1. Introduction
In the last 10 years, research in the field of unmanned aerial vehicles (UAVs) has experienced a vast expansion, which is made possible by the development of aircraft components, primarily micro-electromechanical systems (MEMS) sensors, microcontrollers, batteries, and propulsion components. There are several categories of UAVs that are in different stages of research, development, and utilization. Depending on the categories and purpose, the UAVs are designed from the size of a fighter aircraft (unmanned combat aerial vehicle, UCAV), down to micro aerial vehicles (MAVs) [
1]. Multirotor types of UAVs have the capability of vertical take-off and landing, remain stationary in the air (hover), and flight at a moderate speed allowing them to conduct complex maneuvers which makes them suitable for a wide range of tasks. Different multirotor configurations are intended for missions such as surveillance [
2], inspection [
3], applications in the construction management [
4], agriculture [
5], search and rescue missions [
6], manipulation and interaction with the environment [
7], and others. Conventional configurations are characterized by parallel (planar) and symmetrical arrangement of an even number of rotors, most commonly realized as quadrotor (quadcopter) [
8], hexarotor (hexacopter) [
9], or octorotor (octocopter) [
10].
In the multirotor UAV design process, the requirements of the mission or task as part of the mission need to be considered. The main criterion of multirotor design is the required performance that aircraft should be able to achieve during the flight mission. Given that the propulsion system should provide the thrust required for the movement of the aircraft, respectively achieve the required flight performance, the selection of parameters and components of the propulsion system is the most important and complex step. The configuration parameters, i.e., the geometric arrangement of the propulsion system determines the distribution of the propulsion units generated forces and moments to the control forces and moments of the propulsion system. Since it is possible to achieve the full degree of controllability of the system by selecting the particular configuration parameters [
11], multirotor UAVs are a typical representative of aerial robots due to the possibility of performing precise and complex movements. The dynamics, as well as energy consumption, directly depend on the selected components and the number of propulsion units. It turns out that the propulsion and energy systems are interdependent and when choosing components, it is necessary to maintain a balance with the existing constraints defined by the mission.
Most commonly, the multirotor propulsion system consists of pure electric propulsion (i.e., electric motors fitted with an appropriate propeller and powered by an electrochemical battery). Numerous papers have investigated and presented ways to identify the parameters of multirotor propulsion units [
12,
13]. Regardless of the propulsion type, the physical parameters on the one hand represent the aerodynamic forces and moments generated by the propulsion unit while on the other hand, there are the parameters of energy consumption. In [
14,
15], experimental setups for the identification of parameters are presented, and there are also test stands available on electric propulsion unit (EPU) market [
16]. In addition to experimental identification, research has been conducted with the aim of a more detailed and accurate mathematical description of the rotor [
17]. The efficiency of the propulsion configurations of aircraft with an overlap of propulsion surfaces was also investigated since such rotor arrangements are characterized by a loss of the total thrust force [
18]. Additionally, in [
19,
20] the relation of energy consumption and flight time (autonomy) are considered.
In this paper, a method for characterization of the electric propulsion system is proposed, which is an important step in the multirotor UAV design process. The method was validated using experimental measurements of various EPU setups. The parameter identification procedure is presented which, based on experimental measurements data or manufacturer’s data, results in EPU static maps. Such static maps exactly show the physical parameters in relation to the control (PWM) signal and characterize EPU. The characteristics show aerodynamic forces and moments with respect to rotor angular velocity as well as energy consumption and overall efficiency measure. The proposed characterization method without significant modifications can be applied to the full power range of EPUs, from a few watts to a few tenths of kilowatts which can further lead to parameter estimation, system analysis, and optimization.
The paper is organized as follows.
Section 2 describes multirotor UAV system. The EPU parameters identification procedure and propulsion static maps are shown in
Section 3. The propulsion characterization and comparison are presented in
Section 4. The conclusion follows in
Section 5.
3. EPU Parameters Identification Procedure
Identification of EPU physical parameters is the basis for the characterization of EPU and it is required for the study of propulsion and power source systems. For this purpose, experimental measurements were conducted. The heaviest and most expensive component of EPU is the BLDC motor, which affects the selection of other propulsion components (propeller and ESC). Therefore, it was chosen as the starting point of this study. Additionally, the selection of EPU components must be accompanied by a properly selected energy source, i.e., a battery that poses required voltage and capacity.
Table 1 shows the components used in experimental measurements. Based on the measurement results, identification of the EPU parameters was done and EPU static maps were generated as shown in the following subsections. EPU static maps are then used in the next section to perform the characterization of each EPU and characterization of the entire electric propulsion system.
3.1. Experimental Setup
For the purposes of experimental measurements, it is necessary to select measuring equipment that has sufficient accuracy, resolution, and compatibility with the appropriate software package for data acquisition. There are several established setups [
24] and commercially available measurement systems such as RCbenchmark 1580 [
16] (see
Figure 4) which was used in this research. Measurement of mechanical quantities, i.e., aerodynamic forces (
) and moments (
), takes place using a dynamometer consisting of three load cell sensors. The rotor RPM was measured electrically by a measuring probe that was connected to a single motor phase, and optically through an optical sensor that counts revolutions by detecting a marker mounted on the motor. From electrical quantities, battery voltage (
), electric current (
), and electric power (
) were measured. In addition to the mentioned mechanical and electrical quantities, the motor temperature can be monitored by utilizing an additional temperature sensor. Furthermore, the vibrations of propellers were measured by an accelerometer (embedded within test stand), and it was checked whether their intensity is below the limit that would significantly affect the results. In the case of unfavorable vibrations, or if measured values exceed the defined limit values, the software reports an error and stops the measurement process (safety cutoff) to protect equipment from damage. Additionally, in order to validate the setup, it is important to check if correct RPM is obtained (i.e., with optical rpm probe), and additionally check the validity of electrical measurements by a multifunction logger.
