Prepared by
Page 1
Disclaimer
A Simple Guide to the Direct Analysis Method. How to implement the Direct Analysis Method using Modern Software. Although CSC Inc. takes great care to ensure that any data, information, advice or recommendations it may give either in this publication or elsewhere are accurate, no liability or responsibility of any kind, including liability for negligence, howsoever and from whatsoever cause arising from or related to their use, is accepted by CSC Inc., its servants or agents.
Page 2
Figure 1: Complex buildings with sloping members, hard-to-define floors, and nonorthogonal framing are commonplace today. [Screenshot from Fastrak Building Designer. Model courtesy of Fisher Engineering.]
Page 3
SUMMARY OF THE FULL PAPER FOLLOWS. TO JUMP TO THE FULL PAPER BEGINNING ON PAGE 8 CLICK HERE
In the 2005 AISC Specification for Structural Steel Buildings, AISC introduced significant changes to the stability design requirements. Clear requirements for stability design and three straight-forward methods of satisfying those requirements have been introduced. One of these, the direct analysis method, is the premier method and the focus of this paper. The direct analysis method is the most advantageous because it is not limited in its application, applies to all buildings, and is the most general and accurate approach provided. AISC STABILITY ANALYSIS AND DESIGN REQUIREMENTS The design of all structures for stability must consider: 1. flexural, shear, and axial deformations of members 2. all component and connection deformation that contribute to the lateral displacement of the structure 3. second-order effects (both P- and P- ) 4. geometric imperfections 5. member stiffness reductions due to residual stresses The first three requirements are covered by the structural analysis and the first two are included in nearly all analysis packages. As noted above, three methods to account for #4 and #5 (including provisions relating to #3) are presented in the specification. The Direct Analysis Method is the preferred method. DIRECT ANALYSIS METHOD The direct analysis method is found in Appendix 7 of the 2005 Specification, but will move to Chapter C as the default method in the 2010 AISC Specification. The basics of the direct analysis method include the following: Second-Order analysis: A second-order analysis which considers both P- and Peffects is required. This can be accomplished through a rigorous second-order analysis or by using the approximate method presented in the specification. Initial imperfections: The effects of initial imperfections of the structure geometry are considered either by applying notional loads or directly modeling the imperfections in the geometry of the model. Inelasticity: The axial and flexural stiffnesses of members that contribute to the stability of the structure are required to be reduced to account for inelastic behavior in the members. Effective Length Factor: One major benefit of the direct analysis method is that K=1.0! No more calculating K when determining the nominal strength of columns.
Page 4
HOW TO: SECOND ORDER ANALYSIS Loads applied to a structure will act on the real shape of the structure, including initial imperfections and deflections caused by lateral loads. This introduces flexure into axially loaded members, where the magnitude of the flexure is dependent on the axial load (theP) and the deformed shape (the or ). This causes second-order (nonlinear) behavior known as the P-Delta effect. An analysis considering this effect is required. The second-order analysis can either be a rigorous analysis or the approximate amplified first-order method presented in the specification. The amplified first-order method certainly works well for some buildings, generally those that are simple, regular structures. However, to allow for the greatest range of structure geometry with the highest degree of accuracy in determining internal forces, the best course of action is to take advantage of a rigorous second-order analysis. It would be difficult to apply the approximate methods to the structure in Figure 2. Even smaller projects, with relatively small budgets, often have complex framing requirements that will benefit from a rigorous second-order analysis.
Figure 2: Complex buildings make it difficult to apply approximate second- order analysis methods. Today, even relatively small projects often have complex framing. [Screenshot from Fastrak Building Designer. Model courtesy of Scott Wilson.]
HOW TO: NOTIONAL LOADS To account for the effect of initial imperfections, the specification allows the designer to directly model the imperfections within the model. The set of initial displacements modeled need to consider displacements due to loading and anticipated buckling modes of the structure. Modeling these displacements can be complicated, therefore the alternative of applying notional loads is often best.
Page 5
Notional loads are horizontal forces added to the structure to account for the effects of geometric imperfections. Figure 3 Figure 3 below illustrates a simple version of the concept. The notional load (Ni) is 0.2% of the total factored gravity load (Yi) and applied at each level.
