Thermal Management Of Data Centers Research Articles

Contributions of microlayer to flow boiling heat transfer in the mini-channel

Flow boiling stands out as an efficient mode of heat transfer with significant potential for reducing carbon footprint. Based on a coupled Volume-of-Fluid and Level Set (VOSET) method, this numerical study reveals the microlayer and its evaporative heat flux distributions, as well as the effects of microlayer evaporation and depletion on bubble dynamic behavior, flow pattern, wall superheat distribution, and boiling crisis for the flow boiling in a mini-channel. These aspects are firstly comprehensively reported by the numerical method. Results indicate that the rapid evaporation of a thin microlayer positioned beneath elongated bubbles significantly contributes to heat dissipation during flow boiling in a mini-channel. The heat flux through the microlayer surpasses 300 kW/m2, accounting for approximately 42% of heat dissipation at 200 kW/m2. Furthermore, this contribution increases to over 70% as the heat flux exceeds 500 kW/m2. As a result, in comparison to scenarios without microlayers, the presence of microlayers results in an average heat transfer coefficient (HTC) that is 3.97 times larger within the range of 200–500 kW/m2. Additionally, microlayer evaporation delays the happening of flow boiling crisis. Without microlayers, the wall superheat exceeds 180 K at 500 kW/m2, whereas under microlayer influence, the maximum local wall superheat is only 37.09 K at 1000 kW/m2. With further increases in heat flux, the emergence of large dry patches resulting from microlayer depletion induces the beginning of heat transfer deterioration. Finally, the ongoing coalescence of elongated bubbles necessitates a prolonged period for rewetting dry patches at the exit of the mini-channel, triggering the flow boiling crisis at 1300 kW/m2. The study results provide crucial guidance for the development and application of the flow boiling heat transfer technology for the thermal management of data centers.

A novel integrated flat thermosyphon heat sink for energy-efficient chip-level thermal management in data centers

A novel Integrated Flat Thermosyphon Heat Sink (IFTHS) with enhanced performance is developed in the present study. The IFTHS has a condenser with integrated hollow fins wherein the inner surface of the hollow fins is used for condensation, and the outer surface is used for air-side convective cooling. The thermal performance of the IFTHS is investigated at different filling ratios to find the best-performing filling ratio. The thermal performance of the IFTHS is improved further through an evaporator with minichannels and superhydrophobic condenser. Superhydrophobic condenser improved the performance by 16 % due to quicker condensate return and dropwise condensation. Minichannel evaporator improved the performance by 19 % due to enhanced pool boiling from improved bubble dynamics and increased surface area. The combination of a superhydrophobic condenser and minichannel evaporator improved the performance by 36 %. The overall resistance, maximum temperature, and heat sink weight are reduced by 19–41 %, 8–25 °C, and 44–46 % with IFTHS configurations compared to the baseline conventional flat thermosyphon having a condenser with solid fins. The electronic component lifespan can be increased by 3.73–17.75 times, and the room air cooling load can be reduced by 8.04–25.12 kJ/kg when the IFTHS configurations are used for cooling electronic chips in data center servers instead of the baseline configuration. Thus, the novel IFTHS configurations developed in the present study can contribute to energy-efficient chip-level thermal management in energy-intensive data centers.

Optimization air-conditioning system and thermal management of data center via fan-wall free cooling technology

For data centers located in mild climate areas, the free cooling technology of the fan-wall air supply has great potential. However, there is currently little research on the optimization of air-conditioning system structural parameters and rack heat management in data centers using this technology. This paper studies the optimal design parameters of the internal structure of the air-conditioning system and determines the variation regularity of the inlet and outlet temperatures of the rack through the construction of a similar model for a fan-wall data center. The results show that: (1) The average supply heat index of the sieve with porosity of 60% is reduced by 57.5% and 50.3% compared to others, the average supply heat index of the fan-to-sieve distance of 35cm is reduced by 51.1% and 22.9% compared to others, and the average supply heat index of the fan wall arrangement in one column is reduced by 54.1% and 22.7% compared to others. (2) The two racks closest to the air supply port have the sharpest inlet temperature fluctuations, and the inlet temperature has stabilized from the third rack as the distance increases. (3) Hot spots in fan-wall data centers tend to occur in the lower part of the rack, and as the height decreases, the temperature of the inlet air increases.

