Thermal Energy Storage for Data Center Cooling in Hot Climates

Data centers in hot-climate regions face a persistent challenge: cooling accounts for 40 to 50 percent of total facility power consumption when ambient temperatures regularly exceed 40°C. In the UAE, GCC, and other arid regions, mechanical chillers run at maximum capacity for most of the year, driving up electricity costs and placing sustained demand on utility grids. Thermal energy storage (TES) offers a proven approach to managing this load by decoupling cooling production from cooling consumption, shifting energy-intensive chiller operation to off-peak hours, and providing a buffer against grid disruptions.

This guide examines how TES technologies work in the context of data center cooling systems, the specific advantages they offer in hot climates, and the practical considerations for operators evaluating TES integration in their facilities.

What Is Thermal Energy Storage?

Thermal energy storage systems store cooling capacity produced during one period for use during another. Rather than sizing mechanical chillers to meet the absolute peak cooling demand, a facility with TES can run smaller chillers continuously, building up a reserve of stored cooling that supplements chiller output during peak demand windows.

The principle is straightforward: produce cold when electricity is cheapest or most available, store it, and deploy it when electricity is most expensive or when demand on the grid is highest. The result is lower peak electrical demand, reduced utility demand charges, smaller chiller plant capacity requirements, and an additional layer of cooling redundancy.

TES Technologies for Data Centers

Three primary thermal storage technologies are relevant for data center cooling applications, each with distinct characteristics that suit different operational profiles and physical constraints.

Ice Storage Systems

Ice storage systems freeze water in insulated tanks during off-peak hours, typically overnight. During peak cooling periods, the ice melts and the resulting cold water or glycol solution circulates through the cooling distribution system, supplementing or replacing mechanical chiller output.

The latent heat of fusion gives ice storage a significant density advantage: ice stores approximately 5 to 6 times more cooling energy per cubic meter than an equivalent volume of chilled water. This makes ice storage the preferred option where floor space or outdoor footprint is constrained. A 2,000-ton-hour ice storage system might occupy 150 to 200 square meters, whereas a chilled water tank storing the same capacity would require 800 to 1,000 square meters.

The tradeoff is that ice-making requires chiller supply temperatures around -5 to -7°C, which is 10 to 12 degrees lower than standard chilled water production. This reduces chiller coefficient of performance (COP) during the ice-making cycle, partially offsetting the electricity cost savings from off-peak operation. However, in regions with significant peak-to-off-peak tariff differentials, the net economics remain favorable.

Chilled Water Thermal Storage

Chilled water TES uses large insulated tanks, often stratified designs where cold water (4 to 6°C) sits at the bottom and warm return water (12 to 14°C) occupies the top, separated by a natural thermocline. The tanks charge overnight as chillers produce cold water, and discharge during peak periods as the stored cold water feeds the cooling distribution system.

Chilled water storage integrates naturally with existing chiller plants because it operates at standard chilled water temperatures with no glycol or ice-making modifications needed. This simplicity makes it common in district cooling installations and large campus environments. The major district cooling operators in the UAE, including Empower and Tabreed, use stratified chilled water TES tanks extensively to manage peak demand across their networks.

The disadvantage is tank volume. A chilled water tank storing 10,000 ton-hours of cooling capacity may require a 4,000 to 5,000 cubic meter tank, which needs significant real estate. For greenfield data center campuses with available land, this is manageable. For constrained urban sites, ice storage or phase-change materials may be preferable.

Phase-Change Materials (PCMs)

Phase-change materials absorb and release thermal energy at a specific transition temperature, typically engineered to match the cooling supply temperature of a data center (around 7 to 15°C for different PCM formulations). Unlike ice, which changes phase at 0°C, PCMs can be designed to operate within the normal chilled water temperature range, avoiding the COP penalty of ice-making while still achieving higher energy density than sensible chilled water storage.

PCM systems are still maturing in data center applications. They offer a middle ground between ice and chilled water in terms of energy density and system complexity, but their higher material cost and limited track record in mission-critical facilities mean that most production deployments today use ice or chilled water. That said, PCM-based modules are seeing adoption in edge computing installations where compact form factors and minimal maintenance are priorities.

Why TES Is Especially Valuable in Hot Climates

The business case for thermal energy storage is stronger in hot-climate regions than in temperate ones, for several interconnected reasons.

Year-Round Cooling Demand

In the UAE, data centers require mechanical cooling every day of the year. There is no season where free cooling from ambient air can meaningfully reduce chiller load. This constant demand means that TES operates at high utilization year-round, improving the return on investment compared to locations where TES sits idle during winter months.

Peak Demand Charges and Tariff Structures

UAE utilities including DEWA (Dubai Electricity and Water Authority) and ADDC (Abu Dhabi Distribution Company) assess demand charges based on peak electrical consumption. For a data center that draws 10 MW at peak, a 2 MW reduction through TES-based demand response directly reduces monthly demand charges. Combined with time-of-use pricing differentials where available, the annual savings from TES can reach 25 to 35 percent of total cooling electricity costs.

Grid Stress and Reliability

Summer afternoon peak demand in the UAE can strain generation and transmission capacity. Facilities that shift cooling load off-peak contribute to grid stability and may receive preferential treatment from utilities in terms of connection agreements and tariff negotiations. TES also provides a buffer during grid disturbances: a fully charged ice or chilled water tank can sustain cooling for 2 to 6 hours depending on the system size, supplementing battery energy storage and backup generators in maintaining facility operations.

