Starting lost foam from scratch.

Inside This Story:

* A foundry in Ireland relates its experience in researching whether and how to start-up a lost foam casting operation.

* The basis for its research and comparison was a lost foam cast cylinder head, which was previously low pressure cast. Results on the optimized gating system and mechanical properties are shown.

**********

In 1999, Montupet UK, an aluminum foundry near Belfast, Northern Ireland started a lost foam research and development project with a $6 million budget and a team of 5 engineers and technicians. The goal was to acquire an in-depth understanding of the process and the possible defects associated with it to build generic rules for the design and production of lost foam parts.

Montupet, which produces cylinder heads and alloy wheels via gravity permanent mold has approximately 4000 employees in plants in France, Northern Ireland, Canada and Mexico. The Belfast plant has 1000 employee in 60,500-sq-m, and produces low pressure and gravity cast engine cylinder heads and wheels.

The decision to invest in lost foam came to Montupet following a successful similar venture in 1998 by Saturn and General Motors, who were working toward the production of lost foam automotive engine cylinder heads and blocks for Saturn. Similarly, Montupet hoped to integrate the constraints of the lost foam process with that of an automotive engine application, since the critical areas of the engine cylinder heads could be made more robust through the design flexibility offered by the lost foam process.

However, Montupet needed to convince potential customers that lost foam could he used to produce cylinder heads (especially in Europe, where the cylinder heads are more highly stressed) and that its foundry had enough knowledge and expertise to be a credible supplier for the component.

Montupet had previously conducted research programs on lost foam in partnership with automotive companies in order to compare the type of metallurgy that could be obtained with the process. From these programs, Montupet concluded that:

* high mechanical characteristics can be obtained, with sound components (no shrinkage or cold folds) after extensive optimization of the process;

* their properties are compatible with applications on cylinder heads, blocks and other similar applications;

* some contributions of the parameters were not understood well enough, particularly those that related to the foam. To master the quality on high volume production, in depth process understanding is required;

* a high number of parameters to be put under control exist at every step of the process;

* close interaction of the process parameters, cycle time and size of equipment was required for further study.

With these conclusions, Montupet decided to focus its research on the in-depth understanding of every step of the process. In order to accomplish this, the firm required the complete process in-house and needed to use production size machines to avoid imprecision. The fire1 also required investment in state-of-the-art equipment (including coating and pouring robots) to allow for precision and repeatability.

In 2000, Montupet installed the casting line and equipment. The foam preparation equipment included a pre-expander, foam molding machine, coating robot and drying oven, and foam assembly machine. The casting line consisted of a fully automated seven flask line (with a flask size of 40 x 30 x 36 in.) with horizontal vibration table, pouring robot and sand cooling and return system.

By 2001, the firm was moving full speed on the casting of prototype parts and gathering technical data for marketing purposes. Montupet started quoting lost foam components such as engine cylinder heads and alternator housings the next year, and has now obtained a contract to supply a water jacket outlet for Peugeot-Citroen. All together the research project, including the start-up of the line, cost $6.6 million.

Prototype Cylinder Head

Montupet chose a compact European style 4 cylinder head for automotive to serve as its prototype in experiments (Fig. 1). The cylinder head was similar to an existing low-pressure permanent mold 4 cylinder head already in production at Montupet, which gave the firm a benchmark of the metallurgy and performances that were needed for the highly stressed compact gasoline cylinder head for the European market.

The prototype head, which was redesigned with cast features allowed by the lost foam process such as complex shape internal passageways, weight savings and additional cooling, consists of a 4 slice lost foam pattern. The casting features blind holes 12-22 mm deep and 6 mm in diameter, with an as cast oil gallery 160 mm long, 8 mm diameter and featuring 3 cast in valve inserts with cooling on the intake side and exhaust side. The component also features a 20% weight savings over the permanent mold version. Two for casting alloys were tested--AS7U3 and AS5U3.

The firm wanted to maximize the part quality in the critical areas of the prototype head. Using a plan do control act (PDCA) type wheel (Fig. 2), the research focused on defect characterization using X-ray images, polished sections and dendrite arm spacing (DAS) analysis. At the same time, the casting was instrumented with mold fill probes (or thermocouples) in order to understand the physical mechanisms creating the defects to identify areas for possible improvement.

Montupet carried out several experiments on the prototype head, including:

* optimization of the density of fill/ non-fill areas for each slice to minimize the high density spot;

* different levels of bead fusion and the impact on shrinkage or leaks on the cast part;

* different grades of styrofoam beads;

* test of different coating with differing viscosities to optimize the casting metallurgy;

* test of different sand media, silica or synthetic as mullite to optimize the dimensional aspect;

* solidification simulation to identify defect areas;

* filling simulation to understand the fill pattern of the part and how cold folds could appear;

* vibration analysis of the flasks to understand the impact on dimensional stability.

