Engineered Suspension Systems
We have looked at a lot of webpages designed to get your attention, and convince you to buy suspension parts/pieces/goodies. What we noticed is that very few of these pages describe how the product actually works, what real life results can be expected, or even what the initial design goal was.
I can't speak for everyone, but when I am looking to buy a part intended to improve performance, I want to know how it works!
I put this page together to describe our design process, and hopefully give you some insight into our abilities, and method of design. We're quite proud of the work that we do, and I hope that this page will show you the quality, and overall scope of our design process.
The first thing you might notice is our reference to "Engineered" suspension systems at the top of the page.
We feel it is important to distinguish bewtween an engineered system, and a "Best Guess" collection of parts. Just like an engine, the suspension system on your car is a combination of parts that must work together properly to achieve the best performance.
The key is to quickly and accurately arrive at that optimum combination
At first consideration, a race car suspension appears to be an extremely complex system. While I will not argue that it is simple, I will argue that it can be described quite thoroughly using basic engineering principles that date back to the early 1700's. I'm sure that Sir Isaac Newton never asked himself "How can I make a race car go faster?" He did however give us the tools necessary to ask, and answer that question.
Being a fairly simple minded person, I tend to break complex problems down into simple easy to understand pieces, and find that basic engineering principles allow me to do that.
Over the course of several years I have done just that with the "complex" system we call a racing suspension, and I will now attempt to describe our process of engineering a quality racing suspension system.
So...get yourself a comfortable chair, and follow me through the process.
Throughout the article, there are several words that are highlighted as a link. Clicking any of these will take you to the suspension glossary where you will find a detailed definition of that word. (Even if you're not interested in the definition, you will have a new collection of big words to throw around at the next bench racing session.)
There are basically two types of projects we see here. One is a GT type application where we can start with a clean sheet, and the other is a production based racer, and the class rules that go along with it.
We'll use a production car for an example.
The first step is to detemine what we have to start with. We begin the process by getting the car on the scales to find out what it weighs, and how that weight is distributed. In some cases, we start with a bare tub, and have to make estimations based on the class minimum weight, and how we think the weight will be distributed in the final application.
Once the basic scale weights have been recorded, we remove the existing suspension pieces from the car, weigh all of them separately, and record the unsprung masses. for the front and rear of the car.
After we subtract the unsprung masses from the total mass at either end, we have the front and rear sprung mass.
The next step is to measure the spring rate of the tires. This is critical information, because the spring rate of the tire works in conjunction with the springs and bars on the car to determine the final weight transfer characteristics. We do this the same way you would check the rate of a spring, by plotting displacement vs force, and then calculating the rate.
Now comes the part that is a real pain. Building the geometry model. To do this, we use a program called Susprog. Susprog requires input for all of the suspension pivot points, and upright dimensions. Once that information has been entered, the result is a very detailed model which gives us important details like roll center locations and migration, camber curves, bump steer, ackerman, anti-squat, anti-dive, scrub, tramp, and jacking.
Gathering the required information is referred to as plotting the car. To plot the car, we get it up in the air, remove all the suspension components, and measure x,y, and z coordinates for all of the suspension pickup points, and for the pivot points on the uprights.
This amounts to spending hours, sometimes days with a measuring tape, plumb bob, a handful of lasers and two very sore knees. In most cases, we will bang our head at least once in the process. (We charge extra if we bleed.)
Below is a screen capture of the front suspension of a GT-1 Corvette in Susprog, and below that is another screen capture showing just some of the data that Susprog generates. This information will be used later to either design to a specific goal, or determine how to best deal with the stock geometry.
At this point we have a complete geometric model, several days have gone by, and it is time for the next step.
This is where the fun starts. Being frustrated at the lack of suspension tuning information available, we developed our own simulation software. This is an ongoing project and this program is added to regularly to increase our understanding of vehicle dynamics.
It all started when Chris Harrison, from Harrison Auto Dynamics shared a spreadsheet that he had gotten from the Claude Rouelle vehicle dynamcis seminar. The spreadsheet calculated basic weight transfer in cornering and that got us thinking. After playing with this for a while, we started from scratch to be sure the math was correct, and to improve our own understanding of the basics.
Initially all we had was basic weight transfer data, but we soon felt the need to quantify the effects of the dampers. Over the next few years we crunched a lot of numbers, and listened carefully to some talented drivers until our mathmatical model matched what the drivers had to say about the "personality" of the car.
We now have what we call a dynamic response model which essentially describes how the car reacts to transient inputs and more importantly how the car "feels" to the driver. The modeling has now advanced to the point of being able to closely match the handling characteristics of two completely different vehicles through shock valving, and spring/bar selection.
