Transcription

1

2

3

4

5

6

7 Evolution of Seismic Design Provisions in U.S. Building Codes Ghosh, Dr. S.K. Abstract Seismic design provisions in building codes of the United States have undergone profound and far-reaching changes in recent years. This paper provides an overview of the major trends that have characterized those changes. Trends in the broad areas of seismic input, site classification and site coefficients, triggers for seismic detailing requirements, and performance basis of seismic design are examined. Future trends are briefly commented upon. Seismic Input The seismic input used in seismic design has changed in a number of significant ways in recent years. Through its 1985 edition, the Uniform Building Code (UBC) (1) used a Z-factor that was roughly indicative of the peak acceleration on rock corresponding to a 475-year return period earthquake (an earthquake having a 90% probability of non-exceedance in 50 years). There was only an equivalent lateral force procedure of seismic design. The upperbound design base shear or the flat-top (or constantacceleration) part of the design spectrum was soil-independent; the descending branch or the period-dependent (or constant-velocity) part of the design spectrum varied with 1/T 1/2 and was modified by a site coefficient S; there was no lower-bound design base shear. The Applied Technology Council (ATC) Tentative Provisions (2) in 1978 introduced two spectral quantities: A a = EPA/g, where EPA was the spectral (pseudo-) acceleration divided by 2.5 (the division bringing EPA close to the peak acceleration on rock), and A v = EPV (in./second) x 0.4/12 (in./second), where EPV was the spectral (pseudo- ) velocity divided by 2.5 (a quantity close to the peak velocity on rock). Both quantities corresponded to a 475- year return period earthquake. The National Earthquake Hazards Reduction Program Provisions (NEHRP 1985, NEHRP 1988 and NEHRP 1991 (3)) used the same spectral quantities as seismic input. The acceleration-governed part of the design spectrum was soil-independent, except for a lower plateau for soft soil sites; the velocity-governed part varied with 1/T 2/3 and was modified by a site coefficient S; there was no lowerbound design base shear. The Z-factor of the 1988 UBC became indicative of the larger of two quantities: Aa and Av within a seismic zone. The constant-acceleration part of the design spectrum remained soil-independent, the lower plateau for soft soils was eliminated; the constant velocity part now varied with 1/T 2/3 and was modified by a site coefficient S; a soil-independent minimum design base shear was added in the equivalent lateral force procedure. All of this remained unchanged in the 1991 and the 1994 UBC. The 1994 NEHRP Provisions (3) used soil-modified spectral quantities as the ground motion input. A a was modified by a short-period site coefficient F a, yielding C a ; A v was modified by a long-period site coefficient F v, yielding C v. Ca defined the upper-bound design base shear; C v /T 2/3 defined the descending branch. Thus, the constant-acceleration part of the design spectrum for the first time became soil-dependent. There was still no lower-bound design base shear. The 1997 UBC was similar to the 1994 NEHRP Provisions, except that a single Z-factor was still used to generate shortperiod as well as long-period seismic input. C a of the 1997 UBC was the Z-factor modified by a short-period site coefficient, F a ; C v of the 1997 UBC was the Z-factor modified by a long-period site coefficient, F v. C a defines the flat-top part of the design spectrum; C v /T defines the descending branch. Note the change from 1/T 2/3 to 1/T. Two minimum design base shears are prescribed in the equivalent lateral force procedure one applicable in all seismic zones, the other applicable only in Seismic Zone 4. The higher minimum governs when both values are applicable. The minimum value that applies in all seismic zones is soil-dependent; the other minimum is soil-independent. The 1997 and subsequent NEHRP Provisions (3) and the International Building Code (IBC) (4) use soil-modified spectral accelerations: S DS = (2/3)F a S s and S D1 = (2/3)F v S 1. S s and S 1 are spectral accelerations at periods of 0.2 second and 1.0 second, respectively, corresponding to the maximum considered earthquake on soft rock that is characteristic of the western United States. The maximum considered earthquake has a 2 percent probability of exceedance April 2011 Journal of SEWC 5

8 Evolution of Seismic Design Provisions in U.S. Building Codes in 50 years (an approximate return period of 2500 years), except in coastal California where it is the largest earthquake that can be generated by the known seismic sources. Two-thirds of the maximum considered earthquake replaces the design (500-year return period) earthquake of older codes. S DS and S D1 define a spectral shape that changes from location to location, whereas in the past, the same spectral shape was scaled down from areas of high to low seismicity. The flat-top part of the design spectrum, defined by S DS, is soil-dependent. The descending branch of the design spectrum, defined by S D1 /T, is also soil-dependent. So is the minimum design base shear that is prescribed for all Seismic Design Categories in the equivalent lateral force procedure (Seismic Design Category is discussed later), except that in the 2006 IBC, this minimum has been replaced by a much lower soil-dependent minimum value a change that is to be reversed in the near future. A second minimum base shear is prescribed in the equivalent lateral force procedure for buildings assigned to Seismic Design Categories E and F or for any building located where S 1 = 0.6g. This second minimum is soil-independent. Designing for two-thirds of the maximum considered earthquake provides a uniform level of safety against collapse in that earthquake, 2/3 being the reciprocal of 1.5, the lower-bound margin of safety built into seismic design by U.S. codes (as established by surveys). As long as the 500-year return period earthquake was the design earthquake, the level of safety against collapse in the maximum considered earthquake was non-uniform across the country. This is because in coastal California, the maximum considered earthquake ground motion is only about 1.5 times as strong as the ground motion in a 500- year return period earthquake, whereas in the Midwest and the East, the maximum considered earthquake ground motion may be four or five times as strong as the ground motion in a 500-year return period earthquake. Site Classification And Site Coefficients The 1994 NEHRP Provisions3 brought about a major change in site classification and site coefficients used in seismic design. The new scheme was adopted (with necessary modifications) into the 1997 UBC and has been adopted (again with necessary modifications) into the 1997 and subsequent NEHRP Provisions (3) and the IBC (4). The significant changes from prior seismic design are as follows: 1. Site Classification - The four Soil Profile Types (S 1 through S 4 ) of the 1994 UBC have been replaced by six Site Classes: A through F. In the 1994 UBC, S 1 was rock, S 2 was intermediate soil, S 3 was soft soil, and S 4 was very soft soil. There are now two categories of rock. Site Class A is hard, geologically older rock of the eastern United States. Site Class B is softer, geologically younger rock of the western United States. Site Classes C, D, and E represent progressively softer material. Site Class F consists of material so poor that to be able to design any structure founded on it, a designer must have a site-specific spectrum and must perform dynamic analysis using that spectrum. 2. Site Coefficients - There used to be one soil factor S; now there are two site coefficients: a short-period or acceleration-related Fa, and a long-period or velocitydependent Fv. 3. Dependence of Site Coefficients on Seismicity - Whereas the old S-factor was a function of the Soil Profile Type only (1.0 for S 1, 1.2 for S 2, 1.5 for S 3, and 2.0 for S 4 ), each of the new site coefficients (F a and F v ), in addition to being a function of the Site Class, is also dependent on the seismicity at the site. F a, F v of the 1994 NEHRP Provisions are functions of A a and A v, respectively. C a, C v of the 1997 UBC are both functions of Z. F a and F v of the 1997 and subsequent NEHRP Provisions and the IBC are functions of S s and S 1, respectively. For the same Site Class, the site coefficients F a and F v are typically larger in areas of low seismicity and smaller in areas of high seismicity. This is directly in line with observations that low-magnitude rock motion is magnified to a larger extent by soft soil deposits than is high-magnitude rock motion. 4. Maximum Values of Site Coefficients - While the maximum value of the old soil factor S was 2.0 for Type S 4 soil, the maximum values of F a and F v are 2.5 and 3.5, respectively, in the 1997 and subsequent NEHRP Provisions and the IBC. This requirement results in significant increases in seismic design forces for buildings (particularly taller buildings) founded on softer soils in areas of low seismicity. 5. Basis of Site Classification - Soil Profile Types S 1 through S 4 were qualitatively defined in the UBC. The structural engineer, after reviewing the soils report, typically determined the Soil Profile Type. This is to be contrasted with the new situation where the distinction among the Site Classes must be based on one of three measured soil properties at the site: the shear wave velocity, the standard penetration resistance (or blow count) or the undrained shear strength. If one of a number of given conditions is satisfied at a site, it becomes classified as F. If one of a number of other given conditions is satisfied at a site, it becomes classified as E. Once Class F and Class E, based on the given conditions, are ruled out, soil property measurements need to be undertaken. It is possible for a site to get classified as E, based on property measurements as well. The properties need to be measured over the top 100 ft (30 m) of a site. If the top 100 ft (30 m) is not homogeneous, it must be divided into layers that are reasonably homogeneous, and the properties of those layers measured. The 1997 NEHRP Provi- 6 Journal of SEWC April 2011

9 Evolution of Seismic Design Provisions in U.S. Building Codes sions, the 1997 UBC and the 2000 IBC give formulas by which to arrive at average soil properties over the top 100 ft (30 m), based on those measurements. The IBC, but not the 1997 UBC or the 1997 NEHRP Provisions, permits the geotechnical engineer preparing the soils report to estimate, rather than measure, the soil properties mentioned earlier, based on known geologic conditions. In the absence of measured or estimated soil properties, the default Site Class is D, unless the building official has determined that E or F may exist at the site. Seismic Detailing Requirements Seismic Zones - In the Uniform Building Code, through its 1997 edition, and in seismic codes, standards, and other documents based on the UBC, seismic detailing requirements and other restrictions such as height limits on certain structural systems depended upon the Seismic Zone in which a structure was located. Zones were regions in which the intensity of seismic ground motion, corresponding to a certain probability of occurrence, was within certain ranges. Seismic Performance Categories - Given that public safety is a primary code objective, and that not all buildings in a Seismic Zone are equally crucial to public safety, a new mechanism called the Seismic Performance Category (SPC) was developed in the ATC 3 document, and was used in all the NEHRP Provisions through 1994, and in all codes and standards based on the 1994 and earlier NEHRP Provisions (BOCA/NBC 1993, 1996, 1999 (5); SBC 1994, 1997, 1999 (6), ASCE 7-93 (7), and ASCE 7-95 (7)). In all these documents, the SPC, rather than the Seismic Zone, was the determinant of seismic detailing requirements (and other restrictions), thereby dictating that, in many cases, the seismic design requirements for a hospital be more restrictive than those for a small business structure constructed on the same site. The detailing requirements for Seismic Performance Categories A & B, C, and D & E were roughly equivalent to those for Seismic Zones 0 & 1, 2, and 3 & 4, respectively. Seismic Design Categories - The most recent development has been the establishment of Seismic Design Categories as the determinant of seismic detailing requirements in the 1997 and subsequent NEHRP Provisions (3), ASCE 7-98, ASCE 7-02 and ASCE 7-05 (7), and the 2000, 2003 and 2006 IBC (4). Recognizing that building performance during a seismic event depends not only on the severity of the sub-surface rock motion, but also on the type of soil upon which a structure is founded, the SDC is a function of location, building occupancy, and soil type. For a structure, the SDC needs to be determined twice first as a function of the short-period seismic input parameter, S DS, and a second time as a function of the longperiod seismic input parameter, S D1. The more severe category governs. The 2003 and 2006 IBC permit the determination of Seismic Design Category based on short-period ground motion alone for short buildings that satisfy certain additional criteria. Impact of Changes from Seismic Zones to SPC to SDC - Clearly, the procedure for establishing the seismic classification of a structure has become more complex. Determining the Seismic Zone of a structure simply requires establishing the location of the structure on a Seismic Zone map. Determining the Seismic Performance Category of a structure requires the interpolation of a ground motion parameter on a contour map, based on the location of the structure, determining the use classification of the structure, and consulting a table. The process leading to the establishment of the Seismic Design Category of the IBC for a structure involves several steps, many of which are rather complex. When ATC 3 in 1978 made the level of detailing (and other restrictions concerning permissible structural systems, height, irregularity and analysis procedure) also a function of occupancy, that was a major departure from prior practice. Now, the level of detailing and other restrictions have been made a function of the soil characteristics at the site of a structure in addition to occupancy. This is a further major departure from recent prior practice across the United States a move that has important economic implications that have been discussed elsewhere (9-11). Earthquake design is no longer just a regional concern. In unlikely places such as Atlanta, Georgia, the equivalent of California detailing may be required, particularly on softer soils. Performance Basis Prior to the 1997 NEHRP Provisions - The seismic design provisions of all U.S. codes and similar documents based on the 1994 or earlier NEHRP Provisions, or not based on the NEHRP Provisions, had the following implicit performance bases: (1) For standard-occupancy or ordinary structures, ensure life safety under the design earthquake, which had a 90 percent probability of non-exceedance in 50 years or a return period of 475 years. (2) For assembly buildings or high-occupancy structures, provide enhanced protection of life. (3) For essential or emergency response facilities, improve capability to function during and following an earthquake. It is generally uneconomical and unnecessary to design a structure to respond elastically to the design earthquake. The design seismic horizontal forces recommended by codes are generally much less than the elastic response inertia forces expected to be induced by the design earthquake. Code-designed structures are expected to ensure life safety under design earthquake ground shaking because of their ability to dissipate seismic energy by inelastic deformations in certain localized regions of certain members. April 2011 Journal of SEWC 7

10 Evolution of Seismic Design Provisions in U.S. Building Codes A decrease in structural stiffness caused by accumulating damage and soil-structure interaction also helps at times. The use of seismic design forces prescribed by codes requires that the critical regions of members have sufficient inelastic deformability to enable the structure to survive without collapse when subjected to several cycles of loading in the inelastic range. This means avoiding all forms of brittle failure and achieving adequate inelastic deformability by flexural yielding of members. This is achieved through proper detailing of reinforced concrete beams, columns, beam-to-column joints and shear walls, rules for which are presented in the materials chapters of codes and in materials standards. Enhanced protection of life in high-occupancy structures was provided for in the Uniform Building Code through the requirement of an importance factor of 1.5 for the anchorage of machinery and equipment required for lifesafety systems. The anchorage design forces went up by this factor. Structural observation, which was required for this occupancy category, also played a role. An importance factor of 1.25 for the structure itself, an importance factor of 1.5 for elements of structures, nonstructural components and elements supported by structures, and structural observation requirements together were used as a means of improving the capability of essential facilities to function during and following an earthquake. In ATC 3, in the NEHRP Provisions through the 1994 edition, and in codes based on the NEHRP Provisions predating the 1997 edition, enhanced protection of life in high-occupancy structures (as well as in hazardous and essential facilities) was attempted to be achieved through the device of the Seismic Performance Category, which combined occupancy with seismic risk at the site of a structure. Higher detailing requirements were prescribed for higher seismic performance categories. For essential or emergency response facilities, improved capability to function during and following an earthquake was attempted to be ensured through stricter limits on interstory drift and Subsequent NEHRP Provisions, ASCE 7-98, ASCE 7-02, and ASCE 7-05, 2000, 2003, and 2006 IBC - The performance bases of the 1997 NEHRP Provisions, on which the seismic design provisions of ASCE 7-98 and subsequent ASCE 7 standards and the IBC are directly based, are different from the above. Hamburger (8) has suggested that the performance bases of the 1997 NEHRP Provisions are as illustrated in Fig. 1, reproduced from Reference 8. For ordinary structures, life safety under the design earthquake and collapse prevention under the maximum considered earthquake are ensured by designing the structure for the effects of code-prescribed seismic forces and by conforming to the detailing requirements in the materials chapters. Enhanced life safety and collapse prevention under the same earthquakes are accomplished through the device of the Seismic Design Category (SDC). It may be noted that essential facilities in so-called near-fault areas are assigned to SDC F, while other nearfault structures are assigned to SDC E. The 1997 and subsequent NEHRP Provisions and codes and standards based on them also assign occupancy importance factors, I, of 1.25 and 1.5 to assembly buildings and essential facilities, respectively, to partly achieve the higher levels of seismic performance desired for these structures. The I-values higher than 1.0 have the effect of reducing the effective R-values, permitting less inelastic behavior and, consequently, reduced levels of damage. From SDC A, B to C to D, detailing requirements increase, and the applicability of certain limited-deformability structural systems becomes restricted. In SDC D, height limits begin to apply on certain structural systems, and dynamic analysis as the basis of design begins to be required for certain irregular structures. From SDC D to E to F, detailing requirements do not change. However, height limits often become more restrictive and more and more restrictions apply to irregular structures. Also, structural redundancy must be considered in the design of structures belonging to SDC D, E, and F. According to Hamburger (8), as shown in Fig. 1, current U.S. seismic design provisions are supposed to ensure that ordinary buildings will be immediately occupiable following frequent earthquakes, that essential facilities will remain operational during and following such earthquakes, and that assembly buildings will exhibit performance between the above two. These performance objectives are sought to be met through imposition of limits on the design story drift,, defined as the difference of the deflections of the center of mass on the top and bottom of the story under consideration. Drift limits for highoccupancy buildings are typically more stringent than they are for ordinary buildings; for essential facilities, they are typically more restrictive than those for high-occupancy buildings. The Future Direct performance-based design, where the design professional together with the owner or his representative choose one or more performance objectives (a performance objective is a desired performance level at a particular ground motion severity or seismic demand), and those objectives then directly drive the design, is still in the future of the U.S. codes for new buildings. Such a performance-based approach is already available in a standard for existing buildings (ASCE 41-06) (12). There is little doubt that such direct performance-based 8 Journal of SEWC April 2011

