Casting Design Optimization Driven by Simulation
Casting Design Optimization Driven by Simulation
Casting Design Optimization Driven by Simulation
Dr. B. Ravi, Professor of Mechanical Engineering, I.I.T. Bombay (b.ravi@iitb.ac.in)
T he methods layout of a casting is an important activity
in tooling development. It involves critical decisions
regarding part orientation in mold, parting line, cores,
Further, taking the average difference in the price of a
saleable casting and scrap metal as Rs. 10/kg, and assuming
average rejections in a foundry as 5%, the economic loss
cavity layout, feeders, feedaids and gating system. An caused by defective castings works out to Rs. 500 per tonne
improper layout leads to either poor quality or low yield, of production (in reality this can be much higher, with
1
affecting manufacturing costs and productivity . transport, warranty, and failures during product life).
Casting simulation can overcome the above problems:
virtual trials do not involve wastage of material, energy and
Computer simulation provides a clear insight labour, and do not hold up regular production. However,
regarding the location and extent of internal defects, most of the simulation programs available today are not
ensuring castings are right first time and every time. It easy‐to‐use, take as much time as real trials, and their
however, requires a 3D CAD model of the method accuracy is affected by material properties and boundary
layout (with mold cavities, cores, feeders, and gating conditions specified by users. The biggest problem is the
channels), proper setting of boundary conditions for
preparation of 3D model of the mold cavity with cores,
each virtual trial, and correct interpretation of results.
feeders and gating for every iteration, which requires CAD
AutoCAST software integrates and automates the
skills and takes considerable time for even simple parts.
above tasks, and provides an extremely easy‐to‐use
This also prevents early manufacturability evaluation and
graphical user interface suitable for even first‐time
improvement by product designers, which can benefit
computer users. The mold cavities, feeders and gating
several times more than tooling and process changes.
system are automatically optimized, driven by the
criteria and constraints specified by user. This reduces
the total time for methods design and simulation of a The AutoCAST software developed by Advanced Reasoning
typical casting to less than one hour. Technologies, Mumbai in collaboration with I.I.T. Bombay
provides a single integrated user‐friendly environment for
casting methods design, solid modeling, and simulation2.
It handles both ferrous and non‐ferrous parts, and sand as
well as metal molds. Release 10 incorporates multi‐cavity
Methods design is usually carried out manually on 2D
mold layout, automatic modeling and optimization, and a
drawings of the cast part. Then tooling is fabricated, trial
costing model to compare various layouts (Fig.1).
castings are produced in the foundry, and inspected. If
sample castings contain defects (such as shrinkage or gas
porosity), then the methods layout is modified and the
process is repeated. Each such iteration can take up several
days, affecting regular production. After a few iterations,
the foundry may resort to a ‘safe’ methods design (implying
low yield), or continue with high rejection rates (implying
high scrap or repair cost). This is especially true in the case
of large castings, where the cost of a trial or repair can be
prohibitive.
Assuming a typical foundry develops 50 new castings every
year, each casting requires at least 2 trials, and the average
cost of each trial (tooling modification, melting & pouring,
inspection, and effect on regular production) as Rs. 20,000,
the economic loss works out to be two million (20 lakh)
Fig.1. Casting methods design and simulation software.
rupees per year per foundry.
1
Computer-aided Methods Design
The main input is the 3D CAD model of an as‐cast part:
without drilled holes, and with draft, shrinkage and
machining allowance (Fig.2). The model file can be obtained
from the OEM firm, or created by a local CAD agency.
Various display options such as pan, zoom, rotate,
transparency and measure, are provided to view and
understand the part model. The cast metal and process are
selected from a database. Thickness map is generated. Part
manufacturability (compatibility with the selected process)
is computed and pictorially displayed (Fig.3).
Methods design involves cores, feeders and gating system.
Fig.2. Part property computation. Holes in the part model are automatically identified for
core design, or plugged if they are drilled. Even intricate
holes can be identified by specifying their openings. The
print length is computed based on core diameter and
length (the user can change their values if required), and
the entire core model is automatically created. The
program suggests the number of cavities depending on
the mold size (selected from a customizable library),
considering both cavity‐cavity and cavity‐wall gaps. Then
the part model is automatically duplicated in the correct
locations as per the desired cavity layout (Fig.4).
To facilitate feeder location, a quick solidification analysis is
carried out that identifies feeding zones. The user selects a
suitable connection point close the hottest zone, and the
size of the feeder is computed using modulus principle
(solidification time of feeder slightly more than that of the
Fig.3. Part thickness distribution with sensor. feeding zone). Standard feeder shapes include cylindrical,
oval, spherical‐bottom, cruciform, etc. Other shapes can be
imported. The feeder model is automatically created; the
user can change its dimensions or apply feedaids such as
insulating sleeves and exothermic covers. Chills, padding
and fins can also be created. More feeders or feeders with
multiple necks can be created by specifying their positions.
The gating channels are also created semi‐automatically.
