Transcript ATAS - NovaCast

Adaptive Thermal Analysis System
Metallurgical Process Control for
Gray and Ductile Iron
Derivative and Calorimetric
Thermal Analysis combined with
Applied Artificial Intelligence
Developed by NovaCast AB in cooperation with
Swedish Foundry Association
- provides know-how and technology to winners in the foundry industry 1
© NovaCast AB
Process Control?
Are you a “blind” foundryman?
• Day 1: C = 3,4 %
Si = 2,1 % , 1380 C
Scrap rate = 2,0 %.
• All measured data OK!
• Day 2: C = 3,4 % Si
= 2,1 % , 1380 C Scrap
rate = 12,0 %
• All measured data OK!
The problem is that essential data are not measured and knowledge is lacking.
The foundryman can not see what he should see!
- provides know-how and technology to winners in the foundry industry 2
© NovaCast AB
Factors influencing Metallurgical Quality
•
•
•
•
•
•
•
•
•
•
Chemical composition
Charge materials and composition (Combined & Free C)
Charging sequence and temperature / time ratios
Iron handling and treatment practice
Pouring temperature and inoculation practice
Presence and amount of reaction products
Thermal properties of alloy, mould and core(s)
Mould / Core hardness / Weighting
Methoding (gating, feeding, feeding paths, etc.)
Shake-out time
It is understandable that the properties vary between melts.
- provides know-how and technology to winners in the foundry industry 3
© NovaCast AB
Why chemistry is not enough
•
•
•
•
•
•
Spectrometer only shows amount per element!
No info about compounds!
No info about nucleation properties!
No info about graphite precipitation!
Tramp elements often not analysed
Are spectrometer readings reliable? C, Si
The purpose of ATAS is to open a new window into the metallurgical process. Melts with
identical chemistry can behave completely different when poured! It is the combination of
advanced thermal analysis and chemistry that makes efficient process control possible.
- provides know-how and technology to winners in the foundry industry 4
© NovaCast AB
Metallurgy Related Problems
•
•
•
•
•
•
•
•
•
Macro- and micro shrinkages, porosity
Chill and intercellular carbides
Graphite shape and matrix
Expansion penetration
Certain slag and gas-related defects
Fluidity, castability
Strength, hardness, machinability
Nodularity, Nodule Count
About 40% of all scrap has metallurgical causes!
- provides know-how and technology to winners in the foundry industry 5
© NovaCast AB
Ideal Metallurgical Process Control
•
•
•
•
•
•
•
Measure essential metallurgical parameters
Interpret actual data
Predict final results
Learn from experience
Adjust according to suggested corrections
Advisory system increases knowledge
Increased awareness / Tutorial
ATAS is designed to fulfill these requirements
- provides know-how and technology to winners in the foundry industry 6
© NovaCast AB
Constitution Diagram & TA
Traditional thermal analysis uses tellurium-coated
cups. Carbon precipitates as Fe3C. Easy to detect
temperature arrests and to calculate CEL, C and
Si.
Does not tell us anything about the most important
feature, namely how carbon is precipitated into
graphite.
- provides know-how and technology to winners in the foundry industry 7
© NovaCast AB
Cooling Curves in ATAS
ATAS uses cooling curves from ”gray” samples. It is when carbon is allowed to solidify
as graphite that relevant information can be revealed from the curve.
- provides know-how and technology to winners in the foundry industry 8
© NovaCast AB
The Metallurgical “fingerprint”
Each melt is an “individual”. The cooling curve can be regarded as its thermodynamic fingerprint.
- provides know-how and technology to winners in the foundry industry 9
© NovaCast AB
Learning from observations
•
•
•
•
Induction
... is the process of finding general rules and
combinations by recognizing and combining specific
cases.
Deduction
... is the process of using general rules in order to
achieve an outcome about a specific case.
ATAS uses applied artificial intelligence to discover rules from process data using INDUCTION.
Rules are then applied using DEDUCTION.
- provides know-how and technology to winners in the foundry industry10
© NovaCast AB
Induced Rules in ATAS
These seven rules have been induced from some 100 samples. The inductive learning method was
used during the development of the rules used by the Expert System in ATAS.
- provides know-how and technology to winners in the foundry industry11
© NovaCast AB
Rule-based Expert System
IF
amount of primary austenite is > 20 % (first cond.)