Application of distributed parameter model in thermal management of microchannel loop heat pipe

Microchannel loop heat pipe (MCLHP) possesses the superiority of compact structure and energy saving, which are preferred in the thermal management of data centers and charging piles. In this study, the three-dimensional distributed parameter model combined with an experimental system was presented to investigate the effect of filling ratio, height difference, heat exchanger structure and operating parameters on the thermal transfer performance of the MCLHP system. Specially, pump auxiliary MCLHP was proposed to improve the heat transfer capacity. Simulations combined the distributed parameter model with the response surface methodology showed that the largest heat transfer capacity was 1.402 kW with a filling ratio of 79.7%. Although changing the structural parameters would improve the heat transfer capacity, it would compensate with increased space structure and air resistance. The proposed pump auxiliary MCLHP system could operate stably with the heat transfer capacity of up to 4 kW, which presented a potential application in high heat flux cooling such as charging piles and data centers.

Data center server workload and infrastructure control based on a joint RSM and CFD approach

The rapid expansion of data centers and cloud computing has exacerbated the issue of high energy consumption and localized overheating in servers. Consequently, the identification of effective cooling methods and resource allocation strategies for server rooms has become pivotal in enhancing overall airflow distribution and optimizing thermal performance within data centers. Time-efficient Computational Fluid Dynamics (CFD) tools offer an alternative to resource-intensive experimental measurements. However, it is difficult to monitor the information in real time and optimize the design of DC. In this study, an innovative application of Response Surface Methodology (RSM) is introduced to prediction thermal environment information and optimization performance of data center. In contrast to traditional single-factor optimization methods, this approach incorporates multiple factors, such as server workloads and air supply conditions, as optimization parameters. Furthermore, maximum server temperatures, temperature differences, and other metrics are defined as key performance indicators (KPIs). A predictive model was developed with the aim of offering comprehensive information about the thermal environment of the data center. Moreover, the study qualitatively analyzes the influence of each parameter on various indicators. Fitting equations are solved based on actual conditions to determine optimized configuration schemes in real time for the entire data center or specific scenarios. This approach effectively reduces cooling energy consumption and optimizes data center thermal management. The optimized configuration resulted in a significant reduction of approximately 30% in the maximum server temperature and approximately 20% in the temperature difference. Additionally, the optimization methodology established in this study facilitates the implementation of a real-time control mechanism for data center cooling systems, enabling energy demand and management optimization. The findings of this study offer valuable insights and generalized methods for enhancing the performance of air-cooled servers.

Improving energy and water consumption of a data center via air free-cooling economization: The effect weather on its performance

In the past two decades, data centers have significantly grown in size and number around the globe, leading to substantial energy and water stress wherever they operate. Thermal management typically represents half of the total energy consumption in traditional data centers. This work studies the energy and water savings that air-side free-cooling yields under diverse weather conditions. We modeled a free-cooling system and a conventional data center using MATLAB. The hourly weather data was obtained from a publicly available online repository for the Chilean territory. Using air-side free-cooling for data center thermal management is feasible in various weather conditions. Locations with colder temperatures (< 20 °C) and higher humidity (> 60%) are more favorable as they require minimal use of the chiller and/or humidifier. These conditions are present in BSk’(s) Cold Semi-arid climate with dry summer and oceanic influence, as well as Csb’ Mediterranean climate (warm summer) with oceanic influence. However, locations with drier air (< 50%) and warmer temperatures (> 20 °C), such as those found in ET(w) Tundra climate with dry winter, BWk(w) Cold Desert climate with dry winter, or BSk(s) Cold Semi-arid climate with dry summer, are less suitable for air free cooling. Even under suboptimal energy and water savings, the air-side economizer still performed considerably better than a conventional data center cooling method (using CRAH units).

High energy-density and power-density cold storage enabled by sorption thermal battery based on liquid-gas phase change process

Cold storage is essential for the preservation of food/medical goods, energy-saving of air conditioning, and emergency cooling. However, conventional cold storage in the form of sensible heat or solid-liquid latent heat suffers from the low energy density and large cold loss during long-term storage. To address the problem, a scalable sorption thermal battery (STB) with storage capacity of 30 kWh is designed and fabricated for realizing high energy-density and power-density cold storage by proposing sorption-induced liquid-gas phase change strategy. The STB utilizes the bonding energy of sorption working pair of zeolite 13X-water to achieve cold storage during charging phase, and utilizes the liquid-gas evaporation heat of water to realize cold production during discharging phase. The cold-storage performance of the STB prototype during charging/discharging processes is experimentally investigated at different operating conditions. By optimizing the assembled structure of solid-gas reactor and enhancing the heat transfer performance of liquid-gas evaporator, the STB exhibits high cold energy density up to 114.92 Wh/kg and 26.76 kWh/m3, and high power density of 455.62 W/kg and 106.10 kW/m3, over the conventional cold storage technologies. Correspondingly, the STB demonstrates a good performance in thermal management of data centers with cooling power as high as 150 kW. Moreover, the STB has the distinct advantage of near-zero cold energy loss in long-term cold storage. The proposed scalable sorption thermal battery based on sorption-induced liquid-gas evaporation heat of liquid offers a promising route to realize high energy density cold storage.

