CFD Analysis on the Performance of a Coaxial Rotor
with Lift Offset at High Advance Ratios

A CFD flow solver specifically for rotorcraft, rFlow3D developed at JAXA [22, 23, 24, 25] is used to simulate the aerodynamic performance of the coaxial rotor in forward flight. rFlow3D is not just a CFD solver, but also covers trim analysis and elastic deformation analysis of blades. rFlow3D has been validated to give reliable solutions for many flows associated with rotorcraft.

The basic equations and numerical methods used in rFlow3D have been presented in detail in previous papers (e.g. [10, 11]). Only a brief description of the flow solver is addressed in this paper. rFlow3D employs a finite volume scheme for structured grids that can handle moving overlapping grids. The basic equations are the compressible Naiver-Stokes equations. Several turbulence models can be incorporated depending on the characteristics of the flow field. The coaxial rotor analyzed in this study does not have a Reynolds number large enough to assume that the flow is fully turbulent. Therefore the transition model ($\gamma-Re_{\theta}$ model) [26] is employed to account for the laminar/turbulent transition in the boundary layer. A class of the Menter’s shear stress transport (SST) turbulence model (so-called $k-\omega$ SST or SST-2003) [27] is applied only to the region downstream of the transition point obtained by the transition model.

The trim analysis technique in conjunction with CFD analysis developed by Sugawara et al. [28] is applied to the coaxial rotor system used in this study. This technique is an extension of the trim analysis for single rotor systems [10, 24].

The following six components are selected for trim analysis: the total rotor thrust of the coaxial rotor system $\sum C_{T}=C_{T}^{U}+C_{T}^{L}$, the rolling and pitching moments of the upper and lower rotor $C_{Mx}^{U},C_{Mx}^{L},C_{My}^{U},C_{My}^{L}$, and the yawing moment (torque) balance $\sum C_{Mz}=C_{Mz}^{U}+C_{Mz}^{L}$. The moment coefficients $C_{My}$ and $C_{Mz}$ are defined as

where $M_{y}$ and $M_{z}$ are rotor pitching and yawing moments, respectively. These parameters are adjusted by the six parameters for determining blade pitch angles in Eq. (5), which are $\theta_{0}^{U}$, $\theta_{0}^{L}$, $\theta_{\rm 1c}^{U}$, $\theta_{\rm 1c}^{L}$, $\theta_{\rm 1s}^{U}$ and $\theta_{\rm 1s}^{L}$. The sensitivity matrix of control elements $\theta_{j}$ to target variables $a_{i}$, $\left\{\partial a_{i}/\partial\theta_{j}\right\}(i,j=1-6)$, are numerically calculated using the blade element theory (BET) for a single blade [29]. Note that the sensitivity matrix calculation does not take into account the aerodynamic interference between the upper and lower rotors. Although this interference might affect the target aerodynamic parameters, Sugawara et al. have confirmed that a smoothly converging solution is obtained without taking it into account [28].

Each pitch angle parameter is updated by finding the difference between each aerodynamic coefficient obtained from CFD and its target value and multiplying the difference by the inverse of the sensitivity matrix. The update is performed at every one revolution of the rotor.

The flow around the coaxial rotor without the rotor hub is analyzed using the computational grid shown in Fig. 3. The total number of computational grid points is about 20.9 million, consisting of $125\times 123\times 65$ points on the blade grid, $279\times 221\times 139$ points on the inner background grid, and $211\times 211\times 169$ points on the outer background grid. The blade grid is a moving body-fitted grid, which means that the grid arrangement changes as the blade moves. The computational domain of the outer background grid is set to $\pm 100R$ for each axis. The minimum grid spacing on the blade surface is set to be less than $3.0y^{+}$, where $y^{+}$ is the dimensionless length expressed using the friction velocity at the trailing edge of the tip of the rotor blade at the advance ratio $\mu=0.2$. This minimum grid spacing is a reasonable value that adequately predicts drag using the SST turbulence model [30]. Note that the boundary layer on the blades is not fully turbulent in this study, because the Reynolds number at 75%$R$ is in the range between $1.2\times 10^{5}$ and $6.6\times 10^{5}$ at $\mu=0.52$. For all calculations, the pressure and temperature of the freestream $p_{\infty}$ and $T_{\infty}$ are set to 101.3 kPa and 15^∘C, respectively. For the calculation, one node of the Intel Xeon Gold 6240 processor (18 cores/36 threads, clock frequency 2.60 GHz) is used for parallel computation with 36 threads using OpenMP. The computation time per rotor revolution is 18 hours, and the quasi-steady solution is obtained after about 15 rotations. .

