Feasibility study of modelling a virtual climate chamber with CFD

(1)

Feasibility study of modelling a virtual climate chamber with CFD

Christina Silfwerbrand

Engineering Physics and Electrical Engineering, master's level 2020

Luleå University of Technology

Department of Computer Science, Electrical and Space Engineering

(2)

Abstract

The company tests their industrial tools in a climate chamber to assure that tools meet the ISO-standard requirements and to assure personnel safety. Since it takes time to prepare the physical tool and to get a time slot to test it, this process was a bottleneck. Especially interesting is the temperature testing process, and the question of whether or not the entire tool actually reaches the reference temperature within a time interval has been asked. This is the master thesis project’s core. The time it takes for the entire tool to reach a uniform temperature distribution is called thermal soak time and was unknown at the beginning of the project. The aim was to find the thermal soak time, both for the climate chamber and for a virtual CFD-model of the climate chamber. A simulation model was done in the software Altair HyperMesh and validated with experimental data from the climate chamber on site.

The result of the experiments of the climate chamber showed the impact of all the unknown parameters and contributed to the feasibility study. Temperature cycles were run in the climate chamber to get thermal soak times for increase runs and decrease runs. This resulted in an understanding of how the climate chamber works and its limitations. As for the CFD-model, the simulation model of the climate chamber showed promising results of achieving a thermal soak time for all of the components of the tool. However, there are some limitations to consider when performing simulations in the specified software and with the CFD-model.

In conclusion, the thermal soak time for the best-case scenario with aluminum is determined to be approximately three minutes whilst, for the worst-case scenario, the same time is inconclusive since the simulation did not reach steady state due to time limitation of the project. However, the feasibility study of the climate chamber is complete, and proof of concept is confirmed for transforming a physical climate chamber into a virtual climate chamber. Further studies need to be executed to achieve a robust CFD-model of a virtual climate chamber.

(3)

Preface

This master thesis project is a 30 credit course for the Masters Programme in Engineering Physics and Electrical Engineering with a Master of Science in Computational Methods and Physics at Lule˚a University of Technology. That corresponds to 20 full work weeks during the autumn semester of 2019. This project course is under the Department of Engineering Sciences and Mathematics and has been done at a company in Stockholm.

Many thanks to Mustafa Guducu, simulation engineer at the company, for technical guidance and your expertise during the project. I would also like to thank Thomas Bejefalk Th¨orn, manager for the R&D Elec- tromechanics team, for general master thesis guidance and for giving me the opportunity to do my master thesis at the company. I am truly grateful for it, and I have learned so much during these months. Thank you for your dedication to this project.

I would also like to acknowledge Mokhalad Karim (Consultant to the Electromechanics team) for helping me with guidance for the experimental part of this project. Thank you, I would not have finished this project without your expertise and your ever so helpful spirit.

Thanks to Anna-Lena Ljung at Lule˚a University of Technology for being the examiner and for the sup- port during the project.

Many thanks to Thomas Timan (Lab Engineer) for helping me with preparations for the experimental part of this project. Lastly, I would like to thank the rest of the Electromechanics R&D-team at the company (Anders Nilhav, Daniel Hallberg, Johan Bjurkull, Ola Lindahl, Sebastian Karlsson, and Victoria ¨Osterlund) for all the Friday-pastries and for taking care of me from day one. I could not have wished for a better team to spend every workday with.

Christina Silfwerbrand Stockholm, January 2020

(4)

NOMENCLATURE

Constants

g Gravitational acceleration 9.815 [m/s²] R Individual gas constant 287.05 [J/kg · K]

Subscripts

i, j Counters

in Inlet component

out Outlet component

Variables

A Area [m²]

k Conductivity [W/(mK)]

ρ Density [kg/m³]

d Depth [m]

∇ Divergence [−]

µ Dynamic viscosity [kg/m · s]

v Flow field velocity [m/s]

F Force [m/s²]

h Height [m]

H Heat transfer coefficient [W/m²K]

EK Kinetic energy [J ]

ν Kinematic Viscosity [m²/s]

m Mass [kg]

˙

m Mass flow rate [kg/s]

l Length [m]

U Potential energy [J ]

P Pressure [kg/m · s²]

P r Prandtl number [−]

cp Specific Heat [J/kg · K]

T Temperature [K]

t Time [s]

E_H Thermal energy [J ]

v Velocity [m/s]

V Volume [m³]

Q Volumetric flow rate [m³/s]

w Width [m]

(5)

Abbreviations

AHW Altair HyperWorks

avg Average

CC Climate Chamber

CAD Computer Aided Design

CFD Computational Fluid Dynamics CFL condition Courant-Friedrich-Lewy condition

DAQ Data Acquisition System

HM HyperMesh

hh:mm:ss time in hours:minutes:seconds

min Minimum

max Maximum

R&D Research and Development RANS Reynold Average Navier-Stoke

RH Relative Humidity

IEC International Electrotechnical Commission ISO International Organisation for Standardardisation

Pr Prandtl number

Re Reynolds number

SA Spalart-Allmaras turbulence model

SEK Svensk Elstandard

SIS Swedish Standards Institute

SL SimLab

SST Shear Stress Turbulence model

temp Temperature

ts Time Step

(6)

1 INTRODUCTION

This section gives a brief introduction to the company and the business area followed by an introduction to the master thesis project. The aim and objectives are presented followed by the delimitation of the project.

The company was established 1873 in Stockholm, Sweden, and provides customers with innovative solutions of compressors, vacuum solutions, generators, pumps, power tools, and assembly systems.[1] In 2018, the company is a global and diverse group with approximately 37 000 employees representing different cultures in over 180 countries. The company is divided into several business areas with offices all over the world but with headquarter located in Nacka, Sweden. The organizational structure is seen in figure 1 with the four business areas: Compressor Technique, Vacuum Technique, Industrial Technique, and Power Technique.

Figure 1: Organizational structure of the company with the four business areas divided into divisions.

This master thesis project was done under the division MVI Tools and Assembly System at the Department of Mechatronics in the research and development (R&D) team Electromechanics. The office is located in Nacka, Stockholm, with production in both Tierp, Sweden, and Budapest, Hungary.

The company produces industrial tools that require high tool quality as well as maintaining specific safety standards to withstand the exposed climate conditions and for personnel safety. To verify the robustness of the tool, the requirements were verified by exposing the tools and the parts to the bounding temperatures in a climate chamber. The climate chamber on-site in Nacka was already in use but required further analysis to remove unnecessary uncertainties regarding unknown parameters of the climate chamber. Identifying the critical parameters, was also important to be able to begin a digitalization process of the physical climate chamber and developing a virtual climate chamber. The previous guideline for temperature testing in the climate chamber on-site required higher accuracy and to be adapted to the international standards used by the company. The temperature in the climate chamber was cycled between the specified ISO standard

(9)

temperature levels, but the thermal soak time, i.e., the time it took for the entire tool to reach a uniform temperature, was unknown. Thus, there was a significant uncertainty if the entire tools were exposed to the specified temperatures, and this uncertainty needed to be reduced to ensure that the tools met the set requirements. There was also a time aspect to consider, since not knowing the soak time lead to unnecessary long testing times. So, to improve both the uncertainty of the temperature testing and the testing time, this master thesis project was requested by the company.