Deep reinforcement learning towards real-world dynamic thermal management of data centers

Deep Reinforcement Learning has been increasingly researched for Dynamic Thermal Management in Data Centers. However, existing works typically evaluate the performance of algorithms on a specific task, utilizing models or data trajectories without discussing in detail their implementation feasibility and their ability to deal with diverse work scenarios. The lack of these works limits the real-world deployment of Deep Reinforcement Learning. To this end, this paper comprehensively evaluates the strengths and limitations of state-of-the-art algorithms by conducting analytical and numerical studies. The analysis is conducted in four dimensions: algorithms, tasks, system dynamics, and knowledge transfer. As an inherent property, the sensitivity to algorithms settings is first evaluated in a simulated data center model. The ability to deal with various tasks and the sensitivity to reward functions are subsequently studied. The trade-off between constraints and power savings is identified by conducting ablation experiments. Next, the performance under different work scenarios is investigated, including various equipment, workload schedules, locations, and power densities. Finally, the transferability of algorithms across tasks and scenarios is also evaluated. The results show that actor-critic, off-policy, and model-based algorithms outperform others in optimality, robustness, and transferability. They can reduce violations and achieve around 8.84% power savings in some scenarios compared to the default controller. However, deploying these algorithms in real-world systems is challenging since they are sensitive to specific hyperparameters, reward functions, and work scenarios. Constraint violations and sample efficiency are some aspects that are still unsatisfactory. This paper presents our well-structured investigations, new findings, and challenges when deploying deep reinforcement learning in Data Centers.

Thermal prediction for Air-cooled data center using data Driven-based model

The optimal cooling control of data centers (DCs) relies on the thermal model to simulate and accurately evaluate the temperature distribution of the computer room. Data-driven thermal modeling methods have been widely used in the thermal management of DCs. However, few existing works have comprehensively compared and analyzed the predictive performance and adaptability of various data-driven thermal models. In addition, most of the proposed thermal models are for temperature prediction in steady-state scenarios of DCs, ignoring the study of thermal changes in transient scenarios. Therefore, this work builds a CFD model of a typical data center as a verification platform to compare the temperature prediction performance of six machine learning (ML)-based thermal models (SVR, GPR, XGBoost, LightGBM, ANN, LSTM) in steady-state and transient-state scenarios. For steady-state scenarios, we also explore the impact of sample size, physical layout reconfiguration, and airflow pattern on performance to evaluate the adaptability of the thermal model. For transient scenarios, we design four cooling failure scenarios to verify the ability of the proposed thermal model to capture the dynamic thermal changes of the computer room. Numerous numerical simulation results show that the XGBoost-based and LightGBM-based thermal models outperform other ML models in both steady-state and transient-state scenarios, with prediction error RMSE is less than 1.0 ℃, and the time overhead is less than 10-3 s. In addition, both thermal models are robust to physical layout reconfiguration and poor flow patterns.

DRL-S: Toward safe real-world learning of dynamic thermal management in data center

Deep Reinforcement Learning has been researched for Dynamic Thermal Management in Data Centers. An objective of Dynamic Thermal Management is to minimize power consumption while satisfying a set of hard and soft constraints. However, there is a lack of research on the safety issues of applying Deep Reinforcement Learning in Data Centers during real-world learning, limiting its deployment. To this end, this paper proposes a new method named DRL-S, which can solve both hard and soft constraints simultaneously during the learning stage. In addition to optimizing the main objective, Lagrangian-based Constrained Deep Reinforcement Learning and Reward Shaping enforce policies to satisfy soft constraints through extensive online learning. However, the policy typically fails to satisfy the hard constraints due to the random sampling of actions to encourage exploration and the imperfect policy at the initial online learning stage. To ensure hard constraints are satisfied, we further propose to utilize parameterized Shielding, integrating the approximation of the system dynamics and the projection of the action space to predict the safety of candidate actions and provide backup actions when necessary. Results show that the Lagrangian-based method and Reward Shaping can gradually learn policies to reduce soft constraint violations. The former can better balance the relationship between the main objective and violations by updating Lagrangian multipliers. DRL-S can also effectively avoid extreme temperatures without affecting the normal learning process of vanilla algorithms. The asymptotic power consumption is more than 12% lower than the baseline controller.

