Q&A with Bruce Ellingwood about ASCE 7-02, Section 2.0 - Combinations of Loads
1. How were the strength design combinations developed?
‘The first limit states (or "strength") design approach in the US appeared as an appendix in
the 1956 edition of the concrete standard, ACI 318. In the 1963 edition, the approach was
moved to the main body of ACI 318. By the late 1960s the steel industry had decided to
develop a limit states design methodology for steel that was termed load and resistance
factor design or LRFD (strength, limit states design and LRFD all refer to the same basic
concept). The first version of LRED for steel in the 1970's used different load
combinations than those used in ACI 318. One of the grand men of structural
engineering, Professor George Winter from Comell University, suggested that we
provide the engineering community with a more unified approach. We needed the same
loading combinations for all construction materials, and the logical place to put them was.
the American National Standard A58, "Minimum Design Loads for Buildings and Other
Structures.”
At the time, I was with the National Bureau of Standards and managing the secretariat of
the AS8 committce. In this role, I assumed the responsibility of developing a set of load
combinations for strength design with the assistance of three academic colleagues,
advised by a committee of professional engineers
In the 1970s, the structural engineering community had begun to look toward the field of
structural reliability for tools to improve design, and a lot of work in probabilistic load
modeling had been completed recently. My colleagues and I utilized all of those tools and
techniques, and in the summer of 1979 (a relatively-short period of time considering the
job at hand) the four of us came up with a set of probability-based load combinations
‘The AS8 committee implemented these load combinations by ballot in the 1982 edition
of ANSI AS8.1, Section 2. In the intervening 20 years the load combinations have not
changed much, with a few exceptions.
ASCE took over responsibility for the AS8 standard from ANSI/NBS in 1985. Today,
most of the material specifications have keyed their strength provisions to the load
‘combinations in ASCE Standard 7 so that the engineer only needs to use one set of
combinations for most materials,
Much of steel design still is done using the working stress design method. Supplement 1
to AISC’s 1989 ASD manual (dated Dec. 17, 2001) refers the engineer to ASCE 7's
working stress load combinations. The first steel LRFD specification was published in
1986, and it was in essence 17 years in birth. LRFD adopted the ASCE 7 loading
combinations right off the bat. ACI 318 didn't adopt them until quite some time later.
2. Section 2.3.1 (the introduction for the strength design combinations) states the
following: "Section 2.3.2 shall be used only in those cases in which they [ASCE 7
strength loading combinations] are specifically authorized by the applicable material
design standard."Can you explain the need for this statement?
When determining any safety factors used in design, engineers must first look to the
building code adopted by their local authority for guidance. Material specifications such
as ACI-318 and standards such as ASCE 7 only become law when the building code
adopts them. If the building code or referenced material specifications don't specify the
combinations directly, many will stipulate ASCE 7 loading combinations.
When we published the 1982 edition of ANSI AS8.1 (the predecessor of ASCE 7), we
included this statement because we didn't want engineers to use our combinations without
also considering the resistance side of the picture in checking safety. By putting that
statement in, we tried to make it clear that we had made certain assumptions concerning
the performance of the structural materials and elements against which the combinations
were keyed. We had to do this because the loading factors given in ASCE 7 and the
resistance factors (given in the material specifications) can't be completely uncoupled in
structural reliability analysis and performance assessment. The resistance factors
provided by ACI 318 at the time were keyed to the combinations given in ACI 318, not,
those in ANSI A58.1-1982. This coupling is not apparent in the code language but is
explained in the commentary of ASCE 7-02.
Can you give some examples of how this is applicable to some of the more frequently-
used standards today such as ACI 318 and AISC’s steel design manual?
Today, there isn't as much of a need for this statement because most of the specifications
have keyed into the ASCE 7 requirements. Both the IBC and NFPA 5000 have cross-
referenced ASCE 7 for load combinations. In addition, ACI 318-02, all editions of the
AISC LRFD manuals, and the most recent AF&PA/ASCE 16 are all keyed to the strength
loading combinations of ASCE 7. The AISC ASD manual (with Supplement 1)
references the allowable load combinations of ASCE 7.