1.1 Aim of Master Thesis

The primary mission was to create a stable and robust computational fluid dynamics model (CFD-model) that can be used for future updates of the model and further simulations of other industrial tools. This aim required an understanding and identification of all the important parameters and variables regarding the conversion from a physical climate chamber to a virtual one. The second aim was to define an outline for a thermal soak time matrix as a guide for evaluating the company tools in terms of test time in the climate chamber to primarily improve the usage of the chamber. These aims required an understanding of how personnel works today when tools are temperature tested and decide on which tool to be chosen for this project since over 4500 tools were present in the database. It also required an understanding of the current standards, regarding temperature and humidity, used for tool testing. More specifically, the tasks to be done are listed below:

• Create an initial CFD-simulation model of the climate chamber present at the company

• Validate the simulation model with IEC 60068-2-38 standards for temperature and humidity references with suitable boundary conditions and with additional experiments in the climate chamber.

• Measure experimentally to find suitable boundary conditions since the impact of the fan inside the climate chamber was unknown, resulting in lack of information of the flow rate for example.

• Investigate a thermal soak time guideline

Thus, the purpose of this project was, in short, to perform a feasibility study and to confirm proof of concept for the CFD simulating testing method.

1.2 Delimitation of Project

The main delimitation of the master thesis project was that the company’s tool product catalog consisted of more than 4500 items as of 2019. Since the project had a limited time frame, one tool was chosen and simulated to verify the CFD model.

Information about the fan system in the climate chamber was unknown, and therefore, there was con- siderable uncertainty about the fan’s impact on the result and the system since the flow rate parameter was unknown. A sensor system was developed to decrease some of the uncertainty regarding that boundary condition. However, the sensors introduced another task and uncertainty since the sensors were evaluated too.

The company used two software simultaneously but one software, Altair Hyperworks 2019.2, was used for the simulations to reduce the workload with two separate software systems.

(10)

1.3 Previous work

The climate chamber was already in use at the company at the beginning of the project. Therefore, a rough guideline had been written with a testing time of two hours with a temperature interval between 0^◦C to 80^◦C. These settings were chosen as a result of the most common materials present in the tools, a thermoplastic rubber and different electronic components, operating within the normal temperature range for industrial tools. With sensitive electrical components, an accelerated testing procedure was initiated with a temperature span of −40^◦C to 80^◦C to get comparable results. [2]

1.4 Readers Guide

Section 1: Introduction gives a basic introduction to the master thesis project and what the company had done previously for pre-study to the project.

Section 2: Theoretical Background offers a basic understanding of the physics relevant to the project.

Section 3: Theory behind project specific components presents and explains the mechanical components along with the climate chamber components and material properties.

Section 4: Method introduces the methodology both for the simulation process and with the experiments performed in the climate chamber on-site. Motivations for choices are also presented in this section.

Section 5: Result presents the result from the experiments in the climate chamber and the trends from the simulation of the CFD-model.

Section 6: Discussion gives an insight in limitations of the model and how accurate the result can be interpreted.

Section 7, 8: Conclusion and Recommendations for future work are self explanatory.

(11)

2 THEORETICAL BACKGROUND

Computational fluid dynamics is an important and powerful tool for engineers in all fields. Due to its many applications, the user carefully needs to investigate and choose parameters and models to achieve the correct result. This section briefly explains the theoretical physics behind fluid mechanics and the simulation methods with its models.

2.1 Governing equations

The governing equations used for fluid mechanic cases represent laws for momentum, energy, and mass conservation, which is also commonly known as the Navier-Stokes equations. The conservation laws are presented in equations 1 to 6.

Law of conservation of momentum

This conservation law states that the rate of change of momentum in a region equals the sum of forces in the same region. This relation is described mathematically in equation 1 with Fidenoted as body force per unit mass and Ri as surface forces per unit area and ρvi momentum per unit. By also applying Reynolds transport theorem, the equation correlates to the form of Newton’s second law for a fluid volume: mass times acceleration equals the sum of forces.

Z

V (t)

ρDvi

DtdV = Z

V (t)

ρF_idV + Z

S(t)

R_idS (1)

The equation consists of two volume integrals and one surface integral. To proceed the derivation, Ri, needs to be transformed to a volume integral via the definition of the stress tensor. Mathematical derivations are not part of this project’s scope, so for further references the reader is referred to [3]. The final expression of the stress tensor term is stated in equation 2.

Ri= Ti1n1+ Ti2n2+ Ti3n3+ = Tijnj (2)

The final expression for the momentum equation of an arbitrary fluid volume is summarised in equation 3.

ρDvi

Dt = ρFi+∂Tij

∂xj

(3)

where vi denotes the velocity, ρ the density of the fluid, Fi the forces, and Tij the stress tensor term.

Law of conservation of energy

The total energy of a system is a conserved quantity, which means that regardless of the event, the total energy remains unchanged in an isolated system. The distribution of energy will change form in kinetic energy EK, gravitational potential energy Ug, elastic-spring-potential energy Us, and heat/thermal energy EH. The energy conservation law is summarized in equation 4, where the subscripts i and f denote the initial and final state, respectively.

EK,i+ Ug,i+ Us,i = EK,f+ Ug,f+ Us,f+ EH,f (4)

(12)

Law of conservation of mass

This law states that mass cannot be created or destroyed, only rearranged between components within an isolated system and is expressed in fluid mechanics as equation 5.

∂ρ

∂t + ∇ · (ρv) = 0 (5)

where ρ denotes the density, t denotes the time, ∇ denotes the divergence operator and v the flow velocity field. Thus, for a closed surface in the system, the change of mass in time enclosed by the surface equals the same amount of mass that traverses the surface where the sign denotes the direction. In the entire isolated system, the same reasoning can be applied for the total mass M : the sum of masses in the system remains constant over time.

dM dt = d

dt Z

ρdV = 0 (6)

2.1.1 Navier-Stokes equations

With the law conservation equations, velocity and pressure distributions are achieved, which in turn are used for further calculations of parameters relevant to the specific fluid mechanics problem. The Navier-Stokes equations will take on different forms depending on the flow type along with other parameters to a case to case basis. In this master thesis project, the flow is assumed to be turbulent, with an incompressible viscous flow of air and water in three dimensions. An incompressible flow states that the density remains constant, implying that there is no volume change for a fluid particle. Thus, the Navier-Stokes equations for this case are presented in equations 7 to 9.

∂u

∂t + u · ∇u = −1

ρ∇P + ν∇²u + F (7)

Equation 7 represents the continuity (mass) equation where ρ denotes the density, t denotes the time, u the velocity, p the pressure, ν the kinematic viscosity and F the forces.

