comparison of industry-standard nonlinear dynamic …

11
Tenth U.S. National Conference on Earthquake Engineering Frontiers of Earthquake Engineering July 21-25, 2014 Anchorage, Alaska 10NCEE COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC ANALYSIS METHODS WITH OBSERVED DAMAGE ON A RC BUILDING A. Kozmidis 1 , M. Melek 2 , L. Massone 3 and K. Orakcal 4 ABSTRACT Observations from recent earthquakes such as Chile (2010) and Christchurch (2011) showed vulnerabilities of reinforced concrete buildings with bearing wall lateral force resisting systems. The observed failure patterns included not only web crushing due to shear demand but also flexural failures followed by the buckling and/or fracture of longitudinal wall reinforcement and concrete spalling and crushing. The observed vulnerabilities in modern buildings have raised an intriguing question: “Are the analysis methodologies currently used by the engineering community able to predict the response of shear wall buildings with acceptable level of accuracy?” In order to provide an answer for this question, a building in Chile, which has experienced damage during the 2010 M8.8 Maule Earthquake, was assessed using industry standard nonlinear analysis methodologies. Comparisons of the analysis results with the observed damage patterns were conducted and acceptability of these techniques was investigated to provide a better understanding of the shortcomings of currently used industry standard analysis and design methodologies. Results have shown inconsistencies in the observed damage versus nonlinear time history analysis results specifically in regard to the location of damage and shear failure observations. Better representation of shear-flexure interaction effects, attenuation relationships and soil structure interaction is required for further studies on the topic. 1 Senior Engineer, Arup, Los Angeles USA 2 Consultant, Oklahoma City USA 3 Associate Professor, University of Chile, Department of Civil Engineering , Santiago Chile 4 Associate Professor, Bogazici University, Department of Civil Engineering, Istanbul Turkey Kozmidis A, Melek M, Massone L, Orakcal K. Comparison of industry-standard nonlinear dynamic analysis methods with observed damage on a RC building. Proceedings of the 10 th National Conference in Earthquake Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.

Upload: others

Post on 24-Dec-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

Tenth U.S. National Conference on Earthquake Engineering Frontiers of Earthquake Engineering July 21-25, 2014 Anchorage, Alaska 10NCEE

COMPARISON OF INDUSTRY-STANDARD

NONLINEAR DYNAMIC ANALYSIS

METHODS WITH OBSERVED DAMAGE

ON A RC BUILDING

A. Kozmidis1, M. Melek

2, L. Massone

3 and K. Orakcal

4

ABSTRACT

Observations from recent earthquakes such as Chile (2010) and Christchurch (2011) showed

vulnerabilities of reinforced concrete buildings with bearing wall lateral force resisting systems.

The observed failure patterns included not only web crushing due to shear demand but also

flexural failures followed by the buckling and/or fracture of longitudinal wall reinforcement and

concrete spalling and crushing. The observed vulnerabilities in modern buildings have raised an

intriguing question: “Are the analysis methodologies currently used by the engineering

community able to predict the response of shear wall buildings with acceptable level of

accuracy?” In order to provide an answer for this question, a building in Chile, which has

experienced damage during the 2010 M8.8 Maule Earthquake, was assessed using industry

standard nonlinear analysis methodologies. Comparisons of the analysis results with the

observed damage patterns were conducted and acceptability of these techniques was investigated

to provide a better understanding of the shortcomings of currently used industry standard

analysis and design methodologies. Results have shown inconsistencies in the observed damage

versus nonlinear time history analysis results specifically in regard to the location of damage and

shear failure observations. Better representation of shear-flexure interaction effects, attenuation

relationships and soil structure interaction is required for further studies on the topic.

1Senior Engineer, Arup, Los Angeles USA

2 Consultant, Oklahoma City USA

3 Associate Professor, University of Chile, Department of Civil Engineering , Santiago Chile

4 Associate Professor, Bogazici University, Department of Civil Engineering, Istanbul Turkey

Kozmidis A, Melek M, Massone L, Orakcal K. Comparison of industry-standard nonlinear dynamic analysis

methods with observed damage on a RC building. Proceedings of the 10th

National Conference in Earthquake

Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.