Load cell sensors were used to measure thrust and drag from the propeller. The first sensor with a range of up to 5 kg, is located vertically on the setup and used to measure the thrust of the EPU (
Figure 4a). The base with mounted EPU is attached to the left and right load cell sensors, that are used to measure the drag torque. The strain gauges of load cells are connected to signal amplifiers that are 24-bit analog-to-digital converters (ADCs) integrated into a data acquisition board. According to the diagram in
Figure 4b, ESC, other sensors, and power supply were connected to the setup control board, which connects the PC via USB cable. Prior to the measurements, the dynamometer was calibrated according to the procedure described in the installation documentation [
16]. To verify the measurement of electrical quantities and angular velocity of the rotor, a multifunctional logger PowerLog 6S [
25] was used. Signal acquisition and data storage in .csv format was performed in the software package that comes with the setup.
3.2. Data Acquisition
The throttle signal sent to the ESC drive was a standard 50 Hz PWM signal with “ON” time ranging from 1000 to 2000 μs (dependent on the ESC). It is possible to send individual PWM signals (manual mode) or send varying signals defined in the RCbenchmark software measurement script (automated mode). Since the goal was to automate and unify measurements, an available script was modified to take four measurements for each PWM value within one measurement cycle. The script is changing the PWM signal from minimal to a maximal value and vice versa in discrete steps, twice. Every PWM signal is held constant for 10 s after which the PWM “ON” time increases or decreases by 100 μs depending on the step of the measurement cycle. The flowchart of the setup script is shown in
Figure 5. Upon completion of the measurement cycle, the .csv file is automatically generated in which the rows represent the actual PWM signal sent to ESC and columns represent the measured values (i.e., thrust, current, and others).
3.3. Identification Procedure
The generated .csv files were loaded, processed, and graphically presented using the MATLAB software package. Regarding a relatively large number of measurements, this process was automated with a customized MATLAB script. The flowchart of the MATLAB script is shown in
Figure 6. In the first step, the script finds all .csv files in the defined measurement root folder and in its subfolders. All data from each .csv file is then read and appended into a common data array. Once the array is filled with data from all measurements, raw measurement data are plotted. By averaging measurement data obtained at the same PWM values (and with further data manipulation), output vectors required for EPU static maps were generated and plotted.
Figure 7,
Figure 8 and
Figure 9 show the plot of raw data measurement i.e., revolutions per minute (RPM) in relation to measurement time, and a plot of the static maps for the rotor speed measurement in relation to the input PWM signal. The results of other measurements are shown in the next subsection. Three series of measurements were conducted for selected EPU components (as shown in
Table 1). The first series represents EPU setup with 1806 motors of different motor constants
, the second series represents EPU setup with MN2214 motor and propellers of different geometric characteristics, while the third series represents EPU setup with MN4014 motor and 15″ to 17″ diameter propellers in combination with 4S and 6S LiPo batteries. It can be seen from the figures that lower speeds are expected for motors with lower
, and that diverse ESC may differ in the control signal range.
3.4. EPU Static Maps
Static maps represent processed data and are the first step in characterization. The identified parameters (contained within arrays) are a function of the control PWM signal.
Figure 10,
Figure 11 and
Figure 12 show static maps of aerodynamic thrust force and drag torque with respect to input PWM signal, while
Figure 13,
Figure 14 and
Figure 15 show static maps of electric current and electric power with respect to input PWM signal, for three series of measurements (three different EPU setups).
For the first setup, 5030 and 7024 propellers were tested, where the first two numbers of the designation define the propeller diameter (5″ and 7″), and the other two numbers the propeller pitch angle (3″ and 2.4″). For the second setup, propellers with 9″ to 10″ diameter and various geometric characteristics were selected and for the third setup, propellers with 15″ to 17″ diameter in combination with 4S and 6S LiPo batteries were selected. Generally, it can be seen that EPU setups with lower , paired with larger propeller diameters, achieve higher aerodynamic forces and torques.
Consequently, as the EPU drag torque increases, the power consumption increases. Electrical consumption is extremely important from the aspect of designing a multirotor UAV with a specific purpose (flight tasks). The first two series of measurements were performed using a 12 V power supply equivalent to a 3S LiPo battery (
Figure 13 and
Figure 14). The third series (
Figure 15) was done with 4S and 6S batteries as a power source. The experimental measurement of electrical quantities was additionally verified using a multifunction logger. Higher power EPUs generally operate at higher voltages.
5. Conclusions
In this paper, a method for characterizing the propulsion system of a multirotor UAV was proposed with the aim of component comparison. The procedure for parameter identification and characterization was presented through the process of experimental identification which includes experimental measurements and signal processing.
Experimental measurements were performed utilizing several types of propellers with different diameters and several medium-sized BLDC motors. Raw measurements were used to generate static maps for the EPU. With static maps of the individual EPU, characterization of the whole electric propulsion system can be performed. The proposed method thus allows a comparison of the various electric propulsion components and their individual impact on the aircraft system. Additionally, a characterized propulsion system can serve as a basis for further aircraft parameters optimization. From the presented results, it can be concluded that higher power EPUs are more efficient and thus results in greater flight autonomy while a greater number of EPUs (more than quadrotor configuration) yields more payload capacity and also more flight autonomy but are negatively affected by higher mass from the EPUs.
The proposed method was also validated by comparison with commercially available data for some of the motors for which manufacturer specifications were available. With a reliable estimate of the EPU parameters, an estimate of the overall performance of the aircraft can be made. As an example, hover time was estimated. The methodological approach to the design of multirotor aircraft is the subject of future research.