The value Yi is the total factored gravity load in each load combination on each level. Therefore, the value of the notional loads will change from combination to combination. AISC requires that the notional loads be distributed in plan on each level in the same manner as the gravity load. This would indicate you could apply a single notional load (at each level for each load combination) at the location of the resultant of the total factored gravity loads. Depending on how the loads are distributed on the structure, calculating the location of the resultant can become complicated. The following suggested method is more straightforward and easy to automate within computer software. Calculate the total factored gravity load transferred to each column at each level. Calculate the corresponding magnitude of the notional load for each appropriate load combination. Apply the notional load as a point load at each column at each level. The point loads represent the notional load effect on each column. By applying the loads to all columns and including all framing in the model, the effect of leaning columns is directly included as part of the analysis. This method correctly distributes the notional loads in the same manner as the total factored gravity load and through the 3D analysis allows the effects of those loads to be distributed accurately to the lateral framing. AISC requires the notional loads be applied in the direction that adds to the destabilizing effects. For gravity load combinations you will generally have to apply the notional loads in two orthogonal directions in both the positive and negative sense. You can apply the notional loads in the same direction on all levels. For lateral load combinations, the notional loads are applied in the same direction as the resultant of all lateral loads.
Page 6
HOW TO: STIFFNESS REDUCTION To account for inelastic behavior, a reduction in the axial and flexural stiffness of all members that contribute to the lateral stability of the structure is required. This includes members that deliver the lateral loads to the vertical lateral framing such as collectors, chords, and horizontal bracing. In addition, it is permitted to apply the reduction to all members in the model. Reducing only the lateral members can create artificial distortion of the structure and result in unintended redistribution of forces. The stiffness reduction should also be applied to members of other materials, unless a greater reduction is required by governing codes and specifications. This also will help prevent an unintended redistribution for forces to the members of other materials, since they had not been reduced. The requirements of the direct analysis method are intended for determination of required strengths and they are not intended for serviceability checks. Therefore, when checking the lateral drift of the structure under wind loads, for example, the stiffness reductions need not be applied! CONCLUSIONS In the 2010 AISC Specification the direct analysis method will become the default method and move from Appendix 7 to Chapter C within the body of the code itself. It is clear that if you are not familiar with the features of this method at present you will need to be so in the future. Hopefully, this paper will elucidate these in the context of the use of modern structural engineering software. However, there are a number of questions for which you should seek answers. There are several options for how the direct analysis method is executed within a software package, for example, the type of second-order analysis. What effects on the results does the particular method create (if any)? Is the method limited in its application? If the approximate (B1,B2) method is used, how are the limitations overcome? It is important to know the answers to these questions before you begin designing with the software. I think the overriding question to answer is, Do any of the answers to these questions result in a limit on the types of buildings you can design using the software package? Understanding the requirements of the direct analysis method and how the tools you use implement those methods is key to successful design.
Page 7
A Simple Guide to the Direct Analysis Method How to implement the Direct Analysis Method using modern software
INTRODUCTION In the 2005 AISC Specification for Structural Steel Buildings, AISC introduced significant changes to the stability design requirements. Clear requirements for stability design and three straight-forward methods of satisfying those requirements have been introduced. One of these, the direct analysis method, is the premier method and the focus of this paper. The direct analysis method is the most advantageous because it is not limited in its application, applies to all buildings, and is the most general and accurate approach provided. The direct analysis method includes requirements for a second-order analysis, application of notional loads, and member stiffness reductions. Use of the direct analysis method means several things to design engineers. K = 1.0 Straight-forward design requirements, including how to perform the structural analysis. Improved results with less potential for error, especially when modern software is used. To take advantage of these benefits, decisions must be made about how to implement the method. For example, the most appropriate second-order analysis for the project must be selected. Approximate methods are adequate in many cases. However, consistently using a more general approach will allow the same method to be used for all projects for consistently accurate results. Other details such as how to apply notional loads to the model regarding magnitude, direction, and distribution and to which members to apply a stiffness reduction are also important decisions. The requirements of the specification are summarized in the first two sections of this paper. A how to guide for implementing the direct analysis method forms the remainder. The discussion will provide information vital to making the proper decisions while implementing the direct analysis method to get the best possible results. The information will allow you to refine your current design process and help you evaluate what your software tools are doing for you.
