Comparison of Typical Bridge Columns Seismically Designed With and Without
Abutment Participation Using AASHTO Division I-A and
Proposed AASHTO LRFD Provisions
Richard V. Nutt and Ronald L. Mayes, Consultants
Background and Purpose
In support of the NCHRP 12-49 effort to develop the next generation of LRFD seismic design provisions for bridges, a parameter study was undertaken to determine the impact of the proposed provisions on the design of bridges in the United States. The proposed provisions are not simply a revision of the current Division I-A provisions of the AASHTO Standard Specification or a revision of the seismic provisions in the LRFD Specification. Accordingly, the seismic hazard mapping, the design spectrum, the load factors, understrength factors, and numerous other important design parameters have been changed.
Because the seismic design provisions have, essentially, been rewritten entirely, the purpose of the parameter study is to provide a comprehensive perspective on the changes inherent in the new provisions. The parameter study was divided into two phases, a global design and a detailed design phase. The purpose of this report is to discuss only the detailed design phase which was funded as part of MCEER's FHWA 094 project.
The uses of the parameter study have been two-fold: (1) to benchmark the new provisions' results against those of the existing provisions and (2) to ascertain the effects of key parameters and fine-tune them relative to good engineering practice.
Scope of the Detailed Design Study
The second phase of the parameter study was to develop detailed designs of the primary lateral load resisting elements for relatively simple bridge structures. In this context, 'simple' implies, primarily, that the structure can be considered a single-degree-of-freedom (SDOF) system. The results of these simple designs are likewise compared against the results that the Division I-A provisions would produce. The Division I-A specification was selected as the appropriate benchmark for the detailed design study because it was felt that most agencies are using that specification and not the LRFD specification. Therefore, the best benchmark was judged to be those provisions with which most bridge designers are familiar.
The objective of the detailed design study was to evaluate the results for a relatively small set of near-actual designs. The detailed designs give a more complete view of the resulting member designs and costs than do the more broad comparisons made in the global design phase of the parameter study, although the detailed designs are a relatively limited set of data points. For this reason, the two phases of the parameter study are complimentary.
At this time, the detailed design study includes comparisons for structures founded on reinforced concrete columns with simplified foundation designs for spread footings. Consideration of abutment participation in the longitudinal direction was a main focus of the study.
Currently, the comparisons are based on regular bridge configurations that do not include skew, curves, or other complexities.
Designs have been considered at discrete locations throughout the country. Five locations have been used, which represent a broad geographic range and a broad seismic hazard range. The sites considered are
· Seattle, Washington
· Portland, Oregon
· Memphis, Tennessee
· St. Louis, Missouri
· New York, New York
In addition, only Soil Site Class C has been investigated at this time. This is due to the fact that all the designs compared to date are founded on spread footings, and Type C soil was considered a typical soil on which spread footings would be used.
The 100- and 2500-year seismic coefficients and soil factors from the proposed provisions are included in Table 1. These parameters were used for both phases of the study, global and detailed. For comparison, the acceleration coefficients for the Division I-A provisions are included in Table 2. The site soil factor for Division I-A for all the designs and locations is 1.2.
Table 1. Design Spectral Accelerations and Site Class Coefficients (Soil Class C)
NCHRP 12-49 Seismic Provisions
Site 100-Year Earthquake 2500-Year Earthquake (50% Prob. of Exceed. in 75 Years) (3% Prob. of Exceed. in 75 Years) Ss Fa S1 Fv Ss Fa S1 Fv Seattle, WA 0.314 1.2 0.093 1.7 1.60 1.0 0.56 1.3 Portland, OR 0.176 1.2 0.054 1.7 1.05 1.0 0.35 1.46 Memphis, TN 0.067 1.2 0.015 1.7 1.35 1.0 0.41 1.38 St. Louis, MO 0.06 1.2 0.014 1.7 0.59 1.2 0.19 1.61 New York, NY 0.03 1.2 0.007 1.7 0.42 1.2 0.09 1.7
Table 2. Design Acceleration Coefficients and Soil Type Factors (Soil Type II)
AASHTO Division I-A
Site 500-Year Earthquake
(10% Prob. of Exceed. in 50 Years)
A S Seattle, WA 0.32 1.2 Portland, OR 0.16 1.2 Memphis, TN 0.20 1.2 St. Louis, MO 0.10 1.2 New York, NY 0.15 1.2
Organization of the Report
Chapter 2 includes the detailed design results and comparisons from the designs of simple SDOF bridge structures. The structures considered represent simple viaduct-type structures, in which the tributary mass of a bent is the same as the 'mass' that the bent supports. The comparisons cover both single- and multi-column bents and a range of substructure heights and weights. The different heights and weights are the primary means with which the period of the structures is varied. For the most part, the column size is constant at 4 feet in diameter, although 5- and 3-foot-diameter columns are considered for Seattle and New York, respectively.
