Design of Fungible (Modular) Structures for Data Centers

To view the figures and tables associated with this article, please refer to the flipbook above.

Despite being one of the oldest engineering professions, structural engineering continues to evolve rapidly, much like the technology it supports. Just as Non-Fungible Tokens (NFTs) surged in popularity in 2020 alongside cryptocurrencies, today’s engineers face the challenge of designing adaptable structures in a world where speed and flexibility are paramount. Four years later, the NFT craze has waned, reminding us how quickly trends can shift in the tech industry.
This rapid pace highlights the pressing need for structural solutions that not only support swift deployment but also ensure mission-critical operations through adequate redundancy. To meet these demands, tech companies require robust hardware and infrastructure. Speed-to-market is essential; however, supply chain issues often hinder progress, with lead times for various equipment enclosures stretching to a year or more.

As structural engineers, we tend to view each project as unique, each requiring a bespoke design. Building authorities mandate that a licensed structural engineer design each structure. Tools like the ASCE Hazard Tool calculate environmental loads (wind, seismic, ice, etc.) based on precise latitude and longitude, which complicates the idea of fungible design—where structures are treated as productized solutions.

Fungible design is most commonly seen in generator enclosures, where generator suppliers create standardized containers around their products, allowing for reuse across multiple geographies. However, compliance with local authority requirements is still necessary. Structural design adjustments must be coordinated with various pipes, cables, and vents running through the enclosure. In this scenario, conservatively sizing structural members can streamline the design process and minimize coordination challenges with other engineering disciplines.

The most critical design decision that needs to be made early in the design process, is determination of “Risk Category” for the data center facility. The 2024 International Building Code (IBC) defines this in Table 1604.5, generically represented in Table 1.

The term “hazard to human life” helps us understand that the intent of the building code is first and foremost, life-safety. Although this concept works well in most cases, some facilities have significant consequential losses. Data center facilities are prime examples of this, where consequential losses are much more relevant than life-safety.

AISC’s Engineering Journal recently issued a journal entry titled “Construction Cost Premiums for Risk Category IV SMF Buildings”. The study examined an example building located in Los Angeles and found that a Risk Category IV building is between 6 to 16% more expensive compared to a Risk Category II building. The case studies in this journal focused on the building’s stiffness.

ASCE 7-22 Minimum Design Loads and Associated Criteria for Buildings and Other Structures has added a new Chapter 32 requiring Risk Category III and IV buildings located in tornado-prone regions to be designed to resist tornado loads. Tornado loads may sometimes necessitate weights to be added to the roof structure to act as ballasts against the uplift from tornado loads. The fact that Risk Category III buildings will now be required to be designed for tornado loads may prompt some owners to reconsider the building’s Risk Category selection.

ASCE 7-22 Section 13.1.3 and Table 13.1-1 consider Risk Category when determining requirements for anchorage of non-structural components. The cost premium referenced in the previous paragraph does not account for the additional reinforcements associated with non-structural components.

Since selection of Risk Category affects several other design elements, some Engineers of Record would show two columns: “AHJ Minimum Requirements” and “Owner’s Requirements,” in either the specifications, or General Note section of the drawing set. Vendors on delegated-design items are then instructed to use Owner’s Requirements while permit evaluation will be based on both Authority Having Jurisdiction (AHJ) and Owner’s Requirements being surpassed. The only risk with this approach is that some AHJs may take issue with this approach, thereby delaying the permit approval process.

The discussion on Risk Category naturally progresses to selection of the Importance Factor for the project. Table 1.5.2 on Importance Factor in ASCE 7-16 has been simplified in ASCE 7-22, now only showing Seismic Importance Factor (Ie). This is the life-safety portion: if an owner contemplates the use of Risk Category IV, the building frame’s Ie should be defined as such at a minimum.

Table 1.5.2 references Section 13.1.3 on the Component’s Importance Factor (Ip), where the code requires Ip=1.5 for any life-safety components. The assignment of Ip is where an engineer’s judgement can heavily influence a project’s costs. Ip = 1.5 is required for “Risk Category IV” projects “for the continued operation of a structure.”

