Data centre energy control

Abstract

A method for determining a configuration of a data centre data hall comprises creating model domains having N server cabinets arranged away from edges of the domain, and at least N+1 fan wall units (F

Description

Field of the Invention The field of the invention related to control strategy for data server halls cooling units in data centres. The invention is applicable to but not limited to determining the layout of a data hall with improved energy efficiency. Background of the Invention A data centre is an infrastructure building that holds servers, processors, and other electronic devices in addition to a backup power source such as uninterruptible power supply (UPS). The servers are typically maintained in cabinets, which are known as racks. Within a data centre, these racks are arranged in rows by being placed next to one another. The aisles are created by placing these rows front-to-front and back-to-back. The cooled air can be supplied through these aisles, and they can also be used to provide space for practical uses. Data centre design is categorised into four tiers by the Uptime Institute, each representing different levels of infrastructure redundancy, availability, and resilience. These tiers help determine the reliability and performance of a data centre. A Tier 1 (Basic Capacity) entry-level classification features a single path for power and cooling distribution without redundant components, ensuring 99.671% uptime, which translates to about 28.8 hours of downtime annually. It is suitable for small businesses or non-critical operations where occasional downtime is acceptable. Tier II (Redundant Capacity Infrastructure) in infrastructure still relies on a single path for power and cooling but includes redundant components to enhance reliability. This tier offers 99.741% uptime, equating to up to 22 hours of downtime per year, making it ideal for businesses needing more reliability than Tier I. Tier III (Concurrently Maintainable) represents a significant improvement in ensuring continuous availability. It features multiple power and cooling distribution paths, with only one active at any time, and redundant components. Tier III allows maintenance without disrupting operations, ensuring 99.982% uptime and limiting downtime to just 1.6 hours annually, suitable for businesses requiring continuous availability, whereas Tier 4 does ensure 99.995%. Typical cooling systems provided are based around air cooled chillers feeding chilled water to cooling equipment, such as Computer Room Air Handler (CRAH), Computer Room Air Conditioning (CRAC) and Fan Wall Units (FWU) located within the data halls. The data server halls (server halls in data centre) in data centres are configured in accordance with end-user requirements and are closely monitored to ensure they meet the strict server inlet temperature guidelines laid out in the Service Level Agreements (SLA) which are not normally shared by the tenant, and as a result, the industry typically provides a good degree of margin between the server inlet temperature and the SLA. One of the more popular cooling arrangements in data centres to minimise recirculation problems is the so-called cold aisle-hot aisle containment. These supply cold air and return warm air from the servers and are divided into two regions as shown in the Figure 1. As illustrated in Figure 1, the hot aisle zone 104 includes server racks 130, 132, 134, 136, kitchen area 118, entrance 122, tape library 116, storage 114, and cooling units 120. Similarly, the cold aisle zone 102 includes server racks 106, 108, 110, 112. The hot aisle containment zone 104, separates the back of its respective server rack rows, while the cold aisle containment zone 102 separates the row of racks in front. As such, cold air will be drawn in front of the server racks 106, 108, 110, 112 from the cold aisle containment zone 102 and hot air is exhausted from the back of these server racks. They are positioned with the server rack front facing the cold aisle and the back facing the hot aisle. HAC Ceiling Duct (Hot Aisle Containment Ceiling Duct): are components of a hot aisle containment (HAC) system. They collect hot air released from IT equipment and channel it to the return plenum or cooling system. This setup prevents the mixing of hot and cold air, which boosts cooling efficiency and reduces energy use. Fan Wall Units (FWU) are becoming more popular for cooling because they are more energy-efficient, manage airflow better, and can easily adapt to the needs of modern high-density data centres. Unlike CRAC and CRAH units, FWUs use fans that can move a lot of air without using too much energy, especially when the outside temperature is right for economizer cooling. They create a consistent airflow throughout the data hall, which helps to reduce hot spots and improve hot / cold aisle containment. FWUs are also modular and can be scaled up easily, taking up less space on the floor compared to traditional units. They usually come with electronically commutated (EC) fans that are quieter, more dependable, and simpler to maintain. Plus, with advanced sensors, they can adjust cooling based on the current server loads. Sensors are usually placed in front of the FWUs to control airflow based on the reading they receive, which can be either temperature or pressure reading. Often, a group of the FWU sensors will operate its airflow using the group’s average mean value from their sensors, which ensures that the units perform uniformly in terms of airflow. Figure 2(a) and Figure 2(b) illustrate an example of a typical data hall that utilises FWlls for cooling. A shown the data hall 200 has false ceiling 202, HAC ceiling duct 204, beam 206, riser 208, IT cabinet 210, support column 212, and FWU 214. Figure 2(b) is a cross-sectional view of data hall 200. Showing FWU corridor 216, with FWU 214 in the corridor, separation wall as 218, between the corridor and server cabinets, and the false ceiling 202, and actual ceiling 220. A false ceiling acts as a plenum for containing warm air and provides space for cables, and air ducts systems. Fan wall units 214 are big cooling systems found in data centres, as shown in Figure 3. They feature several fans set up in a wall-like formation, delivering a large amount of low-speed airflow to keep temperatures just right. These units usually work in conjunction with hot and cold aisle containment systems to ensure that IT equipment stays cool and efficient. As shown, the FWU 214 has a supply side 322, and a return side 324. According to the selected specifications, a typical FWU 214 has a cooling capacity of 650 kW, making it suitable for high-density data centre environments. It operates with an airflow rate of 42,250 L / s, ensuring sufficient air circulation across the server racks. The airflow is controlled with a pressure differential of 12 Pa means airflow of the unit controlled based on sensor reading, allowing precise management of air distribution. Usually, these sensors are placed in the white space of the data hall. The Supply Air Temperature (SAT) is maintained at 27.5°C, while the Return Air Temperature (RAT) is 41 °C, resulting in a AT (temperature difference) of 13.5°C. This significant AT indicates efficient heat removal from the IT equipment. The physical dimensions of the FWU are 1,480 mm (L) x 3,960 mm (W) x 3,760 mm (H), highlighting its large form factor, which accommodates multiple fans for airflow regulation. The image shows the FWU Supply side, where cooled air is delivered into the white space, and the FWU Return side, where warm air from the hot aisle is drawn back into the cooling system. IT cabinets 400, as shown in figure 4, commonly known as server racks, are standardized boxes that hold servers, networking gear, and other IT devices. They are built for organized cable management, security, and proper airflow. Typically, these cabinets have perforated doors at the front and back to enhance cooling efficiency. The cabinet 400 supports a variable IT load and utilises dT-based airflow control with a dT setpoint of 13.5 K, ensuring efficient temperature regulation. It is designed with 42U slots, making it compatible with standard full-sized server racks, and has an airflow leakage of 5%, which indicates minimal efficiency loss. With dimensions of 1,380 x 600 x 2,050 mm, the cabinet is structured to provide an optimal balance between space utilisation and thermal management, making it a suitable solution for high-density IT environments. A ll-slot, also known as a rack unit (II), is a standard unit of measurement used to describe the height of equipment that fits into a server cabinet or rack. One II is equal to 1.75 inches (44.45 mm) in height. Server cabinets and racks are typically designed in increments of U-slots to accommodate networking, storage, and computing equipment. For example, a 42U server cabinet (as shown in your image) means the cabinet can hold up to 42 rack-mounted devices, depending on their individual heights. Equipment sizes are commonly 1U, 2U, 4U, etc., where a 1U server takes up 1.75 inches of vertical space, a 2U server takes up 3.5 inches, and so on. This standardized sizing ensures compatibility across different IT infrastructure setups. Overall, while CRAH-based HAC systems remain effective and widely used, FWU single-sided designs offer superior energy efficiency, simplified airflow management, and easier maintenance, making them an attractive option for modern data centres aiming to optimize operational performance. Thermal guidelines in data centres are set by the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) in TC9.9, as shown in figure 5. The Data Centre Power Equipment Thermal Guidelines and Best Practices. However, specific tenants may have their own more stringent requirements. TC 9.9 provides advice on how to improve power equipment thermal compatibility and reliability, referring to the allowable temperature guidelines. When increasing the supply temperatures, it is vital that the room conditions are maintained within the ASHRAE class shown in Figure 06. This shows dry bulb temperature, relative humidity, wet bulb temperature, and dew point temperature. In the chart are regions A1-A4, and a recommended operating zone. According to ASHRAE TC9.9 Reference Card for 2021 Equipment Thermal Guidelines for Data Processing Environments, Class A1 is defined as “typically a data centre with tightly controlled environmental parameters” “and mission critical operations” (ASHRAE, 2021). Whereas classes A2 / A3 / A4 are characterised as IT spaces with