Chapter 3 includes comparisons of longitudinal abutment influence in the seismic design of similar structures to those discussed in Chapter 2. For the most part, two-span bridges are considered in this part of the study, although one four-span structure is included for comparison. The two-span configuration represents one in which the abutments can play a major role in resisting the longitudinal loads induced in the bridge. The abutment role is varied by (1) altering the gap at the end of the structure, which varies from zero gap to infinity (or a large enough gap that no abutment resistance is mobilized), and (2) varying the longitudinal soil passive resistance between 3.5 ksf and 7.7 ksf.
Column Design Observations
Column Strength Control - Column strength, and thus the amount of longitudinal reinforcement, can be controlled by a number of factors in the proposed NCHRP 12-49 seismic design criteria. The column studies demonstrate some interesting trends that are worth noting and that may guide the ongoing development of the criteria.
The use of different "R" factors in the proposed criteria has a surprisingly small impact on a significant number of column designs. The reason for this is apparent when each of the controlling factors for column strength are separated as was done in the study. For Seattle as an example the benefits of an "R" value of 6 or 8 are never realized because the steel requirement for the 100-year earthquake is greater than those for the "R" factor reduced 2500-year earthquake. This is true for taller column heights as well, but as was shown in the study column design begins to be dominated by P-Delta requirements.
In less seismically active areas the effect of the 100-year earthquake is diminished, but minimum steel requirements or the P-Delta criteria usually govern column design rather than the "R" Factor reduced 2500 year forces.
The impact of the "R" factor appears to increase in the single column bent studies as reinforcement ratios are significantly less when an "R" factor of 8 was used. However the full benefit of using a higher "R" factor is never realized due to minimum reinforcement and 100 year earthquake requirements.
The following paragraphs briefly discuss factors affecting column strength.
"R" Factors - It appears that very little benefit can be gained by using high "R" factors. Other column design controls prevent the designer from taking full advantage of these high factors.
100 Year Earthquake - As illustrated in the Seattle multi-column study, the 100-year earthquake tends to be a major factor in column design in areas of high seismicity, particularly the Western United States. In these cases, it often controls over the design for the 2500-year earthquake. This may present a design problem in certain regions, such as parts of California, but it does not appear to prohibit reasonable column designs elsewhere in the United States. Care must be taken not to impose unusually strict requirements on column strength based on this level earthquake.
P-Delta Requirements - The ATC-32 requirements for P-Delta are based on avoiding dynamic instability that can result from a biased response of the column in one direction. This could result in a potential "ratcheting over" of the column in an earthquake of long duration. Because this requirement is so critical to column design and because it is based largely on empirical observations of analytical studies, P-Delta requirements must be carefully chosen.
Minimum Reinforcement Ratios - Currently, AASHTO Division I-A sets the minimum steel percentage at one percent. Even when this percentage is reduced to 0.8 percent, it often controls in this study. In New York, for example, the design of columns is based entirely on minimum steel requirements. Therefore, the setting of a minimum steel percentage is an important issue.
Column Performance - Column performance appears to be satisfactory in all cases studied. The estimated column plastic rotation never exceeded 0.035 for any of the NCHRP 12-49 designed columns. Further evaluation using non-linear pushover analysis is required to confirm this observation.
