Mechanical Engineering Internship: Volvo Trucks

1. The ability to effectively communicate is one of the most important skills an engineer can have. Working for a multi-national company involving engineers, financial managers, and external suppliers, I realized just how critical it is to learn how to effectively communicate my ideas.

2. Taking initiative is just as valuable as being able to finish a project. Working with full-time engineers at Volvo who are often extremely busy, I learned that it can come down to me to follow up with requests or determine the next steps for a project.

Over their lifetimes, trucks experience cyclic loading on many of their components. Volvo designs its parts to withstand these fatigue loads by simulating how they would behave using finite element analysis.

To accurately simulate the fatigue behavior of materials, we would need the fully reversed fatigue data for the material, where the average mean stress of a given cycle is zero. However, most of the data that original equipment manufacturers receive is not in the fully reversed form: there is some amount of pre-loaded tension or mean stress. This is because fully reversed fatigue tests are difficult to conduct because applying a compressive force on a sample introduces a risk of buckling.

Therefore, when we receive non-fully reversed data, we need to apply a mean stress correction model to transform it into a fully reversed format, and several models can accomplish this, such as the Walker equation, the Goodman equation, and the Gerber equation. To determine which model provided the best mean stress correction for polymers, I used two methods:

1. A detailed literature review: I read and summarized over 20 research papers discussing the fatigue behavior of polymers and composites emphasizing mean stress correction. I documented my findings in a research summary. Moreover, I investigated various low-cycle fatigue life estimation techniques and the application of the mean stress correction models to the low-cycle fatigue regime.

2. An experimental data analysis: although we do not often receive fully reversed fatigue data for our materials, we had both the fully reversed and R = 0 fatigue data for PC/ABS and PC/ASA polymers. Therefore, I was able to directly apply each mean stress correction model to the non-fully reversed data and determine which model produced the best correlation with the fully reversed data.

An example set of S-N curves to illustrate the effect of mean stress on fatigue life predictions. The red curve (representing a non-zero mean stress) would predict shorter lifetimes for a component than the blue curve (representing a zero mean stress).

Based on my literature review, it was conclusive that the Walker model provided the best mean stress correction for polymers and composites. However, the value of the best-fit parameter that gave the best correction varied largely between different materials, ranging from 0.28 for PLA to 0.6 for certain grades of PC/ABS.

Based on the literature review and my analysis of the fatigue data that we had for PC/ABS and PC/ASA, I determined that the Walker equation was the best mean stress correction model for polymers and composites. The next major question involved the best-fit parameter: what parameter values would yield the best results? If this value varies largely between different materials (even between different grades of the same material), how do you determine what value to use?

 To help answer these questions, I performed a sensitivity analysis on the Walker best-fit parameter: I varied the parameter from 0 to 1 with increments of 0.01 and determined how well each corrected curve correlated with the fully reversed data. I found that the S-N curves were not very sensitive to changes in the Walker best-fit parameter; for the PC/ABS, values ranging from ~0.4 to ~0.8 produced an S-N curve that had an R^2 value of 0.9 or greater when correlated with the fully reversed curve. I saw similar results in the PC/ASA data when varying the best-fit parameter from ~0.2 to ~0.4.

Based on these results, my final recommendation was the Walker mean stress correction model with a best-fit parameter value of 0.55. This value was chosen because it is backed up by literature and it was shown to provide a good mean stress correction for the PC/ABS material we investigated. The downside is that, for materials that are very sensitive to the mean stress effect such as the PC/ASA (whose best-fit parameter was 0.28), using 0.55 would yield a conservative result.

However, it is important to remember that we are trying to recommend a best-fit parameter for all of our polymers and composites because we rarely receive fully reversed fatigue data for these materials. Because of this, we need a parameter that will provide a good correction in the best of cases and a conservative correction in the worst of cases so that we don't design components that fail before they are predicted to fail.