some environmental parameter controls. A2 has the narrowest requirements, and A4 the widest. Typically, the products designed for these environments are servers, computers and storage products. In traditional data centres, as mentioned above, cold air is provided by the CRAC unit, where heat exchange occurs between the hot air and a coil. However, this traditional type of cooling system is limited to cooling racks producing up to 8 kW each. On the other hand, high-density data centres produce up to 30-40 kW per racks which makes traditional cooling methods inefficient. With Al developments, this can reach up to 10OkW or more per rack. Also, using a traditional CRAC flooded cooling system causes difficulty in maintaining the data centre within the ASHRAE temperature limits. Therefore, new cooling techniques such as liquid cooling should be implemented to cool down the high-density server racks. The cooling process of circulating water is approximately 1000 times better than a fan circulating the air. Watercooling techniques are becoming more common in the high-density data centres. M.K Patterson et al. categorised the liquid cooling implications into three different types as follows: 1) Rear Door Cooling: In this type, the liquid cooled heat exchanger sits in the rear of the cabinet where the server exhausts the hot air. (In this case active or passive cooling methods can be used.) 2) Cooling Pipes to CPU known as Direct Liquid Cooling: In this case, cooling pipes can be taken directly near to the heat source, such as the CPU, of each server and the heat is thereby removed. 3) Upper or Lower Heat Exchanger Configuration: In this case, the heat exchanger can be placed in the top or bottom of the server rack and cold air will be supplied. The water will be circulated through the heat exchanger to remove the heat from the server. Direct-to-Chip (DTC) cooling greatly cuts down on the reliance on traditional air cooling for servers in data halls, but it doesn't completely get rid of it. The amount of air cooling necessary still depends on the design of the DTC system and the overall cooling plan of the data centre. In DTC systems, a large portion of the heat produced by processors like CPUs, GPUs, and RAM is captured by liquid-cooled cold plates that are directly connected to the chips. This technique can capture up to 90% of the overall heat generated by the server through a liquid loop. Even though DTC cooling is effective, components like power supplies, storage drives, and network cards still depend on airflow to get rid of heat. In addition, when the liquid system moves the heat, some still hangs around in the server room. This leftover heat raises the need for air cooling to maintain a balanced temperature. Additionally, good air circulation is crucial for controlling humidity and preventing condensation, which helps keep everything running smoothly. To meet these requirements, many modern data centres use a mixed cooling approach. In this setup, DTC systems take care of the heat from the chips, while low-power air conditioning units deal with the leftover heat and the overall environment. Strategies like hot / cold aisle containment help improve efficiency by managing airflow better. In summary, while DTC cooling cuts down on the need for traditional air cooling, it doesn’t completely eliminate it. The combination of liquid and air cooling is key to ensuring top performance, reliability, and energy efficiency in today’s data centres. Summary of the Invention According to an embodiment there is provided a computer implemented method for determining a configuration of a data hall inside a data centre with improved energy efficiency comprising the steps of: creating a plurality of model domains for the data hall design, each domain having a height, width and depth, and floor and ceiling configuration; wherein each model domain is provided with a representation of a plurality, N, of server cabinets arranged on the floor, away from the edges of the model domain, and a representation of a plurality of at least N+1 fan wall units (FWU) to maintain a defined temperature within the model domain, wherein the FWU units are configured to provide 44% of the cooling requirement of the model domain; wherein the representations of the FWU are arranged in an array of alternating pairs of adjacent FWUs and single FWUs, with an additional single FWU at the end of the array, so there are a pair of adjacent FWU at one end of the array, and two single FWUs at the other end of the array, and the FWUs are all located on a same side of the data hall, spaced away from the server cabinets; on the floor of the data hall, wherein the supply air temperature of each of the FWU is set to be no greater 27.5°C, and the return air temperature of each of the FWU is set to be no greater than 41 °C; splitting the model domain into one or more separate analysis zones; establishing a pressure base mean airflow control for each representation of the plurality of FWUs in the data hall, using data about the airflow control in one or more governing computational fluid dynamic equations for each analysis zone; analysing the one or more analysis zones using the governing equations, to determine temperature and pressure profiles for each of the N server cabinets in the one or more analysis zones, when all of the FWU in the data hall are fully operational, determining temperature and pressure profiles for each of the N server cabinets in the data hall , when at least one of the FWU has failed, independent of the analysis zone that the failed FWU is located in; repeating this analysis for at least one of: different configurations of the server cabinets, different configurations of the FWUs, different airflow control data in the one or more analysis zones of the model domain of the data hall, to obtain results for when the FWU are fully operational, and when there is a failure of at least one FWU in the modal domain; comparing the repeated failure / no failure results from all the analysis zones to provide an output of the optimum configuration of server cabinets and FWU in the model domain of the data hall across the analysis zones to maximise the energy efficiency of the data hall so that the data hall can continue to operate in the event of failure of one FWU. Preferably the pressure-based airflow control is based on pressure difference data recorded via sensors for each representation of the plurality of FWUs in the data hall. Preferably, a set point for pressure based airflow control from the sensors is 12Pa. In a preferred embodiment, the capacity for each zone of the data centre is : 6MW, or a multiple of 6MW. Further preferably, wherein each zone is provided with N or N+ fan wall units. Further preferably, the total redundancy of fan wall units per data hall does not exceed N+2 N+2 wall fan units. Preferably, the plurality of model domains are further provided with a representation of one or more riser elements, and one or more ceiling ducts. Further preferably, the maximum temperature in the temperature profile of the server cabinets is 35°C. In a preferred embodiment of the invention, the governing computational fluid dynamic equations for each analysis zone comprise equations using parameters of temperature, velocity and pressure. Preferably, the governing computational fluid dynamic equations are solved using RANS model. Further preferably, the RANS model is a RANS K-epsilon model. In a preferred embodiment, when there is a failure of at least one FWU, and the temperature of at least one of the N server cabinets exceeds 35°C, due to the FWU failure, the server cabinet has failed and will no longer operate until the temperature of the server cabinet is reduced below 35°C. Preferably, the power supplied to the plurality, N, of server cabinets is distributed across the model domain for the data hall in either alternating high and low density per HAC or equally distributed among each row. Further preferably, total power supplied to server cabinets is less than the total cooling calculations.. These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter. Brief Description of the Drawings Further details, aspects and embodiments of the invention will be described, by way of example only, with reference to the drawings. In the drawings, like reference numbers are used to identify like or functionally similar elements. Elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale. Figure 1: Illustration of standard Hot and Cold Aisle Containment (John N, et. al., 2011): Figure 2(a): illustrates an isometric view of a Data Hall: Figure 2(b) illustrates a sectional View of Data Hall; Figure 3: illustrates a standard Fan Wall Cooling Unit for use in data halls: Figure 4: illustrates a standard Liquid Cooled Server Cabinet for use in data halls; Figure 5: Illustrate ASHRAE allowable temperature ranges (ASHRAE, 2016); Figure 6(a): Shows an example 6MW data hall configuration according to an embodiment; Figure 6(b): Shows an example 12MW data hall configuration according to an embodiment; Figure 6(c): Shows an example 18MW data hall configuration according to an embodiment; Figure 7: Example tables of load distribution and cooling capacity: Figure 8: End Aisle worst case server heat load Figure 9(a): Example Methodology for Research Conducted according to an embodiment; Figure 9(b) is a flow chart for a method according to an example embodiment Figure 9(c) is a continuation of the flow chart of figure 9(b): Figure 10 is an example Methodology for Proposed Strategy Working Principle Figure 11: is an example of zoning of Data Hall into Three zones for 18MW Figure 12: is an example of Zoning of Data Hall into Multiple Zones presented for 12MW Figure 13(a): is a cabinet mean inlet temperature plot for 6MW data Hall at scenario 1Aof Table 1: Figure 13(b): is a pressure plane for 6MW Data Hall at scenario 1A of Table 1: Figure 14(a) is a cabinet mean inlet temperature plot for 6MW data hall at scenario 1B of Table 1: Figure 14 (b): is a pressure plane for 6MW data hall at scenario 1B of Table 1: Figure 15(a): is a cabinet mean inlet temperature plot for 12MW data Hall at scenario 2A of Table 1: Figure 15(b): is a pressure plane for 12MW Data Hall at scenario 2A of Table 1: Figure 16(a): is a cabinet mean inlet temperature plot for 12MW data hall at scenario 2Bof Table 1: Figure 16(b): is a pressure plane for 12MWdata hall at scenario 2B of Table 1: Figure 17(a): is a cabinet mean inlet temperature plot for 12MW data hall at scenario 2C of Table 1: Figure 17(b): is a pressure plane for 12MWdata hall at scenario 2C