Optimization of Gearshift Strategies using Road Information

Transcription

1 Optimization of Gearshift Strategies using Road Information 31st July, 2006 Anders Gaasedal Christensen, s Master s thesis 30 ECTS-point Supervisors Nils Axel Andersen & Ole Ravn Ørsted DTU, Automation Technical University of Denmark DK-2800 Kongens Lyngby, Denmark

2 Cover picture: - Dynafleet illustration & I-shift V 2512 AT Gearbox

3 Abstract Research has intensified in the area of utilizing road information in control of heavy trucks. The demand for fast transportation of goods is rising and this forces the trucks to keep a high cruising speed on the roads. In addition to this, heavy trucks are consuming an enormous amount of fuel. Saving energy - even a few pro mill would therefore be of great importance to the actors in the transportation sector, as well as society in general. This master s thesis describes the task of optimizing gearshift strategies using road information. The aim is to increase the cruising speed and lower the fuel usage. The project concerns two main parts. Firstly a feasibility study combines GPS and a three dimensional road map, and secondly a simulation model to investigate the performance of the improved controllers. The feasibility study shows that it is possible to use this information to read the road ahead of the vehicle, and make a decision to shift gears. The simulation model shows that it is possible to increase the average speed of a truck with more than 1% dependent of the road profile. It further shows that if the speed is increased with 0.5%, it is possible to save more than 1% fuel, decrease gearshifts significantly and increase minimum climbing speeds. Keywords: Automatic Gearbox Optimization, Road Information, GPS, GIS, Map Matching, Truck Model, MVEM. i

4

5 Preface This thesis completes my studies for at Master degree in Electronics. The project has been conducted at Ørsted DTU, Automation at the Technical University of Denmark, supervised by Associate Professor Nils Axel Andersen, and Associate Professor Ole Ravn. The project has been carried out in the period from the 1st of February to the 31st of July The work load is 30 ECTS point. 31st July, 2006, Ørsted DTU, Automation, Lyngby Anders Gaasedal (s001407) anders@gaasedal.dk iii

6

7 Acknowledgement I express my gratitude to all the people who have contributed with valuable help and information through the project. First of all thanks to my two main supervisors, Nils Axel Andersen and Ole Ravn. Thanks also to the Foreman, Jan Meyer Petersen at Volvo Truck Center Denmark A/S, who have contributed with valuable data and background information for the development of model and controller. Thanks to Associate Professor Elbert Hendricks who has placed a large effort within the development of the diesel engine model specially adapted for this project. Thanks also to Associate Professor Niels Kjølstad Poulsen at IMM DTU for guidance in the use of Rule Based Control and Model Predictive Control. Thanks to Erik Hellström, Ph.D. Student at the division of Vehicular Systems at the department of Electrical Engineering, Linköping University, who have provided me with authentic road data for simulations. At last, I want to thank my father Niels Nielsen, truck driver at the haulage company Gustav Larsen & Son A/S who described the problem to me at first, and has contributed with valuable real life test data through the project. v

8

9 Contents 1 Introduction Background Volvo Problem Statement Approach Expected Goals Delimitations Advantages and Disadvantages Outline of the Report - Readers Guide Applied Software System Description Existing Parts Dynafleet Online I-shift Project Components to be Developed Implementation Flow Chart Current Drive Optimization Systems Speed Control Gearbox Control Engine Control General Vehicle Control Summary Road Information GPS Communication with the Receiver GPS Parser Latitude and Longitude to UTM Coordinates User Interface GIS vii

10 4.2.1 TOP10DK KMS Map Based on GPS Measurements Laser Scan Authentic Road Data from Sweden Map Matching Road Profile Test of GPS/GIS System Summary Truck Real Truck - Volvo FM Specifikations Reference Tests Evaluation st Model - Simple Engine Automatic Gearbox Rearaxle/Wheels Vehicle Input - Real Road Cruise Control Output - Data/Gauges Evaluation nd Model - Advanced Engine Clutch Gearbox Evaluation Tests Fuel Usage Climbing Capacity Acceleration Summary Controller Reference Data Truck Data Driver Inputs Inputs Truck States Road Sequence Outputs Controllers viii

11 6.4.1 Normal - Reference Simple - Economy/Power Rule Based for 1st Model Rule Based for 2nd Model Summary Tests and Results Road Information Controllers Comparison Test Detailed Test of Final Controller Summary Discussion Road Information Model Conclusion Future Work 99 Nomenclature 100 Bibliography 107 List of Figures 111 List of Tables 115 A Models 117 A.1 1st Model A.1.1 1st Model - Main A.1.2 1st Model - Engine A.1.3 1st Model - Gearbox A.1.4 1st Model - Differential, Brakes and Wheels A.2 2nd Model A.2.1 2nd Model - Main A.2.2 2nd Model - Engine - Main A.2.3 2nd Model - Engine - Fuel A.2.4 2nd Model - Engine - Turbo A.2.5 2nd Model - Engine - Crankshaft A.2.6 2nd Model - Clutch A.2.7 2nd Model - Gearbox A.2.8 2nd Model - Differential, Brakes and Wheels ix

12 B CD 133 B.1 Contents B.1.1 Report B.1.2 Road Information B.1.3 Truck Model B.2 CD x

13 Chapter 1 Introduction Nowadays, automatic gearboxes have become more and more normal in heavy trucks. The two most important reasons for this is the improvement of drivers comfort and more economic driving. Most drivers are happy to avoid hundreds of gearshifts through a working day, and the haulage contractors are happy to know that the trucks are driven in the most economical way. However, automatic gearboxes suffers from not being able to read the road like the real driver. That makes automatic gearboxes inferior to a normal driver with a manual gearbox in hilly conditions with steep gradients. A normal driver reads the road, and decides if it is possible to reach the top of a hill in that particular gear without loosing a lot of speed, or if it would be more efficient to make a shift down immediately before the gradient to give the engine more torque for climbing the hill, and maintain a higher speed on the gradient. The normal automatic gearbox often handles this situation by using the top gear, trying to climb the hill, and waiting to shift until the engine RPM 1 drops below a certain limit. That causes the truck to lose a lot of speed, and makes it necessary to make the gearshifts when the demand for torque is at its highest. Thus the drivers normally chooses to switch to manual gear control and select the best gear immediately before a certain gradient; thereby making use of the full potential of the engine torque on the whole gradient. Automatic gearboxes for trucks are very expensive, and every time the driver chooses to select gears manually it can be conceived as a waste of money to have bought an automatic gearbox. Since the driver believes he/she can make a better selection of gears than the computer, he/she must be using some senses or earlier experiences that the computer does not have access to. The aim of this project is to develop a system for automatic gearboxes to get closer to the behavior of an experienced driver. 1 Revolutions Per Minute 1

14 1. Introduction 1.1. Background 1.1 Background The background for this project has been formed by several driving hours with my father who is a truck driver. The truck he has been driving for some years is equipped with the new automatic gearbox from Volvo, the I-shift system, which will be explained in detail in chapter 2. This gearbox is mostly a normal manual gearbox but controlled by a computer instead of the driver. The gearbox is very good at selecting the correct gear compared to the load of the truck, the speed and engine RPM. Unfortunately it is not capable of reading the road as the driver can. The problem has been in my thoughts for some time, how to solve the problem and make the gearbox capable of selecting the gears in a more efficient way according to the profile of the road ahead of the truck. Several ideas for using different sensors have been considered in order to find a possible solution. When the idea of using an integrated GPS 2 system together with a 3D digital map of the road came up, contact were established to supervisors at DTU to determine if it could be the subject for my master s thesis, and another contact were established to the Volvo Powertrain Cooperation to ensure that the idea might be useful. As described earlier I had access to a Volvo truck with an automatic gearbox, and the amount of possible equipment to a Volvo truck, including an online transportation system 3 and a GPS system. These factors determined to base this project on the Volvo system. Other large truck manufactures like SCANIA AB, Mercedes-Benz, M.A.N. AG are also using and developing automatic gearbox systems, which are somehow similar to the Volvo system. 1.2 Volvo Volvo was founded in 1927 and in 1928 the first Volvo truck rolled off the production line, at that time a pioneer on the market, shaft driven (compared to other trucks with chain drive), pneumatic tyres (solid rubber tyres were normal at that time) and a fully enclosed cab. Through the years Volvo has been a pioneer in many areas. In the 1950s Volvo was the first to use turbocharged engines, and in the 1960s Volvo introduced tiltable cabs for easier access to 2 Global Positioning System. 3 The Volvo Dynafleet Online system, will be explained in chapter 2 2

15 1. Introduction 1.3. Problem Statement the engine. Nowadays Volvo is still in the front rank of truck production by researching many areas for improving the trucks, including drivers comfort, environmental-, safety- and economical aspects, etc. The idea for solving the problem in this case includes several of these advanced techniques, some of which are already implemented in the Volvo trucks today (to be explained in chapter 2), and the opportunity for testing the system or later implementation will be natural using the Volvo system as background. 1.3 Problem Statement The project is a result from own experiences with automatic gearboxes, and several statements from truck drivers complaining about the inability of an automatic gearbox to read the road, [Magazine, 2006b] and [Magazine, 2006a]. Thereby the problem statement can be defined as: Develop a system that is capable of improving the gearshift strategies for automatic gearboxes by reading the road and selecting the optimal gear sequence for particular gradients. The system will include a combination of GPS and GIS 4 to obtain knowledge of the road in front of the vehicle, and thereby select the optimal gear sequence and the time of shifting. The gear sequence should be chosen by combining the road data and the characters of the vehicle, engine, load, speed, etc Approach The project initiates with an investigation of the GIS/GPS system: whether it is possible to use such a system, and which parts are available on the market. Afterwards the project continues with the development of a simulation model, to be able to simulate the behavior of a truck, driving a road with gradients. The output of the model is likely to be the speed of the truck and the fuel consumption, these are the main areas of concerns. The model will at first be adapted to fit a real truck driving on real road conditions, and then extended with a new gearbox controller to give a natural view of the eventually gained performance Expected Goals The project has several expected goals, both theoretical and practical. The theoretical aspect is to develop a simulation model that can be used to show the behavior of a real truck, driving on both artificial and authentic 4 Geographical Information System - in this case, digital 3D maps. 3

16 1. Introduction 1.3. Problem Statement road sections. The model must also include the properties of the whole drivetrain thereby giving the possibility of creating rules for a controller to improve the gearshift strategies. In the practical aspect it is important to investigate whether it is possible to use a GIS/GPS system to control the gearbox of a truck. The result of the simulation model also includes a practical goal since it is desirable to show a result as close to reality, to give a clear view of the advantages and disadvantages of such a system implemented in real life Delimitations Due to the limited amount of time, some limits have to be set for the project. The probability of making a system which can be implemented in a real truck is impossible, since it would require an adaption of the project to fit the system and software used in real trucks. Besides from that, a lot of tests are required to make sure it will work on a truck and not damage any parts of the truck nor make the vehicle unsafe to drive on real roads, during tests. The investigation of using a GIS/GPS is also limited to some kind of a feasibility study, whether it seems possible to implement and use such a system. A fully functional system will require more data and tests than it is possible in this project Advantages and Disadvantages The project includes several possible advantages and disadvantages using an optimization system, here are the most important. Advantages Higher average speed By using the optimal gear for getting the maximum torque it will be possible to increase the speed on steep gradients. Improved fuel economy It should also be possible to get an overall fuel saving in some cases even though the gearbox is controlled for getting maximum torque. This is possible because the vehicle will not lose as much speed, and speed changes are normally uneconomical. Improved safety in relation to other vehicles If the vehicle can maintain a higher speed on a gradient, the difference in speed compared to other vehicles will be smaller and the risk of accidents reduced. Less wear If gear shifts are made when the load on the drive train is small, wear will be reduced. Better use of the automatic gearbox The automatic gearbox will be used more for automatic shifting, and the capital which is invested in this technology will be better utilized. 4

17 1. Introduction 1.4. Outline of the Report - Readers Guide Improved drivers comfort The driver will be relieved of makeing some gearshifts, specially in hilly terrain. This will increase comfort. Improved safety The driver can fully concentrate on the traffic; it is not necessary to think about selecting the right gear. Disadvantages Increased gearbox wear Making more gearshifts to maintain higher average speed will cause more wear in the gearbox. Poorer fuel economy In most cases the fuel economy will be poorer, since it is uneconomical to run in a lower gear than the highest gear. Price of purchase Since the system is expensive to develop it will be more expensive than existing systems. Not developed for all areas If a 3D map of the road is not available, the system cannot work, and must rely on the standard shifting strategy instead. In some cases the above mentioned factors will affect each other in a positive or negative way. Increased speed will often entail a worse fuel economy. This means, that some kind of trade off between the different aspects will probably be the final solution. Or the final solution will be a number of solutions, each designed to improved different conditions. 1.4 Outline of the Report - Readers Guide Abbreviations and terminologies used through the report are presented in a the end of the report at page 100. List of figures and tables are also to be found at the end of the report, at page 113 and page 115 respectively. A CD is attached at the back of the report, containing the report itself, program code, and other relevant data. References to used literature are marked with [Author, Year] which refers to the Bibliography on page 106 The main chapters of the report contains the following subjects: Chapter 2 - System Description Description of the whole system in two parts, first the current available hardware and software which the system should be built on top of. And second a description of how the new parts for the system should be implemented. 5

18 1. Introduction 1.5. Applied Software Chapter 3 - Current Drive Optimization Systems A summary and description of currently implemented or tested drive optimization systems which uses road information to improve the control of engine, speed, gearshifts, etc. Chapter 4 - Road Information In this chapter an investigation of procuring a usable 3D road map for this purpose has been carried out, both in the matter of using existing data and also with respect to creating maps from scratch with a GPS receiver. Chapter 5 - Truck This chapter describes the development of a detailed truck model used for tests and simulation of the entire system. Chapter 6 - Controller Three different controllers for this system have been developed and tested in this project, the normal system, a simple improvement, and the final solution. Using more or less of the available road data. Chapter 7 - Test Tests and simulations of the system are presented in this chapter, including improvements compared to current systems. The last chapters contains the discussion of the results, conclusion, and ideas for future work and further development of this system. 1.5 Applied Software The development of software modules and models has been carried out on a standard PC installed with Windows XP Service Pack 2. MATLAB version (R14) Service Pack 2 SIMULINK version 6.2 (R14SP2) 6

19 Chapter 2 System Description This project consists of several different parts, some of which are already developed and used in trucks, and some of which are new and developed in this project. The project includes three main parts: Road information which has to be communicated from the new integrated GIS/GPS system to the onboard computer on the truck, to make the truck capable of using the road profile ahead for control of the gearbox. Truck which is the main part to be controlled. In this project it will be a model on a computer. However if the system indicates improved performance the possibility of implementation in real life could be investigated. Controller which by using the information above is capable of calculating the necessary action to be carried out, to make the shifting strategy more efficient, using the road profile ahead. 2.1 Existing Parts As described in the previous chapter, a Volvo truck is chosen as the reference system, because the project will include parts that are already developed and used in Volvo trucks today. Some of those parts will not in the first attempt be used in the project but since they are already developed, the final solution will not be a very expensive and totally new accessory to the trucks. Instead it will consist of already developed parts and an extra module for linking these together in a new way. What currently are useful and available at Volvo is their transport information system - Dynafleet Online, and an automatic gearbox - I-shift. This project will be an extension to these two systems making use of the Dynafleet Online system to control the automatic gearbox. 7