11 Evolution of Seismic Design Provisions in U.S. Building Codes design is the way of the future. Work on performance-based design more sophisticated than what is incorporated in the ASCE 41 document is currently underway in the United States under the ATC 58 project being conducted by the Applied Technology Council. Provisions for direct performance-based design are likely to replace today's code provisions where the performance basis is implicit, rather than explicit. Author Affiliation President, S. K. Ghosh Associates Inc., Palatine, IL, U.S.A., References 1. International Conference of Building Officials, Uniform Building Code, Whittier, CA, 1991, 1994, Applied Technology Council, Tentative Provisions for the Development of Seismic Regulations for Buildings, ATC Publication ATC 3-06, U.S. Government Printing Office, Washington, DC, Building Seismic Safety Council, NEHRP (National Earthquake Hazards Reduction Program) Recommended Provisions for the Development of Seismic Regulations for New Buildings (and Other Structures), Washington, DC, 1991, 1994, (1997), (2000), (2003). 4. International Code Council, International Building Code, Falls Church, VA, 2000, 2003, Building Officials and Code Administrators International, The BOCA National Building Code, Country Club Hills, IL, 1993, 1996, Southern Building Code Congress International, Standard Building Code, Birmingham, AL, 1994, 1997, American Society of Civil Engineers, Minimum Design Loads for Buildings and Other Structures, ASCE 7-93, ASCE 7-95, New York, NY, 1993, 1995, and ASCE 7-98, ASCE 7-02, ASCE 7-05, Reston, VA, 2000, 2002, Hamburger, R.O., Proposed CRDC Seismic Provisions, presented to the International Building Code Structural Committee, Orlando, FL, Ghosh, S.K., Impact of Earthquake Design Provisions of International Building Code, PCI Journal, V. 44, No. 3 (May-June, 1999), pp Ghosh, S.K., New Model Codes and Seismic Design, Concrete International, V. 23, No. 7 (July, 2001), American Concrete Institute, Farmington Hills, MI. 11. Ghosh, S. K., Impact of the Seismic Design Provisions of the International Building Code, Structures and Codes Institute, Northbrook, IL, American Society of Civil Engineers, Seismic Rehabilitation of Existing Buildings, ASCE 41-06, Reston, VA, Fig. 1 Performance Basis of the 1997 NEHRP Provisions April 2011 Journal of SEWC 9

12 Precasting: When, Where, How? Gian Carlo Giuliani, Italy Abstract Conceptual aspects for using the prefabrication and several applications using liner, planar and spatial elements for civil, industrial and tower structures are illustrated in the article with a mention of the contractor's necessary skill and equipment. Keywords Prefabrication, Concrete, Prestressing, Composite Structures, Buildings, Towers. When? Correct structural engineering evolves in a series of steps starting with the definition of the purposes and characteristics of the work, continuing with the analysis and the verifications and ending with the preparation of the construction drawings and specifications. This process is an extremely valid method that lets face and solve in advance the problems linked to the different design choices, without having to introduce modifications at a later date that means forced adaptations and always results in being more expensive. Precasting is one of the basic choices that needs to be made at the very beginning of the design process; the conversion of a conventional structure into a precast one very rarely turns out to be the best choice or completely free of compromises. One needs to decide on precasting straightaway, evaluating the pros and cons before adopting it or, on the contrary, deciding to go for a conventional structure. However it's always a good idea to also evaluate the alternative steel solution for the entire structure or for part of it. When major buildings are dealt with, all aspects of decision about precasting must be carefully considered again at the design stage, because the system and method of assembly and erection generate temporary actions which may require changes of the structural sizing. Not to mention the technical/cost effectiveness of such a decision. The use of precast elements for the load-bearing structures is today commonplace, when applied to floors by using mass produced girders or panels, and for medium span industrial buildings, where there's a wide choice of secondary and primary solutions for roofs and floors and for the whole framework. We should, however, point out that the opportunities to gain and implement considerable skill and mastery in the field of concrete constructions are becoming fewer and fewer and those companies operating in the field of precasting are often the safe-keepers of the remaining experience and the necessary aptitudes. Where? The well-known advantages offered by precasting (factory and onsite) for large structures need not be mentioned here. It is worth to note that, very often, non conventional precasting solutions result in benefits if backed by an adequate constructor skill. In other cases prefabrication is the unique and rational choice, which is certainly very challenging for the design but strongly competitive against the price, the quality and the delivery time of the conventional construction. The precasting can be effected in a factory or in the site; the selection among these solutions depends on a number of conditions, like the element size, the transport distance and easiness, the number of similar pieces to be manufactured and the cost/benefit ratios related with the existence of factory facilities or with the on site construction of a plant and accounting for a balance of the element transport cost within the above said locations. How? To guarantee the final quality of a project, it is often necessary to conceive and design the structure as a precast construction right from the very start. Many types of precasting systems and configurations can be used, being each one suitable for solving different problems and for complying with the site erection constraints, the element transport conditions, the available equipment and know how of the construction firm. The main types of the above said systems can be grouped in the following categories according to the main dimen- 10 Journal of SEWC April 2011

13 Precasting: When, Where, How? sion of the manufactured elements or units: linear, planar and three dimensional; several relevant examples are given in the following. Linear Precasting Long elements are typically factory produced, in many cases by using prestressing also, the typical use is for components of frame skeletons. Verona's Post Office Building The prefabricated beam and TT floor units for the Verona's Post Office building (figures 1a,b) are supported by solid columns of great height (36 m) and limited cross-section (0.60x0.80 m). Prestressing was applied to the columns to assure stability during transportation and erection. Telecom office building in Bologna In other buildings, precasting has proved to be an excellent idea for just a few elements, such as in the Telecom office building in Bologna (figure 2a). Precast cellular beams span 25 m, bear over the corner towers and carry the load of 5 or 7 suspended floors (figures 2b,c); the post tensioning cables are shown in figure 2d. Elevated connections for the new Milano Fair Exhibition Buildings In this case m span, prefabricated, prestressed beams were used as floor elements overpassing public streets; the m span main beams are composed of a self supporting steel truss which is integrated with a posttensioning cable and embedded in concrete in order to keep the over all structural depth within the limit of 1.45 m (figures 3a, b) Planar Precasting Because of their size, two dimensional units are typically prefabricated on site. Prestressed ribbed precast slabs have been successfully used in several projects to create floors with a great load capacity and so avoiding the use of main beams and secondary elements. The advantages of using slabs come from the structural behavior (loads are transferred directly to the columns) and from the monolithic nature of the element. Not to mention the considerably reduced number of production, storage, transportation, erection and in situ assembling operations The disadvantages arise from the need to prepare a precasting plant on site with forms, reaction beams, an accelerated curing unit and so on. The need of suitable equipment for handling and erecting large elements, with unit weights greater than those of normal precast elements, has to be taken into account also. The costs of setting up the above said equipments are fully compensated by the lower cost of producing similar elements. According to our experience for a total surface area to be constructed between 8,000 and 10,000 m2, the solution with in site precast planar units becomes a competitive alternative to precasting and assembling of separate elements. Poli Laboratories building in Rozzano The prefabricated prestressed ribbed slabs for the four story Poli Laboratories building in Rozzano were designed for a superimposed live load of 12 kn/m2 on a 7.20x8.40 m grid line pattern, while keeping the floor depth at 0.60 m only. Figures 4a,b show the plates, which were launched on rails at the ground level and lifted to the top of the precast columns, and the self stressing form prepared on site. Roof units for the Milano City Fiera exhibition center The roof units for the new Milano Fiera City exhibition center span 20 by 20 meters and are designed for the live load capacity of 6.0 kn/m 2 ; the whole structure was precast onsite. Here monolithic plates are being used with single direction ribs and edge beams (figure 5a); the ribs are prestressed with bonded strands, while post-tensioning cables are used for the main beams. Four groups of hydraulic jacks with strand recovery were used to lift the elements into position. A self-reacting form (figure 5b) was used and moved on rails to every new position; the form edge panel was tilted during this launching to allow passing between the columns. Space Precasting The theme of space precasting is a complex one because of the dimensions of the complete structure and of the complicated three dimensional joints which are subjected to groups of in plane and out of plane actions concentrated in a reduced area which is located outside the solid material of the member. In general the structure can be subdivided in sub-elements with linear, planar or space forms to obtain the final space configuration; in any case, the design and the construction of these members, follow criteria which are different from the ones used in the specific above said categories. In general large span roofs are not easily solved with the use of the precast elements currently produced. Shells capable of covering a span of 20 meters are available, but the main beams for similar distances between the columns are too thick and appear unsuitable from the start due to a high roofing load/own weight ratio. April 2011 Journal of SEWC 11

14 Precasting: When, Where, How? The use of secondary and main floor precast members for large spans and the heavy superimposed loads, which are typical in these building layouts, is impossible also. New-concepts in the load-bearing system are needed to optimize the construction system. Examples of prefabrication of three dimensional structures are given in the following. Aeritalia hangar in Turin For the Aeritalia hangar in Turin, composed of several 30 by 30 meters bays, we designed in factory precast thin pre-stressed shells (figure 6a,b) which were fitted with stiffening diaphragms at the ends to create the web and the flanges of the main beams. The shells were aligned on templates at the site and their diaphragms were subsequently post stressed in order to create the edge main beams. By using hydraulic jacks, the whole 30 by 30 m unit was then lifted up to the top of the columns and suspended to the capitals by means of stay cables similar to the ones used for bridges (figure 6c); the completed hangar is shown in figure 6d. Floor units for the Milano City Fiera exhibition center The first floor of this buildings had to be based on 20 by 20 meters bays for the superimposed live load capacity of 15 kn/m2 and for housing, inside the depth of the structure, the air intake and exhaust ducts for the ground floor areas as well as piping and wiring for the at level located stands. An innovative solution was worked out with the concept of a composite plate featuring concrete top and bottom slabs, connected by means of a shear layer composed of steel pipe struts arranged in a 3D truss pattern. In this structural configuration, the connection between the flanges (conventionally created by webs) has been replaced by struts resisting the axial loads created by the three components of the shear action (figure 7a). The space between the two slabs can be accessed for inspection and plant maintenance. The bottom slab is ribbed and prestressed with bonded strands which cross cast iron nodes embedded in the concrete and the pin connected to the steel pipes of the shear layer (figure 7b). The self stressing form was moved on rails to every new bay position (figures 7c,d). Precast glass fiber reinforced concrete elements placed on the cast iron nodes, located where the pipes converge close to the upper surface, constitute the form for casting the upper slab (figure 7e, f). All the ducts, pipes and other equipment were positioned inside the plate (figure 7g) Four groups of hydraulic jacks lifted the completed plate (weighing a total of 4800 kn, including the already installed ducts) into position (figure 7h); the floor plates lay below the roof units (figure 7i) The columns are an integral part of the structural concept: pre-cast, with an octagonal cross-section. 25 meters high and weighing 600 kn; the outer columns are is sight (figure 7k,m). The support of the composite slab was created by the introduction of steel-concrete elements in the column with the necessary recesses to house the bearings, thus allowing for lifting the plates without any overhanging elements and for a reduction of the column bending due to the location within the cross-section of the actions transmitted by the bearings (figure 7j). The completed building is shown in figure 7l. Air traffic control tower at Malaga airport Because of its location, the control tower at Malaga airport has a strong visual impact and is composed of threedimensional shell precast elements for the segments of the six wide body ribs which constitute the structure (figure 8g), supports the vertical loads and resist the seismic and the wind actions. The above said shells were constructed by using the "matching concrete" technology (figure 8a), i.e. casting each element in a form next to another already cast one, to allow for a "dry" joint erection with a thin layer of epoxy resin and reinforced by post-tensioned bars (figures 8b,c) The ramparts for the stairs and the relevant intermediate landings were precast also. A service building with an annular shape is located at the base of the tower and is covered by hypar thin shells which are supported by X shaped prefabricated elements located around the outer facade (figures 8c,d,e). Because of structural reasons related to the limited thickness of the shells and to the relevant rise to span ratio, the hypar roof was cast in place. Air traffic control tower at Barcelona airport In this case, the use of concrete instead of steel was dictated by the client. The shaft of the Barcelona airport control tower is constituted by prismatic elements with axes lying along straight lines which define a hyperbolic paraboloid; the relevant rectangular sections have a radial orientation. The challenge of precasting these elements was even more demanding, because of the geometry of the shaft design featuring helicoid surfaces, and of the necessary detailing of the connections which were engineered with- 12 Journal of SEWC April 2011

15 Precasting: When, Where, How? out any overhanging parts, (figure 9a,b); because of the innovative solution the Client required a full scale test (figure 9c), which confirmed the design assumptions. The erection was performed using the inner self standing aluminum stair structure as a template (figure 9d); the hypar skeleton and the joints got the correct shape (figures 9e,h). The shaft structure supports the vertical loads, including the steel control room (figure 9f,g), and resists the seismic and the wind actions. The two story building at the base of the tower(figures 9i,j,l) is ring shaped and is prefabricated also by using columns, circumference beams, curved facades and sun shading strips cast with a self compacting concrete mix with white aggregate and cement. The completed tower is shown in figure 9m. Conclusions According to the design and construction experience earned with the illustrated examples, precast solutions are very often highly cost effective if the constructor has the suitable level of expertise required. In many cases, while it may be far more exacting in terms of engineering, precasting is the only available choice because it is far more competitive than on site conventional construction with regard to price, the quality of the work and construction time. Author Affiliation Giuliani, Dr. Eng. Gian Carlo, exclusive Consultant Redesco srl Milano/Italy gc.giuliani@redesco.it; Giuliani, Dr. Eng. Mauro Eugenio, exclusive Consultant and General Manager Redesco srl Milano/Italy me.giuliani@redesco.it a) general view of the skeleton (columns, beams, floor units) b) 36 m long precast prestressed columns Fig. 1 a, b Verona's Post Office Building a) Prefabricated Prestressed beams b) Main steel trussconcrete beams Fig.3 Elevated connections for the new Milano Fair Exhibition Buildings a) general view a) prefab.prestressed slabs b). on site self stressing form Fig.4 Poli Laboratories Building in Rozzano b, c) 25 m span cellular beams bearing over the corner towers and carrying 5 suspended floors d) postensioning cables for the cellular beams Fig. 2 a,b,c,d Telecom Office Building in Bologna a) 20 by 20m roof units scheme b) self stressing form Fig. 5 a,b New Milano Fiera City Exhibition Building April 2011 Journal of SEWC 13

16 Precasting: When, Where, How? a) in Factory precast thin prestressed shells Fig.6 Aeritalia Hangar in Turin c) Whole 30 by 30 unit lifted up a) "Matching Concrete" shells b) Basement erection a) Conceptual composition of the multilayer plate c) Start of shaft erection d) Prefabs and shell connection b) Cast iron joint c) Self-stressing form e) Hypar thin shells roofing the base building Fig.8 Malaga Airport Control Tower: f) The completed tower d) Erection of the forms for the upper slab e) Ducts placed inside the plate before lifting f) The roof and the floor erected g) Column steel elements a) Twisted elements d) Erection of the prefabricated elements using the aluminium stair structure as a template c) Prefabricated facade of the base building i) The Exhibition building in operation j) Column on the facade Fig.7 Floor Units for the milano city fiera Exhibition Center d) The assembled prefabricated hypar sk eleton e) Control room bearing Fig.9 Barcelona Airport Control Tower: f) The completed tower (photo by J.Azurmendi) 14 Journal of SEWC April 2011

17 Cements and Concrete Mixtures for Sustainability Mehta P. Kumar Abstract The climate changes, due to man-made global warming triggered by steeply rising volume of greenhouse gases, composed mostly of carbon-dioxide, is a very serious issue that is being addressed worldwide by every major sector of economy. There is a general acceptance of the view that firm measures must be taken without delay to bring down the global carbon emissions to the 1990 level or less during the next 15 years. The focus of this paper is on portland-cement concrete, which is the most widely used manufactured product in the world today. Cement production is not only energy-intensive but also responsible for direct release of nearly 0.9 tonne carbon-dioxide for each tonne of portland clinker, which is the principal component of modern cements. Fifteen years ago, in 1990, the world production of cement was slightly more than 1 billion tonnes. In 2005, it already crossed 2 billion tonnes which means that direct CO 2 emissions from the portland clinker production have nearly doubled. Fifteen years from now, with business-as-usual, the estimated cement requirement would be 3.5 bilion tonnes, and direct CO 2 emissions from cement kilns would triple the 1990 level. Thus, the challenge before the global construction industry is how to meet the buildings and infrastructure needs of rapidly growing economies of the world, and at the same time, cutting down the CO2 emissions attributable to cement consumption to the 1990 level, in conformity with other sectors of economy. Different options for consideration of the construction industry are presented in this paper. The production and use of blended portland cements containing large proportion of complementary cementing materials, such as coal fly ash and granulated blast-furnace slag provide an excellent strategy for immediate and substantial reduction of direct CO 2 emissions associated with the manufacture of portland-cement clinker. Both EU and North American cement standards now permit more than 50 % clinker replacement in composite cements. Furthermore, the use of composite cements and concrete mixtures containing large addition of complementary cementing materials would yield crack-resisting structural elements of radically enhanced durability. High-volume fly ash concrete applications for recently built structures in North America are cited as typical examples of possible CO 2 reduction. Sustainability - An Introduction During the 1990s, it became abundantly clear that industrialization of the world is happening at an unsustainable speed. Among the major sustainability issues of public concern are high rates of consumption of energy and materials, short service life of manufactured products, and lack of space for safe disposal of huge volumes of solid, liquid, and gaseous wastes generated by human activities. Global warming, the cumulative effect of these problems, has emerged today as the most serious sustainability issue of the 21st century. The term, global warming, refers to the greenhouse-gas effect leading to a steady increase in the earth's surface temperature since 1950s. According to a World Watch Institute report, twenty-four of the last 27 years have been the warmest on record. Weather scientists around the world have concluded that a linear relationship exists between the earth's surface temperature and the atmospheric concentration of CO 2, which makes up 85 % of the greenhouse gases. The current CO 2 concentration, about 380 ppm (mg/l) in 2005, is the highest in recorded history (Fig. 1). With business as usual, it is projected to increase at an exponential rate. In 2006, the annual global CO 2 output reached a staggering 30 billion tonnes. Evidence of global warming is not confined to temperature measurements. The following list includes some of the observable effects of the phenomenon: o o o o o A sharp increase in the melting rates of glaciers, polar caps, and ice sheets. Rising ocean levels - a potential threat to coastal populations. Unusual increase in frequency and intensity of rainstorms, flash floods, cyclones, hurricanes, heat waves, droughts, and wild fires. Adverse impact on current sources of agriculture and water. Disruption of the earth's carbon cycle due to changes in the botanical species on land and oceans. April 2011 Journal of SEWC 15

