Anti Roll Bar Design
Please enjoy the 2-minute summary gallery below, or scroll further to explore my work in greater detail
As is readily apparent on this website, the majority of my work pertains to vehicle dynamics modeling and simulation, as well as kinematic design. My skills are not limited to computer coding and dynamic systems, however! One of my other design projects for the 2020 Clemson Formula SAE car was the anti-roll system, and on this page I'd like to walk you through the major steps in the design and development process.
Rendering of the Tiger22 Front Suspension Assembly, including anti roll bars.
Project Scope
Early on in the full vehicle design exploration, the team had decided to pursue a suspension concept with decoupled pitch and roll modes. Typical decoupled suspensions employ a bi-directional spring and damper element, which can introduce cost and complexity to develop. Other teams have worked around this by implementing two roll units, one which actuates in either roll direction, but this is a heavier and larger solution. We decided to use an anti-roll bar (ARB) for the relative simplicity, packaging freedom, and cost to develop. As a bonus, shifting the ARB pickup points to the pitch rockers enabled us to minimize torsion loads across the actuation axes, minimizing weight and compliance.
Design table summarizing the primary considerations behind each concept
Example of a decoupled suspension, with bi-directional roll element, on the Mercedes-AMG Project One.
Source: Topspeed.com, 2017, www.topspeed.com/cars/mercedes/2020-mercedes-amg-project-one-ar167883.html.
Moving forward, I created a list of constraints and criteria to guide the project and design direction. The static roll stiffness targets were calculated from target roll gradient and roll center locations which had already been finalized by that point.
One of the most important functions of anti roll bars are to serve as adjustment tools to fine-tune the cornering balance of the vehicle. Because of this, the criteria to "Maximize adjustment range and resolution" was one of the most important considerations throughout the design process. Before moving forward, it was important to quantify what adequate adjust-ability looked like. Neglecting kinematics, I explored the effect of the lateral load transfer distribution on the steady state handling balance of our vehicle. These results were supplemented by steady state directional control testing data (using SAE J266 test procedure) characterizing multiple setups on an older vehicle. Based on driver feedback of these setups, a target understeer gradient of 0.45 deg/g was selected, with a threshold of 1.5 deg/g before driver complaints became significant.
Combining all of these results, a target lateral load transfer distribution range of 46-56% was selected, with a target adjustment resolution of 0.5% (max allowable 1%). This would provide sufficient range to explore and adjust vehicle balance, and enough resolution to fine tune setups to driver preferences.
Concept Exploration
My first goal was to set up a tool that would enable me to quickly iterate through motion ratios, bar sizes, lever arms, material properties, etc. with automatic results visualization. To do this, I made an excel spreadsheet with our vehicle parameters and basic anti roll bar properties. The basic analysis method is summarized below:
Here is an example of the spreadsheet, pictured below. Notice the key outputs, such as factor of safety, adjustment resolution, static stiffness, and adjustment range. The use of lookup tables made it easy to compare materials quickly and easily.
The following table summarizes the key materials considered. 4130 steel was the preference going into the project, as it aligned with our in-house manufacturing and fabrication capabilities, but I wanted to explore other possibilities as well.
This table summarizes the total of 8 different configuration combinations that I evaluated.
As I moved along, I kept a log of feasible design solutions for each configuration, in order to compare and evaluate them.
In order to better organize my work, I narrowed down to 4 key design concepts, featuring the possible combinations of Torsion Bar Orientation and Lever Arm Style. For each concept, I would find the optimal combination of material selection, bar sizing, and packaging configuration, leaving four final designs to make a final selection. Below are some snippets of my conceptual visualizations, in various states of completion/detail.
This table summarizes the total of 8 different configuration combinations that I evaluated.
At first, I was very hesitant to consider using a blade style ARB concept, due to the increases in complexity and manufacturability. Due to the part geometry, we would have to send the blades out to a machine shop, increasing cost even further. However, I was ultimately left with no choice, as I was struggling to achieve the target resolution I wanted with a traditional arm style.
This illustrates the challenge undertaken with using an ARB for this application. Typically, anti roll bars act as a supplement to the corner springs in generating roll resistance, and account for around 20-40% of the total vehicle roll stiffness. With this suspension concept, the ARB is now responsible for 100% of the roll stiffness. This increases both the strength and resolution requirements of the system. In a traditional ARB, both of these factors contribute to increasing the size and subsequent weight of the overall system significantly, hurting performance and packaging. When using a blade style ARB, however, there is a lot more freedom in the blade geometry to fine tune the range and resolution of adjustment. This is exemplified by the resolution comparison of two concepts below, where green indicates a setting is in the target LLTD range:
Traditional Style
Blade Style
With the blade style concept, however, the more complex geometry makes it much harder to predict stiffness and deflection rates without the use of FEA software. As an approximation, I replicated the finite-element approach in excel, calculating the cross section and stiffness properties at 100 points across the blade, summing the deflections to calculate overall stiffness.
Going off of this test case, it's apparent that this method is not perfectly accurate, but by all means close enough to get me in the ballpark, at which point I can fine-tune the geometry directly in FEA.