AND
liquidus is > (maxliquidus + 10)
AND
first derivative start eutectic < min TES (third cond.)
(second cond.)
THEN Risk for MACRO SHRINKAGE is HIGH (conclusion)
Rules that have been discovered by induction of examples are transformed into so-called
production rules and used in ATAS. This slide shows a typical rule. The threshold values are
stored in the alloy database and can be adjusted to various conditions in the foundry.
- provides know-how and technology to winners in the foundry industry12
© NovaCast AB
Properties vary although chemistry is constant!
Variation Recalescence & R-rate
19
11
9
7
5
1
3
Low eutectic temperature
19
17
15
13
1130,0
11
1135,0
0,0
9
1140,0
2,0
7
4,0
Test #
1165,0
1160,0
TE_low C
1155,0
1150,0
1145,0
1140,0
1135,0
19
17
15
13
11
9
7
5
3
1130,0
1
5
1145,0
3
6,0
17
8,0
1
C
10,0
15
12,0
13
Samples
taken from base iron
1165,0
at1160,0
random intervals during the
1155,0
day.
1150,0
14,0
Test #
- provides know-how and technology to winners in the foundry industry13
© NovaCast AB
Making the Casting Process Reliable
Verify the process by providing information
about essential parameters
Optimize all process stages.
ATAS provides
relevant process data
Adjust the process by allowing accumulative
storage of experience
- provides know-how and technology to winners in the foundry industry14
© NovaCast AB
1st derivative for a good DI
- provides know-how and technology to winners in the foundry industry15
© NovaCast AB
Graphite Precipitation Pattern
- provides know-how and technology to winners in the foundry industry16
© NovaCast AB
1st derivative “bad” DI
- provides know-how and technology to winners in the foundry industry17
© NovaCast AB
Graphite Pattern for a “bad” DI
- provides know-how and technology to winners in the foundry industry18
© NovaCast AB
General Applications for ATAS
• Verify the melt before pouring (Quality Assurance)
• Reduce scrap caused by metallurgical problems
• Improve yield
• Minimize risk for unexpected problems
• Find OPTIMAL process parameters and threshold values
• Reduce costs for charge materials
• Quality control for Inoculants, FeSiMg etc.
• Reduce variations in physical properties
continued..
- provides know-how and technology to winners in the foundry industry19
© NovaCast AB
Specific Applications for ATAS
• Dynamic dosage of inoculant
• Predict % Nodularity
• Predict % Pearlite, BHN and Tensile
• Data for SPC systems
• Educational tool - “Live” learning
• Research & Development
- provides know-how and technology to winners in the foundry industry20
© NovaCast AB
ATAS - the complete system for
Metallurgical Process Control
- provides know-how and technology to winners in the foundry industry21
© NovaCast AB
Adapting ATAS - Phase 1
Taking ATAS into operation starts with about 20 trials to update the alloy database.
- provides know-how and technology to winners in the foundry industry22
© NovaCast AB
ATAS, Ready to Run! - Phase 2
ATAS can be used in production once the alloy database is customized
- provides know-how and technology to winners in the foundry industry23
© NovaCast AB
Accumulate Knowledge - Phase 3
When in production, data are analysed, the alloy database refined and new cases are entered.
Knowledge about the process is accumulated and the prediction accuracy increases.
- provides know-how and technology to winners in the foundry industry24
© NovaCast AB
Economical Consequences
Assume:
Yearly production of 10.000 tons
Scrap reduction: $ 10 to 20 / percent / tonnage
Reduction, 1%, saving :
$ 150.000
Yield improvement: $ 2-4 / percent / tonnage
Improving, 65% to 67%, saving: $ 60.000
$
Reduced consumption of inoculant and Mg-alloy
5% reduction, saving of at least: $ 50.000
Total net savings approx:
$ 25 per ton.
Plus reduced cost for sampling, plus higher customer
satisfaction etc
Yearly savings:
$ 250.000
- provides know-how and technology to winners in the foundry industry25
© NovaCast AB
ATAS makes you
The Scientific Foundryman!
NovaCast AB, Soft Center, S-37225 Ronneby, Sweden
Your partner in Methoding & Process Control
Phone: (46) 45738 63 00 Fax: (46) 457 15622
E-Mail: [email protected]
Web site:http://www.novacast.se
- provides know-how and technology to winners in the foundry industry26
© NovaCast AB