Experimental Characterization of a Direct Contact Latent Cold Storage with Ice

Abstract

Effective thermal management is crucial for Hyperloop vehicles to ensure the reliable operation of onboard systems and to prevent overheating under high-speed and vacuum-like conditions. Due to the near-vacuum environment in which a Hyperloop operates, passive cooling is largely ineffective, making an active thermal management system necessary. This study investigates the application of a direct contact latent heat storage system, which leverages the high energy density of phase change materials. Ice is used as the phase change material and water as the heat transfer fluid, forming a system that avoids emulsion formation and simplifies design by eliminating complex heat exchangers. An experimental setup was used to evaluate the impact of three ice shapes and three flow directions on cooling performance. The results indicate that neither crushed ice nor ice block alone provide the optimal thermal performance for Hyperloop cooling requirements in terms of both effective capacity and dynamic response. Crushed ice offers fives times faster thermal response but has a 42% less packing density, while ice block provides greater thermal mass but responds more slowly to dynamic cooling demands. Therefore, a balance between the two configurations must be identified to combine adequate heat transfer performance with sufficient cooling capacity.

1. Introduction

Hyperloop transport operates in a vacuum-like tube environment, which makes a thermal management system (TMS) indispensable for handling waste heat from components. Passive cooling (e.g., by ambient air) is only possible to a very limited extent in such environments. Hyperloop envisions high-speed transportation within low-pressure tubes that drastically reduce aerodynamic drag. Within these tubes, transport pods are propelled by a linear electric motor and levitated using electromagnetic suspension. The foundational principles of Hyperloop—such as maglev propulsion and operation in low-pressure tunnels—have antecedents in earlier concepts like Swissmetro, a 1970s proposal for high-speed magnetic-levitation transport in underground partial-vacuum tubes developed by EPFL and ETH Zurich [1]. In the Hyperloop pod, components like the linear motor, power electronics, and batteries will generate substantial heat that must be managed in a thin-air environment. According to the Delft Hyperloop study [2], there are three principal approaches for active cooling in a Hyperloop pod: radiation, sublimation of ice, and latent heat storage (LHS). Delft Hyperloop shows that the use of LHS is the most promising solution for a TMS in a Hyperloop. Radiation is only marginally possible because of the small temperature gradient. Sublimation, on the other hand, increases the necessary infrastructure, as it requires condensers upstream of the vacuum pumps. LHS systems utilize the ability of phase change materials (PCMs) to store/release large amounts of thermal energy during phase change (melting/crystallization) while maintaining a constant temperature. LHS systems achieve a high energy density (150–400 J/cm³ and 150–350 J/g) [3].

For an LHS system in a Hyperloop, the choice of PCM is critical. The PCM’s phase change temperature must be in a range that can absorb heat from the pod’s systems. The Hyperloop TMS must regulate the temperature of various components (e.g., motors, electronics) and also provide climate control for passengers. A comfortable cabin temperature around 20–23 °C is desired, so the PCM’s melting temperature should be below 20 °C to effectively absorb heat and keep the cabin and equipment cool. In practice, a PCM melting temperature in the range of about −5 °C to 20 °C is targeted to meet the cooling requirements of the Hyperloop pod’s systems [2]. While the PCM sets the lower thermal boundary of the storage system, heat transfer to the components is managed by an HTF that remains warmer than the PCM’s melting point, ensuring that sub-zero cooling of sensitive components is avoided. Choosing a PCM with a lower melting temperature, such as −5 °C, supports a wider operating range and increases the usable sensible heat in the liquid state, as the PCM can be heated further before reaching its upper operating limit. This effectively improves the overall energy density of the system by maximizing the temperature span over which sensible heat can be stored and extracted. In addition, mass is also a critical constraint in Hyperloop design due to its influence on energy consumption and dynamic performance. In addition, mass is also a critical constraint in Hyperloop design due to its influence on energy consumption and dynamic performance. Therefore, the PCM must possess a high volumetric and gravimetric energy density to maximize thermal storage capacity while minimizing system weight.

Table 1. PCMs with a phase change temperature in the range of −5 °C to 20 °C [3,4].

Substance Class	Material	${\mathit{T}}_{\mathit{PC}}$ (°C)	$\mathbf{\Delta }{\mathit{h}}_{\mathit{PC}}$ (J/g)	$\mathit{\lambda }$ (W/(m K))	${\mathit{c}}_{\mathit{p},\mathit{l}}$ (J/(g K))	${\mathit{c}}_{\mathit{p},\mathit{s}}$ (J/(g K))
Paraffins	Tetradecane	6.0	227.2	0.14 (25 °C)	2.21 (80 °C)	1.68 (0 °C)
	Hexadecane	18.3	235.6	0.34 (0 °C)	2.22 (80 °C)	1.75 (0 °C)
Fatty acids	Hexanoic acid	−3.0	130.8	0.14 (0 °C)	2.25 (80 °C)
	Methanoic acid	8.3	275.5	0.23 (0 °C)	2.20 (20 °C)	1.62 (−22 °C)
	Ethanoic acid	16.6	195.2	0.17 (0 °C)	2.03 (20 °C)	1.38 (0 °C)
Esters	Ethyl tetradecanoate	14.1	173.4	0.16 (26 °C)
Alcohols	1-Decanol	7.0	212.7	0.16 (25 °C)	2.31 (16.75 °C)	2.03 (0.5 °C)
Salt hydrates	Potassium fluoride tetrahydrate	18.5	231.0
	Water	0	333.7	2.26 (<0 °C)	4.23 (≥0 °C)	2.07 (<0 °C)

lists several PCM candidates within the desired temperature range, highlighting the advantage of water in terms of its notably high melting enthalpy.