Column Size Effect - The proposed P-Delta requirements seem to place a reasonable control on column size. In the Seattle multi-column bent study high steel percentages in P-Delta columns suggested the need for larger column diameter. When column size was increased to 5 feet, the steel percentages were reduced to reasonable levels. Also, the results of the New York multi-column bent study may tempt a designer to reduce his column size. This would be prevented for heavily loaded tall columns where reinforcement ratios exceed a practical and ATC-32 codified limit of 4 percent.
Column Over-strength - Column over-strength moments have a major impact on footing design. Footing costs tend to be the major contributor to overall construction cost. Therefore, the use of accurate over-strength moments can have a major cost impact. Because the use of a simplified over-strength factor can often yield overly conservative results, it is recommended that the refined method of calculating over-strength moment be encouraged.
Cost Impact - Except for a few short, lightly loaded columns, the construction cost of the NCHRP 12-49 designed single column bents studied tended to be less than their AASHTO counterparts. This was due to a number of factors including lower "R" factors, gross section properties in the analysis, and lower capacity reduction factors used in the AASHTO Division I-A designs. This is shown in Figure 20 of the parameter study report.
In the case of NCHRP 12-49 designed multi-column bents, construction costs tended to be higher than AASHTO. This was due to the higher "R" factors used for AASHTO multi-column bents and the P-Delta requirements of the proposed NCHRP provisions. This is shown in Figure 21 of the parameter study report. When the P-Delta effects are mitigated by increasing the column size the construction costs are much more in line with AASHTO designs. This is illustrated in Figure 22.
Impacts of Abutments
The following observations were made regarding the results of the abutment parameter studies:
As noted below the anticipated displacements of the deck are an important issue with respect to abutment performance. Figure 3-7 provides a plot of the Spectral Displacements nationwide for a period of 1.0 second. Note that displacements at other periods can be obtained by multiplying the map displacements by the period T.
Longitudinal Displacement of Superstructure - The method used to model the abutment has a large impact on the calculated longitudinal displacements of the superstructure. These displacements are also sensitive to the relative stiffness of the interior support and the level of input ground motion. Figure 4.3 of the parameter study report shows the displacements for bridges with a multi-column bent located in Seattle in which the ultimate passive pressure was assumed to be 7.7 kips/sq.ft. It is apparent that the effect of abutment participation is quite dramatic, reducing the displacement by a factor from between approximately 2 and 6. Except for bridges 1 through 3, which have low dead loads, the amount of assumed expansion joint gap has no effect. This is because in all cases the displacement is large enough to be on the constant force portion of the assumed non-linear abutment response curve. It will be shown later that the ultimate passive pressure is the critical parameter rather than the expansion joint gap.
The impact of relative column stiffness is illustrated in Figure 4.4 of the parameter study report. In this case, all parameters are assumed to be the same as before except the columns are assumed fixed for moment at both ends, and thus the column stiffness is approximately 4 times greater than assumed in the previous example. In this case the relative effect of abutment participation on displacement is considerably less, varying from between 2 and 3 times less than without abutment participation. The effect of gapping on the lighter structures is also more pronounced.
When the level of seismic loading is reduced, as is the case of multi-column bridges in St. Louis, the relative impact of assumed gapping is more apparent. This can be observed in Figure 4.5 of the parameter study report. In some cases the 2" gap never closes and the response is the same as if no abutment participation was present. In almost all cases the displacement response is low enough that the assumed gap makes a difference. Close examination of the displacement results show that displacements sufficient to mobilize the ultimate passive pressure capacity are seldom achieved. This effect is more dramatic when stiffer single column bents are assumed as shown in Figure 4.6 of the parameter study report.