First, the user indicates gate positions on the part or feeder
model. Then the sprue position is decided, and connected
to the gates through runners. Runner extensions are also
automatically created. Either horizontal or vertical gating
system can be designed and modified within minutes. The
program suggests a suitable filling time (which can be
changed by user), accordingly computes the dimensions of
the gating channels, and creates their solid model.
Fig.4. Methods design and its automatic modeling.
2
Automatic Optimization
The mold cavity layout, feeders, and gating models are
automatically optimized within minutes based on quality
requirements and other constraints3. For mold cavity
layout, the primary criterion is the weight ratio of cast
metal to mold material. A high ratio such as 1:2 (cavities
too close to each other) can reduce the heat transfer rate
and lead to shrinkage porosity defects. A low ratio such as
1:8 (cavities too far from each other) implies poor
utilization of mold material and reduced productivity. The
program tries out various combinations of mold sizes and
number of cavities to find the combination that is closest to
the desired value of metal to mold ratio.
Fig.5. Melt jet path and mold filling.
The gating optimization is driven by the ideal mold filling
time, which depends on cast metal, casting weight and
minimum wall thickness. Fast filling leads to turbulence‐
related defects (mold erosion, air aspiration and inclusions).
Slow filling may cause defects related to premature
solidification (cold shuts and misruns). To optimize the
gating design, mold filling is simulated and total fill time is
computed (Fig.5). A layer‐by‐layer filling algorithm takes
into account the instantaneous velocity through the gates
(considering back pressure), and the local cross‐section of
the mold cavity. This gives a fairly accurate estimation of
filling time, while being computationally fast. If the
difference between the ideal and simulated filling time is
more than a specified limit, the program automatically
changes the gating design, creates its solid model, and
verifies the filling by simulation.
Fig.6. Casting solidification simulation.
The feeder optimization is driven by casting quality, defined
as the percentage of casting volume free from shrinkage
porosity. The user indicates a target quality. The program
automatically changes the feeder dimensions, creates its
solid model, carries out solidification simulation (Fig.6), and
estimates the casting quality. The solidification simulation
employs the Vector Element Method, which computes
temperature gradients (feed metal paths) inside the
casting, and follows them in reverse to identify the location
and extent of shrinkage porosity (Fig.7). This has been
found to be much faster than Finite Element or Volume
Method, and usually more accurate too. Feeder design
iterations are carried out until the desired quality is
achieved, or the number of iterations exceeds a set limit.
The user can accept the results, or can modify the feeder
design interactively.
Fig.7. Feed metal paths (temperature gradients).
3
Today AutoCAST is the most widely used casting software in
India with 50 licenses (foundries, engineering and R&D
institutes, and consultants) covering all major cast metals
and processes. Many others have used the software for
benchmarking. Simulation consultants are available across
the country to provide local technical support, ensuring a
smooth transition to computer‐aided methoding.
Summary and Future
Casting simulation can minimize the wastage of resources
required for trial production. In addition, the optimization
of quality and yield implies higher value‐addition and lower
production cost, improving the margins. For widespread
application, simulation programs must be fast, reliable, and
Fig.8. Cost analysis and methods report generation. easy to use. This has been achieved by integrating methods
design, solid modeling, simulation and optimization in a
Finally, the cost of the casting is computed in terms of single software program, and automating many tasks that
amortized tooling, cast metal, other materials (mold, core, otherwise require scientific knowledge and computational
etc.), energy, and labour. Various cost rates and parameters skills. In many benchmarking exercises and simulation
can be set by the user. This enables comparing different clinics (Fig.9), the software has consistently proven its
casting layouts in terms of tooling and manufacturing cost. reliability in predicting internal defects (ex. shrinkage
A detailed methods design report along with an image of porosity) within minutes, often by senior engineers who are
the entire casting is automatically generated, which can be first time computer users. The simulation costs are a
printed or stored for future reference (Fig.8). fraction of the costs of foundry trials, while providing better
and faster insight for casting optimization. A network of
The metal database covers all major alloys (aluminum, local technical support centres and simulation consultants
copper, cast iron, ductile iron, steel, and precious metals) across the country ensures that even SME foundries in
and processes (sand, shell, investment, die casting). It can remote areas can now take advantage of the technology.
be customized to any new metal‐process combination. The goal of castings right first time, every time, in the
shortest time, is within the reach of every foundry.
The software has been developed for standard Windows XP
computers, and performs well on even portable computers.
The graphical interface is designed to minimize the learning
and operation time, and the user is gently guided through
forgotten or wrong steps. Even those without any prior
exposure to computers are able to use the software after a
single day of training. All steps starting from part model
importing to mold, core, feeder and gating system design,
simulation and optimization are completed within one hour
for typical castings.
Fig.9. Casting simulation training and clinic at Mumbai.
Direct benefits include at least 50% reduction in casting
development time and porosity defects. Other benefits References
include yield improvement, faster quotation, handling more