18 Cements and Concrete Mixtures for Sustainability In a series of reports, issued earlier this year by the United Nations Intergovernmental Panel on Climate Change, leading weather scientists of the world have unequivocally stated that global warming is occurring, and that it has been triggered by human activities. They have warned about devastating consequences of global warming if immediate action is not taken by national and industry leaders to reduce the carbon dioxide emissions to the 1990 level or less. Although climate change is a global phenomenon, it has to be tackled in every country individually by each of the major CO 2 emitting sectors of economy, such as power generation, transportation, and energy consumption associated with the use of buildings, and manufacture of structural materials like concrete and steel. According to Kyoto Protocol, proposed in 1990 and signed in 2005 by 141 countries, the signatories agreed to stabilize the greenhouse gas emissions by 2012 to 6 % below the 1990 level. The two largest polluting countries, the U.S. and China, which are responsible for nearly half of the global CO 2 emissions, have yet to show a willingness to commit to any specific goals. However, in 2005, many multinational corporations, State governments in the U.S., and over 400 mayors representing 60 million Americans have signed on to programs that intend to meet or beat the Kyoto targets by In September 2006, the State of California approved the Global Warming Solutions Act according to which, by 2020, California's CO 2 emissions would be reduced to the 1990 level. Concrete Industry's Environmental Impact The subject of environmental impact of the concrete industry is covered by numerous publications across the world including those listed in References (1-6). The embodied energy content, i.e., the sum total of energy required to extract raw materials, manufacture, transport, and install building elements is only 1.3 MJ/kg for 30 MPa concrete, compared to 9 MJ/kg for recycled steel and 32 MJ/kg for new steel. However, being the largest manufactured product consumed in the world, quantitatively concrete represents considerable embodied energy. Worldwide today, approx. 17,000 million tonnes of concrete is being produced annually. Besides natural resources, such as aggregates and water, the concrete industry is a large consumer of cement - a manufactured product directly responsible for high CO 2 emissions. In 2005, according to Cembureau, the global cement consumption was 2,270 million tonnes. Therefore, carbon footprints of the global cement industry are very significant considering the amount of fossil fuels and electrical power consumed for crushing, grinding and transport of materials, and for the 1400 to 1500 C burning operation to make portland clinker - the principal ingredient of hydraulic cements. The scope of this paper is limited to direct CO 2 emissions, of which approx. 6.3 % of the global emissions are attributable to portland clinker manufacture. Co 2 Emissions From Cement Kilns Typically, ordinary portland cement is composed of 95 % clinker and 5 % gypsum, which is a complementary cementing material (CCM) because it enhances the cement performance by improving the setting and hardening characteristics of the product. Depending on the carbon content of fossil fuels used for clinkering, 0.9 to 1.0 tonnes of CO 2 is directly released from cement kilns during the manufacture of clinker. In addition to gypsum, sometimes other mineral additives, commonly known as supplementary cementing materials (e.g., coal fly ash, granulated blast-furnace slag, natural and calcined pozzolans, pulverized limestone, and silica fume) can either be interground with clinker and gypsum or added directly during the concrete mixing operation. Large quantities of these materials are available as industrial by-products. As discussed in this paper, when properly used, the mineral additives have the ability to enhance considerably the workability and durability of concrete. Therefore, these additives too are treated as complementary cementing materials (CCM) in this paper. Global statistics for 1990 and 2005 on cement production, CCM consumption, and direct CO 2 emission attributable to portland clinker manufacture, are presented in Table 1. According to the U.S. Geological Survey records, the world consumption of cement in 1990 was 1,044 million tonnes. From the fragmentary information available it is estimated that, globally, the average clinker factor of cement (units of clinker per unit of cement) in 1990 was 0.9, which means that 940 million tonnes of clinker and 104 million tonnes of CCM were used. Assuming the average CO 2 emission rate as 1.0 tonne CO 2 /tonne clinker, in 1990 the direct CO 2 emission from clinker production were 940 million tonnes. In 2005, due to a gradual increase in the use of CCM, it is estimated that 370 million tonnes of CCM were incorporated into 2,270 million tonnes of cement. This gives a clinker factor of Also, in 2005, due to increase in the use of alternate, low-carbon, fuels for burning clinker, the average CO 2 emission rate dropped to 0.9 tonne per tonne of clinker. This means that, in 2005, 1,900 million tonnes of clinker was produced, with 1,700 million tonnes of direct CO 2 release to the environment. In conclusion, the global cement industry has almost doubled its annual rate of direct CO 2 emissions during the last 15 years. Reducing The Co 2 Emissions Comparing the 1990 and 2005 global CO 2 emissions directly attributable to clinker production (Table 1), the magnitude of the problem becomes at once clear. Not only the annual rate of cement consumption in the world has nearly doubled during the last 15 years but also, at the current rate of economic growth in many developing countries, by the end of the next 15 years the cement requirement is expected to go up to about 3,500 million tonnes a year. Assuming that during the same period the use of CCM in- 16 Journal of SEWC April 2011

19 Cements and Concrete Mixtures for Sustainability creases from 15 to 20 % of the total cement, the global clinker production and CO 2 emission in 2020 would amount to 2,800 million tonnes, and 2,520 million tonnes, respectively. To bring down the CO 2 emission from 2,520 to 940 million tonnes (the 1990 level) involves nearly a two-third reduction in clinker requirement, which is unlikely barring a global catastrophe. In the portland clinker manufacturing process, direct release of CO 2 occurs from two sources, namely the decomposition of calcium carbonate (the principal raw material) and the combustion of fossil fuels. The former accounts for about 0.6 kg CO 2 /kg clinker and the latter kg CO 2 /kg clinker (depending on the carbon content of the fossil fuel); the global average being 0.9 kg CO 2 /kg clinker. Alternate sources of energy other than fossil fuels are being sought but, at present, they are too expensive. Also, there are some cements that do not require calcium carbonate as a raw material (e.g., magnesium phosphate cements) but they are neither economical nor technically feasible for large-scale production. Obviously, it will not be possible to achieve any drastic cuts in CO 2 emission as long as technical and economic reasons favor the use of portland clinker as the major component of hydraulic cements. The golden rule or mantra for successful resolution of all sustainability issues is, "Consume less, and think more." Based on this mantra, the author proposes the following three tools, the simultaneous use of which would enable the cement industry to reduce greatly the direct CO 2 emission attributable to clinker production: 1. Reduce the consumption of concrete: Architects and structural designers must develop innovative designs that minimize the consumption of concrete. Service life of repairable structures should be extended as far as possible by the use of proper materials and methods of repair. Low-priority projects should be postponed or even canceled when possible. Foundations, massive columns and beams of concrete, and precast building components that can be assembled or dis-assembled as needed, should be made with highly durable concrete mixtures described in this paper. 2. Reduce the cementing materials in concrete mixtures: Mix design procedures that involve prescriptive codes (e.g., minimum cement content, maximum w/cm, and much higher than needed strength) lead to considerable waste of cement, besides adversely affecting the durability of concrete. Such prescriptive codes have outlived their usefulness and must be replaced with performance-based specifications that promote durability and sustainability. For example, to achieve durability, it is not the w/c but the cement paste content which should be minimized through optimum aggregate grading, use of plasticizing admixtures, and specifying 56 or 91-day strength for the structural components that do not have to meet a minimum 28- day strength requirement. 3. Reduce the clinker factor of cement: Every tonne of clinker saved would reduce the direct CO 2 release from cement kilns by an equivalent amount. Furthermore, as explained below, concrete products made with cements of low clinker factor are expected to be much more durable when compared to ordinary portland cement products. Imagine if it were possible to enhance the durability of most cement-based products by factor 10 or more, without using any expensive technology and materials! Unquestionably, in the long term, this would serve as an excellent strategy for minimizing the wasteful consumption of cement and other concrete ingredients for general construction. Published literature contains numerous reports showing that high-early strength concrete mixtures used in modern, high-speed, construction often suffer from lack of durability because they are usually made with high content of a cementing material and a high clinker factor of cement. The hardened product contains a heterogeneous cement paste, with weak interfacial bonding, and is vulnerable to cracking from excessive thermal shrinkage and drying shrinkage. According to Reinhardt (7), to minimize the shrinkage, volume of the paste (cement plus mixing water) in concrete should not exceed 290 L/m3. Highvolume fly ash concrete mixtures, described in this paper are made with cements of low clinker factor ( ), and less than 290 L/m3 cement paste content. Therefore, they can be used for making relatively crack-free products of excellent durability without any added cost. What are the Options? As shown in Table 1, compared to the base year 1990, global carbon emissions direct from portland clinker production have already doubled in the past 15 years. If no serious measures are put into place quickly by the world's construction industry, i.e. with business-as-usual it is estimated that the rate of direct carbon emissions from cement kilns will almost triple in the next 15 years (Table 2, Option 1). Table 2 also includes data on two other options, an easy option (Option 2) and a challenging but preferable option (Option 3). Note that Option 1 (business-as-usual) data will be used as a reference point for both Options 2 and 3, that are discussed next. According to Option 2, by 2020, if the global concrete construction industry is able to reduce the concrete consumption by 20% (compared to Option 1) and at the same time increase the CCM utilization to 30% of the total cement, these steps will have the effect of reducing the direct CO 2 emissions from cement kilns to 1,760 million tonnes. This is nearly twice as much as the 1990 emissions rate of 940 million tonnes. According to Option 3, in 2020, the total cementing material (2,100 tonnes) would comprise 1050 million tonnes of portland clinker and the same amount of complementary April 2011 Journal of SEWC 17

20 Cements and Concrete Mixtures for Sustainability cementing materials. In Table 3, estimates of different types and amounts of complementary cementing materials that would be available for use in 2020 are given. Note that coal fly ash is expected to make up 760 million tonnes or nearly three-fourths of the total CCM. Would such a large quantity of fly ash be available in 2020? It is difficult to provide a definite answer, but let us examine the assumptions under which this is possible. In the foreseeable future, fossil fuels will continue to remain the primary source of power generation, and due to the low cost of coal, expansion of the coal-fired power industry will continue in major coal-producing countries such as China, India, and the United States. According to one estimate, approximately 1200 million tonnes of fly ash would be available in It would indeed be a formidable job to ensure that nearly two-thirds of the fly ash produced by coal-fired power plants is suitable for use as a complementary cementing material. This goal can be accomplished, provided the key players, i.e., the producers of fly ash, the consumers of cement and concrete, and individuals or organizations responsible for specifications work together to overcome the problems, discussed below. The power sector of the global economy is the largest single source of carbon emissions in the world. It is estimated that about 7 billion tonnes a year of CO 2 is being released today from the combustion of all fossil fuels, and that the coal-fired power plants alone generate 2 billion tonnes of CO 2. Besides carbon emissions, according to Malhotra (5), coal combustion in 2005 generated approximately 900 million tonnes of solid by-products including 600 million tonnes of fly ash. Due to rapidly changing rates of fly ash production and use in the two large economies of the world, China and India, which meet three-quarters of their electrical power requirement from coal-fired furnaces, accurate data on today's global rates of fly ash production and utilization are not available. However, a rough estimate shows that the current rate of fly ash production is approximately 750 million tonnes/year, and that nearly 140 million tonnes/year is being consumed as an ingredient of blended cements and concrete mixtures. The remaining fly ash either ends up in low-value applications, such as road sub-bases and embankments, or is disposed to landfills and ponds. When used as a complementary cementing material, each tonne of fly ash can replace a tonne of portland clinker. Diverting fly ash from the waste stream and using it to reduce direct carbon emissions from the cement industry is like killing two birds with one stone. Therefore, increasing the utilization of most of the available fly ash as a complementary cementing material is, unquestionably, the most powerful tool for reducing the environmental impact of two major sectors of our industrial economy, namely the cement industry and the coal-fired power industry. In spite of proven technical, economic, and ecological benefits from the incorporation of high volumes of fly ash in cements and concrete mixtures, why does the fly ash utilization rate as a complementary cementing material remain so low? Obsolete prescriptive codes, lack of stateof-the-art information to architects and structural designers, and lax quality control in power plants are among some of the reasons. Also, all of the currently produced fly ash is not suitable for use as a complementary cementing material, however cost-effective methods are available to beneficiate the material that does not to meet the minimum fineness and maximum carbon content requirements - the two important parameters by which the fly ash suitability is judged by the cement and concrete industries (5). Sustainable Cements Sustainable, portland-clinker based cements can be made with 0.5 or even lower clinker factor using a high volume of granulated blast furnace slag (gbfs), or coal fly ash (ASTM Class F or C), or a combination of both. Natural or calcined pozzolans, in combination with fly ash and/ or gbfs, may also be used. Compared to portland cement, the high-volume fly ash and slag cements are somewhat slower in setting and hardening, but they are more suitable for producing highly durable concrete products. Unfortunately, worldwide, the conventional concrete construction practice is dominated by prescriptive specifications that do not permit the use of high volume of mineral additives. Cements containing a high-volume of complementary cementing materials can now be manufactured in accordance with ASTM C a new standard specification for hydraulic cements, which is performance-based. However, in North America significant amount of blended portland cements are not produced, because it is customary to add mineral admixtures at the ready-mixed concrete plants. According to American Coal Ash Association, at present about 14 million of the available 70 million tonnes/ year fly ash is being used as a complementary cementing material in concrete mixtures. Reliable estimates are not available from China and India, however, it is reported that significant quantities of blended cements containing % fly ash, are being manufactured in these countries. The European Cement Specification EN 197/1, issued in 2002, contains 26 types of blended portland cements including three cement types that have clinker factors ranging between 0.35 and Type III-A Cement covers slag cements with % gbfs; Type IV-B Cement covers pozzolan cements with % pozzolans including fly ash, natural or calcined pozzolanic minerals, and silica fume; Type V-A Cement covers composite cements containing % gbfs plus % pozzolans. According to Cembureau statistics for 2005, the consumption of ordinary portland cement in the European Union countries has dropped to 30 % of the total cement produced, whereas blended portland cements containing up to 25 % CCM have captured 57 % of the market share, and 18 Journal of SEWC April 2011