∇ · u = 0 (8)

Equation 8 represents the momentum equation where u denotes the velocity. As for the energy equation, it decouples from the Navier-Stokes equations above for an incompressible flow. By using the thermal energy equation in the definition of enthalpy in a dimensionless form, the final expression becomes as in equation 9.

DT Dt = 1

Re 1

P r∇²T (9)

Thus, equation 9 represents the energy equation where t denotes the time and T the temperature. Further- more, Re denotes Reynolds number and P r the Prandtl number.[3]

Note that variations of the Navier-Stokes equations will occur in other specific cases along with added turbulence modeling equations. The equations presented is an attempt to explain the complexity of the

(13)

fluid mechanics modeling and the physics and equations behind it. Hence, the importance of numerical computations directly performed by CFD-software. The Navier-Stokes equations are expensive to compute since additional terms, i.e., Reynolds stresses, may appear and affect other terms causing a chain-reaction.

One approach to avoid these expensive computations is to average the equations rather than the specific solutions. These equations are known as the Reynolds Averaged Navier-Stokes (RANS) equations. However, the turbulence is omitted in the numerical RANS scheme causing the system of equations to be incomplete.

For completion, a turbulence model is added to the system to complete the equations. However, further derivations of turbulence and Navier-Stokes equations are not part of this master thesis’s scope. For further investigation, the reader is referred to more in-depth articles and books on derivations of the Navier-Stokes equations. [4]

2.2 Reynolds number

As mentioned in the previous section, there are different types of flows. One approach to distinguish flows is in laminar and turbulent flows, which is decided by the Reynolds number: a dimensionless number defined as the ratio between the inertia forces to the viscous forces. This ratio is often interpreted as the dynamic pressure to the shear stress resulting in the formula in equation 10.

Re = ρvL

µ =µ/ρ = ν =vL

ν (10)

where ρ is the density of the fluid, L is the characteristic length, ν is the kinematic viscosity and v is the average velocity. Whether or not the flow is turbulent or laminar is partly determined by the value of Reynolds number.[5]

The inertial forces are dominant in turbulent flows, i.e., flows with high Re, and flow instabilities like chaotic eddies and vorticity are usually generated, causing a non-smooth flow. In laminar flow, though, the viscous forces are dominant to the inertial forces causing the flow to be smooth. The visual difference between the two flow types is shown in figure 2.

Figure 2: Visual difference between laminar and turbulence flows. [6]

2.3 Turbulence modelling

As stated, the Navier-Stokes equations are expensive to solve in their original form, and the averaged RANS-equations will not be sufficient to solve the system of equations since the Reynolds stress tensor

(14)

terms introduce additional unknown parameters. With turbulence modeling, new equations are added to the system to approximate unknown correlations in terms of flow properties. Over the years, many turbulence models have been introduced, creating subcategories depending on the engineering case. To mention a few common ones: algebraic models, one-equation models like Spalart-Allmaras and two-equation models like the well-known k −, k −ω and SST. [7] For this project, the one-equation model Spalart-Allmaras turbulence model has been chosen in the simulation process by recommendations from experts of the used software.

2.3.1 Spalart-Allmaras turbulence model

In aerodynamics and turbomachinery applications, the Spalart-Allmaras turbulence model is one of the most used models due to the proximity the simulation result has to experimental results. [6] Another advantage, compared to the more extensive two-equation models, is that the SA model is computationally faster. How- ever, the SA-model is not calibrated for more general cases since it was developed mainly for aerospace applications and can produce large errors for free shear flows. [8] Nonetheless, there is still no turbulence model that can describe all flow types accurately. Hence, the importance of closely investigating the proper turbulence model and Navier-Stokes equations.

As stated, the derivation of the turbulence model is not part of the scope of this project. However, the basic principle of the SA model is that the additional terms from the RANS system of equations, the Reynolds stresses, are correlated to an eddy viscosity term. The RANS equations, along with the Reynolds stresses and the eddy viscosity term, solve the system of equation. D.C. Wilcox explained in his book Turbulence Modeling for CFD [7] how the eddy viscosity term was used to calculate the Reynolds strain tensor term, even called the Boussinesq approximation. The Spalart-Allmaras model equation is of form according to equation 11 where νT is the turbulent kinetic viscosity, Γ is the diffusion coefficient and Sφ is a source term.

∂

∂t(ρφ) + ∇ · (ρVφ) = ∇ · (Γ ∇φ) + Sφ (11)

2.4 Mesh generation

In CFD-modelling, the flow field around the flow-body is divided into several discrete points with lines connecting the points generating a grid. These flow-field variables grid points are then used by a numerical solver to calculate the discrete partial derivatives. There are several different types of mesh/grids, and depending on which application used in the case, the structure of the grid needs to be adapted to capture the physics. Thus, the mesh density is an important parameter in CFD-modelling since the denser the grid, the more accurate the solver will solve the equations. However, too many grid points will only result in longer computational time without an analogous increase inaccuracy. Thus, engineers need to check both types of mesh as well as the accuracy needed for the specific geometry and case.

Altair software only supported the unstructured grid for this project’s particular problem and settings.

Therefore, the discretization process was sensitive, since the number of mesh elements and the mesh quality might affect the result since the software was sensitive to failed elements in an unstructured grid. A good mesh quality should, therefore, be achieved before initiating simulations to get a stable CFD-model.[4] Some parameters important to observe are listed below in table 1.[9]

(15)

Table 1: Mesh parameters important to the mesh quality

Parameter Explanation

Stretch Ratio between the length of two adjacent cells.

Skew Difference between the cell shape and the shape of an equilateral cell of equivalent volume.

Skew fluent Same reasoning as for skew but in a fluid, i.e., for a 3D element.

Aspect ratio Measurement of the cell’s stretching computed as a ratio of the maximum value to the minimum value of the distance between the cell centroid and the face centroid and the distance between the cell centroid and the nodes.

Fluid aspect ratio Same reasoning as for aspect ratio but for a fluid, i.e., for a 3D element.

These elements parameters’ failure rate depending on the settings chosen in the simulation software: for example, the base size of the element, number of prism layers, the thickness of the prism layer, the total amount of cells, etc.

2.5 Convection

Convection is an essential part of this case since there is movement in the fluid (air in the climate chamber), resulting in convection with surrounding objects, i.e., tool. There are two types of convection: natural and forced. With the presence of an external source, for example a fan, forced convection will take place since the fan forces movement in the fluid. As for natural convection, the main key is density variations in the flow due to temperature variations. A warm fluid has a lower density than a cold fluid resulting in the hot molecules to travel upwards and the cold molecules downwards in the fluid. The convection phenomena between solid and fluid objects are illustrated in figure 3.

Figure 3: Convection between a fluid and solid.

Mathematically, convection is described with the formula stated in equation 12.

Q = hA(T˙ f− Ts) (12)

(16)

where ˙Q is the heat transferred per unit time, A is surface area, h denotes the heat transfer coefficient, Ts

denotes the surface temperature, and Tf is the temperature of the fluid. [10]

2.6 Conduction

When heat is transferred directly through molecular collisions, the heat transfer form is known as conduction.

