News & Insights
Data centers: inside the fabrication-level model
Fabrication-level modeling turns data center designs into virtual twins, preventing expensive field issues and accelerating project delivery.
AI infrastructure has compressed delivery timelines while increasing design complexity. Technology turns over every six months, and projects that miss their market window don’t get a second chance. Yet most traditional design processes weren’t built for this pace.
By the time detailed design wraps, unforeseen problems begin to take shape. Equipment may not fit through openings that were sized conceptually, systems can clash in ways generic models didn’t reveal, and maintenance access can get blocked by installations that looked fine on paper.
The coordination gaps surface as expensive RFIs and change orders, each of which consumes senior engineering time that could go toward new projects. For hyperscalers and developers working on compressed schedules, discovering these issues during construction means blown budgets and missed deadlines. The question becomes whether to solve problems during design when everything’s flexible or accept exponentially higher costs once construction begins.
Why complexity has outpaced traditional methods
The shift from air-cooled to liquid-cooled systems for AI workloads has fundamentally changed what data centers require. While building shell dimensions haven’t changed, the infrastructure inside has grown far more complex.
In current data centers, chilled water piping networks that barely existed in air-cooled facilities now snake through every space. Power densities have multiplied. The number of conveyance systems competing for the same overhead pathways has increased dramatically, while construction timelines have stayed the same or gotten shorter.
This complexity means the rinse-and-repeat era is over. Air-cooled facilities evolved gradually over the past decade, allowing teams to refine a familiar playbook with minor upgrades along the way. However, liquid cooling for AI is uncharted territory where each project introduces new challenges that experience from previous builds doesn’t fully address. Construction crews work on essentially first-of-kind installations without the muscle memory that comes from repetition.
Labor shortages have compounded these challenges across every market, making field installation both more expensive and more difficult to staff. Prefabrication has shifted from optional to essential, but it demands precision that conceptual models can’t provide. Components built off-site require exact connection points, support locations, and assembly sequences to be determined before anything ships. Traditional design approaches that leave these details for construction teams face significant challenges when the field has fewer experienced workers and tighter schedules than ever before.
How fabrication-level modeling changes the equation
Fabrication-level modeling transforms coordination from a handoff problem into a collaborative process. The shift happens through detailed modeling that makes design decisions visible and accessible to everyone who needs to contribute.
Building a virtual twin, not a conceptual representation
Standard modeling practices often treat components as placeholders. For example, when ductwork requires specific fittings and connections, a generic line shows only general routing. Or, in some cases, equipment appears in approximate locations without the detail needed for installation. Support systems for overhead infrastructure can disappear entirely, only to be addressed later by construction teams. The result can pass clash detection while still presenting buildability challenges, because these representations don’t reflect what actually gets installed on site.
Fabrication-level modeling works differently. Every component appears as it will be built with specific fittings, connection points, support assemblies, and prefabricated elements.
With this level of detailed modeling, elements such as fire protection systems include slope requirements that dictate routing. Steel fabrication gets detailed rather than assumed. Overhead supports for ductwork, piping, electrical services, and telecom tray all appear in the model, competing for the same space they will in the constructed building. The model itself serves as a virtual twin, a dress rehearsal of construction rather than a schematic suggestion.
This level of detail enables new conversations among all stakeholders. When construction partners review a fabrication-level model, they see exactly how systems come together. This allows them to provide meaningful feedback informed by local suppliers, field experience, and practical constraints. Fabrication-level models make years of expertise accessible when it matters most.
Seeing how projects come together over time
Creating accurate geometry solves one problem, but construction happens in sequences that create their own coordination challenges.
Virtual sequencing links the 3D model to project schedules, usually in Primavera P6, illustrating how systems are installed over time rather than just their final arrangement. This approach identifies conflicts not visible in static models—such as crews working simultaneously in the same space within overlapping schedule periods, equipment that must arrive before openings are closed, and access routes that work initially but become blocked as installation advances.
Animations serve as collaborative tools rather than strict requirements. Design teams share their vision of how construction could develop and then consult with construction partners to identify any gaps. These discussions help find improvements that lower schedule risks and minimize site congestion. The aim is to develop a common visual reference that allows all stakeholders to participate in sequencing decisions, while keeping changes cost-effective.
Quantity takeoffs become significantly more precise when models include fabrication-level details. Providing early material estimates based on fittings, supports, and assemblies helps teams grasp the bills of materials during the design phase, rather than relying on construction estimates later on.
This accuracy enables cost validation before issuing construction documents and helps quantify how design changes influence project budgets. For instance, modifying a chilled water piping run to eliminate fittings produces measurable savings that can be compared with other options. Consequently, design evolves from a fixed plan to a continuous process of cost optimization.
Catching problems when they’re still cheap to fix
The significance of detailed fabrication modeling is evident in its role in preventing issues. For example, steel fabrication frequently misses details because standard models might omit certain structural elements. Likewise, electrical skids are verified against actual opening sizes during the design phase rather than post-installation.