21 Cements and Concrete Mixtures for Sustainability blended cements with more than 25 % CCM are approaching 10 % of the total cement consumption. Sustainable Concrete Mixtures For reducing direct carbon emissions attributable to portland clinker production, the emerging technology of highvolume fly ash (HVFA) concrete is an excellent example showing how highly durable and sustainable concrete mixtures, with clinker factor of 0.5 or less, can be produced by using ordinary coal fly ash (ASTM Class F or Class C), which are available in most parts of the world in large amounts. The composition and characteristics of HVFA concrete are discussed in many publications and are briefly described below. Note that concrete mixtures with similar properties can be produced by using a high volume of granulated blast-furnace slag or a combination of fly ash and slag, with or without other mineral admixtures. The cementing material in HVFA concrete is composed of ordinary portland cement together with at least 50 % fly ash by mass of the total cementing material. The mix has a low water content ( kg/m3), and a low content of cementing materials (e.g. 300 kg/m3 for ordinary strength and max. 400 kg/m3 for high-strength). The plasticizing action of the high volume of fly ash imparts excellent workability even at w/cem of the order of 0.4. However, chemical plasticizers are often used, when lower w/cem are required. Occasionally, an air-entraining admixture is also included in the mix when protection against frost action is sought. Compared to portland-cement concrete, the HVFA concrete mixtures designed to achieve the same 28-d strength exhibit superior workability without segregating even at slump values of mm. Typically, the concrete is slow in setting and hardening, i.e. develop slightly lower strength at 3 and 7-d, similar strength at 28-d, and much higher strength at 90-d and 1-year. The pozzolanic reaction leading to complete removal of calcium hydroxide from cement hydration products enables the HVFA concrete to become highly resistant to alkali-aggregate reaction, sulfate and other chemical attacks, and reinforcement corrosion (due to very low electric conductivity). Furthermore, the HVFA concrete mixtures are much less vulnerable to cracking from both the thermal shrinkage (less heat of hydration), and the drying shrinkage (less volume of cement paste). Therefore, in addition to very low clinker factor, the ability of HVFA concrete to enhance the durability by factor 5 to10 makes it a highly suitable material for construction of sustainable structures in the future. The author has been involved with many field applications of HVFA concrete that are described in earlier publications (8-11). Three recently built structures in the U.S., with large reduction in CO 2 -emissions resulting from the use of HVFA concrete, are described below. A Hindu Temple, built with concrete members designed to endure for 1,000 years or more, was constructed in Chicago in 2003 (Fig. 2). The superstructure of the temple is composed of some 40,000 individual segments of intricately carved white marble (Fig. 2). Unreinforced monolith slabs are a part of the foundation, supported by 250 drilled piers, 9 m high and 1 m diameter. All structural elements were made with, cast-in-place, HVFA concrete containing 105 kg/m 3 ASTM Type I portland cement and 195 kg/m 3 Class C fly ash, 2 L/m 3 polycarboxylate superplasticizer, and 100 kg/m 3 water. Note that the total cementing material was 300 kg/m 3, the clinker factor was only 0.33, and the w/cem was also The fresh mix had mm slump and showed excellent pumpability, which made it possible to place and finish 400 m 3 concrete for the main prayer-hall slab (22 by 18 by 1 m), in less than 5 hours. Typical compressive strength values at 3-d, 7-d, 56-d, and 1-y were 10 MPa, 27 MPa, 48 MPa, and 60 MPa, respectively. No structural cracks in any concrete member were reported. Also, the chloride penetration permeability, which is an excellent index of long term durability of concrete, was surprisingly low (< 200 coulombs) in 1-year old core samples. A conventional concrete mix would have required 400 kg/m 3 portland cement to achieve similar high-strength. The use of 3,000 m 3 HVFA concrete mix resulted in 900 tonnes of portland cement saving, which corresponds to about 800 tonnes of CO 2 emissions reduction. The Utah State Capitol Building, Salt Lake City, underwent seismic rehabilitation in 2006 (Fig. 3a). Due to heavily congested reinforcement in the foundations, floor beams, and shear walls, a nearly self-consolidating mix containing 160 kg/m 3 ordinary portland cement, 200 kg/m 3 ASTM Class F fly ash, 138 kg/m 3 water, and 1 L/m 3 superplasticizer was used. The clinker factor of this mix was 0.44, and the w/cem was The specified slump and 28-d compressive strength were 150 mm and 27 MPa, respectively. The field concrete showed an average of 225 mm slump and 34 MPa strength. It is estimated that this 4,500 m3 HVFA concrete job, enabled 900 tonnes of reduction in CO 2 emissions attributable to clinker saving. The CITRIS Building at the University of California at Berkeley contains 10,700 m 3 HVFA concrete - the largest volume ever used for construction of a single building. For foundations and mats, a concrete mix containing 160 kg/ m 3 of ASTM Type II portland cement, 160 kg of Class F fly ash, and 123 kg/m 3 water (0.37 w/cem) was used. For heavily reinforced columns, walls, beams, girders and slabs, a concrete mix containing 200 kg/m 3 ASTM Type II portland cement, 200 kg/m 3 Class F fly ash, and 140 kg/ m 3 water (0.35 w/cem) was used. In both cases the clinker factor is The specified compressive strength was d for all structural members except the foundations and mats which were designed for a specified strength of minimum d. Note that the concrete used for reinforced columns achieved 20 MPa 7-d, and nearly d. It is estimated that the choice of HVFA concrete as a structural material for the CITRIS Building resulted in a reduction of 1950 tonnes of direct CO 2 emissions attributable to the low clinker factor of the cementing material. April 2011 Journal of SEWC 19

22 Cements and Concrete Mixtures for Sustainability Economic and Technical Barriers For utilization of high proportions of complementary cementing materials in general construction, human perception appears to be a far more formidable barrier than actual economic and technical barrier. According to Meryman and Silman (12): Sometimes, there is a perception that a "green" material or practice is more costly, but on further examination, it proves no to be so; often it is just a matter of getting on the other side of the learning curve. We must clarify the difference between life cycle cost and first cost, since many sustainable products have better life cycle performance. We need to define the term 'economic' and include the collateral cost of using non-sustainable practices. The use of sustainable cements and concrete mixtures, described in this paper, would undoubtedly produce structural members of high durability. However, a statistical life-cycle analysis is not possible because there are no reliable laboratory tests for quantitative assessment of long-term durability of field structures. Other major barriers are lack of codes of recommended practice and unwillingness of structural designers and engineers to be among the first to champion the use of new materials. Again, according to Meryman and Silman (12): How can an underused material or method become tried, trusted and ultimately the standard? These materials and methods need advocates. As technical professionals, structural engineers can use specifications to communicate a commitment to and confidence in more sustainable choices. By taking responsibility for those practices, we become their advocates. From my own personal experience, I confirm the observations of Meryman and Silman. I have come to the conclusion that it is the hand that writes the specifications which holds the power of leading the concrete construction industry to an era of sustainability. Codes of recommended practice advocated by organizations, such as American Concrete Institute and U.S. Green Building Council, can play an important part in accelerating the sustainability of the concrete industry. For instance, the USGBC pointrating system for new construction has already become a powerful driving force for sustainable building designs. The rating system awards sufficient points for buildings that would consume less energy in their use. A similar emphasis is needed in favor of sustainable materials that produce less CO 2 during their manufacture. By suitably amending the rating system so that some points based on CO 2 emissions reduction are directly assigned for the use of sustainable materials in new construction, the USGBC can help sustainability of the cement and concrete industries. Concluding Remarks The high carbon dioxide emission rate of today's industrialized society has triggered climate change that is potentially devastating to life on the planet earth. To meet the global concrete demand, which was 17 billion tonnes in 2005, two billion tonnes of CO 2 were directly released to the atmosphere from the manufacturing process of portland-cement clinker, which is the major component of modern hydraulic cements. With business-as-usual, the direct CO 2 emissions from portland clinker production, in the year 2020, would triple the 1990 level unless immediate steps are taken to bring down the emissions by making significant reductions in the: (a) global concrete consumption, (b) volume of cement paste in concrete, and (c) proportion of portland clinker in cement. Examples of recently built structures prove that by using high volume of coal fly ash and other industrial wastes as complementary cementing materials with portland clinker, we can produce low cost, highly durable, and sustainable cements and concrete mixtures that would significantly reduce both the carbon footprints of the cement industry and the environmental impact of the coal-fired power generation industry. It seems that the game of unrestricted growth, in a finite planet, by reckless use of energy and materials, is over. Most sectors of the global economy have already initiated action plans to bring down their share of carbon emission to the 1990 level or less, by the year The construction industry is already pursuing the goal of designing and constructing sustainable buildings that consume less energy and resources to maintain. Now, all segments of the construction industry - owners, designers, contractors, and cement and concrete manufacturers - will have to join the new game of building sustainable structures using only sustainable materials. We have the tools to win this game. What is needed now is the will and the individual initiative. To paraphrase John F. Kennedy, "Ask not what others can do. Ask what you can do to promote the use of sustainable construction materials." Acknowledgement The author would like to thank Mason Walters of Forell Elsesser Engineers, San Francisco, for the photographs in Figs. 3 and 4. Author Affiliation Mehta Prof. P. Kumar <pkmehta@berkeley.edu> University of California, Berkeley, U.S.A. References 1. P.K. Mehta, and P.J.M. Monteiro, "Concrete: Microstructure, Properties, and Materials", McGraw-Hill, New York, ACI Board Advisory Committee on Sustainable Development, "White Paper on Sustainable Development", Concrete International, American Concrete Institute, Vol. 27 No. 2, 2005, pp Journal of SEWC April 2011

23 Cements and Concrete Mixtures for Sustainability 3. The Concrete Center of U.K., "Sustainable Concrete", , 18 pages 4. World Business Council for Sustainable Development, "The Cement Sustainability Initiative", Geneva, Switzerland, V.M. Malhotra, "Reducing CO2 Emissions", Concrete International, American Concrete Institute, Vol. 28 No. 9, 2006, pp P.K. Mehta, "Greening of the Concrete Industry for Sustainable Development", ibid., Vol. 24 No. 7, 2002, pp H.W. Reinhardt, "New German Guideline for Design of Concrete Structures for Containment of Hazardous Materials", Otto Graf Journal, FMPA, Univ. of Stuttgard, Germany, Vol. 17, 2006, pp P.K. Mehta and W.S. Langley, "Monolith Foundation Built to Last a 1,000 Years", Concrete International, American Concrete Institute, Vol. 22 No. 7, July 2000, pp D. Manmohan and P.K. Mehta, "Heavily Reinforced Shear Walls and Reinforced Foundations Built with Green Concrete", ibid., Vol. 24 No. 8, 2002, pp P.K. Mehta and D. Manmohan, "Sustainable, High-Performance Concrete Structures", ibid., Vol. 28 No. 7, 2006, pp V.M. Malhotra and P.K. Mehta, "High-Performance, High-Volume Fly Ash Concrete", Supplementary Cementing Materials for Sustainable Development, Ottawa, Canada, H. Meryman and R. Silman, "Sustainable Engineering - Using Specifications to Make it Happen", Structural Engineering International, Vol. 14 No. 3, Aug. 2004, pp *Estimated amounts of CCM used in 1990 and 2005 are 10 % and 15 % of total cementing material, respectively. Table 1 Global Direct CO2 Emission from Cement Kilns (Million Tonnes) Fig. 1 Historical and Future Atmospheric CO2, Based on IPCC Reports (1) Business-as-usual scenario Table 2 Estimates of Global Cement Consumption in 2020, and Direct CO2 Emissions from Cement Kilns (Million Tonnes) Table 3 Estimated Consumption of Complementary Cementing Materials (Million Tonnes) Fig. 2 The BAPS Hindu Temple, Chicago, 2004 High-Volume Fly Ash was Used for Unreinforced Monolith Foundations and Drilled Piers April 2011 Journal of SEWC 21

24 Cements and Concrete Mixtures for Sustainability Fig. 3a Utah State Capitol Building After Seismic Rehabilitation, 2006 High-Volume Fly Ash was Used for Reinforced Foundations, Beams, and Shear Walls Fig. 4a CITRIS Bldg., Univ. of California, 2007 Mat Foundation Under Construction Fig. 3b Utah State Capitol Building After Seismic Rehabilitation, 2006 Excellent Pumpability and Workability of Nearly Selfconsolidating Concrete Mixture Fig. 4b CITRIS Bldg., Univ. of California, 2007 Heavily Reinforced Columns Under Construction 22 Journal of SEWC April 2011

25 When Structures Move Kawaguchi, Prof. Mamoru Abstract In the present paper moving aspects of structures are taken up. In our daily structural design the structures are assumed to be immovable, and most of structural calculations are carried out on the basis of static principles. Although we know that a structure always produces such a movement due to loading that is referred to as deformation or displacement, its magnitude is normally too small to be significant in comparison with the dimensions of the structure, and its effect on the structural behaviors is neglected, the whole phenomenon being treated as static. There are cases, however, where large movements are actually experienced by our structures due to different reasons. Many of them are due to excessive loading and unexpected instability, often leading to collapse of the structures. Some other cases are related to vibration where resonance of structures with external agencies such as earthquakes and wind is a key question. Self-excited oscillation sometimes produces catastrophic and very spectacular motion of structures. Controlled motions can be obtained by adopting isolators to cope with the effects of earthquakes. Dampers which are often incorporated in seismic isolation systems are normally rather still, but motion of tuned mass dampers is sometimes very significant. Structures can be designed to be assembled on the ground and then hoisted to the position. In erection of such structures a big movement is observed as in Pantadome System. Finally those structures which are originally intended to move are described with examples of rocking stones and a flying carp. Keywords: moving structures, collapse, excessive deformation, controlled motion, earthquake isolation, self-excited oscillation, Pantadome system, tuned-mass dampers, pendulum system Undesirable Movements There are unfavorable movements which poor structures have to experience under some undesirable conditions. They are movements due to excessive snow loads, earthquakes, wind, structural deterioration and so on, and those movements have different characteristics due to the natures of the causes. Collapsing Movements due to Snow Loads Structures standing on the principle of arches and domes are sometimes in danger of yielding collapsing movements due to unstable deformation of the compressive members. One of such examples is the dome for a trade center in Bucharest which collapsed in January 1963 (Fig.ures.1 and 2). The dome had a spherical shape to cover a plan of 93.5m in diameter. The dome experienced a huge "snapthrough" deformation, or a deformation from convex to concave geometries under the snow load of 2,000 kn which was less than 30% of the design snow load. Another example of this kind is a collapse of the hanging roof of "Palasport" in Milan which occurred in January 1985 (Figures 3 and 4). It had a circular plan of 128m in diameter. The presumed snow load on the roof at the time of collapse was 1.4 kn/m2 while the standard snow load was 0.9kN/m2. The roof of this velodrome was a saddleshaped hanging roof that should have more sufficient potential strength, but the collapse was caused by the buckling of the ring beam the section of which had been a box section of thin steel plates. In the above examples the structures must have experienced very large movements during the collapse, but no such movements were visually recorded. There are many other examples of structural collapse due to snow loads, but observation records of the collapsing movements are scarce. In general the visual records of collapsing movements of large-span roofs due to snow are difficult to make, since firstly it is not easy to anticipate the time of collapse which often occurs at a lower loading level than in design, and secondly weather and shooting condition are bad because of snow falling and snow drift. Destructive Movements due to Earthquakes Earthquakes make structures produce significant movements which are often destructive. Different from the effects of snow, earthquakes are not loading on the structures but vibrational motion of the grounds on which the structures stand. Therefore the motion induced in the structures by earthquakes is closely related to the vibrational characteristics of the structures, and when the natural periods of the structure are close to those of the prevailing ground motion, the motion of the structures can be destructive. On the other hand this type of motion can often be controlled by means of vibration technology. The ideal case of such a control is seismic isolation, as will be described later.there have been so many destructive motions of structures due to earthquakes, and some of them have been recorded numerically in the form of acceleration data, but visual records of such motions are again very scarce because of the facts that prediction of destructive earthquakes is again very difficult and that pho- April 2011 Journal of SEWC 23

26 When Structures Move tographers are also in danger of their lives during the severe earthquakes. Uncontrolled Movements due to Wind In design of comparatively rigid structures we treat the effects of wind as static loads. When a structure is soft, however, we have to take into account dynamic effect of wind, and motions of the structure due to this effect. Dynamic effect of wind due to disturbance in the wind flow itself is sometimes referred to as buffeting or gusty effect, and resonance of the structure with this effect is often discussed. Another and more important effect of wind is vortex-induced vibration, and still more important is self-excited oscillation or flutter starting from the vortex-induced vibration. In such a motion the structure takes in energy from constant air flow around it to grow the motion until it becomes catastrophic. The collapse of Tacoma Narrows Bridge is explained as the result of such phenomena (Figures. 5 and 6). In general the magnitude as well as the mode of self-excited motion is very big and exceeds our imagination, often being even spectacular. Such motions are comparatively easy to record visually, since the time of strong wind can be predicted, the motion of this kind lasts for relatively long time and the photographer is not always in a dangerous situation. Controlled Movements Seismic Isolation Seismic isolation is a technology to control the response of structures due to earthquake ground motion. The isolation technology is normally applied in combination with energy-absorbing damping systems. The most popular seismic isolation system is the laminated rubber shoes that support the structures. However, there are other isolation systems effective to control seismic motions in more rational manners than laminated rubber system, which will be described in this section. Pendulum Isolators A pendulum system is one of the basic methods of seismic isolation, having the same fundamental principle in common with seismographs. As shown in Figure 7, pendulums used in engineering include (a) simple pendulum, (b) physical pendulum, and (c) translational pendulum. It is well known that the natural period T of a simple pendulum is given as follows with the length of the hanger L, and the gravitational acceleration g. (1) One advantage of a pendulum seismic isolator is that the length of the hanger L is the only parameter governing its natural period, and the mass of the object to be isolated exerts no effect on it at all. Thus, desired periods can be obtained by simply changing the hanger length. This is the greatest advantage of pendulum seismic isolators compared to laminated rubber seismic isolators in which the natural period is determined by the mass and rigidity of isolation structure. The natural period is slightly elongated if the amplitude is made larger. The elongation, however, is minute. Thus, the above equation (1) can be considered valid for all practical cases. This is another advantage of the pendulum seismic isolator compared to the laminated rubber system, of which the deformability is limited. Wide selection of materials is available for the hanger. For example, technology for fireproofing has already reached a mature state if steel is to be used. As discussed above, seismic isolators using pendulum principle possess considerable merit. Considering that seismic isolators must also function as a part of structural support, simple pendulum shown in (a) of Figure7 is obviously difficult to use. Natural period of physical pendulum shown in (b), on the other hand, fluctuates along with the location of center of mass as well as the moment of inertia of the system. Thus, translational pendulum shown in (c), whose natural period is only affected by the hanger length as in the simple pendulum, would be appropriate for use as a seismic isolation device. One possible application of the translational pendulum seismic isolator is for individual floors. A floor suspended from a girder of a building frame as shown in Figure 8 was adopted for the exhibition rooms of an actual museum for pottery and porcelain wares (in Gifu Prefecture, Japan, completed in 2001). The area of the suspended floor is about 1,000 m2, and its mass is about 1,000 tons. Hinges having universal joints are used for the upper and lower ends of the hanger. If the hanger is made to be 4.5 m long, Equation (1) yields a natural period of more than 4 seconds, which is considered sufficiently long for seismic isolation. A series of seismic isolation tests showed that the system was effective to minimize the seismic effects on the floor. Rocking Pendulum Isolators A paddle isolator is based on a rocking pendulum principle, the concept of which has a long history. The first example of isolated foundations in Japan was designed by Ryuichi Oka in 1932, as a column having a spherical end at the bottom and connected to the superstructure via a spherical hinge at the top. Due to rocking motion of the column, the superstructure moves in a trochoidal curve. This concept, however, was impractical as production of spherical elements required considerable skill and manhours, and column design sometimes interfered with overall architectural planning. In the present design rocking motion of the sphere was resolved into the orthogonal com- 24 Journal of SEWC April 2011

