Research on Multi-Stage Battery Detachment Multirotor UAV to Improve Endurance

Abstract

Multirotor UAVs powered by batteries face limitations due to the low energy density of their energy source, which constitutes a significant portion of the total weight. During missions, the high battery mass remains constant, necessitating high required power. This leads to reductions in payload capacity and endurance constraints. This study developed a design tool for multirotor UAVs that sequentially detach used batteries during missions to reduce weight and extend endurance. The developed tool consists of a battery weight prediction model and a required power prediction model. It accurately predicts endurance by considering changes in weight, thrust, RPM, motor-propeller efficiency, and required power at each battery separation point. Using the developed tool, the battery separation technology was applied to a quadcopter with total weights of 7, 15, and 25 kg, and the extended endurances were quantitatively compared. The results showed endurance improvements of 127.3%, 122.0%, and 127.0% for the 7, 15, and 25 kg quadcopters, respectively, compared to using a single battery. In addition, the method was applied to the commercially available industrial UAV DJI Matrice 300 RTK. With a 2.7 kg payload, the two-stage battery configuration extended the endurance by 12.5% compared to the single-battery case. Under no-payload conditions, a three-stage configuration achieved a 16.7% improvement. These results confirm the effectiveness of staged battery detachment even in real-world UAV platforms.

1. Introduction

Multirotor unmanned aerial vehicles (UAVs) are widely employed in diverse applications such as surveillance, delivery, and aerial imaging, owing to their capability for low-altitude, low-speed forward flight and hovering. Recent developments have introduced multirotor UAVs equipped with distributed electric propulsion systems, which significantly reduce noise emissions and offer the advantage of zero carbon output. However, due to the inherently low energy density of batteries, a large proportion of the total weight is occupied by battery mass, resulting in limited endurance and constraints on long-duration or long-range missions. To enhance the operational viability of battery-powered multirotor UAVs, various strategies have been investigated [1,2,3]. Hydrogen fuel cells with higher energy density [4] and tethered power supply systems [5] have been proposed to compensate for the limitations associated with battery performance.

Hydrogen fuel cells generate electrical energy through redox reactions between hydrogen and oxygen, providing the required power for flight. Hydrogen possesses an energy density that is approximately 92% to 170% higher than that of lithium-based batteries, making it suitable for application in long-endurance aerial vehicles [6,7,8,9,10,11,12]. However, the instantaneous power output that can be produced through these chemical reactions is inherently limited. As a result, auxiliary batteries are required to meet high power demands during momentarily high-thrust phases such as take-off and landing. This requirement increases the overall system complexity and contributes to additional weight. Moreover, the need for large storage tanks to contain hydrogen presents significant challenges for integration into small-scale multirotor UAVs.

In the case of tethered power supply systems [5], electrical power is continuously delivered through ground-based external cables, enabling long-duration missions without the need for battery replacement or recharging. However, when the cable length exceeds a certain threshold, the additional mass of the cable and the increase in electrical resistance lead to power losses, ultimately limiting the operational range. Furthermore, during the mission, the cable path must be carefully managed to prevent entanglement with surrounding obstacles or the UAV itself, which imposes significant constraints on the maneuverability and operational flexibility of the multirotor UAV.

Another approach [12] involves charging the UAV battery by harvesting electromagnetic energy from power transmission lines. In this system, the UAV perches on the power line and utilizes the surrounding magnetic field to recharge its battery, offering the advantage of implementation without requiring significant modifications to existing infrastructure. However, this method has several limitations. The amount of power that can be harvested is limited, resulting in long charging times. Additionally, the need for the UAV to physically perch on the power line imposes mechanical stability and safety challenges. Moreover, the applicability of this method is confined to areas with existing power transmission infrastructure, which restricts its operational range.

In contrast to approaches that aim to compensate for the low energy density of batteries, studies have been conducted to enhance endurance by separating depleted batteries from the multirotor UAV during flight. Two primary strategies have emerged: battery replacement and multi-stage battery detachment. The battery replacement concept, proposed by Guetta and Shapiro [13], involves ejecting a used battery from the UAV and receiving a fully charged battery during mission from another UAV. In this system, the mission-performing UAV and the battery-supply and retrieval UAVs operate as a coordinated pair to sustain long-duration missions. This method allows continuous operation while new batteries are being delivered. However, it requires the presence of two separate UAVs: one for executing the mission and another for battery supply and retrieval. Importantly, the battery-supply and retrieval UAVs must carry heavy batteries, which significantly restricts their operational range. Moreover, the delivery and retrieval process demands additional onboard equipment, increasing total gross weight of the UAV and, consequently, its required power. As a result, the overall mission range becomes dependent on the operational range of the battery-supply and retrieval UAVs.

