Development and Evaluation of Alternative Electric Thermostat Design

Arvid Isaksson

Mechanical Engineering, master's level 2018

Luleå University of Technology Department of Engineering Sciences and Mathematics Abstract Introducing an electric thermostat to the thermal management system is a way of actively controlling the temperature of the engine, which has been shown to have several possible gains regarding power, fuel consumption, emissions and engine durability. Complexity, cost and durability are key concerns that have led to no heavy duty truck on the market having an electrically controllable thermostat. This emphasizes the need for exploring alternative solutions that enables electric control of the thermostat according to the needs of heavy commercial vehicles. Several concepts have been generated to solve this problem and a model based approach in Simulink, Matlab and GT Suite was used for the development and evaluation. The most promising concept of combining a BLDC electric motor with a wax body enables electric control with a downsized actuator and full fail-safe function while showing improvements in temperature control performance compared to a traditional wax thermostat. This thesis has increased the knowledge on the subject and could allow for implementing an electrically controlled thermostat in future Scania heavy duty trucks, leading to a more durable engine with lower fuel consumption and emissions.

i Preface This master thesis was written by Arvid Isaksson during the first semester of 2018 as a final course in M.Sc. mechanical engineering at Lule˚aUniversity of Technology. The work was conducted at Scania CV AB in Södertälje, Sweden with Anders Larsson as company supervisor and Jan-Olov Aidanpääas university supervisor. I would like to extend my deepest gratitude to all the colleagues at Scania who have helped and encouraged me during this work. Special thanks to Zoltan Kardos, Mattias Strindlund and all at Scania who have contributed to this project by sharing knowledge and showing interest.

ii Nomenclature

Quantity Description Unit x Linear position of valve [m] mt Mass of valve [kg ] F Linear force exerted on valve [N] Fµ Static friction [N] FS Spring force [N] k Spring constant [N/m ] Finit Spring pretension force [N] rm Mean radius of lead screw thread [m] l Lead or pitch of thread [m] α Helix angle of thread [rad ] N Normal force [N] µ Friction coefficient between thread and nut [−] Fh Hydraulic force [N] ∆P Pressure differential [P a ] p Pressure [P a ] A Area [m2] τ Torque [Nm ] r Radius [m] h Height [m] Tamb Ambient temperature [K] me Mass of engine [Kg ] 2 Ae Surface area of engine [m ] Ce Specific heat constant of engine body [KJ/KgK ] Te Engine temperature [K] Wrad Heat flow out of radiator [W ] m˙ Coolant mass flow over radiator [Kg/s ] Cp Specific heat constant of coolant [KJ/KgK ] n Ratio of fuel energy going to heating the coolant [−] Wfuel Total rate of energy released from burning fuel [W ] Wair Rate of heat transfer from engine to surrounding air [W ] 2 hc Convective heat transfer coefficient [W/m K] Ue Inner energy of engine body [J] m˙ 0 Working point mass flow [Kg/s ] Te0 Target temperature [K] Wfuel 0 Expected average power from engine [W ] 2 hc Convective heat transfer coefficient [W/m K] m˜˙ Deviance from working point mass flow [Kg/s ] T˜e Deviance from target temperature [K] ˙ 2 W˜ fuel Deviance from expected engine power [W/m K] G Transfer function from mass flow to temperature [Ks/Kg ] H Transfer function from engine power to temperature [K/W ] Kp Proportional constant of PIDF controller [−] Ki Integrative constant of PIDF controller [−] Kd Derivative constant of PIDF controller [−] τd Time constant of derivative control term [−] ω Engine rotational speed [rad/s ]

iii Contents

Abstract ...... i

Preface ...... ii

Nomenclature ...... iii

1 Introduction ...... 1 1.1 Background...... 1 1.1.1 Thermal Management System ...... 1 1.1.2 Wax Thermostat Valve ...... 2 1.2 Objective ...... 4

2 Theory ...... 5 2.1 MarketResearch ...... 5 2.1.1 ElectricThermostat ...... 5 2.1.2 Related Technology ...... 6 2.2 ElectricStepperMotor...... 7 2.3 Valve Control Dynamics ...... 8 2.4 Hydraulics...... 10 2.5 Thermal Management System Model and Control ...... 11

3 Method ...... 14 3.1 ProductNeeds ...... 14 3.1.1 ProductCharacteristics ...... 15 3.1.2 ProductSpeciﬁcation ...... 16 3.2 ConceptGeneration ...... 17 3.3 Conceptselection...... 17 3.3.1 Conceptscoring ...... 18 3.4 ModellingandFurtherDevelopment ...... 18 3.5 Functional Prototype ...... 19 3.6 ProjectScope...... 19

4 Results ...... 21 4.1 Hydraulic Actuator Combined With Wax Body Through Cam ...... 21 4.1.1 Hydraulic Dimensioning ...... 21 4.2 BLDC Motor Combined With Wax Body ...... 23 4.2.1 Stepper motor dimensioning ...... 23 4.3 Rough Design of Final Product and Actuator Placement ...... 24 4.4 Prototype...... 24 4.4.1 Prototype Simulation ...... 26 4.4.2 Prototype Assembly ...... 27 4.4.3 Prototype Testing ...... 28 4.5 Simulated Temperature Control in GT Suite ...... 29

5 Discussion ...... 37 5.1 Conclusions...... 37 5.2 FutureWork ...... 38

A Normal Operation ...... 40

B Concept Descriptions ...... 42

C Concept Evaluation ...... 47

iv 1 Introduction Thermal management of the engine is a way to achieve faster warm-up and more optimal working temperatures, resulting in lower emissions, lower fuel comsumption and increased lifespan of engine components. Even though thermal management is an important aspect of the engine performance, the technology used today is often essentially the same as in the early 1900’s. Introducing a smart thermal management system has been shown to have several possible gains on the engine performance[1][2][3]. The thermostat that controls the coolant flow over the radiator is a possible area of improvement by introducing electronically controllable components and smart control [4][5]. A smart thermostat can improve the coolant flow control and allow different control temperatures [6]. It can also accommodate for additional control goals outside of coolant temperature. There exists passenger vehicles which have successfully incorporated a smart thermostat in their engine design [7][8]. Yet no heavy duty vehicles currently employ this kind of technology. A reason for this might be that the benefits of having smart thermal management in the form of an electronically controllable thermostat, does not currently outweigh the added cost and complexity of replacing the simple and robust wax pellet thermostat used today. This emphasizes the need for a solution which offers smart control over the coolant flow while still being robust and cost-effective, offering a solid business case.

1.1 Background Scania does not currently have an electronically controllable thermostat capable of smart control even though the thermal management might be improved by using one. Incorporating such a solution could give improvements in fuel consumption, emissions and durability of engine components. It is deemed that the solutions that exists today can be improved regarding durability, simplicity and cost-eﬃciency.

1.1.1 Thermal Management System The thermal management system handles the heat created by the engine and aims for keeping the engine at temperatures that enhance the engine performance. The traditional thermal management system consists of a cooling circuit which is circulated by a water pump connected to the engine crankshaft. The coolant moves through the engine, keeping it at acceptable temperatures. The coolant then reaches the thermostat which based on its temperature, divides the ﬂow to either be cooled in the radiator or go through the bypass to the pump where it is mixed with the cold radiator coolant and recirculated through the engine. The radiator cools the coolant by having air ﬂow through assisted by a cooling fan which creates forced convection. The working principle of the smart thermal management system is the same but the pump and thermostat can be replaced with electronically controllable components such as an electric water pump and smart valve, as seen in Figure 1.

1 Figure 1: Example of smart thermal management system [2].

The thermal management system for heavy duty trucks may also incorporate retarder cooling in case the truck is outﬁtted with a retarder. The retarder works like a secondary hydraulic brake that is mostly used in downhill driving. When used, high bursts of energy can be introduced into the thermal management system.

1.1.2 Wax Thermostat Valve The wax thermostatic element was introduced to automotive engines in the 1930’s since it was found sub-optimal to constantly cool the engine. The thermostat is designed to not send coolant to the radiator before it is suﬃciently hot, decreasing the engine warm-up time. It then keeps the temperature of the coolant around a favourable working temperature by closing to the radiator as it gets colder and opening as it gets warmer. A common type of wax thermostatic element is the wax-pellet thermostat [9] which uses a wax-pellet which is in a solid state at lower temperatures and thereby closed to the radiator, seen in Figure 2a. When the coolant gets warm, the wax pellet transitions to liquid state resulting in a volume increase which in turn opens the valve and sends coolant to the radiator, seen in Figure 2b. This valve has the advantage of being simple, cheap and robust while still oﬀering the basic function of faster warm-up and temperature control which is why it is still widely used today. Since there is no great way to diagnose its function, it is often exchanged during a trucks lifetime even though it might not always be necessary.

