5.5_Niyama.pdf

Round Robin Study to Assess Variations in Casting

Simulation Niyama Criterion Predictions

Neelesh Jain, Kent D. Carlson and Christoph Beckermann

Department of Mechanical and Industrial Engineering

The University of Iowa, Iowa City, IA 52242

Abstract

The Niyama criterion, a local thermal parameter that is a common output of casting

simulation software packages, is frequently used to predict shrinkage porosity defects in steel

castings. Previous studies indicate that it is a robust parameter that not only predicts the macro-

shrinkage that is visible on radiographs, but also smaller micro-porosity that is usually not

detectable using standard radiographic techniques. The implication of this previous work is that

the Niyama criterion values from a casting simulation may be used not only to provide guidance

in designing shrinkage-free steel castings, but also as a quality measure in a purchase

specification. Before the Niyama criterion can be used in this manner, it is important to establish

a method that assures that the Niyama values are predicted in a reliable and reproducible way

that does not depend on the casting simulation software itself, or on its internal or user settings.

The objective of the present study is to assess variations in Niyama predictions among various

casting simulation software packages and users, for a given casting. Fifteen SFSA member

foundries simulated the solidification of a common casting geometry, using up to four cast alloys

(WCB, CF-8M, CN-7M and M-30C). Niyama criterion results produced from these simulations

are evaluated and compared, and the causes of the sometimes significant variations among the

Niyama values are analyzed. The sensitivity of the Niyama predictions to numerical grids,

location within the casting, variations in the Niyama evaluation temperature, and differences in

thermophysical property data are investigated. In addition, the results from two common casting

simulation software packages are compared using, as much as possible, the same properties and

settings. From the various investigations performed in this study, it is ascertained that differences

in the thermophysical properties of the metal alloy used in the simulation most significantly

affect the Niyama predictions. Therefore, if the Niyama criterion were to be used in a purchase

specification, it must be ensured that a “good” property dataset is used.

1. Introduction

The Niyama criterion, defined as the local thermal gradient divided by the square root of the

cooling rate (i.e.,

Ny = ), is a commonly used output variable in casting simulation

software packages to predict shrinkage porosity defects in steel castings. Shrinkage porosity is

likely to occur if the calculated Niyama value is below a certain critical value. Previous studies [1-

7] by the University of Iowa have shown that feeding distances, for example, can be reliably

predicted using the Niyama criterion. Case studies [8] have indicated that the Niyama criterion

also correlates with the occurrence of some leakage defects in fluid containing low- and high-

alloy castings. Evidence exists [8,9] that the Niyama criterion not only predicts the macro-

shrinkage that is visible on radiographs, but also smaller micro-porosity that is usually not

detectable on standard radiographs used in the foundry industry.

This previous work indicates that the Niyama criterion values from a casting simulation may

be used to not only provide guidance in designing shrinkage-free steel castings, but also as a

quality measure in a purchase specification (in addition to other specifications). By requiring the

Niyama values in an area of a casting to be above a certain critical value, the absence of

shrinkage porosity could perhaps be assured. This would not be unlike setting an ASTM standard

x-ray level requirement for a casting.

Before the Niyama criterion can be used in this manner, it is important to establish a method

that assures that the Niyama values are predicted in a reliable and reproducible way that does not

depend on the casting simulation software itself or its internal or user settings. Ideally, for the

same casting alloy, geometry and process, the same Niyama values should be predicted.

Unfortunately, unlike simple physical measurements (such as temperature), casting simulation is

a complex process that requires much user input. Some of the critical issues in the prediction of

the Niyama criterion value are: (i) some software may provide inaccurate predictions due to the

nature of the numerical approximations made internally; (ii) different software may evaluate the

Niyama criterion differently (e.g., units used, temperatures at which the thermal gradient and

cooling rate are evaluated); (iii) the thermophysical properties of an alloy/mold material, which

are needed in a casting simulation, may not be well established; if different properties are used,

the predicted temperatures and, hence, the Niyama criterion values will be different; (iv) the

casting and boundary conditions may not be accurately known or input (pouring temperature,

mold/metal interfacial heat transfer coefficient, ambient heat transfer, etc.); (v) the choice of the

numerical grid and time steps will depend on the software user and available computing power.

The objective of the study is to conduct a round robin testing program among SFSA steel

foundries to assess the variations in the Niyama predictions among various casting simulation

software packages and users, for a given casting. The study includes four different cast alloys:

WCB, CF-8M, CN-7M and M-30C.

This article details the round robin study procedures, and then presents the results obtained

from the study. For each of the four alloys, the results are compared in order to assess the

variations in the Niyama predictions among the various software packages, as well as among

various users of the same package. In addition, a section that analyzes the sensitivity of Niyama

predictions to position, number of metal cells and Niyama evaluation temperature is also

included. Finally, in order to assess the importance of the modeling and numerical

approximations inherent in casting simulation software, Niyama predictions from different

simulation packages using a common material property steel dataset are compared and evaluated.

Conclusions and recommendations are made from the various investigations and findings in the

study and presented in the last section of this article.

2. Round Robin Test Procedures

Simulation Procedure

This section explains the procedure followed by the participants of the round robin study.