Prior to calculating with a high advance ratio of 0.5 or higher, the calculated aerodynamic performance is verified with wind tunnel test data obtained by Cameron and Sirohi [13] and associated numerical data using two rotorcraft comprehensive analysis tools, the Comprehensive Analytical Model of Rotorcraft Aerodynamics and Dynamics (CAMRAD II) [31] presented by Feil et al. [15] and the Rotorcraft Comprehensive Analysis System (RCAS) [32] presented by Ho and Yeo [17]. Cameron and Sirohi [13] provided extensive data on the aerodynamics performances of the single and coaxial rotors for a wide range of advance ratio ($0.21\leq\mu\leq 0.52$) and lift offset ($LO=0$–0.3). Very recently, Sugawara et al. [28] have validated the numerical results with the CFD flow solver, rFlow3D, for coaxial rotor configurations with various lift offsets at an advance ratio of 0.53, comparing them with experiments and comprehensive analyses. The comparison reveals satisfactory agreement between their results and the other data. Therefore, the aerodynamic performance of the coaxial rotor for various advance ratios with zero lift offset is solely presented in this section.

Table 1 summarizes the conditions for this verification calculation. The rotor speed is 900 RPM. The trim conditions are set to satisfy the target total thrust $\sum C_{T}(=C_{T}^{U}+C_{T}^{L})$ and to maintain the balance of rolling moment, pitching moment, and torque. The lift offset is assumed to be zero. The target thrust is the experimental value for various advance ratios. Note that the thrust is not the target value at the rotor trim in the experiment. In the experiment, the collective pitch angle of the upper rotor was first set to a constant value ($\theta_{0}=8^{\circ}$), and the collective pitch angle of the lower rotor was adjusted to balance the torque. Next, the cyclic pitch angles of the upper and lower rotors were adjusted to match the target value of the lift offset.

Figure 4(a) shows the comparison of various parameters obtained using the rFlow3D with the wind tunnel experiments and the comprehensive analyses. In this comparison, the lift-to-effective-drag ratio (effective lift-drag ratio, $L/D_{e}$) is used as a measure of the aerodynamic performance of the rotor. The effective lift-drag ratio is described as

where the rotor power coefficient as $C_{P}$ and the rotor drag coefficient as $C_{D}$. $\sum$ means the total values such as $\sum C_{P}=C_{P}^{U}+C_{P}^{L}$. $\sum C_{P}$ and $\sum C_{D}$ are defined as

where $P$ represents the rotor power (equivalent to the product of rotor torque $Q$ and rotor angular velocity $\Omega$), and $D$ represents the rotor drag in the forward flight direction. The effective lift-drag ratios obtained by this calculation are in good agreement with the experiment and CAMRAD II calculated values.

Figure 4(b) shows the trimmed pitch angles from this calculation. In the present calculation, the trimmed collective pitch angle for both the upper and lower rotors is approximately 7.5^∘, which is consistent with the results of other analyses in which the trimmed collective pitch angle of the lower rotor was about $7^{\circ}$ with the predetermined collective pitch angle of 8^∘ for the upper rotor. The relation between the longitudinal cyclic pitch angle $\theta_{\rm 1s}$ and the advance ratio $\mu$ also agrees well with Cameron and Sirohi [13] , Feil et al. [15] and Ho and Yeo [17].

Taking into account the difference between the calculated results and the experimental values from the other comprehensive analysis (RCAS) and the error bars in the experiment itself, rFlow3D can be used to properly evaluate the aerodynamic performance of the rotor configuration presented in the following sections.