of Table 1: Figure 18(a): is a cabinet mean inlet temperature plot for 12MW data hall at scenario 2D of Table 1: Figure 18(b): is a pressure plane for 12MWdata hall at scenario 2D of Table 1: Figure 19(a): is a cabinet mean inlet temperature plot for 12MW data hall at scenario 2Eof Table 1: Figure 19(b): is a pressure plane for 12MWdata hall at scenario 2E of Table 1: Figure 20(a): is a cabinet mean inlet temperature plot for 18MW data Hall at scenario 3A of Table 1: Figure 20(b): is a pressure plane for 18MW Data Hall at scenario 3A of Table 1: Figure 21(a): is a cabinet mean inlet temperature plot for 18MW data hall at scenario 3B of Table 1: Figure 21(b): is a pressure plane for 18MWdata hall at scenario 3B of Table 1: Figure 22(a): is a cabinet mean inlet temperature plot for 18MW data hall at scenario 3C of Table 1; Figure 22(b): is a pressure plane for 18MWdata hall at scenario 3C of Table 1; Figure 23(a): is a cabinet mean inlet temperature plot for 18MW data hall at scenario 3D of Table 1; Figure 23(b): is pressure plane for 18MWdata hall at scenario 3D of Table 1; Figure 24(a) is a cabinet mean inlet temperature plot for 18MW data hall at scenario 3E of Table 1; Figure 24(b): is a pressure plane for 18MW data hall at scenario 3E of Table 1; Detailed Description Despite the advantages of liquid cooling systems, the integration of liquid cooling systems into data halls and data centres presents challenges. The initial capital expenditure for liquid cooling infrastructure is higher compared to traditional air-cooling systems, which may deter some data centre operators. Additionally, the complexity of maintenance and the need for specialised knowledge can pose operational hurdles. However, as Al continues to drive up computational requirements, the efficiency gains from liquid cooling are expected to offset these challenges. An embodiment of the invention is concerned with providing a control strategy for use in data centre facilities through the use of FWlls control strategy and zoning their multiple units in single zone and having more than one zone of control in single data hall. Optimising the LC data centre halls to work under conditions when redundant cooling units not working is crucial. The inventors of this application have conducted various different simulations based on key infrastructure elements, including risers, fan wall units (FWlls), IT cabinets, and ceiling ducts in the data halls of the data centre. The choice of a 650- kW per row load distribution in the server racks was driven by sensor placements, ensuring accurate temperature and pressure readings for each of the server racks. Different simulation scenarios examined various alternative failure conditions, focusing on different pressure zones and FWU failures. Key findings indicated that single-row cooling solutions should avoid adjacent FWU failures to prevent temperature spikes and pressure imbalances. The study recommended implementing zonal controls with N+1 redundancy, ensuring that cooling capacity exceeds IT demand by at least 5% to maintain positive pressure in data halls. Advanced computation-based tools were used in this study to evaluate the working principle of server room operations in such these temperatures. Technological uncertainty lay in how to achieve improvements on the industry LC (Liquid Cooled) Data Halls standard for energy efficiency through making alterations to FWU control strategy on air cooling side within data centre cooling systems. The following research investigated airside aspects of cooling optimisation methods. Hence, the study showed that Computation fluid dynamics tools and design optimisation techniques will improve performance of the Data centre beyond the industrial standards benchmark as stated in the Uptime institute (Uptime Institute, 2024). The advance was achieved through computational fluid dynamics (CFD) to validate different arrangements of the FWU, either at a single zone and with multiple zones, and multiple different scenarios including FWU failure conditions and one unit failure per zone or multiple in single zone failure CFD simulations. In the methods as discussed below, a 44% air-cooled, and 56% liquid-cooled data hall is considered. By tackling heat right at its source, the demand on standard FWU is greatly minimised, leading to reduced energy use and improved PUE (Power Usage Effectiveness). The inventors have performed extensive testing and research on the relationship between Liquid cooling FWU control strategy and server inlet temperature using computational fluid dynamics (CFD) modelling. Testing included: varying the MW capacity data hall server LC room from 6MW, 12MW and 18MW with FWU off-coil temperature of 27.5°C (as shown in figures 6(a)-(c)). Preferably, the capacity for each zone of the data centre is : 6MW, or a multiple of 6MW. N+1 and N+2 refers to the redundancy of the system, N+1 means there is 1 additional backup unit so if one FWU unit fails, the system can still continue to maintain the cooling capacity required to keep the data hall within the allowable temperature range. At N+2 there are 2 additional units providing a higher level of redundancy if two units were to fail at once. This also means that a lower MW capacity cooler may be used at N+1 redundancy when compared to N+2 as there are more units running to meet the combined required cooling capacity. Preferably, each zone is provided with N or N+1 fan wall units. Further preferably, the total redundancy of fan wall units per data hall does not exceed N+2. The tables shown in figure 7 highlight key parameters, including IT capacity, liquid cooling (LC%), air cooling (AC%) and cooling infrastructure. In this case, LC refers to the liquid cooling load of 56%, while AC represents 44% of the cooling load. The heat for IT load and cooling is detailed under LC kW and AC kW, with the total load kW representing their combined energy requirement. The number of IT rows, maximum row total IT load (Liquid + Air Cooling) (680kW) remains consistent and whereas minimum rack loading as 500kW (Liquid + Air Cooling), while the number of fan wall units (FWU) increases with capacity, with the "+1" or "+2" indicating redundancy for reliability. Figure 8 shows an example rack power loading layout in a data hall, where each row has a total load of 255 kW. A noticeable feature is the higher load of 17.93 kW on one side end aisle row only is considered to represent the worst loading which could occur in Data Hall. This uneven distribution in only single row in whole data hall is commonly driven by worst-case scenario planning assumed as, a standard practice among data centre operators to ensure infrastructure resilience under peak conditions. The load distribution demonstrates a more dynamic allocation, with some racks operating at 17.93 kW and others at 11.6 kW. This variable load approach results in a 255-kW row total per row and an overall IT load of 2000 kW. Such a configuration aligns cooling capacity with workload variations, improving efficiency and energy management. Key observations include the improved efficiency in distribution, where cooling is directed to high-load areas rather than being uniformly spread. As IT capacity increases from 6MW, 12MW to 18MW, the number of FWU’s scales accordingly to meet the cooling demand. The combination of liquid cooling and air cooling ensures effective heat dissipation, while the flexed load approach optimizes power and cooling for variable workloads, enhancing operational efficiency and system reliability. This design philosophy aligns with modern data centre practices aimed at improving PUE and streamlining airflow management through CFD analysis. As nature of the study focused on the CFD air side only, PUE energy efficiency comparison is not evaluated. This was used to determine whether an increased server inlet temperature could support a sufficient reduction in the energy consumption of the data centre facility and if the increase would remain within thermal guidelines. Figure 9(a): is a flow chart that represents a structured methodology 900 for conducted research: Start, at step 902: Marking the initiation of a systematic investigation into a control strategy for FWU inside data hall is addressed with design suggestions. Literature review step 904: The research reviewed the papers on current cooling techniques, temperature and pressure controls, air conditioning methods, energy optimisation and including the electronic enclosures. The need for further development of high-density liquid-cooled data hall FWU control strategy is also being addressed. Finding Gap in the Field, step 906: Based on detailed literature review key gap is identified. Model Built-Up: Once the research gap is identified, the next step is model built up, step 910, which serves as an initial framework for further refinements and scenario testing in CFD. This preliminary model helps establish benchmarks for comparison on ensure working principles of data hall. Setting Up Boundary Conditions step 912: involves defining constraints and parameters to ensure the model operates within realistic and applicable limits. Setting Up the Number of Scenarios Required step 908: determines the different test cases or scenarios needed to analyse varying conditions and ensure a comprehensive evaluation. This includes the N only and N+R operational conditions. Convergence, step 916: A critical step in the methodology, where the model is checked for stability and consistency in producing reliable results. If the model does not converge, adjustments are made, and the process loops back to refining the baseline model. Once convergence is achieved, the next step is Checking Results Against Project Requirements, step 918, where the outcomes are evaluated to determine if they meet the desired objectives. If the results fail to meet the project criteria, the process moves to changing the boundary condition to identify suitable Scenario, where a different case is analysed until a satisfactory outcome is found. If the results meet the project requirements, the data is Saved and Post Processed, step 924 for further analysis. Finally, the Comparison of Results to Find the Best Solution, step 926 is conducted by analysing the data from multiple tested scenarios and selecting the most effective and optimised outcome. This structured methodology ensures a systematic and iterative approach to research or computational modelling, facilitating thorough testing, refinement, and the selection of the best possible solution. Figure 9(b) is a flowchart of an example process 990 according to an