20 2. System Description 2.1. Existing Parts Dynafleet Online Dynafleet Online is a highly developed Transport Information System (TIS) from Volvo. It is developed to improve logistics and communication between the hauler and the driver. It includes three main parts: Vehicle Management which reads out the vehicle data such as fuel consumption, speed, RPM, etc. It also stores it and transmit it to the hauler for statistic purpose and to monitor the driver. Driver Management which tracks the driving and working time of the driver, to prevent violations of the legal driving time and to maintain service intervals, etc. Transport Management which gives the driver and hauler the position of the truck to make it easier to plan the most efficient route. It also includes a communication module between the driver and hauler. This system also includes traffic reports, fuel stations, workshops etc. to help the driver navigate through traffic as smoothly as possible. For this project the Transport Management part is the most important, since it includes a navigation system based on GPS and digital maps, and a communication part, to send and receive messages and data between the truck and the hauler. As for now, the integrated maps are in 2D, which has to be extended into 3D for this project. Further description of these extensions will follow in chapter I-shift In 2001 Volvo presented the automatic gearbox, named I- shift. This is a fully automated gearchanging system which offers improved driver comfort, reduced fuel consumption and lower gearbox weight. The gearbox is mostly a normal gearbox controlled by a computer instead of the driver. The gearbox is a 12- speed unsynchronized range 1 /splitter 2 gearbox, and the 1 Range is a High/Low gear area resulting in 6 gears from the normal 3 gears. 2 Split is another High/Low gear, but splitting in between gears instead. (Total 12 gears) 8

21 2. System Description 2.1. Existing Parts computer controls the clutch and the engine for smooth up/down shifting. The computer uses different sensors for selecting the optimal gear such as the weight of the truck, gradients for selection of start gear, etc. To give a clear understanding of the normal shifting strategy, the most important parameters are listed here in the section below. Basic Shift Strategy The basic shifting strategy of the I-shift program is rather simple. At launch the gearbox calculates the most efficient gear to use as startup gear ( gear), by measuring the weight of the truck and the gradient of the road. When the truck is accelerating or decelerating, the computer calculates the most optimal gear selection. If it is convenient to shift 1,2 or 3 gears up or down, dependent on the acceleration of the vehicle. If the truck has no load, it prefers to use few gears, e.g If the truck is fully loaded it may use all gears When the engine speed rises above a certain limit a higher gear is selected, or when the engine speed drops below a certain limit a lower gear is selected. These limits for shifting is affected by the acceleration of the vehicle and the throttle level. Additional Strategies A number of additional strategies can affect the selection of gears, here are the most important. Performance Shift Function to give faster and more gentle gear shifts by using the engine brake, the clutch and a special transmission brake in the gearbox. Basic Gear Selection Adjustment The driver can select gears manually during engine braking. Enhanced Shift Strategy Improved strategies for launch and precision driving by controlling brakes and suspension. When gearshifts are performed during engine braking, the wheel brakes are activated to compensate for the lost braking moment. This improves breaking efficiency and comfort. Eco Roll Activation of a special neutral function, lowering the fuel usage. Eco Roll is used on road sections where no engine or brake power is needed. Smart Cruise Control This works together with the brake pilot and ensures that the brakes are not used unnecessary. The system is intended to disable the brakes at the end of descents and let the vehicle run a 9

22 2. System Description 2.1. Existing Parts little faster than desired before the road levels again. In this way it uses the momentum of the vehicle. High Torque Direct Gear In top gear the engine is allowed to produce approximately 200 Nm more than in other gears, because the top gear is the most efficient gear. The top gear couples directly through the gearbox and therefore more efficient. Heavy Duty GCM 3 Controller from Volvo, optimizing gearshifts for heavy trucks of more than 60 tons. Launch Control This system helps the driver to start the vehicle on steep gradients. The breaks are applied automatically until the torque delivered from the engine is large enough to drive the vehicle. Economy/Power Mode The driver has another possibility of controlling the behavior of the gearbox. Right behind the gear lever is a button marked E/P, referring to Economy- or Power mode see figure 2.1. In the Economy mode the gearshifts are executed at the most economical engine speed, which are often rather low. In the Power mode these limits is moved up higher because the engine produces more torque at higher engine speed. The limits are moved upwards both in the shift up and shift down case, that means if the vehicle is accelerating it will stay longer in that gear before the shift. Similar when the truck decelerates because it can not cope with a steep gradient, it will Figure 2.1: make the shift down earlier, i.e., engaging the lower I-shift gear gear earlier. This is not always an economic solution but the driver gets more power for maintaining speed. lever This is actually what this project is about. The aim of the final controller is to get more power by selecting a lower gear when needed compared to the normal shifting strategy. An idea for a simple controller is to automatically activate the Power mode when driving into hilly terrain. This will be implemented and described in chapter 6. A combination of the already implemented gear changing strategies and some new ones will be developed in this project to particularly improve driving in hilly locations. 3 Gross Combination Mass 10

23 2. System Description 2.2. Project Components to be Developed 2.2 Project Components to be Developed To extend the normal system to be able to read the road several components have to be developed, either new or extensions for existing parts. The project contains three superior parts to be developed: Extension for the Dynafleet Online system The normal 2D positioning system in the Dynafleet Online system has to be extended with a 3D map to make the computer capable of reading the road height profile ahead of the truck. This part of the project is described in chapter 4. Truck Simulation Model For the development and evaluation of a new controller, a truck simulation model is needed. The model should reflect reality as close as possible. The model should contain parts like: engine, gearbox, drivetrain, disturbance from road, air resistance, etc. Description and implementation of the model is to be found in chapter 5. Controller The controller can be developed on the background of the model and later evaluated on the model to investigate whether the system is advantageous or not. If the result is positive, and the model can be verified to reflect the reality, there is an incentive to implement this controller in a real truck. Three different controllers are implemented two simple controllers using only a few inputs, and a more sophisticated controller using more inputs for making the decisions. The implementation and description of the different controllers are to be found in chapter Implementation The system to be developed here is implemented as depicted in the flowchart in figure 2.2. The position of the truck on the road is found by the Road Information part, which contains a Map Matching algorithm that combines a GPS position with a 2D digital road map. When the position is found it is compared to a 3D digital road map and the 3D profile of the road ahead is transmitted to the Controller. The controller evaluates the road profile in front of the vehicle, by means of the important physical factors such as gradient, length of gradient, etc. This result, stationary data of the truck (weight of truck, engine size, etc.) and dynamic data of the truck (speed, RPM, gear, etc.) are compared and evaluated, and it is decided if and how the gear shifting algorithm should be affected. The result is transferred to the Truck model and the result of these decisions and the disturbances from the road the truck is driving on, 11

24 2. System Description 2.2. Project Components to be Developed can be measured at the output of the model including information of the fuel usage, average speed, number of gearshifts etc Flow Chart The flowchart is separated in the three main parts referring to the three parts implemented in this project. Besides inputs from the three parts, the model also includes an external input, namely the road on which the truck model is driving. This road is both artificial road sections and authentic road sections. Figure 2.2: Flowchart of system implementation intended here including Road Information, Truck Model and, Controller. 12

25 Chapter 3 Current Drive Optimization Systems Research has already been carried out within this area - automotive systems, based on optimization from 3D road maps. This chapter describes some of the most important investigations in this area. The area is however rather new, because positioning systems like GPS have not been available for public before the 2nd of May 2000, when it was decided to remove the international degradation of the civil GPS signal 1. Since that time several manufacturers of automotive components have started researching in the area of using positioning systems and digital maps to improve different facilities in vehicles. These manufacturers includes many of the large truck manufactures and manufactures of utilities for trucks, these includes: Volvo, Scania, MAN, Ford, Daimler Chrysler, Cummins and Bosch. More research has been carried out at the division of Vehicular Systems at the Swedish Linköping University and the Chalmers University of Technology in Göteborg. A Swedish report [Gustafsson, 2006] has made a collection of references and descriptions of different researched systems. All systems described in this chapter, which are not cited, refer to this thesis. The chapter is intended to give the reader an overview of the different applications which can be optimized using positioning system. Bits and pieces from the different applications has been used in the development of this project. The research mainly concentrates around 3 areas: speed-, gearbox- and engine control. 1 Before that a noise signal causing errors from 0-70 meters were added to the GPS signal. 13

26 3. Current Drive Optimization Systems 3.1. Speed Control 3.1 Speed Control The most popular area for using a road information system is for predictive speed control, via this optimization of fuel efficiency. The speed control includes different methods, the most important are listed here: ACC Normal ACC 2 keeps a certain distance to the vehicle in front, but sometimes it is appropriate to change that distance. MAN has tested a system which increases the speed a little near the end of a descent or near the beginning of an ascent. And similar, decreases the speed a little before the end of a descent or the beginning of a descent, which is claimed to save fuel. Limiting Speed in Curves Heavy vehicles can not maintain full speed in small radius curves, and the cruise control could therefore be modified by means of a positioning system. (This is normally not a problem on highways). Scania and Mazda have described a system to calculate the maximum safe speed to negociate a curve, and thereby affect the cruise control. Adapt Speed to Road Type The possible cruising speed is often limited by the type of road, i.e., small hilly roads with high curvature. BMW and Aisin have tested a system based on information from a positioning system. Predictive Cruise Control This is the most efficient use of a positioning system to find the most economical speed compared to the profile of the road ahead. That means driving like a normal experienced driver. Increase speed a little before an ascent, and lower the speed a little before the end of the ascent. Similarly lower the speed before a descent, and increase the speed a little before the end of the descent. The fact that the most 3 efficient way of using a positioning system is to control the speed of the vehicle with reference to the road profile, makes it the most investigated area. Two new master s theses from the Swedish Linköping University, investigate the performance possible using Predictive Cruise Controllers. Explicit use of road topography for model predictive cruise control in heavy trucks [Hellström, 2005] This project describes the implementation of a MPC 4 controller for 2 Adaptive Cruise Control 3 What investigations shows at this time. 4 Model Predictive Control 14

27 3. Current Drive Optimization Systems 3.1. Speed Control finding the optimal speed profile to save fuel, without lowering the average speed. The system is tested by simulation on a real section of a Swedish highway documented by Scania. The results shows a fuel saving of around 2.5% without changing the average speed. The test is performed on a 40 metric tons vehicle. The fuel savings are obtained by letting the vehicle increase itsspeed a little over the desired cruising speed before gradients, and reducing the speed slowly until the top of the gradient is reached. Similar for descents, the speed is decreased a little before the beginning of a descent, and slowly increased until the end of the decrease., as depicted on figure 3.1. Altitude [m] Speed [km/h] Position [m] Figure 3.1: Predictive Cruise Control Vehicle control using look ahead information [Wingren, 2005] This project is focused on the same problem as above, but the controller in this project is to use a set of simple rules for describing the road profile, and make simple control rules from this. This method is claimed to save from 1.5% to 3.4% fuel. The Predictive Cruise Control - DaimlerChrysler [Lattemann et al., 2004] Representatives of DaimlerChrysler has made a system, also using a 3D digital road map, and a predictive cruise control. The fuel consumption is lowered with 4.1% - 5.2%, with an increase in traveling time of 0.3% - 1.4%. Predictive cruise control systems do have a large potential for saving fuel, between 2%-5% dependent on the system and the road conditions. The disadvantage of the system, compared to traffic conditions is the fact that the system is based on variations in traveling speed, both up and 15

28 3. Current Drive Optimization Systems 3.2. Gearbox Control down. This is only possible if the cruising speed is below the maximum speed limit. Most drivers today are forced to drive as fast as possible. This is required both by the hauler, because time is money and because the driver wants to get home as fast as possible. Trucks are therefore mechanically (or today by software) limited to drive right below 90 km/h. This speed is then the most normal for heavy trucks on highways, whether its legal or not. The Predictive Cruise Control is therefore only effective if it is possible to convince the drivers to drive slower than the maximum speed. 3.2 Gearbox Control Several projects have been carried out in order to improve the shifting strategy of automatic gearboxes by using some kind of road information. Below, the most interesting are listed. Predictive Gearshift Control Volvo Trucks have described a system in a patent, (also described in the report [Gustafsson, 2006]) which uses information on the road ahead and calculates the optimal gear shift strategy with regard to driver chosen parameters, such as emission rates, average speed or fuel consumption. The system uses a GPS and a digital road map to predict the future road profile. The decisions for gear shifts is also based on several internal sensors in the truck. The system also includes engine, turbo and transmission characteristics. This system description is close to the problem statement of this project, but unfortunately it has not been possible to get access to information of this patent. Driving Force Nissan has described a simple system to calculate the necessary driving torque on the road ahead, and from this decide whether it is advantageous to shift gears. Aisin has tested a somehow similar system to the Nissan system, but this system uses simple rules together with a road map to decide whether a gear shift is necessary and most of all fuel efficient. The system also describes a strategy to turn off accessories like airconditioning fan and defogger when full power is needed for an ascent. Road Grade Honda has tested a system to estimate the road grade, by comparing the actual acceleration and the predicted acceleration, this variable affects the gear shifting strategy. A navigation system is used to determine which type of road the vehicle is travelling on. These parameters are used to affect the shifting strategies. The drawback of this system is that it includes no predictivity, but it shows some improvement using the two variables above. 16

29 3. Current Drive Optimization Systems 3.3. Engine Control Learn from Experience IBM has tested a special system where the optimization is based on learning. GPS positions, gearshifts and fuel efficiency are logged when driving, and when the vehicle return to a known road it learns from the previous pass. It uses this information for making a more efficient gearshift sequence on repeated passes. Geographical Position Toyota has tested a system where the gearshifts are affected by the position, if the vehicle is in residential areas, or hilly terrain. Other Implementations Several other systems have been implemented systems to prevent unnecessary gear shifts in corners (2D maps as reference), systems to prevent unnecessary gear shifts near obstacles (2D maps as reference) and systems to shift up if traction is too low for the particular gear. Several systems has been implemented, improving the control of automatic gearboxes in different ways. Unfortunately none of them has any reference to specific measured improvements, which makes it hard to compare these to the results of this project. The amount of research in the area makes it clear, however that this is an interesting area. 3.3 Engine Control Using a positioning system also gives a number of possibilities for controlling the engine, regard to demands for torque, economy and emissions. Several techniques have been tested, based on different levels of detail in the available engine map. EGR Volvo has tested a system controlling the EGR 5 valve, which is an emission control technique. By recirculation a small amount of exhaust gas into the intake of the engine, the nitrogen oxide [NO x ] emissions can be lowered. Unfortunately the EGR system decreases the output power of the engine. The system implemented is used to control the EGR valve before gearshifts, to avoid incomplete combustion and to increase torque on ascents. SCR Scania has tested a system to control the SCR 6 system. SCR is an after threatment of the exhaust gasses, where an urea-based additive (AdBlue) is injected to lower the amount of nitrogen oxides. Since the additive is expensive it is only possible to control the amount of injected AdBlue in accordance with different areas, geographicalor jurisdictional area restrictions. The system is not predictive and 5 Exhaust Gas Recirculation 6 Selective Catalytic Reduction 17