Regardless of which standard the engineer chooses, it is important to use the same
standard throughout the structure. Some designers may be tempted to cherry-pick the
portions of each standard that benefit them the most economically and end up with
something that isn't consistent with the intent of the specifications committee. When a
conversion is done from one specification to another, not everything will transfer over
100 percent congruently
3. For non-buildings such as pressure vessels, how should engineers use ASCE 7?
Pressure vessels are an ASME type of structure, not an ASCE type of structure. The
ASCE 7 standard is not intended to be applicable to pressure vessels; it applies to
buildings and some non-building structures such as lattice towers, signs and chimneys. It
doesn't cover bridges (which are covered by the AASHTO provisions), and it doesn't
cover pressure vessels (which are covered under the ASME boiler and pressure vessel
code). Although ASCE 7 is not intended to be used by those types of structures overall, it
does contain specific provisions such as pressure coefficients for wind loads on tanks(Section 6) and force coefficients for non-structural components and equipment (Section
9). But the load combinations for wind or equipment were not intended to be used with
ASME structures.
4, Section 2.4.1 explains that "Increases in allowable stress shall not be used with the
loads or load combinations given in this standard unless it can be demonstrated that such
an inerease is justified by structural behavior caused by rate or duration of load.”
Why was this done?
Years ago, ASCE 7 permitted working stress load combinations to be multiplied by a
factor of 0.75 when they combined two transient loads, such as live load with wind or
earthquake loads. This so-called "load combination factor" accounted for the low
probability that the building will sce the design live load and wind and/or earthquake load
at the same time.
What was wrong with that? First of all it put a factor of 0.75 on the dead load as well,
which does not vary with time. This makes no sense at all. More importantly, the ASD
edition of the steel specification and some codes at the time permitted engineers to
increase stresses by 4/3 for any combination that included wind or earthquake forces, not
just those that had two transient loads. This led some engineers to believe that the
allowable stress increase did not have the same intent as the 0.75 factor. The fact that
0.75 is the reciprocal of 4/3 added further confusion. Some used the 0.75 factor and the
4/3 stress increase at the same time, thereby accounting for the same phenomenon twice.
What you end up with is essentially no safety factor against wind. This was a major
concern. In the time that I managed the Secretariat of AS8 and in the succeeding years
with ASCE 7, I've received more inquiries on this one issue - whether or how to adjust
the ASD load combinations in this situation - than on all other issues combined.
In ASCE 7-02, the engineer has 3 ASD combinations: 1) D+L, 2) D+W, and 3)
D+.75(L+W), and if the engineer uses them, than he/she is warned not to take the
allowable stress increase except if it was intended for something other than a load
combination effect.
Can you give examples of increases in allowable stress that are permitted per ASCE 7
and those that aren't?
‘The strength of most materials exhibits a rate or duration-of-load effect to different
degrees. For example, under dynamic loads, steel exhibits a modest increase in strength
But wood exhibits a very pronounced rate or duration load effect, to an extent that is not
apparent in the other structural materials. The strength of a wood structural element
loaded to failure in approximately 3 to 5 minutes in a traditional ASTM strength-of-
materials test is substantially different from the strength of the same element that has a
sustained load for 20 years. The NDS has recognized this so-called duration-of-load
effect for many years. The increase in allowable stress in the NDS for combinations
involving short-duration loads (e.g., wind) is not related to load combinations, and so hasa different purpose than the increase of 4/3 in allowable stress that has been historically
used with wind in steel ASD.
5. What does the Wind Directionality Factor (Kd) account for and why can it only be
used with the load combinations in ASCE 7?
‘The wind pressure coefficients and the wind speeds that are given in Section 6 of ASCE 7
are maximum values that are independent of direction. This is significant for two reasons.
First of all, each building component sees its maximum pressure at different orientations
of the building with respect to incident wind. For example, one building orientation may
produce the maximum pressure at the eves of the roof while another orientation produces
‘the maximum pressure at the comers of the walls. As a result, the tabulated wind pressure
coefficients are enveloped values.
Secondly, itis not likely that the typical building will be oriented in such a way that the
maximum wind speed will act in the direction that is most unfavorable for the building,
In some locations of the country, the extreme winds exhibit a strong preferential
direction. A classic example of this is the area on the east slope of the Rockies, where
local orographic features cause a strong directionality effect. In other parts of the country,
there may not be any particularly-strong directional effect.