embodiment. In some implementations, one or more process blocks of Fig. 9(b) may be performed by an apparatus. As shown in Fig. 9(b), process 990 may include creating a plurality of model domains (950) for the data hall design, each domain having a height, width and depth, and floor and ceiling configuration (block 952). For example, apparatus may create a plurality of model domains for the data hall design, each domain having a height, width and depth, and floor and ceiling configuration, as described above. As also shown in Fig. 9(b), process 990 may include where each model domain is provided with a representation of a plurality, N, of server cabinets arranged on the floor, away from the edges of the model domain, and a representation of a plurality of at least N+1 fan wall units (FWU) for temperature control within the model domain, where the FWU units are configured to provide 44% of the cooling requirement of the model domain of the data hall out of 100% liquid cooled (block 952). For example, each model domain is provided with a representation of a plurality, n, of server cabinets arranged on the floor, away from the edges of the model domain, and a representation of a plurality of at least n+1 fan wall units (FWU) for temperature control within the model domain, where the FWU units are configured to provide 44% of the cooling requirement of the model domain of the data hall out of 100% liquid cooled, as described above. As further shown in figure 9(b), process 954 may include where the representations of the FWU are arranged in an array of alternating pairs of adjacent FWUs and single FWUs, with an additional single FWU at the end of the array, so there are a pair of adjacent FWU at one end of the array, and two single FWUs at the other end of the array, and the FWUs are all located on a same side of the data hall, spaced away from the server cabinets; on the floor of the data hall, where the supply air temperature of each of the FWU is set to be no greater 27.5oC, and the return air temperature of each of the FWU is set to be no greater than 41 oC (block 956). For example, the apparatus may determine where the representations of the FWU are arranged in an array of alternate pairs of adjacent FWlls and single FWlls, with an additional single FWU at the end of the array, so there are a pair of adjacent FWU at one end of the array, and two single FWUs at the other end of the array, and the FWUs are all located on a same side of the data hall, spaced away from the server cabinets; on the floor of the data hall, where the supply air temperature of each of the FWU is set to be no greater 27.5oc, and the return air temperature of each of the FWU is set to be no greater than 41 oc, as described above. As also shown in Fig. 9(b), process 956 may include splitting the model domain into one or more separate analysis zones (block 956). For example, the apparatus may split the model domain into one or more separate analysis zones, as described above. As further shown in Fig. 9(b), process 958 may include establishing a pressure temperature base mean airflow control for each representation of the plurality of FWUs in the data hall, (block 958). For example, the apparatus may establish a pressure temperature base mean airflow control for each representation of the plurality of FWUs in the data hall, as described above. As also shown in Fig. 9(b), process 960 may include using data about the airflow control in one or more governing computational fluid dynamic equations for each analysis zone (block 960). For example, apparatus may use data about the airflow control in one or more governing computational fluid dynamic equations for each analysis zone, as described above. As further shown in Fig. 9(b), process 962 may include analysing the one or more analysis zones using the governing equations, to determine temperature and pressure profiles for each of the N server cabinets in the one or more analysis zones, when all of the FWU in the data hall are fully operational, (block 962). For example, apparatus may analyse the one or more analysis zones using the governing equations, to determine temperature and pressure profiles for each of the n server cabinets in the one or more analysis zones, when all of the FWU in the data hall are fully operational, as described above. As also shown in Fig. 9(b), process 964 may include determining temperature and pressure profiles for each of the N server cabinets in the data hall, when at least one of the FWU has failed, independent of the analysis zone that the failed FWU is located in (block 964). For example, apparatus may determine temperature and pressure profiles for each of the n server cabinets in the data hall, when at least one of the FWU has failed, independent of the analysis zone that the failed FWU is located in, as described above. As further shown in Fig. 9(b), process 966 may include repeating this analysis for at least one of: different configurations of the server cabinets, different configurations of the FWUs, different airflow control data in the one or more analysis zones of the model domain of the data hall, to obtain results for when the FWU are fully operational, and when there is a failure of at least one FWU in the modal domain (block 966). For example, apparatus may repeat this analysis for at least one of: different configurations of the server cabinets, different configurations of the FWUs, different airflow control data in the one or more analysis zones of the model domain of the data hall, to obtain results for when the FWU are fully operational, and when there is a failure of at least one FWU in the modal domain, as described above. As also shown in Fig. 9(c), process 968 may include comparing the repeated failure / no failure results from all the analysis zones to provide an output of the optimum configuration of server cabinets and FWU in the model domain of the data hall across the analysis zones to maximise the energy efficiency of the data hall so that the data hall can continue to operate in the event of failure of one FWU (block 968). For example, apparatus may compare the repeated failure / no failure results from all the analysis zones to provide an output of the optimum configuration of server cabinets and FWU in the model domain of the data hall across the analysis zones to maximise the energy efficiency of the data hall so that the data hall can continue to operate in the event of failure of one FWU, as described above. Although Fig. 9(b) and Fig 9(c) shows example blocks of process 990, in some implementations, process 900 may include additional blocks, fewer blocks, different blocks, or differently arranged blocks than those depicted in Fig. 9(b). Additionally, or alternatively, two or more of the blocks of process 900 may be performed in parallel. Computational Fluid Dynamics is the numerical solution used to analyse fluid dynamics problems such as those in the data centre racks. CFD solves the energy, momentum, and continuity equations in the numerical way, using superpower computers. The governing equation of the velocity, temperature and pressure can be described below as: Continuity equation + div(p u) = 0 (1) X, Y and Z momentum equation: The momentum equation is derived from Newton’s second law (F=ma; Force = Mass. Acceleration) X Momentum Equation: + div(puU) = — + div(pgrad u) + Su (2) Y Momentum Equation: + div(pvU) = — + div(pgrad v) + Sv (3) Z Momentum Equation: + div(pwU) = — + div(pgrad w) + Sw (4) Energy Equation: The energy equation is derived from the 1st law of thermodynamics which states that the net heat transferred to the control volume + network done by the control volume = difference in energy control volume. p Cp + div {pCp T if) = — p div U + k div (grad T) + Q (5) The density will be constant for an incompressible flow. So therefore, the term becomes as below: ^=0 (6) dr v ' The equation above represents the steady simulation, the time dependent variable ignored in this case. Where: p - density (kg / m3) u — x direction velocity (m / s) v - y direction velocity (m / s) w - z direction velocity (m / s) p - pressure (Pa) Su - The source term in x-direction. The permeability effect in the X direction in the simulation (in racks and servers). Sv - The source term in y-direction. The permeability effect in the Y direction in the simulation (in racks and servers). Sw - The source term in z-direction. The permeability effect in the Z direction in the simulation (in racks and servers). Cp - is the specific heat (kJ / kg. K) T - temperature (K) U - velocity vector (U =ui +vj+ wk) k - thermal conductivity (W / m. K) Q - energy source term (kW / m3) p. - dynamic viscosity (kg / m. s) Preferably, the governing computational fluid dynamic equations for each analysis zone comprise equations using parameters of temperature, velocity and pressure. Further preferably, the governing computational fluid dynamic equations are solved using RANS model. In an embodiment, the RANS model is a RANS K-epsilon model. Solving the governing equation is very costly, and solving it computationally requires very high-powered computer processors. This method of solving the governing equations is known as direct numerical simulation. With the resources available for this current study, the perfect solution has been proposed in this section. To solve the governing equations for the turbulence model, it is important to consider the fluctuating term. The fluctuating part will be added to each part (which has an averaging part and a fluctuating part) in the governing equations. These new governing equations are called RANS (Reynolds Averaged Navier-Stoke Equations) The CFD analysis consists of 3 different stages such as Pre-Processing, Processing and Post-Processing. The pre-processing stage involves making a domain for a proposed design and meshing the domain. In the meshing process, the domain will be divided into number of cells and the discretised governing equations will solve the solution on each nodal points of the cells divided. The accuracy of the solution depends on the number of cells in the total domain. More cells mean much better accuracy, and vice versa. The processing stage where the 6Sigma CFD Tool will solve the discretised governing equations in each nodal points of the cell (finite control volumes). The solution method will be solved in more detail in an upcoming section. Currently the Software known as Cadence Reality DC (Cadence.com, 2024) In the post processing stage, the processed solution will be analysed in quantitative and qualitative results such as the following 2D-3D