30 3. Current Drive Optimization Systems 3.4. General Vehicle Control therefore it does not take attempt to improve the efficiency of the SCR system during gearshifts or changes in workload. Fuel Maps In some systems, engine outputs are controlled by a so-called fuel map. Cummins has tested a system to use different fuel maps in different locations. Three fuel maps have been implemented: Low emission - for urban areas. Fuel economic - for rural areas. High engine output - for hilly locations. Valve Timing Another way of controlling the output and emission of an engine is to control the valve timing. Volvo has tested such a system. The idea is to change the operating mode of the engine to achieve high output or low emissions. The system is predictive and is adapted to lower emissions at gearshifts and accommodates high torque demands in hilly locations. Turbo Volvo has tested a similar system to that above, to control the geometry of the turbo or change the limit of the wastgate 7 valve. In this way it is possible to eliminate turbo delay and increase the engine output when it is needed. Engine Output Cummins has tested a similar system as above, including more parameters to give either low emission or high engine output dependent on location of the vehicle. Vehicle Accessories Aisin has tested a system to control vehicle accessories like aircondition, defogger, fan, etc. In accordance to road data. When high torque is required for driving, some of these could be switched off. It is clear that Engine Control can be improved by the use of positioning systems. The primary benefits are improved fuel efficiency, less emissions, quicker response in transient conditions and engine output on demand. 3.4 General Vehicle Control Based on a number of the above areas investigated above, a group in Sweden consisting of employees at Scania CV AB, Södertälje, the Royal Institute of Technology, Stockholm, and Linköping University are collaboration on research in the area of vehicle control. 7 The wastgate valve opens when the maximum turbo-pressure is reached and bypassing the turbo. 18

31 3. Current Drive Optimization Systems 3.5. Summary This project [Ivarsson et al., 2006] is part of a large project named IVSS 8 to improve the safety of driving vehicles. The vehicle control project includes several subsystems like the systems mentioned above. 3.5 Summary From the above it is clear that it is possible to improve vehicle control in many ways using digital road maps with more or less detail. In the above sections, different applications are presented separated in three main parts Speed Control, Gearbox Control and Engine Control. These three areas are important, specially in commercial vehicles like heavy trucks. On figure 3.2 the most important applications is listed according to the available details in the digital map. Figure 3.2: Application for GPS/GIS aids in vehicles, according to the amount of detail available in digital maps. 8 Intelligent Vehicle Safety Systems is a program run by the Swedish road administration and the Swedish vehicle manufacturers. 19

32

33 Chapter 4 Road Information This chapter is the first main subject of this project, namely the investigation of procuring a 3D road map to be used as input in the controller for the gearbox. The chapter describes several solutions available to create a map, using existing data or creating maps from scratch with a GPS receiver. This part of the project does not include a final solution instead it is a feasibility study for the use and creation of 3D maps for this purpose. This part includes 4(5) main parts, GPS, GIS (2D and 3D), Map matching and Road Sequence. These are linked together as depicted in figure 4.1. GIS Figure 4.1: Road Information part of the project. data is used as a reference to the real world, first in the 2D version together with a GPS and a map matching algorithm to find the exact position of the truck. After this the road sequence is found using the 3D GIS data. 21

34 4. Road Information 4.1. GPS 4.1 GPS The Global Positioning System (GPS) or the NAVSTAR-GPS 1 (original full title), is a system that is able to calculate a very precise absolute position on the earth, by means of a receiver and a number of satellites. The system measures the distance to different viewable satellites, so that it is possible to calculate the absolute position on the ground with an accuracy of a few meters. The GPS is also capable of calculating the time, speed and direction of the receiver. This project is not highly dependent on a very accurate positioning system, but for other purposes where better accuracy is needed. The GPS system can Where a GPS Figure 4.2: Holux GM-210 GPS receiver be extended to the Differential-GPS (DGPS) system. reference station is used to calculate any deviation from the actual position measured. This deviation holds for a radius of around 200 km from the reference station and the accuracy of the position measurement is around 1 meter.[zogg, 2002] A GPS system has been implemented many times but to make sure it is possible to get this part of the system working, (together with a 3D map), a simple GPS system is implemented. For this purpose a standard GPS receiver is used to test the system, placed at the disposal to the project from the supervisors. The GPS receiver is a Holux GM-210 GPS receiver, with a USB interface Communication with the Receiver Communication with the GPS receiver is made with MATLAB 2. The rest of this project is also implemented in MATLAB, which makes communication and transfer of data simple. A simple program handles the read out of data from the GPS, it is enclosed on the CD in appendix B, MATLAB file OpenGPS.m and read gpsstring.m GPS Parser Data from the GPS are received in different packages that are parsed into usable data. A number of different types of packages are received. They include data of respectively position, speed, heading, satellites in view, time, 1 NAVigation System with Timing And Ranging Global Positioning System 2 MATLAB, a mathematical computing program, developed by MathWorks 22

35 4. Road Information 4.1. GPS etc. These data are parsed in the MATLAB file parse gpsstring.m in appendix B Latitude and Longitude to UTM Coordinates The positions received from the GPS and parser are in the Latitude 3 and Longitude 4 format. For this project this coordinate system is not very appropriate since it is based on degrees, minutes and seconds. Thus the coordinates are transferred to UTM 5 coordinates which are a conversion to a 2D plane. The conversion is rather comprehensive and can be seen in the MATLAB file ll2utm.m in appendix B User Interface To present the result graphically the MATLAB GUIDE 6 has been used. MATLAB GUIDE is also compatible with SIMULINK 7. A test of the GPS receiver is performed by plotting the positions onto a known map. On figure 4.3 a screenshot of the first version of the program can be seen. It reads data from the GPS, and plots it onto a small map which has been aligned with an aerial photo 8 of the surroundings to the building where the receiver was located. Besides from plot the program also validates the Figure 4.3: Program for testing GPS position in a simple way by showing the number of visible satellites. If it is 3 Horizontal lines running around the globe. 4 Vertical lines running from pole to pole. 5 Universal Transverse Mercator Coordinate 6 Graphical User Interface Design Environment 7 Simulation tool for MATLAB 8 Provided by the internet based aerial photo service, Google Earth. 23

36 4. Road Information 4.2. GIS above 4 it makes a green marker, 4 equals yellow, and below 4 equals red. A lot more information can be parsed from the GPS data, but to make an initial test this is satisfactory. The next version of the program includes simple GIS data, (see section 4.2 depicted in figure 4.4. The data from the GPS has now been converted from latitude and longitude to UTM coordinates. Figure 4.4: Program for testing GPS and GIS Altitude, Speed and Heading have also been parsed from the GPS signal, and made visible by the program. Test data can also be seen in section where a short test run is made from Building 326, DTU to Nærum and back again. 4.2 GIS Global Information Systems (GIS) is the appellation of information of landscape, geographical areas, topography, etc. In this project it is a 3D map of the roads where this system is going to be used. 3D maps are known from the normal 2D maps used in the GPS navigation systems commonly used in cars. The only difference is that this map also includes the height above sea level of the roads. From this, gradients can be calculated and used as input to the gearbox controller, whether it is necessary to change the shifting algorithm. This chapter includes investigation and description of 5 different maps: 24

37 4. Road Information 4.2. GIS TOP10DK Create a map from the digital Danish map TOP10DK 9. By combining layers of road and altitude. KMS Create a map by reading data from the online topographical map at KMS 10. Map based on GPS measurements Log GPS data (position and altitude) and create a map from this data. Laser Scan New technology has made it possible to laser scan landscapes from airplanes. Authentic road sections from Sweden Two pieces of authentic road measurements from Sweden have been made available for testing. Further development Discussion of the need for extra development in this area TOP10DK This map has been placed at the disposal of the author by the Danish National Survey and Cadastre, KMS. The data is limited to the local area, Søllerød municipality. The TOP10DK system is a number of different layers, which digitally describes different topographic features. One layer for highways, one for main roads, one for heights, etc. Two of the main themes, the highway theme and the DEM 11 theme, Northern [UTM Zone32] x Eastern [UTM Zone32] x 10 5 Figure 4.5: TOP10DK 2D Highway theme of Søllerød municipality Northern [UTM Zone32] x Eastern [UTM Zone32] x 10 5 Figure 4.6: Part of TOP10DK DEM theme, of Søllerød municipality 9 Topographical, 1:10.000, Map of Denmark 10 Kort & Matrikelstyrelsen 25

38 4. Road Information 4.2. GIS have been combined to find the altitude of every point at the highway. The two themes are depicted in figure 4.5 and figure 4.6. Abstract from the long straight lines in the two themes which are due to the simple rough program developed for showing the results. The combined results can be seen in figure 4.7 where the map has been tilted for better viewing of the elevation of the highway. Again, abstract from the long Altitude [m] x Northern [UTM Zone32] x 10 5 Eastern [UTM Zone32] Figure 4.7: GIS 3D Highway of Søllerød municipality straight lines. Evaluation During the further work with this map, a problem was encountered. The highway theme includes a lot of exit turns and lay ins, which are not necessary for this project. Several algorithms were tested to remove the extra entries, but it proved to be difficult and another solution is preferable. Parallel to the work with the map, an enquiry was placed at KMS to obtain road data of the two steepest gradients on Zealand. Unfortunately it proved to be very expensive, around 15 e/km. This was considered to be too expensive for the problem in hand. Another problem with this solution is that the height of the highway is not that accurate since the height is deduced from the altitude curves of the landscape, and not from the road itself. This problem that might be solved by means of filtering the data, since the altitude of the highway in the landscape in some way can be seen as a lowpassed filtered version of the altitude of the landscape. 11 Digital Elevation Model 26

39 4. Road Information 4.2. GIS KMS The solution for using online mapdata from KMS, is to read out the data manually from an online map. This sounds rather time-consuming, but it appeared to be possible for smaller pieces of road. The occasion for trying this solution was the desire for two certain parts of the highway on Zealand, the two steepest gradients. Ordering those pieces of road directly from KMS cost 100 euros, a large expense taking in consideration this as a feasibility study. KMS has an online 1: map at their website where it is possible to point at a place in the map, and the program will return the UTM position. The height can be deduced from the surrounding contour lines. About 15 minutes of work resulted in a 3D map of the two required gradients, and a few ours of work resulted in data for three pieces of road on Zealand, around 150 km of highway. Route 21 From Holbæk to Folehaven(Copenhagen), interesting because it includes a steep gradient near Holbæk, called Elverdammen. E20 From the Great Belt Bridge to Slagelse, interesting because it includes a steep gradient near Vårby. E47/E55 From Helsingør to Lyngby(DTU), interesting since it is close to DTU and therefore very usable for testing both GPS and GIS. Figure 4.8: GIS 2D Data of Sealand 27

40 4. Road Information 4.2. GIS In figure 4.8 the roads, read off the online map in 3D are depicted, superimposed on a map of Zealand. On figure 4.9 the roads are depicted in 3D, the altitude has been multiplied with 100 to make it more visible. Sealand Altitude x 100 [m] x Northern [UTM Zone32] x Eastern [UTM Zone32] Figure 4.9: GIS 3D Data of Sealand Evaluation The accuracy of these maps is more or less the same as the result from the TOP10DK. The altitude is constructed from landscape height curves, but can be adapted by means of filters. Due to the manual readout of this map faults occur when reading positions or altitudes, but single major faults are easy to find and correct when checking the map. The 2D parts of these maps has been validated by x GPS measurement GIS Northern [UTM Zone32] Eastern [UTM Zone32] x 10 5 Figure 4.10: 2D GIS and GPS from E47/E55 test 28

41 4. Road Information 4.2. GIS plotting GPS positions along them, while driving in a car, see figure The result looks resonably accurate, the GPS positions follows the GIS data of the road smoothly. Validation of the 3D data is a more difficult task, since the only measurement of this is the GPS altitude which is not very accurate. A test has been performed, driving from DTU to Nærum, and back again on the highway. The altitude from the GPS has been logged, and plotted together with the GIS profile of the road. The result is depicted in figure It is not GPS measurement GIS Altitude [m] Northern UTM 32 x 10 6 Figure 4.11: 2D Profile GIS and GPS from E47/E55 test possible to say whether the GPS or the GIS data is the more correct from this test, since real accurate data of the road are not available. It is obvious that the road has not moved between the way out and the way home, which the GPS data shows. The result however shows coherence, primarily in the trend of the curve. Some offset is seen in the altitude but that is normal for GPS measurements. During test driving it was noticed that the GPS positions was more accurate, compared to the visible profile of the road. Therefore the possibility of creating a map, based on GPS measurements was investigated Map Based on GPS Measurements It is possible to log data when driving on roads, although the GPS measurements are affected by noise, several runs on the same road should give a sufficient digital picture of the road profile. The evident advantage of this solution is that it is possible to create maps, very inexpensively, and new roads can be mapped when first used. Another possibility with this system could be on online updating of a global map. If a truck has passed a piece of road which are not mapped 29

42 4. Road Information 4.2. GIS before, it could log the data, and transmit it to an online global database. When a number of trucks have passed that road, a sufficiently accurate map will be available for other trucks Reference GIS GPS measurement Final GIS 35 Altitude [m] Northern UTM 32 x 10 6 Figure 4.12: 2D Profile of map created from GPS measurements at E47/E55 In figure 4.12 the result of this implementation is depicted. The map from section is used as reference (the green line), this profile is updated with several measurements from the GPS (the blue lines), and the final result of the update is the red line. Evaluation This map has proven to be very accurate from a visual evaluation. The map is only based on 5 measurements, which is very little compared to the number of trucks driving on highways. It is not unlikely that in a few days, hundreds of trucks would pass a certain piece of road, and sufficient data would be available for creating a good road map. In a Swedish project [Johansson, 2005] it is also concluded that the GPS receiver is the best source for creation of maps. However the system could probably be improved using other sensors like barometers, torque sensors on driving vehicles, etc. and a method called Sensor Fusion also described in [Jansson et al., 2006]. Use of statistics, Kalman filters, or the like could improve the result Laser Scan A new possibility for the development of an accurate 3D road map has been introduced in the early Airplanes have made a full laser scanning of Denmark to produce digital elevation models. The accuracy should be 15 30

43 4. Road Information 4.3. Map Matching cm. The project has not yet been converted into digital road models, but it should be possible to deduce this out of the data. The only problem of this solution could be the purchase price of maps. [COWI, 2006] Authentic Road Data from Sweden In the master s thesis [Hellström, 2005], a section of the E4 highway in Sweden is used for testing, more precisely the piece of road between Linköping and Jönköping, a total of 127 km, depicted in figure This piece of road has been placed at the disposal of the author by Erik Hellström. The road is measured with an accuracy of 1 meter and provides a very good testing route. The gradient of this piece of road varies between about +/- 4%. 300 Altitude [m] Gradient [%] Position [km] Figure 4.13: E4, Linköping - Jönköping, road profile. 4.3 Map Matching The position found by the GPS does not match the map exactly. Due to the noise on the measurements the position will fluctuate around the real road in the map. To find the actual position of the truck it is necessary 31