2 (a) Thermostat closed at low temperatures. (b) Thermostat open at high temperatures.

Figure 2: Wax-pellet thermostat design [9].

An illustration of how the coolant moves over the thermostat in diﬀerent positions can be seen in Figure 3.

Figure 3: Illustration of wax-pellet thermostat function [10].

Scania currently uses a solution with two separate wax thermostats. In this solution a pilot ﬂow is directed from before the engine, this pilot ﬂow is used to let one of the wax thermostats react

3 to this temperature instead of the temperature after the engine. So the two thermostats reacts to temperatures at diﬀerent parts of the engine, which has been shown to improve the thermal management of the truck.

1.2 Objective The objective of this project is to develop a product that handles the coolant flow control that is currently performed by a wax-pellet thermostat. This includes both generating a concept for the actuator that replaces the wax-pellet thermostat and also developing the basis of its control. The product will then be evaluated through a model based simulation approach to evaluate if it can improve the coolant flow control. Improving the coolant flow control could lead to lower fuel consumption, lower emissions and longer lifetime of engine components. The product should live up to the needs specified by Scania where controllability, robustness and cost will be key factors regarding a future implementation.

4 2 Theory This section presents research on previous solutions within thermal management and related areas and theory used in the development and evaluation of concepts.

2.1 Market Research The heavy duty industry still uses the wax thermostat for coolant ﬂow control but there are several examples of alternative solutions used in the automotive industry and research.

2.1.1 Electric Thermostat Several types of fully electric thermostats have been developed and tested. Schaeffler has made a termal managment module containing a electronic cooling thermostat [8] that is used in Audi’s EA888 TFSI, generation 3 engine [7]. The thermostat is controlled using a electric DC motor connected to a high reduction worm gear which turns the first rotary slide valve and directs the flow. The same electric motor also controls a second rotary slide valve which can stop the flow through the engine completely, speeding up engine warm up. This construction can be seen in Figure 4. The worm gear has the capability of friction locking the rotary slide valve in position. It does however introduce a high sliding friction.

Figure 4: Schaeﬄers electronic thermal management module [7].

Cormerais et al. [6] describes a concept for smart control of the thermal management system using individual poppet valves for each circuit in the cooling system, for example one for the bypass and one for the radiator. All the poppet valves can then be controlled by a mutual circular cam with tracks for each poppet valve. When an actuator rotates the cam, the poppet valves will open and close based on their individual tracks. An illustration of the cam and poppet valves can be seen in Figure 5.

5 Figure 5: Concept with indivdual poppet valves controlled by one cam [6].

There exists more advanced wax thermostats, for example the electrically heatable wax thermostat [11], also called MAP-controlled thermostat, this thermostat adds an electric heater within the wax which when turned on decreases the needed coolant temperature to melt the wax. This adds some degree of electrical controllability to the wax thermostat and allows for safely raising the operating temperature of the engine. Although it can be considered insuﬃcient for more advanced thermal management due to a slow response time and limited control options. The MAP-controlled thermostat can be seen in Figure 6.

Figure 6: Example of MAP-controlled thermostat [10].

2.1.2 Related Technology The hydraulic rotary vane actuator have been used in other applications in the automotive industry successfully, they oﬀer high torque over a limited angular stroke. An example of such an actuator can be seen in Figure 7. The actuator is considered compact and cost-eﬀective [12]. A solution with a hydraulic rotary vane actuator controlling some kind of valve could be suitable. Rose [13] described a invention of this type in 1966.

6 Figure 7: Example of hydraulic rotary vane actuator [12].

2.2 Electric Stepper Motor A stepper motor is a brush less DC motor that is operated in discrete steps, making it suitable for positioning applications among other things. Since there is no brush, the motor requires special drivers to control windings surrounding a permanent magnet. Figure 8 shows a basic case with a polarised center surrounded by windings. By energizing the windings in a specific order, the windings will create a polarity attracting the permanent magnet in the center causing it to rotate. This example has a step angle of 90 ◦ but the most commonly used type is a hybrid stepper where two gears of different polarity is stacked on top of each other slightly off phase, this results in shorter steps where 1 .8◦ is an often used step angle. Since the motor moves in discrete steps, it can be controlled with open loop control if it develops force with a sufficient margin to assume you don’t skip or miss steps. Advanced drivers can also switch the winding polarity in a way that increases the precision, some drivers also have sensor less diagnosing functions which can analyse the function of the motor by measuring the back emf.

7 Figure 8: Illustration of how a stepper motor develops torque and performs discrete steps.

2.3 Valve Control Dynamics A linear valve controlled by a stepper motor through a lead screw was modelled. The parameter to be controlled is the linear position of the valve, this is done by exerting a torque from the stepper motor which is converted to a linear force by the lead screw which moves the valve. The forces acting on the valve is described in Figure (9).

Figure 9: Forces acting on valve.

8 x is the position of the valve with zero corresponding to the radiator being fully closed. The acceleration of the valve can be calculated as in equation (1) where mt is the mass of the linearly moving thermostat parts.

xm¨ t = F − Fµ − FS (1)

F is the linear force exerted from the stepper motor through the lead screw, and Fµ is the static friction exerted mainly from the sealing but also other parts of the valve. FS is the force exerted by the spring which has a certain spring constant k and a initial pretension force Finit , the spring force is described in equation (2).

FS = kx + Finit (2)

By combining equation (1) and (2), the dynamic equation (3) is achieved.

k F − F + F x¨ + x = µ init (3) mt mt

The forces acting on the lead screw is illustrated in Figure 10. No dynamic eﬀects is taken into consideration. The thread can be imagined as a slope with a length depending on the mean radius rm and height being the lead l.

Figure 10: Forces acting on lead screw.

Since equilibrium is assumed, the sum of the forces acting in x and y direction must be zero. Equation (4) shows the result of Fx = 0. where N is the normal force, α is the helix angle and µ is the friction coeﬃcient of theP screw.

τ cos (α) = µN + F sin (α) (4) rm

Equation (5) shows the result of Fy = 0. P τ N = F cos (α) + sin (α) (5) rm

9 When combining equation (4) and (5), the correlation between torque and force can be calculated as in equation (6).

τ (cos (α) − µsin (α)) F = (6) rm (µcos (α) + sin (α))

The ideal force Fideal can be calculated by setting the friction coeﬃcient µ = 0, the eﬃciency is then simply calculated as in equation (7).

F η = (7) Fideal

2.4 Hydraulics Hydraulic actuators works through a pressurized medium which is assumed to be in-compressible. The developed force Fh in a linear actuator is then simply dependent on the pressure diﬀerence ∆ P and area A as shown in equation (8).

Fh = ∆ P A (8)

The torque τ developed by a mechanism at a given radius r is shown in equation (9).

τ(r) = Fhr (9)

A hydraulic rotary vane actuator, illustrated in Figure 11, will have a varying torque over the radius.

Figure 11: Illustration of rotary hydraulic actuator

The area can be deﬁned through the radius and height as a function of the radius h(r), then by combining equation (8) and (9) and integrating over the radius, you get the total torque acting on a hydraulic rotor as shown in equation (10).

rout τ = r∆P h (r)dr (10) Z rin

In a classic hydraulic rotary vane actuator, the height is constant h(r) = h. The actuator might also have several paddles on the rotor like in Figure 7, in this case simply multiply the torque by the number of paddles.

10 2.5 Thermal Management System Model and Control In order to create a basic control for the thermal management system, a rough model was made of how the dynamics of the engine temperature works. Figure 12 shows a simplified version of the engine and radiator circuit of the thermal management system. The radiator is assumed to be ideal, leading to the temperature of the coolant coming out always being ambient temperature Tamb . The pipes are also modelled as ideal with no energy loss, leading to the temperature of the coolant only changing in the radiator and engine. The engine is simplified as a uniform body with a mass and surface area me, Ae, a specific heat constant Ce and a temperature Te. It is assumed the coolant will be the same temperature as the engine directly after moving through it. The thermostat will control a mass flowm ˙ over the radiator.

Figure 12: Temperatures and energy ﬂows in and out of thermal management system.

The energy flows in and out of the circuit are also shown in Figure 12. The radiator will release a energy flow Wrad through forced convection, easily described through equation (11) where Cp is the specific heat constant of the coolant.

Wrad =mC ˙ p(Te − Tamb ) (11)

The burned fuel in the engine will add heat to the engine, simply described as a constant fraction multiplied with the total energy in the burned fuel nW fuel . The engine will also release energy to the surrounding air Wair through free convection. This energy ﬂow can be described as in equation (12) where hc is the convective heat transfer coeﬃcient of the process.