Page 8
1. flexural, shear, and axial deformations of members 2. all component and connection deformation that contribute to the lateral displacement of the structure 3. second-order effects (both P- and P- ) 4. geometric imperfections 5. member stiffness reductions due to inelasticity The first three requirements are covered by structural analysis (typically done within a software package). The first two items are included in nearly all structural analysis packages. There are many types of second-order analyses (item #3) including both direct and approximate methods. The method used varies with the software package. Three methods to account for #4 and #5 (including provisions relating to #3) are presented in the specification. The effective length method and the first-order analysis method are limited to structures with low to moderate second-order effects. The third method, the direct analysis method, is not limited and applies to all steel structures. In fact, it is the required method for structures with large second-order effects. The direct analysis method is described below. The other limited approaches will not be discussed in this paper.
Page 9
already been accounted for in the method. It should be noted that a K value less than one can be used if it is justified by analysis. The following sections of the paper present more details on the requirements and suggestions for implementing them within your design process.
Page 10
SECOND-ORDER WITH THE DIRECT ANALYSIS METHOD In addition to the general requirement of a second-order analysis, the direct analysis method has several more specific requirements. For example, there is a difference in how the analysis is executed based on the design method being used. LRFD: The analysis is performed using LRFD load combinations. ASD: The analysis is performed using 1.6 times the ASD load combinations and the results of the analysis are then divided by 1.6 to get required strengths. Second-order effects on the structure are non-linear and the analysis must be carried out under strength-level loading. This means the analysis must be carried out using load combinations. Load cases cannot be analyzed separately then superimposed in a load combination. The results are not the same. The LRFD load combinations are at strength level. However, when ASD is used, the 1.6 factor is required to get strength-level load combinations (or an adequate approximation) to use in the analysis. In Figure Figure 5 below, the curve represents the non-linear building response. In Step 1, the ASD load combination applied loads, wASD, are multiplied by 1.6. In step 2 the resulting model is analyzed with the amplified loads to yield a set of forces on the members, 1.6RASD. This set of forces is then divided by 1.6 in Step 3 to yield the required strengths, RASD. It is clear that the required strengths, RASD, are greater than the set of forces that results from the analysis under ASD load combinations directly, rASD.
Page 11
Step 1
Step 3
1.6RASD
Figure 5: ASD Load Combination amplification
Required Strength
Note that the approximate B1, B2 method of second-order already includes the 1.6 factor. AMPLIFIED FIRST-ORDER ANALYSIS You can either perform a direct second-order analysis or use the approximate amplified firstorder method presented in the AISC specification (commonly known as the B1-B2 method). The details of applying this method are not presented in this paper. The method has many valid applications and has been used successfully for years, typically for regular buildings with simple lateral framing. However, there are some limitations of the method worth pointing out. Different factors are applied to moments due to lateral translation of the structure (B2) and moments with the structure restrained against translation (B1). For a structure with regular gravity and lateral framing and uniform loading, these effects might be easy to separate (gravity causes no translation and lateral loads cause translation), but for real structures, the separation is not quite so straight-forward. For example, nonsymmetric gravity loading can cause translation within the structure. In cases like this and many others, engineering judgment is required to correctly assign moments to the two categories. The B2 factors can vary at a joint (one value from the story above and one from below) therefore the distribution of moments to the members at that joint can easily become complex and require engineering judgment. Structures with complex geometry can make calculating the factors difficult. Structures with complex lateral framing (such as non-orthogonal frames or a combination of braced frames and moment frames), sloping beams and columns, or
Page 12
where floor levels are not readily identifiable can make the method complex and tedious if not inaccurate. It is recommended in the AISC Commentary to use a more rigorous approach when B1 > 1.2. Therefore, the application of this method may be limited. It should be noted that a value of 1.2 is a significant amplification. B2 is calculated for the entire story. If the structure has a localized area of relative instability, this method could potentially miss this behavior. Given these limits, the range of structures where the approximate method yields accurate results is limited. For example, how does one calculate B1 and B2 for the project illustrated below?
Figure 2: Screenshot from Fastrak Buildings Designer. Model Courtesy of Scott Wilson
Engineering judgment is often required for this method and therefore, generally speaking, more suited to hand calculations since engineering judgment is difficult to program into a software package. If the tools are available, the more general (and often more accurate) rigorous analysis has several benefits. GENERAL SECOND-ORDER ANALYSIS Sophisticated analysis tools with rigorous second-order analysis methods are now readily available to design engineers, making practical the use of the more accurate analyses. A direct second-order analysis frees the engineer of the limitations of the approximate method
Page 13
and often gives a more accurate determination of internal forces and strength level deformations. A rigorous analysis is based on the actual stiffness of the framing and therefore any localized instabilities and the corresponding effects on the surrounding framing will be captured. An engineer should have a thorough understanding of any analysis tool that he uses. The more sophisticated analyses often require a slightly more sophisticated model. More thought is often required while constructing the model to ensure an accurate analysis and provide stability throughout the second-order analysis.
