The demands the automotive gear designer has to considerduring the gear design process have changed. To design a gearthat will not fail is still a challenging task, but now lownoise is also a main objective. Both customers and legalregulations demand noise reduction of gears. Moreover, thequality of the product is more in focus than ever before. Inaddition, the gear design process itself must be inexpensiveand quick. One can say that the gear designer faces a newdesign environment. The objective of this thesis is tocontribute to the answer to some of the questions raised inthis new design environment. In order to respond to the new design situation, the geardesigner must consider new phenomena of gears that werepreviously not a matter of concern. One such phenomenon is anew gear failure type, Tooth Interior Fatigue Fracture (TIFF).As the gear teeth are made more slender in an attempt to reducethe stiffness variation during the mesh cycle, therebypotentially reducing the noise, the risk of TIFF is increased.The phenomenon of TIFF is explored in detail (paper III-VI)through fractographic analysis, numerical crack initiationanalysis using FEM, determination of residual stress by meansof neutron diffraction measurements, testing for determiningmaterial fatigue properties, fracture mechanical FE-analysis,sensitivity analysis and the development of an engineeringdesign method. The main findings of the analysis of TIFF arethat TIFF cracks initiate in the tooth interior, TIFF occursmainly in case hardened idlers, the fracture surface has acharacteristic plateau at approximately the mid-height of thetooth and the risk of TIFF is more pronounced in slender gearteeth. Along with the more optimised gear design, there is atendency for the gear to be less robust. Low robustness, i.e.,great variation in performance of the product, implies a highincidence of rejects, malfunction and/or bad-will, all of whichmay have a negative effect on company earnings. As the use ofoptimisation decreases the safety margins, greater attentionhas to be paid to guaranteeing the products' robustness.Moreover, in order to be cost-effective, the qualities of thegear must be verified early in the design process, implying anextended use of simulations. In this thesis, two robustnessanalyses are presented in which the analysing tool issimulation. The first one considers robust tooth root bendingfatigue strength as the gear is exposed to mounting errors, thesecond one considers robust noise characteristics of a gearexposed to manufacturing errors, varying torque and wear. Bothof these case studies address the problem of robustness ofgears and demonstrate how it can be estimated by use ofsimulations. The main result from the former robustnessanalysis is that wide gears are more sensitive to mountingerrors, while the latter analysis showed that to achieve robustnoise characteristics of a gear it should have large helixangles, and some profile- and lead crowning should beintroduced. The transverse contact ratio is a trade-off factorin the sense that both low average noise levels and low scatterin noise due to perturbations cannot be achieved. <b>Keywords</b>: robust design, Taguchi method, gear, idler,simulations, Finite Element Method, Tooth Interior FatigueFracture, TIFF

<p>The demands the automotive gear designer has to considerduring the gear design process have changed. To design a gearthat will not fail is still a challenging task, but now lownoise is also a main objective. Both customers and legalregulations demand noise reduction of gears. Moreover, thequality of the product is more in focus than ever before. Inaddition, the gear design process itself must be inexpensiveand quick. One can say that the gear designer faces a newdesign environment. The objective of this thesis is tocontribute to the answer to some of the questions raised inthis new design environment.</p><p>In order to respond to the new design situation, the geardesigner must consider new phenomena of gears that werepreviously not a matter of concern. One such phenomenon is anew gear failure type, Tooth Interior Fatigue Fracture (TIFF).As the gear teeth are made more slender in an attempt to reducethe stiffness variation during the mesh cycle, therebypotentially reducing the noise, the risk of TIFF is increased.The phenomenon of TIFF is explored in detail (paper III-VI)through fractographic analysis, numerical crack initiationanalysis using FEM, determination of residual stress by meansof neutron diffraction measurements, testing for determiningmaterial fatigue properties, fracture mechanical FE-analysis,sensitivity analysis and the development of an engineeringdesign method. The main findings of the analysis of TIFF arethat TIFF cracks initiate in the tooth interior, TIFF occursmainly in case hardened idlers, the fracture surface has acharacteristic plateau at approximately the mid-height of thetooth and the risk of TIFF is more pronounced in slender gearteeth.</p><p>Along with the more optimised gear design, there is atendency for the gear to be less robust. Low robustness, i.e.,great variation in performance of the product, implies a highincidence of rejects, malfunction and/or bad-will, all of whichmay have a negative effect on company earnings. As the use ofoptimisation decreases the safety margins, greater attentionhas to be paid to guaranteeing the products' robustness.Moreover, in order to be cost-effective, the qualities of thegear must be verified early in the design process, implying anextended use of simulations. In this thesis, two robustnessanalyses are presented in which the analysing tool issimulation. The first one considers robust tooth root bendingfatigue strength as the gear is exposed to mounting errors, thesecond one considers robust noise characteristics of a gearexposed to manufacturing errors, varying torque and wear. Bothof these case studies address the problem of robustness ofgears and demonstrate how it can be estimated by use ofsimulations. The main result from the former robustnessanalysis is that wide gears are more sensitive to mountingerrors, while the latter analysis showed that to achieve robustnoise characteristics of a gear it should have large helixangles, and some profile- and lead crowning should beintroduced. The transverse contact ratio is a trade-off factorin the sense that both low average noise levels and low scatterin noise due to perturbations cannot be achieved.</p><p><b>Keywords</b>: robust design, Taguchi method, gear, idler,simulations, Finite Element Method, Tooth Interior FatigueFracture, TIFF</p>