Water has a melting point of 0 °C, which is within the desired range and a high melting enthalpy (333.7 J/g). In addition, liquid water has a high specific heat capacity (4.23 J/(g K)), significantly higher than the other PCMs listed (Table 1). These properties mean that water can provide the highest energy density in the relevant temperature range. The high energy density allows for more thermal energy to be stored per unit mass, which directly contributes to reducing the overall weight of the TMS.

In addition to high energy density, the TMS must also ensure a flexible and responsive thermal power output to match varying heat loads from the Hyperloop pod’s systems. According to the feasibility study of Delft Hyperloop, these heat quantities range from 6 kW to 30 MW for a full-sized system [2]. However, a common limitation of LHS is the low heat transfer during the solid state of the PCM due to the change between the dominated heat transfer by convection in the liquid state and by conduction in the solid state. The conventional approach to address this limitation involves the use of heat exchanger (HEX) structure to increase the heat transfer surface [5]. While effective, these solutions introduce additional mass—counterproductive in a system where weight minimization is critical. To overcome this, two alternative strategies can be considered: the development of advanced PCMs with intrinsically higher thermal conductivity or the use of the storage in direct contact, which increases the heat transfer surface without adding additional mass [6]. The latter approach of a direct contact latent heat storage (DCLHS) promises a pathway for enhancing PCM integration in compact, high-performance thermal management systems. In a DCLHS, the heat transfer fluid (HTF) is injected directly with a nozzle into a tank filled with PCM. This direct injection generates numerous small droplets and this results in a large contact surface between the two immiscible media. The large contact surface creates ideal conditions for efficient heat transfer [5]. Owing to the difference in density between the PCM and the HTF, the HTF flows separately through the tank. This creates a complex channel network that acts as a direct contact HEX. As the PCM takes up to 90% of the storage volume, which leads to high energy densities while still allowing the system to be built compactly and simply. Despite the advantages a DCLHS offers, it has not yet been sufficiently researched compared to conventional LHS systems [5,7]. Thus far, only small DCLHS systems have been built for certain PCM/HTF combinations, namely salt hydrates/oil [8], paraffin/water solutions [9], ester/water [6], erythritol/oil [10] and ice/thermal oil [4]. These studies show a clear potential of the system to achieve a high heat transfer rate and high energy density at a system level. Nevertheless, several challenges were identified that do not allow a robust operation of a DCLHS without further improvement. Two major challenge are the emulsification of HTF and PCM and the carry-over of PCM from the storage tank into the HTF cycle, which can cause blockages in the piping system or can damage equipment when the PCM solidifies [6,11]. This problem is exacerbated by higher HTF mass flow rates, which are required to achieve higher power outputs.

In this study, water is employed as both the PCM and the HTF—with liquid water acting as the HTF and solid ice as the PCM. This unique choice eliminates the risk of emulsification and phase separation, since both phases are the same substance. However, the solidification of the storage material has to occur in a separate process as the HTF would also freeze during this operation. In the context of Hyperloop operation, this constraint can be addressed by implementing a system architecture in which the thermal energy storage is fully discharged during transit, followed by either replacement of the storage module or in-situ refilling at terminal stations. This approach allows the recharging (solidification) process to occur independently under controlled conditions at the station, ensuring that the PCM is fully regenerated prior to the next operational cycle. In conventional outdoor transport systems operating in cold climates, freezing of the HTF would typically be a concern. However, this issue is mitigated in Hyperloop applications, as the thermal management system is installed inside the transport pod, which is thermally insulated from the external environment by the surrounding vacuum tube. As a result, the interior remains at a controlled temperature above freezing, making water a practical and justified choice for use in the system. This work focuses on the experimental DCLHS system using ice and water, targeting the specific thermal management demands of a Hyperloop transport system. The goal is to achieve high energy density and rapid thermal response to support peak cooling loads during operation. A critical factor influencing these outcomes is the shape of the ice used in the storage unit. Different ice shapes, such as crushed ice, ice cubes, or a large ice block, affect both the heat transfer rate and the effective utilization of the PCM. Smaller ice particles provide higher surface area for heat exchange, leading to faster thermal response, while larger blocks offer greater volumetric storage efficiency. We hypothesize that ice shape significantly impacts the outlet fluid temperature profile and the total amount of energy extracted. Understanding the relationship between ice shape and system performance offers valuable insights into the internal flow dynamics and thermal interactions within the DCLHS. While the proposed cooling system can certainly be adapted for other applications, its combination of low weight, compact design, and high energy density makes it particularly well-suited for the stringent requirements of Hyperloop systems.

2. Materials and Methods

2.1. Experimental Setup and Materials

This subsection first outlines the construction of the experimental setup, followed by a detailed examination of the various ice shapes used as storage materials.