‘The design wind force can be reduced by the wind directionality factor to account for
these directionality effects. The wind directionality factor was introduced in the 1998
edition for the first time, but the directionality effects were accounted for in other ways
before that, The Kd factor for a rectangular building was built into the probability-based
load combinations that were developed in 1982. In the intervening years more
information became available from wind tunnel testing, and in 1998 we decided to
separate it out and create a table of Kd that is in the 1998 and 2002 editions. At the same
time we increased the wind load factor in Section 2.3 from 1.3 to 1.6. We took 1.3 and
divided by 0.85 (Kd for a rectangular building) which gave us 1.54, and we rounded to
16.
6 Can Kd be used with allowable stress design as well as strength design?
Yes. The Kd factor can be used with the strength design combinations and the allowable
stress combinations.
7. In allowable stress design, 0.6D is used in loading combinations 7 and 8. This change
was first made in the 1998 version of ASCE 7. Why was this change made?
‘The reduced factor on the dead load for the D+W combination in ASD is needed to
account for stability concerns. The 1995 edition of ASCE 7 included a section on
overturning and sliding (Section 2.4.4) that states "Buildings and other structures shall be
designed so that the overturning moment due to lateral forces acting singly or in
combination does not exceed two-thirds of the dead load stabilizing moment . .." Inaddition it uses the load combination D+-W with no load factors. The 1998 edition
climinated the overturning and sliding section and uses an allowable load combination of
0.6D+W instead. The dead load is needed to counteract the overturning forces of the
wind. The 0.6 is the safety factor against overturning and sliding.
We arrived at the 0.6 dead from calibrating the working stress requirements for
counteracting load cases to the probability-based strength design requirements. This is
one area where we knew that working stress design has been unconservative (or unsafe).
To encourage the adoption of the more rational limit states design method, we made a
decision a number of years ago to not do any significant maintenance to the working
stress combinations except in situations where there is evidence that the working stress
combinations are leading to unsafe conditions. In the situation where you have
counteracting loads, there was a demonstrative safety issue involved.
‘There was a spectacular collapse of a cooling tower in the UK in the 1960s. The
commission of inquiry on the Ferrybridge Cooling Towers concluded that the collapse
‘was due to a failure to consider counteracting load effects.
8. What are the differences between ASCE 7-98 and ASCE 7-02? And why were those
changes made?
1) We decided to spell out the load combinations for working stress design rather than
provide them in paragraph form in order to make it more clear what the load combination
requirements are. (Section 2.4.1)
2) We added atmospheric ice loads, as requested by the telecommunication and
transmission industries. Towers and conductors can have significant wind and ice loads,
and we decided to provide guidance.
3) We changed how the live load factors are presented in Section 2.3.2 for combinations
3, 4, and 5. The most conservative companion action factors appear in the combinations,
The less conservative companion action factors appear as exceptions. (The 1998 edition
did the opposite.) This was done to facilitate code implementation, and does not represent
a change in theory.
9. What changes or additions to the load combination Section do you see in the future?
I see three possible additions or changes to the load combination section of future
editions of the standard relative to extraordinary loads, performance-based design and
treatment of load combinations.
Extraordinary Loads
ASCE 7-02, Section 2.5 simply states "Where required . . . strength and stability shall be
checked to ensure that structures are capable of withstanding the effects of extraordinaryevents such as fires, explosions, and vehicular impact." Engineers must refer to the
‘commentary of Section 2.5 for the additional load combinations.
Itis likely that the next edition of ASCE 7 will have a more comprehensive treatment of
extraordinary loads such as blast and fire in the main body of Section two. This is needed
because the U.S. structural engineering community is developing a greater intere
enhanced design procedure for fire. This trend is motivated to a degree by the World
Trade Center collapse, but perhaps more by the desire to eliminate the heavy penalty that
certain types of building construction experience due to the current fire rating procedures
in the United States.
With advances in computation and structural design technology, it is now possible to
design structures for realistic fire exposures instead of qualifying structural components
to pass the ASTM E119 fire test. The ASTM E119 procedures represent early 20th
century technology that aren't based on a rational analysis, but rather on a set of idealized
conditions that don't model reality accurately. In addition, they are based on component
(rather than system) performance.
