Mathematical Modeling of Displacements and Thermal Stresses in Anisotropic Materials (Sapphire) in Cooling

Similar documents
Introduction to Seismology Spring 2008

Using Thermal Boundary Conditions in SOLIDWORKS Simulation to Simulate a Press Fit Connection

Generalized Plane Piezoelectric Problem: Application to Heterostructure Nanowires

Peristaltic Pump. Introduction. Model Definition

Final Design Project: Biodiesel Settling Tank Analysis

Understand basic stress-strain response of engineering materials.

3-D Finite Element Analysis of Instrumented Indentation of Transversely Isotropic Materials

Laminated Composite Plates and Shells

A Finite Element Model for Numerical Analysis of Sintering

Predicting Elastic Properties of Unidirectional SU8/ZnO Nanocomposites using COMSOL Multiphysics

Shape, Convection and Convergence

Constitutive models: Incremental plasticity Drücker s postulate

Prediction of Elastic Constants on 3D Four-directional Braided

A Flexible Scheme for the Numerical Homogenisation in Linear Elasticity

Stress analysis of deflection analysis flexure and obif Vertical Load orientation

Lecture #6: 3D Rate-independent Plasticity (cont.) Pressure-dependent plasticity

FEA Mechanical Modeling of Torque Transfer Components for Fully Superconducting Rotating Machines

3.22 Mechanical Properties of Materials Spring 2008

AEROELASTIC ANALYSIS OF COMBINED CONICAL - CYLINDRICAL SHELLS

Table of Contents. Preface...xvii. Part 1. Level

Elastoplastic Deformation in a Wedge-Shaped Plate Caused By a Subducting Seamount

DESIGN OF LAMINATES FOR IN-PLANE LOADING

arxiv: v1 [physics.ins-det] 4 Jun 2018

Finite Element Modeling for Transient Thermal- Structural Coupled Field Analysis of a Pipe Joint

A Piezoelectric Screw Dislocation Interacting with an Elliptical Piezoelectric Inhomogeneity Containing a Confocal Elliptical Rigid Core

Effect of Temperature on Mechanical Properties and Dynamic Characteristics of the Rotor in High Acceleration Mei-Hui JIA a,*, Cheng-Lin WANG b

Nonlinear Finite Element Modeling of Nano- Indentation Group Members: Shuaifang Zhang, Kangning Su. ME 563: Nonlinear Finite Element Analysis.

Elmer :Heat transfert with phase change solid-solid in transient problem Application to silicon properties. SIF file : phasechange solid-solid

3D and Planar Constitutive Relations

PEAT SEISMOLOGY Lecture 9: Anisotropy, attenuation and anelasticity

142. Determination of reduced mass and stiffness of flexural vibrating cantilever beam

Stresses Analysis of Petroleum Pipe Finite Element under Internal Pressure

MATERIAL REDUCTION & SYMMETRY PLANES

DEVELOPMENT OF SOPHISTICATED MATERIALS USING THE COMPUTER SIMULATION

MODELING OF CONCRETE MATERIALS AND STRUCTURES. Kaspar Willam

Lecture 8. Stress Strain in Multi-dimension

1.103 CIVIL ENGINEERING MATERIALS LABORATORY (1-2-3) Dr. J.T. Germaine Spring 2004 LABORATORY ASSIGNMENT NUMBER 6

Multiphysics Modeling

Buckling Behavior of 3D Randomly Oriented CNT Reinforced Nanocomposite Plate

Two-dimensional ternary locally resonant phononic crystals with a comblike coating

MODELING OF ELASTO-PLASTIC MATERIALS IN FINITE ELEMENT METHOD

MECHANICS OF MATERIALS. EQUATIONS AND THEOREMS

Prediction of Transformer Core Noise

Computational Analysis for Composites

Strain Measurement Techniques for Composite Coupon Testing

Fundamental Solution of 3D Isotropic Elastic Material

SIMULATION AND TESTING OF STRUCTURAL COMPOSITE MATERIALS EXPOSED TO FIRE DEGRADATION

PEAT SEISMOLOGY Lecture 2: Continuum mechanics

AE3610 Experiments in Fluid and Solid Mechanics TRANSIENT MEASUREMENTS OF HOOP STRESSES FOR A THIN-WALL PRESSURE VESSEL

Fundamentals of Linear Elasticity

Simulation and Experimental Validation of Induction Heating of MS Tube for Elevated Temperature NDT Application.

