Development of numerical tool for hybrid manufacturing process for titanium sheet metal forming

The use of tanium in the aerospace industry has grown considerably in recent years in conjunc on with the development of composite aircra�. In this way, improving tanium forming has become an important issue for the industry, both for produc vity objec ves and the ability to deliver basic parts according to the needs imposed by aircra� delivery rates, as well as for cost objec ves. Currently, hot forming of tanium parts can be achieved through two processes: Super-plas c forming (SPF) or Hot Forming (HF). The aeronau cal industry wanted to develop an innova ve process for the manufacture of tanium parts by coupling the HF and SPF processes in order to exploit the advantages of these two technologies. The development of a mixed HF / SPF process will thus not only improve the rates and allow be�er control of the quality of the formed parts (thickness homogeneity), but also, by allowing forming at lower temperatures, this hybrid process presents a large interest at the energy plan. The study was devoted to the development of a hybrid HF/SPF process, carried out at a common temperature, allowing the "pre-forming" of the part in HF mode and the "calibra on" of the part in SPF mode, while respec ng a global cycle me compa ble with the objec ves of the aerospace industry and guaranteeing the quality expected for the final complex part. Improving the performance of the final part requires a development of numerical simula on tool of the forming process. The available simula on tool (ABAQUS/ Standard) must be adapted to define the best simula on strategy according to the simulated parts; moreover, it remains impera ve to determine the input data (material behavior laws of tanium alloys) adapted to the cases to be treated (strain rate and process temperature).

Note: for confiden ality reasons, some informa on will not be communicated.

Super Plas c Forming (SPF):
Superplas city is a plas c deforma on property of a material which results in very high elonga ons at break (typically several hundred percent). This phenomenon occurs at temperatures typically higher or equal to half the absolute mel ng temperature, for some materials only, and for rather low strain rates only.
Industrially, the so-called superplas c process most o�en consists in applying a gas pressure on a blank blocked at its periphery and heated up to a temperature suitable for superplas city. The sheet is then gradually deformed un l it fits the mold. The presence of blank holders preven ng slipping of the sheet, the elonga on also causes a thinning of the part, especially during the free bulging stage.
The main advantage of superplas c forming is the ability to fabricate parts having complex geometries that are impossible to produce with other forming processes. These parts therefore have a high added value and are currently des ned mainly to the avia on industry. However, this method has the disadvantage of presen ng long cycle mes due to the low strain rates required (typically between thirty minutes and two hours to produce a part). Therefore, it is impossible to use it in the context of mass produc on. A second disadvantage is the elevated costs resul ng from these high temperatures needing to be applied for long mes, which demands a lot of energy.

Hot Forming (HF):
At the opposite, the Hot Forming process, which consists in deforming a few millimeters thick sheet metal using a punch, can be used to produce deformed parts in much shorter mes and/or at lower temperatures. It can be related to hot deep drawing in many ways. Indeed, as in hot deep drawing, the sheet metal is forced to adopt the shape of a matrix by the ver cal pressure of a punch. However, the temperature is raised beyond the range of standard drawing, between the third and the half of the absolute mel ng temperature. Furthermore, the HF method is isothermal. That means the tools are heated con nuously at the same temperature as the work piece, which is usually not the case in hot deep drawing. In contrast to deep drawing, there are generally no blank holders in the Hot Forming process. This lack of posi on constraints prevents the blank sheet from stretching, which maintains an almost constant thickness throughout the part.
The main advantage of Hot Forming is a short cycle me (a few minutes) related to higher strain rates than in the case of SPF. However, the complexity of achievable geometries is very limited, with rather small strains due to the low duc lity of the material under such temperatures and strain rates.