As shown in Figure 1, the multi-stage battery detachment method involves equipping a UAV with multiple fully charged batteries and sequentially detaching the depleted ones during flight, thereby reducing the overall weight and extending endurance. The multi-stage battery detachment method was first proposed by Jain et al. [14], who experimentally demonstrated, using a configuration with two batteries, that the multi-stage battery detachment contributes to an increase in endurance. Furthermore, the same study [14] confirmed that stable control of the UAV can be maintained even when approximately 20% of the total weight is suddenly detached downward during flight, causing a vertical shift in the center of gravity, without inducing significant changes in the pitch or roll axes. These results suggest that as the number of battery stages increases and the weight of each individual battery decreases, stable control remains feasible and further improvements in endurance can be achieved. However, the theoretical model they proposed had limitations in accurately predicting the required power and battery weight, which constrained the precise estimation of the optimal number of battery stages and the corresponding mission duration.

This study proposes a conceptual design framework for multirotor UAV systems aimed at improving flight efficiency through a multi-stage battery detachment mechanism. This approach is intended to overcome the limitations of single battery-powered UAVs, which are constrained in performing long-duration surveillance and reconnaissance missions due to limited endurance. The proposed approach is particularly suitable for missions that require long-duration hovering over a fixed location, such as infrastructure inspection, or long-range flights for environmental monitoring or search and rescue. In addition, for missions involving extended hovering, the integration of parachutes with the detached batteries can mitigate impact forces during descent and enable battery reuse, helping offset the potential drawbacks of battery detachment.

Although this strategy involves dropping used batteries mid-flight, in missions that require stationary hovering over a designated area, the drop zone is predictable and localized. Therefore, battery retrieval can be systematically planned and executed, mitigating potential safety concerns. Furthermore, since no additional recovery equipment or complex operational procedures are required, this strategy offers high practical value in operational environments that demand repetitive and immediate deployment.

The design tool is based on Blade Element Momentum Theory (BEMT) to estimate the required power according to the total weight of the UAV at each battery stage and incorporates a battery weight prediction model as a function of battery capacity. This enables a quantitative assessment of thrust variation and required power in response to progressive weight reduction. Based on the comprehensive design tool, the optimal number of battery packs required to satisfy specific mission duration and performance requirements can be determined.

This study investigates the effectiveness and practical applicability of a multi-stage battery detachment strategy using a developed integrated design tool. The analysis is conducted in two main categories. The first category focuses on configurations in which the battery accounts for 75% of the total takeoff weight, a condition known to maximize endurance in multirotor UAV operations. Under this setting, hovering time and effectiveness of the proposed strategy were evaluated for UAVs with total takeoff weights of 7 kg, 15 kg, and 25 kg. The second category addresses configurations equipped with mission payloads, which result in a relatively lower battery-to-weight ratio. The detailed methodology and the results are described below.

First, to assess the strategy under endurance-optimized conditions, a configuration with a 75 percent battery-to-weight ratio was applied. This design reflects typical characteristics found in long-endurance multirotor UAV designs. For a 7 kg UAV, the optimal number of battery stages and corresponding hovering duration were estimated, demonstrating potential to significantly extend flight time.

Second, the same high battery weight configuration was applied to heavier UAVs with takeoff weights of 15 kg and 25 kg to examine the scalability of the approach. These systems are expected to benefit from reduced in-flight mass, potentially improving endurance. However, power requirements increase sharply with total weight, which inherently limits flight time. Typically, such UAVs achieve only 10 to 20 min of endurance under conventional single-battery setups. To quantify possible improvements, the proposed strategy was systematically applied to both platforms, and its effectiveness was evaluated.

Finally, to examine applicability of the proposed strategy to commercial UAV platforms equipped with mission payloads, the multi-stage battery detachment method was implemented on the DJI Matrice 300 RTK platform. In this configuration, the presence of payload reduces the proportion of battery weight relative to the total takeoff mass. Performance was analyzed under both payload and no-payload conditions to quantitatively assess improvement in endurance achieved by applying the strategy to UAVs with reduced battery-to-weight ratios due to payload installation. This analysis confirms practical feasibility of the approach and effectiveness in enhancing endurance across a range of UAV configurations.

