Soderberg Line

In the world of engineering, components often fail not from a single overwhelming force, but from the cumulative damage of millions of smaller, repetitive loads—a phenomenon known as fatigue. Designing against this silent adversary requires more than simply ensuring stress stays below the material's breaking point; it demands a nuanced understanding of how fluctuating loads, composed of both a steady mean stress and a vibrating alternating stress, dictate a part's lifespan. A critical knowledge gap for designers is choosing a failure criterion that matches the specific safety requirements of an application.

This article provides a comprehensive exploration of one of the most fundamental and cautious approaches to fatigue design: the Soderberg line. Across the following sections, you will gain a deep understanding of this powerful model. The "Principles and Mechanisms" section will deconstruct the concepts of mean and alternating stress, introduce the Haigh diagram as a map for fatigue safety, and derive the Soderberg line from first principles, contrasting it with its counterparts, the Goodman and Gerber criteria. Following this, the "Applications and Interdisciplinary Connections" section will bridge theory and practice, demonstrating how engineers use these criteria to design real-world components, account for stress concentrations and manufacturing effects, and even predict lifetimes under complex, variable loads.

Imagine you're designing a critical part for a machine—say, a connecting rod in an engine or a bolt on an airplane wing. You run the numbers and find that the highest stress the part will ever see is well below the material's breaking point. You breathe a sigh of relief. It's safe, right?

Not so fast. In the real world, many components don't just sit there holding a steady load. They vibrate, they shake, they are pushed and pulled, over and over, millions of times. This cyclic loading introduces a silent and insidious mode of failure: ​​fatigue​​. A part can break under a cyclic stress that is much lower than the stress it could handle just once. It's like bending a paperclip. You can bend it once significantly without it breaking, but bend it back and forth, even a little, and it will eventually snap. Our job, as thoughtful engineers and scientists, is to understand this "wobble" and design for it.

First, we need a language to describe this cyclic loading. Any fluctuating stress, no matter how complex, can often be simplified into two key components. Let's say the stress in our part swings between a minimum value, $\sigma_{\min}$, and a maximum value, $\sigma_{\max}$, over and over again.

We can describe this cycle with two numbers:

1. The ​​mean stress​​, $\sigma_m$: This is the steady, average stress that the part is under. It’s the midpoint of the stress fluctuation. $\sigma_m = \frac{\sigma_{\max} + \sigma_{\min}}{2}$

2. The ​​alternating stress​​, $\sigma_a$: This is the amplitude of the stress "wobble" around the mean. It's half the total range of the stress swing. $\sigma_a = \frac{\sigma_{\max} - \sigma_{\min}}{2}$

It turns out that fatigue life doesn't just depend on the peak stress, $\sigma_{\max}$. It's the combination of the steady mean stress and the wiggling alternating stress that dictates how long a part will last. Two components could have the same peak stress but vastly different fatigue lives if their mean and alternating components are different. Separating the load into these two parts is the crucial first step in any fatigue analysis.

So, how do we visualize which combinations of $\sigma_m$ and $\sigma_a$ are safe? We create a map. This map, known in engineering as a ​​Haigh diagram​​, is a simple graph. We plot the mean stress, $\sigma_m$, on the horizontal axis and the alternating stress, $\sigma_a$, on the vertical axis. Each point on this map represents a unique cyclic loading condition.

Our goal is to draw a boundary on this map. Inside this boundary is the "safe zone," where a component can theoretically live forever (or for a very long time). Outside the boundary is the danger zone, where failure by fatigue is expected after a finite number of cycles. The question then becomes: how do we draw this boundary?.

To draw our boundary, we need to plant some flags—anchor points grounded in physical reality. We can find these by looking at the two simplest possible loading cases, which correspond to the axes of our Haigh diagram.