Load combinations during and after fire are significant, and our committee wants to
provide engineers with the guidance they need.
Does the Eurocode have requirements for fire?
Yes, the European community has a more mature attitude toward structural design for fire
conditions. The Eurocodes have a set of load combinations that include fire as an
accidental event. A structural engineer that is interested in designing for a realistic fire
exposure can do a thermo-structural analysis, extract an appropriate load combinations
from Section | of the Eurocode, and check whether the structure as a whole will
withstand an elevated temperature for an extended period of time.
‘The loading combinations in the commentary of ASCE 7-02 are very similar to those in
the Eurocode. [ am a member of the Joint Committee for Structural Safety and other
international standards organizations, and ASCE has a liaison to ISO (the international
equivalent of our ASTM), so ASCE 7 is aware of what the other codes are doing,
Performance-Based Engineering
Performance-based engineering allows engineers to tailor design to a number of different
performance levels rather than just life-safety performance, as currently required in all
US. codes. It is not a new concept, but it has developed a strong momentum in recent
years, primarily in the earthquake engineering and fire areas.
For some structures, it might be appropriate to design for a higher level of performance
under different natural or man-man hazards in addition to designing to the life-safety
event. For example, the specifications for a building might say that the building shall
ensure life safety under a 2,500-year mean-return-period event and shall be capable ofproviding continued service immediately following a 100-year event. If the owner
requires the engineer to design for two (or more) different objectives, the standard must
provide additional tools.
Owners have many reasons for paying extra for better performance. The recent
experiences on the West coast with building performance during earthquakes and
substantial economic losses indicate that many building owners and occupants,
particularly those that invest in high-end properties, expect more than life-safety. They
will demand continued performance under events that may be somewhat less than the
design-basis event, and they are willing to pay an additional fee in order to obtain that,
additional performance.
‘Treatment of Load Combinations
Currently, the load factor for earthquake is set to 1.0 because we map events with a
mean-retur-period of approximately 2,500 years directly. The wind and snow loads have
load factors of 1.6 because their maps traditionally have been based on a 50-year event
rather than extreme events. With load factors, it is difficult to have uniformity in risk
across geographic locations. For example, basic differences in risk due to wind
climatology in hurricane-prone and extra-tropical regions cannot be captured with such a
dull knife as a single load factor. As an alternative, we could map the wind and snow for
a design-basis event with a return period of 500 to 1,000 years, and set the load factors
equal to 1.0. This would be a more rational approach, but would require corresponding
changes to the wind and snow sections of the standard, Structural engineers tend to be a
conservative lot, and we must be sensitive to the number of changes we make to ASCE 7
ina given period of time.
Bio of Author:
Dr. Ellingwood received his undergraduate and graduate education at the University of Illinois at Urbana-
Champaign. From 1975-1986, he held the position of Research Structural Engineer at the National Bureau
of Standards (now the National Institute of Standards and Technology, and led the NBS Structural
Engineering group from 1982 through 1986. He joined the Johns Hopkins University in 1986, was
appointed Chairman of the Department of Civil Engineering in 1990 and was named the Willard and
Lillian Hackerman Chair in 1997. He has served on the faculty of the School of Civil and Environmental
Engineering at Georgia Tech since 2000, and was School Chair from 2000 to 2002. Dr. Ellingwood's
professional interests involve the application of methods of probability and statistics to structural
engineering, with emphasis on the analysis of structural loads, performance of structures under occupancy,
environmental and abnormal load conditions, and development of safety and serviceability criteria for
design. He administered the Secretariat of the American National Standard Committee AS8 (now ASCE
Standard Committee 7) on Design Loads from 1977 through 1984, and directed the development of the
probability-based load criteria for limit states design that appear in ASCE Standard 7 on Minimum Design
Loads, the AISC LRFD Specification, ASCE Standard 16 on LRFD for Engineered Wood Construction,
and ACI Standard 318. He is the author of over 250 publications and reports, and is research and
professional service have received numerous awards from ASCE, AISC and other professional
organizations. He was elected to the National Acedemy of Engineering in 2001