2. Development of Design Tool

In efforts to improve the endurance of electrically powered multirotor UAVs, conventional strategies have primarily involved increasing battery energy density, employing a high-capacity battery, or enhancing propeller efficiency to reduce power consumption, as expressed in Equation (1). However, these approaches are inherently limited by increased structural weight, system complexity, and diminishing returns in efficiency improvements. As shown in Equation (1), $p$ denotes the power required by the propeller at battery, ${\eta }_{motor}$ represents the efficiency of the brushless DC motors, M_n is the number of motors installed on the UAV, and $V$ refers to the nominal voltage of the battery:

$$T={\displaystyle \frac{Energy}{power}}={\displaystyle \frac{C_{battery}V{\eta }_{motor}}{p{M}_n}} \left[\mathrm{h}\mathrm{o}\mathrm{u}\mathrm{r}\right].$$

(1)

To overcome these limitations, this study proposes a multi-stage battery detachment method as an alternative approach. A multirotor UAV employing multi-stage battery detachment begins its mission at maximum takeoff weight and experiences a gradual reduction in mass as depleted batteries are released during flight. Consequently, it is essential to accurately predict the changes in weight at each separation stage, along with the corresponding variations in propeller thrust and power requirements. This is particularly important for battery-powered UAVs, as the battery constitutes a significant portion of the total system mass. Accurate estimation of battery weight is therefore critical for reliable performance prediction. Furthermore, because the staged separation strategy involves repeated prediction of requirements power at each stage, the accuracy and computational efficiency of the prediction model become increasingly important as the number of battery stages increases.

In this study, an integrated analysis tool was developed by combining a requirement power prediction model, based on Blade Element Momentum Theory (BEMT), a widely accepted and efficient method for analyzing rotor performance, with a newly developed battery weight prediction model. The structure of the integrated tool is illustrated in Figure 2. The tool begins by setting the target endurance ($E_{target}$), the total number of battery stages (I), and the initial total weight of the multirotor UAV (W_i).

The first component, the required power prediction model, estimates the required power based on the weight of the UAV (W_i) at a given battery stage (i). The second component, the battery weight prediction model, calculates the required battery capacity (C_battery) [mAh] and battery weight ($W_{battery, i}$) [kg] at each stage i, considering the predicted power and the endurance allocated per battery stage. The battery capacity required at each stage is calculated by rearranging Equation (1), which relates flight time to power consumption, resulting in Equation (2). As shown in Equation (2), $p_i$ denotes the power required per propeller at battery stage i, $t_i$ denotes the flight time assigned to battery stage i.

$$C_{battery}={\displaystyle \frac{p_i{t}_i{M}_n}{V{\eta }_{motor}}} \left[\mathrm{m}\mathrm{A}\mathrm{h}\right].$$

(2)

In this context, the endurance per battery pack is assumed to be uniform across all stages and is calculated by dividing the target total endurance ($E_{target}$) by the total number of battery stages (I), as shown in Equation (3). This assumption was made to clearly evaluate the effect of multi-stage battery detachment by minimizing the influence of other variables.

$$t_i={\displaystyle \frac{E_{target}}{I}} \left[\mathrm{h}\mathrm{o}\mathrm{u}\mathrm{r}\right]$$

(3)

Based on the predicted battery weight at the current stage (i), the total weight at the next stage (i + 1) is estimated using Equation (4). This battery separation process is iteratively performed until the final stage (I) is reached.

$$W_{i+1}=W_i-W_{battery, i} \left[\mathrm{k}\mathrm{g}\right]$$

(4)

Given a battery detachment count I, the total battery weight is calculated by summing the weight of each battery stage. Using this approach, the maximum endurance is predicted for configurations where I ranges from 2 to 500. Among the predicted results, the optimal number of battery stages is selected as the value of I that satisfies the target endurance ($E_{target}$) while minimizing the total battery weight.

2.1. Required Power Prediction Model

For multirotor UAVs powered by batteries and electric motors, accurate prediction of the required power must take into account the combined efficiency of the motor and propeller. This motor–propeller efficiency varies with the total weight of the UAV, and in the case of UAVs employing staged battery separation, the efficiency changes at each separation point due to the corresponding change in system weight.

Multirotor UAVs equipped with BLDC (Brushless DC) motors can adjust their RPM at each battery separation stage to generate the required thrust. In this study, a lookup table was constructed to represent the variation in BLDC motor efficiency, which is commonly used in multirotor systems. To reflect the variation in motor efficiency according to thrust, a lookup table was constructed using thrust-dependent efficiency data provided by the motor manufacturer. This table was applied in the analysis so that motor efficiency dynamically varied with thrust during flight.

