Determination of performance characteristics using FEA-analytical for outer rotor BLDC motor

ABSTRACT


INTRODUCTION
Brushless DC (BLDC) motors are widely used in industrial applications worldwide due to their benefits in terms of high torque, light weight, noiseless operation, and high efficiency [1], [2].Furthermore, the precision of the BLDC motor allows industrial players to reduce their energy consumption and heat generation, resulting in a longer operating life [3]- [6].In high torque applications, the outer rotor BLDC (ORBLDC) motor is one of the best motor candidates due to its high torque density characteristic profile and low manufacturing cost [7]- [11].
To sustain BLDC motor market demand, engineers must design motors with optimal performance but within a short time frame and at an acceptable development cost.Therefore, upfront simulation is very important in motor design process, and the full curve motor performance estimation is one of it [12].The estimation curve is required by the motor designer to run the feasibility design process and compare it against Int J Pow Elec & Dri Syst ISSN: 2088-8694  Determination of performance characteristics using FEA-analytical for … (Saharudin Kamaroszaman1) 2011 the application requirement [13], [14].Since most applications require dynamic motor performance in torque and speed, it is essential to ensure all the operating points fall within the appropriate zone of the curve (continuous or intermittent zones).The electromagnetic characteristics of the motor can be changed by changing the wire size or the number of wires turns in the winding, or both.More torque and lower speed motor performance can be achieved with more turns of smaller wire, while less torque and higher speed motor performance can be achieved with fewer turns of larger wire.This custom configuration can optimize motor performance within product requirements and provide an opportunity for an optimum design solution.Figure 1 shows an example of full curve motor performance, and Table 1 shows the description value.There are several approaches to determine full curve motor performance.The most popular approach is using a template-based electrical machine design tool.The licensed software can be purchased from well-known software providers such as RMxprt from Ansys, MotorWizard from Solidworks and Simcenter SPEED from Siemens.All the template-based software uses analytical approaches and are capable of generating results in a short period of time.The outcome result can be used as a pre-assessment report for the application requirement and to establish an early level of confidence in electrical motor sizing and configuration [15]- [17].The generated model from template-based software can be assessed using a finite element analysis (FEA) solver in 2D or 3D dimension for further analysis, but there are limitations when it comes to complex or unique design geometry conditions [18], [19].
Another popular approach is to use MATLAB/Simulink software to mathematically model the BLDC motor [20]- [22].The motor detailed parameter values on stator, rotor, winding, and mechanical must be configured correctly inside the BLDC block prior to coupling the block with the relevant ideal waveform commutation.The block diagram can be extended to be modelled with a more complex motor control system such as speed control or torque control.However, the issue arises when assumptions must be made when defining parameter values within the BLDC block parameter.In this case, this method can be used when the parameters of the BLDC motor are known, and multiple simulations are performed using different control system parameters.
Last but not least is using FEA analysis to produce full curve motor performance [23], [24].There are two approaches in this process; the first one uses a speed sweep method, and the second one uses a transient run method.Both methods use mechanical motion as a control variable, and the range setting is from a no-load condition to a stall condition.The speed sweep method is simple in a way by just connecting the individual speed points in a steady state condition, whereas the transient run method uses a huge amount of simulation resources but gives the advantage of plotting torque, power, and efficiency within a time function.Although FEA can produce accurate results, it is inefficient in terms of time consumption.
From the preceding discussion, full curve motor performance should be precise at early development because all operating points should fall within the appropriate zoning area.Overestimation prediction will result in increased costs, while underestimation prediction will result in product failure.Thus, an acceptable confidence level toward the prediction curve is required within the agreed project lead time.Currently, there are three popular methods, namely template-based software, mathematical BLDC model using MATLAB/Simulink, and FEA.Template-based software and mathematical BLDC models are relatively quick methods, but these methods are rigid to specific template design geometry models and are unable to deliver correct predictions for complex or custom design geometry.While the FEA method is relatively accurate for all design geometries.However, it takes a long time to complete all simulation configurations.As a result, filling this gap would requires a new prediction method that can provide significant prediction accuracy in any type of design geometries while ensuring deliverable task within an acceptable lead time.To bridge this gap, this paper will present a method to predict the full curve outer rotor BLDC motor performance using a combination of FEA and mathematical equations.The optimum design is gained from the FEA analytical study of the desired motor geometry, and the performance curve is generated from the mathematical equation.The critical torque constant KT in the equation will be derived from FEA results, and the outcome results will be verified against the motor requirements.

RESEARCH METHOD 2.1. Overall flow chart
Figure 2 illustrates the overall methodology flow chart that begins with the generation of design specification and ends with motor performance curve.Section 2.2 elaborates how the design specification was developed.Sections 2.3 and 2.4 explain the methodology for CAD and FEM modelling.The FEA approach was used for all electromagnetic analyses on the stator and rotor, and the best design was chosen based on the target values of Itrans < 6 A and Bstat < 1.8 T. The method allows the motor designer to analyze custom motor designs up to the optimum condition, where the common multiple objectives such as high output power, high efficiency and small volume size could be achieved prior generate the full curve motor performance [25], [26].Furthermore, the electromagnetic analysis produces a torque constant KT value, which serves as the primary constant for calculating full curve motor performance as described in section 2.6.