A Novel Multiscale Evaluation Metric for the Thermal Management of Data Centers

Abstract Tremendous burden is compelled on the energy consumption of data centers with the rapid development of information technology, and a healthy thermal environment is critical for improving the cooling efficiency of data centers. Furthermore, the thermal management of the data center is determined by the evaluation of its thermal environment. However, most of the existing evaluation metrics describe the thermal environment through a single scale, which cannot fully reflect the thermal performance of data centers at different scales. To simplify the evaluation of the thermal environment in data centers, a novel multiscale evaluation metric of the recirculation and hot spots index (RHSI) is proposed in this study. The numerical model of an operating data center is established and verified with the experiments on site. Then, the thermal environment of the data center is evaluated based on the RHSI and the existing metrics. Finally, the applicability of the RHSI is further discussed from different scales in detail. The results show that all bypass airflow, hot recirculation, and local hot spots can be derived from the proposed RHSI simultaneously by comparing with the evaluation metrics of supply heat index (SHI) and rack cooling indicator (RCI) at the room scale, β at the rack scale, and index of mixing (IOM) at the row scale. The proposed RHSI can provide a guiding significance for the thermal management of data centers.

Optimized thermal management of a battery energy-storage system (BESS) inspired by air-cooling inefficiency factor of data centers

Inspired by the ventilation system of data centers, we demonstrated a solution to improve the airflow distribution of a battery energy-storage system (BESS) that can significantly expedite the design and optimization iteration compared to the existing process. A defective cooling system of a BESS decreases the overall operational efficiency and increases the risk of thermal runaway, but current design optimizations rely on a case-by-case approach. The solutions of this fashion are both time-consuming and costly because of the laborious recursive process. The desire to eliminate the time for development motivates us to pursue a more general path for design optimization. In this work, we identified the similarity of geometry between the data center and the BESS, as well as the factors that induced the unbalanced airflow distribution. Inspired by the data-center thermal management, we propose a generalized solution of layout arrangement that we applied to the BESS design. We performed a quantitative analysis of cooling performance and investigated the flow patterns under given operating configurations. After modification, the maximum temperature difference of the battery cells drops from 31.2°C to 3.5°C, the average temperature decreases from 30.5°C to 24.7°C, and the coefficient of performance (COP) increases four-fold. The modification shows an improvement in temperature uniformity, overall temperature and COP. The cooling solution applicable to the general container BESS design demonstrates the enormous potential for an effective and rapid design optimization.

Discussions of Cold Plate Liquid Cooling Technology and Its Applications in Data Center Thermal Management

Discussions of Cold Plate Liquid Cooling Technology and Its Applications in Data Center Thermal Management

Experimental and Numerical Analysis of Data Center Pressure and Flow Fields Induced by Backward and Forward CRAH Technology

Abstract An increasingly common power saving practice in data center thermal management is to swap out air cooling unit blower fans with electronically commutated plug fans, Although, both are centrifugal blowers. The blade design changes: forward versus backward curved with peak static efficiencies of 60% and 75%, respectively, which results in operation power savings. The side effects of which are not fully understood. Therefore, it has become necessary to develop an overall understanding of backward curved blowers and compare the resulting flow, pressure, and temperature fields with forwarding curved ones in which the induced fields are characterized, compared, and visualized in a reference data center which may aid data center planning and operation when making the decisions of which computer room air handler (CRAH) technology to be used. In this study, experimental and numerical characterization of backward curved blowers is introduced. Then, a physics-based computational fluid dynamics model is built using the 6sigmaroom tool to predict/simulate the measured fields. Five different scenarios were applied at the room level for the experimental characterization of the cooling units and another two scenarios were applied for comparison and illustration of the interaction between different CRAH technologies. Four scenarios were used to characterize a CRAH with backward curved blowers, during which a CRAH with forwarding curved was powered off. An alternate arrangement was examined to quantify the effect of possible flow constraints on the backward curved blower's performance. Then parametric and sensitivity of the baseline modeling are investigated and considered. Different operating conditions are applied at the room level for experimental characterization, comparison, and illustration of the interaction between different CRAH technologies. The measured data is plotted and compared with the computational fluid dynamics (CFD) model assessment to visualize the fields of interest. The results show that the fields are highly dependent on CRAH technology. The tile to CRAH airflow ratios for the flow constraints of scenarios 1, 2, 3, and 4 are 85.5%, 83.9%, 61%, and 59%, respectively. The corresponding leakage ratios are 14.5%, 16%, 38.9%, and 41%, respectively. Furthermore, the validated CFD model was used to investigate and compare the airflow pattern and plenum pressure distribution. Lastly, it is notable that a potential side effect of backward curved technology is the creation of an airflow dead zone.

