PBE + Pelicun – FEMA Example B: Why Are Only 5 Columns Assigned per Story in the Component Assignment?

[aminreza70]

Hi,

I am working with the FEMA Example B that comes preloaded in the PBE application integrated with Pelicun.
In the structural component assignment section, I noticed that instead of defining 20 columns per story, only 5 columns have been assigned by the developers for this example.

Could you please clarify why the developers chose to assign only 5 columns per story instead of 20 in this FEMA Example B?
Is this for simplification, modeling assumptions, or another reason?

Thank you very much for your help!

[zsarnoczay]

Hi Aminreza,

Thank you for reaching out. There is indeed an issue with the number of beam-column connections in this problem.

Note that the example in PBE assigns 5 columns per story per direction. That is, a total of 10 columns per story. This is in line with the numbers listed in Table 1-1 in the BD-3.7.15 background report of FEMA P-58. I used those numbers to faithfully reproduce the example problem in the report.

Evaluating and updating numbers from the example will improve its quality, but defeat its purpose as a source of validation through comparison with the results in the BD-3.7.15 report. Having said that, the BD-3.7.15 report is based on the first edition of FEMA P-58 and an old PACT version, which contained several calculation-related bugs that were discovered and fixed after the report was completed. This already makes it difficult to do a proper comparison with Pelicun for validation purposes.

I am inclined to put a comment in the example description about this issue and adjust the example to use the correct (10 per story per direction) component numbers.

Please let me know your thoughts; what would serve you best as a user?

[aminreza70]

Hi zsarnoczay,

Thank you very much for your detailed and helpful response.

In my opinion, using the actual number of columns in the example would be more beneficial for the further development and improvement of the PBE framework. In my own analysis, I assumed 10 columns per story in each direction (a total of 20 columns per story), and I compared the resulting changes using an Excel file.

I also have a question regarding the appropriate way to define the number of columns in PBE based on the actual floor plan of the building I am currently evaluating. The building is located in Italy and was constructed in 1982. It is a reinforced concrete moment-resisting frame structure in both directions, and the engineering demand parameters (EDPs) are available.

Given this setup, how should the number of columns be properly defined in PBE?
For instance, in my case, there are 6 columns in the X direction and 12 columns in the Y direction on each floor. Should I input this directly as 6 and 12 columns per story per direction? Or should I consider the total number of beam-column connections per floor in one direction, e.g., 18? Or is there another recommended approach for defining these quantities?

Furthermore, if we assume that the interior columns in the plan—similar to the FEMA example—do not contribute to lateral load resistance during an earthquake, how should the number of columns per direction be defined in such a case?
For example, in that case, would it be appropriate to consider 6 columns in the X direction and 8 columns in the Y direction?

Additionally, I have another question regarding the FEMA example where the Engineering Demand Parameters (EDPs) are available. Given that EDPs are provided, why do modifications in the FEM and EVT tabs lead to changes in the final results? In projects where EDPs are available, how should the settings in these tabs be configured to ensure the results remain stable and unaffected by such changes?

I would greatly appreciate your guidance on this matter.

Best regards,
Aminreza

[zsarnoczay]

Hi Aminreza,

Thanks for the feedback, I agree with your suggestion and will update the example to use 10 columns per story per direction.

Regarding your Excel file, the 2.2% reduction in the probability of irreparable damage is surprising. The number of columns should not have a direct influence on that probability. Such a large variation between runs often suggests that your sample size might be too small. I recommend investigating this to ensure you have robust results that don't change so much between calculations.

---

Defining the Number of Columns in PBE

For your concrete frame building in Italy, here’s the recommended approach to defining the number of components.

First, I recommend making sure the vulnerability model for ACI 318 SMF (Special Moment Frame) connections provides a realistic model for your building, especially since it was built in 1982. Reviewing the FEMA P-58 background documentation can help you confirm if these components are appropriate for your case. Note that there are other concrete beam-column components in FEMA P-58 that might work better for your case. You can see a complete list here: DLML docs

After confirming the components are suitable, the key is to focus on the number of beam-column connections per floor, per direction.

• It's About Connections, Not Columns: You should count the number of connections in a given direction that can be modeled with the ACI 318 SMF components. The number you input will depend on the design details and how the beams frame into each column.
• Simplifying Component Types: While FEMA P-58 provides several components for various connection sizes (e.g., 'beam on one side' vs. 'beam on both sides'), you'll find that many of them share the exact same fragility functions. In practice, it's often sufficient to classify your connections into just two main categories: small connections (e.g., 24x24) and large connections (e.g., 24x36 or larger). You can then assign the quantity of each type accordingly.
• Handling Non-Lateral Columns: You asked what to do if interior columns don't contribute to lateral resistance. In that scenario, you would simply exclude their connections from your count. For your example, if only the perimeter frames resist lateral loads, you would count the moment-resisting connections along those frames in the X and Y directions.

