1275

SEISMIC EVALUATION AND RETROFIT OF INDUSTRIAL BUILDINGS

Mohamed DAALI1 And Otakar URBAN2

SUMMARY

Performance based design in its actual format is certainly a comprehensive tool for seismic risk minimisation.
What needs to be kept in mind however is that unless structural engineers work hand in hand with other
discipline engineers, contractors and manufacturers, the implementation of performance based design will be
hard. Also, the understanding of non-linear analytical methods, procedures and the inclusion of cumulative
damage methods is of prime importance for good implementation of performance based design.

The use of non-conventional methods, such as coupling building structures to non-building structures,
disconnecting structures that induce mass and/or stiffness eccentricity, the addition of supplemental damping
devices through the use of fluid dampers and fluid damper-springs to reduce the overall response and the use of
snubbers to eliminate pounding between the buildings results in efficient retrofit schemes.

INTRODUCTION

Considerable changes occurred in the seismic requirements of the National Building Codes since their earliest
editions. For instance, earthquake loads specified in the earliest editions of the National Building Code of
Canada [NBC, 1970] are considerably lower than the 1990 and 1995 editions. Sometimes the ratio of base shear
of the latest editions can be as high as twice that of the earliest editions. This often means that buildings built
back then had, in most instances, wind loads governing the design process. It should also be noted that industrial
buildings receive, in general, numerous structural modifications over the years. In some instances, it has been
observed that bracing members were removed to accommodate equipment. In other instances, the basic system
is modified to accommodate specific production requirements; example of this type of interventions is the
removal of a building column for a larger entrance. Other examples of important modifications show up in the
form of additional substantial mass being installed in the structures such as new conveyors, new bunkers, and
new heavy equipment etc. All the above leads to a modified structure with geometrical, stiffness and mass
distribution characteristics that are totally different from the structure originally designed.

Existing buildings have been shown to sustain significant over-stress, local, and global failures when subjected
to moderate and strong intensity earthquakes. The advent of recent earthquakes such as the 1989 Loma Prieta
(California), the 1994 Northridge (California) and the most recent 1995 Hyogo-ken Nanbu (Japan) earthquakes
showed that steel buildings are still vulnerable to moderate and strong earthquakes. The 1995 Hyogo-ken Nanbu
earthquake produced a large number of damaged steel buildings. In fact, 1247 steel buildings were reported to
be damaged, of which 286 had collapsed or were in danger of collapsing during aftershocks [Tremblay et al.,
1996].

PERFORMANCE LEVELS AND OBJECTIVES RELATED CONSIDERATIONS

1
PhD, Hatch Associates Ltd., Ontario, Canada; member of APEO -Association of Proffesional Engineers of Ontario.
2
MSc, Hatch Associates Ltd. Ontario, Canada; member of APEO - Association of Proffesional Engineers of Ontario.
Performance based design has been in use in various engineering disciplines and sectors such as manufacturing
and to a limited extent in structural engineering for many years now. In fact numerous design codes have used
and are still using one form or another of performance based design. Within the context of a limit state design,
two levels of designs exist: a) the first level consists of serviceability limit state, and b) the second level covers
the member strength limit state. The advent of recent strong earthquakes brought Performance Based Design to
the forefront. Performance Based Design and its performance levels and corresponding objectives are well
covered in documents such as ATC-40, FEMA-273 and Vision 2000 of the “Bleue Book”.

The following table shows performance levels and the corresponding basic objectives, essential/hazardous
objective and safety critical objective levels associated with performance based design:

Event Recurrence Probability of Fully Operational Life Safe Near

Frequent 43 years 50% in 30 years O P P P

Occasional 72 years 50% in 50 years Q O P P

Rare 475 years 10% in 50 years Y Q O P

Very rare 970 years 10% in 100 years Y Q O

Where:

O : is for basic objective

Q : is for Essential/Hazardous Objective

Y : is for Safety Critical Objective

From an industrial building owner’s point of view, the major objective is first to protect people within and
around the facility, and then protect the structure for various performance levels. This later objective in most
cases would span from the Fully Operational level to the Near Collapse level. This should be achievable when
planning for a new facility, and would only account for a small percentage of the capital cost. For an existing
structure, this line of thinking will in most cases not be achievable. Protecting against the full spectrum of
performance levels will either be too stringent, exhaustively expensive or not practical since it will hinder
production or operation in the facility. The project team should thus put together a list of priorities that could
include structural considerations as well as non-structural considerations. For instance, a chemical tank falling
and spilling its content might be more dangerous than the consequences of a collapse of a member, which has no
bearing on the overall structural system stability.

For life safety, the collapse performance level is evidently to be achieved at all cost. Also to save life, the
structural engineer should closely work with other engineers to ensure that services inside an industrial building
are designed and manufactured to behave adequately for a specified performance level. For instance, what could
be a tolerable drift from a structural point of view, could also be an intolerable movement for a Carbon
Monoxide pipe that could crack and lead to potentially disastrous consequences. A minor settlement that is
acceptable from a structural point of view may not be acceptable to an in-ground water pipe that is cooling a
furnace.

