Motorcycle Swingarm Redesign
Utilizing generative design to redesign the swingarm for an electric motorcycle, taking the mass, stiffness, and cost of manufacture into consideration.
An individual project completed during 4th year of University
Project Outline
Zero Motorcycles
The project was framed as a feasibility study for a company incorporating generative design into one or more of its products. Zero Motorcycles was the company chosen. They are a manufacturer of electric-powered motorcycles and are the only company to have a full range of them on the market.
Another distinguishing feature of Zero Motorcycles is the customizability of their products. This comes in two forms: when purchasing a motorcycle from them, the option for a premium or standard model, an additional power tank or rapid charger is given; and, for three of their models, the SR range, they also offer 'Cypher Upgrades', such as faster charging, extended range, on-dash navigation, parking mode, and heated grips. These aren't ordered and physically installed, these are instead bought digitally and through software, instantly applied to the bike. These further increase the customizability of their motorcycles and provide an important point-of-difference to competitors.
3D Mesh Model of Zero SR S
Generative Design
Generative design is particularly applicable to Zero Motorcycles. It allows for the relatively simple creation of multiple versions of a component, each providing a different compromise between mass, stiffness, and cost. Giving the user the choice between them further extends the motorcycle's customizability, the company's key selling point.
Beyond the customizability, generative design can provide baseline improvements to the product. It mostly affects a product's cost, mass, and stiffness. The overall stiffness of a motorcycle is crucial to its handling capability, and mass reduction improves acceleration, deceleration, range, and max speed too. In a growing, competitive market, cost is an important objective too. Another plus is the marketability of generative design. It provides a visible, tangible example of the product's innovative nature.
All of this demonstrates the value of generative design for a company like Zero Motorcycles.
Generative Design Process
Chosen Component
To begin the generative design process, a component to be redesigned must be selected.
Zero Motorcycles 'SR' Range Swingarm ('Power Pivot')
For this study, the chosen component is the swingarm. This component connects the rear wheel of a motorcycle to its body, allowing the rear wheel to pivot vertically while maintaining tension in the drive belt.
This swingarm is shared across all three bikes in Zero Motorcycles' 'SR' range. This means a generatively designed replacement would be widely usable.
3D Model of the Swingarm
The original swingarm was isolated from the mesh model of the full motorcycle. This allows for objective comparisons to be easily made between the initial design and the versions created through generative design.
Obstacle and Preserve Geometry
Preserve Geometries Highlighted
'Preserve geometry' is the term for areas of the component that are untouchable to the generative design algorithm. As customizability is essential, the fixing points for the component must be preserved so that parts are interchangeable.
It was later deemed that, to maintain functionality, the mudguard must also be marked as a preserved geometry.
Obstacle Geometries Highlighted
'Obstacle geometry' is the term for regions in space where the generative algorithm is unable to place material. Unlike preserve geometry, these aren't included in the final design.
One use of obstacle geometries was to prevent the generated swingarm from straying into space used by other components such as the rear wheel, brakes, battery, motor, radiators, and drive chain. Another geometry was added to prevent the swingarm from straying too close to the road surface.
Two more obstacle geometries were added to prevent the swingarm from straying horizontally into the area where the rider would naturally place their legs (given regular riding position).
Loading and Structural Constraints
To guide the generative design algorithm, the forces and moments on the structure need to be supplied. For a motorcycle, the loads on the swingarm vary. For this analysis, the loads during the most extreme usage situations were supplied.
Loads during Accelerating and Braking
The force during acceleration is split into three components: the moment due to the rotational acceleration of the rear wheel, the weight of the vehicle, and the linear compressive force that acts as the rear wheel 'pushes' the rest of the motorcycle body forward.
The moment can be easily derived from the motorcycle's maximum torque.
To calculate the weight of the vehicle, the sprung mass is used alongside the proportion of weight over the rear wheel during acceleration.
The 'pushing force' is calculated using the vehicle's maximum acceleration, and the sprung mass plus the weight of the front wheel.
The same process can be used to obtain the loads during deceleration, though the forces and moments act in opposite directions (the rear wheel 'pushing' the bike forward, becomes a 'pull' during braking). One difference is, as the braking force is applied through both wheels, only 40% of the force acts upon the swingarm. The weight distribution of the motorcycle is also drastically different during braking.
Loads during Cornering
The force through the swingarm during cornering depends on the angle of the motorcycle. The thickness of a motorcycle tyre is not negligible so the force is not split evenly between each side of the swingarm. This torsional imbalance is represented by a moment acting about the axis of the swingarm. The reaction and friction forces are determined by the motorcycle's sprung weight and its maximum turning acceleration.
Simplified Geometry
As the complexity of the initial preserve geometry created false stress concentrations and generally slowed down the generative algorithm, they were simplified to cubes that matched the general area. This sped up the design algorithm significantly. After a new component iteration has been selected it is trivial to adjust it to connect to the real design preserve geometries.
Objectives and Manufacturing Methods
As mentioned earlier, the measured objectives are mass, stiffness, and cost. A low mass is desirable and is easily evaluated using the generated component's volume and the material's density. High stiffness is desirable and is automatically calculated by the generative algorithm; when choosing a component iteration for production, FEA can be used to verify this. Low cost is also desirable, but due to the 'optional part', customizability-based business model being suggested, it is lot a large focus. Therefore, it is only estimated comparativly to the original design. The estimate stems from the manufacturing method, material type, and material volume needed.
The options available to the generative design algorithm differ greatly depending on the type of manufacturing method chosen. For this study, a range of manufacturing methods were selected for comparison: 3-axis milling, 5-axis milling, and Metal 3D printing. These all provide their own trade-offs between the range of physical geometries they allow, the cost per part, the range of materials available, and the volume of material needed.
Results
36 versions of the swingarm were created by the generative design algorithm. Below is a selection of the best-performing and most notable component iterations.
Evaluation
The results of this generative design process were mixed. The process performed worse than expected at reducing the component's mass; only one (Outcome No. 17) achieved a significant reduction, and this came at the cost of increased cost per part. However, many iterations (chiefly No. 8 and No. 12) showed far improved stiffness with equivalent mass. These two versions are also distinct from each other in the tradeaoff they find between mass and cost. This array of options with different compromises between the three objectives synergizes well with the customizability selling point. Subjectively, the appearence and novelty of generatively designed components also provide key marketing possibilities, particularly in a field so focused on innovation.
From these standpoints, the process has been a success. However, looking at it from the perspective of direct productaimprovement (specifically mass-reduction) the results aren't as promising as expected. Though, considering the rapid development of generative design algorithms and additive manufacturing methods, repeating the analysis in the near future is likely to yield better results.
The next steps in this study would be using a combination of finite element enalysis and physical testing to validate the assumptions made, particularly when implementing the load cases. A detailed run-down of these assumptions and their verification methods can be found in the full report.
CAD Mock-ups
Outcome No. 8 with simplified preserve bodies replaced by the original geometry
Outcome No. 8 on the model of the Zero SR/S
For access to the full design report, get in touch at Michaelsvanidze0@gmail.com.