Page 2: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

Comparison of Industry-Standard Nonlinear Dynamic Analysis Methods

With Observed Damage on a RC Building

A. Kozmidis1, M. Melek

2, L. Massone

3 and K. Orakcal

4

ABSTRACT Observations from recent earthquakes such as Chile (2010) and Christchurch (2011) showed

vulnerabilities of reinforced concrete buildings with bearing wall lateral force resisting systems.

The observed failure patterns included not only web crushing due to shear demand but also

flexural failures followed by the buckling and/or fracture of longitudinal wall reinforcement and

concrete spalling and crushing. The observed vulnerabilities in modern buildings have raised an

intriguing question: “Are the analysis methodologies currently used by the engineering community

able to predict the response of shear wall buildings with acceptable level of accuracy?” In order to

provide an answer for this question, a building in Chile, which has experienced damage during the

2010 M8.8 Maule Earthquake, was assessed using industry standard nonlinear analysis

methodologies. Comparisons of the analysis results with the observed damage patterns were

conducted and acceptability of these techniques was investigated to provide a better understanding

of the shortcomings of currently used industry standard analysis and design methodologies.

Results have shown inconsistencies in the observed damage versus nonlinear time history analysis

results specifically in regard to the location of damage and shear failure observations. Better

representation of shear-flexure interaction effects, attenuation relationships and soil structure

interaction is required for further studies on the topic.

Introduction

For more than half a century, typical mid to high-rise residential and commercial buildings in

Chile has been constructed using reinforced concrete bearing wall systems. Experiences from

four major earthquakes during this time frame (Valdivia M9.5 1960, Valparaiso M7.5 1971,

Valparaiso M7.8 and Maule M8.8 2010) have shaped the construction and design practice of the

Chilean engineering community. Following the 1985 Valparaiso Earthquake, the 1983 version of

the ACI requirements for reinforced concrete (ACI 318-83, 1983) was adopted, with some

exceptions that notably reduced concrete cover and relaxed boundary element confinement

requirements [1]. The impact of these exceptions on the structural damages observed in some of

the bearing wall systems after the 2010 M8.8 Maule Earthquake is currently under study by the

engineering research community.

Of particular importance is that the 2010 earthquake occurred in a county with a modern seismic

design code as well as good engineering and construction practice. During the earthquake, one

1Senior Engineer, Arup Los Angeles

2 Consultant, Oklahoma City

3 Associate Professor, University of Chile, Chile

4 Associate Professor, Bogazici University, Turkey

Kozmidis A, Melek M, Massone L, Orakcal K. Comparison of industry-standard nonlinear dynamic analysis

methods with obersved damage on a RC building. Proceedings of the 10th

National Conference in Earthquake

Engineering, Earthquake Engineering Research Institute, Anchorage, AK, 2014.

Page 3: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

building suffered complete collapse (Alto Rio Building), whereas numerous buildings have

experienced partial collapse or severe damage. Consequently, performance of the existing

reinforced concrete building stock has served as a platform to provide valuable information on

the vulnerabilities existing in the modern building codes. Large axial stress ratios, relaxations in

the code after 1985 earthquake, poor detailing and wall configuration were some of the primary

causes of damage [2]

This paper attempts to predict the seismic response, using a commonly used nonlinear analysis

software: Perform3D (CSI, 2011) [3], by investigating a tall building, located in Santiago, Chile.

The building structural system, reinforced concrete bearing walls, has suffered substantial

damage during the 2010 Earthquake. The reports and site investigations show that the building

has experienced concrete crushing followed by reinforcement buckling at lower levels during the

earthquake. Herein, we try to answer the question “Can the damage be predicted for buildings

with the industry-standard analysis methods?”. Industry standard nonlinear analysis tools and

software were used to model the building in order to predict the observed damage. Comparisons

and recommendations are provided within this paper.

Chile Earthquake

Chile is familiar with large intensity earthquakes in its history due to its location relevant to

tectonic plates. Nazca plate which situates by the west of Chile has been one of the most active

tectonic plates, moving at an absolute rate of 8cm/year to east [4]. The eastward movement is

causing the plate to be subducted beneath the South-American plate by creating a convergent

plate boundary in Chile. Subduction zone running through shores of Chile is depicted in Figure 1

with a pink line. This particular formation of subduction zone is known to produce largest

earthquakes due to the fact that it allows stress to be built up before the energy is released. These

energy releases were usually occurred in large magnitude in history (1960 Valdivia Earthquake –

Mw 9.5, 1985 Valparaiso Earthquake – Mw 7.8) 2010 Chile earthquake was one of these largest

magnitude earthquakes occurring with a moment magnitude of 8.8 on Richter Scale and a

rupture zone of 500km by 140km. This is the sixth largest earthquake ever recorded. Cities of

Concepcion, Arauco and Coronel, have experienced the strongest shaking with either VIII or IX

on the Mercalli Scale. The epicenter and seismic hazard of the earthquake is shown on figure 1.