To demonstrate the validity of adopting the transition model, the laminar and turbulent distribution of each rotor at advance ratio $\mu=0.53$ is presented in Fig. 5. The transition from laminar to turbulent flow occurs at the blade azimuth angle $\Psi\simeq 315^{\circ}$, and returns to the laminar at $\Psi\simeq 45^{\circ}$. Therefore, it is reasonable in this calculation to use a computational model that takes into account laminar/turbulent transitions.

Figure 6 shows the pitch angles ($\theta_{0}$, $\theta_{\rm 1c}$, and $\theta_{\rm 1s}$) obtained from the trim analysis for each condition. The results for the upper and lower rotors are represented by solid and broken lines, respectively. Note that the trim analysis does not fully converge for the lift offset $LO=0$ with the advance ratio $\mu$ higher than 0.4. As the lift offset $LO$ increases, the upper limit of the advance ratio at which the rotor trim converges increases. The lift offset $LO$ must be increased to achieve high-speed flight of a coaxial rotor.

The results show that the collective pitch angle $\theta_{0}$ increases with an increase in the advance ratio $\mu$ for a constant lift offset. The longitudinal cyclic pitch angle $\theta_{\rm 1s}$ is negative and its absolute value also increases with an increase of advance ratio at the constant $LO$. The increase in the absolute values of the collective pitch angle $\theta_{0}$ and the longitudinal cyclic pitch angle $\theta_{\rm 1s}$ is suppressed with an increase in the lift offset $LO$. On the other hand, the lateral cyclic pitch angle $\theta_{\rm 1c}$ is hardly affected by the advance ratio $\mu$ and the lift offset $LO$. For all conditions, the pitch angles of the upper and lower rotors take almost the same values. The dependence of the pitch angle on the advance ratio and lift offset presented in this study is qualitatively similar to the comprehensive analysis by Ho and Yeo [18].

From Eq. (5), the negative longitudinal cyclic pitch angle suppresses the angle of attack on the advancing side of the blade ($\Psi=90^{\circ}$) and increases the angle of attack on the retreating side of the blade ($\Psi=270^{\circ}$). This control is done to achieve rolling moment balance at a defined $LO$ by reducing lift on the advancing side of the blade and increasing lift on the retreating side of the blade. When a lift offset is applied to a coaxial rotor, the constraints on the rolling moment balance are relaxed, i.e., the lift allowed on the advancing side of the blade is increased. As the lift offset is increased, the blade pitch angle on the advancing side allowed is also increased, so the negative $\theta_{\rm 1s}$ can be set smaller. In particular, it is noteworthy that $\theta_{\rm 1s}$ is almost zero for $\mu=0.4$ and $LO=0.3$.

Under conditions with a high advance ratio ($\mu\geq 0.5$), even with a large lift offset ($LO=0.3$), the negative value of $\theta_{\rm 1s}$ is larger and the collective pitch angle $\theta_{0}$ is also larger. As the advance ratio increases, the lift on the advancing side of the blade tends to become higher and higher, and it becomes increasingly difficult to ensure lift on the part of the retreating side where the reverse flow occurs. Therefore, to reduce the lift difference between the advancing and retreating sides ($\Psi=90^{\circ}$ and $270^{\circ}$) for maintaining rolling moment balance, the lift at the advancing side is suppressed. The collective pitch angle $\theta_{0}$ is increased to obtain the remainder of the required lift by increasing the lift generated in the forward and rear parts ($\Psi=180^{\circ}$ and $360^{\circ}$) of the rotor.

The performance of the coaxial rotor for various advance ratios and lift offsets is summarized in Fig. 7. Figure 7 shows the effective lift-drag ratio $L/D_{e}$ for each condition to clarify the effect of the lift offset on the coaxial rotor.

The effective lift-drag ratio $L/D_{e}$ is the same as the definition Eq. (8) using the sum of the values of the upper and lower rotors. Note that the missing parts of the plot correspond to conditions where the trim analysis does not fully converge.