results, which were presented in results section. In this current research following below methodology flow is used to evaluate the thermal behaviour inside the data hall. Evaluating the max accepted temperature uplift at the cabinet level were evaluated in this study along with the effects on the pressure within the data hall. There has been an iterative process held as benchmark to conduct this study as shown in the flow chart in Figure 10 Cooling units in data centres, known here as Fan Wall Units are typically arranged individually or grouped to manage the thermal environment. This study explores an advanced approach by implementing zoning on top of group controls. Zoning divides the data hall into subgroups of FWUs, allowing precise cooling adjustments based on the specific cooling demands of each zone. The process is detailed in the flowchart of figure 10. This flow chart shows methodology 1000. Step 1002 is the start, followed by step 1004 assign multiple differential pressure sensors to the model. Step 1006 assign equipment to different zones in the model. Step 1008 establish an array of differential pressure centres per zone. Step 1010 set as cooling air supply air control. Step 1012 take average differential pressure sensor readings. Step 1014 compare the readings from the differential pressure sensor to the set point. Step 1016 is the setpoint meant. If yes proceed to step 1018 maintain cooling air flow from the unit. If no proceed to step 1020 is the reading above the set point. If yes proceed to step 1024 cooling unit ramps down if no proceed to step 1022 calling unit ramps up. Steps of 1022 and 1024 returned back to step1012. These steps are further explained below. Steps 1002 and 1004, FWU Units &Differential Pressure (dP) Sensors: FWU units are equipped with multiple differential pressure (dP) sensors to monitor airflow conditions within the data hall. These sensors provide real-time feedback for controlling the cooling output. Preferably, the pressure base mean airflow control is based on pressure difference data recorded via sensors for each representation of the plurality of FWUs in the data hall. Further preferably, the maximum value set for pressure mean airflow control setpoint is 12Pa. Figure 11 illustrates the zoning of a data hall into Three zones presented for 18MW, 1106, 1107 and 1108. In an embodiment, each zone of the data hall is provided with N+2 wall fan units. Preferably, the one or more distinct zones are further provided with a representation of one or more riser elements, and one or more ceiling ducts. Preferably, the maximum temperature in the temperature profile of the server cabinets is 35° C. Preferably, power supplied to the plurality, N, of server cabinets is distributed across the model domain for the data hall in either alternating high and low density per HAC or equally distributed among each row. In an embodiment, total power supplied to server cabinets is less than the total cooling calculations. Figure12 illustrates the zoning of Data Hall into Multiple Zones presented for 12MW and 18MW, 1206 and 1207. Assign Equipment to Zones step 1006: The data hall is divided into zones based on the cooling needs of specific equipment. Each zone includes a dedicated set of FWU units, ensuring targeted cooling delivery to areas with higher heat loads or specific airflow requirements. As presented on Figure 10, the data halls zones can be divided shown in the 106, 107 and 108. Zoning doesn’t cover physical obstruction but just control methods of FWlls using sensors. Whereas Figure 9 represents the same for Data Hall with 12MW capacity. Establish an Array of dP Sensors per Zone, step 1008: Within each zone, an array of dP sensors is established to take a group dP sensor reading. This zoning strategy enables localized monitoring in a specific area of the data hall. Set as Cooling UnitSupply Air Control, step 1010: ThedP sensor arrays are integrated into the supply air control system of the FWU units belonging in their respective zones. This integration ensures that the cooling output responds dynamically to the airflow requirements of each zone. Take Average dP Sensor Reading, step 1012: The system continuously collects dP readings from each zone and calculates an average value. This approach smooths out anomalies and provides a more stable reference for airflow control. In an analysis zone of a data hall there will be multiple dP sensors and their reading will be averaged over that zone of the data hall that the sensor is assigned to. This average value will determine if the FWUs will ramp the airflow up or down to reach its target pressure setpoint. In a preferred embodiment, the target pressure set point is 12 Pa. Compare Reading to Set Point, step 1014: The average dP sensor reading is compared to a predefined set point, representing the optimal pressure for efficient cooling. This step ensures that the cooling system maintains the desired airflow balance. Decision Point: If the Set Point Met, step 1016: If the average reading matches the set point, the system maintains the current cooling unit airflow, ensuring energy-efficient operation without unnecessary adjustments. If it Above Set Point, step 1020: If the dP reading exceeds the set point, it indicates reduced airflow efficiency, prompting the cooling units in associated zone to ramp down, thus reducing fan speed and energy consumption. If the reading falls below the set point, the system ramps up the cooling units to increase airflow and maintain optimal conditions within the zone. This circles back to taking average dP sensor reading. Maintain Cooling Unit Airflow, step 1018: Once the set point is achieved, the system stabilizes the cooling output, ensuring consistent environmental conditions across the data hall. This zoning-based FWU cooling strategy enhances the efficiency and responsiveness of liquid-cooled data centres. By targeting cooling based on specific zone requirements, it ensures optimal thermal management while reducing energy consumption, ultimately contributing to a lower Power Usage Effectiveness (PUE). According to an embodiment a computer implemented method for determining a configuration of a data hall inside a data centre with improved energy efficiency is described. Preferably, the method comprising the steps of: creating a plurality of model domains for the data hall design, each domain having a height, width and depth, and floor and ceiling configuration; wherein each model domain is provided with a representation of a plurality, N, of server cabinets arranged on the floor, away from the edges of the model domain, and a representation of a plurality of at least N+1 fan wall units (FWU) for temperature control within the model domain, wherein the FWU units are configured to provide 44% of the cooling requirement of the model domain; wherein the representations of the FWU are arranged in an array of alternating pairs of adjacent FWUs and single FWUs, with an additional single FWU at the end of the array, so there are a pair of adjacent FWU at one end of the array, and two single FWUs at the other end of the array, and the FWUs are all located on a same side of the data hall, spaced away from the server cabinets; on the floor of the data hall, wherein the supply air temperature of each of the FWU is set to be no greater 27.5oC, and the return air temperature of each of the FWU is set to be no greater than 41 oC; splitting the model domain into one or more separate analysis zones; establishing a pressure base mean airflow control for each representation of the plurality of FWUs in the data hall, using data about the airflow control in one or more governing computational fluid dynamic equations for each analysis zone; analysing the one or more analysis zones using the governing equations, to determine temperature and pressure profiles for each of the N server cabinets in the one or more analysis zones, when all of the FWU in the data hall are fully operational, determining temperature and pressure profiles for each of the N server cabinets in the data hall , when at least one of the FWU has failed, independent of the analysis zone that the failed FWU is located in; repeating this analysis for at least one of: different configurations of the server cabinets, different configurations of the FWUs, different airflow control data in the one or more analysis zones of the model domain of the data hall, to obtain results for when the FWU are fully operational, and when there is a failure of at least one FWU in the modal domain; comparing the repeated failure / no failure results from all the analysis zones to provide an output of the optimum configuration of server cabinets and FWU in the model domain of the data hall across the analysis zones to maximise the energy efficiency of the data hall so that the data hall can continue to operate in the event of failure of one FWU. The study is conducted in three phases, each focusing on the control strategy and zoning configurations. In this comprehensive study, we aim to validate the working principles of thermal zone controls by employing modern Computational Fluid Dynamics (CFD) tools through a structured, phased approach. Phase 1 is dedicated to a single thermal zone configuration, with the objective of establishing a robust baseline. This phase involves a detailed examination of the performance of single-zone control under both adjacent and non-adjacent fan wall unit (FWU) failure scenarios, with a particular focus on analysing cabinet-level inlet temperatures and data hall pressure profiles. Phase 2 transitions to a more complex multiple thermal zone configuration, where all failures are concentrated within a single zone. The goal here is to rigorously evaluate the effectiveness of thermal zone controls by comparing the results against the established criteria from Phase 1. Phase 3 further extends the analysis by assessing the scalability and efficiency of the zoning approach in a longer data hall, with a maximum of one FWU failure per zone. This phase aims to determine the robustness and resilience of the system under distributed failure conditions. Through this comprehensive and methodical analysis, the target is to gain valuable insights into the performance, reliability, and overall effectiveness of thermal zone controls in data hall environments. Several challenges arise, particularly in the context of cooling management and failure response, when data halls are not divided into multiple zones. One significant issue is during the incident of having an adjacent FWU failure, which can lead to substantial