• ​​Anchor 1: The Vertical Axis ($\sigma_m = 0$)​​ What if there's no mean stress? The load swings symmetrically between a positive and a negative value. This is called ​​fully reversed loading​​. Through extensive testing, scientists have found that for many materials (like steel), there is a magical stress amplitude below which the material can withstand an infinite number of cycles. This threshold is called the ​​endurance limit​​, denoted as $S_e$. This gives us our first, undeniable anchor point on the map: the point $(0, S_e)$. Any purely alternating stress below this value is safe..

• ​​Anchor 2: The Horizontal Axis ($\sigma_a = 0$)​​ What if there's no alternating stress? The load is perfectly steady. This is just a simple static strength test. But here, we face a philosophical choice. What do we define as "failure"?

  • ​​The Cautious View:​​ For a ductile metal, long before it actually breaks, it will begin to stretch permanently. This is called ​​yielding​​, and the stress at which it begins is the ​​yield strength​​, $S_y$. In many precision applications, any permanent deformation is considered a failure. If we adopt this cautious view, our anchor point on the mean stress axis is $(S_y, 0)$.
  • ​​The Ultimate View:​​ If we are willing to allow some permanent deformation and are only concerned with preventing the part from actually snapping in two, we would use the material's ​​ultimate tensile strength​​, $S_u$. This is the highest stress the material can withstand before it begins to neck down and break. This gives a different anchor point, $(S_u, 0)$.

This choice between $S_y$ and $S_u$ is not just a technical detail; it's the foundation for different design philosophies, leading to different safety maps.

Let's adopt the most cautious philosophy. We want to design a component that not only lasts forever but also never, ever permanently deforms. Our two anchor points are the endurance limit on the vertical axis, $(0, S_e)$, and the yield strength on the horizontal axis, $(S_y, 0)$.

What's the simplest way to connect these two points to form a boundary? A straight line. This straight line is the ​​Soderberg Line​​..

The equation for this line is beautifully simple, a hallmark of a powerful idea: $\frac{\sigma_a}{S_e} + \frac{\sigma_m}{S_y} = 1$

Any combination of $(\sigma_m, \sigma_a)$ that satisfies this equation lies on the failure boundary. Any point below the line is in the safe zone. The brilliance of the Soderberg criterion lies in its promise: if your design point is under this line, you have guarded against two failure modes simultaneously. You are safe from infinite-life fatigue and you are safe from first-cycle yielding. This is because the line is constructed to ensure that the maximum stress, $\sigma_{\max} = \sigma_m + \sigma_a$, never exceeds the yield strength $S_y$.

This conservatism is not just for peace of mind; it is essential for certain applications. In precision machinery, rotating shafts, or critical aircraft structures, any unexpected permanent change in a component's shape (yielding) can lead to a catastrophic failure of the entire system. In these cases, the Soderberg criterion isn't just a good choice; it's the only responsible choice. It directly addresses the design requirement of "no plasticity, ever".

The Soderberg line is a profound and useful concept, but it's part of a larger family. To truly understand it, we must meet its siblings, who are built on a slightly more adventurous philosophy. What if we chose the ultimate tensile strength, $S_u$, as our static anchor point instead of the yield strength?

• ​​The Goodman Line:​​ If we draw a straight line from $(0, S_e)$ to $(S_u, 0)$, we get the ​​Goodman Line​​. Its equation is $\frac{\sigma_a}{S_e} + \frac{\sigma_m}{S_u} = 1$. Since for most ductile metals $S_u > S_y$, the Goodman line lies above the Soderberg line, defining a larger safe region. It allows for higher stresses, but with a catch: it doesn't protect against yielding. A design that is "safe" by the Goodman criterion might still permanently deform on its first high-stress cycle.

• ​​The Gerber Curve:​​ Decades of experiments have shown that for many materials, even the Goodman line is a bit too conservative. The real failure points tend to follow a curve that lies above the Goodman line. The ​​Gerber Curve​​ is a parabolic fit that also connects $(0, S_e)$ and $(S_u, 0)$. Its equation is $\frac{\sigma_a}{S_e} + \left( \frac{\sigma_m}{S_u} \right)^2 = 1$. It defines the most generous safe region of the three..

