There is continued interest in nonlinear devices for different types of optical signal processing, such as Raman or parametric amplifiers. The small nonlinearity of conventional single-mode fibers sets a major limitation for these devices. A large nonlinearity can be achieved by having a large nonlinear coefficient, a small effective area, or both. Having a small effective area, however, requires efficient coupling to very small core fibers. A novel technique for splicing conventional single-mode fibers to small core fibers is proposed and demonstrated. The coupling efficiency obtained by this technique is considerably improved over that obtained by the butt-joint splice. This technique uses a highly tapered splice in which the field leaves the core and propagates as a fundamental cladding mode before it couples back to the core mode of the small core fiber. At the beginning of the taper the fundamental core mode carries most of the power. Over the down-taper region, the core mode couples to the fundamental cladding mode for which the cladding-air interface plays a major role in guiding the light. Over the up-taper region, the cladding mode is coupled back to the core mode. Fabrication of such a device involves many constraints. Alignment of the cores, the slope of the taper, and the taper length are important issues to ensure that excessive radiation loss does not take place. The theory of tapered single-mode fiber is discussed including adiabaticity criteria, length considerations, mode coupling and wavelength dependence. We use a computational simulation to examine how the field changes from one part of the taper to the other. Variations of the fiber and the field properties along the taper are studied. In this simulation, the tapered region is approximated as a sufficiently large number of cascaded uniform fiber segments of decreasing or increasing diameters. Another analysis based on the conservation of power flow is also provided. Tapered splices were fabricated using two different experimental setups. The experimental setup to verify our theoretical results is shown. The tapering process is thoroughly discussed. The spectrum of a tunable laser passing through a splice shows how modes interact with each other during the tapering process. We successfully fabricated very low loss tapers with extremely small diameters. Tapered splices showed a lower loss than their butt-joint counterparts. Experimental measurements of these tapered splices are presented and discussed. / Ph. D.

This thesis presents the results for measured and predicted rotordynamic coefficients and leakage for a convergent-tapered honeycomb seal (CTHC). The test seals had a diameter of 114.968 mm (4.5263 in) at the entrance, and a diameter of 114.709 mm (4.5161 in) at the exit. The honeycomb cell depth was 3.175 mm (0.125 in), and the cell width was 0.79 mm (0.0311 in). Measurements are reported with air as the test fluid at three different speeds: 10,200, 15,200, and 20,200 rpm; with a supply pressure of 69 bar (1,000 psi), with exit-to-inlet pressure ratios from 20% to 50%, and using two rotors that are 114.3 mm (4.500 in) and 114.5 mm (4.508 in) respectively; this enables the same seals to be tested under two different conditions. The q factor, which is just a simple way to quantify taper is defined as the taperangle seal parameter and is calculated using the inlet and exit radial clearance. Two taper-angles parameters were calculated; q = 0.24 for the 114.3 mm (4.500 in) rotor, and q = 0.386 for the 114.5 mm (4.508 in) rotor. The q = 0.24 condition was compared to a constant clearance honeycomb seal (CCHC q = 0) because both sets of data were taken with the same rotor diameter. The direct stiffness, effective stiffness, and direct damping coefficients were larger for q = 0.24. The CTHC q = 0.24 eliminates the direct negative static stiffness obtained with CCHC ( q = 0). The cross-coupled stiffness and damping also were larger for q = 0.24, especially at low frequencies. Effective damping is one of the best indicators in determining the stability of a roughened stator annular gas seal. The frequency at which it changes sign is called the cross-over frequency. In applications, this frequency needs to be lower than the rotorsystem’s first natural frequency. Otherwise, the seal will be highly destabilizing instead of highly stabilizing. The magnitude of effective damping and the cross-over frequency also increases with q for all frequencies. Constant clearance honeycomb seals have less leakage than convergenttapered honeycomb seals. CTHC ( q = 0.24), has approximately 20 percent more leakage than CCHC ( q = 0). The experimental results for rotordynamic characteristics and leakage were compared to theoretical predictions by the two-control-volume developed by Kleynhans and Childs. All rotordynamic coefficients were reasonably predicted for all cases. The model does a better job predicting the cross-coupled stiffness and damping coefficients rather than the direct stiffness and damping coefficients. Also, the two-control-volume model predicts the dynamic characteristics of CCHC ( q = 0) better, and does not predict well the effective stiffness and damping for CTHC q = 0.386.