27 When Structures Move ponents along the X- and Y-axes on a plane coordinate. Then, a mechanism was designed which assigns the components of the rocking motion in the directions of X- and Y- axes to the upper and lower ends of a column. Paddle isolator was named after a kayak paddle, as the form of the present column resembles it. As shown in Figure 9, a paddle isolator consists of a column provided with "blades" on the top and bottom ends whose curvatures are designed to be orthogonal to each other. This column allows resolution of any horizontal motion in an arbitrary direction into two directions, and the isolated part of the building is free to move in the horizontal direction. Referring to Figure 10, the natural period of a rocking pendulum is determined by the length of the isolation column L and the radius of curvature of a blade R, and can be obtained by Equation (2). One of the features of the rocking pendulum isolator is that its natural period is not governed by the mass of structure to be supported or any mechanical properties of the materials used in the isolator, similar to the translational pendulum isolator. On the other hand, the paddle isolation column may be designed with any length, unlike the translational pendulum isolator. As such, isolating layer may either be placed just underneath the foundation, or the entire ground floor may be designed as an isolating layer. This enables design of seismic isolators having longer periods, which were conventionally deemed impossible. In order to confirm the effect of the paddle isolators, acrylic specimens (shown in Figure 12, the floor panel being 40 cm by 40 cm) were manufactured and tested. The test results indicated fairly constant natural periods for paddle isolators, regardless of the amplitude of the given motion or the mass of superstructure. Furthermore, the torsion movement was hardly observed even when the mass supported by the isolation layer was largely shifted off the center. Figure 11 shows the rates of the observed acceleration responses when the seismic motion based on the records of the actual earthquakes were applied to the vibration table. As shown, response in the upper part of the isolating layer was reduced sufficiently. It was also confirmed that the effect is not influenced by the direction of input seismic motion. Damping Systems Damping systems are often used in combination with seismic isolators, but they are of course used by themselves as well for the purpose of energy absorption. Most commonly used in vibration control are viscous, frictional, hysteretic and tuned-mass dampers. The first three of the above dampers control the vibrational motion of the structures by dissipating the energy in the form of heat, while the tuned-mass dampers transform the energy of their (2) mother systems into the motion of themselves. So the effect of tuned-mass dampers can be visually confirmed by means of scaled model tests, where the transfer of motion from the structure to the dampers is clearly observed. Human-Induced Vibration Soft footbridges often produce significant vibrational motion that is induced by the movement of pedestrians crossing the bridges. It is interesting to note that when the movement of the bridge, especially the transverse horizontal component of which is big enough to be felt by the pedestrians, they are apt to try to secure themselves by tuning their steps to the period of the motion of the bridge, resulting in amplification of the bridge motion. Vibration of Millennium Bridge in London was a typical example, which was solved by incorporating a passive damping system in the bridge. Designed Movements Pantadome System Structures are sometimes designed to move during construction for safe, efficient and economical erection. A patented structural system called 'Pantadome System' was developed by the author with such an idea for a rational construction of spatial structures, and it was successfully applied to the structure of World Memorial Hall completed in Kobe in Pantadome System has since been applied to the Sant Jordi Sports Palace in Barcelona, the National Indoor Stadium of Singapore and some important structures of wide spans realized in Japan. The principle of Pantadome System is to make a dome or a domical structure geometrically unstable for a period in construction so that it is 'foldable' during its erection. This can be done by temporarily taking out the members which lie on a hoop circle. Then the dome is given a 'kinematic mechanism', that is, a controlled movement, like a 3-D version of a parallel crank or a 'pantagraph' which is popularly applied to drawing instruments or a power collector of an electric car (hence the name, 'Pantadome'). By 'folding' the dome in this way, the constituent members of the dome can be assembled on a lower level. The assembly work is thus done safely, quickly and economically, since it can be carried out near the ground level. Since the movement of a Pantadome during erection is a 'controlled one' with only one freedom of movement in the vertical direction, guying cables or bracing members which are indispensable in conventional structures to assure their lateral stability against wind or seismic forces are not necessary in erection of a Pantadome structure. The movement and deformation of the whole shape of the Pantadome during erection are three dimensional and may look spectacular and rather complicated, but they are all kinematically determinate and easily controlled. Three kinds of hinges are incorporated in the Pantadome System which rotate during the erection. Their rotations are all uni-axial ones, and of the most simple kind. Therefore, all these hinges are fabricated in the same way as normal hinges for usual steel frames. April 2011 Journal of SEWC 25

28 When Structures Move Rocking Stones Rocking stones are stones originally created by Nature that are movable by human power, or at least looking to be movable. Here is an example of artificial rocking stone of 36tons that can be moved by a little child. Repetitive push of the child in tune with the period of the rocking stone brings the stone into a motion of slow but significant amplitude. Flying Carp The "KOINOBORI" is a Japanese traditional carp made of fabric which people fly in the breeze in early days of May every year to celebrate the growth of children. The normal size of a KOINOBORI is 2 to 5 meters in length. KAZO is a town in the suburb of Tokyo that has been famous for its production of KOINOBORI since more than one hundred years ago. We can still see excellent craftsmen who handpaint beautiful KOINOBORIS in factories of KAZO City. In 1988 volunteers of KAZO City, who were members of Junior Chamber International, got an idea of fabricating and flying a gigantic KOINOBORI of 100m to advertise their city to the world. However, they did not know how to produce such a huge carp properly. They did not even know if such a monstrous feature might "fly" in the air at all. The author had an opportunity to assist them by establishing the technical basis for the possibility of flying this gigantic fabric fish in the air. He showed by theory as well as experiments that a huge KOINOBORI could be designed to fly in the breeze of the same wind speed at which normal carps fly. By this design theory and structural details a huge KOINOBORI was designed, it was fabricated by the voluntary members of KAZO City, and finally it succeeded to fly elegantly in the sky. Since then flying of the Jumbo KOINOBORI became an annual event of KAZO City, being celebrated in the beginning of May every year. Conclusive Remarks Moving aspects of structures have been described. Although structures are normally regarded stationary, there are many cases where structures move significantly. The most undesirable motion of a structure is the one due to collapsing effects of external agencies such as snow, earthquake, terrorism and deterioration of materials. Structures sometimes produce uncontrolled motion due to wind effects The most dramatic motion is observed when structures show self-excited vibration often started by Kármán Vorticies as in the catastrophic example of Tacoma Narrows Bridge. Since the task of structural engineers is to create strong and safe structures, we should be aware of those undesirable movements and should always try to find the means to cope with those phenomena. Sometimes we can make the motion of structures controlled one. This can be achieved by means of seismic isolation to cope with earthquakes, and general dampers to cope with other vibrational effects. We can also design the structures so that they experience significant movements in the process of construction for the sake of safety, efficiency and economy of construction, as in the example of Pantadome System. Movement is sometimes the intended function to be performed by a structure. One of such cases is an artificial rocking stone of 36tons which the author designed to be moved by a little child. Another example is a huge fabric carp of 100m in length that can fly in the breeze, fabricated by volunteers of a small town in the suburb of Tokyo under the technical guidance of the author. Although it is impossible for those volunteers of the small town to make a jumbo jet aircraft, they could fabricate a carp which is much bigger than Boeing 747, and fly it in the breeze of Kasiserslautern in Germany as well as of their hometown where they have been playing with it for eighteen years. Author Affiliation Mamoru Kawaguchi, KAWAGUCHI & ENGINEERS, mk@kawa-struc.com References 1. A. Beles et al (1966), "Some Observation on the Failure of a Dome of Great Span", st International Conference of Spade Structures, University of Surrey 2. S. Montague (1985), "Milan Roof is Total Write Off", New Civil Engineer, April E.B.Farquharson, et al., "Aerodynamic Stability of Suspension Bridges with Special Reference to the Tacoma Narrows Bridge" Bul. Of Univ. Washington Eng. Exp. Station, No.116, M. Kawaguchi, I. Tatemichi (2000), "Seismic Isolation Systems and Their Application in Space Structures", IASS Symposium on Bridging Large Spans; From Antiquity to the Present, Istanbul 5. M. Kawaguchi, I. Tatemichi (2000), "Characteristics of A Space Structure Seismically Isolated by Rocking Pendulums", IASS- IACM 2000, Fourth International Colloquium on Computation of Shell & Spatial Structures, June 2000, Chania, Greece 6. M. Kawaguchi (2003), "Physical Models as Powerful Weapons in Structural Design", IASS Symposium on Shell and Spatial Structures from Models to Realization, Montpellier, September, 2003 Fig. 1 Bucharest Dome before Collapse Fig. 2 Dome after Collapse or Big Motion 26 Journal of SEWC April 2011

29 When Structures Move Fig. 3 Palasport before Collapse Fig.4 After Big Motion due to Ring Buckling Fig. 5 Oscillating Tacoma Narrows Bridge Fig. 6 Kármán Vorticies in Wind T. Test Fig.7 Different Pendulums Fig.11 The Effect of Isolation Fig.12 Paddle Isolator Fig.8 Concept of Isolated Floors Fig.9 Paddle Isolator Fig. 10 Dimensions of Isolator Fig.13 The Both Paddle Isolators Have the Same Period of Vibration April 2011 Journal of SEWC 27

30 When Structures Move Fig. 14. Non-vertical Hoisting of A Dome Fig. 16 Artificial Rocking Stone in Tsukuba Fig. 15 Rocking Stone in Colorado, USA Fig. 17 Comparative Dimension of 100m long Jumbo KOINOBORI 28 Journal of SEWC April 2011

31 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice Abstract Except for huge unreinforced structures, Weibull's statistical size effect is weak in concrete structures. The size effect source is principally energetic, caused by stress redistribution due to a large fracture process zone size or large cracks formed before maximum load. It is shown that an unbiased statistical analysis of the existing database for shear of R.C. beams without stirrups supports the energetic size effect theory, and that the size effect, albeit milder, afflicts beams with stirrups, too. Known though has this type of size effect been for 24 years, it has been mostly ignored in design codes as well as practice. What are the consequences? -overdesign of many small structures but, more seriously, unacceptable risk for large ones. A tolerable failure probability of engineering structures is 10-6 per lifetime, and collapse statistics indicate that this has indeed been true for small structures. Probabilistic analysis calibrated by a large statistical database confirms this level of failure probability for shear of reinforced concrete beams without stirrups < 0.2 m deep. However, if the size effect is ignored (as in ACI code), the failure probability is shown to increase drastically-to 10-3 for beams 1 m deep (and nearly as much for the unrealistic underestimating formulae of CEB, fib and JSCE). This finding roughly matches several statistics showing that very large structures have been collapsing with a frequency roughly 103-times greater than small ones. This is unacceptable. Now that no longer just a handful of theoreticians, but entire scientific societies and concrete fracture committees (IA-FraMCoS, ASCE-EMD, ACI Comm. 446), are convinced of the inevitability of energetic size effect in brittle failures of concrete structures, the engineering societies ignoring or severely underestimating the size effect in their design codes, and perhaps even design firms, are exposed to serious legal risk when another collapse occurs. Keywords: size effect, structural safety, reinforced concrete, shear failure, fracture, statistical analysis of test data, design codes, legal risk. Introduction Concrete structures much larger than the specimens tested in laboratories are being built in ever increasing numbers. For example, the box girder of the record-span Koror-Babeldaob Bridge in Palau, which collapsed in a brittle shear-compression mode, was 14.2 m deep. The outriggers of the Trump Tower under construction in Chicago are 6 m deep. However the experimental databases collected to establish design code specifications consist mostly of small-size laboratory tests. The mean beam depth in the ACI-445 database (1) on which the shear design in the current ACI standard 318 still rests is only 0.34 m, and in the latest ACI-445 database (2) it is m. In the latter, 86% of the 398 data points pertain to beam depths < 0.5 m and 99% to depths < 1.1 m, and only 1% to depths from 1.2 to 2 m. Since the code-making committees prefer to rely on experiments only, it is thus no surprise that the size effect is not correctly represented. Concrete is an archetypical quasibrittle material whose fracture propagation is characterized by a rather large fracture process zone (FPZ), typically 0.5 m long. This causes that small structures (cross section? FPZ length) fail in a quasi-ductile manner (i.e., with a plastic yield plateau) and exhibit almost no size effect, while very large structures (>> FPZ length) failing in concrete rather than steel behave in an almost perfectly brittle manner, with a steep load drop right after the peak load, and exhibit the strongest possible size effect (3-6); Fig. 1. The size effect has been most studied for shear of longitudinally reinforced beams without stirrups (7, 8), but it also occurs in shear of beams with stirrups, in torsion, punching of slabs, failure of columns, arches and prestressed girders due to concrete compression, failure of anchors and splices, and bar pullout (8, 9). Accumulating experimental evidence (10-20) shows that the size effect causes beams about 2 m deep to fail at loads much lower than that calculated from the design code using the understrength factor ϕ = 0.75 and the required concrete strength (about 30% less than the mean strength). For instance, the largest beam in Toronto tests (3, 4) had a shear strength almost 50% April 2011 Journal of SEWC 29

32 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice lower than what is calculated from ACI design code (21); see Fig. 1(a). This strength reduction due to size effect now causes serious concern about the safety of current design codes and engineering practice. To make the risk of structural failure much smaller than various inevitable risks that people face, the maximum tolerable failure probability is about 1 in a million (22). This value roughly agrees with the frequency of failures experienced for small beams. But for large ones, it has been about 1 in a thousand (23, 24) and could become 1 in a hundred or higher as ever larger beams are being built. Whether or not such unacceptable risk will have to be tolerated depends largely on taking the size effect properly into account. This is an issue of paramount significance for concrete engineering. Risk of Failure in Shear Design The ACI Building Code (21) currently specifies the contribution of concrete to the cross-section shear strength of reinforced concrete members by the formula (which is valid only in psi, lb. and inches). Here f'c is the required compression strength of concrete, d is the beam depth measured from the top face to the longitudinal reinforcement centroid, and bw is the web width. The code formula gives a size-independent average concrete shear strength, v c = V c / b w d (identical to the 'nominal strength' in mechanics terminology). However, ignoring the size effect in Eq. (1) would lead to statistically dangerous designs with insufficient safety margins for large shear-critical concrete beams. We evidence it next. Recently, a database of 398 tests, serving as the experimental foundation for the improvement of current shear design formula, was compiled by ACI subcommittee 445F (2) and named ESDB. Representing an extension of the 1984 Northwestern University database of 296 data (7), it collects 398 shear strength data for longitudinally reinforced three-point-bend concrete beams with no stirrups (the Japanese tests reaching the record depth of d = 3 m were excluded by ACI 445 from this database because the load was uniform rather than concentrated at midspan). In Fig. 2(a), the data in the size range of d from 0.1 to 0.3 m, centered at 0.2 m, are isolated from the database. Within this narrow range, no size effect trend is discernible, and so the data may be treated as a statistical population. The mean and coefficient of variation of the data on within this range are (1) and ω=25% (this relatively high value of ω is the consequence of unsystematic, haphazard, variability of many influencing parameters in the database). The values of and ω suffice for determining the probability density distribution function (pdf) of y in this range. The pdf is assumed to be log-normal. In that regard, note that if the randomness of failure loads of small plastically behaving specimens was caused only by the inevitable inherent randomness of material strength, the only correct choice of pdf would be the normal (or Gaussian) distribution (25). In the case of general design formula, however, the major contribution to the scatter seen in Fig. 2(a) stems from the variation of influencing parameters such as the shear span ratio a/d, steel ratio ρw and steel arrangement, aggregate size da, concrete mix, age of concrete, history of humidity and temperature, etc. The variability stemming from these parameters apparently follows more closely the log-normal than normal distribution. The same pdf is compared in Fig. 2(a) to the series of individual tests of beams of various sizes made at the University of Toronto (3, 4), which have been invoked by some engineers to claim that the size effect may be ignored for d up to 1 m (for this depth, the shear strength of the Toronto beam lies just above where = 0.75 = understrength factor). Now it should be noted that, for the type of concrete, steel ratio, shear span ratio, etc., used in the Toronto tests, their shear strength value lies (in the logarithmic scale) at certain distance a below the mean of the pdf. Since the width of the scatter band in Fig. 2(a) in the logarithmic scale does not vary appreciably with the beam size, the same pdf and the same distance a between the pdf mean and the Toronto data must be expected for every beam size d, including the sizes of d = 1 m and 1.89 m, for each of which there is only one data point. In other words, if the Toronto test for d = 1 m were repeated for many different types of concrete, steel ratios and shear span ratios, humidity and temperature conditions, etc., one would doubtless obtain a pdf shifted downwards, as shown in Fig. 2(a). Now, according to the log-normal pdf shown, the proportion of unsafe 1 m deep beams would be about 40%, while for small beams, it is only 1%. 40% is certainly intolerable. A design code known to have such a dangerous property is unacceptable. A design code ignoring the size effect for beams of d < 1 m will cause the failure probability P f of 1 m deep beams to be about 103-times larger than that of small beams 0.2 m deep. To demonstrate it, one needs to consider also the pdf of the extreme loads expected to be applied on the structure, which is denoted as f(y). Based on the load factor of 1.6 and the understrength factor of ϕ = 0.75, the mean of the pdf of the extreme loads will be positioned as shown in Fig. 2(b). Assuming the individual loads to have the lognormal distribution, their pdf is as shown in Fig. 2(b). Based on the coefficient of variation of extreme loads, here assumed as ωl = 10%, the failure probability may now be calculated from the well-known reliability integral (26-28) (2) 30 Journal of SEWC April 2011