44 4. Road Information 4.3. Map Matching to combine the GIS and the GPS data. This is possible by using a Map matching algorithm. The problem is depicted in figure 4.14, where the raw Figure 4.14: Map matching algorithm. Raw GPS data Figure 4.15: Map matching algorithm. Matched data GPS positions fluctuate around the real road. By projecting the positions to the nearest piece of road with the map matching algorithm, the exact position of the truck on the road can be found, as depicted in figure In this project a rather simple map matching algorithm has been implemented, the algorithm is described in [Dakai Yang, 2003]. The algorithm calculates the matched position by means of simple point to line equations, references in the formulas can be seen in figure 4.16: Figure 4.16: Map matching algorithm From a point P (x 0, y 0 ), find the nearest point D(x 3, y 3 ) on a line going from A(x 1, y 1 ) to B(x 2, y 2 ) where ax + by + c = 0 (4.1) a = 1 b = 1/k c = x 1 + (1 k) y 1 (4.2) k = y 2 y 1 x 2 x 1 32

45 4. Road Information 4.4. Road Profile Based on the equation of perpendicular length from a point to a line, P D is: d = ax 0 + kby 0 + kc (4.3) b ka If PD is the smallest distance from point P to the surrounding roads, then the coordinate of point D is: x 3 = bx 0 + kby 0 + kc b ka y 3 = 1/k x 3 x 0 + y 0 (4.4) This algorithm is simple, but sufficient, if there are no crossing roads etc. For this project it is no problem, but in a later real implementation it would be necessary to include these aspects. Because these problems are already handled in the current GPS navigation systems, it was decided not to work more on this part x Northern UTM GPS measurement GIS road Matched GPS Eastern UTM 32 x 10 5 Figure 4.17: Test of the map matching algorithm In figure 4.17 the result of the map matching algorithm is depicted, the red crosses are the GPS positions, the green line is the road, and the black dots are the map matched positions. When the position gets closer to the road than a specific distance, it starts to match the measured GPS positions on to the known road. This function can be seen in the MATLAB file match map.m in appendix B. 4.4 Road Profile When the absolute position of the vehicle on the road (digital map) is known, it is possible to determine the road profile ahead of the truck. 33

46 4. Road Information 4.5. Test of GPS/GIS System The profile of the road in front can be described by a number of different parameters. The most important are: Gradient of road at present position Distance to steeper gradient Distance to end of gradient Length of gradient Figure 4.18: Description of the road sequence in front of the vehicle These parameters are used by the controller (see section 6), to read the road and determine if it is convenient to make a shift down or up. 4.5 Test of GPS/GIS System The whole system of Road Information has been implemented in a MATLAB GUIDE interface, a screenshot of the program userinterface can be seen in figure The program includes GPS-, GIS-, Mapmatching- and a simple version of Road Profile/Controller implementation. The GPS can be either on-line or off-line (already logged data), matched on the three pieces of road implemented in the GIS section. (Route 21, E20 and E47/E55). The result of the GPS, GIS and Mapmatching algorithm is depicted in the window at the top left. A simple Road Profile/Controller has been implemented by looking at the road gradient ahead. If its above a certain gradient the Max Gear is changed to the appropriate gear for that particular gradient. In the Controller section, this part will be upgraded to use more information about the road and vehicle. The program is implemented and tested to work driving in a normal car, to determine if it is possible to control the gearbox from a GPS/GIS solution. The program can be found in appendix B MATLAB file program 6.fig 4.6 Summary Communication has been established to the GPS, and data can be received and converted in order to combine it with the available GIS data. 34

47 4. Road Information 4.6. Summary Figure 4.19: Program calculating the road sequence ahead of the vehicle The task of procuring a usable 3D map has proven to be difficult. Different solutions has been tested and investigated, and the possibility for creating the map rather than buying one from a second source looks feasible. A simple map matching algorithm has been tested, and the whole section has been implemented into a program with a controller, which can give a command to the driver to shift one gear down, if the gradient of the present road section is too steep. The program has been tested in a driving car, showing expected results. 35

48

49 Chapter 5 Truck This chapter describes the control object to be controlled in this project, namely the truck. Due to the relatively short amount of time available for this thesis, it has not been possible to implement and test the system on a real truck. Instead a simulation model has been developed to match a real truck as close as possible. Another important issue about having a model is the possibility of creating a controller, based on the information contained in a model. This chapter consists of three main parts: Real Truck This is a description of the reference truck which the project is based on. This truck is chosen because it was easy to obtain simple daily running tests to achieve some references for the development of a model. 1st Model - Simple The first model to be developed can be seen as a feasibility study to investigate whether it is possible to control a truck by means of a GPS/GIS system. The main characteristics of a truck will be implemented. And the result will constitute the basis for continuing with a more advanced and more accurate model. 2nd Model - Advanced The intention for this model is to get as close as possible to the behavior of a real truck, to make it plausible that the system would be worth the effort to install in a real truck. The Truck is linked together with the rest of the project as depicted on figure 5.1. The truck receives inputs from the road, (artificial or authentic) and from the controller (will be descriped in chapter 6). The different states and output of the truck are fed back to the controller, for the next control sequence. The truck will as described in the beginning of the chapter be in the form of a simulation model. In the final solution, the system should be implemented on a real truck. 37

50 5. Truck 5.1. Real Truck - Volvo FM9-380 Figure 5.1: Truck part of the project. 5.1 Real Truck - Volvo FM9-380 As mentioned in the Introduction it was possible to get simple driving reference data from normal driving, through my father who is a truck driver. Figure 5.2: Volvo FM9-380 Strawtransport The truck is a Volvo FM9-380, depicted in figure 5.2. It is built for straw transport, 24 bales of 500 kg each Specifikations Main data of the truck: Engine D9A380, 9 Liter, Inline 6, 380 HP, 1700 Nm at RPM Gearbox I-shift VT2412B GSS-AGS, Split/Range, 12 forward, 4 reverse. Rear-axle ratio 1:3.10 Total weight kg Truck weight kg Built

51 5. Truck 5.1. Real Truck - Volvo FM9-380 Detailed engine, gearbox and rear-axle data are supplied by Volvo Truck Center Denmark A/S. The following two models are fitted to match the above truck, by adjusting average fuel/kilometer usage, air drag, rolling resistance, climbing resistance, gearbox behavior, etc Reference Tests A number of tests has been conducted in order to collect reference data for the models. Engine Tests at Volvo Engine tests have been performed at Volvo Truck Center Denmark A/S to measure properties of turbo pressure and engine response time. The truck used for engine testing is depicted in figure 5.3. Figure 5.3: Volvo FM9-300 Test truck at Volvo Truck Center Denmark A/S Unfortunately it was not possible to use a truck similar to the other truck used for driving tests. This truck is only a 300 HP version, the other is a 380HP. The main difference is the turbo pressure limit, which gives the extra 80 HP. But the response time of the engine and turbo are similar. Another test were carried out on a FM9-340HP, unfortunately all data of these tests were lost during the transfer from the truck to a PC, the only data obtained were a number of values written down manually during the test drive, including the maximum turbo pressure. 39

52 5. Truck 5.1. Real Truck - Volvo FM9-380 Maximum turbo pressure From measurements of the FM9-300 and FM9-340, the value of the maximum turbo pressure in the FM9-380 has been estimated for use in the development of models. Truck FM9-300 FM9-340 FM9-380 Turbo Pressure 183 kpa 208 kpa 230 kpa (estimate) Table 5.1: Estimated maximum turbo pressure of FM9 Response time The turbo pressure is logged by the truck in intervals of 0.6 seconds, a plot of the turbo response can be seen in figure 5.4. The graph shows that the Turbo pressure [kpa] Time [s] Figure 5.4: Volvo FM9-300 Turbo pressure. Dynamic response. turbo is at 90% pressure after 2 seconds. This agrees with promotional data, which states that the engine is capable of producing 90% power after 2 seconds. Fuel Usage Average Fuel Usage The average fuel usage has been logged during normal highway driving. Measurements were made over several kilometers, including driving in both directions on different days. Disturbances such as wind speed, atmospheric pressure, etc., are assumed to be minimized. 40

53 5. Truck 5.1. Real Truck - Volvo FM9-380 Average fuel usage Average fuel usage (highway - 85km/h) 2.9 km/liter 3.1 km/liter Max Fuel Usage Tests for determining the maximum fuel usage in the highest gears have been carried out on steep gradients. The measurements were made at 1350 RPM in the middle of the high torque area. Max Fuel, 11 gear Max Fuel, 12 gear 1.0 km/liter 1.1 km/liter Speed and Fuel Usage During Climbing On the highway E20 on Zealand, at Vårby, measurements have been carried out when the truck was climbing the gradient. The gradient has a certain inclination and length which makes the truck shift down once, when driving in Automatic mode. The measurements have been collected by looking at the gauges and digital displays in the truck when climbing the gradient. The measurement starts at the bottom at cruising speed (85km/h) and ends when the same speed has been reached on top of the gradient. Turbo pressure [%] Gear 100 Fuel [km/l] Speed [km/h] Figure 5.5: Speed and fuel tests at Vårby - E20. Automatic - Solid, Manual - Dashed The test was carried out 6 times, 3 times in Automatic mode, and 3 41

54 5. Truck 5.1. Real Truck - Volvo FM9-380 times in manual mode, where gearshifts are controlled by the driver. Due to small changes in the wind condition and load condition, the data show some uncertainty. Mode Min speed Load Wind Automatic Little following Automatic Little head Automatic None Manual None Manual Little head Manual Little following Table 5.2: Speed and fuel tests at Vårby E20. The tests clearly shows that the average speed can be increased by letting the driver control the gearbox, using experience and driver skills. Whether the overall fuel usage is changed is impossible to decide from the tests. That has to be revealed by simulations. Gearbox The gearbox has been tested to determine when the automatic decides to shift gears. The gearbox can be driven automatic in two modes, Economy or Power. Shift down is only performed when the truck is at full load and the engine speed is decreasing. The gearbox has been observed to shift down normally at: Economy mode 1050 RPM Power mode 1200 RPM The gearbox has been observed to shift up normally at: Economy mode 1600 RPM Power mode 1700 RPM Besides this, the gearbox also uses other parameters to control the gear shift. The throttle and load also affect the shifting. If the driver or Cruise Control asks for high power by pressing the throttle, the engine speed for shifting up will be raised. And similarly if the driver releases the throttle, the gearbox will shift up as soon as possible. If the load on the engine is low, the limit for shifting up will also be raised, and the gearbox then decides to shift 2 gears at a time. This is more economic. If the load is high, it will assure that the RPM is held in the high torque area of the engine. The whole strategy is depicted in figure

55 5. Truck 5.1. Real Truck - Volvo FM9-380 Shift up [RPM] Throttle [%] 0 0 Load [%] Figure 5.6: The effect of throttle and load on the gearbox strategy for shifting up. Acceleration To give an overview of the total elapsed time of acceleration including gearshifts, an acceleration test has been carried out. The truck is accelerated from 0 to 80 km/h on plain road, both empty (12 t) and fully loaded (24 t). The throttle is placed at either full normal throttle or Kick-Down which is similar to using the Power Mode described in the previous section. The Kick-Down function normally forces the gearbox to shift one gear down, if it is possible, to deliver extra power on demand of the driver. The result is presented in table 5.3. Weigth Mode Time 12 ton Full (Eco) 33.3 s 12 ton Kick-down (Power) 32.0 s 24 ton Full (Eco) 43.5 s 24 ton Kick-down (Power) 35.5 s Table 5.3: Test of acceleration 0-80 km/h Evaluation The above tests are very useful for matching of the models to the real world. Most of the results are made by visual reading of instruments, which are not very accurate. But due to the limited amount of time available, it was not possible to make all these tests in a special test truck equipped with computers for logging the data. 43

56 5. Truck st Model - Simple All of the reference data of the truck has been incorporated in two MAT- LAB files for the following two models: truck data.m for the 1st model, and truck data new.m for the 2nd model, both to be found in appendix B st Model - Simple A simple model of a truck has been developed and implemented, to get an overview of the system, and test whether it is possible to change the behavior of a truck using inputs from a GPS/GIS system. The model includes vehicle dynamics, road conditions, and drivetrain. The model has been implemented in MATLAB and SIMULINK, see appendix A.1. The model has been implemented using very simple equations and physical properties of the different parts. Using the test data from the previous chapter, the model has been adjusted to match the output of a real truck. The model consists of 7 main parts, linked together as depicted in figure 5.7. Figure 5.7: 1st Model main parts. Engine Simple engine, including most important time constants. Automatic Gearbox Simple automatic gearbox including a timedelay. Rearaxle/wheels Ratios and conversions Vehicle Physical model of vehicle Real Road Road profile, authentic or artificial. Cruise control Cruise control to maintain a cruising speed. Output/Gauges Output data for evaluation. 44

57 5. Truck st Model - Simple Engine The engine has been modeled by a few simple parts. A total torque curve, a power time constant, a turbo with another time constant, a clutch, and a calculation of fuel consumption. The output torque of the engine can be calculated by the formula: T e = Cl (T norm (n) + T turbo (n) T V EB (n) T drag (n)) (5.1) Where T e is the total engine torque, Cl is the clutch which can have the values 0 or 1 controlled by the automatic gearbox, described later. T norm is the torque delivered by the normal engine, T turbo is the extra torque the turbo delivers in its working area and T V EB is the negative torque provided by the engine brake if the speed is above the reference. T drag is the negative torque provided by the internal friction in the engine and n is the engine speed. Normal Engine Torque The torque from the engine is calculated by the formula: T norm = T normmax (n) 1 τ engine s + 1 K throttle (5.2) which is simplified to a 1st order system, where T normmax is the maximal torque to be delivered from the normal engine, found from a reference-table via the engine RPM. τ engine is a time constant for the engine to built up power and K throttle is the position of the throttle, can vary from 0 to 1. Turbo The extra torque from the turbo is calculated by the formula: T turbo = T turbomax 1 τ turbo s + 1 K engine (5.3) which is also simplified to a 1st order system, where T turbomax is the maximal torque to be delivered from the turbo, found from a reference-table via the engine speed. τ turbo is a time constant for the turbo to built up power and K engine is the 1st order delayed engine load from above. Engine Drag When the vehicle is pulling the engine, going down steep gradients, the engine will produce a small negative torque, given by the equation below. T drag = a d n + b d (5.4) 45