As shown in Figure 7, the effective lift-drag ratio $L/D_{e}$ has a quadratic form with respect to the advance ratio $\mu$ and reaches a maximum of around $\mu=0.4$. The value of $\mu$ that maximizes $L/D_{e}$ increases slightly with increasing lift offset. The maximum effective lift-drag ratio $L/D_{e}$ increases as the lift-offset $LO$ increases, improving the rotor’s aerodynamic performance. However, for $\mu\geq 0.6$, the lift-drag ratio $L/D_{e}$ decreases significantly with increasing $\mu$, even if the lift offset $LO$ is set to be 0.3.

The values of the total rotor power coefficient $\sum C_{P}$ and rotor drag coefficient $\sum C_{D}$ for each advance ratio and lift offset are summarized in Figs. 7 and 7, respectively. Both the rotor power coefficient $\sum C_{P}$ and the rotor drag coefficient $\sum C_{D}$ vary with the advance ratio $\mu$ and the lift offset $LO$. Since the thrust coefficient $\sum C_{T}$ is constant for all conditions in this calculation, the variable that causes the effective lift-drag ratio $L/D_{e}$ to change at each advance ratio is either the rotor power coefficient $\sum C_{P}$, the rotor drag coefficient $\sum C_{D}$, or both.

The rotor power coefficient $\sum C_{P}$ shown in Fig. 7 is lower in the condition with a large lift offset than in the condition with a small lift offset. The dependence of $\sum C_{P}$ on $\mu$ is similar to the relationship between the advance ratio $\mu$ and the collective pitch angle $\theta_{0}$ shown in Fig. 6 except for $\sum C_{P}$ at $LO=0.3$ with $\mu=0.6$ and 0.7. It is reasonable to assume that the rotor power decreases as the collective pitch angle decreases because the drag force on the local cross-sectional profile of the entire rotor surface decreases. The results suggest that a coaxial rotor can achieve low rotor power flight regardless of the advance ratio $\mu$ by satisfying the lift-offset $LO$ such that the collective pitch angle $\theta_{0}$ becomes small. Note that even with the large lift offset $LO=0.3$, $\sum C_{P}$ is significantly large at the very high advance ratio $\mu=0.8$. This point will be discussed with the distribution of sectional drag on the rotor, along with the reason why the shaft power is small for $\mu$ in the range of 0.5 to 0.7 at $LO=0.3$.

The rotor drag coefficient $\sum C_{D}$ significantly increases as the advance ratio $\mu$ increases for the high advance ratio $\mu\geq 0.6$, From the definition of the effective lift-drag ratio [Eq. (8)], the rotor drag coefficient $\sum C_{D}$ affects the lift-drag ratio in the same order as the rotor power coefficient $\sum C_{P}$ when the advance ratio $\mu$ is higher than 0.5.

Ho and Yeo [18] showed that setting $LO=0.3$ improved $L/D_{e}$ even under conditions $V>200$ knots ($\mu>0.6$). This was partly due to the significant reduction in shaft power as we also show in Fig. 7. Another factor was that, unlike our calculations, the increase in drag at $\mu\geq 0.6$ was suppressed. The cause of this discrepancy will be discussed using the distribution of sectional drag on the rotor.

For a more detailed study of the effect of the lift offset on the rotor power and the drag, we analyze the vertical force coefficient $c_{z}M^{2}$ and the sectional drag coefficient $c_{q}M^{2}$ for blade element defined in Fig. 8. They are expressed as

where the vertical force is denoted by $f_{z}$, the force in the direction of blade movement by $f_{q}$, speed of sound in freestream by $a_{\infty}$ and blade chord length by $c$.

Figures 9 and 10 show the distributions of the vertical force (lift) coefficient $c_{z}M^{2}$ and the drag coefficient in the direction of blade movement $c_{q}M^{2}$ of the upper rotor for various advance ratios and lift offsets, respectively. It is clear from Fig. 9 that a significant difference between advance ratio $\mu=0.5$ and 0.8 is found in the distribution of the vertical force coefficient $c_{z}M^{2}$ at $LO=0.3$. In $\mu=0.8$, the downforce is observed in large regions at the retreating side of the blade azimuth ($180^{\circ}<\Psi<360^{\circ}$). The region where downforce occurs coincides with the region of reverse flow, whose diameter can be estimated by $\mu R$, the product of the advance ratio $\mu$ and the rotor radius $R$. The maximum thrust at $\mu\geq 0.7$ is obtained at the blade azimuth angle $\Psi$ around 22.5^∘ and 160^∘ due to increasing collective pitch angle $\theta_{0}$, as discussed in section 4.1.

Figure 10 shows that the difference in the sectional drag coefficient $c_{q}M^{2}$ distribution for different lift offsets can be seen in this reverse flow region. As mentioned in section 4.1, the blade pitch angle $\theta$ on the retreating side ($180^{\circ}<\Psi<360^{\circ}$) decreases with increasing lift offset. The projected area of the blade relative to the freestream decreases with decreasing blade pitch angle. The decrease in the projected area of the blade in freestream owing to the increase in the lift offset reduces the local blade drag, resulting in a decrease in the sectional drag coefficient $c_{q}M^{2}$. Therefore, for advance ratio $\mu>0.52$, the rotor drag coefficient $C_{D}$ is reduced for larger lift offsets $LO$.

At high advance ratios, the negative (blue) value region increases significantly on the retreating side ($180^{\circ}<\Psi<360^{\circ}$). In this region, the blade is pushed in the direction of freestream, which has two effects on aerodynamic performance: a decrease in shaft power and an increase in drag in the uniform flow direction. In addition, a very large sectional drag is generated rear of the rotor ($\Psi\simeq 360^{\circ}$), which leads to a significant increase in shaft power. In forward flight, the rotor disk induces a large downwash at the rear of the rotor [29]. This downwash increases the induced drag at the rear of the rotor and increases the torque.

The trimmed pitch angles shown in Fig. 6 indicate that the longitudinal cyclic pitch angle $\theta_{\rm 1s}$ on the retreating side must increase with the advance ratio due to rolling moment balance limitations. The marked non-uniformity seen in the $c_{q}M^{2}$ distribution at the very high advance ratio is due to the nature of this longitudinal cyclic pitch angle.

We use a simple untwisted VR-12 blade in this calculation. The suppression of increase in drag at $\mu\geq 0.6$ reported by Ho and Yeo [18] is probably owing to several differences in rotor configuration and operating conditions. Ho and Yeo employed a coaxial rotor configuration with four twisted blades per rotor for their analysis. Their rotor radius was 8.862 m, much larger than ours of 1.016 m. The rotor speed was reduced with increasing airspeed so that the tip Mach number on the advancing side would not exceed a value of 0.85.

For the improvement of the lift-drag ratio at very high advance ratios, it is necessary to optimize the blade shape to reduce not only the collective pitch angle, which contributes to rotor power but also the cyclic pitch angle on the retreating side, which dominates drag in the reverse flow region.

Feil et al. [16] analyzed the vibratory loads in a coaxial rotor configuration corresponding to Cameron and Sirohi’s experiment as well as our calculations. They showed the lift offset significantly reduced the 2/rev component of thrust variation in the range of the advance ratio of 0.52 or less. Here the mechanism of this suppression of thrust fluctuations is analyzed. Then the discussion is made whether the suppression is also effective under conditions with a higher advance ratio ($\mu\geq 0.6$).

The amplitudes of the 2/rev components for various advance ratios and values of lift offset are compiled in Fig. 15. Comparing the amplitudes with different lift offsets $LO$ for a given advance ratio $\mu$ shows that the amplitudes decrease significantly as the lift offset is increased. In particular, for $LO=0.3$, the 2/rev component is less than 0.1 in the range $\mu<0.6$. For $\mu\geq 0.6$, however, the amplitude increases rapidly with increasing advance ratio, even with a large lift offset ($LO=0.3$).