The earthquake was followed by tsunamis in major cities of Chile, regions of Japan and has also

caused small size damages in shores of San Diego. According the official resources, the total

estimated loss of lives is 525 in addition to 25 people missing. Low to severe damage have

occurred in more than 200 hundred buildings and one of them collapsed completely.

Structural damages have been observed mostly in tall buildings with bearing wall systems.

General damage mechanisms observed in buildings in Chile is due to spalling and crushing of

concrete and buckling of vertical reinforcement at levels close to the ground level. [2] Some of

the later assessments on the damaged structures have shown that typical 20cm horizontal

reinforcement spacing and 90 degree hooks were used at wall ends. This has resulted in a less

dense confinement zones in the wall ends. Use of walls with smaller thickness (15cm-20cm) has

also resulted in comparably small concrete core areas. Further damages of reinforcement

buckling were observed in the web boundaries of T and L shaped walls due to confinement

problems. The causes of the damages being located at areas close to ground level can be

explained by concrete quality.

Page 4: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

Figure 1. Seismic hazard map for Maule earthquake [5]

Building Description and Observed Damage

Benchmark building, is located in the capital city of Santiago which is the highest populated

urban area of the country. The building is 18-story tall with reinforced concrete bearing wall

system as its primary seismic force resisting system. There are also two additional basement

floors below podium level. Concrete with a cubic strength of 30 MPa and conventional steel,

A630-420H, conforming to national code requirements was used in the construction of the

building. Construction of the building was finished in 2007 and the initial design adopted the

Chilean codes after 1985 Earthquake. Percentage of wall to floor area ranges between 2.5-3.5%

which is in line with Chilean construction industry standard of the buildings constructed at the

same era. Typical wall layout for the building includes flanged and non-flanged walls with

uniform reinforcing of 8 to 12 mm diameter rebar and additional reinforcing at ends with varying

rebar diameter from 16 to 32mm. Figure 2 depicts rendering of the building and shows the

typical plan layouts for various floors.

Based on the investigations conducted after the earthquake, there were several areas in the

building which observed nonstructural failure of partition walls. In addition to several

nonstructural damages, the building has also experienced reinforcing steel bar buckling due to

crushing of concrete. Damages have occurred mostly in the regions of webs of T shaped walls at

subterranean levels close to the podium level. This is due to the backstay effect with recessed

walls at basement levels. Per on-site investigations after the earthquake, observed damages for

the building have occurred on gridlines N, Q and U of the first subterranean level. The areas of

damage are depicted in figure 3.

Model Description

The analyses for the investigated buildings were carried out in Perform 3D (CSI, 2011) which

enabled the use of nonlinear analysis methods while explicitly modeling fiber elements with

constitutive material models. Per as-built drawings, use of 90-degree hooks at wall ends was

observed. In the light of aforementioned reason and considering the omission of confinement

Page 5: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

requirements in the existing code, the analysis model assumes unconfined concrete stress strain

curve using Hognestad model. 2.5% of Rayleigh damping was considered in the analysis per

Section 3.2.1 of LATBSDC 2011 Guidelines.

Figure 2. Benchmark building, perspective view (left), typical plan layout of walls of upper

floors (top right), typical plan layout of walls of basement levels (bottom right).