Example-3. Title. Description. Cylindrical Hole in an Infinite Mohr-Coulomb Medium

Design Optimization of an Electronic Component with an Evolutionary Algorithm Using the COMSOL-MATLAB LiveLink

Pressure Vessels Stresses Under Combined Loads Yield Criteria for Ductile Materials and Fracture Criteria for Brittle Materials

Module III - Macro-mechanics of Lamina. Lecture 23. Macro-Mechanics of Lamina

Supplementary Figures

Introduction to Continuous Systems. Continuous Systems. Strings, Torsional Rods and Beams.

The Use of COMSOL Multiphysics for Studying the Fracture Pressure of Rectangular Micro-Channels Embedded in Thin Silicon Substrates

CRITERIA FOR SELECTION OF FEM MODELS.

NUMERICAL ANALYSIS OF HEAT STORAGE AND HEAT CONDUCTIVITY IN THE CONCRETE HOLLOW CORE DECK ELEMENT

Constitutive Relations

DEVELOPMENT OF A CONTINUUM PLASTICITY MODEL FOR THE COMMERCIAL FINITE ELEMENT CODE ABAQUS

Multiphysics Simulation and Optimization for Thermal Management of Electronics Systems

A Semi-Analytical Thermal Elastic Model for Directional Crystal Growth with Weakly Anisotropy

Modelling and numerical simulation of the wrinkling evolution for thermo-mechanical loading cases

Received 23 January 2016; received in revised form 20 May 2016; accepted 21 June 2016 DOI

Lectures on. Constitutive Modelling of Arteries. Ray Ogden

Using MATLAB and. Abaqus. Finite Element Analysis. Introduction to. Amar Khennane. Taylor & Francis Croup. Taylor & Francis Croup,

ENGN 2340 Final Project Report. Optimization of Mechanical Isotropy of Soft Network Material

Coupled CFD-FE-Analysis for the Exhaust Manifold of a Diesel Engine

1. Background. is usually significantly lower than it is in uniaxial tension

Azimuthal AVO and Curvature. Jesse Kolb* David Cho Kris Innanen

COUPLED MAGNETO-MECHANICAL FINITE ELEMENT ANALYSIS OF A POWER TRANSFORMER IN SHORT CIRCUIT CONDITIONS

Seismic Waves in Complex 3 D Structures, 26 (2016), (ISSN , online at

Heat Transfer Modeling using ANSYS FLUENT

Contents. I Introduction 1. Preface. xiii

Reflection of SV- Waves from the Free Surface of a. Magneto-Thermoelastic Isotropic Elastic. Half-Space under Initial Stress

Thermal and hydraulic modelling of road tunnel joints

Back Calculation of Rock Mass Modulus using Finite Element Code (COMSOL)

Multiphysics Simulations for the design of a Superconducting magnet for proton therapy

Analysis of the geometrical dependence of auxetic behavior in reentrant structures by finite elements

Theoretical Manual Theoretical background to the Strand7 finite element analysis system

Modelling Anisotropic, Hyperelastic Materials in ABAQUS

Chapter 6: Plastic Theory

Thermal Analysis. with SolidWorks Simulation 2013 SDC. Paul M. Kurowski. Better Textbooks. Lower Prices.

Constitutive Relations

Contents as of 12/8/2017. Preface. 1. Overview...1

Example Resistive Heating

MODELING THE EFFECTIVE ELASTIC MODULUS OF RC BEAMS EXPOSED TO FIRE

University of Sheffield The development of finite elements for 3D structural analysis in fire

Load Cell Design Using COMSOL Multiphysics

Solving the Inverse Problem of Resonant Ultrasound Spectroscopy on Dumbbell-shaped Compression Samples using COMSOL

SEMM Mechanics PhD Preliminary Exam Spring Consider a two-dimensional rigid motion, whose displacement field is given by

Frequently Asked Questions

Calculating Mechanical Transfer Functions with COMSOL Multiphysics. Yoichi Aso Department of Physics, University of Tokyo

MATERIAL MECHANICS, SE2126 COMPUTER LAB 4 MICRO MECHANICS. E E v E E E E E v E E + + = m f f. f f

A Constitutive Framework for the Numerical Analysis of Organic Soils and Directionally Dependent Materials

Multiphysicss. 1 Indian Institute of. Machining. materials. In. for the anode tool. metal which leads to. form: made to study the.