Page 14
MAGNITUDE The notional load, Ni, is 0.2% of the total factored gravity load at each level, Yi, and can be written as: Ni = 0.002Yi Note that the value 0.002 is equal to 1/500 which is the maximum tolerance for out-ofplumbness in steel structures as indicated in the AISC Code of Standard Practice (see Figure 6). Note that a smaller value can be used if the actual out-of-plumbness of the structure is known.
Gravity loads are defined in the AISC code as Load such as that produced by dead and live loads, acting in the downward direction. The value Yi is the total factored gravity load in each load combination. Therefore, the value of the notional loads will change from combination to combination. Also, the total factored gravity load includes all gravity loads on a level, not just those loads that are vertically supported by lateral framing members. Therefore the effect of leaning columns is included in the notional loads. When using ASD with a direct second-order analysis, the total gravity load (and therefore the notional loads) is multiplied by 1.6. This is the same factor discussed earlier that is required for the second-order analysis with ASD. The loads are not multiplied by 1.6 twice.
Page 15
WHEN TO APPLY Notional loads are required to be added to all load combinations. For combinations including lateral loads, the notional loads are added to the other lateral loads. However, when the ratio of second-order deflection to the first-order deflection ( 2/ 1) is less than 1.51, the notional loads only need to be added to gravity only combinations. DISTRIBUTION Now that you have the magnitude of the notional loads for each level and load combinations, you have to decide where on plan to apply them to the structure. AISC requires that they be distributed on each level in the same manner as the gravity load. This would indicate you could apply a single notional load (at each level for each load combination) at the location of the resultant of the total factored gravity loads, otherwise known as the center of gravity. The direction of the loads will be discussed in the next section. If the gravity loads are distributed uniformly (in plan) on the level, it is easy to calculate where the notional load should be placed. Assuming the lateral framing is symmetric, it is easy to determine what portion of the notional loads each frame resists. See Figure 7 below.
Figure 7: Notional Load for Uniform Loading and Symmetric Lateral Framing
As areas with different gravity loads are introduced to the structure (such as a heavy storage area) the calculation of the center of gravity becomes more complicated. You would more likely add a notional point load at the center of gravity of each loading area. If there are only a few loading conditions and your lateral framing is still simple, the distribution of the notional loads to the frame is still manageable. See Figure Figure 8 below.
The limit of 1.5 applies to structures that have been analyzed using the nominal (unreduced) stiffness of lateral members. If reduced stiffness is used, the limit is 1.7.
Page 16
As soon as a bit more complexity is added to the lateral framing, the distribution of the notional loads to the framing becomes more complicated. See Figure 9 below.
In a real building there are multiple loads on the levels and usually unsymmetrical framing. This makes placing and distributing the notional loads more and more complex. See Figure Figure 10.
Page 17
Figure 10: 'Real' Building with complex loading and Lateral Framing
These more complex situations obviously lend themselves to computer analysis. The following suggested method works well with and without computer software, assuming you are able to accurately distribute the notional loads to the lateral framing by hand. Calculate the total factored gravity load transferred to each column at each level. Calculate the corresponding magnitude of the notional load for each appropriate load combination. Apply the notional load as a point load at each column at each level. See Figure for a sample set of notional loads. Include the set of notional loads in each appropriate load combination. Perform a 3D analysis of the structure for the load combinations. Consider all framing, both lateral and gravity, in the analysis. The point loads represent the notional load effect on each column. By applying the loads to all columns and including all framing in the model, the effect of leaning columns is directly included as part of the analysis. This method correctly distributes the notional loads in the same manner as the total factored gravity load and through the 3D analysis allows the effects of those loads to be distributed accurately to the lateral framing. This
Page 18
method can be easily automated and included as part of the structural analysis.
