Weld quality in the context of process control and internal conditions is studied. Several different alloys are welded including plain carbon steel, high-temperature steels, and several traditional and advanced superalloys. Across all studied weld systems, the following conditions led to stronger welds: higher forces and feedrates, lower temperatures, and moderate or limited upsets. In the best cases, post-weld strengths were nearly equal to basemetal strength. Tradition holds that large and symmetric upsets are necessary for good welds, but this study contradicts that notion. The fundamental requirements for bonding are two sufficiently clean surfaces in intimate contact. Only minimal upset is necessary to achieve that. In welding alloy 718, only 1 mm of feed (or ~0.4 mm of sample upset) was necessary to achieve $>$95\% of basemetal strength. In an advanced superalloy with low ductility, very low upsets were required in order to achieve high joint strength. For this low-ductility alloy, using a containing geometry increased both the internal pressure and ductility of this alloy, leading to a much larger window of sound welding conditions and stronger welds overall. In several dissimilar alloy systems, the relationship between force/feedrate and upset asymmetry varied between each alloy, but a more symmetric upset did not correlate to stronger welds. Advanced process control in FW was also performed with closed-loop temperature control and open-loop predictive cooling rate control. Using this technique, martensitic microstructures associated with a fast natural cooling rate were avoided, and a pearlitic microstructure was obtained. The yield and tensile strength of the weld was not adversely affected, and both were within range of published values for the basemetal.

This study develops a method to simulate a full-scale inertia friction weld with a sub-scale specimen and modifies a direct drive friction stir welder to perform the welding process. A torque meter is fabricated for the FSW machine to measure weld torque. Machine controls are modified to enable a force control during the IFW process. An equation is created to measure weld upset due to deflection of the FSW machine. Data obtained from a full-scale inertia friction weld are altered to account for the geometrical differences between the sub-scale and full-scale specimens. The IFW are simulated with the sub-scale specimen while controlling spindle RPM and matching weld power or weld RPM. The force used to perform friction welding is scaled to different values accounting for specimen size to determine the effects on output parameters including: HAZ, upset, RPM, torque, power and energy of the weld. Increasing force has positive effects to upset, torque, power and energy of the welds, while reducing the size of the HAZ.