cooling capacity loss in the affected local area. In a scenario where FWlls are individually controlled, the units farther from the failed FWU struggle to detect the cooling discrepancy. When a failure occurs, the nearby FWUs attempt to compensate for the lost capacity while those farther away continue to operate as if conditions are normal. This localized response is often insufficient, leading to potential overheating and equipment damage. Grouping FWUs and linking their controls to a single sensor array can mitigate these issues. A unified control system allows all FWUs to respond collectively to a failure, compensating for the lost capacity across the entire data hall. This approach ensures a more balanced and efficient cooling response, reducing the risk of localized overheating. Despite the advantages of group control, challenges remain in longer data halls. When adjacent FWUs fail, units that are significantly distant from the failure site may not effectively deliver cooling to the affected area. Instead, they might oversupply cooling to their immediate vicinity, leading to inefficiencies, potential pressure builds up and cooling imbalances. Results PHASE FLOW - SIMULATION SCENARIOS (Table 1) Phase ID / Scenarios DH Capacity Description Service Level Agreement (SLA) Criteria: 35°C Pressure Readings Across Cold Aisles, Pa Max Min 1A 6MW Normal Operation Pass 14.8 7.3 1B End most Unit Off; Failure Operation Pass 19.1 7.7 2A 12MW Normal Operation Pass 13.0 3.0 2B End most adjacent units Off; Failure Operation Fail 27.7 -16.6 2C Non-adjacent unit Off; Failure Operation Pass 25.8 -4.4 2D 2-zone configuration; Both failed units in same zone Fail 18.0 -9.5 2E 2-zone configuration; Single failed unit per zone Pass 20.3 -2.8 3A 18MW Normal Operation Pass 16.6 6.9 3B End most adjacent units Off; Failure Operation Fail 32.7 -9.3 3C Non-adjacent unit Off; Failure Operation Pass 31.4 -0.8 3D 3-zone configuration; Non-adjacent failed units in same zone Fail 9.9 -9.8 3E 3-zone Control; Singel failed unit per zone Pass 13.1 -1.7 Three different data hall capacities - 6MW, 12MW, and 18MW, were studied via a three-phase approach. These data halls increase in length according to their capacities. 6MWis configured in an N+1 setup, while 12MWand 18MWare in an N+2 configuration, meaning two fan wall units (FWlls) are to fail during failure mode. All data halls are set up to work on a grouped control strategy wherein the performance of all FWlls depends on the average readings of differential pressures placed across the end of the room opposite the FWU supply. For all phases, failing the end most fan wall unit / s did not meet the passing cabinet level inlet temperature criteria. This failure configuration is deemed to be the worstcase scenario since the only thing next to the failed FWlls on the end will be a wall, with no FWU to support it on either side, as opposed to failing the middle FWUs. Figures 13-24 show the results for the various scenarios described in Table 1 above. Figure 13(a) shows temperature plot for 6MW data Hall at scenario 1A, and Figure 13(b) shows pressure plane for6MW Data Hall at scenario 1A. In scenario 1A the data hall has a 6MW capacity, and there is normal operation of all components. As shown the temperature at the server racks is below 32°C, and the pressure ranges from 7.3 to 14.8 Pa. Figure 14(a) shows temperature Plot for 6MW data hall at scenario 1B, and Figure 14(b) shows pressure plane for6MW Data Hall at scenario 1B. In scenario 1B the data hall has a 6MW capacity, and there is a failure of one of the FWU indicated at 1402, at the far-left side of the data hall. As shown the temperature at the server racks is below 32°C, and the pressure range from 7.7 to 19.1 Pa. The increase in the pressure leads to the designation of this scenario as a failure. For Phase 1 - 6MW data hall, failing the end most FWU resulted in IT cabinet mean inlet temperatures well below the 35°C threshold, and pressure readings within the data hall remained below 20Pa. Figure 15(a) shows temperature plot for 12MW data hall at scenario 2A, and Figure 15(b) shows pressure plane for 12MW Data Hall at scenario 2A. In scenario 2A the data hall has a 12MW capacity, and there is normal operation of all components. As shown the temperature at the server racks is below 32°C, and the pressure ranges from 3.0 to 13.0 Pa. This scenario results in normal operation of the data hall Figure 16(a) shows temperature plot for 12MW data hall at scenario 2B, and Figure 16(b) shows pressure plane for 12MW Data Hall at scenario 2B. In scenario 2B the data hall has a 12MW capacity, and there is a failure of the 2no. FWlls indicated at 1602 and 1604, located at the end of the data hall. As shown the temperature at the server racks in some of the server racks 1606, exceeds 32°C, and the temperature in the remaining server racks is below 32°C, and the pressure ranges from -16.6 to 27.7Pa. The increase in the pressure range and temperature for some of the servers in the data hall leads to the designation of this scenario as a failure. Figure 17(a) shows temperature plot for 12MW data hall at scenario 2C, and Figure 17(b) shows pressure plane for 12MW Data Hall at scenario 2C. In scenario 2C the data hall has a 12MW capacity, and there is a failure of the 2no. non-adjacent FWlls indicated in 1702 and 1704, where one is located at the end of the data hall, and one is located more centrally within the data hall. As shown the temperature at the server racks does not exceeds 32°C, and the pressure in the entire data hall ranges from -4.4 to 25.8Pa. The increase in the pressure range for some of the servers in the data hall leads to the designation of this scenario as a failure. Figure 18(a) shows a temperature plot for 12MW data hall at scenario 2D, and Figure 18(b) shows pressure plane for 12MW Data Hall at scenario 2D. In scenario 2D the data hall has a 12MW capacity, with 2 distinct measurement zones. There is a failure of the 2no. non-adjacent FWlls in the same measurement zone, indicated in 1802 and 1804, where one is located at the end of the data hall, and one is located more centrally within the first measurement zone of the data hall. As shown the temperature at some of the server racks in the zone where there is failure of the FWU exceeds 34.9°C, and there is a range of temperatures over the entire data hall between 32°C to 34.9°C. The pressure in the entire data hall ranges from -9.5 to 18.0Pa. The increase in the rack temperature and pressure range for the servers in the data hall leads to the designation of this scenario as a failure. Figure 19(a) shows a temperature plot for 12MW data hall at scenario 2E, and Figure 19(b) shows pressure plane for 12MW Data Hall at scenario 2E. In scenario 2E the data hall has a 12MW capacity, with 2 distinct measurement zones. There is a failure of one FWU in each of the two measurement zones, indicated in 1902 and 1904, where one is located at the end of the first zone of the data hall, and the other one is located in the second measurement zone of the data hall. As shown the temperature at some of the server racks in the first zone where there is failure of the FWU is in the range of 32 to 34.9°C, but the server rack temperatures in the second zone are all still less than 32°C. The pressure in the entire data hall ranges from -2.8 to 20.3 Pa. The temperature and pressure range for the servers in the data hall in this scenario leads to this scenario being considered as a pass. For 12MW data hall, failing the leftmost FWU and its adjacent unit did not result in any failed IT cabinets. However, high pressure (max recorded = 27.7Pa) was observed at the right-most end of the data hall opposite the failed units. Negative pressure was also observed in cold aisles near the failed units (min recorded = -16.6Pa). An alternate failing strategy was conducted wherein the leftmost unit and the third unit to the right failed. This strategy aimed to provide a buffer between the failed units with working FWUs and hopefully support the failed units. IT cabinet performance during this failure improved, and pressure also balanced out, with the highest recorded at 25.8Pa and the lowest at -4.4Pa. Although pressure recorded was still high, there was a reduction in the buildup of negative pressure within the data hall. Figure 20(a) shows temperature plot for 18MW data hall at scenario 3A, and Figure 20(b) shows pressure plane for 18MW Data Hall at scenario 3A. In scenario 3A the data hall has a 18MW capacity and is divided into three measurement zones. In this scenario, there is normal operation of all components in the data hall. As shown the temperature at all the server racks is below 32°C, and the pressure ranges from 6.9 to 16.6Pa. This scenario is considered passing. Figure 21(a) shows a temperature plot for 18MW data Hall at scenario 3B, and Figure 21(b) shows pressure plane for 18MW Data Hall at scenario 3B. In scenario 3B, the data hall has a 18MW capacity, with 3 distinct measurement zones. There is a failure of two adjacent FWUs in the first measurement one, indicated in 2102 and 2104, where these are located at one end of the first zone of the data hall. As shown, the temperature at some of the server racks in the data hall where there is failure of the FWU is, is greater than 34.9°C, for some server racks the temperature is in the range of 32 to 34.9°C, and the remainder of the server racks have temperatures less than 32°C. The pressure in the entire data hall ranges from -9.3 to 32.7 Pa. The temperature and pressure range for the servers in the data hall in this scenario leads to this scenario being considered as a failure. Figure 22(a) shows a temperature plot for 18MW data Hall at scenario 3C, and Figure 22(b) shows pressure plane for 18MW Data Hall at scenario 3C. In scenario 3C the data hall has a 18MW capacity, with 1 distinct measurement zones. In this scenario 3C there is a failure of two non-adjacent FWlls in the measurement zone, indicated at 2202 and 2204, but not adjacent to each other. As shown, the temperature at some of the server racks in the data hall where there is failure of the FWlls is, in the range of 32 to 34.9°C, and the remainder of the server racks have temperatures less than 32°C. There are no server racks with temperatures greater than 34.9°C, despite the failure of two FWlls. The pressure in the entire data hall ranges from -0.8 to 31.4 Pa. The temperature and pressure range for the servers in the data hall in this scenario leads to this scenario being considered as a failure. Figure 23(a) shows a temperature plot for 18MW data Hall