In ASCE’s context of Risk Category (life-safety), this requirement makes sense. When the Risk Category selection is Owner-driven, the definition of Ip becomes more of a risk/cost-efficiency evaluation. Some components could be considered “mission-critical” and should be designed with sufficient redundancy. Consider an ice-cream factory: an owner may require continuous operation of ice-cream storage, but not production. Components required to maintain functionality of the cold storage (such as generators, UPS, etc.) would then be assigned Ip = 1.5 for continuous operation, while components only associated with production would be assigned Ip = 1.0 to maintain cost efficiency.

Table 13.1-1 is where the impact of Ip selection becomes clear. The requirements in Chapter 13 would typically apply to mechanical/electronical components in SDC D, E and F but exemptions may still apply if the component weighs less than 400 pounds (discrete) or 5 pounds/foot (distribution).

The most recent iteration of ASCE 7-22 has moved toward the use of a “multi-period” spectrum, moving away from “two-period” (0.2s and 1s) spectrum definitions. Meanwhile, ASCE 7-22 continues to evaluate SDC based on the two-period spectrum concept through a two-step process:

ASCE 7-22 has significantly revamped Site Class Table 20.2-1 to provide a more granular breakdown of each Site Class. In prior ASCE iterations, SDC and Site Class had similar categories (A through F) which can create misunderstandings among project stakeholders who are less familiar with these concepts. Along a similar note, Section C11.4.3 has also indicated that the Fa & Fv values previously used to convert Ss & S1 into SMS & SM1 are no longer needed and thus eliminated.

Table 11.6-1 and 11.6-2 in ASCE 7-22 is summarized into a single table below. SDC selection depends heavily on the seismicity of the project site and to a lesser extent, the project’s Risk Category. Use of Risk Category IV essentially bumps the SDC up by one bracket, unless the project is located in an area with virtually no seismicity.

The addition of Table 13.1-1 in ASCE 7-22 helps communicate the impact of SDC selection on non-structural components requirements:

The term “positive attachment” is not specifically defined under Chapter 13, but Section 12.1.4 requires a “positive connection” to have “a minimum design strength of 5% of the dead plus live load reaction.”

So, what exactly does this mean when we consider fungibility of modular equipment? One way to analyze this is by analyzing the geography of the project portfolio. In the case of data centers, Table 2 indicates the total Megawatt (MW) for the largest data center regions in the U.S.

A few insights to be gained from this table:

Based on the Table 11.6-1 & 11.6-2 from ASCE7-22, a structure designed with SDS=0.16 and SD1=0.066 aligns well with SDC=A, irrespective of Risk Category designation. IAD (Northern Virginia) represents about 37% of the data center market, and likely falls within this threshold.
Designing a structure based on seismic design criteria of SDS=0.2 and SD1=0.13, would cover about 82% of the data center market. Risk Category selection starts affecting cost at this point, due to its impact on SDC and thus seismic bracing requirements. Bumping up SDS to 0.33 and SD1 to 0.133 may cover more geographies, without a significant impact on costs.

Consistent with the Pareto principle, the “vital few” geographies will be unique to each site and the concept of a fungible design starts to become very expensive. At this point, site-specific design parameters will likely be required: designing a project with PDX’s seismicity will be very different compared to a project with LAX’s seismicity.

For Engineers of Record and Owners Reps, the concept of fungibility is essential for ensuring flexibility of equipment purchased. Enhancing fungibility of modular equipment and enclosures enables project owners to easily reallocate procured assets to different projects as priorities shift, expediting procurement lead-times. Design checks are likely still required by the AHJs but should be simpler since the design criteria is likely conservative enough to satisfy the localized loads imposed on the structure.

Use of energy dissipation devices can be used to achieve a more efficient design while maintaining fungibility. Displacements should be identified ahead of time where base isolation is used, to make sure any mechanical/electrical distribution routes are properly supported. Dampers can also be a good way to limit displacements while maintaining fungibility.

From a product vendor’s perspective, adopting fungible design is advantageous to minimize non-productive hours. Fungible products can be standardized across various clients and geographies and total inventory can be reduced since parts of fungible products will be more interchangeable.

Fungibility can be enhanced by analyzing critical project design criteria. Structures designed for Risk Category IV can be used for other Risk Categories and may be desirable in small structures. Analysis of project portfolio for seismicity and weather conditions (such as snow, rain, wind/tornado) design criteria can be optimized for enhanced fungibility. ■