On our Haigh diagram, for any given mean stress $\sigma_m > 0$, the allowable alternating stress will always follow this order: $\sigma_{a, \text{Soderberg}} \sigma_{a, \text{Goodman}} \sigma_{a, \text{Gerber}}$

This hierarchy represents a spectrum of conservatism. Soderberg is the most cautious, guarding against both fatigue and yield. Gerber is the least cautious, providing the best fit to experimental fracture data but offering no inherent protection against yielding. Goodman lies in between..

For example, consider a steel with $S_y = 600\,\text{MPa}$, $S_u = 900\,\text{MPa}$, and $S_e = 400\,\text{MPa}$. If it is subjected to a mean stress of $\sigma_m = 220\,\text{MPa}$, the Soderberg criterion predicts it will fail, while Goodman predicts it is just barely safe, and Gerber predicts it is comfortably safe. The choice of criterion is a crucial engineering decision that reflects the design's specific requirements..

A beautiful feature of this entire framework is its adaptability. We built it around the concept of an infinite-life endurance limit, $S_e$. But what about materials like aluminum or magnesium alloys, which don't have one? Their S-N curves continue to slope downward; more cycles always mean a lower fatigue strength.

Does this mean our elegant maps are useless? Not at all! The logic is robust. We simply replace the idea of an "infinite life" with a specific ​​target life​​, say, $10^6$ cycles. We then go to the material's S-N curve and find the fatigue strength for that specific life, which we'll call $S_f$.

Now, we just repeat the same process. We anchor our lines and curves at the point $(0, S_f)$ on the vertical axis.

• The ​​Soderberg line for a life of $N$ cycles​​ connects $(0, S_f)$ to $(S_y, 0)$.
• The ​​Goodman line for a life of $N$ cycles​​ connects $(0, S_f)$ to $(S_u, 0)$.

Instead of a single map for infinite life, we now have a whole atlas of maps—a constant-life diagram for any lifespan we choose. This allows us to make precise predictions and designs even for materials that will eventually wear out. The fundamental philosophies remain: Soderberg prevents yielding for that target life, while Goodman and Gerber are concerned with fracture. The inherent beauty and unity of the underlying principles shine through..

Now that we have explored the principles behind mean stress and its effect on fatigue, we might be tempted to leave these neat lines and parabolas in the abstract world of graphs. But that would be a terrible mistake! For these are not mere academic doodles; they are the tools of creation, the very compass by which engineers navigate the perilous seas of stress and strain to build a world that endures. The responsibility is immense—to design an aircraft wing that can flex for millions of cycles, a bridge that can bear the ceaseless rhythm of traffic, or an artificial heart valve that must beat billions of times without fail. Fatigue is the silent, patient adversary in this endeavor, and the concepts we've discussed are our first and most crucial line of defense. Let's embark on a journey to see how these ideas come to life, connecting the dots between physics, engineering, and even statistics.

Imagine you are an engineer designing a simple steel tie-rod. You know the material's properties—its yield strength $S_y$, its ultimate strength $S_u$, and its endurance limit $S_e$. You also know the service conditions: the rod will be under a steady (mean) tensile stress $\sigma_m$ and will also vibrate with an alternating stress of amplitude $\sigma_a$. The fundamental question is: will it last forever?

Here, our family of fatigue criteria—Soderberg, Goodman, and Gerber—offer different philosophies for answering this question.

The ​​Soderberg line​​ is the most cautious philosopher of the group. It operates on a strict principle: a component must not only resist fatigue fracture for an infinite number of cycles, but it must never, ever yield, not even on the very first load cycle. By tying its safety envelope to the material's yield strength $S_y$, the Soderberg criterion guarantees absolute dimensional stability. For applications where any amount of permanent deformation is catastrophic—think precision instruments or components with tight tolerances—the Soderberg criterion is the undisputed choice.

