Results and Conclusion

From these results, in particular, the copper calculations confirm that this process of measuring thermal diffusivity of various metals can be very accurate. This however starts to loose accuracy as our initial conditions start to break down. This can be seen significantly in the brass measurement, which is off by almost 13 sigma. This sample was significantly thicker than the other two metal samples, in addition it had the smallest width of the three samples, thus the heat would spread to edges of the sample very quickly. Once the edges of the sample were no longer at a fixed temperature the data could no longer be used. Brass also has the smallest thermal diffusivity coefficient of the three metal samples. This effects the initial condition of having a delta function for the initial thermal profile. Because the heating element was a quarter inch steal rod rather than a point source it would have been impossible to achieve this goal. However, this is a fair assumption when the initial ratio of the area of the heating element to the width of the thermal profile is very small. This ratio is largely effected by the thermal diffusivity and thermal conductivity. The lower theses two values are the greater the value of the ratio.

The grade of aluminum was also unknown to us. The calculated value was compared to pure aluminum, and if the sample was some other grade of aluminum alloy, this would affect the thermal diffusivity of the material significantly. For example, if the sample used was aluminum 6061, an alloy composed of 2 to 5 percent of other elements. These impurities in aluminum have the effect of changing the thermal conductivity to 6.4•10^(-6) m^2/s. This would improve our calculated value of aluminum to be 0.89 sigma from the accepted value of aluminum 6061.

The X axis of the unidirectional carbon fiber corresponds to the path along the fibers, while the Y axis is across the fibers. Meanwhile the X and Y axis should produce equivalent values for the 3k woven carbon fiber. Originally, we had hoped to impose the accuracy of the thermal diffusivity calculating process from the metal onto the thermal diffusivity calculations for the carbon fiber. However, due to the inaccuracy from the aluminum and brass samples, the accuracy of the calculated thermal diffusivity for the carbon fiber cannot be determined. In addition to not being able to credit any accuracy to the calculated carbon fiber values, the Y-axis data for the unidirectional carbon fiber was not well represented by a Gaussian distribution. This can be obviously seen by the example below. However, the Y-axis did defuse at a constant rate and future tests could be conducted to find a better model that accurately represents the data. Attempts could also be made to more closely achieve the required initial conditions required by the governing equation, for the width of the initial thermal profile for the Y-axis was about an inch, with the quarter inch steal rod being used as the heating element, this made the ratio previously discussed very poor for our model.