| OCR Text |
Show 20 In light of this discussion, the error in the inferred temperatures shown in Fig. 6 and 8 is believed to be 75 K for the highest temperatures. The overall accuracy is estimated to be ± 3 percent. Conclusions The principal results can be grouped into five areas. 1) The requisite alignment necessary for folded BOXCARS phasematching for reasonably high spatial resolution (5 mm) can be maintained over the long path lengths (>10 m) that are often required for industrial applications. Fiber optics, which are convenient to use in these environments but can lead to signal loss due to f/no. mismatch at the fiber-spectrograph interface,37 were not necessary since the divergence of the exiting CARS signal is small enough to allow transfer over similar distances with little loss. 2) The fluctuating index of refraction gradients produced by combustion instabilities typically cause signal losses of an order of magnitude requiring signal averaging for high spatial resolution configurations. 3) A one percent mass concentration of potassium seed in the gas stream has little effect on the ability to recover analyzable CARS signatures even though the visible luminously of the gas stream greatly increases. 4) The introduction of coal fly ash particles has two effects. First, the amount of scattered pump radiation increases greatly and can cause a high nonuniform background to the CARS signal if the radiation scattered in the forward direction is allowed to enter the spectrograph. This was not a problem in the data taken in the test stand, however it was troublesome in data taken in a laboratory particle burner.29 A second and potentially more serious problem is the particle-induced laser breakdown which can cause a large nonresonant CARS signal. Although this can be eliminated by simply reducing the intensity of the focused radiation, doing so without an accompanying increase in interaction length leads to unacceptably low signal levels. If the nonresonant component to the CARS signal is less than 50 percent of the total CARS signal peak, however, the temperature can be recovered from the N2 signature, albeit with less precision. 5) A nonlinear least-squares fitting program was used to assess the error in the inferred temperatures. The effect of uncertainties in the Stokes radiation center frequency, bandwidth of the pump radiation, and the nonresonant susceptibility lead to a composite |
