Semiconductor Water Supply: Focus on Reliability, Not Just Volume, Including Extreme Drought Scenarios

It was pointed out that water measures for large-scale semiconductor clusters need to shift from focusing on securing volume to focusing on supply reliability, since semiconductor manufacturing processes require large amounts of ultrapure water for cleaning and water of different quality depending on use, such as cooling. Because semiconductor production requires large volumes of water and water of varying quality for each process, simply securing the planned amount is not enough; the factors to consider must include not only quantity but also raw water quality and supply stability.

Accordingly, to keep semiconductor production lines operating even under extreme drought and climate change conditions, a policy direction was proposed that goes beyond merely securing the planned required amount and instead links multiple water sources, expands reclaimed water, and uses AI-driven flexible dam operations. It was also suggested that related systems need to be improved, as converting existing water sources for power generation and agriculture to industrial use creates cost burdens and water rights adjustment issues.

At a forum held on the 14th at the National Assembly Members' Office Building, Kang Bu-sik, a professor in the Department of Infrastructure Construction Engineering at Dankook University, presented potential demand by combining water demand from three advanced-industry projects using published project plans and his own calculation standards in the forum titled "The Republic of Korea's Semiconductor Future Map and Infrastructure Strategy Part.2 Water." The projects considered were the Yongin semiconductor cluster, the Chungnam region AI and semiconductor industrial complex, and the southwestern semiconductor industrial complex. If these projects are built as planned and operate at full capacity, the required water volume was estimated at about 2.237 million tons per day. Specifically, the Yongin semiconductor cluster was estimated to need 1.337 million tons per day, the Chungnam region about 250,000 tons, and the southwestern region 650,000 tons. The Chungnam region includes semiconductor and display industrial complexes and AI data centers, and Kang estimated the figures based on the scale of the projects and other factors.

The 2.237 million tons cited by Kang is not the government-confirmed supply volume or current usage, but potential demand premised on future business plans. In particular, total demand may change depending on the actual facility scale and operating plans. Kang explained that if daily per-capita water use is assumed to be about 373 liters, 2.237 million tons corresponds to the daily water use of 6 million people.

A forum titled "The Republic of Korea's Semiconductor Future Map and Infrastructure Strategy Part.2 Water" was held at the National Assembly Members' Office Building on the 14th, and participants posed for a commemorative photo.

At the forum, experts consistently pointed out that in discussions about securing water for semiconductors, it is necessary to distinguish between the amount that can be secured in plans and actual supply capacity.

Kim Sang-dan, a professor in the Department of Environmental Engineering at Pukyong National University, said that the calculation showing that "650,000 tons" can be secured is separate from the ability to supply that amount continuously. He also explained that calculating semiconductor industrial water based only on average annual inflow is not sufficient.

Kim said it is necessary to review whether supply is possible during extreme drought periods. He also said the discussion should be premised on existing supplies for residential, industrial, agricultural, and river maintenance water.

He then explained that if new industrial water demand is added, it is necessary to analyze changes in reservoir levels at each dam, as well as changes in the supply capacity of the entire watershed.

There is concern that water sources within the same region may be simultaneously affected by prolonged drought, meaning that securing multiple dams does not proportionally increase risk diversification. Accordingly, it was explained that simply increasing the number of dams is not enough to sufficiently spread drought risk.

Kim said that rather than simply adding up the supply capacity of individual dams, it is necessary to assess the supply reliability of the entire system, reflecting simultaneous drought across multiple water sources. He also said it is necessary to examine whether using water designated as emergency water sources during drought as part of regular industrial complex demand could reduce separate emergency response capacity.

Professor Kang said it is necessary to shift from a plan centered on securing required volume to a system centered on supply reliability and water security. He also noted that it is difficult to apply a single method to securing water for industrial complexes because demand, available water source types, water quality, and existing uses differ by industrial complex.

As a solution, a multi-source system was proposed in which multipurpose dams, power generation dams, agricultural reservoirs, and treated wastewater reuse are operated in coordination. A method of combining different types of water sources and operating them like a single supply network was mentioned as an alternative.

An image of water for semiconductor manufacturing processes was presented. Stable water supply was presented as a key infrastructure for semiconductor production. The image was generated using AI.

The article explained methods to improve supply stability by treating water supply for semiconductor production as a core foundation. It said that rather than continually building new dams, the more important task is to efficiently store and reuse existing secured water resources.

Professor Kang proposed separately collecting and storing water that must be released during floods and water discharged downstream during power generation if it can be used. He suggested installing retention facilities, pump stations, and diversion pipelines at appropriate locations for this purpose.

This approach was described as an operating method that stores water in normal times and uses it as industrial water during droughts and emergencies. The flow is that collecting and storing released water becomes a means of securing industrial water in emergencies.

There was also an opinion that it is necessary to increase the use of reclaimed water generated from sewage and industrial complexes. Reclaimed water is relatively less affected by rainfall changes and has the advantage of being well suited for building a stable supply source mix.

Kim said treated wastewater reuse and reclaimed water from industrial complexes are water sources relatively less affected by climate, and that reclaimed water should be considered as a major pillar of the industrial water supply portfolio rather than as a backup measure used only in emergencies.