In this calculation, one rotor consists of two blades. The thrust at any given azimuth angle $\Psi$ is the sum of the lift forces of the blade located at $\Psi$ and the blade located at $\Psi+180^{\circ}$. That is, as can be seen in Fig. 11, the thrust fluctuation at azimuth angles $\Psi$ of around $0^{\circ}$ and $180^{\circ}$ (or around $90^{\circ}$ and $270^{\circ}$) is almost the same value, and its magnitude is the sum of the lift forces at front and rear of the rotor (or on the advancing and retreating sides of the rotor).

The large amplitude of the 2/rev component is due to the large lift generated near the front and the rear of the rotor in the forward direction at the high advance ratio with the small lift offset (see Fig. 9). This results in a large thrust force at azimuth angles $\Psi=0^{\circ}$ and $180^{\circ}$. The thrust is small at azimuths $\Psi=90^{\circ}$ and 270^∘ because the lift is almost vanished on the advancing and retreating sides of the rotor.

Figure 15 shows the relationship between the longitudinal cyclic pitch angle $\theta_{\rm 1s}$ and the amplitude of the 2/rev component for all 15 results obtained in this calculation. The amplitude of the 2/rev component can be correlated by the longitudinal cyclic pitch angle $\theta_{\rm 1s}$. Among the rotor controlling angles ($\theta_{0}$, $\theta_{\rm 1c}$, and $\theta_{\rm 1s}$), the longitudinal cyclic pitch angle $\theta_{\rm 1s}$ is the factor varies the magnitude of the pitch angle (i.e. lift) on the rotor’s advancing and retreating sides. Reducing the longitudinal cyclic pitch angle to zero, i.e., increasing the lift of the advancing side blade, is effective not only in improving the lift-drag ratio but also in reducing the 2/rev component fluctuation.

Note that the 2/rev component of vibratory aerodynamics loards may be reduced by adjusting the first blade crossing angle of two rotors [16, 17]. This point will be discussed in the future.

The 4/rev component is analyzed from the present calculations given various lift offsets at advance ratios higher than 0.5. The major source of this component is due to blade crossing events. Passe et al. [20] showed impulsive air loads corresponding to blade passage in an 8-bladed coaxial rotor (4 blades per rotor) in their calculations with a single lift offset given for each advance ratio in the range $\mu\leq 0.41$, according to the X2 Technology Demonstrator (X2TD) experiment.

Figure 16 shows the amplitude of the 4/rev component of the thrust fluctuation. The 4/rev component of the total thrust shown in Fig. 16 indicates that the lift offset reduces the amplitude at the advance ratio $\mu=0.52$. As the advance ratio increases, the amplitude increases even with a large lift offset. Notice that the amplitude for $\mu=0.8$ with $LO=0.3$ in total thrust fluctuation is larger than for $\mu=0.7$ with $LO=0.2$, where the amplitudes of 2/rev component are the same with each other (see Fig. 15). This implies that the 4/rev component is not just a harmonic of the 2/rev component. Figure 16 shows that the amplitudes of the thrust fluctuations for the upper and lower rotors are nearly equal for $\mu=0.52$, whereas the amplitude of the lower rotor is larger at $\mu>0.6$. The cause of this difference is discussed later using Fig. 18.

Returning to the azimuth dependence of the total thrust (Figs. 11 and 12), the thrust drops sharply at azimuth angles $\Psi=0^{\circ},90^{\circ},180^{\circ}$, and 270^∘. These azimuth angles are when the upper and lower rotor blades pass through each other. When the blades pass through each other, the pressure field on the pressure surface of the upper rotor blade interferes with the pressure field on the suction surface of the lower rotor blade. As a result of the interference, the difference between the positive and negative pressure on each blade becomes so small that the lift force decreases. This reduction in lift causes fluctuations in thrust due to the blades crossing each other.