This form is also a vital part of the physics of this investigation since the tool and sensors are in contact with other objects and materials. The basic principle of conduction is that an area of kinetic energy will transfer thermal energy to a lower kinetic energy area. The particles with higher kinetic energy have a higher speed than the low kinetic energy particles, resulting in the slower particles will increase their kinetic energy as a result of collisions with the higher kinetic energy particles. The cross-section and travel path are two important parameters since the greater the exposed surface area is, the more heat is lost as well as, the greater the size of an object, the more energy is required to heat it. Other factors that affect the heat conduction process are temperature gradient and physical material properties. The rate of conduction is calculated via equation 13.

Q =kA(T_hot− T_cold)

d (13)

where Q is the heat transferred per unit time, k denotes the thermal conductivity, A the area of the heat transfer, Thot and Tcold denotes the temperature of the hot and cold region respectively and d denotes the thickness of the heat transfer barrier.

2.7 Thermal soak time

For many products, electrical components are the most sensitive regarding failures of repetitive temperature changes in a testing process.[11] The majority of the company’s tools consist of variations of electrical components. Therefore these conditions need to be carefully tested and verified in real-life experiments to assure the result. The thermal soak time is achieved by running a temperature cycle and observing the time it takes for a tool to reach a uniform temperature distribution.

Thermal soak time is the holding time of an object at a constant temperature until there is a uniform distribution of heat throughout the material. The uniform distribution is achieved by soaking the object for a definite time interval in a fluid. As the temperature is uniformly distributed, the object has achieved its soaking time, which is represented as a temperature cycle with a heating process, soaking time, and cooling process.

A graph over a simplified idealized thermal soak time cycle is seen in figure 4. However, in non-simplified non-idealized conditions, there will be ripples and disturbances in the output signal. The red dashed line represents the soak time, i.e., the time for the entire tool to uniformly reach the specified temperature. The soak time was unknown at the beginning of the project and was intended to be simulated in the software Altair Hyperworks 2019.2.

(17)

Figure 4: A simplified idealised thermal soak time graph to describe the general theory behind the project.

X-axis represents the time in hours and the y-axis represents the temperature in degree Celsius.

2.8 Relative Humidity

Relative humidity (RH) is described as the ratio of the current absolute humidity to the highest possible absolute humidity, i.e., how much water vapor is in the air compared to how much water vapor the air could hold at the current temperature. Thus, relative humidity is temperature-dependent, which means that the amount of moisture in the air, at 20^◦C and 50% RH, does not equal the amount of moisture in 10^◦C and 50% RH. As the air temperature is reduced, the ability to hold moisture is reduced, too, resulting in the lower the temperature, the higher the RH for a given amount of water contained in the air. [12]

2.9 Simulation stability

By observing the output from the simulation, the stability of the solution can be studied as all of the components in the temperature graph settled to the final temperature: when all component sensors have stabilized to the final temperature, the temperature in the system has stabilized too. This theory is displayed in figure 5.

(18)

Figure 5: Temperature sensor plot over the simulation goal.

Moreover, to achieve convergence, the mass flux at the inlet and outlet should stabilize: the mass flux difference at the inlet and outlet should be zero. With this statement, the continuity equation is also verified.

∆ ˙m =mout˙ − ˙min= 0 (14)

The pressure also need to be stabilised to reach convergence. Like the mass flux, the pressure at the inlet needs to stabilise to reach a zero change of the inlet and outlet, explained in equation 15.

∆p = p_out− p_in= 0 (15)

A convergence limit is set in the simulation settings in the software, and relevant parameters in the physics of the simulation model need to reach a residual less than the convergence limit. By default, Altair HyperMesh has a limit of 0.001, and a convergence limit of this value is considered a good result, according to Altair application specialists.

2.9.1 CFL-conditioning

A Courant-Friedrichs-Lewy (CFL) condition is used to stabilize numerical methods for convection and wave phenomena. The CFL condition states that the distance information travels during the specified time step within the mesh must be less than the distance between mesh elements. [13] The CFL condition is a tool to check for stability in the model. However, it should be combined with other, more restrictive stability conditions. The derivation starts with considering the linear convection problem of a quantity, u, and a velocity a.

∂u

∂t + a∂u

∂x = 0 (16)

(19)

By applying a first order explicit upwind scheme and performing a Taylor series expansion, the dimensionless quantity, the Courant number, is derived to a final expression as in equation 17.

C = a∆t

∆x (17)

To avoid a negative numerical viscosity the Courant number should be less or equal to 1; C ≤ 1. [14] Thus, by lowering the Courant number, the condition can be satisfied by either lowering the timestep or a coarser mesh, in accordance to equation 17.

(20)

3 THEORY BEHIND PROJECT SPECIFIC COMPONENTS

This section explains the physical components in the experimental setup and how the components work. The international standards used at the company are also presented, followed by a section for material properties.

3.1 International Standards

SIS (Swedish Standards Institute) is the main supplier of external standards for the company. Both IEC- and ISO-standards are channeled via an agreement with SIS and downloaded for internal use.[15] Excerpts from the international standards can be downloaded for private use from SIS’s website.

3.1.1 IEC - International Electrotechnical Commission

IEC 60068 is a framework for methods for environmental testing of electronic products and equipment.

The primary use for the IEC 60068 framework is to assess the performance ability under environmental conditions such as extreme cold and dry heat. The standard, IEC 60068-2-38, consists of three parts, but only the second part was used in this project, which represented the testing procedures. The other two parts represent a general guidance part and supporting document guidance. The IEC 60068-2-38 standard was released in 2009 and is the latest edition.[16] Sweden has its national organization, Svensk Elstandard (SEK), which is responsible for the standardization within the electric component field as well as coordinate international and European standardization. SEK is said to be the Swedish national committee of IEC.

Figure 6: For reference, the IEC standardisation temperature run the temperature sensors were initially tested against.

(21)

3.1.2 ISO - International Organisation of Standardisation

ISO is a framework for standards for products not containing electrotechnical components, i.e., for example, sound or vibration standards. All the company tools are verified by an ISO-standard, as well.

3.2 Mechanical components

3.2.1 Tools

In the product database of the company over 4500 tools are present ranging from battery drills to hydraulic tools to electric assembly tools.[17] The materials vary from tool to tool but are generally made of mostly a thermoplastic rubber and a type of steel. Two common tools produced by the company are shown in figure 7 below.

(a) A battery-driven drill from the catalogue (b) A vertical grinder from the catalogue Figure 7: Example tools from the company’s tool catalogue.

3.2.2 Climatic test chamber

The climatic test chamber was of model VC³4034 from V¨otsch Industrietechnik GmbH. The exterior of the test chamber is seen in figure 8a, and the interior of the test chamber is seen in figure 8b. Specifications of the climate test chamber used in this project are listed in table 2 below.

(22)

(a) Climatic test chamber, model VC³ 4034 from V¨otsch Industritechnik GmbH.

(b) Interior of the climatic test chamber with some important parts marked.