Wair = hcAe(Te − Tamb ) (12)

The inner energy of the engine body Ue can be described as in equation (13).

Ue = TemeCe (13)

The ﬁrst law of thermodynamics states that energy can not be lost or destroyed, meaning that the total rate of energy entering and leaving the thermal management must be the same as the rate of change for the internal energy of the engine U˙e, shown in equation (14)

11 U˙e = nW fuel − Wrad − Wair (14)

The changing variable in U˙e will be Te, when combining equations (11), (12), (13) and (14), a non-linear dynamic model of the thermal management system is achieved, shown in equation (15).

T˙emeCe = nW fuel − mC˙ p(Te − Tamb ) − hcAe(Te − Tamb ) (15)

To simplify the model and control, it is linearised by using Taylor ﬁrst order linearisation around a working point equilibrium wherem ˙ =m ˙ 0, Te = Te0, T˙e = 0 and Wfuel = Wfuel 0. Te0 is chosen as the target temperature and Wfuel 0 is the expected mean engine power.m ˙ 0 is found by solving f(m ˙ 0, T e0, 0, W fuel 0) = 0. Equation (16) shows the resulting expression form ˙ 0.

nW fuel 0 − HcAe(Te0 − Tamb ) m˙ 0 = (16) Cp(Te0 − Tamb )

The Taylor ﬁrst order linearisation then gives the linear model shown in equation (17) where m˜˙ = ˙ m˙ − m˙ 0, T˜e = Te − Te0, T˜e = T˙e and W˜ fuel = Wfuel − Wfuel 0.

˙ nW fuel 0 T˜emeCe + T˜e = −Cp(Te0 − Tamb )m˜˙ + nW˜ fuel (17) Te0 − Tamb

A transfer function from the input m˜˙ to the output Te can easily be found by using the Laplace transform. The transfer function for the system is shown in equation (18). Wfuel is considered as a measured disturbance since it will not be controlled.

Cp(Te0 − Tamb ) 1 Te(s) = − m˙ (s) + nW fuel = Gm˙ (s) + HW fuel (s) (18) nW fuel 0 nW fuel 0 sm eCe + sm eCe + Te0−Tamb Te0−Tamb

The control will be a PIDF feedback controller which controls the flow by reacting to the error between the target temperature and measured temperature. The PIDF controller consists of a integrative part, a proportional part and a derivative part with a first order filter. The PIDF controller is expressed as in equation (19) where Kp is the proportional constant, Ki is the integrative constant, Kd is the derivative constant and τd is the time constant of the derivative part. e(s) is the measured error.

1 s m˙ (s) = e(s)( Ki + Kp + Kd ) (19) s τds + 1

A feedforward control is used to adjust the mass ﬂow input based on the current engine power. This control simply multiplies the measured W˜ fuel with a proportional controller derived from solving m˙ (s) from equation (18) for Te(s) = 0. The thermal management system model with feedback and feedforward control is shown in Figure 13.

12 Figure 13: Block diagram of thermal management system model and control.

13 3 Method This project follows an adjusted version of the product development process described in [14]. This was due to the project being done by a single individual instead of a project group, and to allow more time to be spent on stages that are considered more crucial in this project. The process is well suited for creative and innovative product development and covers all stages from planning to product testing. An iterative approach [14] was used for all phases where new information, calculations and simulations were used to go back and improve the product. After the planning was done, the needs of the product were defined and used to make a product specification. Based on this, concepts were generated through different brainstorming methods. A concept was then chosen for further development based on the product specification. The chosen concept was then developed using a model based approach where simulations and calculations are used to design and dimension a product that lives up to the product specification. The product testing was performed through simulations and a functional prototype that was tested with Scania’s flow rig. Figure 14 shows a summary of the product development process.

Figure 14: Product development process based on [14]

3.1 Product Needs The needs of the coolant ﬂow control are to be expressed in such a way that no speciﬁc type of solution is favored [14]. Need statements and their relative importance were derived from statements made in interviews with Scania employees and information found in internal and external documents. The importance is given on a scale of 1 to 5 where 5 is very important and 1 is not important. The need statements are listed with explanations in Table 1.

14 Table 1: Need statements and explanations for the coolant flow control. Needs Statements - Coolant Flow Control Need Need statement Imp. Explanation no. 1 Robust design 5 The solution should be reliable and durable during all operating conditions. 2 Controllable 5 The solution should offer some degree of control over the coolant flow. 3 Fast response time 1 The solution should quickly be able to adjust the flow to properly adapt to fast changes in driving conditions. 4 Cost-effective 5 The solution should have a low cost compared to its performance. 5 Varied temperature set 3 The solution should be able to keep points the temperature around different set points. 6 Reasonable size and 3 The size and weight of the solution weight should be reasonable in order for it to fit within the current system. 7 Good precision 2 The solution and its control offers sufficient precision in the control of the coolant flow. 8 Low flow losses 4 The solution should minimize the flow losses in the cooling circuit. 9 Lower fuel consumption 5 The solution should either directly or and emissions indirectly through more efficient control of the thermal management system achieve lower fuel consumption and emissions. 10 Efficient 2 The solution should not consume too much power during operation. 11 Smooth temperature con- 3 The temperature fluctuation around trol the set point should be kept smooth and at low amplitude to optimize efficiency and durability of engine components.

3.1.1 Product Characteristics A number of characteristics for the product are listed as needs and desires in Table 2. These characteristics are used to deﬁne the functionality of the product and also to evaluate concepts where no meaningful metric can be used or evaluated within the scope of this project. The functionality needs are functions that must be included in a viable concept while the functionality desires are favourable but can be skipped with proper motivation.

15 Table 2: Product Characteristics. Functionality - Needs • Must allow for some form of electric control, this includes both the actuator and the required sensors to control it. • Must have a fail-safe function where full cooling or sufficient function can be achieved even if the electrical components fails. • Must avoid plastic deformation during operation. • Must not contaminate coolant. • The solution must be able to develop the required force to direct the flow during normal operating conditions. Functionality - Desires • Valve design should minimize flow losses. • Should convey information of its performance to the user to ease diagnosing and performing maintenance on the actuator. • Should be easy to perform maintenance on. • Should be easy to produce and assemble. • The temperature control should aim to minimize oscillations, both frequency and amplitude. • Leakage through valves should be minimized. • Should have higher control precision at small flows.

3.1.2 Product Specification A product specification was made based on the identified needs of the product. The product specification consists of metrics which can be objectively measured and connected to a fulfilment of the need statements made in Table 1. The metrics are also given an importance based on the needs they are connected to where 5 is very important and 1 is not important. The product specification can be seen in Table 3. The metrics are defined during specific operating conditions explained in Appendix A.

16 Table 3: Product specification. Product Specification Metric Need Metric Imp. Explanation no. no. 1 1 Product life- 5 The product should be durable enough to last the time entire lifetime of the truck. Expressed in hours h or km of normal operation before failure. 2 3 Response time 1 The solution must be able to perform the full stroke within a certain time under normal operating conditions counting from when the signal to move is first sent. Expressed in seconds s. 3 4 Product cost 5 The solution can not exceed an estimated cost expressed in SEK . 4 2,5,11 Temperature 3 The solution must be able to keep the temperature control toler- within a certain tolerance from any reasonable set ance point, expressed in ±◦C. 5 6 Geometric con- 3 The solution should stay inside some geometric constraints straints expressed in m3. 6 6 Weight 3 The solution should have a low weight expressed in kg . 7 3,7 Actuator volu- 2 The actuator must control the volumetric flow metric flow con- within a certain percentage tolerance of the total trol accuracy flow expressed in ±%( m3s−1). 8 8 Pressure loss 4 The actuator and valve design should minimize the over thermostat pressure loss expressed in MP a at a flow of 300 l/min. 9 9 Radiator path 4 When the radiator path is closed, the leakage to the leakage radiator must be minimized. Expressed in leakage cooling effect kW . 10 9,10 Actuator power 2 Actuator should be energy efficient, expressed in av- consumption erage power consumption W during normal operation.

3.2 Concept Generation A key factor for a successful concept generation is an open mindset, there are no bad ideas. It is important to explore a wide range of solutions both internally and externally, quantity is often regarded as more important than quality in the early stages [14]. Some external inspirations that have been used during the concept generation can be seen in Section 2.1. For the internal concept generation, individual brainstorming has been the main method where ideas are sketched down on paper. Also the morphological matrix method [14][15] has been used in order to combine partial solutions and ﬁnd new combinations. The process was repeated through two iterations.

