Talking Turbines

Home / Talking Turbines

takntubThe wake of its progress will carry investment casting forward for many years to come. In fact, Howmet, PCC, GEAE, and Pratt & Whitney are already riding that wake in another agreement, along with the U.S. Air Force, Allison Engine Co. in Indianapolis, and Lockheed-Martin Aeronautical Systems in Marietta, Ga.

Started in 1995, it is called the Engine Supplier Base Initiative (ESBI), and it is planned to run six years. Part of ESBI’s objective is to efficiently utilize newly developed technology, such as that produced by ICCA, in a continuing campaign to strengthen the U.S. industrial base.

Processing simulation is seen by ICCA and ESBI as a precompetitive tool that will benefit the investment casting sector by enabling its participants to raise their level of collaboration. This means that competition continues as usual. While all will benefit to some extent from the tools, each is free to use them in their own way for their own purposes.

For example, the development and validation of a process simulation code requires that simulated output be compared with the actual behavior of castings. Such tests are carried out on production components cast at Howmet and PCC, and the results are shared with all of the program participants.

The details of the casting setup, the casting procedure, and the formulation of input to the simulation are not revealed. Such details for a specific production turbine blade are kept between the OEM (GEAE or Pratt & Whitney) and the foundry that casts it (Howmet or PCC). In this manner, proprietary information is handled in the same way that it was before ICCA.

To make a casting, a molten alloy is poured into a ceramic mold whose internal cavities form the shape of the desired part, and solidification is allowed to proceed as the metal cools down. Solidification begins in the coolest regions, usually at the mold walls, and the solid grows into the liquid as a more or less finely divided, periodic array of tree-like projections called dendrites.

The expanding dendrites consume the liquid as they lengthen, branch out, thicken, and run into one another. The feature size (the dendrite spacing) of the dendrite network in the finished casting depends on the solidification rate, but it is certainly microscopic (on the order of 100 [[micro]meter]), according to William Boettinger, a metallurgist at the National Institute for Standards and Technology (NIST), Gaithersburg, Md.

During solidification, three types of regions can be roughly defined: an expanding volume of continuous solid, a diminishing volume of continuous liquid, and a region between them that is a composite of dendrites and interdendritic liquid.

In the absence of special actions, a cast part will be polycrystalline, its composition highly nonuniform, and it will likely contain shrinkage porosity and other defects. Through special efforts it is possible to control the solidification and thereby exert an influence on the structure and properties of the resulting solid.

Turbine blades, for example, are frequently made to solidify as single crystals, because a single-crystal blade can more readily withstand the high temperatures in today’s gas turbine aircraft engines. Boettinger points out that the utility of a casting simulation is its capacity to predict how casting variables, such as gating design or solidification rate, will influence crystal structure and defect formation.

With such a tool, engineers can evaluate casting designs without wasting material, energy, and time on test castings.

The standard approach to casting simulation is based on the finite element method, whose origins lie in the aircraft industry of the early 1950s. In this method, a computer-based representation of the part to be cast is divided into a continuous network of solid sub-regions, typically tetrahedra.

These sub-regions are the so-called finite elements, the network is called a mesh, and the network junctions are called nodes.

Starting with a set of boundary conditions and initial conditions for the system, the quantity of interest (temperature, stress, or fluid velocity) is computed inside each element of the meshed casting model in such a way that the system of solutions is continuous at the nodes. The output is superimposed on the solid model in some graphical manner (color coding) so as to present a convenient visual display of the simulation.

Depending on the quantity or property to be simulated, a macro- or micromodeling regime is identified. Heat flow, fluid flow, and stress are subjects of the macromodeling regime, while crystal grain structure, nucleation, dendrite structure, and solute redistribution are issues in the micromodeling regime.

Of the two, macromodeling is more highly developed and has been put to use, while micromodeling is still preliminary.

Howmet’s Mueller points out that by the time you account for the metal casting, mold shell, and an inner core, you have 13 parameters for which you must have accurate values. These include the heat of fusion for the alloy, as well as the individual densities and specific heats of the alloy, shell, and core.

Through ICCA, an accurate database was created by developing good procedures to measure the needed thermophysical and thermomechanical constants.

Both Howmet and Precision Castparts use an FE code called ProCAST, from UES for their casting simulations.

UES was funded through ICCA to develop the appropriate coupling of three macrocodes and enhanced automatic meshing capabilities.

Simulations on the macrolevel can be put to good use in predicting the possible locations of defects. For example, simulated time/temperature profiles can predict the formation of isolated hot spots where late solidification might lead to shrinkage pores.

The alloys used for turbine blades (nickel-based superalloys) and other jet engine castings contain several alloying elements. In contrast to a pure, unalloyed metal, they solidify over a range of temperatures. The challenge of microstructure simulation starts with the determination of this range as a function of composition, as well as the amount of heat released by the freezing process.

In Boettinger’s words, “one must compute the enthalpy vs temperature curve during solidification. This can be done using a database that describes the underlying thermodynamics and phase diagrams.”

A large database is needed, and the microstructure simulation code uses this database, along with some coupling back to the thermal macrocode, to model such things as grain nucleation, dendrite formation and growth, and the concentration gradients produced in the vicinity of growing grains and/or dendrites.

In turn, notes Boettinger, such defects as second-phase grains and freckles can be predicted. The science of freckle formation has been especially well demonstrated through the utility of ICCA micromodeling results. The simulations clearly predict a process called thermal-solutal convection, which is said to be the cause of freckling.

There are two approaches to micromodeling – deterministic and stochastic. Deterministic models give the same answer every time and nucleate crystals on every node.

Stochastic models take the FE mesh, subdivide it, nucleate grains on these subdivisions on a random basis, then grow the grains at variable rates in a realistic manner. UES is working with both approaches.

ESBI is a joint U.S. Air Force/industry cooperative agreement that began in 1995 and is slated for six years. The Air Force Research Laboratory supplies 80% of the funding, and the program management team is led by Howmet.

What is truly unique about it is that the entire investment casting supply chain has now been brought to the table, bringing together suppliers, engine makers, an airframe builder, and the customer – the U.S. Air Force.

ESBI is structured on three areas of concentration:

* Technology processes focus on manufacturing technologies and methods to reduce process variability, cycle time, and waste.

* Business practices and policies focus on definition of a common culture, simplification of audits, and the standardization of various testing specifications.

* Electronic data utilization is similar in concept to the standardization that was set up under ICCA for engine OEMs to deliver CAD information to casting foundries, but it is more comprehensive.

Leave a Reply

Your email address will not be published. Required fields are marked *