Figure 3. Observed damage at gridline Q (left) [6] and damage key on the wall layout (right)

Wall Modeling

Walls are modeled with industry standard fiber modeling techniques to capture stress

strain relationship throughout each fiber element. Fixed fiber sections are considered at the edges

to capture the behavior precisely. Typical macro modeling of structural walls has evolved

throughout the years to include custom response characteristics. Recent macro modeling

techniques such as multiple-vertical-line element model, MVLEM (initially proposed by

Vulcano et al. later improved by Orakcal, Wallace and Conte), uses multiple fibers connected to

rigid base to predict flexural behavior. Each fiber has its own response characteristic due to its

force-displacement relationships. Perform 3D software uses a similar approach to model

nonlinear flexural behavior of concrete walls. Although flexural modeling approach of the

industrialized software is similar to that of the MVLEM, modeling of wall shear response

Page 6: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

incorporates more flexibility. Perform3D allows the user to model inelastic or elastic shear

material separately from the flexural fiber model, whereas the MVLEM elements incorporate

horizontal springs at which shear deformations are concentrated. Based on the observed damages

which are governed by flexure and due to the height of the building, use of elastic shearwall

material was considered in the Perform3D analysis model. [7]

Coupling Beam Modeling

Post-1985 buildings in Chile mainly include non-structural components in between walls.

This is due to the poor performance of slender coupling beams in 1985 earthquake [8]. In the

benchmark building, walls are connected through the slab for most of the cases. However,

building also includes few coupling beams per as-built drawings. Although the failure

mechanism did not include failure of coupling beams, for the sake of industry-standard approach

of modeling, coupling beams are explicitly modeled in this study by lumped plasticity approach.

Modeling parameters of plastic hinge rotation and residual strength ratios for these coupling

beams are considered based on Table 6-18 of ASCE41-06.

Seismic Hazard

Considering there is no record close to the building, couple of various records varying in

peak ground acceleration was used for the analysis. Figure 4 shows three acceleration records

that are applied onto the building. Histories are obtained from records provided from different

ground motion stations in central Santiago and Maipu which is considerably closer to the site of

the benchmark building. Figure 5 shows spectral accelerations of each record. Given the soil

conditions, it is less likely that the building behavior is well represented by the Santiago record.

Figure 4. Ground motion records, TH18 Santiago-Conjunto Villa Andalucia (top), TH19

Santiago-Conjunto Villa Andalucia-corrected (middle) and TH2021 Maipu-Centro de

Referencia de Salud [3],[9]

Page 7: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

Figure 5. Spectral accelerations of hazards for each direction

Analysis Results

Nonlinear dynamic analyses on the building show that some walls on the building undergo

failure of concrete during earthquake. However the building is able to dissipate energy. Coupling

beams have performed well with varying plastic hinge rotations of 0.004 to 0.01 radians. This is

in line with the observed damage that no major energy dissipation has occurred and no damage

observed for coupling beams.

Story Drifts

Figure 6 shows building drift time-history for each record. It is observed that maximum residual

drift in the building is approximately 0.1%. This value would fall into Immediate Occupancy

performance criterion per structural performance levels defined in Table C1-3 of ASCE41-06.

Figure 7 depicts the extrema of inter-story drifts on the building with transient drift limit states

shown per ASCE41-06. Drift values stay within 1% which can conclude to building satisfying

drift requirements for Life Safety performance level. Although buildings in Chile are not

designed for any probabilistic or deterministic earthquake definition, these drift results are

satisfying an industry-standard performance level under DBE level earthquake.

Figure 6. Building drifts vs time history plots

Page 8: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

Figure 7. Interstory drifts with limit states

Energy dissipation

Analyses have shown the inelastic energy dissipation is ranging between 20-50% of total energy

dissipation for different records. It is also noted that vast amount of the ductility in the system is

provided with shear walls where inelastic energy dissipation of these components is larger than

coupling beam energy dissipation of the building. Figure 8 shows the representation of energy

dissipation for each record.

Figure 8. Energy dissipation chart for the building

Strain Gauges

Axial strain outputs on the structural wall elements have shown areas where compressive strains

have exceeded crushing strain in level 1 and below. These numerous wall locations are depicted

in figure 9.

Page 9: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

Figure 9. Zones of concrete crushing at Level 1 per analysis Shear Wall Behavior along Gridline Q

The structural wall located along gridline Q, which has observed concrete crushing and rebar

yielding at wall base, has been further investigated to provide an answer to identifying damage.

Shear demands on the wall were compared with capacity per information provided on as-built

drawings. The results shown in Figure 10 states that shear demands on the lower levels of the

wall close to the podium exceed capacity.