3.3 Concept selection It is important to make an objective assessment of the generated concepts by using the identified product needs. The process of selecting a concept is iterative, where concepts are narrowed down through concept selection and new concepts with a higher level of detail is generated through combinations and new ideas. The level of detail on the concept selection is highly based on the detail of the concepts and it is only the final selection that uses the full scale product specification as using a high detailed selection on low detail concepts may give a false sense of security. It is important that all concepts are at a similar level of detail to allow for a fair comparison.

17 Concepts are first eliminated based on if they fulfil all the product characteristics needs, seen in Table 2. To make an initial assessment and narrow down the number of concepts, a screening is used where concepts are evaluated relative to a well defined reference concept on a scale of +, 0, and - where + indicates better performance, 0 indicates similar performance and - indicates worse performance on each metric. Some metrics are not used in the screenings due to them being hard to evaluate with a low level of detail and/or having a low importance.

3.3.1 Concept scoring The concept scoring uses the complete product specification to make an objective choice on which concept to proceed with. Each concept will be given a score on each metric on a scale of 1 to 5. 5 means that the solution should perform around the ideal value for the metric or better and 1 means that it barely makes the marginal value for the metric. If a concept is deemed to not make the marginal value for a metric, it is eliminated. The evaluation is based on preliminary calculations and simulations for dimensions and performance. For factors such as product lifetime, product cost and to some degree the flow and control factors. The evaluation for these metrics was instead based on results from previous similar solutions and extensive discussions with experienced Scania engineers. The final score for a concept is calculated by multiplying the score of each metric with the importance of the metric and then summing the score on all metrics.

3.4 Modelling and Further Development To further develop the chosen concepts and evaluate the effect of different design choices, a model based approach is used. Simulink and its Simscape library which is very good for modelling multi domain physical systems is used to model the chosen valve and actuator and the basis of the control. To evaluate the thermal management performance and compare it to the old wax thermostat. An existing engine and thermal management system model in the program GT Suite is used together with the reference data explained in Appendix A. GT Suite is a simulation tool for multi domain physical systems often used in the automotive industry. The model is very detailed and is assumed to be sufficiently correct, an overview of the model is seen in Figure 15. Matlab will be used for data post processing and general calculations regarding design and dimensioning.

18 Figure 15: GT Suite model used for simulating thermal management performance.

Catia will be used to design and evaluate 3D models for the concept and prototype, this enables using existing Scania Catia models as reference for placement and design.

3.5 Functional Prototype The prototype is meant to validate the basic function of the concept, meaning that little regard will be given to size, shape and weight to allow for an easy and quick development and manufacturing of the prototype. The functions that will be tested are the ability to accurately position the valve and control the ﬂow. The ﬂow rig that exists does not have warming and cooling of the liquid which makes testing of the thermal control impossible, atleast during this project.

3.6 Project Scope The main focus of the project is to generate and evaluate alternative concepts that can give better thermal management through electric control. To achieve this within the time frame, some simplifications have been made. • No detailed production or cost analysis will be done, this will instead be evaluated based on common sense and input from Scania employees. • No CFD will be done to evaluate flow characteristics or fatigue of concepts. This will instead be based on previous evaluations of similar solutions and common sense. • The GT Suite model is assumed to be correct and no changes will be made to it except for evaluating the effect of adding the modelled electric thermostat. • Only the performance during the reference path described in Appendix A will be evaluated during this project even though the performance during other operating conditions might be important at a later stage. • Electric motors will only be specified in terms of performance specifications as designing an electric motor from scratch would be very time consuming. It can be assumed electric motors will be purchased from a sub supplier.

19 • The ﬁnal design of the chosen concepts will only be presented as a rough overview regarding size, shape and placement and how it can be implemented with the rest of the thermal management system and engine.

20 4 Results The concept generation resulted in 6 final concepts described in Appendix B. These were evaluated according to the product needs and the concept which performed best was concept D, hydraulic actuator combined with wax body through cam. Concept A, BLDC motor combined with wax body had a very similar performance and they use the same valve. Therefore it was decided to also evaluate the effect of using a BLDC electric motor to control the valve. See Appendix C for the complete evaluation and motivations. For the functional prototype, it was chosen early to evaluate the electric concept since off the shelves components can be used at a higher degree. The further development also showed that the concept with a BLDC electric motor should have better performance on several metrics, leading to it being the focus of the later part of the development and evaluation efforts in this project.

4.1 Hydraulic Actuator Combined With Wax Body Through Cam The concept is a hydraulic solution combined with a wax body. The actuator consists of a rotor with a single wing and a stator. Pressurised engine oil is distributed from the center of the rotor by using an integrated valve controlled by a variable force solenoid. When pressurised oil enters one of the sides, it will induce a torque in one direction. Actuators of similar have achieved a precision of ±0.1◦ so the risk of not achieving sufficient control with a actuator of this type is small. The actuator rotates a cam which pushes down the sleeve valve and opens the flow to the radiator. The cam profile can be tailor made to achieve higher precision in the low flow control region where the majority of the control is performed. A wax body which lies in the middle of the coolant flow from the engine is also connected to the sleeve valve in a way such that it can work in parallel with the hydraulic actuator. The working principle of this design is that the hydraulic actuator will be dimensioned to work in the low flow region where the absolute majority of the control is needed. The wax body is set on a high opening temperature where it opens when full cooling is needed and fine electric control is deemed obsolete.

4.1.1 Hydraulic Dimensioning The required linear force for the actuator was based on the spring characteristics of the lower wax body and sleeve valve design. Measurements of spring force depending on deformation can be seen in Figure 16. Assuming the actuator will work in the low ﬂow region, using the fully deformed spring force of 225 N plus a worst case axial seal friction assumed to be 75 N should give a reasonable dimensioning with a 300 N linear force. At idle conditions, the engine oil pressure is 1 bar and the actuator must be dimensioned for this pressure. If the rotor wing and stator wall each requires 30 ◦ of working space, then only 300 ◦ can be used used for eﬀective work.

21 Spring Force Depending On Deformation 250

200

150

Spring Force 100

0 0123456789 Spring Deformation [mm]

Figure 16: Spring force as a function of spring deformation for wax body and sleeve valve design.

A space eﬃcient solution is to dimension the cam with the same radius as the hydraulic actuator. To simplify the initial dimensioning, the cam is also assumed to have a total stroke of 10 mm with a constant incline over 300 ◦. By assuming force equilibrium, the required torque to achieve 300 N linear force can be calculated for each radius, shown in Figure 17a. For each actuator radius, a height is calculated that gives the required torque according to equation 10. If the rotor center is assumed to be 10 mm in radius and the outer walls are assumed to be 10 mm thick, the resulting volume for each radius can be seen in Figure 17b. The red dot in Figure 17b indicates the minimum volume for the chosen force and wall thickness.

300 N linear force Volume depending on working radius 0.5737 0.3

0.5736

0.5735

0.25 0.5734

0.5733 volume [l] 0.5732 0.2 Required Torque [Nm] 0.5731

0.573

0.5729 0.15 20 40 60 80 100 120 140 160 180 200 20 25 30 35 40 45 50 Radius [mm] Radius [mm] (a) Required torque to achieve a 300 N linear force (b) Volume as a function of radius.

Figure 17: Required torque and volume depending on radius.

A variable force solenoid is also added, it will have a diameter of 55 mm and a height of 33 mm, assuming it will be the same size as for the variable cam phaser. A rough model of the concept was modelled to evaluate size and placement, it was chosen to place the cam in the middle to simplify the connection of the oil supply to the hydraulic actuator. The model can be seen in Figure 20, note the tube like design that is achieved by having similar radii’s on actuator, cam and solenoid.

22 (a) Overview of size and design of hydraulic actuator. (b) Colour coded section view.

Figure 18: Overview of size, design and function of the hydraulic actuator.

4.2 BLDC Motor Combined With Wax Body The concept uses the same wax body and valve design as the hydraulic concept. A BLDC electric motor in the form of a stepper motor is used to control the position of the valve by moving the motor in discrete steps. This allows fine position control without having to add sensors and the relatively high torque that can be achieved with a stepper motor eliminates the need for a large reduction. A lead screw is used to convert the rotary motion of the stepper motor into a linear motion of the sleeve valve. The lead screw mechanism is not the most efficient but it offers high precision in a compact format and can eliminate the need of a gearbox depending on the chosen parameters. The working principle of this design is the same as for the hydraulic concept where the actuator is dimensioned to control in the low flow region and the wax body is meant to handle the valve movement when high or full flow to the radiator is needed.