33 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice where R(y) is the cumulative probability density distribution (cdf) of structural resistance. Upon evaluating this integral, one finds the following failure probabilities: for beams 0.2 m deep: deep: for beams 1.0 m The failure probability of 1 in a million corresponds to what the risk analysis experts generally consider as acceptable (22-24). But 1 in a thousand is unacceptable. So, if the size effect in beam shear were ignored for beams without stirrups up to 1 m deep, the probability of failure for 1 m depth would be about 103-times greater than for 0.2 m depth. This would be unacceptable. If there should be any difference, it should be in the opposite sense because, for large beams, the consequences of collapse are generally more serious than for small ones. Statistical Analysis Overcoming Bias in the Database Sound arguments for a realistic design formula capturing the size effect on shear strength of beams must be based on fracture mechanics, verified by properly designed experiments, and statistically calibrated by a broad database. For many engineers, though, a purely statistical evidence, with no use of mathematics and mechanics, is most convincing. Such evidence can be, and has been, readily provided for many design problems where experiments are easy to perform through the entire range of all parameters. But the problem of size effect is different. In the case of size effect, it is financially prohibitive to conduct experiments through the entire range of beam depths of practical interest, which spans from 0.05 m to perhaps 14 m (the latter being the depth of the record-setting box girder in Palau, whose compression-shear collapse must be partly attributed to size effect). Obtaining statistics and covering by experiments the full range of influencing parameters other than the size (or beam depth) has been easy for small beams, but is almost impossible for very large ones. Thus it is not surprising that the existing ACI database has major gaps and a strong subjective statistical bias caused by crowding of the test data in the smallsize range, scant data in the large size range, and no data at all for the largest sizes of practical interest (depths >2 m). Consequently, simple bivariate statistical regression of all the points of the ACI-445F database yields a misleading trend (29). Eliminating the bias is important for a realistic update of the code provisions currently under consideration for the design codes of many countries. The size effect is defined as the size dependence of the nominal strength of structure when geometrical similarity is maintained and all the parameters other than the size are kept constant. In the case of beam shear, the size may measured by the beam depth d, the nominal strength of structure may be taken as the average concrete shear strength in the cross section, v c, and the parameters that must be kept constant comprise all the concrete properties (including the maximum aggregate size d a ), the longitudinal reinforcement ratio ρw, and the shear span ratio a/d (a = distance of the load from the support). If the entire database on size effect in beam shear were to be obtained in one testing program in one laboratory, a sound statistical design of size effect experiments would dictate choosing the same number of tests in equally relevant size intervals and maintaining within all the size intervals the same means and distributions of parameters ρw, a/d, d a,, over their entire practical range. This condition is far from satisfied by the existing database. But there is no other choice. So the question is how to minimize the statistical bias in regard to the size effect. From the size effect viewpoint, this database has a bias of two kinds: o Kind 1. Crowding of the data in the small size range - 86% of the 398 data points pertain to three-point-loaded beams of depths less than 0.5 m, and 99% to depths less than 1.1 m, while only 1% of data pertain to depths from 1.2 to 2 m. o Kind 2. Strongly dissimilar means and distributions, among different size intervals, of the subsidiary influencing parameters, particularly the steel ratio ρw, shear span ratio a/d, and the maximum aggregate size d a. To reach any meaningful statistical conclusion on the size effect, both kinds of bias must be filtered out. Statistical Regression of Size Effect We want to isolate the trend of size effect from a database governed by multiple variables. The standard way to do that is to carry out multivariate least-square nonlinear regression in which all the parameters are optimized simultaneously. This is the approach which was pursued in previous work (29, 30). There is another way, though. It does not lead to multivariate regression, yet makes the statistical trend conspicuous without any mathematics. To this end, an unbiased (i.e., objective) procedure of data filtering is required. Let us subdivide the range of beam depths d of the existing test data into 5 size intervals (vertical strips in Fig. 3(ac)). They range from to 0.15 m, from 0.15 to 0.3 m, 0.3 to 0.6 m, from 0.6 to 1.2 m, and from 1.2 to 2.4 m. In the ACI database, these intervals contain 26, 251, 80, 38, and 3 data points, respectively; see Fig. 3(a-c). Note that the borders between the size intervals are chosen to form a geometric (rather than arithmetic) progression because what matters for size effect is the ratio of sizes, not their difference (to wit, from d = 0.1 to m, the size effect is strong, from 10 to m negligible). The chosen intervals are constant in the scale of log d, and this is also needed for another reason - in the plot of y = versus d (Fig. 3), the database is heteroscedastic (i.e., April 2011 Journal of SEWC 31

34 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice has a variance density decreasing with size), but transformation to the plot of log versus log d renders the database almost homoscedastic (i.e., of uniform variance density), which is necessary for meaningful regression analysis (31). Fig. 3(a-c) shows the restricted (filtered) data points by bigger circles, and those filtered out by the tiny circles. The problem with the distribution of subsidiary influencing parameters in the full database is graphically documented by Fig. 4(a,b), in which the diamonds show their means in the individual size intervals, and the error bars show the span from the minimum to the maximum retained value (Fig. 4(c-f) shows the same plots achieved by filtering the database). In Fig. 4(a), the mean of ρ w is in the second interval nearly 7-times larger than in the last interval (and almost 2-times larger than in the fourth interval). In Fig. 4(b), the mean of a/d is in the third interval 30% larger than in the last interval (and 10% larger than in the fourth interval). Obviously, such differences among size intervals must completely distort size effect statistics. To filter out the effect of influencing parameters other than d, each interval of d must include only the data within a certain restricted range of ρ w -values such that the average would be almost the same for each interval of d. Similarly, the range of a/d and da in each interval must be restricted so that the average and would also be about the same for each interval of d. The filtering of data must be done in an objective manner (i.e., with no human preference). To this end, a computer optimization algorithm has been formulated. It progressively deletes from each interval, one by one, the data points in each size interval that lie at the top and bottom margins of the ranges of ρ w, a/d and d a, until uniformity of each subsidiary influencing parameter throughout all the intervals is optimally approached. 40% Because, as generally agreed, the effect of the required concrete strength f' c is adequately captured by assuming the shear strength of cross section, v c, to be proportional to, we do not need to restrict the range of f' c and may obtain the ordinate of data centroid in each interval by averaging, within that interval, not the vc-values but the values of that fall into the aforementioned restricted ranges of ρ w, d/a and d a. As seen in Fig. 3, there are only three test data in the size interval spanning 1.2 to 2.4 m. The first has the longitudinal steel ratio of ρ w = 0.14%, the second 0.28% and the third 0.74%. The extremely low ρ w of the first two makes it impossible to find similar data in other intervals of d. For example, the minimum ρ w is 0.91% within the first interval of d, and 0.46% within the third interval. Therefore, one must consider the size range from to 1.2 m. Formulating a statistical optimization algorithm for database filtering (to be presented in a forthcoming journal article), one finds 7, 68, 17, and 36 data points within the admissible ranges for each interval of d (ideally, of course, the number of data in each interval should be the same, and the fact that it is not shows that complete elimination of statistical bias is impossible; nevertheless for obtaining reliable means, 7 data certainly suffice). After filtering, the mean values of ρ w for the restricted ranges are 1.51%, 1.5%, 1.5%, and 1.5%, the mean values of a/d are 3.45, 3.33, 3.33 and 3.23, respectively, and the mean values of da are 16.8, 17.0, 16.8 and 16.5 mm. This provides data samples with minimum bias in terms of ρ w, a/d and da. The data centroids for each interval are plotted as the diamond points in the plot of versus log d (Fig. 3(d)). We see that, despite enormous scatter in the database (Fig. 3(d)), the trend of these centroids is quite systematic. Under the assumption that the statistical weight of each size interval centroid in Fig. 3 is the same, the foregoing procedure is used to obtain the optimum least-square fit of these 4 centroids with the classical size effect law (type 2 energetic size effect law (32)), which was proposed for beam shear in 1984 (7) and recalibrated in 2005 (30), and is written here as where C, d0 = free constants to be found by the fitting algorithm (for reasons of proper weighting, it is best to conduct nonlinear regression with a nonlinear optimization subroutine, although a linear regression in transformed variables is possible and acceptable (9)). The resulting fit of the centroids (the solid curve) is seen to be quite close; it gives, for predicting the mean strength, a very small coefficient of variation of errors, namely ω= 2.5% = standard deviation of the optimum fit curve from the centroids, divided by the data mean (for individual beams, ω is, of course, much larger). This coefficient of variation characterizes the uncertainty in the mean strength of many structures of a given size rather than in the strength of an individual structure, which is of main interest for design. The negative curvature of the trend of the centroids confirms the theoretically predicted (8) gradual transition from quasiplastic behavior for small sizes to perfectly brittle behavior for large sizes. The trend of the last two centroids roughly matches the theoretical prediction of the slope -1/2 of the final asymptote of the size effect curve, given by vc ~ d -1/ 2 (7, 13, 29, 30, 32) (which is a property unanimously endorsed as fundamental in 2004 by ACI Committee 446). Using the same statistical algorithm, let us now increase the average steel ratio for each interval to 2.5%. The fitting of the centroids is shown in Fig. 3(e). The asymptotic slope of -1/2 is confirmed and the negative curvature is obvious. To increase the size range, one may further include one point from the largest size interval spanning 1.2 to 2.4 m, namely the Toronto beam with = 0.74%; see Fig. 3(f). w Admittedly, one data point is too little, but nothing more exists because of the cost of testing very large beams. Then the same procedure as above is followed and, for the other 32 Journal of SEWC April 2011

35 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice 4 intervals of d, one finds 1, 2, 5, and 15 data points for which the means of ρ w in the interval of d are 0.91%, 0.94%, 0.94%, 0.91% and 0.74%, while the mean of a/d (= 2.9) and the mean maximum aggregate size da (= 10 mm) are the same for each interval. The coefficient of variation of errors of mean prediction now is ω= 5%, and the size effect trend is very clear. Again, the trend agrees well with the asymptotic slope of -1/2 and with the energetic size effect law (solid curve, Fig. 3(f)). Now, an important point to note is that, for different averages, and, the trend of the interval centroids is the same, and closely matches the size effect law. This demonstrates objectivity of the data filtering approach. Also note that the present statistical results lend no support to the previously proposed power laws based on Weibull's statistical theory (33). Neither do they lend any support to the asymptotic size effect which is implied by an alternative model (34, 35) based on MCFT (Modified Compression Field Theory) (besides, an exponent of magnitude >1/2 is impossible thermodynamically as well as from the viewpoint of material strength randomness (8, 9, 36)). Variance of Individual Data Via Weighted Regression: Kinds 1 and 2 of bias afflict not only the mean trend of the full database, but also its scatter. The scatter may be measured by an unbiased coefficient of variation of the errors of the optimum fit curve compared to the individual data points. This is the error that must be considered for safe design. It can be ascertained by one of two methods: 1) One method is a simple bivariate nonlinear regression of our filtered restricted database, in which the kind 2 bias is already suppressed. To suppress the kind 1 bias, one needs to give the same weight to the data in each size interval i, regardless of the number mi of the points that fall into that interval. This may be achieved by assigning to the data in each interval i the normalized weight. Nonlinear regression, i.e., the minimization of the weighted sum of square deviations from the size effect law, then yields the coefficient of variation of 22.3% for the filtered database with =1.5%, and 23.6% for that with = 2.5% (Fig. 5). 2) The other method, which is the standard one, is a multivariate weighted nonlinear regression of the entire database. Compared to the first method, there is the complication that, instead of filtering the database, one must judiciously select the mathematical function describing the dependence of the parameters C and d0 of the size effect law for shear strength on the subsidiary influencing parameters ρ w, a/d and d a, and then optimize simultaneously the coefficients of all these functions by minimizing the variance of errors. Proper choice of these functions suppresses the kind 2 bias. The kind 1 bias is in ref. (30) minimized by weighting the data points in inverse proportion to the value of a smoothed histogram of the number of tests versus size. The result is quite similar to the first method - the coefficient of variation is 19.0%, after transformation to the variable. However, the range from the minimum to the maximum value of each subsidiary parameter (Fig. 4) fluctuates, from one size interval to the next, more than in the first method (ideally, the range should be the same for all the intervals, and the fact that it is not introduces some extra measure of bias, which cannot be removed although it probably is small). The effect of data weighting can further be clarified by Fig. 5(a,b) where the solid curves are the bivariate nonlinear regression curves of the interval centroids, with the same weight on each centroid. As one can see, almost undistinguishable curves (dashed ones) are obtained by the weighted nonlinear bivariate statistical regression of all the data points in the restricted (filtered) database. An unweighted regression of the same data points is shown in Fig. 5 by the dash-dot curves, and, as we can see, the dash-dot curve is again hardly distinguishable from the regression curve of the centroids in Fig. 5(a), but is very different in Fig. 5(b). One reason for this difference is that the vertical ranges of the restricted data in the individual size intervals, marked by vertical bars, are in Fig. 5(a) nearly symmetric with respect to the centroid curve, but not in Fig. 5(b). Another reason is that the restricted database in Fig. 5(a) is roughly homoscedastic, while that in Fig. 5(b) is not. For comparison, the coefficient of variation of the multivariate nonlinear regression conducted on the entire database (2) is 15% if the data are weighted, and 17% if unweighted. When only the 11 beams deeper than 1 m are considered, the coefficient of variation is 14% if these data are weighted and 16% if unweighted. As we see, the weighted regression gives a better prediction for the scatter of shear strength of large beams. Size Effect For Concrete Beams With Stirrups Although much information exists on the size effect on reinforced concrete beams without shear reinforcement, there is little information on the size effect in shear failure of beams with minimum or heavier shear reinforcement (stirrups). Many engineers are of the opinion that beams with minimum or heavier stirrups exhibit no size effect. However, this opinion is incorrect and would lead to unsafe designs for large structures. Computational simulations, and even the limited experimental evidence that exists, reveal that stirrups do not eliminate the size effect. They only mitigate it. According to the analysis by Ba ant (37), the energetic size effect law (32) remains valid and the effect of stirrups is to increase the transitional size d0 (9). Avoidance of size effect would require elimination of post-peak softening on the load-deflection diagram, and this could be achieved only if the concrete were subjected to triaxial confinement with all negative principal stresses exceeding in magnitude several times the uniaxial compression strength. The test series conducted by Walraven et al. (19) clearly show that there is a strong size effect for deep beams with April 2011 Journal of SEWC 33