Figure 10. Shear demand versus capacity on wall at Gridline Q

Although the analysis showed deficiency in the shear behavior, axial strains measured from the

Page 10: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

analysis model state minor yielding of steel reinforcement with strains lower than concrete

crushing strain at reverse cycles. To further the investigation on the capacity of the gross

structural wall section, PM capacity of the wall at first subterranean level is calculated. Figure 11

shows some critical data points on the PM curve of the wall. The capacity is exceed by nearly

5% on the first subterranean level whereas the demands on the lower level do not exceed

capacity. Considering the use of concrete crushing strain at 0.003 in conventional designs and to

form this PM curve, 5% overcapacity would not generate local crush of this wall.

Figure 11. PM curve of wall at crushing strain 0.003 on Gridline Q subterranean level 1

Comparison of Model Results with Observed Damages

As mentioned earlier in the paper, damages occurring at basement levels were compared with

strain gauge information provided by the building model. Although actual damages were mostly

concentrated along the transverse axis of the building at first subterranean level, analysis results

showed that concrete crushing occurs at several locations, at which concrete damage was not

observed during the earthquake. Another contradictory outcome of the analysis was that there

was no shear failure observed on the building, whereas one of the damaged walls that is assessed

more in detail have shown that the shear capacity of the section is not adequate. This outcome

raises the question of the impact of shear on initiating the flexural failure considering they are

not independent behaviors.

Conclusions

The analysis results presented in this paper have shown that the building dissipates hysteretic

energy to the range of 50%, with crushing of concrete in several areas. Although there were

various damages predicted by the analysis, locations of these damages did not coincide with the

observed damage locations. It is also seen that damages have been highly varying in between

different seismic hazards measured in discrete stations.

Considering the damages occurring in subterranean levels, further investigations need to be

conducted on the effect of soil on basement levels. More accurate results may be obtained with

better representation of soil structure interaction after conducting series of sensitivity studies on

Page 11: COMPARISON OF INDUSTRY-STANDARD NONLINEAR DYNAMIC …

the bathtub analogy of the subterranean soil. Further refinement of seismic hazard based on

consideration of attenuation relationships and frequency content of the ground motion can help

to better predictions on the damage. This paper did not include any attenuation relationships

during the analysis which might have led to conservatism in the approach. The lack of a

connection between shear and flexure failure mechanism is also providing a setback for more

realistic predictions. Shear-flexure interaction of the failure mechanism will further need to be

investigated.

This study will be succeeded by comparing damages in several other benchmark buildings to

provide a stock of data while taking into account attenuation relationships, shear flexure

interaction effect and soil structure interaction.

References 1. NEHRP Consultants Joint Venture. Comparison of U.S. AND Chilean Building Code Requirements and Seismic

Design Practice 1985-2010. NIST GCR 12-917-18, 2012.

2. Wallace JW, Massone LM, Bonelli P, Dragovich J, Lagos R, Lüders C, Moehle J. Damage and Implications for

Seismic Design of RC Structural Wall Buildings. Earthquake Spectra Volume 28 No. S1, 2012.

3. Boroschek LR, Contreras V, Kwak DY, Stewart JP. Strong Ground Motion Attributes of the 2010 Mw 8.8

Maule, Chile, Earthquake. Earthquake Spectra Volume 28 No. S1, 2012.

4. CSI Perform-3D V5-0. Computer and Structures, Inc. Berkeley, CA,2011.

5. U.S. Geological Survey (USGS),2010. USGS Earthquake hazards program: Magnitude 8.8 Offshore Maule,

Chile, available at http://neic.usgs.gov/neis/eq_depot/2010/eq_100227_tfan/neic_tfan_w.html.

6. Idiem, Inspeccion Post Sismo Del 27 de Febrero de 2010 Edificios Sol Oriente I y II. Chile, 2010

7. Orakcal K, Wallace JW and Conte JP, Flexural Modeling of Reinforced Concrete Walls – Model Attributes.

ACI Structural Journal Title No. 101-S68, 2004

8. Massone LM, Bonelli P, Lagos R, Lüders C, Moehle J, Wallace JW. Seismic Design and Construction

Practices for RC Structural Wall Buildings. Earthquake Spectra Volume 28 No. S1, 2012.

9. Civil Engineering Department of University of Chile, Chile, 2013. Earthhquakes of Chile, available at

mhttp://terremotos.ing.uchile.cl