Design of Experiment (DOE) is a statistical method for identifying and optimizing various process factors that affect final product specifications. In a simple example there may be two different factors that might impact the final spec, with two possible levels for each. By performing four different experiments we can identify significant levels and optimal settings for each factor. It becomes increasingly complicated to evaluate the relative impact of greater numbers of process factors and their possible settings.
In these more complex situations, Carestream Contract Manufacturing often conducts a DOE to identify which factors significantly affect final results of the precision coating process. A Design of Experiment makes the testing process more efficient by using statistics to figure out what matters and eliminate the need to conduct endless individual experiments.
A well-designed DOE can reveal how different process factors interact with each other to produce statistically significant changes in the final result. When used to establish a baseline, a DOE can generate a knowledge base about the interaction between a product and the coating process. Then if something changes down the line – such as raw material, formulation, or final specification for example – the data helps engineers predict the impact of new parameters.
In other cases a customer might need a specific optimal design value for a particular specification (e.g., coating thickness, peel strength, level of curl), so we use the DOE to generate empirical data. We then run an additional set of experiments to find optimal settings to reach the final spec as precisely as possible.
For example one customer recently came to Carestream with seven factors they believed might be contributing to results for two different final specs. After initial screening to decide which factors were statistically significant, we narrowed it down to three factors and conducted a DOE. We tested each factor using statistical software at different levels to find the optimal run factors. We then conducted a set of experiments and analyzed the results to determine which factors had significant impacts. The DOE also enabled us to examine interactions between different factors. In this case lamination was a leading factor so we examined the temperature of the rubber roll, temperature of the steel roll, nip pressure and tensions.
Another customer asked Carestream to stop a curing process at a specified point. To narrow down the parameters contributing to that specified cure stage we studied the temperature in different oven zones, airflow (how open the fans were or should be), drying time and other factors.
In instances where a product is not meeting desired specifications, Carestream may use a DOE to troubleshoot. We first conduct a root cause analysis and if we determine that the problem has more than one root cause, DOE is a useful tool in fine-tuning and troubleshooting the necessary processing factors.
In one case, to perfect the coating laydown for a product, we worked with the customer to create a DOE for percent solids, coating laydown and drying time. We first performed the root cause analysis to determine possible contributing factors. We analyzed the factors that could be changed and by how much, and created a DOE to help us optimize the process.
While not all customers and processes require a DOE, Carestream views it as a critical step in implementing good statistical thinking and Six Sigma practices for processes with tight specification parameters. Carestream technicians and engineers are experienced with Six Sigma, so we effectively implement DOE where there is a need. The DOEs conducted by Carestream enable process refinement performance to help move products to commercialization more quickly, saving time and money.