The ​​Goodman and Gerber criteria​​ are the more daring pragmatists. They argue that as long as the component doesn't ultimately break apart, a tiny, localized amount of yielding might be acceptable. They anchor their safety envelopes to the material's ultimate tensile strength $S_u$, which is always higher than the yield strength. This opens up a larger "safe" operating region, allowing for designs that are lighter and more materially efficient.

But this pragmatism comes with a crucial warning, a beautiful trap for the unwary designer. Imagine using the Gerber criterion to find the maximum mean stress a part can handle for a given vibration amplitude. The calculation might cheerfully announce a large, safe $\sigma_m$. The part is safe from fatigue, yes. However, if the peak stress of the first cycle, $\sigma_{\max} = \sigma_m + \sigma_a$, exceeds the material's yield strength $S_y$, the part will stretch like taffy and be ruined before fatigue even has a chance to get started! This reveals a profound truth: the true safe operating zone is a territory carved out by multiple constraints. It is bounded not only by the chosen fatigue curve but also by the simple, inviolable law of first-cycle yielding.

Our idealized tie-rod is a useful fiction. Real-world components are riddled with features: bolt holes, grooves for retaining rings, sharp corners, and changes in diameter. These are not just geometric details; they are stress concentrators.

Think of stress flowing through a part like water in a channel. A smooth, uniform channel allows for smooth, uniform flow. But place a large rock in the middle, and the water must rush around it, accelerating to high speeds at the rock's edges. A geometric feature does the same to the flow of stress. A hole or notch forces the lines of stress to crowd together, and the local stress at the edge of the feature can be many times higher than the nominal, or average, stress in the part. This amplification factor is the theoretical stress concentration factor, $K_t$.

Now, here is where materials science adds a wonderful twist. It turns out that in fatigue, not all materials "feel" the full brunt of this theoretical concentration. Due to microscopic plastic deformation at the notch tip, the material can blunt the sharpness of the stress, effectively resisting the concentration. This material-dependent effect is captured by the ​​notch sensitivity​​, $q$. A tough, ductile material might have a low $q$, while a brittle, high-strength material might have a $q$ close to 1, meaning it feels the full force of the notch.

The effective stress concentration for fatigue, known as the fatigue strength reduction factor $K_f$, beautifully combines pure geometry ($K_t$) and material behavior ($q$):

When an engineer analyzes a real part, they must first calculate the local stresses at the notch root, $\sigma_{a,\text{local}} = K_f \sigma_a$ and $\sigma_{m,\text{local}} = K_f \sigma_m$. It is this higher, localized stress state that must be plotted on the Haigh diagram and checked against the Soderberg or Goodman criteria. Failure is a local event; it almost always begins at the point of highest stress.

So far, we have assumed that a component is stress-free until we hang a weight on it. This is rarely the case. Stresses can be locked into a material during its very creation, a permanent legacy of the manufacturing process. These are called ​​residual stresses​​.

A welded joint, for instance, cools unevenly. The regions near the weld want to shrink but are restrained by the cooler surrounding metal, leaving them in a state of high tension. This tensile residual stress, $\sigma_r$, is a ghost in the machine—a stress that exists without any external load.

The principle for dealing with this is beautifully simple: linear superposition. A stable residual stress acts just like an additional mean stress. The effective mean stress a material point experiences is the sum of the applied mean stress and the residual stress:

The residual stress does not change the stress amplitude $\sigma_a$, because it is a static, unchanging field. A high tensile residual stress from welding can be so dangerous because it dramatically shifts the operating point on the Haigh diagram into a less safe region, using up a large portion of the material's fatigue life before it even sees its first service load.

But what man can put in, man can also control. This leads to one of the most powerful applications in manufacturing engineering: creating beneficial residual stresses. Processes like ​​shot peening​​—essentially bombarding a surface with tiny spherical media—create a layer of high compressive residual stress. This compressive stress acts as a protective shield. It effectively pushes the operating point down on the Haigh diagram, often into the compressive mean stress region where cracks struggle to grow. This is why critical components like aircraft landing gear and engine connecting rods are often shot-peened; we are pre-loading them for a longer, safer life.