Numerical analysis for thermal management of data center with phase change material

PurposeThe purpose of this paper is to analyze the performance of a data center and thermal management by using phase change material (PCM). Numerical studies were conducted for two dimensional model of data center and installation of PCM at different locations.Design/methodology/approachFinite volume method was used for the unsteady problem, while impacts of air velocity and PCM location on the flow field, thermal pattern variations and phase change dynamics were evaluated. Three different locations of the PCM were considered while air velocity was also varied during the simulation. Thermal field variations and cooling performance of the system for different PCM location scenarios were compared.FindingsIt was observed that the installation of the PCM has significant impacts on the vortex formation, thermal field variation within the system and its performance. The left, right and top wall installation of the PCM changed the thermal patterns near the heat cell of the data centre. The phase change process is fast for the upper wall installation of the PCM, while the discrepancy of the melt fraction dynamics between different air flow at this position is minimum. The case where PCM placed in the upper wall at the highest air velocity is the best configuration in terms of heat storage. The utilization of PCM and changing its locations provide an excellent tool for thermal management and cooling performance of data centre.Originality/valueResults of this study can be used for initial design and optimization of cooling systems for thermal management of data centers while the importance of the high-performance computing becomes very crucial for the advanced simulations in different technological applications.

Ultra-Compact Microscale Heat Exchanger for Advanced Thermal Management in Data Centers

Abstract This study is focused on the experimental characterization of two-phase heat transfer performance and pressure drops within an ultracompact heat exchanger (UCHE) suitable for electronics cooling applications. In this specific work, the UCHE prototype is anticipated to be a critical component for realizing a new passive two-phase cooling technology for high-power server racks, as it is more compact and lighter weight than conventional heat exchangers. This technology makes use of a novel combination of thermosyphon loops, at the server-level and rack-level, to passively cool an entire rack. In the proposed two-phase cooling technology, a smaller form factor UCHE is used to transfer heat from the server-level thermosyphon cooling loop to the rack-level thermosyphon cooling loop, while a larger form factor UCHE is used to reject the total heat from the server rack into the facility-level cooling loop. The UCHE is composed of a double-side-copper finned plate enclosed in a stainless steel enclosure. The geometry of the fins and channels on both sides are optimized to enhance the heat transfer performance and flow stability while minimizing the pressure drops. These features make the UCHE the ideal component for thermosyphon cooling systems, where low pressure drops are required to achieve high passive flow circulation rates and thus achieve high critical heat flux values. The UCHE's thermal-hydraulic performance is first evaluated in a pump-driven system at the former laboratory of heat and mass transfer (LTCM)-Swiss Federal Institute of Technology of Lausanne, where experiments include many configurations and operating conditions. Then, the UCHE is installed and tested as the condenser of a thermosyphon loop that rejects heat to a pumped refrigerant system at Nokia Bell Labs, in which both sides operate with refrigerants in phase change (condensation-to-boiling). Experimental results demonstrate high thermal performance with a maximum heat dissipation density of 5455 (kW/m3/K), which is significantly larger than conventional air-cooled heat exchangers and liquid-cooled small pressing depth brazed plate heat exchangers. Finally, a thermal performance analysis is presented that provides guidelines in terms of heat density dissipations at the server- and rack-level when using passive two-phase cooling.

Optimization and Inverse Design of Floor Tile Airflow Distributions in Data Centers Using Response Surface Method