58 5. Truck st Model - Simple Normal engine torque Extra turbo torque Total engine torque Drag Torque Torque [Nm] RPM Figure 5.8: Engine torque of a Volvo FM9, 380HP The engine torque is depicted in figure 5.8. It should be mentioned that the total torque is from the Volvo description of the engine, the two other curves are created by a simple qualitative guess and may differ from the real facts. But the overall result of this system is an engine which has a working area with some time delays, including the time constants in the turbo. The low area engine speed differs a lot from the real world, but is made in this way to simplify the whole clutch mechanism, i.e. the truck is able to start from 0 km/h with 0 RPM in 1st gear. VEB An Engine brake (VEB - Volvo Engine Brake) has been implemented to brake the vehicle if the speed increases above the desired reference speed. n 1 T V EB = k step T V EBmax n max τ V EB s + 1 (5.5) Which is a simplified 1st order system, where k step is a constant k step [0; 3] referring to 0 brake or step 1-3, 3 is most braking torque. T V EBmax is the maximal breaking torque to be provided by the engine brake, which equals 80% of the normal driving torque on this engine at max engine speed, depicted in figure 5.9 The engine is lagged by the τ V EB time constant, which equals the engine time constant. 46

59 5. Truck st Model - Simple Brake Torque [Nm] VEB step 1 VEB step 2 VEB step RPM Figure 5.9: Engine brake torque of a Volvo FM9, 380HP Fuel Consumption The fuel consumption is calculated in a rather simple way [Hellström, 2005]. fc = n K throttle n cyl n r (5.6) Where fc is the fuel consumption in liter/sec, n cyl is the number of cylinders and n r is the number of crankshaft revolutions per stroke. The SIMULINK engine block can be seen in appendix A Automatic Gearbox The gearbox is mainly a system to multiply the torque delivered from the engine. The gears are selected by a simple algorithm which shifts the gear up, when the engine speed reaches 1650 RPM and shifts down when it drops below 1050 RPM. This is a simple algorithm but sufficient for the first model. A gear efficiency has been included in the model, and the efficiency is as depicted in figure 5.10 [Hellström, 2005]. The efficiency is highest in the top gear. The gearbox is also equipped with a time delay to simulate the time it takes to shift gears, meanwhile a command is sent to the engine to set the output torque to 0 (activated clutch). The gearbox controller can also be given a command to force a shift up- /down from the controller described in chapter 6. The SIMULINK gearbox 47

60 5. Truck st Model - Simple 15 Ratio [x:1] Efficiency [%] Gear Figure 5.10: Gear ratios of an I-shift VT2412B block can be seen in appendix A Rearaxle/Wheels The rearaxle and wheels are a conversion from the engine/gearbox torque to a force acting on the road. The rearaxle includes a gear ratio and an efficiency constant. The model includes wheel brakes to make it possible to keep the speed down when driving downwards on steep slopes. Otherwise the overspeed gained could be use to climb the next hill and influence the result. The total equation of the rearaxle and wheels: ω w = T gi r η r k b B r w F w (5.7) J w Where ω w is the angular acceleration of the wheels, T g is the torque delivered from the gearbox, k b is a constant parameter, B is the brake control signal B [0, 1]. r w is the wheel radius, F w is the total force acting on the wheels and J w is the wheel inertia. i r is the rearaxle ratio and η r is the rearaxle efficiency. The SIMULINK wheel block can be seen in appendix A Vehicle A simple total longitudinal acceleration vehicle model has been developed, inspired from the article [Mangan et al., 2003] and [Kiencke & Nielsen, 2000]. The model is derived from the two general equations: F = m a (5.8) 48

61 5. Truck st Model - Simple and T = J ω (5.9) Where m is the mass of the vehicle, F is the total driving force, a is the resulting acceleration of the vehicle, T is the Torque, J is the moment of inertia and ω is the angular acceleration. The total driving force is a combination of the force from the driveline and the opposing forces acting against movement of the vehicle. F = F dl F rr (5.10) Running Resistance The running resistance consist of 3 forces: F rr = F ro + F ae + F cl (5.11) Where F ro is the rolling resistance, F ae is the aerodynamic drag and F cl is the climbing resistance and downgrade force. Rolling Resistance The rolling resistance is a product of the deformation of the tire on the contact patch to the road surface. The total resistance can be calculated using this equation: F Ro = f m g cosα (5.12) Where f is the rolling resistance coefficient, here is as reference used [Hellström, 2005] and later adjusted to match the real vehicle. g is gravity (9.81m/s 2 ), m is the mass of the vehicle and α is the longitudinal road gradient. Aerodynamic Resistance The aerodynamic resistance is calculated using the equation: F ae = 0.5 σ c w A v 2 (5.13) Where σ is the air density, 1.19kg/m 3 at 100m altitude and 20 C. c w is the vehicle drag coefficient. This is different for every vehicle but here the value 0.6 is used. [Hellström, 2005]. A is the front area of the vehicle which for the Volvo FM9 truck is 9.75m 2. 49

62 5. Truck st Model - Simple Climbing Resistance The climbing resistance which is by far the largest force acting on the vehicle can be calculated from: F cl = m g sinα (5.14) Driveline Force The force acting from the driveline can be calculated from the expression: F dl = T e i g i r η g η r r w (5.15) Where T e is the torque delivered by the engine, i g is the gear ratio of the gearbox. η g is the gearbox efficiency. Complete Vehicle Model The complete vehicle model including forces, mass and moments of inertia can now be stated: r w v = J w + mrw 2 + η r i 2 rη g i 2 (η r i r η g i g T e gj e ) k b B 0.5c w Aσr w v 2 mgr w (fcosα + sinα) (5.16) Input - Real Road The model is simulated driving on authentic or artificial road sections. During simulation the acceleration of the vehicle is integrated to give the speed and again to give the driven distance. This distance is used to look at the maps of a certain road and find the gradient of the road, thus it is possible to drive on the real roads with the model. Besides several constants the model is also fed with a reference speed for the Cruise Control Cruise Control The Cruise Control is implemented to make the vehicle maintain a certain speed. The cruise control is implemented as two simple PID 1 controllers, one for the throttle and one for the brakes. It has not been possible to obtain the real values for these controllers, instead the controllers are tuned to give a fast and damped response. Both of the controllers are implemented in this form: ( G c (s) = K p ) τd s + 1 (5.17) τ i s α cc τ d s + 1 This is actually not a PID controller but a PI-lead controller. 1 Proportional-Integral-Derivative controller 50

63 5. Truck st Model - Simple Throttle Controller The input to the controller is the deviation of the speed compared to the reference speed and the output is the throttle position to the engine. The parameters are: K p = 10, τ i = 10, τ d = 3, α = The controller has also been equipped with an anti-integrator-windup since the control signal (throttle) is limited. Brake Controller The input to the controller is the deviation of the speed compared to the reference brake speed (a few km/h higher than the reference cruising speed), and the output is the pedal position to the brakes. The parameters are: K p = 0.2, τ i = 10, τ d = 3, α = The controller has also been equipped with an anti-integrator-windup since the control signal (brake) is limited Output - Data/Gauges The output of the model is presented in two different ways: online and offline. During the simulation, the states of the model are presented on a number of gauges depicted in figure 5.11, which gives a clear overview of the simulation. After the simulation is carried out the result is saved in a number of files for offline inspection. Figure 5.11: Gauges from the simulation model, similar to the gauges in a Volvo Truck. 51

64 5. Truck nd Model - Advanced Evaluation This model is a rather simple implementation and is only tested with a simple controller (many details are left out). However the model gives an overall idea of the performance using a system to read the road in front of the vehicle. The most important problem with this model is the simulation of fuel consumption. This is a very rough and inaccurate function and is the primary reason for constructing a more advanced and realistic model. The SIMULINK diagram of the whole 1st model can be seen in appendix A nd Model - Advanced The 1st model developed in the previous section is based upon simple basic knowledge of the driveline of a truck. The engine is based on a certain torque, and a number of time constants to reflect reality. This model is good for simple test of the system and to give an overview of which parameters a controller for the gear shifting strategy affects. No changes are needed for improving the equations describing the vehicle itself. The input and output to the model are also the same as in the first model. Three parts of the model, concerning the driveline are improved in the 2nd model: Engine A more accurate engine model is needed where the correct time constants and small secondary effects can bring the result closer to reality. One of the most important parameters to improve is the fuel consumption. In the first model, the fuel consumption is a calculated value based on engine speed and throttle position. The more correct way of simulating the fuel consumption is to base the model on energy considerations. Clutch In the first model the truck can be started from 0 RPM, that is not possible in real life. In the first model this also means that the engine model is adapted to have a working area from 0 RPM to 2500 RPM. This needs to be improved since a real engine can normally not work below approximately. 600 RPM. Therefore a real clutch is implemented in this model to make it possible to start the model from 0 km/h. Gearbox The gearbox in the first model was a simple ratio conversion with a time delay of a fixed time to shift gears. In the second model the whole shifting sequence is incorporated to give a more accurate response. The gearbox normally controls the engine to make the engine speed correct for engaging the next gears, this takes some time and 52

65 5. Truck nd Model - Advanced consumes a small amount of fuel. second model. These parts are included in this Engine The new engine model is based on the MVEM (Mean Value Engine Model) which is a dynamic description of the states in the engine by means of a set of nonlinear differential equations. The model uses mean values instead of instantaneous values, and is simplified so that it is just sufficient to describe the time development of the most important engine variables. In spite of the simplification the model is capable of describing the conditions in an engine with a very high accuracy, and similar models are used by both Delphi Automotive and Bosch for control of air/fuel mixture and throttle/power. In this project the model is simplified and adapted to fit a large diesel engine with the main emphasis on fuel consumption, time constants and torque produced. Introduction The MVEM modeling of turbocharged engines, requires two main dynamic systems. The crank shaft dynamics and the air supply dynamics. In a diesel engine the fuel supply to the cylinders is controlled according to the available air supply in the intake manifold. The purpose of the model in this thesis is to model a heavy duty turbocharged diesel engine during medium to high load operation, normally used in highway operation. This simplifies the task of modeling significantly, specifically the dynamic air supply is easier to model during medium to high engine operation. The need for modeling of the air supply dynamics relies on the fact that the output in a turbocharged engine is strongly affected by the air flow dynamics. The fact that a turbocharger has a finite reaction time, makes the modeling of air dynamic very important. The turbocharger is a compressor driven by an exhaust turbine on operation range of more than RPM. Acceleration and deceleration of the turbocharger shaft takes a lot of time and is thus not negligible. Thus this must be taken into consideration in the investigation of optimum vehicle operation in hilly locations. The effect is especially visible during gearshifts where the engine is moved from high torque to zero torque and back again. Here the turbocharger time constant is the most important variable to affect the period of time when the engine is not delivering full torque to the driveshaft. Fortunately the model required in this project can be based on a foundation which already exists in the form of MVEMs already constructed and which are available in the literature. The model is primarily based 53

66 5. Truck nd Model - Advanced on earlier work done on the modeling of very large ship diesel engines, described in [Hendricks et al., 1984] and more detail in [Hendricks, 1989] as well as later work on turbocharged common rail diesel engine models for cars [Hendricks et al., 2004]. Two other important references has been used: the detailed compressor model from the SAE paper [Hendricks et al., 2005] and the port air mass flow model from [Hendricks et al., 1997]. Further assistance with the model has been provided from local experts at DTU, [Hendricks & Sorenson, 2006]. In order to construct this simplified diesel engine model, bits and pieces from the references above have been combined as appropriate for this special task. The model has been adapted and checked against available data of the real engine. None of the data on the engine has been supplied by the manufacturer, thus the required MVEM has been constructed only on the basis of speed and torque curves provided in promotional material, simple measurements carried out on a daily running truck, and measurements on a truck with the VCADS 2 at Volvo Truck Center Denmark A/S. Figure 5.12: Primary parts of the engine The functionality of the engine is depicted in figure 5.12, a more detailed flowchart can be seen in appendix A.2.2 The fuel is fed into the cylinders, together with an amount of air. This gives a certain efficiency which gives the amount of torque delivered on the crank shaft. Some power are also let out through the exhaust which drives the turbocharger. The turbocharger provides the engine with a higher pressure when more fuel is fed into the engine, this means more air/fuel into the crankshaft, which means more torque on the crankshaft. Crank Shaft Speed Dynamics In [Hendricks et al., 1997] the crank shaft speed equation is written: ṅ = 1 In ( (P f + P b ) + H u η i m f (t τ d )) (5.18) 2 Vehicle Computer Aided Diagnostic System 54

67 5. Truck nd Model - Advanced where n is the crankshaft speed, I is the moment of inertia of the engine, plus eventually load. P f (n) is the loss of friction in kw and the power supplied to the load is P b (n) in kw and m f is the fuel mass flow kg/s. The frictional losses are obtained from a normally used expression described in [Heywood, 1988] which can be scaled with the engine displacement. The expression in [Heywood, 1988] is based on a liters engine, thus the frictional losses can be determined by scaling: P f (n) = V ( d n n 2) (5.19) V ref where V ref = 1.275L is the reference engine, V d is the displacement of the large engine (in this case 9.4 liters). The parameters in equation 5.19 have been obtained from measurements at MEK 3, DTU. The last term in the right parenthesis in equation 5.18 includes the heating value of the fuel, (i.e. the power included in the fuel) H u = 43kJ/kg. The indicated efficiency η i in the same place is also obtained from experiments conducted at MEK, DTU. This is a typical value of the efficiency of a modern diesel engine. It is given by the equation: η i (Φ, n) ( = (Φ 0.45) 2) (5.20) This approximation is possible because the indicated efficiency is only weakly dependent on the engine speed in the medium/high load region. Φ is the fuel/air equivalence ratio, given by the expression: Φ = m f L th m ap (5.21) where m f is the fuel mass flow, m ap is the port air mass flow, and L th = 14.7 is the stoichiometric air/fuel ratio. Equation 5.20 is a very flat parabola in θ. The rate at which air can be supplied to the four cycle engine is given by the port air mass flow, which is given by the speed density equation: m ap = V d 120RT i (e v p i ) n (5.22) where V d is the engine displacement volume, R is the gas constant, T i is the temperature of the air in the intake manifold. (e v p i ) is the volumetric efficiency times the intake manifold air pressure (i.e. the air charge per stroke) described in [Hendricks et al., 1997]. The aircharge per stroke derived in [Hendricks et al., 1997] is given by a simple universal equation: 3 Department of Mechanical Engineering e v p i = s i (n) p i + y i (n) (5.23) 55

68 5. Truck nd Model - Advanced where s i (n) and y i (n) are only very weak functions of the engine speed and are effectively constant. Typical values found for these parameters of a diesel engine are s i = 1.05 and y i = The last parameter in equation 5.18 is the time delay τ d from the injection of fuel until the torque increase on the crank shaft, is given by the equation: τ d = 60 n ( ) n cyl (5.24) where n cyl is the number of cylinders. The SIMULINK engine crankshaft block can be seen in appendix A.2.5. Intercooler The intake manifold pressure and temperature used above are determined by the intercooler, which most turbocharged engines are equipped with. The purpose of the intercooler is to lower the temperature of the intake air, and thereby increase the air density. The intercooler is cooling the intake air by the ambient air when driving. This is necessary because when the turbocharger is increasing the intake pressure the air is also heated up. In principle the intercooler is a dynamic system, like the crank shaft speed state equation and the compressor pressure. But since the engine is assumed to operate near maximum load, and for the sake of simplicity the intercooler can be assumed to operate instantaneously and an algebraic equation can be used to describe it instead. The intake manifold pressure and temperature are given by: p i = 0.916p c (5.25) and T i = 0.229T c (5.26) where p c and T c are the compressor outlet pressure and temperature. These equation are derived by measuring on a real turbocharged engine, described in [Hendricks et al., 2004]. The measurements was carried out in a lab and an air to water intercooler was used instead, but the results are similar to an air to air intercooler used in trucks. The intercooler was found to give a constant efficiency of 70% over the entire operating range of the engine. The purpose of including the intercooler as a static model is to give the model the required functionality. In principle a dynamic model could be implemented if required. A dynamic model of an air to air intercooler has been derived in [Müller, 1997] and [Müller et al., 1998]. 56