Low cycle fatigue due to cumulative damage and its consequent cracks in connections of moment resisting
frames could have equal disastrous consequences from a strong motion earthquake compared to consequences
from a repeated number of moderate motion earthquakes. For this reason, models for cumulative damage
assessment [Daali & Korol, 1995, and Daali, 1995] should be part of non-linear static analysis and non-linear
dynamic time-history analyses.

2 1275
 STATIC AND DYNAMIC ANALYSES

Often times, the evaluation and assessment of existing industrial building starts with a visual study of drawings
or the structure itself followed with rule-of-thumb rough checks. The basic objective in this phase is to identify
the system weaknesses such as lack of secondary load paths, lack of effective systems (bracing) that could
transfer loads to the foundations, distribution of mass and stiffness.

In consulting offices, it is accepted practice to design new structures and evaluate/assess existing structures using
two and three-dimensional equivalent static analyses. The analysis uses code specified lateral loads applied to a
structure modelled with an elastic linear stiffness. The results of this linear elastic static analysis will be internal
forces and displacements whose magnitudes are approximately those of the design earthquake. However, it
should be mentioned that the nature of industrial type of structures presents plane and vertical irregularities that
makes the use of the static analysis inappropriate in most cases.

For the reasons above mentioned linear time history dynamic analysis and linear response spectrum analysis
found their way in design offices. To use these two methods, the structure is also modelled with an elastic linear
stiffness. The advantage of these methods of analyses over the static linear analysis resides in a better
distribution of forces that takes into account system irregularities and mass distribution. But the shortcomings
reside in the fact that material and geometrical non-linear effects and cumulative damage are not included.

Since the advent of performance based design, non-linear static (or pushover) analysis and dynamic analysis
became the tools of choice for analysis and evaluation for the identification of areas where there is a lack of
stiffness and ductility, or weaknesses in the member or connection. The FEMA-273 and ATC-40 are two
documents where modelling techniques, guidelines for use of ground histories and acceptance criteria can be
found. The static non-linear procedure is used on structural models with non-linear moment-rotation (or force-
deformation) relationships. Gravity loads are first applied, then the lateral loads are progressively increased and
a load-deformation envelope for the whole structure is obtained. The non-linear time history dynamic analysis is
similar to the linear time history dynamic analysis except that non-linear force-deformation relationship are
included in the modelling and that cumulative damage models are also ideally included.

STATE-OF-THE-ART

The relationship governing the conservation of energy [Uang and Bertero, 1988] dictates that the energy input
resulting from a strong motion earthquake is absorbed through the sum of the recoverable elastic energy, the
kinetic energy, the damping energy and the irrecoverable hysteretic energy. If a structure does not have the
ability to dissipate energy through hysteretic energy, then it should either have sufficient strength to resist the
earthquake forces or have adequate damping to dissipate the input energy

Supplemental damping can be introduced in a structure by the use of mechanical devises. The latter dissipate
energy through friction, yielding of specific components, motion of a piston into a fluid or a material which is
similar or by viscoelastic action. To achieve this, numerous devices can be found in the market. In fact as early
as the 1890’s fluid dampers have been used to attenuate recoil in artillery guns [see Taylor and Constantinou,
1994]. Other examples of dampers can be found in the friction-damper device proposed by Pall and March in
1982, the Taylor fluid dampers which have been in use since 1955, in commercial and aerospace/defense
projects and the Jarret viscoelastic elastomers dampers.

RETROFIT SCHEMES

3 1275
The preliminary assessment and evaluation of existing buildings dating from 30 years and older often reveals
that the lateral resistance of the structural system is only a small portion of the code base shear loads.
Furthermore, considerable irregularities would also exist in mass and stiffness distribution.

To enhance the lateral resistance of an existing industrial building, very often it will be a negotiating process
whereabouts the structural engineer proposes retrofit schemes and receives feedback from the plant management
and production people. The process is not easy and would require a lot of imagination.

For instance a continuous multi-bay crane runway could be rigidly linked, tied with dampers or snubbers to the
main structure. Two parts of the building separated by an expansion joint could also be linked together with
snubbers and dampers strategically placed to force the parts of the building to move in-phase and thus avoid out-
of-phase motion.

Among other methods of retrofit for existing industrial buildings could be the addition of external structures that
are in line or enveloping the existing ones. Also, a modification of shear connections into moment resisting
connections in girder to column connections would help in securing alternative load paths. The addition of
bracing coupled with friction or fluid dampers would definitely be of great help in dampening the earthquake
motion.