To account for the variation in propeller efficiency with respect to changes in RPM or required thrust, Blade Element Momentum Theory (BEMT) was employed. BEMT combines Momentum Theory and Blade Element Theory. In Momentum Theory, the flow around the propeller is modeled as a streamtube, and the analysis assumes a one-dimensional, inviscid, incompressible, and irrotational flow. Under these assumptions, thrust is generated as a result of the change in momentum of the fluid flowing through the streamtube—from its entry point, across the propeller disk, and to the exit point.

Blade Element Theory divides the propeller into a series of virtual elements and predicts the lift and drag generated at each element by accounting for the local geometric and aerodynamic flow conditions. The thrust and torque can then be calculated using the lift and drag forces at each element, following Equations (5) and (6). By incorporating the induced velocity predicted from Momentum Theory, three-dimensional effects, which are not inherently captured in Blade Element Theory, can be indirectly taken into account [15].

$$dT=N_b(dL\cos \varphi -dD\sin \varphi )$$

(5)

$$dQ=N_b(dL\sin \varphi +dD\cos \varphi )$$

(6)

At the tip region of an actual propeller, strong tip vortices are generated, leading to a significant reduction in thrust due to an increase in induced velocity—this phenomenon is known as the tip loss effect. To account for this, the present study incorporates the Prandtl tip loss function, defined in Equations (7)–(9), into the BEMT framework to correct for tip loss. In these equations, $\sigma$ denotes the solidity, $\lambda$ is the inflow ratio, $r$ is the blade radius, $C_{l_{\alpha }}$ is the lift curve slope,  $N_b$ is the number of rotor blades, $\varphi$ is the inflow angle, and $F$ represents Prandtl tip loss function [15].

$$F={\displaystyle \frac{2}{\pi }}{\cos }^{-1}\left(e^{-f}\right)$$

(7)

$$f={\displaystyle \frac{N_b}{2\left(\frac{1-r}{r\varphi }\right)}}$$

(8)

$$\lambda \left(r\right)={\displaystyle \frac{\sigma C_{l_{\alpha }}}{2\left(\sqrt{1+\frac{32F}{\sigma C_{l_{\alpha }}}\theta r}-1\right)}}$$

(9)

To validate the power requirement prediction model, the results were compared with experimental data measuring the Figure of Merit (FM) under various thrust conditions [16,17]. Mantay et al. [16] conducted experiments at the Langley Research Center of NASA by mounting the UH-1 main rotor on a rotating test stand. Berry et al. [17] performed tests under both hover and forward flight conditions using a scaled UH-1 helicopter main rotor in a 4 × 7 m wind tunnel facility.

In this study, the FM was predicted under hover conditions across a range of thrust levels and was validated against experimental results [16,17]. FM is a nondimensional parameter that evaluates rotor efficiency by comparing the ideal power to the actual power consumed, and it is defined as shown in Equation (10).

$$Figure of merit={\displaystyle \frac{Ideal power}{Autal power}}={\displaystyle \frac{\frac{C_T^{1.5}}{\sqrt{2}}}{\kappa \frac{1}{2}\sigma {\int }_0^1{C}_l{r}^{2}dr+\frac{\sigma }{2}{\int }_0^1{C}_d{r}^{3}dr}}$$

(10)

Figure 3 presents the FM values predicted by the developed power requirement model in comparison with the experimental data [16,17], over a thrust coefficient range of 0.001 to 0.0045. The predicted FM agreed with the experimental results within 5% across the entire thrust range.

2.2. Battery Weight Prediction Model

To determine the optimal number of battery stages for a multirotor UAV employing staged battery separation, it is essential to accurately predict not only the total battery weight but also the weight associated with the battery capacity at each stage. In particular, since battery weight increases nonlinearly with capacity, it is important to account for this characteristic in the analysis.

In a previous study [14], the battery weight was predicted based on a constant energy density of 130 Wh/kg for a quadcopter weighing less than 1 kg, assuming a linear relationship between battery capacity and weight. However, as shown in Figure 4, actual batteries consist not only of energy-storing cells but also of additional components such as circuits and wiring for current control and interconnection, as well as structural elements for securing the cells. When battery capacity decreases, the weight of the energy-storing cells decreases linearly, but the circuits and structural components maintain a minimum weight and volume. As a result, the proportion of energy-storing cell mass relative to the total battery weight decreases with lower capacities. Therefore, applying a constant energy density across the entire capacity range, as in [14], may lead to overestimation or underestimation of battery weight in certain capacity ranges, ultimately reducing the accuracy of the prediction.