36 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice a/d < 2 (Fig. 6(a)) to which the strut-and-tie model is applicable. As is well known, if the failure is triggered by the compression crushing of concrete strut, it typically exhibits size effect (8). For slender beams with a/d > 2, two test series are found in the literature: 1. tests conducted by Bhal (11) in 1968 in Stuttgart, in which the shear span ratio is a/d = 3, the shear reinforcement is heavier than the minimum requirement, and the size range is almost 1 : tests conducted by Kong and Rangan (20) in 1998 in Perth, in which the shear span ratio is a/d = 2.4, the shear reinforcement is heavier than the minimum requirement, and the size range is 1 : 3. When plotted in the logarithmic scale (Fig. 6(b,c)), it can be seen clearly that, in both data sets, the shear strength markedly decreases with increasing beam depth. The asymptotic size effect trend of slope -1/2 does not contradict these test results. Extensive finite element simulations based on the crack band model have also been carried out to investigate whether the shear failure of beams with stirrups exhibits a size effect. The beam geometry considered in these simulations is same as in the Toronto tests (3, 4, 18). Computations are run for geometrically similar beams of depths 0.47 m, 1.89 m, which is the size of Toronto test, and 7.56 m. The stirrups and longitudinal bars are assumed not to slip (although the bond slip was found to play only a minor role and tend to intensify the size effect). The mesh and the computed cracking pattern at maximum load are shown in Fig. 7(a), and the simulated loaddeflection diagrams are shown in Fig. 7(b), for all the sizes. The diagram for d = 1.89 m (the size tested in Toronto) shows the peak load of 283 kips. This is very close to the value measured in Toronto. The yield plateau observed in this test is also well reproduced by the simulation. However, for the largest beam simulated, the yield plateau disappears and the load descends steeply right after the peak. Fig. 7(c) shows the dependence of the average beam shear strength vn = V/bw d on beam depth d, and Fig. 7(d) shows the same for the average shear strength vc = Vc/bw d contributed by concrete (Vc = V - Vs, Vs = As fy d/s; As, s = stirrup area and spacing). Compared with the concrete beams without stirrups tested at same laboratory, the transitional size d0 shown in Fig. 7(c,d) is significantly increased. These plots document that a strong size effect exists also in the beams with stirrups, although it is pushed into larger sizes. The asymptotic slope of -1/2 is seen to remain. Together with the experimental evidence, the finite element simulations clearly demonstrate that the shear reinforcement, whether minimum or heavier than minimum, is unable to suppress the size effect. It mitigates the size effect in the larger size range, but not enough by far to make it negligible. Some Catastrophic Collapses With A Role Of Size Effect The overall safety factor µ, although not used in the current codes, is defined as the mean of failure test data divided by the mean (or unfactored) design load. The part of µ of concern here is the understrength factor. Besides the overt understrength factor j characterizing the brittleness of failure mode, there also exists a covert understrength factors jf due to the design formula error and jm due to the material randomness (38). Consequently, for shear failure of longitudinally reinforced concrete beams without stirrups, the overall safety factor currently is µ ~ 3.8 for small sizes and µ~1.7 for large sizes. The former is totally dominated by the live load, and the latter is totally dominated by the self weight. In the latter case, the neglected size effect factor has been considered (38) as 2. In view of the scatter in Fig. 3, the individual overall safety factors vary within 2.3 to 6 for small sizes, and 1.05 to 2.8 for large sizes. The very large values of these safety factors are doubtless one reason why, despite the neglect of size effect, there have not been many more structural collapses than actually experienced. These large values also reveal that, in concrete engineering (by contrast to aeronautical engineering), a single error in design or construction is usually not enough to bring the structure down. The size effect factor for normal concrete structures can hardly be more than 2, and so the size effect alone would rarely suffice to cause the collapse if the material strength and formula error have nearly mean values. To produce collapse, the material strength and formula error must simultaneously have values of small probability, far from the mean. Thus, at least two, and typically three, simultaneous mistakes or lapses of quality control are needed to make a concrete structure collapse. This makes it easy for an investigating committee to blame collapse entirely on the other factors and ignore the theoretically more difficult size effect. For example, in the case of catastrophic sinking of Sleipner oil platform in a Norwegian fjord in 1991 (Fig. 8(a)), which was due to shear failure of a thick tricell wall, there were three simultaneous mistakes. Besides two mistakes recognized by government forensic committee, the necessity of strength reduction of about 34% due to the size effect was pointed out but omitted from the conclusions. Of major interest for the size effect theory is the 1996 collapse of the Koror-Babeldaob Bridge in the Republic of Palau (Fig. 8(b)). This prestressed box girder had the world record span of 241 m when it was built in In addition to the erroneous initial prediction of creep and shrinkage deflections and apparently inappropriate remedial prestressing, one would have to expect a major strength reduction due to size effect on the compressionshear failure seen in the photograph. Analysis of this collapse would offer a unique opportunity to check and cali- 34 Journal of SEWC April 2011

37 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice brate the size effect theory but, incredibly, all the technical information was sealed after litigation by a court verdict. Scientific ethics demands this verdict to be reversed, in the interest of progress (imagine, e.g., that all the technical information on the collapse of Tacoma Narrows Bridge were suppressed). Another reason why structural collapses have not been more numerous is that most codes, unwittingly, hide a partial (thought imperfect) protection against size effect in an excessive value of the load factor for self-weight, which is 1.4 for the self-weight acting alone, according to the current ACI code. In small structures, the self-weight is a negligible part of the load, and so the value of self-weight load factor does not matter. But in a very large bridge, self-weight alone is the decisive loading case. Now, how could the self-weight be 40% larger than assumed in design? This is inconceivable (except as a sabotage). At most it could differ by a few percent. So very large structures are penalized by almost 40% compared to small ones. This way most codes give a covert protection against the neglect of size effect (39). But such covert protection is insufficient, by far, for very large structures. It also exhibits an incorrect trend from the viewpoint of size effect (39), as well as other wrong features. E.g., it gives greater protection to unprestressed or normal concretes compared to prestressed or high-strength concretes, because they lead to heavier structures, although the opposite should be the case because they are much more brittle; it gives too little protection to columns compared to beams, etc. This covert size effect should be eliminated and replaced by introducing the proper size effect in the code formulae. The Question Of Legal Liability Of Concrete Societies In the face of ever increasing diversification of science, it is nowadays impossible for the code-making committees, typically composed of the best and most renowned engineers, to follow in detail all the recently solidified scientific advances relevant to the building code article or recommended practice that they are developing. Nevertheless, keep informed they must. A quarter century ago, when the experimental data were scant and scattered, and only a handful of scientists espoused a coherent scientific theory, it was entirely plausible and defensible for concrete societies to ignore the size effect. When a failure attributable to size effect occurred, they could not be held liable. Not any more. The experimental evidence has become undeniable and the theoretical basis solid. Virtually all the researchers in fracture mechanics of concrete and entire research-oriented societies and committees in this field (e.g., IA-FraMCoS, ASCE-EMD, ACI Committee 446) have no doubt that a significant non-statistical size effect exists in all the brittle failures of concrete structures. Consequently, ignoring the size effect for the sake of simplicity, or even sanctioning a simplistic or partial consideration of size effect that is now known to imply a significantly increased risk of failure of large structures, is no longer acceptable. It might expose concrete engineering societies to legal liability when another catastrophe occurs. Conclusion At the dawn of this century, the size effect in brittle failures of concrete structure has become an established fact. It is time to introduce it into the design codes and practice. Ignoring it will cause large structures to be failing with the frequency of about one per thousand or more, instead of less than one per million as generally considered tolerable for engineering structures. The human society must not be knowingly exposed to such a risk. Acknoweldgment: Financial support from the U.S. Department of Transportation through the Infrastructure Technology Institute of Northwestern University under Grant No A210 is gratefully acknowledged. Author Affiliation Bazant, Zdenek P.: Ph.D., Dr.h.c. mult., S.E. McCormick School Professor and W.P. Murphy Professor of Civil Engineering and Materials Science, Northwestern University, 2145 Sheridan Road, Tech A135, Evanston, Illinois 60208; z-bazant@northwestern.edu Yu, Qiang: Ph.D, Research assistant, Northwestern University; qiangyu@northwestern.edu References 1. ACI-ASCE Committee 326 (1962). "Shear and diagonal tension" ACI Journal, Proceedings Vol. 59, pp (Jan.), (Feb.), (March). 2. Reineck, K-H, Kuchma, D. A., Kim, K. S. and Marx, S.(2003). "Shear Database for Reinforced Concrete Members without Shear Reinforcement" ACI Structural Journal, V. 100, No.2, March-April 2003, pp Podgorniak-Stanik, B.A. (1998). The influence of concrete strength, distribution of longitudinal reinforcement, amount of transverse reinforcement and member size on shear strength of reinforced concrete members M.A.Sc. Thesis, Department of Civil Engineering, University of Toronto, Angelakos, D, Bentz, E.C., and Collins, M.P. (2001). "Effect of concrete strength and minimum stirrups on shear strength of large members" ACI Structural J, 98 (3), pp Bazant, Z.P., Sener, S. and Prat, P.C. (1988). "Size effect tests of torsional failure of plain and reinforced concrete", Materials and Structures (RILEM, Paris), 21, pp Regan, P.E. (1986). "Symmetric punching of reinforced concrete slabs", Magazine of Concrete Research, V. 38, N. 136, September 1986, pp Bazant, Z.P., and Kim, J.-K. (1984). "Size effect in shear failure of longitudinally reinforced beams" Am. Concrete Institute Journal, 81, ; Disc. and Closure 82 (1985), pp Bazant, Z.P. (2002). Scaling of Structural Strength. Hermes Penton Science, London; 2nd updated ed., Elsevier April 2011 Journal of SEWC 35

38 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice 9. Bazant, Z.P., and Planas, J. (1998). Fracture and Size Effect in Concrete and Other Quasibrittle Materials. CRC Press, Boca Raton and London. 10. Leonhardt, F. and Walther, R. (1962) "Beiträge zur Behandlung der Schubprobleme in Stahlbetonbau" Beton-und Stahlbetonbau (Berlin), March, 54-64, and June, pp Bhal, N.S. (1968)?ber den Einfluss der Balkenhöhe auf chubtragfähighkeit von einfeldrigen Stalbetonbalken mit und ohne Schubbewehrung. Dissertation, Universität Stuttgart, Stuttgart, Germany. 12. Bazant, Z.P., and Sun, H-H. (1987). "Size effect in diagonal shear failure: Influence of aggregate size and stirrups" ACI Materials Journal, 84 (4), pp Bazant, Z.P., and Kazemi, M.T. (1991). "Size effect on diagonal shear failure of beams without stirrups" ACI Structural Journal 88 (3), pp Iguro, M., Shioya, T., Nojiri, Y. and Akiyama, H. (1985). "Experimental studies on shear strength of large reinforced concrete beams under uniformly distributed load" Concrete Library International of JSCE, No.5 (translation from Proc. JSCE, No. 345/V- 1, Aug. 1984), pp Shioya, T., Iguro, M. Nojiri, Y., Akiyama, H., and Okada, T. (1989). "Shear Strength of Large Reinforced Concrete Beams" Fracture Mechanics: Application to Concrete, SP-118, V.C. Li, and Z.P. Ba ant, eds., American Concrete Institute, Farmington Hills, Mich., 1989, pp Shioya, T. and Akiyama, H. (1994). "Application to design of size effect in reinforced concrete structures" Size Effect in Concrete Structures (Proc., Japan Concrete Institute International Workshop, Sendai), H. Mihashi, H. Okamura and Z.P. Ba ant, eds., E\&FN Spon, London, pp Kani, G.N.J. (1967). "How safe are our large reinforced concrete beams?" ACI Journal, 58(5), pp Lubell, A., Sherwood, T., Bentz, E., and Collins, M.P. (2004). "Safe Shear Design of Large, Wide Beams" ACI Concrete International 26 (1), pp Walraven, J.C. and Lehwalter, N. (1994). "Size effect in short loaded beams in shear" ACI Structural Journal, V. 91, No. 5, pp Kong, P.L. and Rangan, B.V. (1998). "Shear strength of high performance concrete beams" ACI Structural Journal, 95(6), pp ACI Committee 318 (2005). Building Code Requirements for Structural Concrete (ACI ) and Commentary (ACI 318R- 05), Am. Concrete Institute, Farmington Hills, Michigan. 22. Duckett, W. (2005). "Risk analysis and the acceptable probability of failure" The Structural Engineer, August, pp Melchers, R.E. (1987). Structural Reliability, Analysis & Prediction. Wiley, New York. 24. NKB (Nordic Committee for Building Structures) (1978). Recommendation for loading and safety regulations for structural design. NKB Report, No Bazant, Z.P., and Pang, S.-D. (2007). "Activation energy based extreme value statistics and size effect in brittle and quasibrittle fracture". J. of the Mechanics and Physics of Solids 55, Ang, A.H.-S., and Tang, W.H. (1984). Probability concepts in engineering planning and design. Vol II. Decision, Risk and Reliability, J. Wiley, New York. 27. Madsen, H.O., Krenk, S., and Lind, N.C. (1986). Methods of structural safety. Prentice Hall, Englewood Cliffs, N.J. 28. Haldar, A., and Mahadevan, S. (2000). Probability, Reliability and Statistical Methods in Engineering Design, J. Wiley & Sons, New York. 29. Bazant, Z.P., and Yu, Q. (2005). "Designing against size effect on shear strength of reinforced concrete beams without stirrups. I. Formulation" J. of Structural Engineering ASCE, 131 (12), pp Bazant, Z.P., and Yu, Q. (2005). "Designing against size effect on shear strength of reinforced concrete beams without stirrups: II. Verification and Calibration" J. of Structural Engineering ASCE, 131 (12), pp Ang, A.H.-S., and Tang, W.H. (1975). Probability concepts in engineering planning and design. Vol I. Decision, Basic Principles, J. Wiley, New York 32. Bazant, Z.P. (1984). "Size effect in blunt fracture: Concrete, rock, metal" J. of Engrg. Mechanics, ASCE, 110 (4), pp Weibull, W. (1939). "A statistical theory of the strength of materials" Proc. Royal Swedish Academy of Eng. Sci., pp. 151, Vecchio, F.J., and Collins, M.P. (1986). "The modified compression field theory for reinforced concrete elements subjected to shear" ACI J.. Proc., 83 (2), pp Collins, M.P., Mitchell, S., Adebar, P. and Vecchio, F.J. (1996). "General shear design method" ACI Struct. J., 93(1), pp Bazant, Z.P. (2004). "Scaling theory for quasibrittle structural failure", Proc., National Academy of Sciences 101 (37), pp (inaugural article). 37. Bazant, Z.P. (1997). "Fracturing truss model: Size effect in shear failure of reinforced concrete" J. of Engrg. Mechanics, ASCE 123 (12), pp Bazant, Z.P., and Yu, Q. (2006). "Reliability, brittleness, covert understrength factors, and fringe formulas in concrete dessign codes" Journal of Structural Engineering, ASCE, Vol. 132, No. 1, pp Bazant, Z.P., and Frangopol, D.M. (2002). "Size effect hidden in excessive dead load factor." J. of Structural Engrg. ASCE 128 (1), Fig. 1 Size Effect Tests of (a) Shear Failure; (b) Torsional Failure; and (c) Punching Shear Failure 36 Journal of SEWC April 2011

39 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice Fig. 2 (a) Probability Distribution of Shear Strength of Beams from 10 to 30 cm Deep, Based on the ACI-445F database, Compared to Toronto Data; (b) Failure Probability for Small Beam and 1 m Deep Beam. Fig. 4 Interval Centroids and Spread between the Maximum and Minimum Value Fig. 4 Interval Centroids and Spread between the Maximum and Minimum Values of Reinforcement Ratio rw and Shear Span Ratio a/d ; (a,b) for Full ACI-445F Database; (c,d) for Restricted Database with Mean rw 1.5%, a/d 3.3; (e,f) Ditto but rw 2.5%, a/d 3.3; (g,h) Ditto but rw 0.9%, a/d 3.0. Fig. 3 ACI-445F Database and Statistical Regression of Centroids of Test Data Subdivided into Intervals of Equal Size Ratio. (a-c) Full Database (the data retained are shown by larger circles and those filtered out in various cases by tiny circles); (d-f) Filtered Restricted Data Giving the Indicated Combinations of Uniform Mean Values of Subsidiary Parameters, Their Centroids and Regression Curves. Fig. 5 Regression Curves Corresponding to Weighted Fitting (Dashed Curves), Unweighted Fitting (Dash-dot Curves) and Fitting on Centroids (Solid Curves) for Filtered Database of (a) Average Steel Ratio = 1.5%; and (b) Average Steel Ratio = 2.5%. April 2011 Journal of SEWC 37

40 Consequences of Ignoring or Mis-Judging the Size Effect in Concrete Design Codes and Practice a) Sleipnerplatform, 1991 b) Palau bridge, 1996 Fig. 8 Examples of Catastrophic Failures of Concrete Structures. (a) Sleipner Oil Platform, 1991; (b) Koror-Babeldaob Box Girder in Republic of Palau, Fig. 6 Size Effect Test of Concrete Beams with Stirrups. (a) Deep Beam with a/d = 1; (b) Slender Beam with a/d = 3; (c) Slender Beam with a/d = 2.4. Fig. 7 Computational Simulations of Toronto Beam with Minimum Stirrups. (a) Mesh and Cracking pattern at failure; (b) Load-deflection Curves Generated by Simulations; (c) Size Effect Fitting of the Total Shear Strength; (d) Size Effect Fitting of the Concrete Shear Strength 38 Journal of SEWC April 2011