Montupet also carried out a simultaneous development of coating and coating removal techniques, developing a quantitative measurement method for residual coating. Trials with coatings and beads were carried out with the suppliers' technical assistance.

Optimizing the Gating System

One example of Montupet's dual approach of defect analysis and part instrumentation is its optimization of the gating system. The firm developed the first version of the downsprue for the cylinder head with accelerator slots and in-gates with an 18-mm diameter. Shrinkage was observed on the flame face with an ATSM level of 3. Thermal instrumentation confirmed by initial simulation results lead the firm to believe that the small gates were freezing too early and that the flow of metal through these gates was insufficient to feed the bolts properly. A second version of the gating, which increased the diameter of the gating stubs from 18 to 26 mm (Fig. 3), was used to eliminate the shrinkage. The length of the in-gates also decreased by 10 mm.

Though the fill pattern did not change drastically, shortening the length of the gates allowed a faster fill. The increase of the gating stubs allowed hotter metal to feed through the gate and eliminated the shrinkage on the flame face. With this gating, the maximum shrinkage observed was approximately ASTM 2, which was located in non-critical areas of the prototype head. This shrinkage level was further reduced through coating and foam density optimization.

The filling pattern and the porosity observed at the end of the solidification was compared with the simulation results, so that Montupet could have an accurate prediction of the filling phase. In particular, the filling of the pattern needed to be accurately represented to identify possible locations of cold fold defects, in particular due to converging fronts. Montupet carried out a comparison of the filling and temperature maps observed on castings. The filling simulation identified that the metal fronts joining together had the potential to create cold fold defects at the bottom of the part. Subsequent analysis on the cast part showed brown marks indicating metal front converging in this area but no actual cold folds on the part.

The fill pattern observed on the simulation matched the test, although some discrepancies were observed on the overall part fill time and temperature at given thermocouple locations. The firm is currently carrying out more research to compare the simulation with instrumentation using different coating viscosities.

Material Properties

The gating optimization followed by the experimental trials on the coating and foam fusion allowed Montupet to reach the metallurgical properties as seen in Table 1. Montupet then tested a number of lost foam prototype heads with cast-in inserts on an internal thermal bench (Fig. 4) that mimics the hot and cold cycles of an engine test. While on the thermal bench, the prototype withstood 12,000 cycles equivalent to 1500 hr of engine high/low regime without any crack seen on the fire face. The process requires 1500 hr for a diesel cylinder head application.

Similarly, both the low pressure and prototype heads were tested on a hydraulic bench that would mimic the cycling pressure on the fire face. Both heads showed no cracks on the fire face after 120 hr of tests at a maximum pressure of 70 bars.

These tests showed that the metallurgical properties generated by the lost foam process, coupled with the design flexibility the process offers, could equal or surpass the results obtained on these benches with the low-pressure cast cylinder head.

Understanding Lost Foam

Compared to other processes, lost foam offers an increased design flexibility that can counteract the effects of a slower solidification due to the sand environment. Also, the tools are available to the designer to predict the metallurgical quality for a given geometry during the design phase using lost foam simulation software.

Montupet found that the optimized lost loam design (combined with an optimized metallurgy using a dual approach of instrumentation, simulation and trials/parts characterization) can produce a level of quality capable of withstanding highly stressed applications. Montupet obtained the same mechanical strength as a low-pressure cylinder head through lost foam casting.

Since these tests were carried ()Lit on the lost loam prototype head, Montupet has been able to:

* have generic design rules to integrate the advantages/constraints of this process, including the use of simulation software;

* design a generic development methodology and tools geared to obtain an in depth understanding of defect creation at each process step for any new part.

While Montupet is only now starting to obtain production orders for lost foam casting, the firm expects to see a return on investment in the near future as it markets the process to potential customers not aware of its benefits.

Table 1. The Average Mechanical Characteristics Obtained on the
Prototype Cylinder Headx

            Ultimate Tensile    Yield       Elasticity
               Strength        Strength

Fire Face       190 MPa        150 MPa    0.85% (average)
Pillars         170 MPa        140 MPa    0.55% (average)

For More Information

"Optimizing Mold Making for Lost Foam Casting, "J. Bast, W. Hopf, L. Sacharuk and T. Hahn, 2003 AFS Transactions (03-161).

"Practical Control of Lost Foam Coatings in Foundry Operations," J.S. Parish, D.J, Jason and J.L. Meloni, 2003 AFS Transactions (03-153).

D. Boudignon is the R&D manager for Montupet UK, Belfast. He has been working on the lest foam protect since 1998.