A new battery weight prediction model was developed to indirectly account for the structural components of batteries. The model was constructed based on the weight variation observed with respect to battery capacity, using data collected from 249 commercially available lithium-polymer batteries listed on online marketplaces. A new battery weight prediction model was developed to indirectly account for the structural components of batteries. To determine the most accurate approach, both cell-specific models and a unified model using battery energy (Wh) as the independent variable were developed and compared. A total of 249 commercially available lithium-polymer batteries were collected from online marketplaces, covering 3-cell (3S), 4-cell (4S), and 6-cell (6S) configurations to ensure applicability across a wide range of multirotor UAV platforms. For each configuration, the relationship between battery weight and capacity was analyzed, and predictive models were formulated accordingly. The performance comparison showed that the cell-specific models consistently outperformed the unified model. For the 3S configuration, the standard deviation of prediction difference was 0.034 for the cell-specific model, compared to 0.037 for the unified model. In the 4S case, the cell-specific model achieved a standard deviation of 0.038, while the unified model resulted in 0.042. Similarly, for the 6S configuration, the prediction difference increased from 0.056 in the cell-specific model to 0.060 in the unified model. These results indicate that the cell-specific models yield higher predictive accuracy. In particular, for lightweight batteries, the prediction difference using the unified model was found to increase by up to 13%, highlighting the importance of using configuration-specific models for precise weight estimation in such cases. Figure 5 presents the correlation between predicted battery weights and the weights specified in the manufacturer catalog across different cell configurations. Each subplot uses the specified weight for the x-axis and the predicted weight for the y-axis. The solid diagonal line represents an ideal 1:1 agreement, and the surrounding gray band indicates the ±3σ range calculated from the residual standard deviation. Most data points are located within this range, indicating high prediction accuracy of the model. Differences in the spread across configurations are attributed to variations in form factor and packaging methods.

Accordingly, the cell-specific models were adopted as the final prediction framework in this study. Equations (11)–(13) present the derived expressions, where x represents the battery capacity.

$$W_6=-7\times {10}^{-10}{x}^2+1.4\times {10}^{-4}x+0.033$$

(11)

$$W_4=-1.8\times {10}^{-9}{x}^2+10\times {10}^{-5}x+0.012$$

(12)

$$W_3=-1.5\times {10}^{-9}{x}^2+8\times {10}^{-5}x+0.011$$

(13)

Figure 6 presents the relationship between battery capacity and battery weight for 3S, 4S, and 6S lithium-polymer batteries. The dotted lines represent the battery weight predictions based on the constant energy density used in the previous study [14], while the solid lines indicate the results from the proposed battery weight prediction model. The actual battery weights are also plotted as data points. As shown in Figure 5, the predictions made using a constant energy density, regardless of cell configuration, tend to diverge from the actual values as battery capacity increases. In particular, for the 4S 8000 mAh battery, the model in [14] overestimates the actual weight by 56.49%. In contrast, the proposed battery weight prediction model demonstrates high accuracy, with coefficients of determination (R²) of 0.96, 0.97, and 0.996 for 3S, 4S, and 6S batteries, respectively.

2.3. System Verification

The objective of this study is to verify the reliability of the integrated design framework by replicating the experimental analysis previously conducted to evaluate the feasibility of staged battery detachment. This verification requires input data including the motor efficiency defined in Equation (2), the propeller characteristics described in Equation (10), and the battery weight and capacity specified in Equations (11)–(13).

The previous experiment [14] involved a quadcopter with a total weight of approximately 0.98 kg without any payload, equipped with four propellers and two batteries, each weighing 0.19 kg with a capacity of 2.2 Ah, resulting in a total battery mass of 0.38 kg. One battery was detached at 0.155 h (9.3 min), and the remaining battery sustained flight until 0.38 h (22.8 min). Since the manufacturers of the motor and propeller used in the experiment [14] did not provide efficiency curves or performance specifications, there were limitations in conducting precise numerical validation. To address these constraints, a representative propeller was designed to match the thrust and electrical power requirements observed in the experiment. The total hover power requirement was set at 159 W, and the resulting propeller configuration reflected weight, power, and efficiency variations before and after battery detachment. The propeller design also aimed to reflect the operational condition changes caused by battery separation.