41 Damage Identification of Structures Through Simple and Measurable Indicators Raghu Prasad B.K, Lakshmanan N, Gopalakrishnan N and Muthumani K Abstract Simple, fast and robust strategies are required for estimating the health of a structure using globally available response data. The paper proposes a first stage structural health monitoring methodology using natural frequencies and static deflections as damage indicators. Equations and methodologies are proposed and validated in the paper, numerically and experimentally towards damage estimation. Identification methodology developed in this paper is applicable to distributed smeared damage, which is typical of reinforced concrete structures. The idea is that the first stage monitoring can be done for a large number of vulnerable structures in a remote manner and the features extracted from the data should help in determining whether any second stage detailed investigation is warranted. Key words: Structural health monitoring, Damage Indicators, Bridges and Buildings. Introduction A civil engineering structure is a plant (structure or system with a governing set of linear or differential equations) which takes some input (force) and produces some output (response). The input could be purely random or deterministic. The relationship between the input to output (I/O) defines the physical phenomenon that governs its structural behavior. This I/O relationship is termed as flexibility (or stiffness) in a static structure, compliances (with frequency dependence) in a dynamic structure or an impedance in an electrical circuit. In many of the situations when a built structure or existing mechanical component is to be assessed, the I/O relationship has to be generated whole or part. The problem of identifying this I/O relationship is termed as "system identification". This is typically an "inverse problem". A good example is a pile foundation, which is suspect for its load carrying capacity (which cannot be inspected visually) and its state of well being could only be inferred through its response to a set of dynamic or static forces. Contrary to a forward problem, the peculiar challenges that are associated in a system identification and the associated inverse problem are: (a) Unwanted signals come in the form of noise both in the input force as well as in the response and this may be purely Gaussian or ground-loop electrical noise or due to other sources (temperature, humidity etc.) (b) There may be less number of measurements and more number of un-known variables (under-determined). (c) Non-uniqueness of the response, where many possible inputs could give rise to the same response. (d) There may be more number of measurements and less number of un-known variables (over-determined). (e) Less sensitivity of the structure or part of the structure in response to impressed forces. (f) System-non-linearity, friction and other causes, which are difficult to quantify in precise mathematical terms. (g) Uncertainty in modeling boundary conditions, joints, and opening-closing cracks (which shows different stiffness at each half-cycle) and so on. Previous Investigations on Damage Assessment Damage-induced variation of static deflection in a beam is used to identify the concentrated damage through an inverse analysis and closed form expression by Caddemi and Morassi [1] and Caddemi and Greco [2]. Interestingly, the procedure developed by them, requires measurement of displacements at both the sides of a concentrated crack to evaluate the crack position. A Monte-Carlo simulation procedure is also used to evaluate the effect of Gaussian distributed measurement errors on the damage evaluation procedure. The paper by Buda and Caddemi [3], also on similar lines, outlines a strategy for damage identification for concentrated damage in an Euler-Bernoulli beam. The static damage identification problem is reduced to a Fredholm integral equation of the second kind characterized by a Pincherle-Goursat kernel by Paola and Billelo [4] in another note-worthy work. A simple and elegant formulation by Choi [5] and Choi et. al [6], presents an elastic damage load theorem, which indirectly states that change in deflection due to damage is maximum near the vicinity of April 2011 Journal of SEWC 39

42 Damage Identification of Structures Through Simple and Measurable Indicators damage and absolute maximum when the position of load, damage and measurement locations coincide. The available literature on system identification and damage detection using only dynamic measured data is huge. A comprehensive survey is presented by Doebling et al. [7], who have reviewed the numerous technical literatures available on damage detection through vibration testing. Hassiotis et. al. [8], Hassiotis [9] outline a method based on first order perturbation and optimization theory to compute the damage from measured natural frequencies. Identification of damage locations in plate-like structures using strain modal approach is proposed by Li et. al [10], using bending moment index and residual strain mode shape index. A combined static and dynamic approach for damage identification using curvature mode shape and strain frequency response function is proposed by Yam, et. al [11]. The combined static and dynamic data or static-only data for damage detection are less. Various structural properties of anisotropic composites are identified using probability of detection (POD) concepts and boundary element numerical simulation, using static data by Rus et. al [12]. Use of wavelets in damage identification using experimentally derived deflection through digital photography by Rucka and Wilde [13] is an interesting study. Nejad et. al [14] have used an optimization method that minimizes the difference between the load vector of the damaged and undamaged structure using static noisy data for damage identification. A static-dynamic combined damage detection technique, without prior information of intact structure is used by Vanlanduit et. al [15]. Sensitivity studies on static deflection curvature, curvature mode shape and strain frequency response function on damages have been used by Yam. et. al [16] in yet another interesting study. Another combined static-dynamic system identification study using measured damage signature and predicted damage signature is through the work of Wang et. al [17]. Recently wavelet and radial basis function (ANN) based damage identification from dynamically measured parameters is proposed by Lakshmanan et al. [18,19] and Raghuprasad et al. [20, 23]. In a previous work Lakshmanan et. al [21,22] have proposed a smeared damage model and the sensitivity matrix of static deflection due to damage (reduction of flexural EI) formulated and solved as an over-determined sets of equation. Definition of Damage A structure is deemed to have been damaged if the structure, after un-loading could not return to its original state and there is a permanent deformation with loss of energy. Damage is defined, as per International Standards Organization, "as an unfavorable change in the condition of a structure that can affect the structural performance". A structural member can suffer varying degrees of damage due to reasons such as over loading, environmental ageing, corrosion, poor quality of construction, fatigue crack growth under cyclic loading, and creep. On many occasions, a maintenance engineer is required to take decisions regarding the repair and improvement of the damaged structure. The viability of repair has to be weighed with the cost of new replacement and this is governed by the state of damage suffered by the structure. Estimation of the magnitude of damage, location and its spread thus plays a crucial role in the repair methodology to be adopted. Also, residual strength and remaining life depend on the magnitude and position of damage. Structural health monitoring (SHM) is a field of science dealing with the mathematics, algorithms, instrumentation, data-analysis and feature extraction methods for damage identification of structures. SHM provides an engineer with mathematical, structural and electronic tools such that damage could be measured and quantified. This is analogous to a doctor provided with various diagnostic tools for identifying the health of a human being. The methodology adopted for identification should be simple, both in terms of investigations involved and the instrumentation. Researchers have proposed various methodologies including damage identification from mode shapes, wavelet-based formulations and optimizationbased damage identification and instrumentation schemes and so on. These are technically involved but require more measurements for all critical bridges/buildings, where the sheer volume of number of structures to be investigated is enormous. Ideally, structural health monitoring has to be carried out in two stages: (a) Stage-1: Remote monitoring of global damage indicators and inference of the health of the structure. Instrumentation for this stage should be less, simple, but at critical locations to capture the global damage in a reasonable sense. (b) Stage -2: If global indicators show deviation beyond a specified threshold, then a detailed and localized instrumentation and monitoring, with controlled application of static and dynamic loads is to be carried out to infer the health of the structure and take a decision on the repair and retrofit strategies. Damage Identification from Natural Frequency Change In this section, another method based on contours of equal frequency change, named, "Iso_Eigen_value_change contours" is proposed. It is shown that the intersection of two or more contours with ratios of frequency changes between multiple pairs of frequencies shall give the position and extent of damage. It may be proved that the ratio of the changes in natural frequencies for any pair of modes (say N and M-th modes) remains constant irrespective of the magnitude of damage, '?'. However, the criterion is valid for a single contiguous damage and many researchers have applied this methodology to concentrated damage idealised by a rotational spring. The extension of this method for a distributed damage covered here is novel and unique. 40 Journal of SEWC April 2011

43 Damage Identification of Structures Through Simple and Measurable Indicators Necessary equations for drawing Iso_Eigen_value_change contours are developed using first-order perturbation theory. Analytical validation of these curves is carried out for simulated damage values. Fig.1 shows the parameters that define a damage. Damage Identification From Iso_Eigen_Value_Change Contours of contours. Following physical in-sights can be obtained from the general contours : Contours are symmetrical indicating that a symmetrical damage affects the frequencies equally. The boundary line shows that the extent of damage, on the Y-axis shall at best be twicehe centre of damage location, on the X-axis. Fig. 1 Simply supported beam with a reduced EI for portion of its length Generally, damage due to over-loading of a structure, affects certain definite length of the structure, which is contiguous. The distribution of damage, over this affected length could be in some known variation, say, less at ends of damage zone and more at the centre of the damage. Hence identification of the parameters that define a single contiguous damage is more relevant as generally damage does not happen in some abrupt variation along a structural element. The knowledge of at least three frequency changes are essential for retrieving the unknown parameters, which define a damage. Hence the strategy is to eliminate? by taking the ratio of the change in frequencies for two natural modes. Ratio of the changes in Eigen values are obtained after modifying and re-writing relevant equations as, hence it is inferred that the ratio of changes in Eigen values, for two modes 'n' and 'm' (squared frequencies) are independent of the magnitude of damage and are functions of only the central position of damage and its extent. Indirectly, the method is to find out the ratio of variation in frequencies as the normalizing factor (1) which is a constant for any pair of frequencies. Having eliminated a variable (β ), out of three, it is possible to construct curves such that the variation of Eigen-change-ratios can be visualized. Varying values of which give rise to constant Eigen-change-ratios for a simply supported beam are plotted in the form Contours at the top most point indicate a wide-spread constant damage and all frequencies are affected equally and hence the ratio is 1.0. Contours closer to support are around 3 to 4, indicating that high frequencies are affected more by off-centric damage. As a converse of point (d), contours closer to mid-span, show smaller values especially for small amounts of damage. This exercise on the theoretical damage identification study involves certain example validation problems which are solved using these contours. The example validation problems (D1 to D2) solved are as follows : ( is the magnitude of damage, kept as 0.10) (a)d1: (b) D2: 2 n 2 m For example, in case(a) 2 2 m n for the pairs of 2-1, 4-2, 3-1 and 4-1 frequency ratios are 1.897, 0.129, and respectively. The intersection point of all the contours indicate the position and extent of damage, which in this case works out to be 0.25 and 0.1 respectively (Fig. 2 (a) ). Figures 2 (b) shows the results of the remaining case. After the geometric details of the damage are got, it is easy to extract the magnitude of damage. The least square estimation can be adopted, for this over-determined set of equations, (with four equations and one un-known), which in this case works out to be It is to be observed that, for symmetrical damages, the intersection point is also a point of zero slope and hence the contours are tangential to each other. An extra frequency information may be required to resolve this issue or additionally, it is also proposed to have a static based system identification procedure, which shall resolve a symmetric and an un-symmetric damage, in conjunction with dynamic measurements. Damage Estimation Through Deflection Measurements At The Load Application Point April 2011 Journal of SEWC 41

44 Damage Identification of Structures Through Simple and Measurable Indicators A simplified formulation based on flexibility approach is written, using only deflection measurements and through monitoring of the change in deflection (ratio of change in displacement to the original displacement). Certain conditions are also imposed such that the value of measured deflection is more and hence the errors are less. These include Deflection is measured only at the loading point is used in the inverse analysis. Ratio of the change in deflection is used instead of the absolute value of deflection. and ud are subscripts for damaged and un-damaged states. Here 'α' ( 0 < α< 1)is the remaining ratio of EI after a damage of 'β EI' has occurred (1-α = β). The factor is termed as the modified damage factor ( β* ) It is to be noted that the strain energy and subsequently deflections are proportional to the modified damage factor and not to the damage factor themselves. Total strain energy (U) due to flexure can be summed up for various regions of the un-damaged Euler-Bernoulli beam (Fig. 3) as in the following expression U ud 2 M dx 2EI 2 M dx 2EI 2 M dx 2EI 2 M dx 2EI A B C D 2 M dx 2EI (2) Fig. 3 Simply Supported Beam taken for Static System Identification Studies Reckoning 'x' from the right-hand side support and the change of deflection is sought below the applied load itself, (5) (6) Fig. 2 (a,b) Prediction of Damage Position and Extent from Iso_Eigen_change Contours (a) D1, (b) D2 For the damaged beam, only the region of damage is replaced with an α EI. (3) This can be modified and written as, U U d U U d ud ud U U ud ud 1 1 C 1 2 M dx 2EI C C 2 M dx 2EI 2 M dx 2EI C C 2 M dx 2 EI (4) 2 M dx 2EI Where, M = moment at any section, EI = flexural rigidity ; d where is the incremental increase in deflection at the 'i-th' node due to a damage magnitude βj at the 'j-th' element. If there are 'N' damage sites in the beam, the effect of all the damages can be simply summed up and the above equation can be written as, For a particular case of a uniform and widespread reduction in EI, the equation can be modified and the effects on both the left and right sides of the load are added and the resulting expression is, The above equation reinforces the well-known fact that for a uniform decrease in EI, throughout the beam, by a factor of α, deflection increases by a facor of 1/ α and change in deflection is ((1/ α)-1) The ratio of increased deflection to the original deflection, (7) (8) (9) 42 Journal of SEWC April 2011

45 Damage Identification of Structures Through Simple and Measurable Indicators for the same loading can similarly be written as, (10) In a system identification procedure, for deriving the system parameters from the known values of ratio of changes in deflections from the original deflection values, above equation can be re-cast into a matrix form with un-known vector composed of and known vector of, both related by a coefficient matrix depending on the geometry of load position and damage. The matrix 'A', termed as the sensitivity matrix, is constructed relating the sensitivity of damage at element 'j' to an increased deflection at node 'i'. The vector 'b' is the un-known, modified damage factor (β*) at various elements and vector 'c' is the known change in deflection, expressed as the ratio of original deflection, occurring at node 'i' due to the effect of damages in all elements and which can be measured. Fig.4 shows the results of the numerical study with simulated and retrieved damages. Conclusions Iso_Eigen_value_Change contours constructed from Cawley-Adams criteria are the effective means of identification of damage in terms of damage position and extent. It is possible to estimate the damage magnitude, as an over-determined problem, once the position and extents are known. These contours are more tolerant to errors in measurements and up to 5% deviation, a region of intersection is identified rather than a point and the centre of the region is the likely damage location. The method of damage identification from static measurements improve, if drive-point deflections are used (load and deflections at the same point), rotations derived from deflections are not used in the formulations and more redundancy is incorporated in measurements. Acknowledgement The paper is published with the approval of Director, CSIR-SERC and his constant support and encouragement are gratefully acknowledged Author Affiliation Raghu Prasad. B.K, Professor, Indian Institute of Science, Bangalore, Lakshmanan. N, Scientist, CSIR-SERC, Chennai, Gopalakrishnan. N, Corresponding Author, Muthumani. K, Scientist, CSIR-SERC, Chennai References 1. Caddemi, S and Morassi A, "Crack detection in elastic beams by static measurements", Intl. Jl. of Solids and Structures, 44 (2007), 2. Caddemi, S and Greco A, "The influence of instrumental errors on the static identification damage parameters for elastic beams", Computers and Structures, 84(2006), Buda, G and Caddemi, S, "Identification of concentrated damages in Euler-Bernoulli beams under static loads", Jl. of Engg. Mech., ASCE, 133 (2007), Paola, M D and Bilello, C, "An integral equation for damage identification of Euler-Bernoulli beams under static loads", Jl. of Engg. Mech., ASCE, 130 (2004), Choi, I Y, Lee, J S, Choi, E and Cho, H N, "Development of Elastic damage load theorem for damage detection in a statically determinate beam", Computers and Structures, 82 (2004), Choi, I Y, "Damage identification techniques for bridges using static response", Ph. D thesis, Hanyang university, Doebling S.W., Farrar C. R., Prime M. B., Shevitz, P. W., "Damage Identification. Health monitoring of structural and mechanical systems from changes in their vibration characteristics - A literature review", Los Alamos National Laboratory, Los Alamos, New Mexico, (1996). 8. Hassiotis, S. and Jeong G.D., "Assessment of Structural damage from natural frequency measurements", Computers and Structures, 49, (1993), Hassiotis, S., "Identification of damage using natural frequencies and Markov parameters", Computers and Structures, 74, (2000), Li, Y Y, Cheng L, Yam L H and Wong W O, "Identification of damage locations for plate-like structures using damage sensitive indices: strain modal approach", Computers and Structures, 80 (2002), Yam, L H, Li, Y Y and Wong W O, "Sensitivity studies of parameters for damage detection of plate-like structures using static and dynamic approaches", Engineering Structures, 24 (2002), Rus, G, Lee S. Y and Gallego R (2005), "Defect identification in laminated composite structures by BEM from incomplete static data", Intl. Jl. of Solids and Structures, 42 (2005), Rucka, M and Wilde, K, "Crack identification using wavelets on experimental static deflection profile", Engineering Structures, 28(2006), Nejad F.B, Rahai, A and Esfandiari a, "A structural damage detection method using static noisy data", Engineering Structures, 27(2005), Vanlanduit, S, Parloo, E and Guilaume P, "Combined damage detection technique", Journal of Sound and Vibration 266(2003), Yam L.H, Li Y.Y and Wong W.O, "Sensitivity studies of parameters for damage detection of plate like structures using static and dynamic approaches", Engineering structures, 24 (2002), Wang X, Hu N, Fukunaga H and Yao Z H, "Structural damage identification using static test data and changes in frequencies", Engineering structures, 23 (2001), Lakshmanan, N, Raghuprasad, B.K., Muthumani, K., Gopalakrishnan, N., and Basu D. (2007). "Wavelet analysis and April 2011 Journal of SEWC 43

46 Damage Identification of Structures Through Simple and Measurable Indicators enhanced damage indicators", Smart Structures and System- An Intnl Jl., 3, Lakshmanan, N, Raghuprasad, B.K., Muthumani, K., Gopalakrishnan, N., and Basu D. (2008). "Damage evaluation through radial basis function network (RBFN) based artificial neural network scheme", Smart Structures and System- An Intnl Jl., 4, Raghuprasad, B.K., Lakshmanan, N, Gopalakrishnan and N., Muthumani, K. (2008), "Sensitivity of Eigen values to damage and its identification", Structural durability and health monitoring, 4(3), Lakshmanan, N, Raghuprasad, B.K., Muthumani, K., Gopalakrishnan, N., and Basu D. (2009). "Identification of reinforced concrete beam-like structures subjected to distributed damage from experimental static measurements", Computers and Concrete- An Intnl Jl., 4 (1). 22. Lakshmanan, N, Raghuprasad, B.K., Gopalakrishnan, N., Sathishkumar, K., and Murthy S.G.N., Detection of contiguous and distributed damage through contours of equal frequency change, Journal of sound and vibration (2010). 23. Raghuprasad, B.K., Lakshmanan, N, Muthumani, K. and Gopalakrishnan N., "Enhancement of damage indicators in wavelet and curvature analysis", Sadhana, 31(4), (2006), Fig. 4 Various Damage patterns for damage identification for a simply supported beam 44 Journal of SEWC April 2011

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68