Shaped Metal Deposition (SMD) is a near-net shape prototyping system patented by Rolls-Royce, licensed to the University of Sheffield and operated in collaboration with Footprint Tools.
The process allows complex parts to be built directly from the CAD model with minimum finishing. The system creates components from the base up in a layer-wise fashion, depositing weld material without the need for tooling. Complex parts can be made with improved material properties, and it is possible for hybrid components to be created. The main uses of SMD are one off parts, rapid prototyping, repair and complex or hybrid components.
State of the Art:
The SMD rig consists of a robot with a TIG (Tungsten Inert Gas) welding head and a manipulator, which are housed inside a sealed chamber. The chamber has two openings, one in the roof so that a crane can be used to load and unload large components, and an access door for entry into the chamber for maintenance or removal of smaller components. In addition to this, there is a vacuum pass box which is placed to one side of the glove chamber which allows consumables such as tungsten electrodes to be loaded into the cell without having to return the entire chamber to ambient. To isolate the SMD system from surrounding machine tool vibrations, pads are placed under the base of the cell; this eliminates deposition accuracy errors. Other than this, no special foundations are required and the system can be moved round the workshop as necessary. Electrostatic interference isolation is also used between the controller and the robot.
The system employs cold wire TIG deposition, using a tungsten cathode to weld the chosen material in an inert argon atmosphere to prevent the substrate, electrode and part from reacting with atmospheric gasses. This is a very stable process, chosen for the maturity of process control. The TIG welding head is attached to a 6-axis robot with a working envelop of 1.61m and a repeatability of 0.08 mm moving in concert with a 2-axis manipulator.
The gas purity in the chamber must be 99.999% before welding takes place. This typically takes 10 volume changes to achieve and must be maintained vduring the weld; a minimum consistent supply is essential for long builds. Once used the argon can be safely vented outside via an extraction system, or re-circulated via a scrubber system, which helps to fully utilise argon usage and further reduce costs.
For health and safety reasons, independent oxygen monitors are fitted inside the chamber; one at floor level and one near the roof. As well as monitoring the oxygen content during the welding process to ensure that desirable conditions are met, these also ensure that it is safe to enter the chamber after return-to-ambient, and that levels near the floor where argon collects are also safe. Real-time monitoring of the weld is possible through the use of a water-cooled vision system which allows the welding arc (electrode, contact tip) to be viewed (through filters) and the size of the bead and weld pool to be monitored. Thermal cameras can record the part temperature distribution.
The weld robot path is generated by reversing the NC milling path. Features can be built in any orientation, negating the need for support structures. The simulation software imports the output from the source programming package (ProEngineer), then checks that there are no interference problems or singularities, before outputting a part program that can be fed directly to the robot.
Enhancing the state of the art through modelling and control:
Accuracy in the part depends on the thermal stresses induced during the welding process. Controlling the heat transfer during deposition would reduce residual stress and hence the distortion. Process modelling can also be used to predict the expected distortion and to compensate during deposition. Factors such as the way that temperature changes with the changing deposition geometry must also be considered.
Wall thickness is controlled by the current, travel speed and wire feed rate and also to some extent by the wire thickness. Travel speed is the resultant of the rotation and tilt of the table and the movement of the robot. The faster the travel speed, the thinner the wall thickness, but the speed is limited to values where a consistent arc is produced. These parameters all need to be modelled and optimised using a mechatronic approach mixing mechanism simulation, FE structure analysis and control/command boxes so that a control system can be produced, freeing the operator from constant monitoring of the weld. If the correct parameters are used, the deposited components will be fully dense. Therefore, care must be taken over the choice of weld parameters, to prevent porosity due to contamination or to poor choice of wire feed rate, excessive speed, insufficient wetting, or high gas turbulence.
The resultant material properties are a function of the techniques used and depend on solidification kinetics and heating and cooling rates. The gradient properties produced when creating parts through hybrid methods or joining dissimilar materials also need to be investigated.
The robotic system used for SMD was modelled to allow system behaviour to be analysed and stiffness and precision to be improved by tuning the robot controller. Different modelling techniques (rigid, super-element and full finite element) were used to create a library of parts for the robot and a master model was created, which picks the most appropriate element type for the analysis.
It was found that a mixed model using rigid models for all components except the hand, arm and wrist which used super-elements and the link bar which was a full finite element model, allowed the dynamic behaviour of the system to be investigated whilst reducing computational time. This mechanical model was useful to analyse system behaviour and vibration modes.
Topology optimisation was then used to redesign the extension arm and reduce vibrations caused by low vibration frequencies. This led to reduced spatter during deposition
A mechatronic model was also developed which includes the force controllers for each joint and which was used to calculate the optimum robot control sequence for deposition. When creating parts, it is important that the tool-tip follows the trajectory smoothly. The joint positions created using the mechatronic model eliminate tool-tip oscillations and overshoot and reduce transient errors leading to a fast rise time