This is done by relating the damping to the mass and inertia of the sprung and unsprung masses, and the energy storage of the springs, bars, and tires. By treating the suspension as a simple dynamic system, with a defined natural resonant frequency, and damping ratio, we can quantify the dynamic response of the system.
The importance of this is that it has allowed us to move known good combinations from one car to another, regardless of masses, motion ratios etc. After doing this enough times we were able to dial in on some combinations that maximise grip and tire compliance for any vehicle.
As a pure analytical tool, it has allowed us to try a number of "what if" scenarios on the computer screen before ever trying it on a vehicle. For instance, we used it to determine the effects of the roll center height and the roll axis on the vehicles transient response. While not initially designed for this purpose, we quickly realized that we could quantify the effects numerous variables through accurate mathematical modeling.
By now, you're probably tired of all this, and would like to know what comes next.
What comes next is the process of deciding what we can do to correct the basic deficiencies of the vehicle within the rules. This amounts to a lot of head scratching and "what if" scenarios punched into both Susprog, and our own software. While we will not describe all the steps we go through, or all the decisions that are made throughout the process, (We need to keep something to ourselves!) there are a few basic issues that we will discuss here. The first is ride height. The car must be set to the minimum allowable ride height, either determined by the rules, or the requirement that the underside of the car not grind itself to pieces as it goes around the track.
Once the car is "lowered" in Susprog, we have a clear picture of the mess we are left with, and we set out see how we can fix it. In some cases, absolutely nothing can be done within the class rules (Limited prep EP Miata for instance) and we have to make the best of what we have, with a combination of spring, bar, and shock. In other cases we can effect geometry changes by fabricating new pieces.
Let's consider the case where we cannot fix the geometry first.
In many cases the roll axis is way out of whack, and we have to "trick" the car into behaving with appropriate shock valving. While this is not as good as fixing the geometry, we can normally make a car quite well behaved by using damping forces to affect the cars basic transient response. This is not as complex as it sounds, because our dynamic response model will tell us quite clearly what is happening and how the dampers will affect it.
In addition to the basic dynamic response, we have to consider the geometry,and limit the roll, dive, and squat to acceptable levels. The first step is to determine how much nose dive we can accept under braking. This is determined based on the bump steer and front camber curve. For instance, if the car has a very steep camber curve at the front, we will need to severely limit nose dive under braking to keep the front tires flat on the ground. This determines the front spring rate. How do we determine that? Simple...we enter a target nose dive number, and rate of deceleration into our program, and it spits out the answer for us!
This number is based on calculations that refer to the center of gravity height, wheelbase, and anti dive geometry which comes from Susprog.
The key is in knowing just how much nose dive the car should have, and that is a matter of experience. (Unfortunately, our program does not do everything for us.)
The same considerations/calculations are applied to the squat of the suspension at the rear of the car under acceleration, and we now have the basic front and rear spring rates. What's left is to determine how much anti roll bar is required to limit the final roll angle to an acceptable level, and to achieve proper front to rear weight transfer distribution which determines overall balance.
Just as with the squat and dive calculations, we simply enter a few target numbers and let the computer do the work.
Once this process has been completed, we will have determined spring and bar rates required to achieve a neutral balance in a steady state condition. The next step is to determine a damping curve for the front and rear suspension to maximise grip, and a achieve a neutral balance in transient conditions.
Note: It is quite common for a car to have a different balance in steady state, and transient conditions. This always leads to a situation where the car oversteers in some corners, and understeers in other corners. In many cases, the car will even change balance in a single corner. For instance, the car pushes on entry, and is loose on exit, or vice versa. For that reason, we always strive to match the steady state and transient balance. The result is better overall grip, and more importantly, a car that gives the driver confidence to drive at the limit.
This damping curve will then be converted to a standard force vs velocity curve, and the shocks will be valved to match the curve.
Unfortunately this process still requires a lot of trial and error. It is not uncommon to change the valving a half dozen times or more before we hit the target curve, and hopefully walk away from the dyno with our sanity intact. (And our hair and clothes drenched with shock oil.)
Once all this is complete, we have defined the basic "personality" of the car, and have a reasonable starting point for initial track testing. It would be ridiculous to say that all this number crunching has resulted in a car that will be perfect right out of the box, but it will result in a car that is very driveable, and not too far from its ultimate ideal setup.
As you can see, the days of spending half a season gettng a car sorted out are a thing of the past. In many cases, the car can be dialed to its final configuration within the adjustment range of the bars, and the dampers without further re-valaving, or bar changes.
So there we have it. The basic process of "tuning" a suspension system on the computer.