69 5. Truck nd Model - Advanced Turbocharger State Equation The turbocharger is driven by the energy in the hot exhaust gasses from the cylinders. In [Hendricks et al., 1984] and [Hendricks, 1989] it has been shown that the outlet pressure can be modeled by a first order system. The exhaust power is a function of the amount of fuel fed to the engine. If it is assumed, that the power available to drive the turbine is P exhx (kw ) then the compressor outlet pressure can be found by: p c = 1 ( p c + f (P exhx, n)) (5.27) τ c where f (P exhx, n) = c 1 (n) P exhx + c 2 ( ) = n n P exhx 0.2 (5.28) where the constants c 1 and c 2 have been adjusted according to the compressor characteristic map in [Hendricks et al., 1984]. A compressor characteristic map is a plot of the compressor pressure ratio against the temperature adjusted compressor mass flow (here m ap ). The time constant τ c is found by: τ c = p 1 κ c m ap (5.29) where κ = c p /c v = 1.4 (The adiabatic constant) for air. The time constant should be in the order of 1-3 sec at maximum mass flow [Hendricks, 1989]. In equation 5.28 the adjustment of constants and functions is carried out by looking at the maximum volumes and pressures to be handled by the compressor for at certain engine. The required air mass flow are judged from the engine displacement and the equivalence ratios which must obtain at the highest engine output. The linear load lines in the compressor map in figure 2 in [Hendricks et al., 2005] has been found to work generally at medium to maximum engine loads, related to the engine speed. The parabolic dependence of the slope on the engine speed has also been found to be generally true. The power to drive the turbo, P exhx, must be a proportion of that used to drive the engine. An effective efficiency can be used to set this proportion. This efficiency should be a factor in the order of 1 η i 2 = = 26.5% of the fuel used to feed the engine. If m f is the fuel flow used to drive the engine, then the power can be expressed as: P exhx = H u m f (5.30) 57

70 5. Truck nd Model - Advanced where the factor gives a scaling factor from exhaust power to the mass flow (linear) equation, here for the target engine of 9.4 liters. During this project it has proven that it is possible to adjust the constants in equation 5.28 and equation 5.30 to make the model capable of describing the target engine in the entire upper operating range. The compressor in this project is assumed to be working at the top of its operating range, and can therefore be assumed to operate at a more or less constant efficiency. A reasonable guess for its outlet temperature can thus be given assuming an efficiency of about 73%. The outlet temperature is then described by ( T c = T c (η c, p rc, T a ) = T a 1 + Y ) c η c = T a + T a η c ) (p κ 1 κ rc 1 (5.31) where η c = 0.73, p rc = p c p a and p a,t a are respectively the ambient pressure and temperature. The SIMULINK engine turbo block can be seen in appendix A.2.4. Adjustment of Model - Static The entire engine model has been implemented in SIMULINK, see appendix A.2.2. To Benchmark and adjust the model, it has been loaded with an infinitely large moment of inertia without any torque load. In this way it is possible to measure the full torque of the engine without the engine speed increasing. Thereby a torque curve can be made. The engine parameters can then be adjusted to fit the torque curve of the real engine. The adjustment of the engine includes several steps: Fuel function The output of the engine is controlled by the amount of fuel fed to the engine. From the promotional engine data at the Volvo website, the power curve can be deduced and the required amount of fuel calculated. Turbo pressure When the fuelling function is correct, the turbo pressure can be adjusted to match the real engine data. Efficiency When the turbo pressure is correct, the efficiency of the engine can be adjusted to produce the correct torque and power. Fuel Function The fuel function is calculated to produce the right amount of power. By combining equation 5.18 and equation 5.19 and inserting points from the 58

71 5. Truck nd Model - Advanced Fuel Mass flow [kg/s] Maximum Fuel Minimum Fuel (Idle) RPM Figure 5.13: Fuel Function power curve, a number of fuel values for different engine speed can be calculated. With the MATLAB curvefit function a fuel function can be found: m f = n n (5.32) This function works from around 800 RPM and up, below this a linear function has been fitted to supply the engine with the correct amount of fuel. The minimum fuel fed to the engine is the idle fuel which means the necessary amount of fuel to overcome the internal friction in the engine at idle speed. Now the engine is actually adjusted to match the target engine in regards to power and torque. Now the turbo pressure has to be adjusted to match the target engine to make sure the turbo has the right influence on the engine. Afterwards the efficiency of the engine might have to be adjusted slightly to make the power and torque fit again. The SIMULINK engine fuel block can be seen in appendix A.2.3. Turbo pressure At Volvo Truck Center Denmark A/S, a test drive was conducted in order to measure the maximum turbo pressure. The truck was equipped with the VCADS software for monitoring the important sensors in the truck. The maximum turbo pressure was measured, peaking at p cmax = 230kP a. The measurements can be seen in figure 5.14 as the red curve. The blue line shows the simulated values after adjustment of the constants in equation When the turbo pressure is confirmed to be correct the actual efficiency of the engine can be found. It will probably deviate a bit compared to the 59

72 5. Truck nd Model - Advanced Charge Pressure [kpa] Simulated Max, Volvo D9A RPM Figure 5.14: Maximum turbo pressure value used in the fueling function. After a few iterative adjustments it is Power [kw] Volvo D9A380 Simulated RPM Figure 5.15: Engine power Torque [Nm] Volvo D9A380 Simulated RPM Figure 5.16: Engine torque possible to make all the values fit. And the simulation compared to real values can be seen in figure 5.15 and figure The result of the simulations also shows the real efficiency of the engine depicted in figure 5.17, which shows that the overall efficiency is around 45%. The fuel/air equivalence ration can be viewed in figure 5.18 which shows an overall value of around Φ [0.6; 0.7] in the working area which is normal of diesel engines at high loads. Normally Φ [0.1; 0.7]. Adjustment of Model - Dynamic The model also needs some adjustment to fit the dynamic response of the real engine. The real engine is capable of delivering 90% of the requested 60

73 5. Truck nd Model - Advanced Efficiency (Eta i ) RPM Figure 5.17: Engine efficiency η i Equivalence ratio (Theta) RPM Figure 5.18: Equivalence ratio θ torque after 2 sec, as seen in figure 5.4. This can be achieved by adjusting the time constant in the turbocharger equation Another parameter to take into account is the fuel injection. The fuel injection can not be changed momentarily since it will cause the equivalence ratio Φ to move outside the allowable boundaries and make the efficiency and torque drop rapidly, because the air supply can not change rapidly see figure The fuel has to be lagged to follow the lag in the air supply. A time constant related to the time constant of the turbo has therefore been added to the fuel input. The results of the dynamic adjustment can be seen in figure 5.19 and figure 5.20 as the red curves, compared to the blue curves without the timelag. Since the efficiency of the engine is kept as even as possible during changes in the operating point, the rise time of the torque produced is not affected significantly compared to the model without the fuel lag. The result can be seen in figure

74 5. Truck nd Model - Advanced Fuel Mass flow [kg/s] Turbo pressure [kpa] Torque [Nm] Without timedelay With timedelay Time [s] Figure 5.19: Engine dynamic response - Torque. It is seen that the response with the fuel time delay is not remarkable slower than the response without the time delay. 0.5 Efficiency [eta i ] Equivalence ration [theta] Without timedelay With timedelay Figure 5.20: Engine dynamic response - Efficiency. The efficiency is kept high by using a time delay in the fuel injection. 62

75 5. Truck nd Model - Advanced Clutch A clutch has been implemented in the model to give the possibility of starting the truck from 0 speed. A clutch is a rather simple device to model, in practical it consists of three different modes: Disengaged When the clutch is fully disengaged, no torque is transmitted through the clutch. Slipping Going from disengaged to engaged, or opposite can be defined as slipping where the torque is determined primary by the dynamic friction in the disks in the clutch. Engaged When the clutch is engaged, the full torque is transmitted through the clutch. Dynamic Model When the clutch is disengaged the torque transmitted via the clutch T cl is zero: T cl = 0 (5.33) When the clutch is slipping the torque transmitted via the clutch is[glielmo & Vasca, 2000]: T cl = kf n sign(n n g ) (5.34) where k = 4R cl µ d /3 (5.35) R cl = 0.3[m] is the equivalent disk radius, µ d = 0.4 [Glielmo & Vasca, 2000] is the dynamic friction coefficient. F n [N] is the force on the clutch. n and n g are the rotational speed of the crankshaft and the input axle in the gearbox. When T cl becomes higher than the load torque T l the clutch is engaged, and the torque transferred is equal to the torque delivered to the clutch from the engine. T cl = T e (5.36) and n e = n g (5.37) The clutch has been implemented in the SIMULINK model which can be seen in appendix A

76 5. Truck nd Model - Advanced Gearbox The gearbox is primarily a number of different gear ratios to convert the torque from the engine to the rear axle. The gearbox implemented includes three special functions, to imitate the real gearbox: Start-up gear The gearbox calculates which gear it is possible to use as a start up gear, by means of weight of the vehicle and gradient of the road. Decision of gear shift The gearbox follows a number of rules about when to shift gears. Shifting sequence The shifting sequence is implemented as the real gearbox, i.e. the engine is used to control the speed of the gears to make them synchronize when shifting. Start-up Gear The startup gear is selected automatically. It can vary from 1-6 dependent on the load in the initial start up moment. Based upon the torque the engine is capable of producing at 800 RPM, the highest possible start up gear is selected compared to the torque from the weight of the vehicle and gradient of the road. Decision of Gear Shifts As described in section the gearbox has a number of different rules for which affects the gearshift. These are affected by the throttle position and the load on the engine. These rules are all implemented in the SIMULINK model, to use the shifting strategy depicted in figure 5.6. The Economy/Power mode can be selected to affect the shifting strategy. Shifting Sequence One of the most important extensions of the gearbox compared to the first model is the sequence of shifting a gear. This includes a number of states as depicted in figure 5.21: The shifting sequence includes 8 steps, all implemented in the SIMULINK model, which can be found in appendix A.2.7, and in the MATLAB file gear shift function.m which can be found in appendix B. Command to shift gears A command from the computer to shift gears is received. Request zero torque A request for zero torque in the gearbox is parsed on to the throttle control, i.e. cut fuel to idle level. 64

77 5. Truck nd Model - Advanced Figure 5.21: Gearbox shifting sequence Wait Wait for the torque to drop to zero. Shift to Neutral Disengage the old gear, go to neutral. Change RPM for new gear ratio If the new gear is higher than the old gear, the engine needs to decrease RPM. This is performed by cutting the throttle and engaging the engine brake to make the RPM decrease rapidly. If the new gear is lower than the old gear, the engine needs to increase RPM. This is accomplished by increasing the throttle. Wait Wait for the correct engine speed. Engage new gear When RPM is correct, engage the new gear. Put on torque again The shift is completed and torque transfer can be reestablished. The whole sequence is completed within a few seconds, but the torque to drive the vehicle is zero during the shift, which causes the vehicle to loose speed especially if it is carried out on a steep gradient. The sequence is depicted in figure 5.22 and figure 5.23, respectively shift down and shift up Evaluation Compared to the first model, the 2nd model is more alike the real driveline in a truck. Specially the fuel usage is now very accurate, because the whole driveline is based on the input of fuel, compared to the first model where the fuel was calculated separately. Delays from fuel input and turbo has now been modeled in more details, and on the basis on more tests. One of the forces about the 2nd model is the ability to be adapted to different engines. In this case it has been adapted to a 9.4L 380HP engine, however it is easy to adjust the model to fit another diesel engine, only the fuel function and turbo pressure has to be adjusted to match the new engine. The whole SIMULINK diagram of the 2nd model can be seen in appendix A

78 5. Truck nd Model - Advanced Torque [Nm] RPM Throttle 1000 Gear VEB Neutral Engine Gearbox Time [s] Figure 5.22: Gear shift sequence - shift down Torque [Nm] RPM Throttle 1000 Gear Engine Gearbox Time [s] VEB Neutral Figure 5.23: Gear shift sequence - shift up 66

79 5. Truck 5.4. Tests 5.4 Tests A number of tests have been carried out in order to evaluate the two complete developed models against the available real data. The tests includes: Fuel consumption Average fuel consumption test during highway driving. Climbing capacity The models climbing capacity on known gradients compared to the reference truck. Acceleration Acceleration time of the truck including all gearshifts Fuel Usage The fuel tests has been carried out at the available GIS data of Route 21 from Holbæk to Copenhagen. Parameters like aerodynamic drag and rolling resistance have been tuned to match the real fuel consumption. Reference truck 1st model 2nd model 3.1 km/l 3.1 km/l 3.1 km/l Table 5.4: Average fuel tests, Route tons, 85 km/h Climbing Capacity Both models are tested on the gradient near Vårby at the E20. The minimum speed on the gradient has been compared to the real truck. Reference truck 1st model 2nd model 65 km/h 63.5 km/l 64.8 km/h Table 5.5: Minimum speed test, E tons, 85 km/h Acceleration Both models are tested in relation to acceleration from 0 km/h to 80 km/h, to compare the time used for gearshifts and acceleration on the real truck. It should be noticed that the 1st model is not capable of shifting more than 1 gears at a time, thus it takes more time for the 1st model to accelerate. It should also be taken into account, that the reference test includes some disturbance due to the road where the real truck was tested. 67

80 5. Truck 5.5. Summary Weight/Mode Reference truck 1st model 2nd model 12 t/eco 33.3 s 44.0 s 34.0 s 12 t/power 32.0 s 41.5 s 31.7 s 24 t/eco 43.5 s 56.5 s 42.0 s 24 t/power 35.5 s 52.0 s 36.0 s Table 5.6: Acceleration test, 0 km/h - 80 km/h 5.5 Summary Three real trucks has been examined and tested in regards of several parameters. On the basis of these tests and promotional data, two models has been developed. The models are adapted to respond as the real trucks, specially in regards to torque, speed and fuel consumption. The first developed model was based on the need of a feasibility study, and proved to be useful in tests of a simple controller. The second controller is an improved version of the first model, specially in regards to fuel consumption, behavior of engine and gearbox complexity. 68