Coupled electromagnetic, thermal and stress analysis of large power electrical transformers

Design and Analysis of Various Microcantilever Shapes for MEMS Based Sensing

Transcription:

Mathematical Modeling of Displacements and Thermal Stresses in Anisotropic Materials (Sapphire) in Cooling Timo Tiihonen & Tero Tuovinen September 11, 2015 European Study Group with Industry, ESGI 112, LUT, Sept 7-11, 2015 Problem setup by Jari Järvinen Silicom Ltd. Abstract In this study we are mathematically modeling a sapphire crystal growth by heat exchanger method (HEM). In the system, solid material is initially at temperature close to its melting point temperature, and the cooling takes place until the solid is at room temperature. Cooling is organized with decreasing heater power (heat source) in the system. We have modeled the system that is isolated thermally in every boundary except at the bottom and a cone, and there we have defined constant heat out-flow. Sapphire belongs to the hexagonal system and rhombohedra single crystals. In our analysis, the aim is to study displacements and thermal stresses for isotropic/anisotropic 3D and axi-symmetric solid material (sapphire) during the cooling? In particularly, we will answer the reliability issues related to how valid the axi-symmetric approximation compared to 3D computations. We will consider the validity of the isotropic assumption compared to the thermally and structurally anisotropic case. The numerical simulations are implemented by using Comsol Multiphysics software. 1 Introduction Sapphire crystal growth with heat exchanger method is schematized in Figure 1. The process takes place in an isolated chamber filled with inert gas. Due to high temperatures involved (over 2000C) radiativeheattransferispredominant in the gas. As our focus here is primarily in the prediction of thermally induced stresses in the crystal we will focus on modeling the crystal and assume its thermal environment known. By simple estimation it can be concluded that the role of gravitation (self weight and related forces against supporting surfaces) is 1

Figure 1: Sapphire Crystal Growth by Heat Exchanger Method (HEM). quite minor compared to expected level of thermal stresses. This, together with the fact that sapphire shrinks during the cooling faster than the supporting materials, enables us to ignore also the surrounding vessel from mechanical considerations. Hence the modeling can be restricted to sapphire alone. The modeled sapphire crystal is sketched in Figure 2. It has three regions: the seed crystal, the expanding cone and the cylindrical part that will be the actual produced crystal to be processed further. In what follows we briefly summarize the governing equations with focus on the material characteristic properties, sketch the simulation scenarios and summarize the main findings. Sample numerical data/graphs are included as appendix. Additional information related to the topic can be found in [Buchanan, 2000], [Chen et al., 2014], [Dobrovinskaya et al., 2009] and [Fang et al., 2013]. 2

2 Governing equations 2.1 Heat transfer in solids Modeling the heat transfer in solids, we have used the following representation C p @T @t = r (krt ) (1) where C p denotes the heat capacity (constant in our study), k the heat conductivity (likewise treated as constant and isotropic in our study) and T is temperature. We have assumed that system can be controlled outside using heaters so that boundary conditions in producing chamber can be considered thermally insulated. The equation for thermal insulation is defined as n ( krt )=0 (2) The computations have been executed for two separate boundary conditions for cooling, heat flux and constant out-flow. However, we have reported only the results from out-flow condition. The equation for heat flux boundary condition is described as follows n ( krt )=h (T ext T ) (3) and boundary conditions for constant out-flow can be represented as n ( krt )=q 0. (4) T ext is external temperature and we have assumed that to be 273K. 2.2 Solid mechanics Modeling of mechanical deformation and stress analysis, we have considered the following equation r + F v =0 (5) in the domain. F v stands for volumetric force (gravitation) that is ignored in our study. One-side coupling with thermal effects is done by using following condition = C :( T ) (6) where = 1 ru +(ru) T and T = (T T ref ). (7) 2 and denotes the (matrix of) heat expansion coefficients that is slightly anisotropic in our case. Our aim was to disturb the stresses of the system as little as possible using prescribed displacement and other constraints that are necessary for numerical convergence. In particular, we have used spring foundation approach where we 3