Tapered waveguides are often used as spot-size converters and power dividers. In general, radiation in these devices is hard to analyze by either BPM or FD-TD methods. We apply the full eigenmode expansion technique (FEMET) to study the propagation of an adiabatic dielectric tapered waveguide. We often assume that wave propagation in an ideal ¡§adiabatic¡¨ dielectric waveguide suffer no reflection nor radiation loss. It is especially true for the fundamental mode propagation in an ¡§adiabatic¡¨ dielectric waveguide. Thus, the fundamental mode of the input waveguides will be converted to the corresponding fundamental mode of the output waveguide whenever the two are connected by an adiabatically tapered waveguide. However, the higher-order modes do not always propagate through the tapered waveguide when the output waveguide does not support that particular guiding mode. It is interesting to predict such radiation phenomena and to observe them in a numerical experiment. In this thesis we consider the titled straight waveguide (TSR) as our test example. Since TSR has an exact solution in its natural coordinate system, we can study computational characteristics of FEMET via TSR examples. Using FEMET and FD-FD method, we carefully examine mode evolution, conversion radiation and reflection of many quasi-adiabatic tapered waveguides. Finally the apparent visual radiation angles are defined and computed as function of taper angle core/cladding indices and incident mode order for both TE and TM mode cases.

Predictions are presented for a selected compressor using longer hole-pattern seals with L/D ratios from 0.5 to 2.5. Results were obtained for back-to-back and in-line compressors with the seal located at mid-span and at 82% of rotor span respectively, considering different seal lengths, radial seal clearances, as well as constant clearance and convergent-tapered seal geometries. Predictions of the synchronous rotordynamic coefficients and leakage were estimated using a code developed by Kleynhans and Childs with zero preswirl and constant pressure ratio of 0.5. This code does not include moment coefficients; which can affect the results. Results of all configurations show an increase of stiffness and damping coefficients with increasing seal length. In addition, a significant reduction in leakage (approximately 47 percent) as L/D increases is exhibited for constant clearance and convergent-tapered hole-pattern seals. For the back-to-back compressor, the stability analysis predicts that the system is stable for all speeds and L/D ratios. In fact, the rotor cylindrical-bending mode becomes more stable with lengthening the seals, for both constant clearance and convergent-tapered hole-pattern seals. For constant clearance seals (Case A), the synchronous response at mid-span show a critical speed at 8,000 rpm (cylindrical-bending mode) for all L/D ratios, while a reduction of 85 percent in the peak response is exhibited as L/D increases. Case B, in which the radial clearance is increased as L/D increases to have the same leakage as case A, slightly increases the synchronous response of the model compared to case A. For convergent-tapered seals (Case C), the synchronous response at mid-span shows a higher critical speed (9,000 rpm) for all L/D ratios, and a larger reduction (89 percent) in peak response with increasing L/D, compared to Case A. However, the magnitude of the peak response is larger for convergent-tapered seals than that for constant clearance seals, for all L/D ratios. For in-line compressor, the stability analysis predicts two critical speeds at 6,000 (conical mode) and 18,000 rpm (first bending mode) respectively. Both modes are predicted to be stable for all speed and L/D ratios. Synchronous response at the mid-span for Case A shows the peak response at the first critical speed is slightly reduced as L/D increases while the response at the second critical speed is increased for most of the cases. In addition, the second critical speed is reduced from 18,000 to 13,000 rpm, which is not a concern because it remains above the running speed. This was also the trend for convergent-tapered hole-pattern seal. In addition, the increase of radial clearance in Case B slightly increases the amplitude of vibration, compared to Case A.