Abstract Computational fluid dynamics (CFD) has become a popular tool compared to experimental measurement for thermal management in data centers. However, it is very time-consuming and resource-intensive when used to model large-scale data centers, and often unrealistic for real-time thermal analyses. In addition, it is prohibitive to use CFD for the optimization process where thousands of designs need to be generated. Floor tile airflow distribution control is a key technique for maintaining a sufficient cold air delivery to variable thermal loadings of server cabinets. Regular practice of deploying a set of identical floor tiles may not result in the best solution for airflow uniformity through tiles. In this paper, an optimization procedure based on response surface methodology (RSM) is proposed to find the best arrangement of mixed-porosity floor tiles for different targeted tile airflow distributions. Fast-approximation RSM based on radial basis function (RBF) allows thousands of designs to be generated for the optimization process which uses a genetic algorithm as its main solver. The method shows proven success in maximizing floor tile airflow uniformity, and in the inverse design optimization where various tile airflow distribution topologies, i.e., linear, parabolic, and sinusoidal shapes are targeted. For the considered data center and aisle configuration, the improvement over the all-uniform-tile design is 55% in terms of standard deviation to the average tile airflow rate, whereas 90%, 91%, and 94% in root-mean-square error (RMSE) for the linear, parabolic, and sinusoidal floor tile airflow distribution objectives, respectively.

Datacenter Thermal Monitoring Without Blind Spots: FBG-Based Quasi-Distributed Sensing

Computing performance and hardware integrity are tightly linked to adequate thermal management in datacenters. In addition, a large share of its electricity bills comes from cooling systems, so there is a pressing need for enhanced thermal management systems. This paper proposes a fiber Bragg grating (FBG) quasi-distributed sensing device as an alternative to electronic sensors. The main focus is to demonstrate that a FBG-based solution allows thermal measurement at critical non-monitored regions (blind spots) within a server in an acquisition rate of 1 Hz, with thermal sensitivity of 8.757 pm/°C, a thermal resolution smaller than 1 °C and spatial resolution and multiplexing needed to monitoring diverse blind spots where there is virtually no room for adapting conventional sensors. In addition, we also monitor spots where there are already internal sensors next to them (referential spots), such as central processing unit (CPU) and graphics processing unit (GPU). As a result, a simple FBG sensor calibration technique is proposed using those referential spots. The technique proposed can be scaled out to monitor a large number of elements across large areas in a datacenter environment, since quasi-distributed FBG sensing signal can be transmitted over longer optical fibers than those employed in optical distributed temperature systems (DTS).

Thermal Prediction for Efficient Energy Management of Clouds Using Machine Learning

Thermal management in the hyper-scale cloud data centers is a critical problem. Increased host temperature creates hotspots which significantly increases cooling cost and affects reliability. Accurate prediction of host temperature is crucial for managing the resources effectively. Temperature estimation is a non-trivial problem due to thermal variations in the data center. Existing solutions for temperature estimation are inefficient due to their computational complexity and lack of accurate prediction. However, data-driven machine learning methods for temperature prediction is a promising approach. In this regard, we collect and study data from a private cloud and show the presence of thermal variations. We investigate several machine learning models to accurately predict the host temperature. Specifically, we propose a gradient boosting machine learning model for temperature prediction. The experiment results show that our model accurately predicts the temperature with the average RMSE value of 0.05 or an average prediction error of 2.38 °C, which is 6 °C less as compared to an existing theoretical model. In addition, we propose a dynamic scheduling algorithm to minimize the peak temperature of hosts. The results show that our algorithm reduces the peak temperature by 6.5 °C and consumes 34.5 percent less energy as compared to the baseline algorithm.

An open source fast fluid dynamics model for data center thermal management

Although computational fluid dynamics (CFD) has been widely adopted to improve data center thermal management, the high computational demand limits its applications, such as multivariate optimal design and operation. Fast fluid dynamics (FFD), which has been applied for fast airflow simulation, shows great potential. However, few research applied FFD for optimal design and operation of data center thermal management. This research improves the FFD model for data centers and conducts a comprehensive evaluation and demonstration. First, the FFD model is improved by solving the advection and diffusion equations together using an upwind scheme instead of a semi-Lagrangian advection solver in the conventional FFD model. Second, new features for data centers are added, such as a pressure correction method to simulate plenum airflow and dynamic boundary conditions for IT racks. The new FFD model is first validated with two indoor environment cases and the results show that the new FFD model has slightly better overall prediction accuracy and faster speed compared to the conventional FFD model. It is also observed that both FFD models achieve acceptable accuracy, except for a few localized disparities with experimental data, which might be due to simplified handling of turbulence viscosity near the boundaries. Furthermore, validation with a real data center shows that the FFD model achieves a similar level of accuracy as CFD when compared to the experimental measurements with some level of uncertainties. It is then demonstrated for data center optimal design and operation, which saves 53.4–58.8% of annual energy while still meeting the thermal requirements. With a much faster speed and comparable accuracy compared to CFD, the FFD model parallelized on a graphics processing unit is promising for practical model-based data center early design and operation.