have defined very weak springs in the upper corners of the domain. conditions for spring foundation can be expressed as Line n = k A (u u 0 ) (8) where k A is spring constant per unit area. The anisotropic elastic properties are expressed in matrix notation, also called Voigt notation. To do this we take advantage of the symmetry of the stress and strain tensors and express them as six-dimensional vectors in an orthonormal coordinate system (e1, e2, e3) as [ ]= 11, 22 33, 23, 13, 12 1, 2, 3, 4, 5, 6 (9) and [ ] = 11, 22, 33, 2 23, 2 13, 2 12 1, 2 3, 4, 5, 6 (10) Then the stiffness tensor Ccan be expressed as 2 3 c 1111 c 1122 c 1133 c 1123 c 1131 c 1112 c 2211 c 2222 c 2233 c 2223 c 2231 c 2212 C = c 3311 c 3322 c 3333 c 3323 c 3331 c 3312 6 c 2311 c 2322 c 2333 c 2323 c 2331 c 2312. (11) 7 4 c 3111 c 3122 c 3133 c 3123 c 3131 c 3112 5 c 1211 c 1222 c 1233 c 1223 c 1231 c 1212 Sapphire is transversely isotropic material. That is, it is symmetric with respect to a rotation about an axis of symmetry. For such a material, if e 3 is the axis of symmetry, Hooke s law can be expressed as 2 6 4 1 2 3 4 5 6 3 2 = 7 5 6 4 C 11 C 12 C 13 0 0 0 C 12 C 11 C 13 0 0 0 C 13 C 13 C 33 0 0 0 0 0 0 C 44 0 0 0 0 0 0 C 44 0 0 0 0 0 0 1 2 (C 11 C 12 ) 3 2 7 6 5 4 1 2 3 4 5 6 3 7 5 (12) For our analysis we split C to isotropic part C iso and anisotropic C an for which C an 33 =C 33 -C 11 C an 13 =C 13 -C 12 and (13) C an 44 =C 44 -C 66. This allows convenient parametrization of the level of anisotropy as C( ) =C iso + C an. (14) 4

Similar anisotropy is present also in the thermal expansion that differs in e 3 -direction from the others. For Sapphire the main anisotropy relates to C13 an that is about 25% of the corresponding isotropic value. For other components of C the anisotropic effects are smaller. Transversely isotropic material allows axisymmetric treatment only if the axis of symmetry and the axis of anisotropy coincide. Otherwise a non axisymmetric effect appears even if all the loading is axisymmetric. To estimate the significance of this effect we have to consider the change in material properties expressed in the axisymmetric frame when the plane of isotropy is tilted by angle. The components of the stiffness tensor in the rotated configuration are related to the components in the reference configuration by the relation c pqrs = l pi l qj l rk l sl c ijkl (15) where l ab are the components of an orthogonal rotation matrix [L( )]. Tilting of the material can be modeled by rotation in, say yz-plane. By definition the rotation leaves isotropic properties intact and for the considered anisotropic part the leading order of changes is sin( )sin( ) for small values of. Hence for relatively small rotations the axisymmetric approximation can still be used quite safely. 2.3 Simulated scenarios We first simulate the system in axisymmetric setting. The material is initially at constant temperature (2000C) and will be cooled with constant rate either from the seed crystal boundaries only or from seed crystal and the conical surface with controlled distribution of heat fluxes simulation parameter. Other boundaries are taken as thermally insulated. For the mechanical part the system is fixed with very weak springs from the boundaries of the uppermost surface to avoid artificial fixture induced stresses in the thermally loaded zone. It should be noted that any kind of support of the seed crystal (fixing either the z-displacements or all displacements at the lowermost surface) leads to stress peaks at the fixed boundary. The following scenarios were simulated: equal cooling power to seed crystal and cone isotropic material anisotropic material 75% of cooling to seed crystal, 25 % to cone surface isotropic material anisotropic material 5