2.1.1. Experimental Setup

A schematic view of the setup can be seen in Figure 1. In

Table 2. List of components in the setup.

Position	Component	Brand
1	Thermostat	HUBER Variostat
2	Heat exchanger	Unbranded Plate heat exchanger
3	Pump	Unbranded MP-15 RM
4	Bypass
5	Expansion vessel	Unbranded 24l expansion vessel
6	Compact cold storage
7	Water connection

, the most important components are listed, and in

Table 3. Measurement points in the setup.

Label	Measured Quantity	Sensor Type	Measurement Uncertainty
${\dot{m}}_{CL}$	Massflow cooling loop	AF-E 400 (Krohne, Basel, Switzerland)	±0.8% of measured value + 0.2% of full scale (35 L/min full scale)
$p_{CL}$	Pressure cooling loop	Manometer Typ 707 (Haenni, Frauenfeld, Switzerland)
$\mathsf{\Delta }{p}_{CCS}$	Pressure difference CCS	Deltabar PMD75 (Endress+Hauser, Reinach, Switzerland)	±0.05%
$T_{out}$	Outlet temperature CCS	PT100 (Albert Balzer AG, Dornach, Switzerland)	±0.15 °C at 0 °C
$T_{in}$	Inlet temperature CCS	PT100 (Albert Balzer AG, Dornach, Switzerland)	±0.15 °C at 0 °C
$T_T$	Temperature of the thermostat

, the measurement points are included. Via the thermostat (Offenburg, Germany), the temperature of $T_{in}$ in the cooling loop (CL) is set. The thermostat heating loop (HL) is connected by a heat exchanger to the CL. A LabView (2021 SP1) program controls the cooling power of the storage by adjusting the flow rate through the pump. The bypass was manually operated toward the end of the experiments to help the pump and system maintain its cooling power. In the CL, only water flows through the pipes, where the HL consists of a water–glycol mixture. By flowing through the DCLHS, referred to in this study as the compact cold storage (CCS), the water cools down and is heated up again in the HEX. An expansion vessel is installed to compensate for the melting ice and the increasing water temperatures.

The CCS is rotatable and the pipes of the inflow and outflow of the storage can easily be disassembled. This makes it simple to change to different flow directions in the CCS. The CCS consists of a double-walled transparent pipe with an inner diameter of 44 mm, an outer diameter of 75 mm and a length of 850 mm (Figure 2). The schematic view in Figure 3 shows the dimensions of this tank and the later explained freezing process of the ice block.

The experimental setup includes several measurement points. The mass flow rate and pressure are measured before the water enters the CCS. To detect storage clogging, the pressure difference between the inlet and outlet is measured. The inlet and outlet temperatures of the CCS were also measured. The thermostat temperature was only used as a set point. The temperature was set, and the internal controlling of the thermostat was used to keep the temperature $T_T$ at a constant level.

The accuracy of the measurements depends on the sensor specifications summarized in Table 3. While temperature measurements are precise, the massflow sensor introduces a larger relative uncertainty at low flow rates due to its fixed offset. Nonetheless, these uncertainties are systematic and consistent, allowing for reliable comparison between different test cases.

2.1.2. Shape of Storage Material

Three different ice shapes were generated and used as storage material in the CCS system. These are presented in Figure 3. The ice cubes (Figure 3a) were produced using the ice-cube maker PC-EWB1187 (ProfiCook, Kempen, Germany) ice machine, set to its smallest size setting, yielding cylindrical ice pieces with a diameter of 15 mm and a height of 30 mm. The crushed ice (Figure 3b) was obtained by crushing the ice cubes, resulting in randomly sized fragments ranging from approximately 5 mm to 30 mm in length. The ice block (Figure 3c) was produced directly inside the CCS setup. As shown schematically in Figure 3, this was achieved by inserting a stainless steel pipe (6 mm diameter) into the CCS, which was connected to the thermostat. By setting the thermostat temperature $T_T$ to −20 °C, the water surrounding the pipe gradually solidified into an ice block. After producing the ice block, the stainless steel pipe was removed to carry out the experiments.

The different ice shapes were produced in a separate process due to the specific system configuration, in which water serves as both the PCM and the HTF. In this context, the packing density—defined as the ratio of the ice volume to the total available volume (Formula (4))—was determined for each ice shape. For the ice block, a high packing density of approximately 98% was achieved, calculated based on the known geometry of the CCS and the volume displaced by the freezing pipe. For the ice cubes and crushed ice, packing densities of 47 ± 2% and 56 ± 2%, respectively, were measured by filling a predefined volume—with a diameter similar to the actual CCS—with each ice type, recording the corresponding mass and calculating the volume using the density of ice (916.8 kg/m³ [4]). The reported packing densities represent average values based on three repeated measurements for each ice shape. These packing densities significantly influence the volumetric energy storage capacity and the thermal exchange characteristics of the system.

2.2. Methods

Each measurement with ice cubes and the crushed ice was prepared in the same four-step process.

1.

Precooling: Before each measurement, the CCS was filled with water, and then the setup was cooled to 5 °C. For this, the pump was running at a constant speed of 60% of its maximum flow rate and the thermostat delivered the cooling power. The reason for this step is to prevent the ice from melting too fast in the preparation phase of the measurements.