District Cooling Integration

The UAE has the largest district cooling infrastructure in the world. Many district cooling plants already incorporate TES. Data centers connected to district cooling systems benefit indirectly from the provider's TES capacity, which smooths supply and reduces the risk of capacity constraints during extreme heat events. For operator-owned cooling plants, adding TES capacity mirrors the approach that district cooling operators have proven at scale.

Sizing and Integration Considerations

Integrating TES into a data center cooling plant requires balancing several variables against the facility's operational profile.

Full Storage vs. Partial Storage

In a full storage design, the TES system is sized to handle the entire peak cooling load, allowing all chillers to shut down during peak periods. This maximizes demand charge reduction but requires large tank capacity and higher capital investment.

In a partial storage design, the TES system supplements chillers during peak periods. Chillers run at reduced capacity while the TES discharges to meet the remainder of the load. This is the more common approach for data centers because it requires smaller tanks, lower capital cost, and still captures most of the demand charge and energy cost savings.

Factor	Full Storage	Partial Storage
Peak demand reduction	100% of cooling load	30-60% of cooling load
Tank volume	Large	Moderate
Capital cost	Higher	Lower
Demand charge savings	Maximum	Significant
Payback period	5-8 years	3-5 years
Cooling redundancy	Hours of standalone capacity	Supplemental buffer

Controls and Automation

TES operation requires integration with the building management system (BMS) or data center infrastructure management (DCIM) platform. The control system must manage charge and discharge cycles based on tariff schedules, predicted cooling demand, weather forecasts, and real-time chiller efficiency. Predictive algorithms that account for the following day's weather forecast and IT load profile can optimize charge schedules to minimize total energy cost while maintaining cooling reliability.

Space and Structural Requirements

Chilled water TES tanks are heavy structures. A 4,000 cubic meter tank filled with water weighs approximately 4,000 metric tons, requiring appropriate foundation design. For existing facilities, structural assessments may limit tank size. Ice storage systems with their smaller footprint offer a practical retrofit path for facilities that cannot accommodate large water tanks.

Greenfield data center developments in the UAE increasingly incorporate TES capacity into site master plans from the outset, allocating dedicated areas adjacent to chiller plants for tank installation. This is particularly common in the industrial zones of Abu Dhabi, Dubai South, and KEZAD where land is available at reasonable cost.

TES and PUE Impact

Thermal energy storage does not directly reduce the total cooling energy consumed. The same amount of heat must be removed from the IT load regardless of when the chiller operates. However, TES improves power usage effectiveness (PUE) through several indirect mechanisms.

• Nighttime chiller operation at higher COP: Chillers running at night benefit from lower ambient wet-bulb temperatures, improving condenser performance and increasing COP by 10 to 20 percent compared to afternoon operation.
• Reduced chiller cycling: Continuous operation at optimal loading is more efficient than cycling chillers to match variable load.
• Lower parasitic loads: Smaller chiller plants and reduced condenser fan operation during peak hours reduce auxiliary power consumption.

For a facility with a baseline PUE of 1.5 in a hot climate, TES integration combined with optimized chiller scheduling can contribute to a PUE reduction of 0.03 to 0.08, moving closer to the 1.3 to 1.4 range that represents good practice for high-density data centers in hot climates.

Case Study: District Cooling TES in the UAE

Empower, the world's largest district cooling services provider headquartered in Dubai, operates chilled water TES tanks at multiple plants across the emirate. Their Business Bay plant includes stratified TES capacity exceeding 30,000 ton-hours, allowing the plant to shift a significant portion of its chiller operation to nighttime off-peak hours while meeting afternoon peak demand from stored chilled water. This approach reduces peak grid demand by thousands of kilowatts across the network and provides cooling continuity during scheduled maintenance windows.

For data center operators, connecting to district cooling systems that incorporate TES provides the economic benefits of storage without the capital investment and operational complexity of building and maintaining a dedicated TES system. Operators who build their own cooling plants can replicate the same principles at facility scale, with partial storage systems offering the most accessible entry point.

Combining TES with Other Energy Strategies

Thermal energy storage works best as one component of a comprehensive energy management strategy for hot-climate data centers. Complementary approaches include:

• Battery energy storage (BESS): While TES handles the cooling load, battery systems can shift IT power consumption and provide grid services, together reducing total peak demand across the facility.
• Solar integration: In the UAE, solar generation peaks in the afternoon when cooling demand is also highest. Pairing rooftop or adjacent solar with TES allows the solar output to power IT loads while stored cooling handles the thermal load, minimizing grid draw during the most expensive hours.
• Waste heat recovery: For facilities exploring waste heat recovery, absorption chillers driven by recovered heat can charge TES systems, creating a pathway to reduce both grid dependency and cooling costs.

How Rax Approaches Thermal Efficiency in UAE Hosting

Rax designs its UAE data center operations with thermal efficiency as a core infrastructure principle. That means evaluating TES integration for new capacity, working within DEWA and ADDC tariff frameworks to optimize chiller scheduling, and providing tenants with competitive power rates that reflect the operational efficiencies achieved through smart cooling management. For operators seeking colocation in the UAE and GCC where cooling costs represent a substantial portion of total hosting expense, thermal energy management is a meaningful differentiator.

For operators evaluating the broader UAE power and tariff landscape, thermal energy storage is one component of a comprehensive cooling and power strategy that also includes chiller plant design, humidity management, and long-term energy procurement planning.