Design specification flowchart
Figure 3 shows the design specification flowchart, which begins with the selection of the target application and followed by data collection.The required torque and speed were then determined before the flux density and current limit values were confirmed.For good design practice, the flux density value must be kept below the saturation flux value of 1.8 T and the current value has to be kept below the allowable air convection cooling limit of 6 A [27].Based on this information, a design review was required to determine whether the specification value could be proceeded with or required modification on material limit selection.After establishing a feasible design specification as per Table 2, the values will be used as the top-level requirement figure that must be met throughout the design process.

CAD modelling flowchart
As shown in Figure 4, the selection of motor design intent is the initial step in the CAD modeling process, followed by component design.At this stage, the overall number of components could be determined, and the material utilized needs to be identified for an early cost estimation.The assembly process must be considered from the subassembly level all the way up to the final assembly level.During the CAD review, the interference and clearance surface condition will be validated, and any necessary adjustments will be made at component design or sub-assembly design.Upon check all the interaction surface conditions, the CAD model is ready with designated parameter as per Table 3 and variable component name as per Figure 5.At this stage the CAD model is good to be used in FEM modelling.The selection of the stator slot and pole number is fixed based on the finding that the 18 s/20 p configuration delivers the highest output torque with the lowest torque ripple when compared to the 12 s/10 p and 9 s/8 p configurations [28].

FEM modelling flowchart
Figure 6 shows the FEM modelling flowchart that begins with structure modelling process, where the rotating and non-rotating component are declared accordingly.Then the meshing process can be performed by zoning area and element calculation can be proceeded after that.The magnetic analysis only can run after declare material inside FEM software with appropriate configuration.Parameter change must be performed if change in permeance model is not sufficient.

Determine winding turn and winding resistance
Prior proceeding with the full curve performance calculation, there are two constants required, one is torquing constant KT that can be determined from electromagnetic analysis and another one is total winding resistance R, which can be deduced from winding turn estimation.To estimate the turn, an approximate available area for winding is required and this area can be calculated from cross sectional of Single Slot Area  Slot , as per defined in Figure 7.

Structure
Were,  Slot : single slot area (m 2 ),  Wind : copper wire cross sectional area (m 2 ),  : slot fill factor,  O : resistivity copper wire (Ω/m), and  Wind/Slot : resistance per slot (Ω).The total winding resistance  can be derived from two phases winding resistance  Phase in series circuit condition as per (5).

Calculation full curve performance
The proposed structure design in Figure 5 is simulated using various sizes of Wstat, Hmag and Lstack.The analytical simulation begins with CAD and FEA modelling of the stator portion.Under this condition, only the stator winding is energized using three types of stator winding turns, N and the relevant target stator current Istat in order to achieve 600 At and 1400 At of magnetomotive force, Fm.The outcome value of stator flux density Bstat is plotted against Istat and the decision will be made based on the curve gradient condition.Next, the selected Wstat size in the stator portion will be combined with the rotor portion, and the Hmag and Lstack values will be determined using an ideal DC electromagnetic analysis.The static torque, Tstatic, and Bstat will be plotted against the stator current Istat and the judgement will be made based on the lowest Istat at maximum Tstatic with the condition that Bstat does not exceed 1.8 T. After that, the correlation curve among Bstat, transient current Itrans, and electromagnetic transient torque Ttrans is generated using an ideal sinusoidal BEMF current profile.The final design must meet the requirements of Itrans less than 6 A and Bstat of less than 1.8 T. Table 4 shows all the variables used in this research paper.Upon the completion of analytical analysis, an appropriate graph torque versus current could be established and the gradient could be represented by the torque constant, KT in (6).
The motor performance curves for speed, ω, current, I, mechanical output power, P could be described in ( 7)-( 9).

Analytical analysis
In this section, the motor has been set to various current loading conditions.Section 3.1.1focus on the Wstat variable at the stator portion, and section 3.1.2focus on Hmag and Lstack variables at the rotor portion.Then, the optimum motor design can be determined and the torque constant KT can be derived from FEA analytical graph for full curve motor performance.

Electromagnetic analysis stator
Two Fm conditions are used in this section, 600 At and 1400 At as per Figure 8.The analysis was performed under no loading conditions and with no interaction between stator and rotor.The 15 mm Wstat is easily saturated when the current reaches 6 A, but for the 25 mm Wstat, it is hard to be saturated even when the current reaches 28 A. For 19 mm Wstat, the stator teeth start to be saturated at 7 A and show an acceptable flux density condition for current less than 7 A. By plotting Bstat versus Istat as per Figure 9, the electromagnetic characteristic curve gets less stiff when a large value of Wstat is used, but more current injection is required to generate sufficient stator flux density.This is due to a lack of winding turns around the teeth and a lack of space.A similar situation occurred with a small value of Wstat, where stator teeth could be easily saturated even with a small value of current.As a result, a 19 mm Wstat is the ideal size to be used due to sufficient electromagnetic energy with a current of less than 6 A. The data shows an early confidence level toward 19 mm Wstat and is good to be used in the next section analysis.