Proper equipment placement is essential because it enables maintenance teams to easily access filters and perform routine repairs. This helps avoid access problems that could be costly and time-consuming to fix later.
Cable bus installations show how precise this method can be. These electrical systems require high accuracy but can greatly reduce field work compared to traditional conduit-and-wire setups.
Many teams may avoid them because small coordination mistakes can halt the installation, and conceptual models often lack sufficient detail for confident work. Accurate modeling of the fabrication process captures the required geometry, turning a potential problem into an efficient, cost-effective solution.
Involving the right people at the right time
The stakeholders whose feedback is most crucial often have less interest in working with technical models. Operations personnel understand the long-term functioning of facilities but are not experts in modeling. Fabricators are aware of what can be prefabricated efficiently, but they usually get involved after the design phase ends.
This disconnect can leave coordination gaps. For instance, field supervisors, with their installation experience, could help avoid coordination issues. However, they typically join projects only after construction has started.
Fabrication-level models make information accessible to these stakeholders through visual representations that don’t require technical expertise to understand. When everyone can see how systems connect, where access exists, and how installation sequences unfold, the conversation shifts from abstract coordination concerns to concrete problem-solving. Getting this input at 30 or 50 percent design means incorporating it easily rather than forcing changes after permits and construction documents lock decisions in place.
This process gives construction teams clear starting points rather than generic placeholders, making their jobs much easier. The process is more team-oriented because design teams can admit they don’t know everything and feel empowered to seek input from people with on-the-ground experience.
Production staff handles detailed modeling, which alters the resource balance. Senior engineers tend to constrain the project due to their high rates and frequent reallocation between priorities. Delegating detailed modeling to production staff early on is more cost-effective and yields better results, as production resources are more readily available than those of senior engineers.
What adoption requires
Two concerns consistently surface when considering fabrication-level modeling. Both stem from understandable assumptions about how detailed coordination affects project delivery.
Design timelines stay comparable
With ever-tightening deadlines, scheduling is always the primary concern. Does creating detailed models prolong the design process and push back project initiation?
Experience shows that design timelines remain similar to traditional methods, generally around 14 weeks, though they vary with scope and complexity. The difference lies in how effort is allocated and what is delivered at the conclusion.
Traditional workflows may rush conceptual models, causing months of managing field coordination issues during construction. Fabrication-level modeling reverses this approach. By allocating production resources early to create detailed coordination plans, issues that would otherwise require significant senior engineering time decrease. While this shift in timelines works from reactive problem-solving to proactive development, the total duration stays roughly the same.
Collaboration replaces territorial concerns
Construction teams often see detailed design models as a form of territorial encroachment—believing they are a way for designers to control means and methods or to seize work typically handled by contractors. However, this concern quickly fades once teams observe how fabrication-level modeling works in real-time.
The fabrication-level approach explicitly avoids prescribing construction methodology. Design teams offer detailed models as starting points, not mandates, and actively seek input from construction partners who bring field experience and regional knowledge that designers lack. The models create a foundation that contractors can refine and build upon rather than recreating from scratch.
Construction teams have often found that collaboration produces practical deliverables rather than generic placeholders that add little value to fabrication and installation.
How Salas O’Brien can help
Your data center projects need coordination that keeps pace with the speed your timelines demand. Salas O’Brien brings Virtual Design and Construction expertise developed across hundreds of data center facilities, with deep understanding of the liquid-cooled AI infrastructure driving today’s most complex builds.
We approach fabrication-level modeling as a partnership rather than a service delivery. Our teams integrate construction partners, vendors, and operators early in design when their input creates the most value.
The detailed models we develop give contractors starting points rather than requiring them to recreate. They provide operations teams with facilities they understand from day one because they participated in development rather than inheriting documentation after turnover.
The result is often fewer RFIs consuming your senior engineering resources, minimal change orders extending schedules and budgets, and construction teams working from your models rather than discarding them. Most importantly, you get facilities online faster without sacrificing the coordination quality that keeps them running reliably once they’re live.
Contact our data center team to discuss how fabrication-level modeling can compress your delivery timeline: [email protected]
For media inquiries on this article, reach out to [email protected].
Paul Ehvert, PE
Paul specializes in design team project management for hyperscale, AI, and colocation data center programs across the United States. With over fifteen years of experience, he leads large, multidisciplinary design teams spanning architecture, engineering, construction, and vendor partners to deliver complex mission-critical facilities on aggressive schedules. Paul’s focus is on owning design execution, aligning scope, schedule, and deliverables across multiple firms, driving decisions, and maintaining accountability from early concept through IFC and construction support. He also leads Salas O’Brien’s Virtual Design and Construction practice, leveraging VDC as a design management tool to advance models to fabrication-level detail, enabling high-confidence coordination, prefabrication, accurate quantities, and schedule-driven design alignment. Paul serves as a Principal at Salas ’O’bBrien. Contact him at [email protected]