several individuals in measuring and evaluating the dynamic properties of the structures involved in this review: H. Gavin; S. Yuan; K. Jacob; E. Pekalis; N. Barstow; S. Horton land the Lamont/Doherty Earth Observatory at Columbia University (which provided the geophones, accelerometers and recording instruments) “The original research at UCB was mainly financed by the University of Califomia at Berkeley; RCRC (Task 51); with additional funds provided by the following firms: Rosenwasser/Grossman; Rose Associates; S&A Concrete Co., Inc.; Timko Contracting Corp.; and Harry Macklowe Real Estate Co., Inc. (all of wnom are located in New York Ct) | would also lke to acknowledge the tremendous contribution Prof. Jack Moehle and his students (the UCB Research Team) provided toward the advancement of flat stab construction, BRIEF BIOGRAPHICAL SKETCH Jacob 8. Grossman, FACI, PE, is partner in the firm of Rosenwasser/Grossman Consulting Engineers, P.C., New York, NY. The fim specializes in high-rise construction. He has been a member of this firm for 38 years. Mr. Grossman is currently a member of the Reinforced Conorete Research Council; a past member of Conerete Material Research Council; a past member of ACI Committee 318, Standard Building Code ‘and is currently a consulting member to this Committee. He is and has served as a member of several other Technical Committees at ACI and has also served on the ACI Board of Direction. Mr Grossman received the 1987 ACI Maurice P. van Buren Structural Engineering Award and in 1989 he received the Alfred E. Lindau Award. Mr. Grossman holds degrees from the University of Calfomia at Los Angeles and the University of Souther California. 31BUILDING A BH] pus-rounparion. #2. STORIES, Oh’ BTORY HEIGHT, ‘TOTAL HEIGHT ABOVE GRADE: © 537-0", Bh SLAB THICKNESS, n = ” oous £ BOZTONSS, PT. ROCK: FOUNDATION, STORIES, ee BEBY# STORY HEIGHT. TISTAL “HEIGHT ABOVE GRADE = 454’0% SECs SLAB THICKNESS,108 a dine of _symmetr: a 96x 48] 12.8% 64 af 6.46.4 6.4 6.4 9.6 x 4.8 96x 48 PLAN = x ol Q “pinned” support (typ. SECTION A-A (All units are in inches, 1 inch = 25.4 mm) Fig. 1 Layout of Test Slabof Tet Sab Fig 2(9) Bottom Stee Mat 28 Fit. 2(a) Top Stet Mat of Test SlabLATERAL-LOAD STIFFNESS (kip/in.) 120: 90-4 30+ 1/400 Equivalent Frame (EFM) ~8- Effective Width — Cracked FM ~ “Cracked Torsion Element —- EFM Vanderbilt and Corley -h EFM =Crecked Slab Beam ond Torsion Element 1/800 1/200 LATERAL DRIFT — NS “Tue Lateral Suiffaetes of UCB Tes Sab (NS Dir.) > 4160 120-4 40-4 LATERAL—LOAD STIFFNESS (kip/in.) s-- Effective Width — Finite Element 7 ivolent, Frome (EFM) “EFM =Cracked Slab Beam eM “Cracked Torsion Element EFM =Venderbitt ond Corley eM “Cracked Slob Beam ond Torsion Element 1/400 1/800 1/200 LATERAL DRIFT - EW ~The Lateral Sitseses of UCB Test Sab (EW Dir.) Fig & 4oFig. 4 Layout of Test Slab at UCB [1] Fig. 2 (a) Top Steel Mat of Test Slab at UCB [1] Fig. 2 (b) Bottom Steel Mat of Test Slab at UCB [1] Fig. 3 Lateral Drift - NS Direction comparing “TWR” with methodologies reviewed in UCB [1] Lateral Drift - EW Direction comparing “TWR” with methodologies reviewed in UCB [1] Fig. B-1 West Elevation of Building A Fig. B-2 East Elevation of Building B Fig. B-3 Lower Typical Floor of Building A 32Fig. B-4 Lower Typical Floor of Building B TABLE B1 MEASURED VS. COMPUTED PERIODS AT AMBIENT CONDITIONS TABLE B2 MONITORING BUILDING A During Wind Storm March 4, 1993 Table-1 Chronology Of Tests On The Flat Plate Tested At UCB [1] Table-2 Measured Global Lateral Stiffness At Peaks [1] Table - A ‘Sttinoss Adustmants for Noch Hal Slab-Column Connectone Table - B Ratio of Cracked / Gross Column Momort of nara Re lath Table - C ‘Stfinose Agustmonts Ri," for Efacive Pane! Within Norh Hall of Tast Sib 33Table - D Sensitivity Review - Comparing Shear in Columns. (),/ 400) Table - E ‘Sensitivity Review - Comparing Shear In Colurons.(h,/200) Table - F ‘Sensitivity Review - Comparing Shear In Colurane. (h,/200) 34