Project

Hydro Power

Hydroelectric power plants and data centers both use water as their main component, respectively for power generation and hardware cooling. This section aims to address the hypothesis of how much energy could be generated if we were to utilize the same amount of water that a data center consumes for electricity generation.

The most commonly used hydroelectric power system is the impoundment system [1], which involves holding water in a reservoir and then moving it from a higher to a lower elevation. This system generates electricity by passing large volumes of water through turbines, utilizing the force of gravity to effectively convert potential energy into kinetic energy, ultimately into electrical energy, with an astounding efficiency of around 90% [2].

https://www.energy.gov/eere/water/types-hydropower-plants
https://www.sciencedirect.com/topics/engineering/hydroelectric-energy

The power generated from such a system could be calculated using the following equation, where $\eta$ represents the efficiency, $m$ is the mass of water, $g$ is the gravitational acceleration, $\Delta h$ is the distance between the water surface and the turbine (commonly referred to as head), and $t$ is the time.

$$\begin{array}{rl}P& =\eta \frac{mg\Delta h}{t}\\ & =\eta \frac{\rho Vg\Delta h}{t}\end{array}$$

Referring to the numbers in the analysis part, and the fact that there are around 8000 data centers in the world [3], their total annual water consumption (TWC) and average water consumption (AWC) are as follows,

https://cloudscene.com/region/datacenters-in-asia-pacific

$$\begin{array}{rl}& \text{TWC}=206+51.5=257.5\text{ billion gallons}\\ & \text{AWC}=\frac{257.5}{8000}\approx 32.2\text{ million gallons}\approx 122\times 10^3{\text{ m}}^3\end{array}$$

Assume we utilize the water in a medium-sized impoundment system with a head of around 100m [4]. And taking $\rho$ as $\text{pu}1000kg/m3$, $g$ as $\text{pu}10m/s2$, we can obtain the annual energy generated from water consumed in a single data center,

https://www.sciencedirect.com/topics/engineering/hydropower-plant

$$\begin{array}{rl}E& =Pt\\ & =\eta \frac{\rho Vg\Delta h}{t}t\\ & =0.9\times 1000\times 122\times 10^3\times 10\times 100\\ & \approx 1.11\times 10^{11}\text{ J}\\ & \approx 30500\text{ kWh}\end{array}$$

We could use this number to compare with different type of building, for example, a grade A office in Hong Kong consumes $\text{pu}398MJ/m2\approx \text{pu}111kWh/m2$ annually [5]. And since the average floor space of offices is around $\text{pu}10000ft2\approx \text{pu}929m2$, we can see that a typical office would consume $111\times 929=103119\text{ kWh}$ annually. This is roughly 30% of what the water consumed in a data center can generate.

https://ecib.emsd.gov.hk/index.php/en/energy-utilisation-index-en/commercial-sector-en
https://www.statista.com/statistics/912675/hong-kong-office-sizes-in-central-and-kowloon-east-by-floor-space/

Take another example, The Hong Kong University of Science and Technology consumes 99 million kWh of energy in the 2022-23 academic year [6]. Here, the energy produced by water in a data center is roughly 0.03% of the energy consumption of our university.

https://sust.hkust.edu.hk/progress-and-performance/progress/energy

https://www.sciencedirect.com/topics/engineering/hydropower-plant

Commercial/Cities

Apart from the above hypothesis, we could directly compare the water consumption between data centers and cities.

In Hong Kong, from 2022 to 2023, the annual consumption of water is 1074.53 million cubic meters [1], which is equivalent to the water consumption of around $\frac{1074.53\times 10^6}{122\times 10^3}\approx 8800$ data centers, it is surprisingly close to the total number of data centers in the world. In other words, the annual consumption of water for all data centers in the world combined is equivalent to that of Hong Kong.

https://www.wsd.gov.hk/en/publications-and-statistics/pr-publications/the-facts/index.html

If we take a closer look into the per capita water consumption in Hong Kong, it is around 130L per person per day [2], which is equal to $\frac{130}{1000}\times 365=\text{pu}47.45m3/year$. Therefore, the water consumption of a data center can support $\frac{122\times 10^3}{47.45}\approx 2600$ people living in Hong Kong.

https://www.waterconservation.gov.hk/en/why-save-water/virtual-water/index.html

If we assume a building estate with 30 stories, and each story consisting of 10 flats and each flat with 3 people, then the water consumption a data center can effectively supply $\frac{2600}{30\times 10\times 3}\approx 3$ buildings.

Conclusion

While the exponential need for information processing drastically increases the water consumption of data centers, technological advancements shed light on remedying this issue. Along with the rising awareness of sustainability, companies have already started to shift their focus toward reducing environmental impact. Industry leaders in cloud computing such as Google aims to use alternatives to freshwater like industrial water, wastewater, and seawater [1], with AWS claiming that 2.4 billion liters of water are returned to the communities [2]. In addition, the implantation of renewable energy allows these companies to manipulate water while consuming cleaner electricity, which also reduces their indirect water consumption, we can already see this in Google where they are matching 100% of their electricity consumption with purchases of renewable energy [3].

With the data and comparisons shown in this report, we can grasp a clearer picture of how managing water consumption is vital to the environment. In the meantime, our focus on sustainability is paving the way for the reduction of water consumption in data centers, as well as driving the world into a more sustainable future.

https://blog.google/outreach-initiatives/sustainability/our-commitment-to-climate-conscious-data-center-cooling/
https://sustainability.aboutamazon.com/products-services/the-cloud?energyType=true
https://www.google.com/about/datacenters/cleanenergy/

Unused
https://www.datacenterfrontier.com/sustainability/article/21438279/aws-targets-water-use-in-its-cloud-data-centers

Poster

A1 = 4 x A3