4.2.1 Stepper motor dimensioning The functional demands are the same as for the hydraulic solution. The stepper motor will also have to compress the spring whose characteristics is shown in Figure 16 and handle the seal friction. A stepper motor with open loop control also needs a safety margin in order to not miss or skip steps. This margin was initially chosen to be 2, which resulted in the stepper motor and lead screw being designed to deliver 600 N linear force. For a ﬁnal product, a weaker motor with high reduction should be suitable due to high demands on precision and low demands on speed and eﬃciency. A stepper motor with reduction used for the HVAC system of the truck has been evaluated for this application as a reference, this motor can be seen in Figure 19.

23 Figure 19: Stepper motor with reduction used for the HVAC system of the truck.

The HVAC actuator develops a nominal torque of 0.6 Nm at 320 Hz, or in worst case 0.3 Nm at low voltage and high temperature. The large reduction leads to a step giving a 0 .056◦ turn which is good for precision. If the the lead screw is dimensioned according to equation (6), choosing a lead screw with 10 mm diameter and 2 mm lead would lead to a sufficient force of 550 N, but a 100 second response time which is far outside of the product specification. Leading to the conclusion that a stronger and faster motor will need be used in the final product although the HVAC motor will still be used as reference in this project. The current HVAC actuator is quite cheap and small and it is deemed reasonable to assume a stronger version will also stay within the product specification regarding product cost and size.

4.3 Rough Design of Final Product and Actuator Placement The modelled hydraulic actuator and HVAC actuator were placed on a modular thermostat housing. Surrounding engine components were added to see how well the concepts fit on the engine and it could be seen that the HVAC actuator fits very well with no collisions and easy cable management. The hydraulic concept takes up considerably more space and does not fit as well on top of the thermostat housing, it also collides with a surrounding engine component. The final design of the electric motor concept will not be done in this project but having a more cylindrical design could allow for placing it deeper in the thermostat housing, saving even more space.

4.4 Prototype It was chosen to build a prototype for the BLDC motor combined with wax body concept. Testing equipment exists that can easily be modiﬁed to test the function of the concept. Figure 20a shows a 3D model of the complete functional prototype and Figure 20b shows which parts have been developed and manufactured or bought speciﬁcally for this project.

24 (a) Complete functional prototype for ﬂow control. (b) Newly developed and bought parts.

Figure 20: Functional prototype of electric thermostat combined with wax body.

To achieve the functional demands, a stepper motor with driver was bought. It was chosen to use no gearbox in the prototype to simplify the construction and the stepper motor was dimensioned with this in mind. An off the shelf stepper motor PD42-4-1140-TMCL with driver was chosen to get it running fast. It has a 0.7 Nm holding torque and a 1 .8◦ step angle. The torque curve for the stepper motor is presented in Figure 21. The stepper motors rotary motion is converted into a linear one through a trapezoidal lead screw in steel and bronze nut with a diameter of 10 mm and a lead of 2 mm. The resulting specifications are listed below. • 32% lead screw efficiency based on equation (7) and a conservative friction coefficient of 0.15 for lubricated steel screw to bronze nut. The screw is also self locking since the efficiency is negative if calculated for when the nut is pushing. • 710 N linear force on valve. • 0.01 mm linear resolution. • Torque curve shows that a speed of 250 rpm can be achieved without dropping below 0.6 Nm or approximately 600 N, corresponding to a linear speed of 8 mm/s.

Figure 21: Stepper motor torque curve

25 4.4.1 Prototype Simulation To further test that the performance of the prototype would be suﬃcient, it was modelled and tested in Simulink using Simscape components and a simple control. A simple driver controls the position of a stepper motor by sending pulses at a ﬁxed frequency until the stepper motor reaches the set position. A lead screw converts the rotation into a linear motion which moves the modelled sleeve valve. The valve is modelled with a pre tensioned spring with characteristics derived from the data shown in Figure 16. The friction is set to be 75 N static friction and a 1000 N/(m/s) viscous friction to get some damping in the system although the true viscous friction is hard to assess. Hard stops is also added to make sure the valve will behave realistically also in the extreme positions. Figure 22 shows the Simulink model used to simulate the stepper motor valve control.

Figure 22: Simulink model of stepper motor controlling sleeve valve.

Data from real stepper motors is used for the simulation, like for example winding voltage supply, resistance and inductance. The performance of the motor can then be evaluated to see if it performs as expected and most importantly if it can position the valve accurately without losing track of the position. Figure 23a shows the simulation results when using the chosen 0.7 Nm stepper motor PD42-4-1140 at 340 Hz, which corresponds to a linear speed of 3.4 mm/s. It shows good position tracking at suﬃcient speeds. Figure 23b shows that the motor will not be able to perform the full stroke of 10 mm at a higher frequency of 400 Hz, losing track of its position. 400 Hz corresponds to 120 rpm, showing that this might be a conservative simulation since the torque curve in Figure 21 still shows good performance at this speed.

Stepper Positioning Performance, 0.7Nm, 340Hz Stepper Positioning Performance, 0.7Nm, 400Hz

10 10

8 8

6 6

4 Sleeve Valve Current Position 4

Position [mm] Sleeve Valve Lower Target Position [mm] Sleeve Valve Upper Target Sleeve Valve Current Position Sleeve Valve Lower Target 2 2 Sleeve Valve Upper Target

0 0

012345678 01234567 Time [s] Time [s] (a) Positioning with 340 Hz frequency. (b) Positioning at 400 Hz.

Figure 23: Positioning performance for 0.7 Nm stepper motor at diﬀerent frequencies.

A smaller version of the same motor PD42-3-1140 was also simulated to evaluate the eﬀect of using a lower torque. The motor develops 0.44 Nm at up to 400 rpm according to its torque curve. Figure

26 24 shows that it can handle the entire stroke with good positioning although it can’t handle as high frequencies as the stronger motor.

Stepper Positioning Performance, 0.44Nm, 170Hz Stepper Positioning Performance, 0.44Nm, 200Hz

10 10

8 8

6 6

4 4 Sleeve Valve Current Position Position [mm] Position [mm] Sleeve Valve Lower Target Sleeve Valve Current Position Sleeve Valve Upper Target Sleeve Valve Lower Target 2 2 Sleeve Valve Upper Target

0 0

0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 Time [s] Time [s] (a) Positioning with 170 Hz frequency. (b) Positioning with 200 Hz frequency.

Figure 24: Positioning performance for 0.44 Nm stepper motor at diﬀerent frequencies.

For both motors, the simulation seems to give lower possible frequencies than what is stated in their torque curves, this could be due to the use of a simple driver in the simulation. The driver has a constant voltage supply to the windings which leads to a lower current at higher frequencies as the back emf increases with rotational speeds along with the induction in the coils damping current changes. As the current amplitude decreases, so does the torque. Stepper drivers often have a constant current supply that regulates the voltage to achieve more torque at higher speeds. Figure 25 shows the current during the positioning in Figure 23a at 340 Hz, note how the current only is at the speciﬁed 2 A when the motor is standing still. The 0.7 Nm stepper motor seems like a good choice for the prototype as it performs suﬃciently well even in a conservative simulation.

Current 0.7Nm, 340Hz 2

1.5

0.5 Current [A]

-0.5

-1 012345678 Time [s]

Figure 25: Current during positioning of valve at 340 Hz.

4.4.2 Prototype Assembly The assembly of the prototype was performed as soon as all parts had arrived. Some minor ad- justments were necessary to improve the ﬁt and performance of the prototype. A key aspect that was considered when assembling the prototype was to properly align and lubricate the rotating and

27 sliding parts to minimize friction and stress. Figure 26 shows the parts that were developed and bought in this project fully assembled.

Figure 26: Prototype fully assembled.

This assembly was then mounted in the existing valve testing equipment, using spacers to ensure that the actuator is the right height above the valve and can perform the full stroke. The positioning performance both with and without ﬂow was excellent and no sign of the actuator losing synchronisation could be found during the testing.

4.4.3 Prototype Testing The testing was done in an existing Scania flow rig. The motor speed was set to 5 mm/s and was positioned using micro stepping with a resolution of 32 micro steps in a full step, giving it a resolution of 3200 steps per millimetre or 0.003125 mm/step. Before the testing, the actuator was tuned to make sure that the starting position of the motor is just before the radiator starts to open. The valve position was then gradually increased and the flow was measured for each position. The pressure was measured before the thermostat, in the bypass and in the radiator. The flow was measured before the thermostat and in the radiator, note that the bypass flow is simply calculated by knowing that the sum of the bypass and radiator flow is equal to the thermostat inlet flow. The data was sampled during static conditions and a mean value of the last 100 samples were used to decrease the effect of evenly distributed disturbances. An engine speed of 1150 rpm was used. During the testing, smaller increments were used when positioning the valve in the low flow region and were then increased as a higher flow and opening was achieved. The resulting coolant flow through the radiator depending on valve opening can be seen in Figure 27. It is worth noting that the precision and resolution of the flow measuring equipment was limited which might have affected the results, especially in the low flow region.