at scenario 3D, and Figure 23(b) shows pressure plane for 18MW Data Hall at scenario 3D. In scenario 3D the data hall has a 18MW capacity, with 3 distinct measurement zones. In this scenario 3D there is a failure of two non-adjacent FWUs in the third measurement zone, indicated at 2302 and 2304, where these are located in the final right most measurement zone, but not adjacent to each other. As shown, the temperature at some of the server racks in the data hall where there is failure of the FWU is, is greater than 34.9°C, for some server racks the temperature is in the range of 32 to 34.9°C, and the remainder of the server racks have temperatures less than 32°C. The pressure in the entire data hall ranges from -9.8 to 9.9Pa. The temperature and pressure range for the servers in the data hall in this scenario leads to this scenario being considered as a failure. Figure 24(a) shows a temperature plot for 18MW data Hall at scenario 3E, and Figure 24(b) shows pressure plane for 18MW Data Hall at scenario 3D. In scenario 3E the data hall has a 18MW capacity, with 3 distinct measurement zones. In this scenario 3E there is a failure of a FWU at the end of the second measurement zone, as well as a failure of a FWU at the end of the third measurement zone, indicated at 2402 and 2404, As shown, the temperature of all the server racks in the data hall where there is failure of the FWU remains less than 32°C. The pressure in the entire data hall ranges from -1.7 to 13.1 Pa. The temperature and pressure range for the servers in the data hall in this scenario leads to this scenario being considered as passing. In all of these scenarios, when there is a failure of at least one FWU, and the temperature of at least one of the N server cabinets exceeds 35°C, due to the FWU failure, the server cabinet has failed and will no longer operate until the temperature of the server cabinet is reduced below 35oC. It is during failure mode that these data halls experience either failing IT cabinet mean inlet temperatures or high pressures across the cold aisles, with the exception of 6MW model. Following the same failing configuration for 18MW data hall, results showed similar effects in terms of IT cabinet mean inlet temperatures and data hall pressure distribution. Failing the left-most FWU and its adjacent unit unsatisfactory IT cabinet temperatures and high pressure build up, both positive and negative, on both ends of the data hall. Failing non-adjacent units provided better IT cabinet temperatures and positive pressurization within the data hall however majority of the room was still exceeding 25Pa. For Phase 2, multiple-zoning control for fan wall units was introduced to the 12MWand 18MW data hall. For 12MW data hall, seven fan wall units were assigned to the first zone 1206, and six units were assigned to the second Zone 1207 as shown in figure 12. For the 18MW, three zones were implemented with the first two Zones, 1106 and 1107, having six fan wall units and the third Zone, 1108, having seven fan wall units as shown in figure 11. Due to 12MW-Zone 1206 and 18MW-Zone 1108 having seven units assigned to each, the two fan wall units were chosen to fail in these Zones. For the 12MW data hall with Zone 1206 experiencing two non-adjacent failures, the configuration resulted in worse IT cabinet mean inlet temperatures. However, in terms of pressure, it was more evenly distributed compared to Phase 1’s single-zone configuration. The highest pressure was recorded to be 18Pa. For the 18MW data hall, Zone 1108 experienced two non-adjacent failures. This configuration yielded similar results to the 12MWdata hall, with IT cabinet mean inlet temperatures worse than the single-zone configuration. However, pressure was more evenly distributed, with the highest recorded pressure being 9.9Pa. In Phase 3 - 12MW data hall, having one fan wall units fail in Zones 1206 and 1207 yielded net positive pressurization inside the data hall with all IT cabinet mean inlet temperatures below 35°C. Highest and lowest recorded pressure for this data hall in this configuration was 20.3Pa and -2.8Pa. In 18MW data hall, having one fan wall units fail in Zones 1107 and 1108 as in figure 24(a) yielded net positive pressurization inside the data hall with all IT cabinet mean inlet temperatures below 35°C. Highest and lowest recorded pressure for this data hall in this configuration was 13.1 Pa and -1.7Pa. The analysis of different data halls (6MW, 12MWand 18MW) configurations highlights the critical role of multi-zone control in ensuring both temperature compliance and realistic pressure conditions in data centres. While single-zone configurations may meet the 35°C temperature threshold under normal operation, they result in excessive pressure variations during fan wall unit (FWU) failures. This issue arises primarily due to the absence of pressure relief mechanisms, which are commonly employed in real-world data centre setups. Pressure relief dampers (PRDs) are crucial in data centres for managing airflow and maintaining stable pressure within the data hall. They automatically open or close based on pressure differences, preventing excessive positive or negative pressure that could affect equipment performance. In multi-zone setups, PRDs between zones and external walls maintain consistent pressure, ensuring safe and efficient data centre operation. Single-zone configurations demonstrate significant limitations, particularly under failure conditions. Although they maintain acceptable inlet temperatures, the lack of pressure relief dampers causes unreasonably high positive and negative pressure build-up across the data hall. Such conditions make single zone designs impractical for real-world applications, as they fail to provide the stable environment required for reliable data centre operations. As CFD work on closed loop programme volume conservation equation, so it did not predict this real-life pressure relief damper behaviour. In contrast, multi-zone control offers a more effective solution by distributing pressure evenly across the data hall. The study showed that multi-zone configurations in both the 12MW and 18MW setups significantly improved pressure balance under failure conditions, with maximum pressure readings of 20.3Pa and 13.1 Pa, respectively. This zoning approach ensures that, even during FWU failures, IT cabinet inlet temperatures remain within acceptable limits, safeguarding the performance and longevity of critical equipment. Effective failure management further underscores the advantages of multi-zone design. Distributing failure units across multiple zones enhances system resilience, as evidenced by the superior pressure balance and temperature control observed when one FWU failed per zone. Additionally, implementing multiple 6MW zones, rather than a single large zone of 12MW and 18MW, provides a scalable and robust solution for real-world data centre operations. Although the present invention has been described in connection with some example embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the scope of the present invention is limited only by the accompanying claims. Additionally, although a feature may appear to be described in connection with embodiments, one skilled in the art would recognize that various features of the described embodiments may be combined in accordance with the invention. In the claims, the term ‘comprising’ does not exclude the presence of other elements or steps. Furthermore, although individually listed, a plurality of means, elements or method steps may be implemented by, for example, a single unit or processor. Additionally, although individual features may be included in different claims, these may possibly be advantageously combined, and the inclusion in different claims does not imply that a combination of features is not feasible and / or advantageous. Also, the inclusion of a feature in one category of claims does not imply a limitation to this category but rather indicates that the feature is equally applicable to other claim categories, as appropriate. Furthermore, the order of features in the claims does not imply any specific order in which the features must be performed and the order of individual steps in a method claim does not imply that the steps must be performed in this order. Rather, the steps may be performed in any suitable order. In addition, singular references do not exclude a plurality. Thus, references to ‘a’, ‘an’, ‘first’, ‘second’, etc. do not preclude a plurality. Acronyms: UPS- Uninterruptible Power Supplies CRAH- Computer Room Air Handler CRAC- Computer Room Air Conditioning FWU- Fan Wall Units SLA- Service Level Agreements EC- Electronically Commutated SAT- Supply Air Temperature RAT- Return Air Temperature △T- Temperature Difference HAC- Hot Aisle Containment PUE- Power Usage Effectiveness CAC- Cold Aisle Containment DTC- Direct-to-Chip CFD- Computational Fluid Dynamics dp- Differential Pressure PRD- Pressure Relief Dampers LC- Liquid Cooled AC- Air Cooled RANS- Reynolds Averaged Navier-Stoke Equations ASHRAE- American Society of Heating, Refrigerating and Air-Conditioning Engineers CPU- Central Processing Unit References - Uptime Institute. (2019). Data Center Efficiency Metrics and Standards. - Uptime Institute. (2023). Large data centres are mostly more efficient, analysis confirms - Uptime Institute blog. Available at: https: / / journal.uptimeinstitute.com / large-data-centers-are-mostly-more-efficient-analysis-confirms / (Accessed: 03 February 2024). - John N, Kevin B, Victor A (2011),’ Impact of Hot and Cold Aisle Containment on Data Centre Temperature and Efficiency’ pp. 1-14 - Uptime Institute. (2018). Data Centre Site Infrastructure Tier Standard - Versteeg H K, and Malalasekera W (2007),’ An Introduction to computational fluid dynamics: the finite volume method. 2nd edition. Pearson education limited, ISBN 978-0-13-127498-3. - Cadence.com. (2024). Cadence Reality DC Design, [online] Available at: 5 https: / / www.cadence.com / en_US / home / resources / product-briefs / cadence- reality-dc-design-pb.html [Accessed 3 Feb. 2024], - R. Sethuramalingam, A. Asthana, S. Xygkaki, K. Kiu, J. Eduardo, S. Wilson, C. Bater, (2023). Energy Demand Reduction in DataCentres using Computational Fluid Dynamics. 10 - ASHRAE. (2021). 2021 Equipment thermal guidelines for data processing environments. https: / / www.ashrae.orq / file%20library / technical%20resources / bookstore / suppl emental%20files / therm-gdlns-5th-r-e-refcard.pdf - ASHRAE (2016). ASHRAE TC9.9 - Data Centre power equipment thermal 15 guidelines and best practices