Our simple sine wave loading is a nice melody, but real service loads are a chaotic symphony. Consider the stress history of a car's axle bouncing over potholes or an airplane wing encountering turbulence. The stress goes up and down, with large cycles mixed in with small ones. How can we apply our simple criteria to this mess?

The answer is an ingenious algorithm called ​​Rainflow Counting​​. Imagine the stress history graph turned on its side, looking like a series of pagoda roofs. Now, let it rain. The algorithm lays down a set of rules for how water flows down these roofs, cleverly pairing up individual peaks and valleys to identify every closed stress-strain loop hidden within the complex signal.

This powerful tool allows us to decompose a chaotic stress history into a simple list of individual cycles, each with its own amplitude $\sigma_{a,i}$ and mean stress $\sigma_{m,i}$. From here, the procedure is a masterpiece of engineering synthesis:

1. For each cycle in the list, a mean stress correction (like Goodman) is used to calculate an "equivalent fully reversed amplitude," which is the amount of damage that cycle would do if it had zero mean stress.
2. Using the material's baseline S-N curve, we find the number of cycles to failure, $N_i$, for that equivalent amplitude.
3. We then invoke the ​​Palmgren-Miner rule​​. Think of the component's life as a budget. Each cycle "spends" a fraction of that budget, equal to $1/N_i$. We sum these fractions for every cycle in the history. Failure is predicted when the total damage sum reaches 1.

This complete workflow, from raw signal to a final life prediction, is the heart of modern fatigue analysis and is embedded in the sophisticated software used to design nearly every vehicle and machine we rely on.

Our journey so far has treated material properties like $S_y$ and $S_u$ as fixed, deterministic numbers. But the real world is a dance of variation. No two batches of steel are perfectly identical. The strength of a material is not one number, but a statistical distribution—a bell curve with a mean and a standard deviation.

This forces us to ask a more sophisticated question. Instead of "Is it safe?", we must ask, "What is the probability of it being safe?". This shifts our perspective from deterministic design to ​​probabilistic design and reliability analysis​​.

By treating material strengths as random variables, we can use the tools of calculus and statistics to see how their uncertainty propagates into the final safety margin. The limit-state function, which defines the boundary between safe and failed states, itself becomes a random variable. We can then compute its mean and standard deviation. The ratio of the mean safety margin to its standard deviation gives us a ​​reliability index​​, $\beta$. This single number is a powerful measure of robustness. A design with $\beta=3$ has a failure probability of about 1 in 1000. A design for an aircraft component might require $\beta=6$, corresponding to a failure probability of 1 in a billion. This approach allows engineers to quantify risk and make informed decisions, a crucial capability when the consequences of failure are high.

We began with simple lines on a page—Soderberg, Goodman, Gerber. We saw them as an engineer's compass for designing durable machines. But as we looked closer, these lines transformed. They became windows into the geometry of real parts, into the hidden legacy of manufacturing, and into the chaotic symphony of real-world service loads. They connected the design engineer's desk to the statistician's probability charts.

And in the end, we find the deepest connection of all, linking back to the fundamental physics of matter. Why is the Gerber parabola often a better fit for ductile metals than the Goodman line? The answer lies in the microscopic world of crack tips. In some conditions, the rough faces of a microscopic fatigue crack interfere with each other as the load cycles, propping the crack open. This phenomenon, called ​​crack closure​​, shields the crack tip from the full effect of the stress cycle and makes the material less sensitive to the mean stress—a behavior that the gentler curve of the Gerber parabola happens to mimic beautifully.

So, these are not just arbitrary curves. Their shape is a macroscopic echo of microscopic events. They represent a profound unity between the abstract world of mechanics, the tangible world of materials science, the practical world of manufacturing, and the statistical world of reliability. They are a testament to our ability to distill complex physical reality into models of elegant and powerful simplicity.