2.

Heating of the thermostat: After the precooling, the pump was turned off and the thermostat was set to the temperature of the experiment.

3.

Filling the CCS: The CCS was emptied and filled with ice cubes or crushed ice. Then, 5 °C water was added to fill the remaining volume. It was assumed that all voids were completely filled with water.

4.

Pressurize the setup: With the water connection, the laboratory setup was filled again and pressurized to an overpressure of 0.7 bar. The residual air in the CCS was vented in this same step.

After these steps, the measurements were started. With the ice block the process was slightly different. The ice block was already produced in the CCS. The thermostat was set to 5 °C for 3 min to start the ice melting around the metal pipe. This made sure that the pipe could be removed. After the removal of the pipe, the thermostat was set to the experiment temperature. The system was pressurized, and the residual air was vented. Each experiment was concluded once the temperature difference across the system decreased to the point where the nominal cooling power could no longer be maintained, despite further increases in mass flow. This condition marked the limit of effective capacity under the given configuration.

2.2.1. Flow Directions

With the rotatable CCS, the influence of the flow direction in the CCS was also tested. In Figure 4, these three directions are illustrated with the directions of the arrows in and out of the storage. In the results, the flow direction where the water enters from the top is called downwards. The opposite, where the entry is at the bottom, is called upward. The last option is called horizontal.

2.2.2. Experimental Plan

Table 4. Experiments conducted on the experimental setup; secondary tests are marked in gray.

Type of Ice	Power (W)	Thermostat Temperature (°C)	Flow Direction
Ice cubes	100	15	Upwards
	100	15	Horizontal
	120	15	Downwards
Crushed ice	100	15	Upwards
	100	15	Horizontal
	120	15	Downwards
	200	25	Downwards
	300	35	Downwards
Ice block	100	15	Upwards
	100	15	Horizontal
	100	15	Downwards
	200	25	Downwards
	300	35	Downwards

provides an overview of the experimental matrix, detailing the combinations of ice shapes, thermal power, thermostat setpoint temperatures, and flow directions tested. In the primary experiments, tests were conducted at 100 W (or 120 W, depending on flow configuration) for all three ice shapes. For the ice cubes and crushed ice, it was observed that the downward flow configuration resulted in an insufficient flow rate for stable pump operation at 100 W, necessitating an increase to 120 W. Two test runs were performed for each configuration.

Based on the results of the primary experiments, it was determined that the thermal behaviour of the ice cubes was similar to that of the crushed ice but with a lower effective capacity. This difference is primarily attributed to the lower packing density of the ice cubes, which reduces the volumetric energy storage potential. As a result, ice cubes were excluded from the secondary experiments at higher power levels (200 W and 300 W), and the focus was placed on the more effective ice shapes.

Furthermore, analysis of the primary results showed that the downward flow direction yielded superior thermal performance compared to the other to flow directions. Therefore, all secondary experiments were conducted exclusively with the downward flow configuration to ensure consistent and optimal system behaviour.

2.2.3. Calculations

To evaluate the performance of each of the flow directions and the ice shapes, several parameters and indicators are calculated. The data to calculate the parameters come from measurements from the experimental setup in Figure 1 or were obtained from the literature and listed in Table 1.

Cooling Power

The cooling power $\dot{Q}$ (W) is determined from the measurement points in the CL of the experimental setup [12].

$$\dot{Q}={\dot{m}}_{CL}·c_{p,l,water}·(T_{in}-T_{out})$$

(1)

In this equation, ${\dot{m}}_{CL}$ (kg/s) denotes the massflow rate in the CL, and the value of the specific heat capacity $c_{p,l,water}$ (J/(kg K)) was obtained in the literature and can be seen in Table 1. The temperature difference was taken from the inlet to the outlet with $T_{in}$ and $T_{out}$ (°C).

Effective Capacity of the CCS

By summing up all cooling power and multiplying them by the measurement time interval, where $t_0$ (s) is the start time and $t_{total}$ (s) is when the experiment was stopped, the effective capacity Q (J) is calculated [12].

$$Q={\int }_{t_0}^{t_{total}}\dot{Q}dt$$

(2)

The measurement interval $dt$ was fixed by the experimental setup to 2 s.

Power-to-Capacity Ratio

The power-to-capacity ratio $PtC$ (1/h) is used to compare the average cooling power $P_{avg}$ (W) extracted from the storage with the theoretical storage capacity $Q_{100}$ (Wh) [13]. This ratio also allows conclusions to be drawn about the discharge time of the system:

$$PtC={\displaystyle \frac{P_{avg}}{Q_{100}}}$$

(3)

The theoretical capacity $Q_{100}$ is calculated separately for each ice shape, as it depends on the packing density $\varphi$ (−), which determines the effective volume fraction of ice $V_{ice}$ (m³) within the volume of the storage $V_{storage}$ (m³):

$$\varphi ={\displaystyle \frac{V_{ice}}{V_{storage}}}$$

(4)

$$Q_{100}={\rho }_{ice}·V_{storage}·\varphi ·(c_{p,s}·\mathsf{\Delta }{T}_s+\mathsf{\Delta }{h}_{PC}+c_{p,l}·\mathsf{\Delta }{T}_l)+{\rho }_{water}·V_{storage}·(1-\varphi )·(c_{p,l}·\mathsf{\Delta }{T}_l)$$