For industrial buildings where a section of the building induces high eccentricity in mass and rigidity, it might be
beneficial to disconnect or isolate the later from the former. Particular attention should be paid to the retrofit of
structures built of heavy clay brick masonry walls with a soft and weak first story. At first glance, the use of stiff
bracing seems to be a suitable solution to brace the first story bays. This, however, would create a strong first
story susceptible to transfer higher forces to the higher story, eventually damaging the clay brick masonry walls.
Although rods and flat bars tension-only bracing are known to exhibit a rather poor behaviour in older buildings,
tension-only bracing if properly detailed can behave in a satisfactory manner. One such example of such
behaviour is the tension-only bracing used in a building, which was subjected to the recent Hyogo-Ken Namibia
earthquake [Tremble et al. 1996].

CASE STUDY

When the original smelter building in province of Quebec, Canada was built in 1952, it accommodated five
furnaces and was 420 feet (128 m) long. As the requirements for steel production grew, the building was
gradually extended so today it houses nine electric arc furnaces and is about 1000 feet (305 m) long. The
building is 110 feet (34 m) wide and 70 feet (21 m) high and is separated in longitudinal direction by a 2-inch (5-
cm) expansion joint. Crane aisle on West Side and furnace aisle on the East Side run for the whole length of the
building. Furnace aisle accommodates the furnaces, electrode floor, and materials handling, such as conveyor
galleries and bunkers. Lateral loads are resisted by moment resisting frames in the transverse direction and by a
braced system in the longitudinal direction.

The crane runway consists, longitudinally, of two continuous moment resisting frames, separated by a two-inch
(50-cm) expansion joint. The crane runway is attached to the main structure transversally above the height of the
crane columns by a series of shear links.

Attached to the building along the long side are transformer houses constructed of a steel skeleton frame infilled
with concrete masonry blocks in the first story and a one-foot thick clay brick masonry walls in the stories
above.

Static and three dimensional linear dynamic analyses were first used to evaluate the resistance of the smelter
building when subjected to seismic lateral loads determined per the Canadian Building Code [NBC 1995]. This
was followed with non-linear push over and times history analyses. The evaluation of the building concluded
that the building could resist the NBC seismic loads in the transverse direction but was significantly weak in the
longitudinal direction. Only 36% of the seismic loads could be resisted in the longitudinal direction. Several
weaknesses that could possibly lead to a partial or total collapse of the building were identified. The most
important among these were the lack of secondary load paths, the lack of ductility, lack of effective systems
(bracing) that could transfer loads to the foundations and finally various local weaknesses such as absence of
moment connections.

4 1275
An independent reviewer specialising in the field of earthquake engineering first reviewed the studies performed
by the consulting team. Then, within the context of performance based design approach, design levels and
performance objectives were targeted and numerous retrofit schemes were reviewed for implementation.
Finally, the upgrading option that was selected included the following items:

- Addition of a new tower coupled with the original building,

- Disconnection of the transformer houses from the smelter building,

- Modification of shear connections into moment connections,

- Connecting the independent crane runway framing system to building column lines, and

- Addition of vertical bracing.

To further enhance the response of the smelter building under earthquake forces, it was decided to introduce
supplemental damping devices coupled with bracing system in the critical areas of the building. To eliminate
“pounding” of the north and south parts of the smelter building, snubbers and damper units were placed into the
building expansion joints. The latter were subjected to a rigorous testing program (Fig. 1) in the manufacturing
plant before being shipped and installed in the building.

Fig. 1 Testing of a damper for a smelter building.

REFERENCES

5 1275
ATC (1996), “Seismic Evaluation and retrofit of Concrete buildings”, Volume 1, ATC-40 Report, Applied
Technology Council, Redwood City, California.

Daali, M.L. and Korol, R.M. (1995), “Low Cycle Damage Assessment in Steel Beams”, Structural Engineering
and Mechanics, An International Journal, Vol. 3, Number 4, pp. 341-358.

Daali, M.L. (1995), “Damage Assessment in Steel Structures”, 7th Canadian Conference on Earthquake
Engineering, Montreal, Canada.

FEMA (1997), “NEHERP Guidelines for the Seismic Rehabilitation of Buildings”, Building Seismic Safety
Council for the Federal Emergency Management Agency, Report No. FEMA 273, Washington, D.C.

NBCC (1990), National Building Code of Canada, National Research Council of Canada.

Pall A.S. and March C. (1982), “Response of Friction Damped Braced Frames.”, J. Struct. Engrg., ASCE, 108
(6), 1313-1323.

SEAOC (1997) “Vision 2000 in Blue Book”, Recommended Lateral Force Requirements and Commentary,
Published by Structural Engineers Association of California.

Taylor D. and Constantinou M.C. (1994), “Test Methodology and Procedures for Fluid Viscous Dampers Used
in Structures to Dissipate Seismic Energy”, Technical Report.

Tremblay R. et al. (1996), “Seismic Design of Steel Buildings: Lessons from the 1995 Hyogo-Ken Nambu
Earthquake.”, Canadian Journal of Civil Engineering, 23:727-756.

Uang C.M. and Bertero V.V. (1988), “Use of Energy as a Design Criterion in Earthquake-Resistant Design.”,
Report No. UCB/EERC-88/18, University of California, Berkeley.

6 1275