We propose a full factorial DOE on baking the best banana bread based on the
ripeness of banana used, temperature baked at, and time that it was baked at. For the
levels chosen, the ripeness of banana is; low=not ripe (no brown spots), high=ripe (mostly
brown). For baking temperatures; low=325° F, high=350° F. Finally, the levels for baking time
are; low=60 minutes, high=70 minutes. Our yield is the quality of the banana bread or how
much the judge enjoys eating it. The judge will score the bread on a scale of 1-10. There will
be only one judge for the bread so we will not need to eliminate bias.
To determine our standard error, we will bake each combination, of which there are
8 since we are running a 23 full factorial design. We plan on calculating the standard error
with the t * s = SE equation, which could be computed after doing all the runs. We would
then compare the standard to the effects from each category to determine if the factor is
significant or not.
Beyond using the standard error, we plan on creating pareto plots and using the
regression model to see how the factors affect each other. We will eliminate irrelevant
factors on a step by step basis from the pareto plot as to not miss any important factors or
interactions. Generating the regression analysis in Minitab will allow us to see if factors are
important and if we need to add or subtract a factor for a better yield. To ensure we get the
best results we will use the optimization feature in Minitab as well and maximize our yield
or our score from the judge.
We would follow up with a t test. We would be using a reference average of the
judge’s favorite banana bread that his mom makes as a reference. We will compare the two
averages and see if they fall within the range we are looking for and see if the maximized
bread makes the cut or needs further testing. The second study will have the judge eat a
slice of each banana bread, the control and the maximized from the previous study. The
score he gives to each will act as their average. Randomization is just to eliminate bias, but
we will use the control bread as the average to compare to.
For the third study, the one at a time study we would test the ripeness and keep the
baking time and temperature constant. This would allow us to better determine if the
bananas have much of an effect on the taste when it is baked at a typical optimal time and
temperature. We would use bananas at 4 different levels for this. The first being a green
banana, then a yellow one with no brown spots, then a banana with brown spots but still
yellow, then a mostly brown banana. If they are all baked at the same time and temperature
this would give a more thorough look at how the ripeness affects both the taste and
structure of the bread.
The workload needed would be 8 runs for the first experiment, the DOE. And then 4
runs for the one at a time component. For the DOE it would be a bit simpler through since
you would only need to make two batches of batter for the two levels of ripeness and then
use them accordingly. You wouldn’t need to make more than two bowls of batter since the
other factors are about the actual cooking time not the ingredients. This would eliminate
most of the work, you could even put some of the samples in the oven at the same time if
they are the same temp and time. This would cut back on a fair amount of time by the oven.
Overall, the work load here is rather minimum compared to most DOE.