Figure 8: The exterior and interior of the climatic test chamber used in this project.

Table 2: Relevant specifications of the climate chamber VC³ 4034.

Parameter Value Entity

Area inlet 3.315×10⁴ mm²

Area outlet 1.056×10⁵ mm²

Diameter left entry port 50 mm Diameter right entry port 125 mm

Humidity range 10 to 98 %

Interior depth 760 mm

Interior height 750 mm

Interior width 580 mm

Temperature range −42 to 180 ^◦C

Test space volume 335 l

To perform measurements entry ports were placed on both sides of the climate chamber to be able to connect the sensors within the climate chamber to a measuring device. These entry ports are also marked in figure 8b above. The setup of the right entry port with an anemometer measurement device connected to the climate chamber is seen in figure 9 below. The entry port was isolated with a silicon foam plug with a small slit, where the anemometer cable exited.

(23)

Figure 9: Right entry port viewed from the right hand side

The airflow in the climate chamber is illustrated in figure 10. The airflow entered the climate chamber at the inlet under the metal floor and continued via a channel, bounded by a metal plate, to the front side of the chamber and up in the chamber’s free space. A fan extruded air at the top backside of the chamber to maintain constant pressure and humidity. A pressure compensation opening was also located on top of the entire test system to reassure constant pressure, and a water tank was placed under the chamber to regulate the humidity. For efficiency, the condensed demineralized water was filtered and reentered in the chamber, which meant that the test procedure could be left for a longer period. For negative temperatures, the chamber regulated the humidity and air extrusion to prevent frost on the inside of the chamber and on the humidity sensor. The entry port isolation foams also prevented the evaporator to ice up.

Figure 10: Air flow in the climate chamber

(24)

The climate chamber test space was made of high-gloss stainless steel, material number 1.4301, and the floor of material number 1.4404. [18] The climate chamber was run by a monitor ”Simpac Controller,” where a manual or automatic option for the program could be chosen. The manual program resulted in only humidity and temperature options, while the automatic option resulted in choosing or adding a program, for example, temperature cycling.

3.2.3 Outlet fan

The continuity equation for the mass flux, ˙m, in equation 18, was used for the relation between the inlet velocity and outlet velocity and is based on the law of conservation of mass in equation 5. The velocity from the extrusion of the fan is referred to as outlet velocity and denoted vout.

˙

m_in=m˙_out⇔ ρin· Ain· vin= ρ_out· Aout· vout (18) where ρ was the density of the fluid at the inlet and outlet, A was the area of the inlet and outlet and v denotes the velocity at the inlet and outlet. The subscripts in and out denote the inlet and outlet, respectively. By using the proper conditions for the continuity equation, specified in equation 19a to 19c, the mass flux equation was rewritten to equation 20. The area of the outlet was assumed to be a rectangle with the height, hout, and width, wout. The inlet area was also a rectangle with the height, hin, and width, w_in.

ρin= ρout, (19a)

A_in= w_in· h_in, (19b)

Aout= wout· hout, (19c)

Eq.(18) ⇒ win· hin· vin= wout· hout· vout (20) Solving for the velocity of the extrusion from the fan, vout in equation 21.

v_out= v_in· w_in· hin

wout· hout

(21) The flow rate of the fan was calculated with equation 22 along with measurement data from experiments on the fan.

Q_out = A_out· v_out= A_out· v_in· win· hin

w_out· hout

(22)

3.2.4 Anemometer

An anemometer was purchased to measure the inlet and outlet velocities in the climate chamber. The range, resolution, and accuracy of the measurement device are seen in table 3. The anemometer had art.nr. 48710 from Kjell&Company.

Table 3: Parameters and range of the anemometer used.

Parameter Temperature [^◦C] Velocity [m/s]

Range -30 to 60 0.3 to 45

Resolution 0.1 0.1

Accuracy ±1.5 ± 3%

(25)

3.2.5 Humidity sensor

The relative humidity is specified in percentage in a measurement cycle, and the humidity sensor is seen in figure 11. As the humidity is measured, the white cord is placed in the drain tube, and the humidity is obtained in real-time. The humidity sensor’s sleeve, i.e., the white cord, is continuously wetted by a pump distributing the right amount of water depending on the temperature.

Figure 11: Close up of the humidity sensor in the climate chamber.

3.2.6 Temperature sensors

Temperature sensors were used for measuring the temperature convergence for verification of the simulation model. Since the temperature on surfaces was supposed to be measured, contact temperature sensors were chosen. This type of sensors required physical contact with the objects and used conduction to monitor temperature changes via a voltage difference. K-element thermocouple sensors that operate in the temperature range of −200^◦C to 1250^◦C that used Nickel-Chromium and Nickel-Aluminium as conductors were chosen for the experiment. These types of thermocouple temperature sensors are constructed of glass fiber insulation and a welded exposed junction to improve the response time.[19]

The Seebeck Effect is the process in which a thermocouple sensor measures a voltage difference and converts it into a temperature scale. Thus, a thermocouple sensor consists of two wire legs from different metals joined together to form a junction. The measuring junction is connected to the surface, which temperature is going to be measured and the reference junction is connected to a body of known temperature. When the measuring junction is placed in contact with a hot or cold surface, a potential difference occurs, which is converted via a data acquisition system into a temperature. A simple illustration of the thermocouple wire setup is shown in figure 12 below.

(26)

Figure 12: Sketch over the thermocouple setup where J1 is the measuring junction, J2 is the reference junction, the blue dotted line is one metal and the red dashed line is the other metal. V1 and V2 represents the voltage of the two wires and Vout represents the output potential difference.

The output potential,Vout, is calculated as a potential difference between the two wires.

Vout= V1− V2 (23)

The temperature is then converted to Celsius or Kelvin via the software to the data acquisition system.

Temperature sensors were glued onto and into a tool to check if the temperature of the sensors on the tool followed the same trend as for the simulation and the reference temperature cycle in the climate chamber. If the experiment were successful, the temperature of each component would respectively turn to the specified temperature within a given tolerance from the simulation, the same trend as in figure 5. The tolerance from the IEC 60068-2-38 standard that needed to be met was ±2 K for the temperature and ±3% for the relative humidity.

3.2.7 Data acquisition system

The data acquisition system (DAQ) used was of model 34972A from supplier Keysight and could hold up to 48 sensors. The corresponding software, BenchLink Data Logger, has several options depending on which type of sensors used in the experiment to obtain the correct entity. The datasheet stated accuracy of ±1^◦C for type K-thermocouples, which is less than the IEC 60068-38 standard limit of ±2K. [20]

3.3 Material properties

The ideal gas law, in the equation below, states that material properties of fluids and gases are temperature- dependent for properties like density and pressure.

p = ρ · R · T (24)

Where p is the pressure, ρ denotes the density of the material, R is the individual gas constant, and T is the temperature of the material. However, the pressure remains constant in the climate chamber during a temperature cycle leaving the relation between density and temperature to be the only changing variable.