Figure 17 shows the change in contour maps of the pressure $p$ as the blades pass through each other for the advance ratio $\mu=0.52$. These pressure distributions are for cross sections through $r=0.75R$ perpendicular to the $r$-axis for each phase angle $\Psi^{U}$. Figure 17(a) corresponds to the case $LO=0.1$ and Fig. 17(b) to the case $LO=0.3$. The effect of lift offset on the mutual interference of pressure fields when blades pass through each other is discussed using this figure. The crossing events are labeled as from 1st to 4th crossing in the order of azimuth angle of the upper rotor ($\Psi=0^{\circ}$, 90^∘, 180^∘ and 270^∘). The contours are displayed using the pressure coefficient,

where $p_{\infty}$ denotes the static pressure in freestream.

Focusing on the 1st crossing, the high-pressure region of the upper blade for $LO=0.1$ (Fig. 17(a)) shrinks as the low-pressure region of the lower blade approaches. This high-pressure region recovers after the blades pass through each other. In contrast, the pressure coefficient for $LO=0.3$ (Fig. 17(b)) has a small variation range for both the upper and lower blades. The small pressure variation around the blades at $LO=0.3$ is due to the small rotor thrust required at this crossing angle ($\Psi=0^{\circ}$) (see Fig. 9). The above trend is also observed for the third crossing of the upper and lower rotor blades at azimuth angle $\Psi=180^{\circ}$, where the pressure field for $LO=0.1$ is slightly different from the 1st crossing.

In the case of 2nd crossing, the upper blade is located on the advancing side of the rotor, while the lower blade is located on the retreating side of the rotor. At $LO=0.1$ shown in Fig. 17(a), the pitch angle $\theta$ of the lower rotor blade is so large that the low-pressure region above its suction surface enhances the low pressure below the upper rotor blade after the crossing. In contrast, for $LO=0.3$ shown in Fig. 17(b), the pitch angle $\theta$ of the lower blade is small. Both the upper and lower rotor blades do not show any change in the pressure field during the blade crossing event. The pressure interference weakens as the lift offset $LO$ increases.

At the 4th crossing, the upper rotor blade is located on the retreating side of the rotor, while the lower rotor blade is located on the advancing side of the rotor. Comparing Figs. (a) and (b) in Fig. 17, the low-pressure area above the lower rotor blade is larger for $LO=0.3$ than for $LO=0.1$, which also has a larger effect on the upper rotor blade pressure field. This is the cause of the steep thrust drop seen at azimuth angle $\Psi=270^{\circ}$ shown in Fig 11.

In summary, lift offset suppresses thrust fluctuations when the rotor blades cross in the front and rear locations. It also suppresses the fluctuations when the upper blades are on the advancing side. In contrast, lift offset slightly amplifies thrust fluctuations when the upper blade is on the retreating side. In general, as the pitch angle increases, the pressure distribution around the blade becomes more pronounced. Therefore, keeping both the collective and cyclic pitch angles small is effective in weakening pressure interference during blade crossing.

Finally, we discuss the 4/rev component at $\mu=0.8$. Figure 18 shows the difference in the vertical force coefficient $C_{n}M^{2}$ between the upper and lower rotors at the advance ratios $\mu=0.52$ and 0.8 with the lift offset $LO=0.3$. Note that it represents the difference between the upper and lower rotors at the same azimuth angle. The red color means the $C_{n}M^{2}$ of the upper rotor is larger than that of the lower rotor, and the blue color means the opposite. The sharp blue line at $\Psi=270^{\circ}$ corresponds to the thrust drop of the upper rotor blade at the 4th blade crossing as discussed in Fig. 16. Compared to $\mu=0.52$, for $\mu=0.8$, $C_{n}M^{2}$ of the lower rotor is substantially larger (blue-colored regions in Fig. 18) at $\Psi=90^{\circ}$ and $180^{\circ}$ than the value of the upper rotor. Complex distribution of mixed red and blue is seen around $\Psi=0^{\circ}$. Since this is the rear of the rotor disk, the complex distribution may be attributed to the flow disturbance generated by the front blades. These features lead to an increase in the 4/rev component at $\mu=0.8$ as well as the difference in amplitudes for the upper and lower rotors.