81 Chapter 6 Controller This chapter is a description of the control strategy developed in this project, and of the implemented and tested controllers. First of all, this part is linked together with the rest of the project as depicted in figure 6.1. The controller includes reference data, two main inputs and one output. Figure 6.1: Controller part of the project. Reference data The controller uses a number of reference data to adjust the shifting strategy, that includes a number of data of the truck, engine, weight, aerodynamic properties, etc. and some inputs from the driver whether the gearbox should optimize for power or economy. Inputs The first main inputs are the current state of the truck, i.e. the speed, engine speed, gear, etc. The other input is the road profile from the GPS/GIS system which the controller uses for predicting if a gear shift is advantageous. Output The output are the commands to be sent to the gearbox, (or the computer controlling the gearbox) if the gear shift strategy should be changed. 69

82 6. Controller 6.1. Reference Data The different controllers implemented in this project includes more or less of the above parameters. 6.1 Reference Data The reference data are the inputs that the controller uses and which do not changing normally during driving. This includes parameters such as the truck characteristics and reference values. These data are used to give a reference for the controller of when the gradient is too steep to climb using the present gear, or if the truck can handle the gradient only by its momentum. Other reference data are the inputs from the user like the desired reference speed, shifting strategy and Economy/Power mode Truck Data The most important truck parameters are: Weight The weight of the truck influences the rolling resistance negatively, but also contributes to a higher momentum. Engine power/torque Torque the engine can produce i.e. which gradient the truck can climb without making gearshifts. Time constants How long does a gearshift takes, how much speed will the vehicle loss during a shift. Gear ratios When is it most efficient/economic to shift from one gear ratio to another, compared to the profile of the torque curve. Aerodynamic The aerodynamic parameters of the truck affects the necessary amount of torque to overcome the aerodynamic resistance Driver Inputs Shifting strategy The driver is in some way given the possibility of deciding which program the controller should use, if it should shift most economical or most powerful, or somewhere in between. Reference speed The reference speed is also affecting the strategy since more speed entail more aerodynamic resistance, but also more momentum. 70

83 6. Controller 6.2. Inputs 6.2 Inputs Truck States The inputs to the controller are provided from the truck itself, that means data like the present speed, the present RPM, etc. Speed The speed should be kept at the reference speed, but when it goes below, the controller has to be affected, since the necessity for looking far ahead on the road is changed, if the truck is climbing a hill with 40 km/h instead of the reference of 85 km/h. Engine speed The current speed of the engine is important, to let the controller decides if it is advantageous to shift gear to produce more torque. Gear The current Gear is also important, since this affect the hill climbing capacity of the truck Road Sequence The road ahead is fed to the controller in the form of position and gradient data. These data are combined with the other input data, and the different rules in the controller decides when to change strategy. 6.3 Outputs The decisions made by the controller can be transmitted to the truck in two ways, either as direct orders or as an adjustment of present strategies. Direct orders Direct orders for a particular gear can be transmitted to the gearbox when the controller decides from the map, that it would be advantageous. Max Gear Another solution is to change the maximum allowed gear to use. In some way it will give the same result as the above solution, but the advantage is that if the truck is not running at top speed, i.e. because of weather conditions, the flow on the road or alike, the system will not affect the gearbox. Change limits A solution to move the limits of when to shift gear has also been implemented. This solution has the advantage that it is working like parsing a suggestion to the gearbox and not a direct order. This gives a more noise resistant system because if the truck is not reacting exactly like the controller has predicted, it will still shift when it reaches the new limit of shifting. No matter if its a little 71

84 6. Controller 6.4. Controllers earlier or later. It can also be seen as a safer way of using the gearbox because the gear shift can never be ordered on too high RPM or alike. The gearbox will always wait to shift until it reaches the new desired RPM limit which are also defined in between the safe limits. 6.4 Controllers Three controllers have been implemented and tested in this project. They are all compared to the reference system, which is the normal gearbox control. The three controllers are: Simpel - Economy/Power The first controller is very simple. It makes use of the difference in the Economy/Power mode by switching to Power mode, when the gradients of the road ahead increases above a certain limit. Rule based for 1st model A very simple rule based controller is implemented for the 1st simple simulation model. When the gradient of the road rises above a certain limit it changes the maximum allowed gear. Rule based for 2nd model A more advanced rule based controller has been developed for the 2nd model. This controller uses a number of inputs to decided which shifting strategy should be used. This controller can also be adjusted by the driver to give more or less power or better economy Normal - Reference The reference performance for all the controllers, is the standard system implemented in the I-shift gearbox. This system is described in detail in section Basically this system uses some predefined limits for the RPM of the engine, when these limits are reached either up or down a gearshift is carried out. However the limits are not permanent, but are affected by the throttle level and load on the truck. This function can be seen in figure 5.6. The new controllers are evaluated up against this reference, however it should be taken into account that this reference is not 100% correct since it has not been possible to acquire the real data of the gearbox due to patent security. The gearbox has been based of the result of different tests and promotional data of the real truck and gearbox Simple - Economy/Power The first simple controller to test is the use of the two different standard modes, Economy or Power, normally selected by the driver on a small button behind the gear lever. As described in section the normal engine speed 72

85 6. Controller 6.4. Controllers where the gearbox shifts down is increased from 1050 to 1200 RPM, likewise the limit for shifting up is increased from 1600 to 1700 RPM Rule Based for 1st Model The first simple rule based controller is developed for a feasibility study, whether it is advantageous to use the GPS/GIS data to control the gearbox. It only includes one simple rule and is of the type Max Gear as described in section 6.3. The gradient of the road is compared to a simple calculation of the climbing capacity of the truck by taking the weight and engine torque of the vehicle into account. If the gradient of the road on the actual position is above a certain level, the gearbox is forced to shift one gear down. If the gradient exceeds a larger limit, the gearbox will shift two gears down. This is a very simple strategy but has been proven to give a result that shows the system is working Gradient [%] x 10 4 Gear Position [m] x 10 4 Figure 6.2: E20, Vårby, gradient and gearshift strategy On figure 6.2 the gradient on the road near Vårby at E20 is shown, by the blue curve. The red curve shows the maximum gear on the hill, meaning that the gear sequence the truck should use up the gradient is Rule Based for 2nd Model The 2nd rule based controller for this project is developed on the background of several rules, used to improve the gearshift strategy with a number of different goals. 73

86 6. Controller 6.4. Controllers Unfortunately the physics do not allow optimizing a controller to give both optimal economy and optimal power, since these two requirements are contradictory. If the most economic solution is preferred it is to the detriment of the power. And opposite, if full power is preferred, it is to the detriment of the fuel economy. Mostly the normal controllers are based on giving the best fuel economy, thus the controller implemented here are mostly focused on maximizing power for climbing steep gradients as fast as possible. Afterwards the area in between the two solutions has been investigated to give the driver the ability of choosing between the performance criterions. Most Power/Best Economy The dynamics of a diesel engine gives a number of characteristics that can give a number of simple rules of which gear the engine should use for delivering the most torque, or the best economy [Jacobson, 2004]. These boundaries are highly valuable for the gearbox controllers. In figure 5.13 the fuel consumption when producing maximum torque is depicted and in figure 5.16 the maximum torque is depicted. These curves can be used for deriving at which engine speed it is appropriate to shift gears compared to economy or power. In figure 6.3 the delivered torque to the rearaxle in 11th, and 12th gear 11 gear 2000 Torque RPM (12. gear) Figure 6.3: Torque output in 12th and 11th gear gear is depicted, the RPM at the x-axis is refers to the 12th gear. The RPM of the 11th gear has been multiplied with a factor equal the difference in gear ratio between 11th/12th gear. The intersection between the two curves gives the RPM at which a gear down should be performed, if maximum torque is requested. The RPM at the intersection is 1477 RPM between the 12th and 11th gear. Similarly in figure 6.4 the function of torque delivered per fuel unit has been plotted against RPM of the 11th gear and 12th gear. The RPM at 74

87 6. Controller 6.4. Controllers x gear 11 gear Torque/fuel RPM (12. gear) Figure 6.4: Economy in 12th and 11th gear the x-axis again referring to the 12th gear. That means, where the lines are cross the shift down should be performed. The figure shows that the most economic RPM to shift down from 12th to 11th gear is In table 6.1 the similar up/down limits are shown. Power Economy Gear Up Down Up Down Table 6.1: Power and Economy engine speed for the drive modes. Optimal shift up and down. It is clear to see that the Economy mode is always lower that the Power mode. Which also complies to the result of using the Power/Eco button in the real truck, which mainly increases the RPM in the Power mode. It is also seen that a normal gearshift should always be performed be- 75

88 6. Controller 6.4. Controllers tween those limits. Never below the Economy limits and never above the Power limits, that will cause both worse fuel economy and less power. The one exception to this rule will be during acceleration where some gears are skipped, thus RPM will go higher to accommodate a gearshift of two gears. A table similar to table 6.1 could be calculated for 2 or more gears in a shift. But for this controller only one gear has been taken into account. Maximum Gradient The truck is capable of handling a certain gradient, in every gear. If the gradient exceeds this limit, it should be considered if another gear should be selected. The maximum gradient to be handled by the truck is straight forward to calculate from the equations describing the resistance forces acting on the vehicle, described in section 5.3 F dl = F ae + F ro + F cl (6.1) i.e. at maximum gradient the driveline force is equal to the sum of resistance forces (Aerodynamic, Rolling and Climbing). The total equation of the maximum gradient (with a minor simplification in the dependence of gradient in the rolling resistance, causing an error of 0.1%). θ = sin 1 T e maxi gi rη gη r r w 0.5σc w Av 2 fmg (6.2) m g As example, for a speed of 85 km/h the maximum gradient the truck can climb in top gear is 0.95%, and in 11th gear it can climb 1.48%. All gradients smaller than the maximum climbable gradient in the top gear are not important for the controller. Truck Momentum The next step to consider is the momentum of the truck. Some of the gradients in figure 6.5 are so small that the truck will force the gradient with only a small decrease in the speed. Smaller than the decrease in speed by making a gearshift or smaller than the driver is willing to accept. An estimate of the resulting decrease in speed can be calculated and if this is more than the accepted limit, the limits for the gearbox can be changed to expedite a shift down. The estimate can be computed from the momentum of the truck and the integral of the gradient. Momentum of truck: p = mv (6.3) where p is the momentum of the truck in [kg m/s] or [Ns], m is the mass of the vehicle, and v is the velocity in [m/s]. It is thus given, that the change in momentum gives the change in speed: 76

89 6. Controller 6.4. Controllers 5 Gradient [%] Position [km] Gradient exceed [%] Position [km] Figure 6.5: [Top] Gradient of the road between Linköping to Jönköping. [Bottom] Gradients that exceeds the maximum climbable gradients of the truck in 12th gear. FM9-380HP, 40 tons. From the 2nd law of Newton it is also known that: p = m v (6.4) F = p t (6.5) where F is the resulting force acting on the vehicle, and t is the time in seconds. Combining equation 6.4 and equation 6.6 gives: The fact that: v = F t m t = s v (6.6) (6.7) where s is the distance in meters. Taking the integral of a gradient in regards to the distance gives a total expression for the loss of speed climbing a certain gradient: gend g v = start F (s)ds m (6.8) v where g start and g end are the beginning and end of the gradient. On figure 6.6 this result has been depicted. It can be seen that the speed of the vehicle will drop with approximately 55 km/h on the steepest gradient if 12th gear is used and if a gearshift is introduced it can be decreased to 77

90 6. Controller 6.5. Summary around 40 km/h. Now it is fairly straight forward to decide how much the speed is allowed to drop before a gearshift should be considered. This gear 11 gear Estimated speed decrease [km/h] Position [km] Figure 6.6: Estimated decrease in speed on steepest gradients between Linköping and Jönköping, using 12th or 11th gear. FM9-380HP, 40 tons. parameter could be decided by the driver, but tests has shown that the relationship between speed decrease and economy/power is not simple. And tests should therefore clarify what the optimal limit for a speed decrease should be. These will be carried out in chapter 7. The 2nd rule based controller is implemented in the MATLAB file road section calc.m which can be found in appendix B. 6.5 Summary This chapter has dealt with 3(+ reference) different controllers, two simple controllers, and one more advanced, based on rules derived from the advanced model. The first simple controller, based on the Economy / Power mode is very simple and makes use of the fact that the engine produces a higher torque at a higher RPM, but the economy is worse. The second simple controller uses information about the gradient of the road at that section it is driving of. If the gradient is above a certain limit, a shift down is imposed. The last advanced controller also uses information about the gradient of the road, but it looks up the road, and estimates the loss of speed on the imminent gradient. If that loss of speed is more than an acceptable limit, the shifting strategy of the gearbox is changed. 78

91 Chapter 7 Tests and Results This chapter describes tests and associated results of the different parts of the project. The test is divided in two main parts plus a summary in the end: Road Information Test of the GIS, GPS and a very simple controller. This is mainly the parts that should be installed in a real truck to make use of the system. Controllers The two models are tested with different controllers to measure the performance of using the improved system. Parameters like speed, fuel consumption and gear shifts have been measured and compared to the reference model. Summary A summary of the tests, to give a clear overview of the gained performance compared to reality and the difference of model 1 and Road Information This part of the system is as described in chapter 4 mostly a feasibility study, since some parts of the system are already installed and used in normal vehicles. This part is only superficial implemented to test whether it is possible to supply a controller with road profile data, and the associated tests are carried out in order to test the overall functionality. Thus accuracy of GPS, GIS, map matching algorithm etc. are not tested. Test/Evaluation of this part of the project is carried out by combining GIS data, GPS data and a very simple controller for the gearbox, and test whether these parts can work together and produce a result that supports the possibility of using a 3D map for control of a gearbox. The whole implementation is depicted in figure 7.1 where the test vehicle (car) is running along highway E47/E55 heading south, with a velocity 79

92 7. Tests and Results 7.1. Road Information around 100 km/h. The green markers in the upper left plot are the actual GPS positions, and the black dots are the matched positions on the real blue highway. Based on this, the program finds the altitude and gradients Figure 7.1: Program calculating the road sequence ahead of the vehicle of the road in front of the vehicle, and plots this in a contour plot on the lower right plot. The position of the vehicle is the black dot heading to the right, which means up the hill. Right above this plot, a small label indicates the maximum gear to be 11th gear instead of the normal 12th gear in the vehicle, i.e. the program decides it is more efficient to drive up this particular gradient in 11th gear. The functionality of this part of the project is evaluated, the program is capable of tracking the GPS position to a digital map, finding the gradients of the road ahead of the vehicle, and based on this give a simple command to the driver to shift down one or two gears. With the accuracy and implementation time of this parts it is a very satisfactory result, and the possibility of developing a real useful system seams possible. 80