(27)

3.3.1 Air

So, the density of air graphs used for the simulations is seen in figure 13a. The specific heat, conductivity, and dynamic viscosity of air are also temperature-dependent and are seen in figure 13.

(a) Plot of the temperature dependence of density for air.

hejhejhejhej

(b) Plot of the temperature dependence of specific heat for air.

(c) Plot of the temperature dependence of thermal conductivity for air.

(d) Plot of the temperature dependence of dynamic viscosity for air.

Figure 13: Graphs of the temperature properties of air in the temperature range from −10^◦C to 100^◦C.

(28)

3.3.2 Water

The density of water graphs used for the simulations is seen in figure 14a. The specific heat, conductivity, and dynamic viscosity of water are also temperature-dependent and are seen in figure 14.

(a) Plot of the temperature dependence of density for water. hejhejhejhej

(b) Plot of the temperature dependence of specific heat for water.

(c) Plot of the temperature dependence of dynamic viscosity for water.

Figure 14: Graphs of the temperature properties of air in the temperature range from 0^◦C to 100^◦C.

(29)

4 METHOD

4.1 Choice of tool

As of September 2019, the company’s tool database contained of approximately 4500 tools. Since time was the limiting factor, and there was no pre-study done on how long it would take to simulate a temperature cycle, only one tool was chosen for verification of the simulation model. The tool chosen was a pneumatic pistol drill with simple geometry and no electrical components to simplify the modeling process.

4.2 Software

Altair Hyperworks 2019.2 was used as a resource for the CFD simulations, and Altair HyperWorks subse- quently consists of several different software. The workflow of Altair is seen in figure 15 and the purpose of the software described in table 4.

Figure 15: Flow chart over the workflow of Altair.

Table 4: Altair softwares and their usage.

Software Purpose Inspire Concept design

Geometry modifications SimLab Initial meshing

Defining fluid volume Load case setup HyperMesh Refining mesh

Advanced solving settings AcuSolve Temperature/flow solver HyperView Post-processing tool

(30)

4.3 Simulation setup

The climate chamber was modeled as a three-dimensional box in Altair Inspire with settings from section 4.3.1 and boundary conditions from section 4.5 along with an imported Creo CAD design of a tool within.

(a) 3D climate chamber box model (b) 3D box model in the x- and y-plane Figure 16: 3D box model made in Inspire with the CAD model of the tool. The interior tool is to showcase the placement of the CAD design.

The inlet was assumed to have a uniform inlet velocity on a rectangular area. The outlet area was perforated, as seen in figure 17, and was assumed to be 80% of the total area of the rectangular outlet.

Figure 17: Outlet with perforated area

(31)

4.3.1 Simulation settings

The simulation model was set to SI entities, i.e., length scale in meter, mass in kilograms, time in seconds and thermodynamic temperature in Kelvin. By using equation 10, a Reynolds number of approximately 70 000 was calculated, which indicated a turbulent flow. Thus, the SA turbulence model was chosen in Altair HyperMesh. The simulation settings for the final model of the virtual climate chamber modelled in HyperMesh are listed in table 5.

Table 5: Simulation settings for climate chamber model in Altair

Parameter Setting Value Entity

Walls Adiabatic No flux —

Outlet Hydrostatic pressure 0 [P a]

Body force Gravitation -9.815 ˆy [m/s²]

Inlet Velocity 8.0 ˆz [m/s]

Simulation time Max iterations 2500 —

Iteration steps Initial time increment 0.5 [s]

Temperature equation Advective diffusive — —

Radiation equation None — —

Flow state Transient state — —

Turbulence model SA — —

4.3.1.1 Forced convection

The simulation model was subject to forced heat convection since the velocity of the fluid was accelerated with an inlet velocity in the climate chamber, i.e., an external source. [10]

4.3.1.2 Transient state

Since the thermal soaking time was one of the primary missions to obtain, a transient state was used in the simulations to be able to observe how the temperature distribution changed over time.

4.3.2 Geometrical simplifications

To simplify the meshing process, geometry simplifications of the CAD model were implemented. The results of the simplifications are displayed in figure 18d and 18b compared to the original CAD model in figure 18c and 18a.

(32)

(a) Original CAD drawing of tool. (b) Tool after geometry simplifications.

(c) Original CAD interior of tool.

(d) Tool after geometry simplifications. The blue internal parts are the internal air and the green part is the tool.

Figure 18: Tool before and after geometry simplifications to get an efficient meshing process.

(33)

4.4 Meshing process

The 3D-model was built in Inspire and imported as a Parasolid to HyperMesh (HM), where the meshing process was initiated. Firstly, the important surfaces on the model were closely investigated, and relevant Mesh Controls implemented to preserve surface shapes and mesh entities. The grading factor was set to 1.2, i.e., the adjacent element was allowed to be 20% larger than the current element to achieve a stable tetrahedral (TET4) mesh. A list of the mesh settings is seen in table 6. The settings in table 6 resulted in approximately 300 000 elements in the model.

Table 6: Mesh settings for the 3D-model.

Setting Value Entity

Grading factor 1.2 [%]

Surface mesh average element size chamber 40 [mm]

Surface mesh average element size tool 1 [mm]

Surface mesh average element size inlet & outlet 2.5 [mm]

Volume mesh average element size chamber 40 [mm]

Volume mesh average element size tool 1 [mm]

Boundary layer condition All surfaces except

inflow & outflow

—

First boundary prism layer thickness 0.1 [mm]

Boundary layer growth rate 1.2 [%]

Final layer height to Base Ratio 0.8 —

Number of boundary layers 5 —

With the function called Quality index in the HM software, the mesh quality was verified and modified. A built-in function Cleanup tools, seen in the figure above, was used in the Quality index function to improve the mesh quality. With that function, badly placed nodes could be moved in the surface mesh step to improve the element and mesh quality before volume bodies were created. If the mesh quality, i.e., ratio failed elements, was less than tolerance of 1%, the mesh was said to be independent, and no further mesh studies needed to be performed.

The result of the mesh quality cleanup is seen in table 7.

Table 7: Result of meshing process and cleanup process of the mesh quality.

Setting Elements failed before cleanup

Elements failed after cleanup

Percentage failed elements

Stretch 0 0 0%

Skew 81 0 1%

Aspect ratio 107 326 0%

The final mesh of the climate chamber model is seen in figure 19.

(34)

Figure 19: A cross-section of the mesh of the handle of the tool in red and its surrounding fluid in purple with boundary layers in the transition area between materials.

4.5 Boundary conditions

4.5.1 Inlet and outlet velocities of the climate chamber

The inlet velocity and outlet velocity were measured experimentally with an anemometer placed in the climate chamber since these boundary conditions were unknown.

Outlet velocity measurements: To verify the temperature independence of the outlet velocity, a cycle with seven temperatures was performed. The anemometer was placed in the center on the rack at a distance of 0.15 m from the outlet area. The seven velocities were measured in real-time as the temperature cycle program ran. After confirmation of the temperature independence, the actual outlet velocity was measured in real-time with the anemometer. The anemometer was now placed on the oven rack 0.12 m from the outlet in the middle of the width of the climate chamber with the display connected to the outside via an entry port, and the outlet velocity could be determined.

Inlet velocity measurements: Then the anemometer was placed at the inlet, closest to the floor of the climate chamber. The same procedure was used for the outlet velocity. The inlet and outlet velocity were then determined, and boundary conditions for the simulation model obtained.

(35)

4.5.2 Wall conditions

The walls of the climate chamber were assumed to be adiabatic, i.e., there was no flux or heat transfer with the outside surroundings of the fluid. The inner walls, i.e., tool walls and internal air walls, were assumed to have no flux.

4.6 Experiment

4.6.1 Thermocouple sensors

The thermocouples used, seen in figure 20, were of model RS PRO Type K Thermocouple 2m Length, 1/0.3mm Diameter with order code 814-0156 and operated in temperature range −60^◦C to 350^◦C.

Figure 20: Thermocouple sensors used for the experiments.

4.6.2 Verification tool

The tool chosen for the verification experiment is displayed in figure 21a. This choice was based on several factors, including the presence of no electrical components, the size, materials, and the ability to drill holes for the temperature sensors.

(36)

(a) A pneumatic aerospace tool was chosen for the experiments The drilled holes were used to fit the sensors in the product.

(b) The sensors mounted and secured on the tool with superglue and thermal resistance glue.

In total, 12 sensors were superglued into the tool and then secured with the thermal glue Thermally Con- ductive Oxime Cure RTV from the supplier Electrolube. After 24 hours, the thermal glue had solidified, and the experiment could proceed. The final look of the tool is displayed in figure 21b.

4.6.3 Process in the climate chamber

The verification tool was mounted inside the climate chamber and the sensors connected to the DAQ. The IEC 60068-2-38 standard program, seen in figure 6, was slightly modified according to table 8 to fit the testing time slot and then implemented in the climate chamber. This cycle was executed to assure the temperature sensors’ performance and how well the temperature in the tool followed the known IEC temperature testing cycle. A dash in the time-column meant that the climate chamber optimized the temperature increase to reach the reference temperature as fast as possible.

(37)

Table 8: Modified IEC temperature cycle for the climate chamber experiment Step Temp [^◦C] Tolerance [K] RH [%] Tolerance [%] Time [h:min]

0 25 ±2 93 ±3 initial

1 25 ±2 93 ±3 1:00

2 25 to 65 ±2 93 ±3 2:00

3 65 ±2 93 ±3 3:00

4 65 to 25 ±2 88 ±3 1:30

5 25 ±2 92 ±3 0:20

6 26 ±2 93 ±3 0:05

7 26 to 65 ±2 93 ±3 1:55

8 65 ±2 93 ±3 3:0

9 65 to 25 ±2 85 ±3 2:0

10 25 ±2 93 ±3 0:30

11 25 ±2 93 ±3 1:30

12 25 to −10 ±2 88 ±3 0:30

13 -10 ±2 0 – 3:00

14 −10 to 25 ±2 93 ±3 0:50

15 25 ±2 93 ±3 2:00

(a) Interior setup with the tool and the attached temperature sensors exiting via the right entry port.

(b) The entire experiment setup with the climate chamber, the DAQ and a computer to retrieve the data.

Figure 22: The verification experiment setup.

(38)

After the entire IEC 60068-2-38 temperature cycle was performed, parts of the cycle were initiated to achieve the maximum temperature increase and decrease of the climate chamber to get the experimental thermal soaking time of the sensors. Thereafter, the following cycles, described in table 9 and 10, were performed.

Table 9: Temperature increase cycle for the experiment in the climate chamber.

Step Temp [^◦C] Tolerance [K] RH [%] Tolerance [%] Time [h:min]

0 25 ±2 93 ±3 initial

1 25 to 65 ±2 93 ±3 min

2 65 ±2 93 ±3 0:20

3 65 to 25 ±2 88 ±3 min

4 25 ±2 93 ±3 1:00

Table 10: Temperature decrease cycle for the experiment in the climate chamber.

0 25 ±2 93 ±3 initial

1 25 to −10 ±2 88 ±3 min

2 -10 ±2 0 – 0:30

3 −10 to 25 ±2 93 ±3 min

4 25 ±2 93 ±3 0:10

After analyzing the data, modifications in the temperature cycles were executed and another set of experiments were performed according to tables 11 and 12.

Table 11: Temperature increase cycle for the second set of experiment in the climate chamber.

0 25 ±2 93 ±3 initial

1 25 to 65 ±2 93 ±3 min

2 65 ±2 93 ±3 1:30

3 65 to 25 ±2 88 ±3 min

4 25 ±2 93 ±3 1:00

Table 12: Temperature decrease cycle for the second set of experiment in the climate chamber.

0 25 ±2 93 ±3 initial

1 25 to −10 ±2 88 ±3 min

2 -10 ±2 0 – 1:30

3 −10 to 25 ±2 93 ±3 min

4 25 ±2 93 ±3 1:00

(39)

4.7 Final CFD simulation of the tool

With the same settings as previous stated in table 5, the final simulation for the verification tool was initiated according to table 13. Due to time limitations, only the temperature increase cycle was simulated and is presented below.

Table 13: Temperature increase cycle for the final simulation of experiment in the tool in the virtual climate chamber model.

Time Step Temp [^◦C] RH [%] Duration [hh:mm:ss] Real time [hh:mm:ss]

0 25 93 initial initial

2 25 93 00:00:01 00:00:01

1804 25 to 65 93 00:15:00 00:15:01

12604 65 93 01:30:00 01:45:01

14404 65 to 25 88 00:15:00 02:00:01

21604 25 93 01:00:00 03:00:01

The temperature cycles in table 13 would result in a total of 21604 time steps. Each time step takes approximately 320 seconds to iterate over resulting in a total simulation time, for one temperature cycle only, according to equation 25 below.

21604[ts] · 320[s] ≈ 6920000[s] ≈ 1922[h] ≈ 80[days] (25) Thus, for the entire temperature cycle, i.e., both one temperature increase and decrease cycle, it would take roughly six months with the capacity of a normal laptop. However, taking into account the use of a cluster, the simulation time will be reduced drastically. Since this master thesis project has a limited time frame, a shortened version of the temperature cycles was derived to capture the thermal soak time parameter, essential to the project.

Table 14: Temperature increase cycle for the simulation to capture the thermal soak time for the tool.

Time Step Temp [^◦C] RH [%] Duration [hh:mm:ss] Real time [hh:mm:ss]

0 25 93 initial initial

1 25 93 00:00:01 00:00:01

3 25 to 65 93 00:00:01 00:00:02

500 65 93 00:16:40 00:16:42

1500 65 93 00:33:20 00:50:02

As for the temperature decrease cycle, the same time steps ratio was used but an temperature decrease from 25^◦C to -10^◦C. Two different scenarios were run: one with a thermoplastic rubber representing a worst case heat transfer scenario and one with aluminium representing the best case of heat transfer.