Figure 2: Geometry and mesh used in computations. The temperature histories were the same for all comparable cases as the temperature evolution does not depend on mechanical deformations (in our simplified model) and hence presented only once. Between the scenarios we compared the following quantities: peak stress stress distribution in the symmetry plane stress distribution on the surface distribution of total displacement All these were evaluated after 3600 s of cooling. Moreover the evolution of temperature and stress on the symmetry axis was monitored for longer time. For 3D simulations the above scenarios were repeated and in addition we monitored stress distribution in a cross section and minima and maxima of stresses on the circumference as indicator of the numerics induced loss of symmetry. Finally for both cooling scenarios the 3D simulation was done for anisotropic material that was rotated 90 degrees (changing y- and z- coordinates). 3 Numerical considerations Figure 2 presents the used geometries and defined mesh for the computations. Numerical results have been simulated using Comsol Multiphysics software and many of the predefined configurations have been used. Meshes have been build using definition finer and solver configuration have been accepted as it is by default. Computational time used in axi-symmetrical setup was around 20 s and full 3D computing takes time approximately 10 minutes. Some of the used parameters are represented in the table 1. 6

Property Name Value Unit Density 3980 kg/m 3 Thermal conductivity k 35 W/(m K) Coefficient of the thermal expansion 7 8 10E 6 1/K Heat capacity at constant pressure C p 800 J/(kg/K) Table 1: Material parameters. 4 Analysis of the results and conclusions We have presented the main results in the Table 2. In our experiments we found up to 8% difference in the stress peak between isotropic and anisotropic material models. As a summary we can conclude that anisotropic material model should be used. If the material is aligned with the symmetry axis, the simulation can be easily done with axisymmetric model. In the full 3D-model we noticed that the grid effects cause artificial asymmetry and hide the anisotropic effects partially. As a whole 3D simulation would require quite fine grid to be able to give reliable results. The results depend, naturally, quite strongly on the boundary conditions. More information of actual cooling strategies would be needed. The simulated scenarios are quite extreme in that respect concentrating the heat flux to a small part of the crystal boundary. More even distribution of heat flux is likely to produce lower peak stresses. Due to strong shrinking of sapphire even minor blocking of deformations induces significant stresses. Hence for realistic modeling the supports and contact to them must be modeled carefully (contact surfaces allowing detachment during shrinking). Situations where the material anisotropy is not aligned with symmetry axis were studied in two ways. Numerically we studied the case where crystal is turned 90 degrees. In this case clear symmetry loss was anticipated. We noticed over20% asymmetry in peak stresses and higher Von Mises stress levels. For small deviations between the material and symmetry axes we concluded with sketchy perturbation analysis that for small rotations loss of symmetry is proportional to sin 2. Hence, given the challenges in 3D simulations, it may be more reliable to use anisotropic axisymmetric model even if the positioning of the seed crystal and its orientation are not perfect. As a summary of conclusions is that material anisotropy should be taken into the model. It has 10% effect to z-displacements and may effect simulations with realistic supporting mechanisms. Moreover, it effects to peak stresses. Material anisotropy can be implemented in axisymmetric code and it is not overly sensitive to perfect alignment to symmetry axis. However, there is no immediate need for 3D model. First of all, it is costly compared to axisymmetric model is same level of reliability is aimed to. Secondly, there are no dominant 3D effects present in the system apart from growing in wrong direction or from Grasshoff convection in melt phase. 7

Simulation Cooling Isotropy Peak stress Displacement Axisymmetric 50 Isotropic 3,07E+07 1,16E-04 Axisymmetric 50 Anisotropic 3,05E+07 1,26E-04 Axisymmetric 75 Isotropic 4,72E+07 1,41E-04 Axisymmetric 75 Anisotropic 4,71E+07 1,52E-04 3D 50 Isotropic 3,51E+07 1,21E-04 3D 50 Anisotropic 3,25E+07 1,32E-04 3D 75 Isotropic 5,36E+07 1,47E-04 3D 75 Anisotropic 4,97E+07 1,57E-04 5 Discussion Table 2: Peak stresses and displacements. Because this was small workshop research related to event of mathematical modeling of industrial problems, it is convenient to express some notes about the our modeling process. At the first, we focused on the question of comparison between axisymmetric and full 3D computations. The starting point was 3D simulation of thermal expansion induced stresses. Using extreme scenarios, we estimate the role of the (asymmetric) anisotropy. The lack of practical knowledge of the dimensions of real system drives us to working with world record scale dimensions. Next, we build up axisymmetric model with smaller dimensions. After some analytical approximations, we succeeded to estimate semi-realistic parameters to heat flows etc. Practically the estimation relies on the information that the product took one week time to cool down. In the workshop, we had limited time resources, and therefore some details are excluded from this report. 8