For consistency in aerodynamic modeling and to isolate the effects of battery configuration from aerodynamic influences, the NACA 0012 airfoil was adopted for the blade section. Aerodynamic coefficients were obtained through RANS-based CFD simulations at Reynolds numbers corresponding to low-speed UAV propeller operation. These conditions reflect the typical operational regime of small UAV propellers and have been validated in previous studies [18,19,20]. These data were incorporated into the BEMT framework, enabling accurate prediction of thrust and power performance.

A rotor radius of 0.203 m was selected to maintain identical disk loading (thrust-to-disk area ratio) with the experimental setup. The propeller geometry was configured with a solidity ratio of 0.2 and a linear twist distribution with a root pitch angle of 10°, yielding the desired thrust of 0.25 kg per rotor for a quadcopter configuration. This configuration matched the experimental maximum takeoff weight and power requirement.

Given the absence of motor-specific efficiency data in the previous study [14], three alternative motors—Antigravity 4004 KV300, MN2806 KV400, and AIR GEAR 450 II KV920—were selected for analysis. These motors, while differing in kV rating and recommended propeller size, are capable of delivering the required thrust and are representative of commercially available options. Efficiency data for these motors were used even if obtained under different propeller configurations, to assess sensitivity in power prediction. Their efficiency profiles, extracted from manufacturer data, are presented in Figure 7.

Accordingly, the predictive accuracy of the integrated analysis model was validated by comparing the battery mass estimated using the proposed model with the reference values reported in the previous study [14]. The model incorporated the initial battery detachment time and total flight duration, and the resulting required power and battery mass predictions for the three motor configurations are presented in Figure 8.

As illustrated in Figure 8, the model accurately predicted electrical power during the initial flight phase, prior to battery detachment. This confirms the suitability of the BEMT-based propeller design framework under conceptual-level constraints. After the battery detachment at 0.155 h (9.3 min), the maximum difference between the predicted and measured power reached up to 7%. This discrepancy is primarily attributed to the inability to precisely replicate the experimental conditions, mainly due to variations in motor efficiency—up to 12.2% depending on operating conditions as shown in Figure 7—and differences in propeller geometry.

Despite these limitations, the battery mass predictions across all motor cases were nearly identical, with less than 1% variation. The first battery was estimated to be 2.2 Ah and 0.187 kg, while the second was estimated at 2.4 Ah and 0.201 kg. Compared with experimental data, the first battery prediction exhibited a 0.003 kg (1.6%) underestimation, and the second showed a 0.011 kg (5.8%) overestimation. The total battery mass prediction difference was limited to 2.1%, demonstrating the high predictive accuracy of the model.

It is noteworthy that the current model does not explicitly incorporate certain loss mechanisms, including ESC efficiency degradation, motor performance variations due to temperature rise, and battery voltage drop associated with C-rate dynamics. Nonetheless, the model exhibits robust predictive capability under conceptual design constraints. Future enhancements will integrate these secondary effects to improve accuracy and generalizability.

3. Results

3.1. The Effective Number of Batteries

This study investigated the hover endurance performance of a commercially relevant 7 kg multirotor UAV by applying various numbers of battery stages. For this configuration, the MN7005 KV230 motor was used, and the propeller design parameters were as follows: a radius of 0.3 m, a solidity of 0.07, and a linear twist angle of 25°. To clearly demonstrate the effectiveness of the battery separation technique, the study was conducted without any payload. Figure 9 shows the variation in required power with respect to flight time for different numbers of battery stages, assuming a target endurance (E_target) of 2.5 h (150 min). As shown in Figure 9, based on the simulation tool, the single-battery configuration achieved a maximum endurance of approximately 1.1 h (66 min) no-payload. However, when the multi-stage battery detachment method was applied, the required power decreased in a stepwise manner at each battery detachment point. For configurations with up to 100 battery stages, the system was able to satisfy the target endurance (E_target) without exceeding the MTOW, thereby meeting the mission requirements. In contrast, when the number of battery stages increased to 500, the system failed to meet the mission constraints. In particular, as shown in Figure 10, which presents the total battery weight according to the number of battery stages, the total battery weight decreased as the number of battery stages increased up to 13. However, beyond 13 stages, the total battery weight began to increase again. This result contrasts with the prediction made in a previous study [14], which suggested that increasing the number of battery stages would continuously improve endurance performance.