---

FEM and EVT Panel Settings

You also asked why modifications in the FEM and EVT panels change the final results when EDPs are already provided.

The short answer is: they shouldn't.

When you import EDPs from an external calculation, the PBE tool is designed to ignore all inputs from the panels prior to the Damage and Loss (DL) model. If you're certain that making changes in the FEM or EVT panels is affecting your results, please provide more information about your workflow, and I’ll be happy to investigate it.

Adam

[aminreza70]

Dear Dr. Zsarnóczay,

I sincerely appreciate the time you took to answer my questions. I am truly grateful for your continued support.

Regarding the change in the probability of irreparable damage, following your advice, I increased the number of samples and re-ran the analysis. As a result, the value changed to 0.77%.

In terms of column definition, I had previously reviewed the column types based on the link you had kindly shared. After revisiting the information in light of your recent explanation, I believe that the most appropriate component ID for this Italian building project is B.10.41.091a/b.

Considering your note about the importance of the number of beam-to-column connections and the fact that all columns participate in the seismic response, I defined the columns as follows: instead of modeling 18 columns, I modeled 36, in order to explicitly define each connection.

Alternatively, one could model only 18 columns and assign the connections selectively to some of them while omitting their counterparts. However, based on your explanation, I believe the first method is more accurate and appropriate.

Additionally, I have a kind request. Since I have already studied and implemented the FEMA example in the software, I would be truly grateful if you could share another example that was conducted using similar software. Ideally, the example would include EDP values and be published in a peer-reviewed paper, so that I can examine validated results. If you consider it appropriate, this example could also be added to the PBE analysis set. This would allow me to compare the results with credible data and gain a deeper understanding. In any case, I will gladly follow whatever approach you recommend and am fully at your disposal.

Once again, thank you very much for your guidance and support.

Sincerely,
Aminreza

[aminreza70]

Dear Professor Zsarnóczay, I just wanted to kindly follow up on my last comment from about two weeks ago. I would greatly appreciate it if you could take a look when you have a chance. Thank you very much for your time.

…

On Wed, Jul 30, 2025 at 3:33 AM Adam Zsarnoczay ***@***.***> wrote: *zsarnoczay* left a comment (NHERI-SimCenter/PBE#72) <#72 (comment)> Hi Aminreza, Thanks for the feedback, I agree with your suggestion and will update the example to use 10 columns per story per direction. Regarding your Excel file, the 2.2% reduction in the probability of irreparable damage is surprising. The number of columns should not have a direct influence on that probability. Such a large variation between runs often suggests that your sample size might be too small. I recommend investigating this to ensure you have robust results that don't change so much between calculations. ------------------------------ *Defining the Number of Columns in PBE* For your concrete frame building in Italy, here’s the recommended approach to defining the number of components. First, I recommend making sure the vulnerability model for *ACI 318 SMF (Special Moment Frame) connections* provides a realistic model for your building, especially since it was built in 1982. Reviewing the FEMA P-58 background documentation can help you confirm if these components are appropriate for your case. Note that there are other concrete beam-column components in FEMA P-58 that might work better for your case. You can see a complete list here: DLML docs <https://nheri-simcenter.github.io/DamageAndLossModelLibrary/dl_doc/damage/seismic/building/component/FEMA%20P-58%202nd%20Edition/B/B.10/B.10.41/index.html> After confirming the components are suitable, the key is to focus on the number of *beam-column connections per floor, per direction*. - *It's About Connections, Not Columns:* You should count the number of connections in a given direction that can be modeled with the ACI 318 SMF components. The number you input will depend on the design details and how the beams frame into each column. - *Simplifying Component Types:* While FEMA P-58 provides several components for various connection sizes (e.g., 'beam on one side' vs. 'beam on both sides'), you'll find that many of them share the *exact same fragility functions*. In practice, it's often sufficient to classify your connections into just two main categories: small connections (e.g., 24x24) and large connections (e.g., 24x36 or larger). You can then assign the quantity of each type accordingly. - *Handling Non-Lateral Columns:* You asked what to do if interior columns don't contribute to lateral resistance. In that scenario, you would simply *exclude their connections* from your count. For your example, if only the perimeter frames resist lateral loads, you would count the moment-resisting connections along those frames in the X and Y directions. ------------------------------ *FEM and EVT Panel Settings* You also asked why modifications in the *FEM* and *EVT* panels change the final results when EDPs are already provided. The short answer is: *they shouldn't*. When you import EDPs from an external calculation, the PBE tool is designed to *ignore all inputs from the panels prior to the Damage and Loss (DL) model*. If you're certain that making changes in the FEM or EVT panels is affecting your results, please provide more information about your workflow, and I’ll be happy to investigate it. Adam — Reply to this email directly, view it on GitHub <#72 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BH36D2PAR34B6NMXST3H2OL3LADV3AVCNFSM6AAAAACCASEPLWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTCMZUGQZDSMBSHA> . You are receiving this because you commented.Message ID: ***@***.***>