(5)

Here, $\mathsf{\Delta }{h}_{PC}$ (J/kg) represents the phase change enthalpy. The sensible heat in the solid and liquid phases is calculated using the specific heat capacities $c_p$ (J/(kg·K)) and the corresponding temperature ranges $\mathsf{\Delta }{T}_s$ and $\mathsf{\Delta }{T}_l$, which cover the relevant intervals from −10 °C to 20 °C. The ice block was considered over the full range from −10 °C to 20 °C, as it had been pre-cooled to −10 °C during preparation. In contrast, ice cubes and crushed ice were considered from −1 °C to 20 °C, since they were taken directly from the collection bin of the ice machine, which did not provide subzero storage and thus limited the minimum temperature to approximately −1 °C. The −1 °C starting temperature was confirmed through spot measurements of the ice immediately after loading. While small temperature variations may exist due to partial warming at the surface, the average temperature across the ice volume was consistently close to −1 °C, justifying its use in the theoretical capacity calculations. The symbols ${\rho }_{ice}$ and ${\rho }_{water}$ represent the densities (kg/m³) of ice and water, respectively, which are used to determine the mass content based on the volume and packing density. The size of the CCS is shown in Figure 3, and the material properties are listed in Table 1.

Table 5. Theoretical capacity $Q_{100}$ of each ice shape.

Type of Ice	Pack Density (%)	Mass Ice (kg)	Mass Water (kg)	${\mathit{Q}}_{100}$ (Wh)
Ice cubes	47	0.56	0.68	81.1
Crushed ice	56	0.66	0.57	90.8
Ice block	98	1.16	0.03	142.2

shows the theoretical storage capacity of each ice shape with the corresponding mass of ice and water.

Utilization Factor

A measure for evaluating how large the effective capacity Q is compared to the theoretical calculated capacity $Q_{100}$ is the utilization factor $Uf$ (−). It can be determined by directly comparing the capacities:

$$Uf={\displaystyle \frac{Q}{Q_{100}}}$$

(6)

This approach considers the ratio of the actual and theoretical capacities.

Dimensionless Time

To enable a consistent comparison between experiments of varying durations, the dimensionless time $\tau \ast$ (–) was introduced. It is defined as the ratio of the real time $t_{\left(t\right)}$ (s) to the total duration of the experiment $t_{total}$ (s):

$$\tau \ast ={\displaystyle \frac{t_{\left(t\right)}}{t_{total}}}$$

(7)

This normalization allows for a direct comparison of time-dependent performance indicators across different test conditions by mapping all experiments onto a common temporal scale ranging from 0 to 1.

2.2.4. Heat Losses

To calculate the heat losses of the CCS, it was filled with 73 °C water. The ambient temperature was stable at 22 °C. With a thermocouple in the water, the temperature $T\left(t\right)$ (°C) was recorded over time t (s). This curve is, in theory, described by

$$T\left(t\right)=T_{amb}+\left(T_0-T_{amb}\right)·exp\left({\displaystyle \frac{k·A_{CCS}}{m_{water}·c_{p,l,water}}}·t\right)$$

(8)

which is Newton’s law of cooling. Here $T_0$ (°C) is the starting temperature, $T_{amb}$ (°C) the ambient temperature and k (W/(m² K)) is the overall heat transfer coefficient [14].

$$A_{CCS}=0.85·0.044·\pi +{\left({\displaystyle \frac{0.044}{2}}\right)}^2·\pi ·2=0.12 {\mathrm{m}}^2$$

is the surface where the CCS is losing heat, and the mass of water

$$m_{water}={\rho }_{water}·V_{water}=998·{\left({\displaystyle \frac{0.044}{2}}\right)}^2·\pi ·0.85 = 1.29 \mathrm{kg}$$

is calculated by the volume inside the CCS $V_{water}$ (m³) and the density of water ${\rho }_{water}$ (kg/m³). k is calculated through Equation (8) by a curve fit. Its value is 4.5 W/(m² K). In Figure 5, the raw data and the fitted curve (based on Equation (8)) are presented.

By using the law of heat transfer

$${\dot{Q}}_{loss}=k·A_{CCS}·\left(T_{amb}-T_{ice}\right)=4.5·0.12·\left(22-0\right) = 12.0 \mathrm{W}$$

(9)

and by assuming the ice temperature $T_{ice}$ to 0 °C and using $T_{amb}$ of 22 °C, the heat losses ${\dot{Q}}_{loss}$ (W) are calculated. These heat loss calculations are intended to provide a rough estimate of the thermal losses from the CCS, primarily to assess their order of magnitude. While the thermal properties of the insulation material—such as thermal conductivity—can vary with temperature and influence the overall heat transfer coefficient k, these effects were considered negligible for this estimation. Since the purpose is to capture general system behavior rather than precise values, such simplifications are acceptable. Moreover, the characteristic behavior described by Newton’s law of cooling remains valid, and the same exponential trend would apply in the case of heat gain from the environment during an actual experimental run.