The procedure detailed here was explained in a description sheet, which was provided to all

round robin study participants. The valve geometry shown in figure 1 was selected as the casting

to be used by all participants in this study. To provide a sense of scale, the diameters of the valve

flanges are about 23 inches, and the diameter of the riser is about 9 inches. CAD files for the

mold box, riser and valve geometry were created and provided to all participants for use in their

simulations, in order to ensure that all participants began with the same geometry. It was decided

that participants would perform a solidification-only simulation for this study (i.e., no filling

simulation), in order to remove any variation caused by differences among filling simulations.

588 mm

223 mm

606 mm

(b)

(a)

(c)

(d)

Figure 1. Views of the valve geometry

The participating organizations were instructed to use pep-set sand mold properties,

assuming the use of hot topping on the riser. They were asked to conduct the simulations on as

many of the following four metal alloys as they had material property data for: WCB, CF-8M,

CN-7M and M-30C. The following table lists the superheat values (difference between the initial

metal temperature specified in the simulation and the liquidus temperature of the alloy) that the

participants were instructed to use while conducting the simulations.

Table 1. Superheat values for the alloys included in the study

Alloy

Superheat value (°C)

WCB

CF-8M

100

CN-7M

100

M-30C

100

Preparation of the Results

The results were collected in the form of Niyama contour plots given at cross-sectional slices

(in the XZ-plane, see figure 1c) at three locations in the valve, as shown in figure 2. In order to

produce the plots, the participants were asked to use their best judgment in choosing a scale for

the Niyama plots.

Position A (Mid-Flange)

Position B (Mid-Body)

Position C (Mid-Flange)

Figure 2. Slice locations at which the results were requested

In addition to the preparation of the Niyama plots, the participants documented necessary

simulation details in a simulation data sheet that was provided to them. The details included the

name of the simulation package used, number of metal cells or metal elements used for the

simulations, and the units of the Niyama criterion in the simulation package used. Also, for each

alloy simulated, participants provided the values of the liquidus temperature ( T ), the

temperature at which the metal is 100% solid ( ), and either the temperature at which the

Niyama criterion was evaluated ( T ) or where in the solidification range ( ) the Niyama

criterion was evaluated (for example, 10% of the solidification range above the temperature at

which the metal is 100% solid).

liq

sol

T −

liq

sol

3. Niyama Plots from The University of Iowa

This section displays Niyama plots at the slices indicated in figure 2, for all four alloys,

obtained using MagmaSoft at The University of Iowa. Table 2 shows characteristic temperature

values that were used in the simulation of each alloy. The first three columns are taken from

material datasets, while the fourth is a user-specified value. The WCB dataset represented in

table 2 is the standard MagmaSoft WCB dataset (named ‘GS24Mn4’). The CF-8M dataset was

developed by the present researchers using the software package IDS, developed by Miettinen et

al ., [10,11] which simulates the microsegregation and phase transformations that occur during low-

alloy and stainless steel solidification. IDS also calculates all of the other material properties

required as input for casting simulation software (density, thermal conductivity, etc.). The CN-

7M and M-30C datasets were developed by the present researchers using the software package

JMatPro, [12] which calculates the solidification path and all casting-relevant material properties

for a given alloy composition, using thermodynamic databases for certain classes of alloys.

Experimental temperature measurement data for these alloys, indicating liquidus and 100% solid

temperatures, was used to fine-tune parameter settings in the JMatPro simulations. [5,7] The

Niyama criterion evaluation temperatures listed in the last column of table 2 are the MagmaSoft

default values for each alloy, where the default Niyama evaluation temperature is 10% of the

solidification interval above the 100% solid temperature [i.e.,

(

)

−

sol

liq

sol

Table 2. Temperature values used by The University of Iowa

Liquidus

Temperature,

(°C)

Solidification

Range,

Niyama Evaluation

Temperature,

(°C)

100% Solid

Temperature,

(°C)

Metal

Alloy

(

)

T −

(°C)

liq

sol

WCB

1519

1412

107

1423

CF-8M

1430

1320

110

1331

CN-7M

1393

1300

1309

M-30C

1303

1193

110

1204

For The University of Iowa simulations, all four of the alloys listed in table 2 were paired

with the sand database ‘FURAN’ from MagmaSoft, which models a resin-bonded silica sand.

The heat transfer coefficient specified between alloy and mold for all alloys was ‘C800’ from the

MagmaSoft database, which is a constant heat transfer coefficient of 800 W/m 2 -K. All four

simulations were based on the same numerical grid, which was generated using MagmaSoft’s

automatic grid generation with approximately 2 million control volumes. This produced a mesh

that contains a total of 1,956,930 computational cells, of which 439,039 are cells located in the

casting or riser (‘metal cells’) and the remainder of the cells are located in the mold (‘mold

cells’). The initial temperatures used for each alloy were determined by adding the superheat

values listed in table 1 to the liquidus temperatures listed in table 2.

Niyama Plots for WCB

Figures 3a to 3c show Niyama plots for WCB. These plots were obtained at the cross-

sectional locations (“slices”) indicated in figure 2, in the center of each flange and the center of

the valve body. The units for the Niyama values shown below are (°C-s) 1/2 /mm. The scale of 0

to 1.4 shown in these figures was chosen because low Niyama values in this range are known to

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: