Advertisement
Articles
Advertisement

Modeling Processes Depends on Your Application

Fri, 10/25/2013 - 11:05am
Tim Studt

Plot of temperature in the symmetry section of a weld reveals that without any magnetic field (top) the weld takes on a wine glass shape, while with a magnetic field (bottom) shape forms a V. Image: ComsolThe modeling and simulation of various manufacturing processes is important because, in many cases, it’s impractical or even impossible to measure the specific operating parameters involved that contribute to the resulting products. This is particularly true in high-temperature processes like blast furnaces or the welding of large metal structures such as those used in shipbuilding and reactor vessels. Trial and error has been the traditional methodology used to create critical high-temperature processes in the past, which has been inexact at best and could be potentially dangerous to manufacturing personnel in the worst cases.

In blast furnace processes, the parameters that equipment designers have to contend with include molten metal temperatures near 1,500 C; varying metal consistencies; molten metal furnace runner systems consisting of a metal frame with multiple concrete liners and cast iron covers; substantial corrosion, erosion and thermal shock effects in the concrete portion of the runner systems; multiple fluid flow conditions including that from superheated air and the molten metal; cycling through varying length production runs and, of course, validation and verification of any software-based models. Simon Chiartano, a blast furnace equipment designer at Terres Refractaires Du Boulonnais (TRB), Nesles, France, developed a model of this system utilizing the Comsol Multiphysics system and its Heat Transfer Module.

As in all models of these types, certain initial conditions must be established for the individual components, along with a comprehensive set of boundary conditions and materials characteristics which are crucial to obtaining accurate results. Based on their previous experience, Chiartano’s design team established some modeling interactions to simplify the model without affecting the results. For example, they eliminated the effects of moving air in the runner system, which they calculated would not add significant information. For long-term process models, they also set the air temperatures to a value that they knew would be a stabilization value.

“There are some variations between the different runs of the molten metal due to the blast furnace process and also raw materials,” says Chiartano. “But these variations have only a small effect in our model because the boundary conditions chosen take into account that the molten metal will stay approximately the same.”

Plot of temperature in the symmetry section of a weld reveals that without any magnetic field (top) the weld takes on a wine glass shape, while with a magnetic field (bottom) shape forms a V. Image: ComsolTo verify their resulting model, the designers took thermal images of an actual cast iron runner roof and found that the temperature was in reasonably good agreement with the model. They then opened the runner roof and measured the temperature of the liner which, when taking into account the rapid temperature drops due to the opening, was also in good agreement with the model results, with an accurate calculation within 10 C. The resulting model also gave the designers considerable new understandings of what is going on within the roof runner system and realizing that their previous systems had been considerably over-designed, allowing new systems to be designed that are less expensive and more maneuverable.

In the welding of thick metal structures, design engineers have to contend with material distortion during the welding; stability of the weld surfaces; voids within the weld; non-uniformities within the weld due to differing materials and post-weld cooling rates. Marcel Bachmann at the BAM Federal Institute for Materials Research and Testing, Berlin, Germany, utilized the Comsol Multiphysics system to develop a magnetic field system that could be applied to a laser welding process to eliminate many of these welding inconsistencies and provide more uniform welds. In particular, they wanted to reduce the effects of the Marangoni effect where the welding high-temperature effects cause a flow of metal directed from the center of the weld toward the outer boundary due to temperature-dependent surface tension, rather than the desired flow into the depth of the weld cavity or keyhole.

“The simulations we created were steady-state with no temporal dependence on the welding result on the specified parameter set,” says Bachmann. “The parameters for numerical simulations are not associated with the welding process directly. Instead, especially in welding-based fluid flow simulations, some assumptions are chosen and lead to a numerical picture of the real process (i.e., the weld-based keyhole evolution in the real world is a highly dynamic process that definitely influences the flow field). In the simulations, we used a fixed cavity instead. We simplified the model until we reached a steady-state process that is much easier to solve numerically.”

The process for creating this model is different from the process created for Chiartano’s blast furnace application. “It always depends on what you want to simulate,” says Bachmann. “For real welding applications, the process becomes more unstable the higher the laser penetration depth is. Also, we found a strong dependence of the weld surface appearance on the used shielding gas supply, that cannot be predicted with our model as we do not consider the gas flow and assume flat surfaces. Our multiphysics model was more devoted to the interaction of a laser melt pool with external forces (i.e., magnetic fields). In other investigations, mechanical stresses are simulated that depend on the temperature field (i.e., blast furnace—these can be computed very well.”

Advertisement

Share This Story

X
You may login with either your assigned username or your e-mail address.
The password field is case sensitive.
Loading