The effect of multiple interior supports is shown in Figure 4.7 of the parameter study report, which simulates the response of a four span bridge. A comparison of these results with those of the corresponding two span bridges shown in Figure 4.3 of the parameter study report shows that the unrestrained displacements of the two structure configurations are similar. However, abutment participation is not nearly as effective in reducing displacements in the multi-span bridge. Unrestrained displacements are typically between 2 to 3 times restrained displacements compared to 2 to 6 times for the two-span bridges. Because multi-span bridge displacements are higher than the corresponding two-span bridge in all cases where abutments are participating, the ultimate passive pressure is always mobilized, regardless of the gap size. Therefore, the assumed ultimate passive pressure capacity will be critical to overall displacement response.
The effect of the assumed ultimate passive pressure is shown in Figure 4.8 of the parameter study report for two-span, single column bent bridges located in Seattle. As would be expected, the higher the ultimate passive pressure, the lower the longitudinal displacement. The ratio of these displacements to those where no abutment participation is assumed is shown in Figure 4.9 of the parameter study report.
In locations of lower seismic loading, such as St. Louis, the passive pressure has pretty much the same effect as in areas of higher seismic activity, although the longitudinal displacements are much less as shown in Figure 4.10 of the parameter study report. The ratios of these displacements to the displacements calculated without abutment participation are shown in Figure 4.11. These ratios are only slightly less than those for Seattle, which were shown in Figure 4.9 of the parameter study report.
Column Design - Longitudinal displacements dictate the magnitude of column moments and drifts, and therefore effect the amount of reinforcing steel that is required. Also, the reduced displacements caused by abutment participation will mitigate P-Delta design requirements and thus effect the design of tall, heavily loaded columns. This is clearly shown in Figure 4.12 of the parameter study report for multi-column bent bridges in Seattle designed with an assumed passive pressure of 7.7 kips/sq.ft.. Notice that reinforcing ratios in many columns designed without abutment participation exceed the practical, and ATC-32 codified, limit of 4 percent, while columns designed with abutment participation require only the minimum amount of reinforcement. As shown in Figure 4.13 of the parameter study report, an increased "R" factor has only a marginal affect on the shorter and/or more lightly loaded columns. Therefore, it will either be necessary to consider some abutment participation in certain bridge configurations, or make the column dimensions larger.
P-Delta affects are generally not as critical in the single column bent bridges, which have columns that are approximately 4 times stiffer than the multi-column pinned top piers, on a column-by-column basis. This is illustrated in Figure 4.14 of the parameter study report. Such single column bents will attract a greater percentage of the load because of their increased stiffness, and thus in some cases these columns may require more reinforcement than corresponding multi-column bents when abutment participation is present. This is readily apparent for bridge number 7, which has short, heavily loaded columns. As can be seen in Figure 4.15 of the parameter study report, these columns can easily be designed in all cases if an "R" factor of 6 is used, even when a relatively low ultimate passive pressure of 3.5 kips/sq.ft. is available.
Cost Impact - It is interesting to compare the cost of a typical column/footing unit designed by the current Division I-A criteria with the cost of one designed by the proposed NCHRP 12-49 criteria using various "R" factors. For bridges with single column bents located in Seattle, P-Delta effects do not control design in the examples studied, and construction costs of NCHRP designs are generally equal to or less than comparable AASHTO designs. The one exception is for the short, lightly loaded columns of bridge number 1. This trend is illustrated in Figure 4.16 of the parameter study report and even this case is mitigated when an "R" factor of 6 is used as shown in Figure 4.17 of the parameter study report. Further reductions in the "R" factor do not effect this case due to requirements imposed by the 100-year earthquake.
A much different trend is observed for multi-column bents as illustrated in Figure 4.18 of the parameter study report. When abutment participation is ignored, the NCHRP construction costs are much higher than AASHTO. A higher "R" factor will only effect the shorter and/or lighter bridges. This is because of the P-Delta requirement that controls several of the NCHRP designs, and absence of a P-Delta requirement in AASHTO. When abutment participation is included, the costs of AASHTO and NCHRP designs are similar.
In lower seismic zones, such as New York, all designs are usually controlled by minimum column steel requirements and thus costs are similar. This is true even when abutment participation is ignored.