Predictions are presented for a selected compressor using longer hole-pattern seals with L/D ratios from 0.5 to 2.5. Results were obtained for back-to-back and in-line compressors with the seal located at mid-span and at 82% of rotor span respectively, considering different seal lengths, radial seal clearances, as well as constant clearance and convergent-tapered seal geometries. Predictions of the synchronous rotordynamic coefficients and leakage were estimated using a code developed by Kleynhans and Childs with zero preswirl and constant pressure ratio of 0.5. This code does not include moment coefficients; which can affect the results. Results of all configurations show an increase of stiffness and damping coefficients with increasing seal length. In addition, a significant reduction in leakage (approximately 47 percent) as L/D increases is exhibited for constant clearance and convergent-tapered hole-pattern seals. For the back-to-back compressor, the stability analysis predicts that the system is stable for all speeds and L/D ratios. In fact, the rotor cylindrical-bending mode becomes more stable with lengthening the seals, for both constant clearance and convergent-tapered hole-pattern seals. For constant clearance seals (Case A), the synchronous response at mid-span show a critical speed at 8,000 rpm (cylindrical-bending mode) for all L/D ratios, while a reduction of 85 percent in the peak response is exhibited as L/D increases. Case B, in which the radial clearance is increased as L/D increases to have the same leakage as case A, slightly increases the synchronous response of the model compared to case A. For convergent-tapered seals (Case C), the synchronous response at mid-span shows a higher critical speed (9,000 rpm) for all L/D ratios, and a larger reduction (89 percent) in peak response with increasing L/D, compared to Case A. However, the magnitude of the peak response is larger for convergent-tapered seals than that for constant clearance seals, for all L/D ratios. For in-line compressor, the stability analysis predicts two critical speeds at 6,000 (conical mode) and 18,000 rpm (first bending mode) respectively. Both modes are predicted to be stable for all speed and L/D ratios. Synchronous response at the mid-span for Case A shows the peak response at the first critical speed is slightly reduced as L/D increases while the response at the second critical speed is increased for most of the cases. In addition, the second critical speed is reduced from 18,000 to 13,000 rpm, which is not a concern because it remains above the running speed. This was also the trend for convergent-tapered hole-pattern seal. In addition, the increase of radial clearance in Case B slightly increases the amplitude of vibration, compared to Case A.

No description available.

No description available.

Current procedures for estimating effective length factors for web-tapered members rely heavily on the use of charts and graphs. This makes them difficult to implement using a computer. In addition, they are often based on unrealistic assumptions. In cases where these assumptions are not satisfied, design errors may result. This investigation proposes a modification to an effective length factor expression developed by Lui (1992). This modification allows the expression to be applied to web-tapered members with good accuracy. A derivation of the proposed expression is presented, and the results obtained by applying the expression to a range of frames are compared to the results obtained from second-order finite element analyses. Calculations involved in using the expression are presented. / Master of Science

Tapered beams and columns are often used in single story gable framed steel buildings for reasons of economy. By varying the resistance to bending in similar proportion to the bending moments, more economical structures can be obtained. The beam and column connection, or knee area, is generally subject to the greatest bending moments. It is therefore comprised of the deepest sections of the tapered members, which also possess the least resistance to shear buckling. The web element's stress distribution in this region of relatively complicated geometry is unknown. For this reason, web stiffening plates are commonly used to brace the slender web elements against elastic shear buckling. The design and proper installation of these relatively small elements, while proven to be effective, is also costly. Because it is desirable to remove the stiffeners, the shear behavior of unstiffened tapered members near the moment connection was the primary focus of this study. Four knee area specimens were tested to failure under simulated gravity load conditions. The specimens were analyzed according to the AISC shear provisions for prismatic members. The appropriateness of a modified shear force, which accounts for the influence of inclined flanges, and the role of initial web imperfections were examined as well. Finally an analysis method which most consistently produces conservative results is proposed. / Master of Science

Abstract Optical fiber communications have been become one of the most popular researches since 1970s. In this field, there are many studies on the coupling between semiconductor lasers and fibers and many conclusions are demonstrated. In this thesis we build a 2-D numerical model to simulate energy coupling between laser diodes (LDs) and tapered optical fibers. Our model is based on the spectral domain integral equation (SDIE) formulation which is derived from Maxwell equations and the principle of mode matching. Through this numerical model we will be able to show the field distribution in LD waveguide junctions. We may also use this tool to study the coupling parameters such as the separation distance and tapered fiber geometry.