28 Electric Thermostat Flow Control at 1150 RPM 100

Percentage of total flow going to radiator [%] 10

0 10 20 30 40 50 60 70 80 90 100 110 Valve Lift [%]

Figure 27: Resulting ﬂow depending on valve opening in Scania ﬂow rig at 1150 rpm.

The pressure drop over the thermostat behaved as expected where a lower opening and higher flow results in a higher pressure drop. The maximum pressure drop naturally occurs when the valve is completely closed to the measured flow path. The pressure varied more than the flow and had some dynamic effects that were not captured using the mean value data gathering method. These dynamic effects were not examined further as they were deemed to be outside the scope of this project.

4.5 Simulated Temperature Control in GT Suite To evaluate the temperature control of the concept, it was modelled in GT Suite and added to the detailed engine and cooling system model. The modelled thermostat consists of a electrically controlled part that reacts to the measured temperature after the engine and a wax body part that reacts to the coolant ﬂowing past the thermostat. A logic operator has also been added to properly model the interaction between the wax body and electric actuator where the one which requests the highest opening controls the ﬂow. The modelled valve is shown in Figure 28. To see the interaction between the wax body and electric actuator, they were modelled with target temperatures that allow them to work together during the reference track. The reference track should in reality be easy enough for the electric actuator to handle alone, achieving even better temperature control.

29 Figure 28: Thermostat function modelled in GT Suite.

The modelled valve is placed in the complete thermal management system model as shown in Figure 29.

Figure 29: Placement of modelled thermostat.

The wax body is modelled with the same opening characteristics as the existing wax thermostat. The opening areas to the radiator and bypass have been adjusted to match those of the prototype. The initial opening temperature is set to 96 ◦C with full opening at 102 ◦C, it is modelled with a hysteresis leading to it starting to close at 100 ◦C and fully closing at 91 ◦C. The electric control is modelled with GT suites PID block with a target temperature of 92 ◦C and a feedforward controller that measures the engine power Pe = τω and adjusts the requested mass flow. A first order order filter with a time constant of 0.7 is also added to simulate the delayed response of the electric actuator. The parameters of the PID controller was set by creating the model from equation (18) in Matlab and using Matlabs pidtune tool to get PID parameters for different crossover frequencies, it was chosen to only use a PI controller since it decreases the complexity and showed better results. When using higher cross over frequencies, the control becomes twitchy with a highly pulsating flow control and small but fast oscillations in engine temperature. The crossover frequency was gradually decreased

30 until the GT simulation showed a smooth ﬂow control with fewer oscillations in engine temperature while still maintaining good tracking at a crossover frequency of 0.1 rad/s . The feedforward proportional control was derived as described in Section 2.5, but the constant n was iterated some to ﬁnd the value that gives the best tracking, resulting in n = 0 .2. The resulting temperature control performance is compared to Scania’s current double wax thermostat solution where both use the operational data described in Appendix A and initial conditions where ambient temperature is 25 ◦C and coolant temperature is 92 ◦C. The temperature of the engine for both thermostats is shown in Figure 30. Note the occasional temperature spikes where the retarder is suddenly activated. The temperature dip shortly before the 1500 second mark is from an extra intense retarder activation that makes the wax body take over the temperature control.

Engine Temperature 100

Temperature [Degrees Celsius] 84 Electrically Controlled Thermostat 82 Reference Scania Truck

80 0 500 1000 1500 2000 2500 3000 Time [s]

Figure 30: New electrically controlled thermostat compared to old wax thermostat.

The requested valve lifts for both the electric actuator and wax body is shown in Figure 31. The electric actuator requests a mean lift of 0.36 mm and a max lift of 1.05 mm during the reference track while keeping a good temperature tracking. This gives merit to the idea of being able to signiﬁcantly downsize the electric actuator by combining it with a wax body that can handle the few cases where full cooling is needed.

31 Thermostat Valve Lift 7

6 New thermostat - Electric actuator New thermostat - Wax body 5

Lift [mm] 3

0 0 500 1000 1500 2000 2500 3000 Time [s]

Figure 31: Valve lift during operation.

The temperature sensor for the electric thermostat is placed before the retarder, leading to only the wax thermostat sensing and reacting to the temperature spikes in the coolant coming from the active retarder. Figure 32 shows the temperatures that the electric and wax part of the thermostat reacts to.

32 Thermostat Temperature 110

105

100

90 Temperature [Degrees Celsius] 85 Temperature sensed by wax thermostat Temperature sensed by electric thermostat 80 0 500 1000 1500 2000 2500 3000 Time [s]

Figure 32: Comparison of sensed temperatures for the wax part and electric part of the thermostat.

The resulting mass ﬂows to the radiator is shown in Figure 33, the mass ﬂows to the radiator for the old wax thermostat is also added for reference.

33 Mass Flow Radiator 2.5 Electrically controlled thermostat Reference Scania Truck

1.5

Mass flow [Kg/s] 0.5

-0.5 0 500 1000 1500 2000 2500 3000 Time [s]

Figure 33: Mass ﬂow to radiator for new and old thermostat.

The electrically controlled thermostat achieves a mean temperature of 91.95 ◦C with a root mean square error (rmse) of 0.81 from its target temperature 92 ◦C. Meanwhile the old wax thermostat has a mean temperature of 93.78 ◦C, since no exact target temperature can be pinpointed, that mean is considered as the target. Leading to a rmse value of 1.68 from that mean, showing that the concept with electric control oﬀers a signiﬁcant improvement to the temperature amplitude control. When examining the frequencies in the temperature signals from Figure 30 using the Fast Fourier Transform method, it can be seen that the NCG wax thermostat actually has a higher mean frequency of 0.033 Hz while the electric thermostat has a mean frequency of 0.011 Hz. The frequency spectrum of both signals is shown in Figure 34.

34 Frequency Spectrum Comparison 600 Electrically controlled thermostat Reference Scania Truck 500

400

300 Magnitude

200

100

0 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 Frequency [Hz]

Figure 34: Temperature frequency spectrum comparison.

If the electric actuator fails, the wax body will take full control of the flow. Offering full cooling functionality although suboptimal. This allows for safely operating the vehicle while planning a well thought out service session that doesn’t disrupt the transport schedule. The failsafe functionality can be seen in Figure 35 where the electric actuator is shut off after 1000 seconds, notice that the temperature is kept at acceptable levels allowing for continued operation.

35 Engine Temperature 105 Electrically controlled thermostat failing at 1000 seconds Reference Scania Truck

100

Temperature [Degrees Celsius] 85

80 0 500 1000 1500 2000 2500 3000 Time [s]

Figure 35: Simulated fail of electric actuator at 1000 seconds to show failsafe operation.

36 5 Discussion The developed concept of having an electrically controlled actuator with a wax body shows good promise in regards to fulfilling the product specification and future implementation. The hydraulic concept is however deemed to be less space efficient, especially when considering both oil supply and cable management. This should be due to the fact that the hydraulic actuator must be designed for the lowest pressure which is around just 1 bar at idling conditions. At higher engine speeds, the pressure can reach above 4 bar, giving 4 times the needed torque. The BLDC electric motor actuator shows more promise with smaller dimensions and consistent performance. The hydraulic actuator is also limited by the fact that the engine must the turned on for it to work. The cam solution does offer a good way of translating the rotating motion into a linear one while increasing the precision in areas where it’s needed and the life time of cam mechanisms have been proven to be well within the margins through their use in other parts of the engine. However it seems that the uncertainty of when the cam is actually engaged might negate the precision. It is also deemed that the linear precision should be sufficient even without a tailored opening characteristic, especially when accounting for dimensioning for a limited stroke and having a considerable reduction. The lead screw mechanism offers a compact and accurate translation from rotational motion to linear motion. The low efficiency might however result in high wear, which makes the life time of the mechanism questionable. A gearbox and rack and pinion solution might be more favourable than originally thought since it can offer low wear and accurate translation at a reasonable price. The size should also still stay within the margins. The prototype simulations shows that the dimensioning should be sufficient and that good valve position control can be achieved with this concept. The simulation follows the specification of the tested motors at lower speeds but loses torque at higher speeds faster than specified for the motors. As shown in Figure 25, this should be due to not having a adjustable voltage driver which increases the supply voltage to achieve a constant current at higher speeds. Implementing such a driver in the simulation was deemed to be outside the scope of this project. The modelled valve in GT Suite behaves as expected and seems to be properly modelling the control of the electric actuator, wax body and their interaction. Most of the advantages of a electrically controlled thermostat lies in the future where any data available in the truck can be used to improve the control and the target temperature of the engine can be optimized and changed for different conditions. Even with a very simplified system model and simple control, the temperature control performance is improved a great deal with both lower amplitude and frequency compared to the old NCG wax thermostat. A more advanced model and control should definitely result in additional improvements. Another important finding from the GT Suite simulation is how little the electric thermostat actually has to open to achieve good thermal control. During the used conditions, the actuator could have been dimensioned for only performing 10% of the total stroke with good results. The wax body control during the intense retarder sequence properly shows how the wax thermostat can handle the drive cases which the downsized electric actuator can not, although there is a lot of room for optimization of their interaction. It might be that the operating conditions during this reference track should be handled solely by the electric actuator in the final product. The measuring equipment was not optimized for measuring low flows, giving some uncertainty on how accurate the results are in this region. The general characteristics of the flow control should be pretty reliable though and the valve positioning was very robust and accurate.