As shown in Figure 9, when the number of battery stages increases to 100 or 500, the system weight continuously decreases during flight, similar to fossil fuel-powered aircraft, and the required power gradually decreases as well. In the case of 100 stages in Figure 9, where the required power is reduced and energy consumption becomes minimal as the number of stages increases, the energy density-based approach predicts the battery weight and endurance as the most ideal values. Accordingly, the experimental study [14] predicted that if an unlimited number of batteries could be added, continuous weight reduction would allow for maximized endurance. However, this study conducted an analysis using a battery weight prediction model developed to account for the impact of structural components on overall battery mass. In battery systems, non-energy components such as wiring and circuit boards must be included to enable power supply, distribution, and control functions. These non-energy components cannot be easily reduced in size or weight due to manufacturing limitations, even when the battery pack is miniaturized. As a result, the proportion of these components in the total battery weight increases as the battery size decreases. Consequently, when the number of battery stages exceeds a certain threshold, endurance begins to decrease rather than increase. This indicates that employing a large number of low-capacity batteries may lead to a reduction in total battery capacity and an increase in total battery weight, ultimately resulting in decreased system efficiency.

A theoretical analysis was conducted to identify the optimal conditions by applying various numbers of battery stages to a 7 kg multirotor UAV. The analysis showed that, for a 7 kg quadcopter with a target endurance of 2.5 h (150 min), configuring 13 detachable battery stages resulted in a 127.3% improvement in flight performance compared to a single large battery, while reducing the total battery weight from 5.25 kg to 5.11 kg—a decrease of 2.7%. These findings suggest that there exists an optimal battery stage configuration depending on the mission requirements, and highlight that such an approach serves as an important design strategy for reducing system weight and increasing the payload capacity available for mission equipment.

3.2. Heavy-Weight Multirotor UAV

As the weight of a multirotor UAV increases, both the required thrust and power also increase, leading to a reduction in endurance. In particular, for multirotor UAVs with a large MTOW, the required power becomes significantly higher due to the added weight of components such as motors, frames, ESCs, and payloads, while the relative proportion of battery weight decreases. Figure 11 presents a scatter plot of operating multirotor UAVs, showing the relationship between their total weight and endurance. It can be observed that heavier multirotor UAVs generally exhibit shorter flight durations compared to lighter ones. Notably, multirotor UAVs weighing over 10 kg that are capable of flying for more than 25 min are rarely found.

Building on the significant improvement in endurance observed when applying the battery separation technique to a 7 kg multirotor UAV, this study further analyzed the changes in endurance resulting from the application of the technique to heavy multirotor UAVs. The objective was to assess the feasibility of applying the battery separation strategy to high-MTOW platforms.

The battery separation technique was applied to 15 kg and 25 kg multirotor UAVs to predict the optimal number of battery stages from a range of 2 to 500. For the 15 kg UAV, the MN701S KV280 motor was used along with a propeller having a radius of 0.29 m, solidity of 0.19, and a linear twist angle of 27°. For the 25 kg UAV, the P80 KV120 motor was applied with a propeller of 0.38 m radius, 0.13 solidity, and a twist angle of 28°. The variations in required power over endurance for each case are presented in Figure 12, respectively. As a result of applying staged battery separation, the endurance increased by 122% and 127% for the 15 kg and 25 kg UAVs, respectively, compared to using a single battery. For both weight classes, the total endurance remained below 2 h (120 min), which is shorter than that of the 7 kg UAV. However, compared to the single-battery configuration, the application of battery separation technology demonstrates a substantial improvement in endurance, even for high-MTOW multirotor UAVs.

The optimal number of battery stages was determined to be 13 for the 7 kg UAV, 20 for the 15 kg UAV, and 27 for the 25 kg UAV. Under these conditions, the total predicted battery weights for the 15 kg and 25 kg UAVs were 11.11 kg and 18.65 kg, respectively. Compared to the 7 kg UAV using 13 battery stages, the increase in UAV weight resulted in a corresponding increase in the optimal number of battery stages for the heavier platforms.

The reason for the increase in the optimal number of battery stages in heavy-lift multirotor UAVs is as follows. First, when multiple batteries are used and the depleted batteries are jettisoned early in the flight, a significant weight reduction effect is observed. In the case of heavy UAVs, high required power results in larger capacity and heavier batteries for each stage. The weight of a single battery stage was estimated to be approximately 0.39 kg, 0.55 kg, and 0.69 kg for 7 kg, 15 kg, and 25 kg UAVs, respectively. Due to the higher weight of a single battery in heavy UAVs, the first battery separation alone leads to a reduction in required power by approximately 11% for the 15 kg UAV and 8% for the 25 kg UAV. This reduction contributes to an overall decrease in total weight and is expected to improve overall flight efficiency.