After sufficient iterations and evaluations, I was able to narrow down to a handful of optimized concepts. The final comparison is shown below:
The best solution turned out to be a hybrid solution, using a steel torsion bar with titanium blades. Using as much steel as possible reduces manufacturing cost and effort, using titanium only when necessary. The titanium itself makes for an ideal blade material due to it's ductility, high UTS, and excellent fatigue properties. The blade style configuration came out on top as it enabled me to hit my resolution target in a "short bar" configuration, while the traditional bars could only be made to work in "long bar" layout, which would be heavier and harder to package. Even though this solution will end up being more expensive to manufacture due to the blade geometry, the superior performance and significant weight savings justify its selection.
Physical Design
With the design concept finalized, next it was time to finalize some of the physical details of the system. One of the first big decisions was to select an adjustment concept. With a blade style ARB, it is especially crucial to create a away to easily and quickly make consistent and repeatable stiffness adjustments. I came up with two main concepts, which are pictured below:
The clocking plates concept offered a more compact solution with higher resolution. However, it has a lot more moving parts, and is more complex to adjust. It is easy to imagine someone working under the car getting flustered trying to align that center plate. The index plate concept, on the other hand, still offers adequate resolution while being easier to adjust and manufacture accurately. Ultimately I chose the index plate.
The next decision I had to make was on a mounting concept. Our 2020 vehicle uses a steel spaceframe chassis, so it was easiest to design a welded steel mount, with additional bolt-ons as necessary. I also wanted to use a lubricated brass bushing on the torsion bar to avoid binding during operation. The best way to incorporate this bushing was using an aluminum bearing block, bolted to the steel mount. I came up with, again, two concepts to compare:
The split clamp design, though not illustrated in the image, is actually split across the aluminum block, allowing for quick and easy assembly. It also introduces an element of adjust ability, as the bolts can be tightened or loosened to adjust the clamping force on the bushing. The pillow block style mount has a much tighter manufacturing tolerance, but saves weight from the reduced hardware. The steel triangular tabs pointing up come from a piece of laser cut and bent sheet metal, and are welded to the steel mount. This solution is ultimately harder to manufacture accurately and has less margin for error (something to always consider in FSAE), while offering minimal benefit, so I ultimately went with the split clamp design.
Another design issue introduced by the blade style design is the mounting of the ARB drop links. It is common to use rod ends on the drop links in order to adjust preload, but this creates a single shear joint when bolted to the end of the blade.
I ultimately came up with three options. Option one was to simply leave the blade end unsupported. This was not favorable, as the bending moments at the end of the blade were significant enough to raise fatigue issues. Option two was to machine an aluminum spacer to support the bending moment on the blade, leaving the joint to act in tension. This was deemed an improvement, but not reliable enough due to the risk of spacer separation creating a sudden bending load. The ultimate decision was to go with option three, which consisted of a feature machined onto the blade. Even though this would increase the manufacturing complexity, it provided the most reliable bending support for the bolted joint, minimizing shear and bending loads into the blade.
This introduced a concern about stress concentrations. A comparison carried out in FEA showed a migration of the peak stress areas, and resulted in a 5.4% increase in peak stress on the blade face. This was still well within the fatigue limits of the blade, and deemed acceptable
In addition, extensive clamp load analysis was carried out in order to ensure the integrity of the bolted joint.
In order to ensure as much manufacturability as possible, the blade geometry was arranged in a way such that the main body only required two operations to create: A turning operation to create the circular profile, followed by a milling operation to cut the side profiles.
Step 1: Starting from rod stock, the following profile can be machined on a lathe
Step 2: The side profile can be milled out without extra fixturing operations. From there, the final detail features such as the threads are straightforward.
Structural Analysis
After design concept were finalized, it was time to turn to FEA to fine tune the blade geometry and ensure the reliability of all components in the assembly. Every component was designed to withstand a fatigue life of 10^6 or more cycles. Included below are a few images describing the analysis conditions used.
Comparing FEA results to stress and deflection predictions, the correlation was... not great. Initial comparisons using individual blade designs correlated within 25%, but as blade design complexity increased, the accuracy of my spreadsheet predictions decreased. This does not come as a surprise, and the intention was never to create a final design using only excel. For the full assemblies, the complexity involved with the various joints and connections made correlation even worse.
Ultimately this was not a concern. The excel spreadsheets served their purpose well in helping me narrow down on a final design solution concept, and I could generate the final design iterations in CAD with FEA analysis to take me home.
Final Design
With some final tweaks and adjustments done manually in CAD, I was able to generate a blade geometry that, when measured in the full assembly, generated the desired target range and resolution of stiffness adjustments. As shown in the table below, I was very nearly able to achieve my target LLTD adjustment resolution of 0.5%, while maintaining the full desired adjustment range.
The final suspension balance and adjustment characteristics are summarized below:
Overall, the final design with the blades is a bit more complex to manufacture (and expensive) than I would have liked. However, the end result is an assembly that maintains superior adjustment characteristics even despite the challenge of generating 100% of the total vehicle roll stiffness. Every other component is easy to manufacture and contributes to a simple, compact and lightweight package that is easy to adjust, Please enjoy the images of the final design below:
