Solar2Power Data Centers Globally
In today’s technology-driven era, the voracious appetite for data consumption shows no signs of slowing down. Consequently, data centers must persistently cater to this escalating demand. In the following article, we shall delve into the global electricity consumption of data centers and present a viable solar power solution through a theoretical case study. This approach demonstrates the potential for data centers to significantly reduce their electricity expenses.
Global electricity consumption for data centers
According to an article on energyinnovation.org, precise data regarding the worldwide energy consumption of data centers is hard to come by. A 2011 study by Koomey remains one of the most frequently cited references in this field. Koomey’s analysis employs a “bottom-up” mathematical approach, considering the installed inventory of IT equipment within data centers, thereby enabling an estimation of the overall energy needs. In 2010, it was estimated that globally, data centers accounted for 1.1-1.5% of global electricity usage. (Marcacci, 2020)
By 2018, data demands had surged by a factor of ten, and while many asserted that energy usage had more than doubled, a “bottom-up” analysis indicated a more modest increase of approximately 6% in electricity consumption. The latest estimates take into account recent data on IT equipment, featuring improved technology, software, and efficiency that enables more streamlined and efficient data storage compared to older hardware. For instance, enhanced server virtualization software permits multiple thread operations to be executed on a single server instead of multiple ones, significantly reducing power consumption. Moreover, advancements in server room equipment cooling have resulted in superior airflow and reduced reliance on air conditioning systems. It was concluded that by 2018, data centers used around 205 TWh, which is 1% of global consumption, slightly less than originally estimated in 2010. (Marcacci, 2020)
The possibilities of solar power
Solar power is a promising solution, primarily due to the abundant sunshine it leverages. The Algarve region, renowned for its 300 days of annual sunshine, has become a prime location for both residential and commercial solar energy solutions. However, when considering data centers, it’s crucial to acknowledge their continuous 24/7 operation requirement.
To sustain a data center solely on solar power, would necessitate a substantial battery storage system capable of functioning during nighttime and overcast, stormy days. While this is technically feasible, current battery technology does not yet make it a financially viable option. Consequently, the more practical approach for data centers is to operate in tandem with the grid, utilizing battery storage as a backup for grid power outages.
Typically, grid power outages seldom persist for more than 12 hours. Therefore, data centers should have sufficient battery backup for this duration to ensure uninterrupted operations.
A theoretical case study
A typical large-scale Tier 3 data center would use around 100 MW of power. So if we convert this to Killowat hours, throughout a 24-hour day, it would use 2400MWh for the day.
As mentioned previously, for this study we will try and offset this without the use of battery storage as it would not be feasible.
Data Center requirement – 100MW solar system
Solar Farm required panels – 175 500x570Wp Bifacial panels
Area of a 570Wp Bifacial SHARP Panel with spacing – 2,58 SQM
Required Land Area – 4.515 Million SQM (450 Hectares)
With the above requirements, we could fit approximately 175 500 panels requiring 450 hectares of usable space at ground level. However, we need to adjust for spacing between rows, so ideally, 600 hectares of land would facilitate a solar project of this capacity.
Solar power output
Let us assume 8.5 Hours of optimal sunlight per day using 175 500x 570W panels.
Solar Output = 175 500 Panels x 570W rated power x 8,5 Peak Daily Sunlight Hours x 0.8(losses) = Approximately 682MWh of energy provided daily on average.
A 100MW data center operating for 24 hours would require 2 400 MWh of electricity daily.
Therefore, a solar farm supplying 100MW of electricity can power a large-scale data center for approximately 8.5 hours, offsetting its overall daily cost by 28.4%.
Calculation –
682MWh(maximum supply from panels) / 2400Mwh(data center usage) x 100 = 28.4%
Estimated Cost
Estimated breakdown
Price/MW: 700 Thousand Euros
100MW farm total – approximately 70 – 90 million Euros
(Rough estimation according to a 48MW project done by Greenvolt Group)
Estimated savings breakdown
Currently, the yearly average electricity cost in Portugal is €90 per MWH.
So for a data center running at 100MW capacity daily, would require 2400MWH of electricity over 24 hours which translates to €216 000 Euros daily and €78 840 000 yearly.
With the proposed 100MW solar farm, we determined we could offset the overall required electric consumption from the grid by 28.4% daily.
Therefore, the daily average electricity expense of €216 000 would drop by €61 344 to an overall average of €154 656 daily. The yearly cost would now be €56 449 440.
So if the total estimated savings is (€61 344 x 365) €22 390 560 yearly, it would take approximately 4 years to pay off the solar farm investment of around €80m.
Conclusion
In conclusion, the exploration of solar power as a sustainable energy solution for data centers presents a compelling case. With the global electricity consumption of data centers on a consistent rise, the need for innovative and eco-friendly alternatives is paramount. The theoretical case study focused on a large-scale Tier 3 data center operating at 100MW and showcased the potential of solar power to significantly reduce electricity expenses.
While the complete reliance on solar power alone may not be currently economically feasible due to limitations in battery technology, the integration of solar farms in tandem with the grid emerges as a practical and efficient approach. The theoretical case study revealed that a 100MW solar farm, requiring approximately 600 hectares of land, could offset a large-scale data center’s electricity consumption by 28.4%. This reduction in daily expenses translates into substantial yearly savings, making the investment in solar infrastructure financially viable over time.
The estimated cost of the proposed 100MW solar farm, ranging from 70 to 90 million Euros, may initially seem substantial. However, the potential yearly savings of around €22.4 million indicate a relatively quick return on investment. In this scenario, it would take approximately four years to recoup the initial investment, making solar power an economically attractive and environmentally responsible choice for powering data centers.
As the world continues to grapple with the challenges of energy consumption and environmental sustainability, the integration of solar power in data centers exemplifies a forward-thinking and economically sound solution. This theoretical case study serves as a testament to the potential for solar energy to not only address the rising demands of data centers but also contribute to a more sustainable and resilient energy future.
Disclaimer: We strive to provide accurate and up-to-date information, but we make no representations or warranties of any kind, express or implied, about the completeness, accuracy, reliability, suitability, or availability concerning the information, products, services, or related graphics in this blog post. Any reliance you place on such information is strictly at your own risk.
References
Marcacci, S. (2020) How much energy do data centers really use?, Energy Innovation: Policy and Technology. Available at: https://energyinnovation.org/2020/03/17/how-much-energy-do-data-centers-really-use/ (Accessed: 15 November 2023).
Greenvolt connects the 48 MW of Tábua Solar Power Station (2023) Greenvolt. Available at: https://greenvolt.com/tabua-solar-power-station/ (Accessed: 15 November 2023).
570WP / NBJD570 (no date) Sharp. Available at: https://www.sharp.eu/monocrystalline-solar-panels/570wp-nbjd570 (Accessed: 15 November 2023).