Save
Money with Better Tech
Improvement on Level 2 Model
Temperature Prediction
Significance of Accurate
Temperature Prediction
After a
product order is received in
a mill,
the metallurgist will
schedule a steel grade (if
the order didn't specify it) and a
set of production stages
(reheating, rolling,
controlled cooling, etc.) to
fulfill the order. The
temperatures in the start
and end of each stage are
among the most critical
parameters for the
metallurgist to schedule,
because the metallurgical
transformations and finish
steel properties are tightly
related to the transaction
temperatures. If the Level 2
model is not sufficiently
accurate, the actual
temperatures in the
production would be off the
schedule and the final
product properties would be
off the targets. In
addition, due to the
different temperatures
between the actual and
predicted (planned) ones, the
actual deformation could be
under higher force and
torque, and this could even damage
the equipment. Selected
examples for the problem are
discussed below.
Examples of the
problems caused by a wrong
temperature
-
Wrong
mechanical properties, often
poorer ones. Wrong
rolling temperature leads to
the rolling process to be
performed in a wrong
metallurgical region and
this may produce different
microstructures from
planned. During the
production scheduling, the
metallurgist would fully
utilize metallurgical
transaction (controlled
rolling, precipitation,
etc.) to produce preferred
microstructure and finish
properties. Due to the
system error, a wrong
rolling temperature in each
rolling stage, and
consequently maybe also a
wrong initial temperature
for controlled cooling, lead
to different microstructure
and poor mechanical
properties. The rolled steel
may not achieve the target
properties.
-
Wrong
force. Wrong temperature
leads to a wrong force and
may damage the rolls and
other equipment. This is
particularly the case in the
roughing rolling stage. For
finishing passes, accurate
force prediction is critical
to a good product shape. See
Significance of Accurate
Force Prediction and
Product Defects and Level 2
Model Error.
-
Other
unexpectations. For example,
due to temperature error, the rolling
in the finish pass may actually
be performed in the
two-phase region, as in a
case of a Metal Data client (Fig. 1).
In the last two passes, with
a decreasing temperature the
flow stress went lower.
This created large force
error (sometimes over 40%)
and caused very poor product
geometry.
Fig. 1: Measured flow stress at strain 0.3 and strain rate 10/s
Temperature Model
Improvement
Many Level 2 systems only
use the primary model to
calculate temperature. The
primary model takes into
account the heat transfer in
conduction, convection,
radiation and heat
generation during rolling
and controlled cooling.
During the calculation the
coefficients for those heat
transfer types are used as
constants rather than
variables depending on
operational parameters (temperature,
speed, etc.).
However, it is the secondary model that contributes greatly
to high accuracy of temperature prediction. We have a long
list of the secondary models to predict heat transfer
coefficients in conduction, convection and radiation, and the
ratio of thermal energy to deformation energy. As
examples, the convection heat transfer coefficient can be
calculated with the relative speed of air/water and the
rolling stock, and the heat transfer coefficient between work
roll and stock also depends on many parameters and often varies from
2500W/m2K to 10,000W/m2K. Many systems
treat the heat transfer coefficient between work roll and
stock as a constant in doing the temperature learning. If the
influence factors are integrated into the learning, it will
greatly increase the temperature prediction accuracy. In
addition, the specific heat is strongly temperature dependent.
If austenite - ferrite phase transformation occurs, the
specific heat will sharply increase to more than twice as high
and then sharply drop.
Most
Level 2 systems also ignores
thermal energy involved in
the phase transformation or
other metallurgical
processes.
Most
thermal properties are
temperature dependent though
many Level 2 system treat
them as constant values.
Please see
http://www.Meta4-0.com/hit
for background knowledge and
data.
What we can do for you
For your Level 2
temperature model
improvement we may help in following areas:
-
To analyze your history data by accessing your log files or database data to
identify operational problems. We have a list of programs that can be used to
analyze your data. To
save your money, we may
provide our web-based
Remote Diagnosis
system for you to find
your problems yourselves.
-
To establish optimized system learning for your Level 2 model. For
example, many Level 2
systems use the heat
transfer coefficient
between the roll and
steel as learning
parameter. It would be
more efficient to
redesign this learning
logic by using our
secondary models because
the heat transfer
coefficient depends
on rolling process
parameters.
-
To apply accurate
high-temperature
property data and our
secondary thermal models
in your Level 2 system.
Most modifications may
be made only in the
input log
files or input databases,
so we may not need to
make so many
modifications to your
Level 2 source code.
Metal Data Resources on
Level 2 Model Improvement