(40)

5 RESULT

5.1 Experiments in the climate chamber

5.1.1 Inlet and outlet velocity measurements

Two separate velocity measurements with the anemometer verified the temperature independence of the inlet and outlet velocity. The result is presented in table 15 for the outlet and table 16 for the inlet.

Table 15: Outlet velocities in the climate chamber for seven temperatures.

Temperature [^◦C] Velocity [m/s]

-10 1.4

0 1.5

10 1.7

25 1.5

30 1.5

40 1.6

50 1.6

As a conclusion, the outlet velocity was assumed to be constant.

Table 16: Inlet velocities in the climate chamber for seven temperatures.

Temperature [^◦C] Velocity [m/s]

-10 6.3

0 6.4

10 6.5

25 6.5

30 6.4

40 6.7

50 6.7

As a conclusion, the inlet velocity was assumed to be constant.

Thereafter, measurements for the actual velocities were performed for an ambient temperature of 25^◦C and the results is displayed in table 17. The anemometer for the inlet velocity was placed at the inlet and for the outlet velocity the anemometer was placed 0.12 m from the outlet.

Table 17: Result from inlet and outlet velocity measurement.

Inlet velocity: 8.0 m/s Outlet velocity: 2.5 m/s

To verify the continuity equation, equation 20 was used together with the inlet and outlet velocities and the result concluded in table 18.

(41)

Table 18: Verification of the velocity measurement.

Parameter Fan/Outlet Inlet

Area, A [m²] 0.1056 0.03315

Velocity, v [m/s] 2.5 8

Flow rate, Q = A · v [m³/s] 0.264 0.265

The flow rate, i.e. continuity equation, is verified with two decimals accuracy. Thus, the velocity measurements are verified.

5.1.2 Temperature cycles

The result from the five temperature cycle runs in the climate chamber is displayed in the sections 5.1.2.1 to 5.1.2.3 below.

5.1.2.1 IEC 60068-2-38 temperature cycle

Firstly, the result from the temperature sensors test was compared to the IEC standardization temperature cycle to verify the correct behavior of the temperature sensors and if any modifications needed to be done.

Figure 23: Result from the IEC 60068-2-38 temperature cycle showing the reference temperature of the chamber in light blue and the sensors’ temperature in green and under the green plot.

Thus, the temperature sensors followed the IEC standard temperature cycle which resulted in continuation with the verification experiments without any modifications of the temperature sensors.

(42)

5.1.2.2 Temperature increase cycle

(a) Initial temperature increase run. The light blue line represents the reference temperature in the climate chamber.

(b) Temperature increase run 2 with slight modifications of the longest possible soak time for the sensors. The light blue line represents the reference temperature in the climate chamber.

Figure 24: Temperature increase runs

By studying the output data and graphs, a thermal soaking time in the climate chamber in a hot temperature cycle run can be determined to be at least 45 minutes when the tolerance of ±2K is taken into account.

(43)

5.1.2.3 Temperature decrease cycle

(a) Initial temperature decrease run. The light blue line represents the reference temperature in the climate chamber.

(b) The second temperature decrease run with slight modifications to achieve a thermal soak time for the climate chamber. The light blue line represents the reference temperature in the climate chamber.

Figure 25: Temperature decrease run

By studying the output data and graphs, the thermal soak time in a cold temperature cycle run can be determined to be at least 1 hour to reach the reference temperature and 30 minutes when the tolerance of

±2K is taken into account.

(44)

5.2 Simulation of the virtual climate chamber

5.2.1 CFD-simulations of tool

The result of the final simulation of the tool is shown in figure 26 below.

(a) Temperature graphs of the components in the simulation model for the case when the tool consist of only aluminium. The components have stabilized to the reference temperature, the inlet temperature.

(45)

(b) Temperature graphs of the components in the simulation model for the case when the tool consist of only thermoplastic rubber.

Figure 26: Temperature graphs for all components in the tool for the worst and best heat transfer case.

Figure 26 shows the heat transfer differences between the best case and the worst-case scenarios. For the best case, with aluminum as material, the entire tool achieved a steady-state temperature distribution within 500 time-steps. Opposite to the worst case, with thermoplastic rubber as material, the tool did not reach a steady-state distribution within 2000 time steps, four times as many time steps as for aluminum. Thus, the thermal soak time is extremely material dependent, and the thermal soak time for the thermoplastic rubber is inconclusive since the temperature has not stabilized yet. However, a trend can be interpreted but not confirmed.

The temperature sensors within the tools were also placed in the simulation model. Figure 27 shows the temperature plots over the aluminum case and the thermoplastic rubber case. The same trends as in figure 26 are displayed. For comparison purposes, only the first 500 time steps are displayed in the thermoplastic rubber graph, figure 27b.

(46)

(a) Temperature graphs of the temperature sensors for the case when the tool consist of only aluminium.

(b) Temperature graphs of the temperature sensors for the case when the tool consist of only thermoplastic rubber.

Figure 27: Temperature graphs for the temperature sensors for the worst and best heat transfer case.

(47)

Thus, for the same amount of time-steps, the temperature sensors in the thermoplastic rubber have reach a temperature of maximum 319K compared to the aluminium case, in which all temperature sensors had reached the reference temperature of 338K far within 500 time steps.

5.2.1.1 Convergence study of final CFD-model

The result of the convergence study for the final CFD-model is concluded in four graphs below. The entire tool is said to be in aluminum, representing the best-case scenario for heat transfer.

(a) Output of the mass flux of the inlet in red and outlet in purple.

(b) Output of the pressure of the inlet in red and outlet in purple.

(c) Temperature convergence of the components in the model with the inlet in black, outlet in purple, fluid in orange, mean of internal air walls in brown, shell of the tool in green and tool in dark blue, air internal back in light blue.

air internal front in pink.

(d) Close-up of the components as the temperature in- creased with the inlet in black, outlet in purple, fluid in orange, mean of internal air walls in brown, shell of the tool in green and tool in dark blue, air internal back in light blue.

air internal front in pink.

Figure 28: Convergence plots over the important parameters.

(48)

Figure 28a and 28b showed convergence for the mass flux and the pressure, resulting in the continuity equation for the mass flux is achieved. Figure 28c showed convergence for all components as all parts in the climate chamber is successively heated to the reference temperature. Lastly, figure 28d verified the simulation model since the components are heated in the proper order, i.e., the inflow regulated the temperature increase and is treated as the reference temperature. The fluid and the outlet are heated firstly and should reach the reference temperature the fastest, while the internal air components inside the tool are heated lastly. This statement is also summarized in table 19.

Table 19: Summary of the time for the components to reach the reference temperature.

Component Time to reach reference temp [hh:mm:ss]

Inlet 00:00:03

Outlet 00:00:35

Fluid/CC air 00:00:41 Tool shell 00:03:36

Tool 00:03:40

Internal air wall 00:03:45 Internal air front 00:03:50 Internal air back 00:04:00

Note, that all the components are a mean of all nodes in the respective component. The simulation model is verified for the material aluminium.

Feasibility study of modelling a virtual climate chamber with CFD