5.1 Conclusions The work performed during this project has resulted in several viable concepts for an electrically controlled thermostat. Where the concept of having a BLDC motor combined with a wax body shows the most promise. Combining the electric actuator with a wax body allows for a substantial

37 downsizing as shown by the simulated temperature control in Section 4.5. The wax body can then act as an extra actuator that handles the few cases where higher cooling effects are needed. The wax body also gives the feature of having full fail safe function even if the electric actuator would fail. The simulated temperature control also shows improvements in following a set temperature and lowering the frequency of the temperature oscillations compared to the old wax thermostat solution even when using a basic control. This should lead to improvements in engine performance and durability. Especially as the electric thermostat opens up new possibilities where the thermal management system can be used actively instead of passively to optimize the engine performance. The dimensioning in Section 4.2 and prototype simulations in Section 4.4.1 shows that the forces and speeds needed for accurate positioning of the valve is no problem for a reasonably sized BLDC electric motor. The actuator performance for the functional prototype was as expected better than in the simulation and it showed that this solution can be used for accurate flow and temperature control while staying at reasonable dimensions. From the evaluation of the design and placement of the final product in Section 4.3, a suggestion for a modular solution that is space efficient and requires very few unique components is presented.

5.2 Future Work There is a lot of work left in order to turn a promising concept into a final product ready for production. This project have managed to evaluate some general design choices, but for the final product each part of the design has to be refined further. There is room to optimize the BLDC motor specifications and design, this includes looking at both stepper motors and servomotors in combination with gearbox and mechanisms for translating rotational motion to linear motion. Develop a more advanced control that can adapt to different conditions and use more data in the truck to improve its performance, like for example the look-ahead function could tell that a hill is coming up soon which the controller can then adapt to. The interaction between the wax body and electric actuator can also be optimized such that the electric actuator gets to work undisturbed during most operating conditions. The ability to diagnose the thermostat is quite important. This includes both the function of the electric actuator and the wax body. The possibility of diagnosing the function of the thermostat through measuring the back emf from the BLDC electric motor or some other method should be evaluated. Although the smaller electric actuator and modular design indicates a good performance on the product cost metric. A detailed cost analysis must be performed in order to ensure that the product does in fact offer a good business case. Optimizing the product for cheap manufacturing and assembly is also an important need that can help achieve a good business case. CFD and FEM should be performed to ensure that all components are durable enough to withstand the entire specified life time. The coolant flow over the thermostat can also be evaluated and optimized to decrease flow losses.

38 References [1] David J. Allen and Michael P. Lasecki. “Thermal Management Evolution and Controlled Coolant Flow”. In: Vehicle Thermal Management Systems Conference & Exposition . SAE International, 2001. doi : https://doi.org/10.4271/2001-01-1732 . url : https://doi. org/10.4271/2001-01-1732. [2] John R. Wagner et al. “Smart Thermostat and Coolant Pump Control for Engine Thermal Management Systems”. In: SAE 2003 World Congress & Exhibition . SAE International, 2003. doi : https://doi.org/10.4271/2003-01-0272 . url : https://doi.org/10.4271/2003- 01-0272. [3] John F. Eberth et al. “Modeling and Validation of Automotive “Smart” Thermal Management System Architectures”. In: SAE 2004 World Congress & Exhibition . SAE International, 2004. doi : https://doi.org/10.4271/2004-01-0048 . url : https://doi.org/10.4271/2004- 01-0048. [4] Andrew A. Kenny, Cyril F. Bradshaw, and Brian T. Creed. “Electronic Thermostat System for Automotive Engines”. In: SAE International Congress and Exposition . SAE International, 1988. doi : https://doi.org/10.4271/880265. url : https://doi.org/10.4271/880265. [5] Matthieu Chanfreau et al. “The Need for an Electrical Water Valve in a THErmal Manage- ment Intelligent System (THEMIS TM )”. In: SAE 2003 World Congress & Exhibition . SAE International, 2003. doi : https://doi.org/10.4271/2003-01-0274 . url : https://doi. org/10.4271/2003-01-0274. [6] Mickael Cormerais et al. “ACT Valve: Active Cooling Thermomanagement Valve”. In: SAE 2014 World Congress & Exhibition . SAE International, 2014. doi : https://doi.org/10. 4271/2014-01-0632. url : https://doi.org/10.4271/2014-01-0632. [7] R Wurms et al. “Innovative technologies in current and future TFSI engines from Audi”. In: (Jan. 2011). [8] Michael Weiss. “Hot & Cold Schaeffler’s Thermal Management for a Reduction of up to 4%”. In: (Apr. 2014). [9] Y. Kuze. Wax-pellet thermostat . US Patent 4,948,043. 1990. url : https://www.google.se/ patents/US4948043. [10] Mahle Aftermarket Thermostat description . http://www.mahle-aftermarket.com/media/ media-global-&-europe/products-and-services/thermostats/thermostat_broschuere_ en_final.pdf. Accessed: 2018-02-01. [11] M. Kurz and R. Saur. Electrically heatable thermostatic valve for a coolant circulating system of an internal-combustion engine . US Patent 5,385,296. 1995. url : https://www.google. com/patents/US5385296. [12] “Smart Phasing”. In: Solving the Powertrain Puzzle: 10th Schaeffler Symposium April 3/4, 2014 . Ed. by Schaeffler Technologies GmbH & Co. KG. Wiesbaden: Springer Fachmedien Wiesbaden, 2014, pp. 156–170. isbn : 978-3-658-06430-3. doi : 10.1007/978-3-658-06430- 3_10. url : https://doi.org/10.1007/978-3-658-06430-3_10 . [13] Henry B Rose. Fluid pressure rotary vane actuator . US Patent 3,237,528. 1966. [14] Karl T. Ulrich & Steven D. Eppinger. Product design and development . New York, USA: McGraw Hill, 2012. isbn : 978-007-108695-0. [15] Asa˚ Wikberg-Nilsson, Asa˚ Ericson, and Peter Törlind. Design : process och metod . 2015, p. 237. isbn : 9789144108858.

39 A Normal Operation Normal operation is deﬁned as the truck driving along the reference path from Stavsj¨oto Sillekrog seen in Figure 36. Operational data for this trip is available and can be used in the detailed simulation model in GT-suite to see performance of the old thermostat and also to evaluate and compare the performance of the concepts which are further developed and modelled.

Figure 36: Reference path goes from Sillekrog along e4 to Stavsj¨o.

The data which is used from the reference path are listed below. • Engine load [ Nm ] • Vehicle speed [ km/h ] • Engine speed [ RP M ] • Retarder brake eﬀect [ kW ] The data is sampled once a second, the engine load and rotational speed is shown in Figure 37 which can also be used to calculate the engine power.

Engine Load Reference Engine Speed Reference 3000 1800

2500 1600

2000 1400

1500 1200 1000 1000 500 Engine Load [Nm] Engine Speed [rpm] 800 0

-500 600

-1000 400 0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500 Time [s] Time [s] (a) Engine load during reference path. (b) Engine speed during reference path.

Figure 37: Engine characteristics during reference path.

The vehicle speed and retarder effect is shown in Figure 38, the retarder effect in particular will affect the temperature control.

40 Vehicle Speed Reference Retarder Effect Reference 90 45

80 40

70 35

60 30

50 25

40 20

30 Retarder Effect [W] 15 Vehicle Speed [km/h]

20 10

10 5

0 0 0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500 Time [s] Time [s] (a) Vehicle speed during reference path. (b) Retarder eﬀect during reference path.

