You are on page 1of 3

BOOK REVIEW

Displacement-Based Seismic Design of Structures


M. J. N. Priestley, G. M. Calvi, and M. J. Kowalsky 2007. IUSS Press, Pavia, Italy, 721
pp., $198.69

Reviewed by Graham H. Powell,a) M.MEERI


It is rare for a book on structural engineering design to be revolutionary, but I believe
this is such a book. If you are involved in any way with seismic resistant structural design, this should be on your bookshelf, and you should read at least the first three chapters.
Displacement-based design is not new. The concept was introduced years ago, most
notably by Paulay and Priestley in New Zealand, combined with the concept of capacity
design. The term displacement-based design is rather generic. The main thrust of this
book is actually Direct Displacement-Based Design (DDBD), where direct means that
a final design is obtained with little or no iteration. The book covers DDBD in depth and
detail, and it includes worked examples for a variety of building and bridge structures.
If you are not already familiar with displacement-based design, the first chapter will
give you a lot to think about. The authors make a persuasive case that there are serious
flaws in the conventional force-based method for seismic design. Using simple examples, they show that the strength of a structure is a relatively unimportant parameter.
They also show that the force-based procedure (using R factors) is often illogical, and
generally provides an inconsistent level of protection against damage. They proceed, in
the rest of the book, to turn the force-based design process on its head, emphasizing the
importance of displacement and ductility rather than force and strength.
The second and third chapters do a thorough job of outlining the design process. In
my opinion, Chapter 3, which describes the overall approach, gets bogged down in too
many details, so it can be heavy going, but it is well worth the effort. The rest of the
book (all 700 pages of it) contains a huge amount of information and presents a variety
of worked examples in considerable detail. That part of the book is quite daunting, and
you are likely to study it in detail only after you decide to apply DDBD to an actual
project.
DDBD is very different from conventional force-based design. For force-based design, the steps are essentially as follows: (1) estimate member sizes and stiffnesses and
the structure mass (2) calculate the vibration periods, (3) using an acceleration response
spectrum and the vibration periods, calculate spectral accelerations and hence lateral
loads, (4) for combined gravity and lateral loads, calculate strength demands on the
structural members, (5) design the members to satisfy these demands, and finally (6) if
the member sizes or stiffnesses are significantly different from the estimates, iterate
from Step (1). In a steel frame, member stiffness depends on member size, so if the sizes
chosen in Step (5) are different from those estimated in Step (1), iteration is needed.
Hence, this is not strictly direct design. For a reinforced concrete structure, the mema)

Emeritus Professor of Civil Engineering, University of California at Berkeley

555

Earthquake Spectra, Volume 24, No. 2, pages 555557, May 2008; 2008, Earthquake Engineering Research Institute

556

GRAHAM POWELL

ber stiffness is usually based on the gross concrete section (more on this below), and in
Step (5) only the reinforcement may be designed, with no changes in the estimated
member sizes. In this case iteration is not needed, and this is direct design.
For DDBD, the steps are essentially as follows: (1) using capacity design principles,
choose the components in the structure that are allowed to yield (essentially, choose a
plastic mechanism for the structure, for example a strong-column-weak-beam mechanism for a frame), (2) choose the relative strengths of the yielding components (for example, the relative strengths of the beams over the frame height), (3) choose a design
displacement (lateral drift) for the structure, (4) estimate the member sizes and calculate
the structure mass, (5) estimate the yield displacement (see below), (6) given the design
displacement and the yield displacement, get the displacement ductility, (7) given the
displacement ductility, and knowing something about the nonlinear behavior of the
structure, estimate the effective damping ratio for the structure (as the amount of yield
increases, the amount of dissipated energy increases, and hence also the effective damping ratio), (8) using a displacement response spectrum, the damping ratio and the design
displacement, get an effective period of vibration (note that this is the reverse of the way
that an acceleration spectrum is used in force-based design), (9) given this period and
the structure mass, calculate the required effective stiffness for the structure, using the
well known equation T = 2M / K, (10) multiply the effective stiffness by the design
displacement to get the base shear, (11) given the base shear, use the plastic mechanism,
the relative strengths of the yielding components, and equilibrium to calculate the actual
strengths of the yielding components, and finally (12) using capacity design principles,
calculate the strengths of the non-yielding components. It may be necessary to iterate
from Step (4), or to change the choices in Steps (1) through (3), but usually the design
will be close to a final design.
Step (5), which requires estimating the yield displacement, is a key step. An important assumption for DDBD is that member stiffness is proportional to member strength.
This is essentially true for steel beams and columns. As the book shows, it is also true
for reinforced concrete frames and walls. (This means that the commonly used method
of basing stiffness on the gross concrete section is inaccurate, since this assumes that the
stiffness is independent of the amount of reinforcement, and hence independent of the
strength.) If member stiffness is proportional to strength, it follows that the yield displacement can be estimated quite accurately knowing the yield strain for steel and the
member dimensions. As a consequence, a displacement-based design can be obtained
with little or no iteration, and DDBD works for both steel and concrete structures. It is
interesting to note that this is the opposite of force-based design. When force-based design is used, iteration may not be needed for concrete structures because stiffness is assumed (incorrectly) to be independent of strength. In displacement-based design, if stiffness is (or is assumed to be) independent of strength, the yield displacement depends on
the strength, the estimation of yield displacement at Step (5) is more of a guess, and
iteration is needed to get the member strengths.
The above summary of the DDBD steps leaves out a huge number of details, which
is one reason why the book is 700 pages long. The summary shows, however, that the
process is very different from force-based design, in concept and in detail.
The end result of the design process can be a final design. For a large structure, how-

BOOK REVIEW: DISPLACEMENT-BASED SEISMIC DESIGN OF STRUCTURES

557

ever, I suspect that the designer will often use this as an initial design, to be refined using
nonlinear dynamic analysis. An initial design can, of course, be obtained by other methods, including force-based design (as is often done in current practice). The advantage of
using DDBD is that the initial design is likely to be very close to a final design, and any
refinements are likely to be minor. A force-based initial design may be very different
from a final design, and is likely to require much more effort to refine.
The book is full of points that are interesting and sometimes counterintuitive. One
such point concerns the distribution of strength among the members of a structure.
When force-based design is used, this distribution is determined by the relative stiffnesses of the members. For example, if a concrete structure has two shear walls with
equal thicknesses but different widths, and if bending controls, the stiffness of a wall
based on the gross concrete section is proportional to the cube of its width. Hence, if the
widths of the two walls are, say, in the ratio 2:1, the bending strengths from force-based
design will be in the ratio 8:1. In displacement-based design, the designer is, in principle, free to choose any distribution of strength between the two walls. The book shows
that the best performance is obtained if the strength distribution is 4:1 (based on the
square of the wall width). One of the examples in Chapter 1 of the book explains why
this is the case.
If I have a criticism of the book, it is that the design procedure is based on hand
calculation, and it is not easy to adapt it for a computer. Before the concepts in the book
can be applied widely and to complex structures, I think it will be necessary to implement them in computer programs. It will be interesting to see how long this takes to
happen.
This book is not the easiest read, but the authors have done an outstanding job of
covering the topic. Given the limitations of force-based design, and given recent improvements in nonlinear analysis capabilities, it is inevitable that designers will move
toward a displacement-based approach. I expect this book to have a major impact on
design practice.
(Received 1 May 2008; accepted 1 May 2008

You might also like