2.2.5. Data Processing

To enhance the clarity and comparability of the experimental data, all recorded time-series variables were subjected to a preprocessing step prior to analysis. A rolling average filter with a window size of 5 samples was applied to each variable to suppress high-frequency noise and reduce short-term fluctuations. This smoothing step ensures that the primary trends and characteristic behaviours of the system are more clearly visible in the resulting plots.

3. Results

In this section, the measurements and thus the results and comparisons are shown. The corresponding findings and conclusions are then followed in Section 4 and Section 5.

3.1. Primarily Experiments

To gain an initial overview of the system performance, a series of experiments were conducted using all ice shapes and flow directions at nominal cooling powers of 100 W and 120 W. Figure 6 presents a representative example of these tests, illustrating the time-dependent behaviour of key operational parameters: cooling power, inlet and outlet temperatures, and mass flow rate in the CL. In this experiment, the cooling power (blue line) was controlled by adjusting the mass flow rate (red line). The inlet temperature, denoted as $T_{in}$ (green line), gradually increased to the setpoint defined by the external thermostat. The outlet temperature $T_{out}$ (purple line) initially decreased as the system began discharging and thermal energy was effectively absorbed from the ice. As the ice approached complete melting, $T_{out}$ started to rise, indicating the depletion of latent heat storage and a transition to sensible heat storage in the remaining liquid phase. This is shown, for example, in Figure 6 at around 25 min.

Figure 7 gives an overview of the mean outlet temperatures of the CCS compared with the effective capacity for all primary experiments performed. The mean outlet temperature for each experiment was calculated based on the time interval during which the nominal cooling power was consistently maintained. Across all tests, outlet temperatures ranged from 2.6 °C to 10.7 °C, while effective storage capacities varied between 36 Wh and 123 Wh. The maximum theoretical capacity of the CCS system, which could be reached with the ice block, is 142 Wh, as presented in Table 5. The results clearly indicate that the highest effective capacities were achieved with the ice block configuration, while the ice cubes consistently showed the lowest capacities. This trend corresponds well with the packing density of the different ice shapes: larger and more uniform geometries like ice blocks allow for higher packing densities within the storage volume. Conversely, loosely packed geometries such as ice cubes or crushed ice result in lower packing densities and thus reduced capacity.

In addition to ice shapes, the flow direction was found to significantly influence performance. Among all tested configurations, the downward flow direction resulted in the highest effective capacities across all ice shapes. The greatest variation in capacity was observed in the horizontal flow orientation for all three ice shapes. Moreover, the mean outlet temperature was strongly affected by ice shape. Configurations with smaller particles, such as crushed ice and ice cubes, consistently achieved lower outlet temperatures due to the larger surface area available for heat exchange. This effect was most pronounced in the downward flow direction, where both ice cubes and crushed ice exhibited outlet temperatures approximately 3 K lower than in the horizontal and upward flow cases. The ice block also showed this trend, though the differences between flow directions were less pronounced. Given that low outlet temperature and high effective capacity are critical performance parameters for meeting the cooling requirements of the Hyperloop application, the downward flow direction, along with the crushed ice and ice block configurations, were selected for further investigation in the secondary experiments. The ice cubes were excluded from further testing due to their lower effective capacity and thermal behavior closely resembling that of the crushed ice but with no significant advantages in performance.

3.2. Secondary Experiments

In the secondary experiments, the focus was placed on the crushed ice and ice block configurations, both operated under the downward flow direction, which had demonstrated the most favorable thermal behavior in the primary experiments. These tests were conducted at increased cooling power levels of 200 W and 300 W, building upon the results of the primary experiments performed at 100 W and 120 W. Figure 8 provides a comprehensive overview of the mean outlet temperatures and effective storage capacities for both the primary and secondary test series. The results from the secondary experiments clearly show that increasing the set cooling power leads to higher effective capacities. In addition to capacity, the mean outlet temperature is also affected by the increased thermal load, though the extent of this influence varies with the ice shape. For the crushed ice, the outlet temperature remains relatively stable, even at higher cooling powers, with values consistently around 3 °C. In contrast, the ice block shows a more pronounced sensitivity to cooling power: as the set power increases from 100 W to 200 W and 300 W, the mean outlet temperature rises by approximately 3 K.

3.3. Effective Capacity

With the experimental setup, only constant power profiles from 100 W to 300 W were tested, as listed in Table 4. The different nominal powers had an influence on the effective capacity, as seen in Figure 8, and with a higher nominal power, a higher effective capacity was reached. The effective capacity is also influenced by the direction of the flow. With horizontal, the results showed significant variations for all types of ice. This dispersion was also observed during testing, as the melting behavior varied considerably across different tests. The flow direction upwards had a negative impact on the effective capacity. For example, with the ice block and upwards, the effective capacity was 83 Wh, which is 58% of the theoretical capacity. On the other hand, with ice block and downwards, the lowest effective capacity from the experiments was 116 Wh, which is 82% of the theoretical capacity. The positive impact with the flow direction downwards is also present for crushed ice.

