In recent years, the use of robots for cleaning dust, leaves, and other debris from the surface of solar panels has become increasingly important due to the growing adoption of photovoltaic systems. The efficiency of solar panels is highly dependent on the cleanliness of their surfaces, as accumulated dirt can significantly reduce energy output. Therefore, developing efficient and reliable cleaning robots is crucial for maintaining optimal performance in solar power installations. One key challenge in designing these robots is minimizing their weight to reduce energy consumption during operation, especially for mobile components like the support frame of the walking mechanism. Excessive weight can lead to higher power demands and increased risk of structural failure, such as fractures or deformations. In this article, I present a lightweight optimization design methodology for the support frame of a solar panel cleaning robot, combining topological optimization and response surface methodology to achieve significant mass reduction while ensuring structural integrity.
The support frame is a critical component of the walking mechanism in a tracked solar panel cleaning robot, as it bears the weight of the robot and transmits forces from the drive system. Traditional design approaches often rely on trial-and-error or conservative safety factors, resulting in heavier structures that consume more energy. To address this, I propose an integrated approach that first uses discrete topological optimization to remove inefficient material from the frame, followed by size optimization based on response surface models and multi-objective genetic algorithms. This method allows for a systematic reduction in mass while controlling stress levels, making it ideal for applications where weight and durability are paramount, such as in solar panel maintenance robots.
The support frame is typically made from aluminum alloy 6061, a material chosen for its favorable strength-to-weight ratio and corrosion resistance, which is essential for outdoor applications like solar panel cleaning. The material properties include a density of 2700 kg/m³, Young’s modulus of 69 GPa, Poisson’s ratio of 0.33, and an allowable stress of 245 MPa. The frame connects components such as the drive wheel base, driven wheel base, and motor, forming part of the tracked locomotion system. Since the support frame accounts for a significant portion of the robot’s total weight, optimizing its design can lead to substantial energy savings and improved performance. In my work, I focused on parameterizing the frame’s geometry to enable precise control over key dimensions during the optimization process.
To begin the lightweight optimization, I first conducted a finite element analysis (FEA) to understand the stress distribution and deformation under operational loads. The support frame primarily experiences two types of forces: the gravitational load from the robot’s weight and the axial force from the synchronous belt transmitted through the drive and driven wheel bases. The axial force, denoted as $F_Q$, is calculated using the following formulas derived from belt drive theory:
$$ F_Q = \frac{1000 P_d}{v} $$
where $P_d$ is the design power in kilowatts and $v$ is the belt speed in meters per second. The design power is given by:
$$ P_d = K_A P_m $$
Here, $P_m$ is the nominal power, and $K_A$ is the service factor, which accounts for operational conditions such as daily usage time. For a solar panel cleaning robot operating about 5 hours per day, $K_A$ is typically 1.2 based on standard tables. The belt speed is computed as:
$$ v = \frac{\pi d_1 n_1}{6 \times 10^4} $$
with $d_1$ being the pitch diameter of the small pulley in millimeters and $n_1$ its rotational speed in revolutions per minute. In my design, the nominal power is approximately 0.07 kW, and the belt speed is set to 0.15 m/s, resulting in an axial force $F_Q$ of 560 N. Additionally, the gravitational load $F$ on the support frame is determined by summing the weights of all attached components, which amounts to about 300 N for the entire robot, or 150 N per side for the walking mechanism.
I imported the 3D model of the support frame into Ansys Workbench for FEA, applying appropriate constraints and loads. The mesh was generated using free meshing with local refinements, resulting in 36,204 elements and 60,824 nodes. The stress analysis revealed a maximum equivalent stress of 11.794 MPa and a maximum deformation of 0.017627 mm, both within safe limits compared to the material’s allowable stress. However, the stress contour plot indicated regions of low stress, suggesting opportunities for material removal without compromising strength. This insight guided the subsequent topological optimization step.
Topological optimization is a powerful tool for identifying inefficient material distributions in a structure. I employed a discrete topology optimization approach using the Shape Optimization module in Ansys Workbench, which iteratively removes elements with low stress contributions to achieve a target mass reduction. The goal was to reduce the mass by 20%, a balance between weight savings and structural integrity. The optimization results highlighted specific areas, labeled as region C, where material could be removed due to low load-bearing efficiency, while region B was recommended for retention as it carried higher stresses. Based on this, I modified the support frame geometry by eliminating material from region C, resulting in a redesigned model. A subsequent FEA on the modified frame showed a maximum stress of 11.371 MPa and a mass reduction, confirming the effectiveness of the topological optimization for initial lightweighting.