Figure 38: vehicle speed and retarder eﬀect.

Figure 39 shows an example of simulation results when the GT suite model is used with the reference data and diﬀerent initial conditions.

Simulated Coolant Temperature After Engine - Warm Start Simulated Coolant Temperature After Engine - Cold Start 110 110

100 100

90 90

80 80

70 70

60 60

50 50 Temperature [Degrees Celsius] Temperature [Degrees Celsius] 40 40

30 30

0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000 Time [s] Time [s] (a) Temperature with warm start. (b) Temperature with cold start.

Figure 39: Temperature after engine with current wax thermostat and diﬀerent initial conditions.

41 B Concept Descriptions This appendix describes the concepts that were generated and evaluated in the last iteration of the concept generation. Concepts which perform a similar task have been preliminary dimensioned to have the same basic performance to allow for a fair comparison. It is important to note that the performance and dimensions can be changed in the detail design based on additional research and results as they are rough estimations of what might be required of diﬀerent valve types.

Concept A - BLDC Motor Combined With Wax Body This concept is based on a concept which has two wax bodies controlling a single sleeve valve. In the original concept, one wax body is in the center and reacts to the temperature at the thermostat. The other wax body is at the top and reacts to a pilot flow. This concept replaces the upper wax body with an electric actuator in the form of a BLDC electric motor with reduction. The actuator and wax body works in tandem where the one which requests the biggest opening is in control. Since the absolute majority of the flow control is done in the low flow region, the actuator can be downsized to only deliver a smaller opening while the wax body can give full cooling in the cases where that is necessary. The concept is illustrated in Figure 40.

Figure 40: Illustration of concept A.

Concept B - Wax Thermostat With BLDC Controlled Side Flow Valve This concept utilizes a wax thermostat to control the bulk of the flow while a smaller side valve is used to adjust the flow and get better thermal management. The side valve is controlled through a BLDC electric motor with reduction which drives a rack and pinion mechanism that directs the flow to either the radiator or the bypass. Since the valve controls a smaller flow, the BLDC motor can be downsized compared to concepts which must control the entire flow. The concept is illustrated in Figure 41.

42 Figure 41: Illustration of concept B.

Concept C - Hydraulic Actuator Combined With Wax Body Through Rack and Pinion This concept is the same as concept A except the actuator is a hydraulic rotary vane actuator connected through a rack and and pinion to the rest of the mechanism. The actuator is similar to the hydraulic cam phaser solution that exists today and has two rotor wings to develop more torque at a shorter angular stroke. The hydraulic control could be the same as for the hydraulic cam phaser with a central valve controlled by a Variable Force Solenoid (VFS). An illustration of the concept can be seen in Figure 42.

Figure 42: Illustration of concept C.

43 Concept D - Hydraulic Actuator Combined With Wax Body Through Cam Very similar to concept C but has only a single rotor wing which gives it less torque but a longer angular stroke. A cam transfers the torque into a linear force moving the sleeve valve. The cam can also be tailored to give a desired opening characteristic. The concept is illustrated in Figure 43.

Figure 43: Illustration of concept D.

Concept E - Wax Thermostat With Stepper Controlled Side Flow Valve This concept uses the same principle as concept B, but it instead uses an electric stepper motor to control the side valve. A stepper motor develops more torque at the cost of speed and using no gear reduction is reasonable in this case. A stepper motor can also be used without position feedback due to its ﬁxed step angle, but if it skips or misses a step the system will not know. The concept is illustrated in Figure 44

44 Figure 44: Illustration of concept E.

Concept F - Hydraulic Actuator Controlling Ball Valve This concept is based on using a hydraulic rotary vane actuator to control a ball valve which in turn will be able to control the full ﬂow through the thermostat. The rotor has two wings to develop more torque at the cost of a shorter angular stroke and it is placed directly on top of the ball valve. Seals will be needed to avoid oil leaking into the water and a spring will be used to return the valve to full cooling in case of failure. The ball valve can also be designed to give a good opening characteristic. The concept is illustrated in Figure 45.

45 Figure 45: Illustration of concept F.

46 C Concept Evaluation Six concepts were evaluated in the ﬁnal scoring, these concepts can be seen in Appendix B. The scoring results are presented in Table 4.

Table 4: Final scoring.

Metric Metric Imp. Concept A Concept B Concept C Concept D Concept E Concept F No. Product 1 5 5 2 5 5 2 5 lifetime 2 Response time 1 4 5 3 2 4 3 3 Product cost 5 3 3 4 4 4 1 Temperature control tolerance motor 4 3 3 4 3 5 4 3 Temperature gradient control radiator 4 3 3 4 3 5 Geometric 5 3 3 3 2 2 3 1 constraints 6 Weight 3 3 4 2 2 4 1 Actuator volumetric 7 2 3 2 2 4 2 5 ﬂow control accuracy Pressure loss over 8 4 4 3 4 4 3 4 thermostat 9 Radiator path leakage 4 5 2 5 5 2 5 10 Actuator power consumption 2 4 5 3 3 3 2 Total Score 124 94 115 127 94 104 Ranking 2 5 3 1 5 4 Continue? Yes No No Yes No No

Evaluation of Concept A This concept is similar to concept C and D except for the fact that it has a BLDC electric motor as an actuator. The lifetime of a BLDC electric motor is good and the rest of the thermostat should have a good lifetime since the thermostat can work in tandem with the actuator, possibly enabling some form of diagnosing of the wax thermostat function. The response time should be quite fast and due to the downsizing, the price should be moderate. It should be possible to achieve a good thermal control for all normal cases and the size and weight should be reasonable. The sleeve thermostat should have a low pressure loss and leakage and the BLDC electric motor is very eﬃcient.

Evaluation of Concept B This concept is regarded as cheap due to the possibility of downsizing the actuator. To fully function, it requires a working wax thermostat since it controls the majority of the ﬂow, diagnosing the function of the wax thermostat is hard in this solution which might lead to it often being replaced. The solution will be relatively small and light due to the smaller actuator. The wax thermostat and side valve is expected to have a moderate pressure loss and having two pathways increases the risk of leakage to the radiator. The BLDC motor is small, fast and very eﬃcient which gives a small power consumption.

Evaluation of Concept C The actuator type has been shown to be very sturdy in other applications and the wax thermostat working in tandem gives fail safe function if either of them fails. The response time should be moderate due to the hydraulic components being slower than pure electric and the actuator type has been shown to be quite cheap giving a good chance of the solution staying within the product specification. The hydraulic actuator should give a good control but it lacks the possibility of finer control in the low flow region due to the rack and pinion connection. The sleeve thermostat valve type is expected to have a smaller pressure loss than the poppet valves and allow very low leakage to the radiator. The power consumption should be moderate due to low parasitic losses and small movements.

47 Evaluation of Concept D This concept has similar performance as concept C on most metrics but the single rotor wing gives a more compact actuator since more of the inside area can be used for the angular stroke compared to a double rotor solution, the response should be slightly slower due to the longer stroke though. The cam also allows for ﬁner control since it can be tailored to give a desired opening characteristic.

Evaluation of Concept E This concept is the same as concept B except it replaces the BLDC motor with an open loop stepper motor. This should be cheaper but gives less speed and control due to the ﬁxed step angle and risk of missing or skipping a step. A stepper motor might also have a higher idle current draw which increases the power consumption and heat generation.

Evaluation of Concept F Both the ball valve and actuator is expected to last the entire life time of the truck although the fail safe is less advanced than for some of the other solutions with wax thermostats. The response time should be similar to concept C due to both using essentially the same actuator although this concept must develop a higher force since it controls the entire flow. The ball valve can be tailored to give a finer flow control at low angles and being able to control the entire flow might be advantageous on some occasions. It is also on the limit regarding size due to the sizeable valve and actuator. The ball valve should have a low pressure loss and good sealing and the power consumption should be higher than for the part flow control solutions.

Concept to Further Develop Concept D shows great promise and also works well with Scania’s modularity mindset. Therefore it seems like a clear choice for further development. Concept A also performed well and came very close in score. It is very similar and if it would be used with the cam such as in concept D, only the actuator differs. This concept and most importantly the electric actuator solution for the wax thermostat and sleeve valve control will therefore be examined further together with concept D. Concepts B and E have the same basic function as concept A but the design is less appealing in several areas as shown by the scoring in Table 4 which is the reason for not examining these concepts further. Concept C is very similar to D but the cam solution and single wing rotor is deemed to be the better choice for control and size. Concept F is the only full flow control concept and it is deemed that the gain in controllability for being able to control the entire flow does not outweigh the added cost and size since so much of the control is done in the low flow region.