Furthermore, the internal structure of a battery comprises both energy-storing components and non-energy components. Since the weight of non-energy components does not scale linearly with battery capacity, excessive subdivision of the battery into smaller units can result in a relatively high proportion of non-energy mass. This limits the potential for further system weight reduction and may lead to a deterioration in overall system efficiency as the number of battery stages increases. Therefore, employing batteries with a capacity above a certain threshold helps reduce the mass fraction of non-energy components and mitigates the efficiency degradation typically associated with an increased number of battery stages.

3.3. Applying Commercial Multirotor UAV

To evaluate the effectiveness of the multi-stage battery detachment method, a performance analysis of the DJI Matrice 300 RTK was conducted using a representative propulsion system configuration. The MN7005 brushless motor was selected as a substitute for the actual motor, and a custom-designed propeller was applied based on a NACA0012 airfoil cross-section, with a radius of 0.267 m, a solidity of 0.16, and a twist angle of 27.5°, matching the publicly available specifications of the Matrice 300 RTK. This setup was used to simulate hover performance under both payload and no-payload conditions while varying the number of battery stages from 2 to 500. The results showed that, with a payload installed, the total battery weight began to increase beyond three stages, indicating that utilizing two stages is optimal in such cases. In contrast, without a payload, battery mass started to rise after the fourth stage, suggesting that three stages yield the best performance. This difference is primarily attributed to the presence of various mission equipment—such as cameras, video transmitters, receiver modules, telemetry systems, LiDAR, and ultrasonic sensors—which reduce the proportion of battery mass relative to the overall system weight. For instance, in a previously analyzed 7 kg quadcopter with 13 battery stages, the batteries accounted for approximately 75% of the total system weight. This configuration is representative of long-endurance UAVs, which are typically designed with lightweight frames and minimal payloads to maximize battery capacity. These findings suggest that, when multirotor UAVs carry heavy mission equipment, the proportion of battery mass relative to the total weight decreases. As a result, even with the application of staged battery detachment, the overall weight remains high due to the payload, leading to persistently high-power demand and a relatively smaller improvement in hover time compared to the no-payload condition. As shown in Figure 13, under payload conditions, the single-battery configuration provided a maximum hover time of 28.8 min, which increased to 32.4 min when transitioning to a two-stage system. This improvement resulted from a mass reduction of approximately 1.56 kg (17.3%) following the first-stage battery drop, which led to a 51.38 W (23.2%) decrease in power demand. In contrast, under no-payload conditions, where the battery accounts for a larger proportion of the UAV’s total weight, the benefits of staged detachment became more pronounced. The hover time increased from 48 min with a single battery to 56 min with a three-stage system, representing a 16.7% improvement.

These results imply that in order to maximize hover time, a higher proportion of battery mass relative to the system weight leads to greater benefits from staged detachment. Consequently, reduced power requirements and increased endurance can be achieved. Specifically, under no-payload conditions, the increased relative proportion of battery mass contributed to an increase in both the optimal number of battery stages and the maximum achievable hover time.

4. Conclusions

In this study, an integrated analysis tool was developed by combining a BEMT-based power requirement prediction model with a battery weight prediction model, taking into account the variation in UAV weight at each battery separation point. Using the developed tool, a study was conducted on the staged battery separation method as a means to overcome the inherent limitations of battery-powered multirotor UAVs, particularly their short endurance caused by the low energy density of batteries and the inability to reduce battery weight during flight.

For the widely used 7 kg UAV, applying 13 battery stages resulted in a maximum endurance improvement of 127.3%. For heavier multirotor UAVs, applying 20 and 27 battery stages to 15 kg and 25 kg platforms, respectively, led to endurance increases of 121.0% and 130.7%. These results confirm that the staged battery separation technique is highly effective in significantly enhancing endurance.

An endurance analysis of the DJI Matrice 300 RTK with a 2.7 kg payload showed that the single-battery configuration achieved 28.8 min of flight time, while the staged battery separation method extended it to 32.4 min, resulting in a 12.5% improvement. Under no-payload conditions, endurance increased from 48 min in the single-battery configuration to 56 min in the three-stage configuration, representing a 16.7% improvement. These results indicate that the benefit of staged battery detachment becomes more significant as the battery accounts for a larger proportion of the total system weight, whereas its effectiveness may be limited in UAVs equipped with heavy mission payloads.

Since the scope of this study is limited to conceptual design, future work will involve the development of an optimal battery detachment mechanism that incorporates parachutes and physical separation devices. Based on this, the resulting improvements in hover endurance will be further analyzed.