Now let's consider a situation where we do not have our hands tied by the rules.
In the case where we can make some geometry changes, we really get to have some fun because we get to build parts, instead of sitting in front of the computer.
In many racing classes, some geometry modifications are allowed. Quite often McPherson struts can be fabricated from scratch, or suspension pickup points can be moved, uprights can be modified, etc. In this case we start by determining just how much we can "move" the important pivot points. In some cases the rules determine that, and in other cases the physical restraints of the car are the determining factor. For instance how much can we lower the bottom ball joint before it contacts the wheel?
To answer these questions we use another piece of software called Solidworks, which is a 3D modeling program. Solidworks allows us to build a 3 dimensional model of any part we can imagine, and it even allows us to build a complete suspension which moves just like the real thing. Once the suspension is modeled, we can grab the upright with the mouse and move it in any direction within the restraints imposed by the suspension links themselves. In other words, we can build a complete suspension system, and run it through its full range of travel without ever leaving the office.
This allows us to check for interference, (Does the control arm contact the wheel limiting full turn radius?) determine the mass of the parts, and visualize what we have in mind before we ever cut a piece of steel. While the modeling is rather time consuming, it actually saves time and material, and results in a better part in the long run. Additionally once the part is designed, it generates blueprints which simplifies building the parts and gives us a permanent record so that we can quickly reproduce these parts in case of crash damage.
Below is a few screenshots from Solidworks showing a work in progress. It is the rear half of a GT-1 Corvette, and a close up of the Mumford Link assembly on that same car.
Notice the upper link running right through the chassis? This is one of the benefits of 3D modeling. In this case, the pickup points on both the rear end housing, and the chassis were determined in Susprog. The numbers looked good there, but the image on the screen makes it clear that something must be changed before the concept is committed to steel.
3D modeling offers additional benefits that make it invaluable. The most important being Finite Element Analysis, or FEA. The glossary will give further details, but the quick description is that FEA performs very accurate stress analysis on nearly anything that can be modeled. By using FEA we can "stress" the parts in the computer and see if they are tough enough for the job, or maybe much heavier and stronger than they need to be.
More importantly it shows the stresses in the part, and is an invaluable guide to proper design. Quite often a part can be sufficiently strong in most areas, but have just one point that is over stressed which will lead to eventual stress cracking, or maybe even catastrophic failure.
Additionally it shows the displacement for a given load, so we know if the part is stiff enough to keep the wheels in proper alignment, or if it will flex to the point of affecting the geometry.
Below are two screenshots showing the result of FE analysis on a bellcrank which is part of the Mumford linkage shown in the previous picture. The applied loads are equivalent to the car cornering at 3 g's (Twice the expected average)
The graph on the right relates stresses to the color seen on the screen. Note the high stress area in red at the crotch of the bellcrank. It is clear that this is the weak spot on the bracket, and if it were overstressed, we could expect failure to start here. Also note that the stresses in the rest of the link are relatively low. Due to the shape of the part, much of it is thicker/stronger/heavier than it needs to be. To use thinner material would result in the stresses being too high in the crotch of the bellcrank.
A better option would be to use thinner material, but to extend the crotch out further, giving it a larger radius, or possibly even change the shape to a basic triangle with rounded corners and eliminate the crotch all together.
Whatever the case, the existing stress distribution is known, and making further changes to improve/lighten the part is a piece of cake. Most importantly, we know the part is strong enough to do the job without suddenly breaking and putting the car and driver into a wall, or another vehicle.
This plot shows the displacement for the applied load. Referencing the chart on the right, we can see that maximum displacement is a little less than .005". A small enough number that we don't need to worry about flex in this application.
By now your eyes are probably burning from reading all this fine print. That's OK, because we're nearly done. Once the design work is all completed, and we know that our geometry, balance, and transient response is good, and the parts have been designed, checked for clearance, and been analyzed/optimized using FEA, we hit the print button, carry the blueprints back into the shop, and start cutting and welding!
At some point we will include pictures of the fabrication process, but chances are you've all seen that before.
We will probably add to this page as time goes on, but for now, I think the point has been made, and you are left with a good feel for our design process, and capabilities.
A race car suspension is a simple mechanical device, and by applying the basic laws of physics, we can gain a fairly thorough understanding, and hopefully offer a superior system.
Our level of understanding improves with every vehicle and we try to make every system better than the last. We will continue to improve our understanding, much in the same way that a quality driver constantly works to improve his skill. In the racing game, if you're not moving forward, you are moving backwards.
Comments? Questions? Use the Contact link on the right menu bar to send us an email.
Thanks for reading.
Paul Yaw - Yaw Power Products