Claims

1. A computer implemented method for determining a configuration of a data hall inside a data centre with improved energy efficiency comprising the steps of: creating a plurality of model domains for a data hall design, each domain having a height, width and depth, and floor and ceiling configuration;wherein each model domain is provided with a representation of a plurality, N, of server cabinets arranged on the floor, away from edges of the model domain, and a representation of a plurality of at least N+1 fan wall units (FWU) to maintain a defined temperature within the model domain, wherein the FWU units are configured to provide 44% of cooling requirements of the model domain;wherein the representations of the FWU are arranged in an array of alternating pairs of adjacent FWUs and single FWUs, with an additional single FWU at an end of the array, so there are a pair of adjacent FWU at one end of the array, and two single FWUs at the other end of the array, and the FWUs are all located on a same side of the data hall, spaced away from the server cabinets; on the floor of the data hall, wherein supply air temperature of each of the FWU is set to be no greater 27.5°C, and return air temperature of each of the FWU is set to be no greater than 41 °C;splitting the model domain into one or more separate analysis zones; establishing an airflow control based on average pressure for each representation of the plurality of FWUs in the data hall, using data about the airflow control in one or more governing computational fluid dynamic equations for each analysis zone;analysing the one or more analysis zones using the governing equations, to determine temperature and pressure profiles for each of the N server cabinets in the one or more analysis zones, when all of the FWU in the data hall are fully operational, determining temperature and pressure profiles for each of the N server cabinets in the data hall, when at least one of the FWU has failed, independent of the analysis zone that the failed FWU is located in;repeating this analysis for at least one of: different configurations of the server cabinets, different configurations of the FWUs, different airflow control data in the one or more analysis zones of the model domain of the data hall, to obtain results for when the FWU are fully operational, and when there is a failure of at least one FWU in the model domain;comparing repeated failure / no failure results from all the analysis zones to provide an output of a configuration of server cabinets and FWU in the model domain of the data hall across the analysis zones to maximise the energy efficiency of the data hall so that the data hall can continue to operate in the event of failure of one FWU.

2. The computer implemented method as claimed in claim 1 wherein the pressure-based airflow control is based on pressure difference data recorded via sensors for each representation of the plurality of FWUs in the data hall.

3. The computer implemented method as claimed in claim 2 wherein a setpoint for pressure-based airflow control from the sensors is 12Pa.

4. The computer implemented method as claimed in any preceding claim wherein the capacity for each zone of the data centre is: 6MW, or a multiple of 6MW.

5. The computer implemented method as claimed in any preceding claim wherein each zone is provided with N or N+1 I fan wall units.

6. The computer implemented method as claimed in any preceding claim wherein the total redundancy of fan wall units per data hall does not exceed N+2.

7. The computer implemented method as claimed in any preceding claim wherein the plurality of model domains are further provided with a representation of one or more riser elements, and one or more ceiling ducts.

8. The computer implemented method of any preceding claim wherein the maximum temperature in the temperature profile of the server cabinets is 35°C.

9. The computer implemented method of any preceding claim wherein the governing computational fluid dynamic equations for each analysis zone comprise equations using parameters of temperature, velocity and pressure.

10. The computer implemented method as claimed in claim 8 wherein the governing computational fluid dynamic equations are solved using RANS model.

11. The computer implemented method as claimed in claim 9 wherein the RANS model is a RANS K-epsilon model.

12. The computer implemented method as claimed in any preceding claim wherein5 when there is a failure of at least one FWU, and the temperature of at least oneof the N server cabinets exceeds 35°C, due to the FWU failure, the server cabinet has failed and will no longer operate until the temperature of the server cabinet is reduced below 35°C.10 13. The computer implemented method as claimed in any preceding claim whereinpower supplied to the plurality, N, of server cabinets is distributed across the model domain for the data hall in either alternating high and low density per HAC or equally distributed among each row.15 14. The computer implemented method as claimed in claim 12, wherein total powersupplied to server cabinets is less than the total cooling calculations.IntellectualPropertyOfficeApplication GB2505270.5Search report under Section 17 of the Patents Act 1977Date search completed: 20 October 2025Claims searched: 1-14International classificationSubclass and subgroup Valid from G06F30 / 18 01 / 01 / 2020 G06F30 / 28 01 / 01 / 2020Field of searchWorldwide search of patent documents classified in the following areas of the IPC:G06FDatabases used in the preparation of this search report:SEARCH-NPL; SEARCH-PATENTDocuments considered to be relevantPatent literatureCategory Relevant claims Document of relevance A - CN 118862736 A HENGHUA DIGITAL TECH GROUP CO LTD,Intellectual Property Office is an operating name of the Patent Office www.gov.uk / ipo37Non-patent literatureCategory Relevant Document of relevance claimsCategoriesLetter or symbol Description X Document indicating lack of novelty or inventive step.Y Document indicating lack of inventive step, if combined with another document of the same category. & Member of the same patent family. A Document indicating technological background. P Document published on or after the priority date but before the fling date of the present application. E Earlier application published on or after the filing date of the present application.

Images