Electromagnetic analysis rotor
The selected 19 mm Wstat is used with variable parameters from Hmag, Lstack and Istat.In every graph, there is a common trend in Tstatic and Bstat.The Tstatic increases when the Lstack and Istat are increased.However, the increment in the Bstat value is contributed by Istat alone.This indicates that the flux density saturation at Bstat is independent of the stack length.When performing a comparison among all four graphs, the electromagnetic energy increases when the Hmag is increased.The 14 mm Hmag in Figure 10(d) can produce up to 125 Nm of torque at 7 A current when using a 50 mm Lstack.The vice versa situation occurs on Bstat in Figures 10(a)-(c), where the curve is shifted downward when Hmag is increased, and this gives further margin toward higher Istat loading conditions.After going through all the available configurations, the preferable solution is to use 14 mm Hmag with a 30 mm Lstack.The design could achieve target output torque (Tstatic > 35 Nm) at the lowest current (Istat = 3 A DC), and the flux density is below the saturation level at 1.8 T (Bstat = 1.2136T) as shown in Table 5.This configuration has been selected for further analysis in the transient condition.The preferable configuration is further analysis using the BEMF current profile that simulates the design in transient conditions.Figure 11 shows the analytical performance among Ttrans, Bstat, and Itrans.The model delivers the 35 Nm torque at 2.62 A (Itrans) current and 1.59 T (Bstat) flux density as shown in Table 6.The transient Itrans value is lower than the static Istat value because all three phases are energised according to waveform commutation, whereby in static analysis, only two phases are energised and the remaining one phase remains ideally zero.This finding allows for an early judgement to be made on the worst condition without spending too much resources on transient simulation.As a result, the selected model of 14 mm Hmag and 30 mm Lstack is accepted for further assessment in full curve performance analysis.

Full curve motor performance
In order to determine the full curve function in terms of torque value, the torque constant KT in (1) has to be calculated first from the gradient curve of torque versus current in Figure 11.In this case, the torque constant KT value is determined from the FEA transient analysis result based on a series of transient current loadings and using the custom design model, which is close to the actual motor condition.There are another two constants that are required prior to the determination of full curve performance, BEMF constant Ke and total resistance R. The BEMF constant Ke can be calculated directly from the torque constant KT value, and the total resistance R can be estimated from the winding space area in Figure 7 and then calculated using (5) to (9).Upon having all these three constants, the full curve motor performance of speed, current, and output power can be generated as per Figure 12.Based on the curve generation, the motor design is capable of delivering a maximum output power of 703 W with a free running speed of 113 RPM and a stall current of 22.6 A. In terms of rated condition, the design is capable of meeting the target requirement with a rated speed of 96 RPM, a rated current of 3.2 A and a rated output power of 357 W. Table 7 shows the summary finding, and the outcome result shows promising data for the application used.

CONCLUSION
There are several methods to simulate the overall motor performance, but the design limitations of template-based software can lead to inaccurate judgement and early design errors.Too much reliance on FEA analysis will cause a long lead time in motor design development.Hence, an appropriate balance approach between FEA and mathematical equations has to be considered.Therefore, this paper presents a generic method for using FEA for analytical analysis prior to the generation of full curve motor performance using a mathematical equation.The critical torque constant KT value in equation 1 is determined from the FEA result, and the total resistance R value is estimated from the winding space area in the actual motor design condition.These two constants are determined as precisely as possible prior to the generation function of full curve motor performance for speed, current, and output power in terms of torque value.All the rated conditions, including free running speed, maximum output power, and stall current, are verified as a whole package for design review prior to design fabrication and provide an opportunity for any necessary improvement at an early design stage with minimum cost impact.As a conclusion, this research paper shows a methodology that can be used by any motor designer to produce the optimum motor design prior to the first prototype model creation.The leverage benefit can be gained in robust design as well as low-cost development.

Figure 1 .
Figure 1.Example full curve performance of BLDC motor

Figure 8 .
Figure 8. Stator flux density Figure 9. Stator flux density characteristic Figure 10 shows the analytical performances of Tstatic, Bstat and Istat.The plotted graphs are separated based on the dedicated values of 8 mm Hmag, at Figure 10(a), 10 mm Hmag, at Figure 10(b), 12 mm Hmag at Figure 10(c), and 14 mm Hmag, at Figure 10(d).All the performance analyses are using DC injection.

Figure 11 .
Figure 11.Characteristic of selected model Figure 12.Full curve motor performance

Table 1 .
Example parameter values from the curve

Design Specification End Electromagnetic Analysis on Stator Electromagnetic Analysis on Rotor -Static Analysis - -Transient Analysis- Generate Motor Performance Curve Using Mathematical Equation
Determination of performance characteristics using FEA-analytical for … (Saharudin Kamaroszaman1) 2013

Table 2 .
Design specification

Table 5 .
Model selection

Table 6 .
Transient analysis result

Table 7 .
Summary motor performance