DIRECTION Now we have a good method for calculating and applying notional loads in the proper location in the model. The next step is to determine in which direction or directions to apply them. The requirement in the AISC Specification is to apply the notional loads in the direction that adds to the destabilizing effects. For gravity load combinations you will generally not know which direction is critical and different directions will be critical for different members within the framing. Therefore you will have to apply the notional loads in two orthogonal directions in both the positive and negative sense. (See Figure Figure 12) You can apply the notional loads in the same direction on all levels. Therefore, you will now have four load combinations for each gravity load combination each with the notional loads applied in a different direction.
Page 19
For lateral load combinations (when notional loads are required) the notional loads are applied in the direction that adds to the destabilizing effects. This would be in the same direction as the resultant of all lateral loads. So if you are adding notional loads to a wind load case where the wind loads are applied in the +X direction, the notional loads would also be placed in that direction. If you have a combination of X and Y wind loads (as is often required by ASCE 7), then you would apply the notional loads in the resultant direction of the wind loads. See Figure 1Figure 3 for an illustration of simple examples.
Page 20
The flexural stiffness is multiplied by a factor, b (less than or equal to one), which is dependent on the value of the axial load in the member. To avoid the iteration required to calculate this factor, the specification allows the designer to use a value of b = 1.0, if an additional set of notional loads equal to Ni = 0.001Yi are included. In addition, analyses including other materials become more straight-forward since b does not need to be calculated for those members. MEMBERS TO REDUCE Members that contribute to the lateral stability of the structure must have the stiffness reduction applied. This includes members that deliver the lateral loads to the vertical lateral framing such as collectors, chords, and horizontal bracing. In addition, it is permitted to apply the reduction to all members in the model. Reducing only the lateral members can create artificial distortion of the structure and result in unintended redistribution of forces. For example, Figure 1Figure 4 shows the undeformed shape of the sample framing. The sample consists of a moment frame on the right side with a leaning column on the left. If only the lateral members are reduced, the relative shortening of the moment frame columns under vertical loads will be greater than that of the gravity column. The (exaggerated) deformed shape shown in Figure 1Figure 5 will result. This could cause vertical load to shed to the gravity column, leaving the axial load in the lateral framing underestimated. Reducing all members in the model will keep the relative deflections closer and the redistribution of forces to a minimum.
Page 21
The stiffness reduction should also be applied to members of other materials, unless a greater reduction is required by governing codes and specifications. This also will help prevent an unintended redistribution for forces to the members of other materials, since they had not been reduced. SERVICEABILITY The requirements of the direct analysis method are intended for determination of required strengths and they are not intended for serviceability checks. Therefore when checking the lateral drift of the structure under wind loads, for example, the stiffness reductions need not be applied! Similarly, the effects of notional loads need not be included in drift checks either.
Page 22
As discussed above, b requires iteration to determine its value. b applies to members whose flexural stiffness contributes to the stability of the structure and has significant compressive load. Based on these criteria, there are several cases where you can reasonable set b = 1.0 as a starting point for the iteration process. Beams in moment frames since they generally have low axial load or braced frames since the flexural stiffness does not contribute to the overall stability. Braces since the flexural stiffness does not contribute to the overall stability. Columns in braced frames since the flexural stiffness does not contribute to the overall stability. By taking the slightly conservative approach of assuming b is 1.0 and increasing the notional loads as described above, even these decision points can be avoided.
CONCLUSIONS
The current AISC Specification (2005) includes the direct analysis method. The next AISC specification comes out in 2010, most likely late in the summer. One significant change is that the direct analysis method will become the default method and move from Appendix 7 to Chapter C within the body of the code itself. It is clear that if you are not familiar with the features of this method at present you will need to be so in the future. Hopefully, this paper has elucidated these in the context of the use of modern structural engineering software. However, there might need to be a number of questions for which you should seek answers. There are several options for how the direct analysis method is executed within a software package. An example would be selecting the type of second-order analysis that will be used. What effects on the results does the particular method create (if any)? Is the method limited in its application? If the approximate (B1, B2) method is used, how are the limitations overcome? Does the second-order analysis include both P- and P- ? It is important to know the answers to these questions before you begin designing with the software. Other details you should understand are: Are the notional loads calculated automatically and if so how are they applied and in what direction? What members have a stiffness reduction and what effect does this have on the distribution of forces in your model?
Page 23
I think the overriding question to answer is, Do any of the answers to these questions result in a limit on the types of buildings you can design using the software package? Understanding the requirements of the direct analysis method and how the tools you use implement this method is key to the whole process but I dont have to tell you that.
Page 24
Page 25