Hybrid Process (HF+SPF):
Today, increasing produc on rates requires to produce complex parts with short cycle mes. That is why a process combining HF and SPF is being considered. This process is thus divided into two separate steps. The fast performing, comparable to HF, is performed by the movement of a punch, without a matrix but with blank holders. It is immediately followed by the second step of blow forming like in superplas c forming [1] (Figure 1). The hybrid process allows combining the advantages of HF and SPF processes. Indeed, the preforming step significantly reduces the total cycle me compared to gas forming only, while the SPF comple on allows achieving more complex parts than HF.
The few reported studies on the hybrid process [2,3] have demonstrated the ability to produce successfully parts of aluminum and tanium alloys, at temperatures lower than those usually used in SPF. In addi on, these parts were all formed with a significantly reduced cycle me compared to SPF. Forming cycles, which typically last between 60 and 120 minutes in SPF, were achieved within reduced me cycle. Our aim is here to characterize the influence of the fast HF preforming on the SPF step, with a more systema c inves ga on of the associated metallurgical aspects.

Material modelling:
Tes ng: Hot-forming condi ons: The tension tests have been performed in Arts et Mé ers ParisTech, using A GLEEBLE. The specimen is heated to the test temperature with a hea ng speed of 10 ° C/s, followed by a stable phase of one minute to obtain a nearly homogeneous temperature in the central zone of the specimen (Figure 2). It is then deformed un l it broke. Finally, a cooling of the specimen was done with a speed of about 50 °C /s). The specimen is a�ached through 2 copper jaws ( Figure 2). A type K thermocouple is welded to the center of the specimen to control the temperature in its center. An extensometer is mounted at the most heated zone to follow the movement and to control the rate of deforma on there.
Super Plas c forming condi ons: The tension tests were carried out at Arts & Me ers ParisTech laboratory using a horizontal tensile test machine (figure 3) equipped with a tube furnace (four halogen lamps), an argon flow, a K-type thermocouple welded in the center of the specimen, a displacement sensor and a force sensor (1000 N).

Iden fica on of behavior law:
In order to be able to simulate the mixed HF / SPF forming in a single step, it was necessary to iden fy a behavior law that covered the en re HF + SPF domains; it is called "the unified law".

Principle of iden fica on:
The law chosen is that of Norton-Hoff (N-H): (Eq. 1) The equivalent law in ABAQUS \ Implicit is the CREEP law whose expression is as follows [4]: Where is the uniaxial equivalent creep strain rate, is the uniaxial equivalent deviatoric stress and is the equivalent creep strain.
A, n and m are the material parameters. The reverse method was used for parameters iden fica on. Figure 4 shows the comparison between experimental and analy cal curves for the different strain rates and for the two steps (HF + SPF). Generally, the analy cal iden fica on of the parameters presents a sa sfactory level of confidence with respect to the experiment.  For high strain rates, we find a peak of stress on the experimental curves at about 10% of strain. This phenomenon is linked to microstructural changes at the grain level. The difference between the analytical model and the experimental test is maximal for the highest strain rate of .

Process simula on:
Numerical simula on condi ons: For confiden al reasons, the tooling used for the results valida on is not shown in this study. The die, the punch and the blank holder are considered as rigid bodies. The blank is modeled as a deformable body with using quadra c shell elements. The chosen element size is 10 mm. This value has been op mized a�er several calcula ons taking into account the calcula on me and the accuracy of the results.
For HF phase, the clamping is applied by blank holder and the stamping is carried out by punch displacement. For the SPF phase, the op mized pressure law is presented by the graph in the figure 5.
The fric on coefficient is 0,15. The model is composed of three main steps: Step 1: apply a clamping force; Step 2: perform the HF phase with applying the displacement of the punch; Step 3: block the movement of the punch; apply embedding of the nodes at the gasket level; apply the pressure cycle to perform the SPF phase.

Results:
Thickness distribu on: The measurement of the thickness at different points of the part indicates that the thinning is very homogeneous throughout the useful part zone with an average value of 10%. Figure 6 shows the numerical predic on of the thickness at the end of the HF phase. A thinning of 10% can be seen. At the end of the SPF phase (figure 7), the thinning reaches its maximum with a local value of 21% at the corners of the part whose material is strongly deformed. On the rest of the useful area of the part, the overall maximum thinning is about 16%.
These numerical predic ons have been verified experimentally through several tests and it has been found that the maximum overall difference is about 5%. This has been considered sa sfactory for industrial applica on.