Materials Algorithms Project
Program Library

Data Library MAP_DATA_THERMAL

Provenance of code.
Purpose of code.
Description of subroutine's operation.
References.
Any additional information.
Keywords.
Download source code.
Links.

Provenance of Source Code

Hala Salman Hasan and Mathew Peet,
Phase Transformations Group,
Department of Materials Science and Metallurgy,
University of Cambridge,
Cambridge, U.K.

Top | Next

Purpose

Provides data on thermal conductivity in steel. The authors used the data to create a neural network model of the thermal conductivity in steel.

Top | Next | Prev

Description

The TAR file database.tar.gz contains four data files. These are present as the worksheets in the spreadsheet versions.

databases-compiled-data.csv all the data collected before training our neural network (data from matweb.com and Smithall metals Reference Book .
databases-used-for-model.csv pruned and sorted data for training.
databases-final-test.csv data collected after training to test the model (data from Holman - heat transfer).
databases-committee_prediction.csv contains gamma-prime phase lattice parameters (in ］gstroms) for each alloy.

Description of databases-compiled-data

Column 1: "Label"; Line Label - depending on alloy, e.g. pure_01, pure_02, Steel_01
Column 2: "Fe"; Amount of Iron in weight percent - calculated as balance.
Column 3 - 20: "C","Mn","Ni","Mo","V","Cr","Cu","Al","Nb","Si","W","B","Ti","Co","P","S","N","Zr"; Elements Carbon, Manganese, Nickel, Molybdenum, Vanadium, Chromium, Copper, Aluminium, Niobium, Silicon, Tungsten, Titanium, Cobalt in weight percent.
Column 21: "T / 'C"; Temperature in Celcius.
Column 22: ""Thermal Cond / W/m/K",; Thermal conductivity in Wm^-1K^-1
Column 23: "Reference"; Source for the data.

Description of databases-used-for-model

Column 1: "Fe"; Amount of Iron in weight percent - calculated as balance.
Column 2 - 14: "C","Mn","Ni","Mo","V","Cr","Cu","Al","Nb","Si","W","Ti","Co"; Elements Carbon, Manganese, Nickel, Molybdenum, Vanadium, Chromium, Copper, Aluminium, Niobium, Silicon, Tungsten, Boron, Titanium, Cobalt, Phosphorous, Sulphur, Nitrogen, Zironcium in weight percent.
Column 15: "TC"; Temperature in Celcius.
Column 16: "K",; Thermal conductivity in Wm^-1K^-1
Column 17: "label",; Line Label - depending on alloy, e.g. pure_01, pure_02, Steel_01

Description of databases-final-test

Same as the databases-used-for-model but final column is reference.

Description of databases-committee_prediction

Comparison of the target values of thermal conductivity against the values generated by the model.

Column 1: "Cond"; Experimental value of the thermal conductivity.
Column 2: "Predicted"; Value predicted using the final committee model of the neural network.
Column 3: "Error"; Error of the prediction.
Column 4: "Error-with-sigma-nu"; Error of the prediction including the sigma_nu model percieved level of noise.

Top | Next | Prev

References

Smithall metals Reference Book Ed: W.F. Gale, T.C. Totmeier Eight Edition Publisher Elsevier/ASM 2004.
P. Kardititas, M-J Baptiste, Thermal and structural properties of fusion related Materials, https://www-ferp.ucsd.edu/LIB/PROPS/PANOS/ss.html.
MATWEB, Material property Data, https://www.matweb.com/.
J.P.Holman, Heat Transfer, 8th Edition 1997 McGraw-Hill Companies.

Top | Next | Prev