However, raw water quality was presented as an important variable in semiconductor processes. It was explained that not all water used in semiconductor processes needs to be ultrapure water, but if the raw water for producing ultrapure water contains large amounts of substances that are difficult to treat, the burden of the purification process increases and costs rise.

As a measure, it was suggested that rather than simply increasing reuse rates, water quality requirements should be distinguished by process, and reclaimed water should be used for purposes such as cooling water, where relatively high water quality is not necessary.

The forum also raised the idea of applying AI to dam operations. Current dam operations are based on pre-set operating standards such as flood-season restriction levels and are run by reflecting weather forecasts, inflow volumes, and reservoir conditions. Kang said that climate uncertainty is increasing and proposed the need for more dynamic adjustment of operating standards by comprehensively analyzing weather, inflow, and industrial demand with AI.

The Dynamic Rule Curve method adjusts a dam's appropriate water level based on AI analysis of weather forecasts, expected inflows, and industrial water demand. If heavy rainfall is expected, it secures storage space in advance, and if insufficient rainfall is expected later, it stores additional water. The concept is focused on optimizing flood response and water supply at the same time.

This approach can also be extended to operating multiple dams and reservoirs as a single system when they are linked. In that expanded form of coordinated operation, water intake, storage, and release from each source are adjusted according to weather, inflow, and industrial demand.

However, in actual large-scale industrial water supply projects, the issue of industrial water pricing becomes prominent. Accordingly, the task was raised of determining who will bear the cost of large-scale industrial water supply and how much they will bear.

In the case of power generation dams, taking stored water for industrial use before power generation may reduce power output and cause losses. Reducing the discharge volume from power generation dams may also lead to lower power generation and losses. On the other hand, taking and storing downstream release water after power generation has different effects and requires separate review.

If agricultural water is converted for industrial use, costs may arise to secure substitute water sources for existing users. In addition, the industrial benefits of stable water supply for semiconductor production lines may be greater than those of ordinary water use. Ultimately, when determining industrial water pricing and cost sharing, power generation losses, compensation costs for existing water users, and differences in industrial-sector benefits must all be considered together.

Professor Kang proposed that when setting industrial water prices and cost burdens, it is necessary to integrate power generation losses, the benefits of industrial water supply, and conversion and transport costs. He also said that for high-value-added industries, the beneficiary-pays principle should be applied more actively.

Professor Kang viewed industrial water for semiconductors as both a public good and a core production factor for a national strategic industry. He therefore proposed reviewing a pricing system that preserves public character rather than treating it in the same way as residential water.

Professor Kang said this pricing system needs to reflect economic value and supply reliability. Current tariff-setting standards are structured around the type of water supplied, such as raw water and treated water, rather than the final point of use, and under this structure the economic benefits of industrial water use are not directly reflected in tariffs.

An official from the Korea Water Resources Corporation also said it is necessary to review ways to reflect the economic benefits of industrial water in the tariff system. The official also suggested reviewing how to allocate infrastructure costs for water used in high-value-added industries and proposed linking industrial benefits with cost burdens.

It was pointed out that the existing water management system alone has limits in the legal and institutional sense when responding quickly to the water demand of large-scale advanced industries. To secure water for advanced industries, it must be assumed that power generation dams and agricultural reservoirs will be used to supply industrial water, and the need to adjust existing uses and water rights was raised as a related task.

In this process, it was considered necessary to establish standards for compensation for power generation losses and for existing users. It was also noted that promoting multi-source linkage and the storage and reuse of floodwater and power generation discharge water requires a clear legal basis to support related projects.

Regarding the legislative approach, Professor Kang suggested establishing special provisions for strategic industry water use within the "Special Act on Strengthening and Protecting the Competitiveness of National Advanced Strategic Industries" rather than creating a separate special law. He also proposed revising related laws and regulations together, including the Framework Act on Water Management, the Dam Construction and Management Act, the River Act, and the Rearrangement of Agricultural and Fishing Villages Act.

Another issue raised was aligning the timing of semiconductor facility construction with the timing of water infrastructure development. To this end, it is necessary to specify the maximum and average water demand of each industrial complex and the actual timing of production facility operation, and the schedule from water source securing to permitting, pipeline construction, and operation should be linked.

It was also emphasized that simply securing the required amount in plans is not enough. There was also a demand that supply systems capable of supporting production facilities even in the event of extreme drought be verified in advance.

Professor Kang said that water-securement measures should be pursued not only through technical means such as multi-source linkage, expanded use of reclaimed water, and AI-based water resource operations, but also together with institutional reforms related to the economic value of water, cost sharing, water rights, and conversion of use. He also explained that fab operating rates in semiconductor processes need to be synchronized with the timing of the water supply system, and said it is necessary to establish a supply-demand system and operational framework centered on supply reliability rather than on securing volume.

Source: IT DAILY · Kim Byung-joo
Original: https://www.itdaily.kr/news/articleView.html?idxno=241016

References

• https://www.itdaily.kr/news/articleView.html?idxno=241016