Speaker | Transcript |
| Na Liu | Hello, everyone. Today, we would like to present using definitive design...screening design to guide lubricants formulation development in 14 lubricants. My name is Na Liu. I'm a scientist at Fuchs Lubricants. |
| Adam.Kyrouac | And I'm Adam Kyrouac. I'm a chemist at Fuchs Lubricants and to start off this presentation, let's establish little a background in forging. |
| It is used to make items you use every day, from car parts to kitchen utensils. Forging is defined as the shaping of a metal using heat |
| and compressive force and it usually requires a special lubrication to help perform this process without damaging equipment or the final product. |
| Now the most common additive used in forging lubricants is graphite, due to the fact that it provides excellent lubricity at high temperatures and pressures. |
| However, graphite comes with main drawbacks, the biggest two being how cleanliness...we call it the cleanliness problems it causes and crystalline silica it contains. |
| Generally kind of this problem stems from the dirty look it causes from |
| airborne graphite. |
| And it can cause electrical problems too, because it can get into electrical systems from that. As for the crystalline silica this a problem with natural graphite as it can...it can create...it can contain |
| crystalline silica that can cause respiratory problems if inhaled. In today's market most customers are looking away to move away...looking to move away from that because of the problems that causes. As part of our next generation forger product, we decided to begin to develop a new graphite... |
| new graphite-free oil-based forge lubricant. And to do that we need to know what additives, whether by themselves or in combination with other additives, have a significant |
| result on the COF...will not have negatively affecting the film's coverage and viscosity. And to do that, we decided that definitive screen design would be an excellent way to begin this project. Now I'll hand off to Na to discuss DSD. |
| Na Liu | Thank you, Adam. Since the first publication, definitive screening design attracted a lot of attention in many applications. |
| It has been proved to be a very effective and efficient way to scoring many factors in a relative small design. |
| So in definitive screening design, there is never a complete confounding with two factor interactions, so we can easily evaluate the secondary interactions. |
| It also includes runs that are not in second-level designs, so we can use it to estimate square terms. |
| For this study, we started with eight different chemicals and five responses. Using the DSD platform in JMP, we decided on a design of 24...21 runs. |
| The main factor powers are all above .09. And the nature of the DSD ensured very small correlations among main factors and secondary level interactions as well as quadratic terms. |
| DSD really allows us to explore not only the main factors, but also the secondary interactions and curvature in the responses. |
| DSD, stepwise and generalize regression |
| to compare and identify the optimal model. For stepwise method, we feed models from one to 10 term. All models are compared by R squared and AICc values. |
| The final models are selected for each response with consideration of the R squared, the AICc values, as well as appearance frequency of each term. |
| So, for instance, we compared the 10 models for viscosity and took a very careful look at the terms. |
| The R squared increased, while the term number increased in stepwise model. However, we must balance it with the AICc value to avoid overfitting issues. |
| The other factor we considered is the appearance frequency of each term. It is preferred to have terms considered significant in multiple models. |
| After the comprehensive consideration, the SDS model was selected for the viscosity data fitting. So for the rest of the responses, the same process was applied. |
| From here, I would like to hand back to Adam to talk about the findings from our study. |
| Adam.Kyrouac | Thanks, Na. The definitive screen design proved to be a very efficient method in identifying additives which would have significant impact on the properties of interest. |
| Going into this DOE, it was thought that additive PS300 and additive LL80 were going to provide the best results, as they have been used in other lubricants we have made and had good results. |
| However, they turned out to have no effect or even strong negative effects on our responses. For instance, let's look at the predictive profile for PS300. |
| You can see it causes very significant increase in viscosity and a significant decrease in film density, but offers virtually no help with COF. |
| Additive LL80 had no effect on any responses at all. On the other hand, there is little known about additives K65 and KL, |
| but they turned out to provides some of the better results to our responses. If you look at the predictive profiler you can see that additives K65 and KL both provide excellent contribution to the film density. |
| Lastly, we see some secondary interactions with additive LS11 and additive MP. Additive MP is the most interesting to the secondary interaction because, while it positively affects the COF, it also negatively affects the film density. |
| Using the information obtained from this DOE, we will perform a mixture design using the selected components to explore the optimal design space as our next step for this project. |
| With that, we come to the end of our presentation. We would like to thank Stephanie Oats from Fuchs Lubricants for her help and her extensive subject matter expertise in forging lubricants and Jerry Fish from JMP for his technical support on the design. |
