Abstract
By the year 2025, there will be 200+ zettabytes of data in cloud storage. The servers handling this cloud storage space live in data centers. Since we live in the era of data, demand is surging, in fact, a 50% growth is expected by 2030. Data centers are energy consuming monsters, hence their lifecycle has a huge energy footprint. But how much is this footprint exactly over the lifecycle? This is the question this study aims to uncover exhibiting the steel structure of our MACAD Studio project, ‘System on a Chip’ a modular data center (the Reykjavik version).
Structural System
Our modular data center is based on an aggregation strategy, where the module components are populated according to a main building axis, this is where the main operation corridor of the facility aligns to at the same time. The principle is that, space programs can plug into this main corridor and expand. At the ground, we have a structure working together with MEP and storage. On a typical floor, we have our data halls and their necessary office and operations control.
The modular system is developed focusing on easy assembly and disassembly, a uniform raster and a simple, but versatile steel structural system that can adapt to any kind of module aggregation. The connectivity to the main structural frame is achieved with inter-module connections. These structural components have good energy dissipating capability and offer good scaling opportunities.
Energy Consumption – Steel H Sections
The main structural frame of the facility is built using HEA sections. The following infograph shows the energy consumption calculation for these steel members (steel beams and columns) per ton of steel, starting from the raw materials (iron ore and alloying elements) ending with transporting the steel members to site.
The total energy consumption related to the journey of steel to the site is as per the below table, the final figure is identified after taking the average of the value ranges.
Energy Consumption – Construction of the Steel Frame
As described in a former section, the building is built up from 4.5×4.5×4.5m modules, which means that the structural frame is fairly uniform. Accordingly, the energy consumption of the construction process can be calculated for a single module, then the amount multiplied by the number of the modules. Based on ‘Platforms in Practice’ whitepaper by Bryden Wood, the main construction steps for a module with the energy consumption of the main equipment used per step are as follows:
The sum of the energy consumption related to these steps is 3178 kWh/module.
Energy Consumption – Data Center Operation
The aggregation definition rely on several parameters that we can control. The position of the central line, its rotation and direction, the position of the entrances, the offset, maximum height that can be set and the cooling capacity, the percentage of the data halls, the MEP space, the office space, the control room, the storage and circulation. We also control the energy. The IT capacity has to be less than the IT cooling capacity. We considered that there is an IT cooling capcity for the entire building that we evaluated as 1 MW per 5000 SQM, the FAR, the white and grey space ratio, the number of generatiors that we have, the number of chillers that we need and how many control operators can sit in our space.
After running 200 or more iterations, we filtered the potential design options to achieve the specified targets based on the relationship to functional areas and the two main metrics: White-Grey ratio and FAR in this case. Accordingly, the above design was selected, with the related metrics. In terms of annual energy consumption it is 66MW.
The Math
Conclusion
Construction related energy consumption is only a small fragment compared to the operation related one for a data center. Smart construction techniques that save energy for the building’s initial construction and yield further savings for the renovations and expansions that are expected in a building’s post-handover lifecycle, can only offset the energy footprint very little. The really problematic part remains the operation, and the demand for it is surging.
According to Mack DeGeurin in an article published at Gizmodo, “large language models like ChatGPT are energy-intensive, requiring massive server farms to provide enough data to train the powerful programs. Cooling those same data centers also makes the AI chatbots incredibly thirsty. New research suggests training for GPT-3 alone consumed 700,000 liters of water. An average user’s conversational exchange with ChatGPT basically amounts to dumping a large bottle of fresh water out on the ground.”
Curbing the numbers, therefor, mainly depend on people – are those useless selfies stored in the cloud really required? Are those searches and ChatGPT conversations really required? The power to change the trends are in our hands.