Following the topological optimization, I proceeded with size optimization using response surface methodology (RSM) to fine-tune the frame’s dimensions. RSM is a statistical technique that constructs approximate models of system responses based on experimental data, enabling efficient optimization with multiple design variables. In this case, I selected seven key dimensions of the support frame as design variables, denoted as $D_1$ to $D_7$, representing parameters such as thicknesses and lengths. Their initial values and bounds are summarized in the table below:
| Design Variable | Initial Value (mm) | Lower Bound (mm) | Upper Bound (mm) |
|---|---|---|---|
| $D_1$ | 50 | 45 | 55 |
| $D_2$ | 12 | 10 | 14 |
| $D_3$ | 6 | 5.4 | 6.6 |
| $D_4$ | 50 | 45 | 55 |
| $D_5$ | 50 | 30 | 70 |
| $D_6$ | 12 | 9 | 15 |
| $D_7$ | 100 | 70 | 130 |
To build the response surface model, I used a Box-Behnken experimental design, which generated 57 sample points within the design space. For each point, I performed FEA simulations to compute two objective responses: the mass of the support frame (denoted as $m$) and the maximum equivalent stress ($\sigma$). The mass response was chosen to minimize weight, while stress was constrained to remain below the allowable limit of 245 MPa, ensuring reliability for solar panel cleaning tasks. The data from these simulations were then fitted to a second-order polynomial regression model, expressed as:
$$ y = \beta_0 + \sum_{i=1}^{n} \beta_i z_i + \sum_{i=1}^{n} \beta_{ii} z_i^2 + \sum_{i < j} \beta_{ij} z_i z_j + \epsilon $$
where $y$ represents the response (mass or stress), $z_i$ are the design variables, $\beta$ coefficients are determined by least squares regression, and $\epsilon$ is the error term. For the mass response, the fitted equation after variance analysis was:
$$ m = (100.43 + 3.878D_1 – 1.46197D_2 + 4.38318D_4 – 0.361662D_5 – 2.08853D_6 + 0.0554D_1D_5 – 0.0554D_3D_5 – 0.0481668D_4D_5 – 0.0554D_6D_7) \times 10^3 $$
This model showed high significance with an $R^2$ value close to 1, indicating a good fit to the simulation data. The stress response was similarly modeled but is not detailed here for brevity. The optimization problem was formulated as a multi-objective minimization: reduce mass while keeping stress under the allowable limit. Mathematically, this can be expressed as:
$$ \min f(D) = m(D) \quad \text{subject to} \quad \sigma(D) < [\sigma] $$
where $D = (D_1, D_2, D_3, D_4, D_5, D_6, D_7)$ is the vector of design variables.
To solve this multi-objective optimization, I employed a non-dominated sorting genetic algorithm (NSGA-II), a popular evolutionary algorithm that finds Pareto-optimal solutions. NSGA-II works by maintaining a population of candidate designs, evaluating their fitness based on the objectives, and using selection, crossover, and mutation operators to evolve better solutions over generations. The algorithm converged after 10 iterations with an initial population of 500, yielding a set of Pareto-optimal designs that represent trade-offs between mass and stress. From this set, I selected a solution that balanced both objectives effectively, leading to the optimized design variables shown in the table below:
| Design Variable/Response | Before Optimization | After Optimization |
|---|---|---|
| $D_1$ (mm) | 50 | 45.033 |
| $D_2$ (mm) | 12 | 13.947 |
| $D_3$ (mm) | 6 | 6.0375 |
| $D_4$ (mm) | 50 | 45.184 |
| $D_5$ (mm) | 50 | 54.879 |
| $D_6$ (mm) | 12 | 14.854 |
| $D_7$ (mm) | 100 | 125.4 |
| Mass $m$ (g) | 388.89 | 297.12 |
| Max Stress $\sigma$ (MPa) | 11.371 | 10.364 |
The results demonstrate a significant improvement: the mass of the support frame decreased by 24.6%, from 388.89 g to 297.12 g, while the maximum equivalent stress reduced by 1.007 MPa, from 11.371 MPa to 10.364 MPa. This stress level is well below the allowable limit of 245 MPa, ensuring that the optimized frame can safely withstand operational loads during solar panel cleaning. The reduction in weight directly translates to lower energy consumption for the robot’s movement, enhancing its efficiency and battery life—a critical factor for autonomous systems deployed in remote solar farms. Additionally, the lower stress indicates improved durability, reducing the risk of fatigue failure over time.
To validate the optimization, I performed a final FEA on the optimized support frame model under the same loading conditions. The stress distribution was more uniform compared to the initial design, with no localized high-stress concentrations. The deformation remained minimal, confirming that the structural stiffness was preserved. This outcome underscores the effectiveness of combining topological optimization and response surface methodology for lightweight design. The topological step provided a macroscopic guide for material removal, while the RSM-based size optimization refined the dimensions to achieve precise performance targets. Such an integrated approach is particularly valuable for complex components like those in solar panel cleaning robots, where weight savings must not compromise reliability in harsh environmental conditions.
In conclusion, this study presents a comprehensive lightweight optimization design methodology for the support frame of a solar panel cleaning robot. By leveraging discrete topological optimization and response surface models with multi-objective genetic algorithms, I achieved a substantial reduction in mass without sacrificing structural integrity. The optimized frame is 24.6% lighter and experiences lower stress, contributing to improved energy efficiency and longevity of the robot. This methodology can be extended to other components of solar panel maintenance systems or similar lightweight-critical applications, offering a systematic framework for achieving optimal design trade-offs. Future work could explore dynamic loads or incorporate fatigue analysis to further enhance the robustness of solar panel cleaning robots in real-world scenarios.
The success of this optimization highlights the importance of advanced engineering techniques in renewable energy infrastructure. As solar panels continue to proliferate globally, efficient maintenance solutions like cleaning robots will play a vital role in maximizing energy output. By reducing the weight of key components, we can develop more sustainable and cost-effective robots that operate with minimal energy input, aligning with the broader goals of green technology. I believe that methodologies like the one described here will become increasingly standard in the design of solar panel support structures and other renewable energy systems, driving innovation in the field.