References [Buchanan, 2000] Buchanan, G. R. (2000). Vibration of truncated conical cylinders of crystal class 6/mmm. Journal of Vibration and Control, 6(7),985 998. [Chen et al., 2014] Chen, C.-H., Chen, J.-C., Chiue, Y.-S., Chang, C.-H., Liu, C.-M., & Chen, C.-Y. (2014). Thermal and stress distributions in larger sapphire crystals during the cooling process in a kyropoulos furnace. Journal of Crystal Growth, 385,55 60. [Dobrovinskaya et al., 2009] Dobrovinskaya, E. R., Lytvynov, L. A., & Pishchik, V. (2009). Sapphire: material, manufacturing, applications. Springer Science & Business Media. [Fang et al., 2013] Fang, H., Pan, Y., Zheng, L., Zhang, Q., Wang, S., & Jin, Z. (2013). To investigate interface shape and thermal stress during sapphire single crystal growth by the cz method. Journal of Crystal Growth, 363, 25 32. 9

APPENDIX Temperature Figure 3: Temperature fields using full 3D computations. Surface and cut-plane. 50/50 cooling. 10

Figure 4: Temperature evolution versus time. The behavior will be similar until the end. 50/50 and 75/25 cooling. 11

Figure 5: Temperature fields in surface and cut-off plane using full 3D computations. 25/75 cooling. 12

Von Mises Stress Figure 6: Isotropic and anisotropic Von Mises stress fields using axial symmetric approach. 50/50 cooling. 13

Figure 7: Isotropic and anisotropic Von Mises stress fields using axial symmetric approach, 3D-view. 50/50 cooling. 14

Figure 8: Isotropic and anisotropic Von Mises stress fields using full 3D computations. 50/50 cooling. 15

Figure 9: Isotropic and anisotropic Von Mises stress fields in cut-off plane using full 3D computations. 50/50 cooling. 16

Figure 10: Isotropic and anisotropic Von Mises stress fields using axial symmetric approach. 25/75 cooling. 17

Figure 11: Isotropic and anisotropic Von Mises stress fields using axial symmetric approach, 3D-view. 25/75 cooling. 18

Figure 12: Isotropic and anisotropic Von Mises stress fields using full 3D computations. 25/75 cooling. 19

Figure 13: Isotropic and anisotropic Von Mises stress fields in cut-off plane using full 3D computations. 25/75 cooling. 20

Figure 14: Asymmetric and symmetric anisotropic Von Mises stress fields using full 3D computations. 25/75 cooling. 21

Figure 15: Asymmetric and symmetric anisotropic Von Mises stress fields in cut-off plane using full 3D computations. 25/75 cooling. 22

Figure 16: Asymmetric anisotropicvon Mises stress fields using full 3D computations. 75/25 and 50/50 cooling. 23

Figure 17: Asymmetric anisotropic Von Mises stress fields in cut-off plane using full 3D computations. 75/25 and 50/50 cooling. 24

Figure 18: asymmetric and symmetric anisotropic Von Mises stress cut using full 3D computations. 25

Displacement Figure 19: Isotropic and anisotropic displacement fields using axial symmetric approach. 50/50 cooling. 26

Figure 20: Isotropic and anisotropic displacements using full 3D computations. 50/50 cooling. 27

Figure 21: Isotropic and anisotropic displacement fields using axial symmetric approach. 25/75 cooling. 28

Figure 22: Isotropic and anisotropic displacements using full 3D computations. 25/75 cooling. 29

Figure 23: Asymmetric and symmetric anisotropic displacements using full 3D computations. 25/75 cooling. 30

Figure 24: Asymmetric anisotropic displacements using full 3D computations. 75/25 and 50/50 cooling. 31

Evolutionary behavior Figure 25: Isotropic and anisotropic Von Mises stress evolution in the middle line of the system using full 3D computations. 50/50 cooling. 32

Figure 26: Isotropic and anisotropic Von Mises stress evolution in the middle line of the system using full 3D computations. 25/75 cooling. 33

Figure 27: Asymmetric and symmetric anisotropic Von Mises stress evolution in the middle line of the system using full 3D computations. 25/75 cooling. 34

Figure 28: Asymmetric anisotropic Von Mises stress evolution in the middle line of the system using full 3D computations. 75/25 and 50/50 cooling. 35