Outlet Temperature of CCS

The outlet temperature behavior varied significantly depending on the ice shape and flow direction. For the crushed ice, the outlet temperature exhibited a consistent and characteristic pattern across different experimental conditions. As shown in Figure 8, particularly for the downward flow direction, the mean outlet temperature stabilized around 3 °C. In this configuration, two distinct outlet temperature plateaus were observed (see Figure 9): an initial level at approximately 1.5 °C, which occurred following a rapid temperature drop, followed by a transition to a second, higher level at around 3.5 °C. This stepped behavior was also observable, though less distinctly, in the horizontal and upward flow directions, where outlet temperatures stabilized at around 3.5 °C and 6 °C, respectively.

In contrast, the ice block configuration exhibited a distinctly different thermal response. The outlet temperature was more sensitive to both flow direction and nominal cooling power. At 100 W, the mean outlet temperature was around 7 °C, while at 200 W and 300 W, it increased to approximately 10 °C, as shown in Figure 8.

In the downward flow configuration with 300 W cooling power (see Figure 9), the ice block setup exhibited a two-stage outlet temperature profile. Initially, the outlet temperature remained elevated at around 18 °C, followed by a rapid decline to a quasi-stationary value near 10 °C. This transition occurred approximately five minutes into the experiment—about 20% of the total duration—and coincided with a sharp drop in both mass flow rate and outlet temperature. This behavior was linked to the detachment of the ice block from the tank walls and its subsequent upward flotation, which was visually confirmed via the transparent design of the CCS. Figure 10 captures this moment, showing a visible cold water layer (colored blue) beneath the floating ice block.

Following this event, the outlet temperature stabilized around 8–10 °C, and a cold water layer formed at the bottom of the tank, similar to the behavior observed in the crushed ice configuration under downward flow. However, unlike the crushed ice, no distinct temperature plateaus were observed, and the ice block configuration showed a more continuous rise in outlet temperature toward the end of the experiment.

3.4. Comparison $Uf$ and PtC

In Figure 11, the comparison of the utilization factors for the ice block and crushed ice configurations demonstrates that both operate within a similar performance range. As ice cubes were only tested during the primary experiments at a single cooling power level, they were excluded from this comparison. Across all experiments, the utilisation factor varies between 0.80 and 0.98 for both ice shapes. This indicates that a significant portion of the available thermal energy was effectively extracted during each discharge cycle, regardless of the ice shape, as the utilization factor quantifies the proportion of the effective capacity that is actually utilized relative to the maximum theoretical capacity of a given ice shape.

However, the power-to-capacity ratio is consistently higher in the experiments involving crushed ice. This outcome is expected, as both configurations were operated at identical nominal cooling powers, while the crushed ice exhibits a lower theoretical capacity due to its reduced packing density. By definition, the power-to-capacity ratio increases when a smaller theoretical energy capacity is discharged with the same fixed power level (see Equation (3)).

Additionally, a clear correlation is observed between the power-to-capacity ratio and the utilization factor. As the power-to-capacity ratio increases, indicating a more intensive use of the storage system relative to its theoretical capacity, the utilization factor also increases in both cases.

4. Discussion

Based on the previously presented results, this discussion section examines the power and capacity of the CCS, as well as the thermal characteristics of its outlet temperature.

4.1. Power and Capacity

The results from the previous experiments clearly demonstrate that the ice cubes configuration yielded the lowest effective capacity, which aligns with expectations due to its lower packing density.

When isolating the influence of flow direction, independent of ice shapes, it was found to have a dominant impact on capacity. The horizontal flow configuration exhibited the largest variability in capacity within a single flow direction. This inconsistency is primarily attributed to buoyancy-driven effects: during the initial stages of operation, ice tends to accumulate at the top of the storage unit while water flows beneath it, reducing effective thermal contact and stratification. Additionally, uncertainties in experimental preparation—such as the variation in freezing time and handling of the ice block contributed to capacity fluctuations in all three flow directions. In contrast, the preparation and loading of ice cubes and crushed ice were more reproducible, leading to improved consistency across experiments.

Among all tested flow directions, the downward configuration consistently resulted in the highest effective capacities across all ice shapes. This enhancement is attributed to favorable thermal stratification, enabled by the anomalous density behavior of water, which reaches a maximum at 4 °C [15]. Cooler water sinks to the bottom of the CCS, promoting efficient layering and heat transfer. Such stratification cannot form in horizontal or upward configurations, negatively impacting performance in those cases.

From a thermal regulation perspective, crushed ice provided more stable outlet temperatures. The formation of random flow channels facilitated uniform heat exchange and allowed for more controlled cooling power delivery. In contrast, the ice block offered a lower surface area-to-volume ratio, limiting heat transfer. Although the contact area between water and ice gradually increased during melting, it remained substantially lower than that of crushed ice. Consequently, the ice block exhibited higher outlet temperatures and required increased mass flow rates to maintain the same cooling power. In downward flow conditions, the accumulation of a liquid water layer beneath the ice block acted as a thermal buffer, stabilizing outlet temperatures and contributing to smoother power regulation.

A positive correlation between the power-to-capacity ratio and the utilization factor was observed. This trend is partially explained by thermal losses from the CCS, which were not included in the calculation of effective capacities. The estimated heat loss of the CCS is 12.0 W (see Equation (9)). Since experiments with higher nominal cooling power and therefore higher $PtC$ typically had shorter durations, they incurred lower total heat losses, resulting in relatively higher $Uf$ values.

It should be noted that this heat loss estimate accounts only for surface losses of the storage tank. Additional losses in the circulation system—such as from piping, valves, and the pump—were not included and would further reduce the actual delivered cooling energy. As a result, if thermal losses were not accounted for, the reported effective capacities would slightly overstate the actual energy delivered by the CCS. However, since such losses would also occur in a real application, their presence does not compromise the validity of the comparison. Moreover, all configurations were tested under comparable ambient conditions and durations, ensuring that the relative differences between them remain meaningful and consistent.

Both crushed ice and ice block configurations achieved similar utilisation factors, indicating that a comparable proportion of their respective theoretical capacities was extracted. However, it is important to note that this comparison is based on the individual theoretical capacities of each shape. If the comparison were instead made using a common maximum theoretical capacity (e.g., that of the ice block), the crushed ice would yield a lower $Uf$ due to its inherently lower packing density.

4.2. Outlet Temperature

Across all experiments, the downward flow direction consistently produced the most stable outlet temperatures and exhibited the least variability between repeated runs. This effect can be attributed to the anomalous density behavior of water, which facilitates the formation of a thermally stratified layer. Due to the density differences between ice and liquid water, the ice naturally accumulates at the top of the CCS, while water near 4 °C—the temperature at which water reaches its maximum density—settles at the bottom of the tank [15]. This stratification effect contributes to the stabilization of outlet temperatures, particularly evident in the crushed ice configuration, where a plateau around 4 °C is observed in the second half of the discharge cycle. The crushed ice configuration inherently provides a larger surface area for heat exchange compared to the ice block, which enhances thermal responsiveness and enables more effective regulation of the outlet temperature under varying cooling power conditions. In the downward flow configuration, the outlet temperature of the CCS using crushed ice remained largely unaffected by changes in set cooling power, further demonstrating its thermal robustness. The CCS with crushed ice reached a steady-state outlet temperature in less than 5% of the total measurement time, while the system with an ice block required more than 20% of the duration to stabilize. This indicates that ice blocks respond more slowly to changes in thermal load—a potentially critical limitation in applications with rapidly fluctuating cooling demands, such as in a Hyperloop environment. For actual TMS in a Hyperloop, implementing a mass flow regulation strategy would be a suitable approach to actively control outlet temperatures and adapt to varying cooling demands in real time.

5. Conclusions

In this study, a laboratory-scale CCS was experimentally investigated and characterized with respect to flow direction and ice shape to gain an initial understanding of which configurations may be most suitable for implementation in a TMS for a potential Hyperloop application. Among the tested parameters, flow direction was identified as a dominant factor influencing system performance. The downward flow direction consistently yielded the lowest outlet temperatures and the highest effective capacities across all ice shapes. These findings suggest that the downward configuration promotes favourable thermal stratification within the CCS, enhancing overall heat transfer efficiency.

In particular, the combination of downward flow with crushed ice proved highly effective. This configuration maintained a low and stable mean outlet temperature of approximately 3 °C, even under varying nominal cooling powers. Such thermal stability under different load conditions is advantageous for TMS applications where dynamic cooling demands are expected.

While the ice block configuration offered a higher effective capacity due to its high packing density of 98%, it was less responsive to changing thermal loads. The outlet temperature in ice block experiments exhibited a stronger dependence on the set cooling power and showed delayed stabilization. These characteristics suggest a slower dynamic response, which could be a limitation in systems requiring rapid thermal adjustment. As a result, crushed ice in the downward flow direction is identified as the most favourable configuration for applications requiring both high thermal responsiveness and consistent outlet temperature, reaching steady-state conditions approximately five times faster than the ice block configuration. However, due to its lower packing density of 56%, crushed ice offers less overall cooling capacity compared to the ice block, which provides greater thermal storage potential but responds more slowly to temperature changes. Therefore, an optimal balance between the fast thermal response of crushed ice and the higher cooling capacity of the ice block must be found to meet both dynamic and sustained cooling demands effectively.

To build upon these findings, future research should aim to optimize the properties of ice-based cooling media to close the performance gap between crushed ice and ice block. This includes identifying an intermediate particle size and volume distribution that improves packing density and overall energy storage capacity, while still maintaining the fast thermal response observed with crushed ice. Additionally, improved flow distribution strategies should be explored to enhance stratification and promote uniform heat transfer within the CCS.

Further investigations into long-term cycling stability, melting dynamics under realistic TMS load profiles and the scaling of system volume are recommended. Expanding the storage volume of the experimental setup may help mitigate boundary effects, which could have influenced the thermal performance measurements and system behavior in this study. Additionally, the random size distribution of the crushed ice may have affected the results; therefore, future tests should also consider using ice samples with more uniform and defined shapes to better isolate and understand shape-related performance differences.

When scaling up the CCS for implementation, boundary conditions, surface-to-volume ratios, and other real-world effects—such as mechanical vibrations—will differ from the controlled laboratory setup. These factors can influence melting dynamics, thermal stratification, and overall system behavior. While the modular design of the CCS supports scalable adaptation, further investigations are needed to understand how these geometric and physical influences affect performance on larger scales.