93 7. Tests and Results 7.2. Controllers 7.2 Controllers A total of four controllers (1 reference, and 3 new improved) has been implemented in chapter 6. In this part the controllers has been tested against each other, to measure the gained performance. This section is divided in two parts, first a comparison test and after that a detailed test of the final advanced controller implemented on the 2nd model Comparison Test The comparison test includes 5 different test scenarios, including both models and three controllers. All the test includes: 1st Model, reference Due to the minor differences between the 1st model and 2nd model, both of the models are used as reference to the different controllers. 1st Model, E/P controller The simple Economy/Power controller used on the 1st model. 1st Model, simple controller The simple rule based controller derived for the 1st model. 2nd Model, reference Reference for the controllers applied to the 2nd model. 2nd Model, E/P controller The simple Economy/Power controller used on the 2nd model. 2nd Model, simple controller The simple rule based controller derived for the 1st model, but used on the 2nd model 2nd Model, Advanced controller The advanced rule based controller specially derived for the 2nd model. All of the controllers are tested on 5 different pieces on road. E20, including one large gradient, E47/E55 - a rather flat road. Route 21, including two steep gradients. E4 real road data from Sweden, including more steep gradients tested in both directions. 81

94 7. Tests and Results 7.2. Controllers E20 Speed [km/h] Reference E/P Simple Rule Position [km] Figure 7.2: E20-1st Model, 85 km/t, 40 tons Speed [km/h] Reference 40 E/P Simple Rule Advanced Rule Position [km] Figure 7.3: E20-2nd Model, 85 km/t, 40 tons Controller Min. speed Avg. speed Fuel usage Gearshifts 1st Model [km/h] [km/h] [%] Reference E/P Simple Rule Controller Min. speed Avg. speed Fuel usage Gearshifts 2nd Model [km/h] [km/h] [%] Reference E/P Simple Rule Advanced Rule Comments Driving E20 east, from The Great Belt Bridge to Slagelse East, includes one very steep gradient near Vårby. As seen in the figures 7.2 and 7.3, the minimum climbing speed can be improved and thereby the average speed. In one case the advanced controller on the 2nd model, it is possible to lower the fuel usage a little without using more gearshifts. 82

95 7. Tests and Results 7.2. Controllers Route 21 Speed [km/h] Reference E/P Simple Rule Position [km] Figure 7.4: Route 21-1st Model, 85 km/t, 40 tons Speed [km/h] Reference 40 E/P Simple Rule Advanced Rule Position [km] Figure 7.5: Route 21-2nd Model, 85 km/t, 40 tons Controller Min. speed Avg. speed Fuel usage Gearshifts 1st Model [km/h] [km/h] [%] Reference E/P Simple Rule Controller Min. speed Avg. speed Fuel usage Gearshifts 2nd Model [km/h] [km/h] [%] Reference E/P Simple Rule Advanced Rule Comments Driving Route 21 east, from Holbæk to Copenhagen, features three significant gradients. As seen in the figures 7.4 and 7.5, the minimum climbing speed can be improved and thereby the average speed. In one case, the E/P controller on the 2nd model, features a lowered fuel usage. But in regards to speed this solution is not superior. 83

96 7. Tests and Results 7.2. Controllers E47/E Speed [km/h] Speed [km/h] Reference 30 E/P Simple Rule Position [km] 50 Reference 40 E/P Simple Rule Advanced Rule Position [km] Figure 7.6: E47/E55-1st Model, 85 km/t, 40 tons Figure 7.7: E47/E55-2nd Model, 85 km/t, 40 tons Controller Min. speed Avg. speed Fuel usage Gearshifts 1st Model [km/h] [km/h] [%] Reference E/P Simple Rule Controller Min. speed Avg. speed Fuel usage Gearshifts 2nd Model [km/h] [km/h] [%] Reference E/P Simple Rule Advanced Rule Comments Driving E47/E55 north, from Lyngby to Helsingør, features two significant gradients, and a number of smaller gradients on the limit of being significant. As seen in the figures 7.6 and 7.7, the minimum climbing speed can be improved and thereby also the average speed, but not as much due to the rather short gradients. All tests shows an increased fuel usage. 84

97 7. Tests and Results 7.2. Controllers E4 Linköping - Jönköping Speed [km/h] Speed [km/h] Reference 30 E/P Simple Rule Position [km] 50 Reference E/P 40 Simple Rule Advanced Rule Position [km] Figure 7.8: E4 Linköping - Jönköping - 1st Model, 85 km/t, 40 tons Figure 7.9: E4 Linköping - Jönköping - 2nd Model, 85 km/t, 40 tons Controller Min. speed Avg. speed Fuel usage Gearshifts 1st Model [km/h] [km/h] [%] Reference E/P Simple Rule Controller Min. speed Avg. speed Fuel usage Gearshifts 2nd Model [km/h] [km/h] [%] Reference E/P Simple Rule Advanced Rule Comments Driving E4, from Linköping towards Jönköping, features a number of smaller gradients. As seen in the figures 7.8 and 7.9, the minimum climbing speed can be improved a little and also the average speed using the 2nd model. All tests shows an highly increased fuel usage compared to the gained performance of speed. 85

98 7. Tests and Results 7.2. Controllers E4 Jönköping - Linköping Speed [km/h] Speed [km/h] Reference 30 E/P Simple Rule Position [km] 50 Reference E/P 40 Simple Rule Advanced Rule Position [km] Figure 7.10: E4 Jönköping - Linköping - 1st Model, 85 km/t, 40 tons Figure 7.11: E4 Jönköping - Linköping - 2nd Model, 85 km/t, 40 tons Controller Min. speed Avg. speed Fuel usage Gearshifts 1st Model [km/h] [km/h] [%] Reference E/P Simple Rule Controller Min. speed Avg. speed Fuel usage Gearshifts 2nd Model [km/h] [km/h] [%] Reference E/P Simple Rule Advanced Rule Comments Driving E4, from Jönköping towards Linköping, features two very steep gradients, and a number of smaller gradients. As seen in the figures 7.10 and 7.11, the minimum climbing speed can be improved significantly and the average speed too. Speed is significantly increased by the advanced controller even the number of gear shifts is the lowest. Fuel usage is fluctuating between the different controllers. 86

99 7. Tests and Results 7.2. Controllers Evaluation To give an overview of the average performance of the different controllers, all the results from the past test has been put together in table 7.1. The table Model Min. speed Avg. speed Fuel usage Gearshifts and Increase Increase Increase Increase Controller [%] [%] [%] [%] 1st - E/P st - Simp. Rule nd - E/P nd - Simp. Rule nd - Adv. Rule Table 7.1: Combined results of tests, FM9-380, 40 tons. shows that the results obtained using different controllers are of variable characteristics. It is hard to determine which of them are the best. The E/P controller gives surprisingly positive results, taken into account that this controller is only using the road information to determine if the truck is in hilly locations. The Simple Rule controller also performs good but the drawback of this controller is the high amount of gearshifts, 4 to 5 times as many gearshifts as the normal controller. The Advanced Rule controller gives a good average result. It gives the highest minimum speed which are one of the main goals of this project, but it also gives the second best average speed improvement. The drawbacks are the high fuel usage and high amount of gearshifts, close to twice as many as normal. The Advanced Rule controller has an advantage compared to the other controllers. It has more possibilities of being optimized and it takes more parameters into account when deciding which gears to use. Therefore this controller is further tested in the following section Detailed Test of Final Controller The advanced rule based controller is tested in regards to a number of different parameters to investigate how parameters as speed, and weight influences the performance of the controller. Also the allowed decrease in speed, to be determined by the driver is tested, to investigate speed and fuel consumption. The highway E4 from Jönköping to Linköping, a distance of 127 km has been used for testing the controller. This road is made of real measurements and includes a number of small and large gradients. 87

100 7. Tests and Results 7.2. Controllers Reference Speed The dependence of the vehicle speed has been investigated to see if the system performs equal at different speed or not. The speed has been varied from 80 to 90 km/h which are the normal range of highway cruising speed of trucks Min. speed [%] Avg. speed [%] Fuel usage [%] 2 1 Gearshifts [%] Reference speed [km/h] Reference speed [km/h] Figure 7.12: Variation in vehicle cruising speed from 80 km/h - 90 km/h. Driving Jönköping to Linköping, 40 tons, allowed speed drop 5 km/h. Comments From the results in figure 7.12 it can be seen that the performance of the controller is affected by the reference speed. The increased minimum speed is higher when the reference speed is high, due to the increased momentum of the vehicle. The increased average speed is highest at low speed, due to the larger potential of extra torque of the engine at that speed. The fuel usage is increased at all tests. Gearshifts are the same at low speed, but more than doubled at high speed. 88

101 7. Tests and Results 7.2. Controllers Weight The vehicle weight has been varied to investigate if the system performs best if the truck is heavy or light. The weight of the truck has been varied from 16 to 48 tons. Which are normal weight of trucks. (48 tons, is maximal weight in Denmark. 60 tons is maximal weight in Sweden which could have been used, due to the fact that the road section is Swedish. However it is not normal only to use 380HP for 60 tons) Min. speed [%] Avg. speed [%] Fuel usage [%] Gearshifts [%] Weight [tons] Weight [tons] Figure 7.13: Variation in vehicle weight from 16 tons - 48 tons. Driving Jönköping to Linköping at 85 km/h, allowed speed drop 5 km/h. Comments From the results in figure 7.13 it is clear that the largest potential of the system is at high vehicle weight. Fuel and number of gearshifts are fluctuating but a small tendency in the results shows better performance in regards to fuel despite of the increased use of gearshifts around 30 tons. It might be due to some kind of match between 30 tons and 380 HP. 89

102 7. Tests and Results 7.2. Controllers Allowed Speed Decrease The controller uses a driver input to decide when to shift gears. If the predicted speed loss is more than a specified limit, the controller forces a gearshifts. This limit has been tested from 1 km/h below reference speed to 20 km/h below reference speed. Min. speed [%] Avg. speed [%] Fuel usage [%] Gearshifts Allowed speed drop [km/h] Allowed speed drop [km/h] Figure 7.14: Variation in allowed speed drop from 1 km/h - 20 km/h. Driving Jönköping to Linköping at 85 km/h, 40 tons. Comments From the results in figure 7.14 it can be seen that there is a clear relationship between the performance of the system compared to the allowed speed drop. If a smaller speed decrease is allowed, the average and minimum speed is increased, but to the detriment of the fuel usage and number of gearshifts. Likewise if a larger speed drop is allowed, the increase in speed is smaller but the fuel usage and gearshifts is lowered. It is interesting to see that the fuel usage can be decreased with more than 1 % and gearshifts decreased with 25% and still the average speed is increased with more than 0.5% and the minimum speed is increased with 8%. 90

103 7. Tests and Results 7.3. Summary 7.3 Summary Road Information The tests shows that it is possible to use an integrated GIS/GPS system to feed the gearbox controller with information about the road ahead of the vehicle. However it has proven to be difficult to procure a 3D map, and the result of the GIS/GPS is therefore on the basis of that. No coinciding data has been available of the road sections. Controllers It is clear that the performance of the vehicle can be changed using road information as input to the controller. It is also clear from the test of the advanced controller, that some kind of tradeoff between the wish for speed and economy has to be made. The controller performance also changes when the reference speed and vehicle weight changes. The differences between the performance of the 1st and 2nd model is negligible, but it is hard to defend to use the 1st model on this basis. Due to fact that large assumptions has been made, and the model is developed on the bases of weak foundations. 91

104

105 Chapter 8 Discussion 8.1 Road Information This part of the project has been carried out mostly as a feasibility study. It includes simple working versions of GIS, GPS, Map Matching, Road Sequence and a simple Controller for testing. The system is tested on surrounding highways and the results described in chapter 7 shows that the system is working as expected. The available GIS and GPS data is interacting through a Map Matching algorithm and a controller gives a simple output to the driver of the vehicle to shift gears when the road profile requests it. This part of the project has not been further developed, due to the fact, that many of the parts are already implemented in far more sophisticated versions in the present GPS systems used in vehicles today. However the extension of the map from 2D to 3D has been investigated and procuring a suitable 3D map seems to be the largest task. The possibility of creating a map from available GPS data when driving on roads has been investigated and shows positive results. An online global map updating function, where vehicles measures and transmits data to a global map facility, seems possible. The controller tested in this part is very simple, but should be extended with the functionality of the advanced controllers, described in chapter 6. Implementing the advanced controllers would make it possible to make tests in real life by an open-loop solution in a truck (command send to driver only). To test if the system is deciding the optimal way to handle gradients compared to the defined control-actions. Later on the system could be implemented in closed loop, controlling the gearbox directly. 93

106 8. Discussion 8.2. Model 8.2 Model During the project, two models has been implemented to test the performance of the system. It was not possible to test the system on a real truck, instead the two models has been implemented and adjusted to reflect a real truck as close as possible. Real data and tests on real truck has been carried out in order to create enough reference data to implement a good model behaving like a real truck. In the first attempt, a simple model was created based on simple equations and basic knowledge of trucks. This model was used as a feasibility study of the whole project to see whether the result of the implemented system reacted and responded like intended and described in the problem statement. The result of the first model was positive, and the new improved second model was developed to make the model and achieved results as close as possible to reality, to make it plausible that this system would perform similar in real life. This model is tested with a number of controllers, spanning from very simple to advanced rule based controller. The result of these tests, described in chapter 7, somehow reflects the expected results set in the beginning of the project. The result shows that it is fairly easy to improve the average speed of a driving truck, and easy to improve the minimum climbing speed of trucks. However the amount of fuel used to achieve this result is fluctuating from higher fuel consumption than normal to lower fuel consumption than normal. It indicates that if you instead of using the system for getting the maximum possible speed increase, tune the system for a smaller speed increase it is actually also possible to save fuel. For a 40 ton vehicle traveling at 85 km/h it is possible to decrease the fuel consumption with more than 1 %, decrease the gearshifts with 25% and still maintain an average speed 0.5% higher than normal. The minimum climbing speed on the steepest gradient is 8% higher than normal. If this holds for a real truck the system should be very advantageous to install. The system will improve the constant speed of the truck, lower the driving time, lower the fuel consumption and improve safety on steep gradients by letting the truck move faster. The forte of the implemented controllers is the question about safety. Often automatic systems reveals a safety issue if the system do not respond as intended, if the system is fed with wrong data or disturbances drowns the real signals. In this case the only problem that could arise, is an unintended gearshift which can not harm anything. The controller is implemented in such a way that it can only adjust some limits for shifting and not the shifting itself. Thus gearshifts will only be performed if the truck is actually responding as 94

107 8. Discussion 8.2. Model the controller expects. If the controller decides that a gradient in the front is too steep it will change the limits for shifting, but if there are no gradient, and the engine load and RPM are not showing any signs of a gradient, the gearshift will not be carried out. Similar if there is a gradient, and the system has not noticed it, the gearbox can always rely on its basic shifting strategy. In regards to the cost of implementing this system, it should be compared to the saved fuel and time using this system. If the system is capable of saving 1% fuel in average, which means more than 500 liters a year on a standard truck. With a cost of more than 1 e/liter, it means more than e 500 saved a year. The increased speed /decreased time is difficult to measure in money, since most of it probably would be to the benefit of the driver, but principally 0.5% time is saved, and 0.5% saved salary of a truck driver a year could be more than 20 hours, equals approximately 300 e. The decreased number of gearshifts could save a visit on the workshop. Overall there should be enough gained savings to compensate the extra price of the system. And finally the environment would be exposed to less pollution. The positive results of the project makes a strong case for proceeding with an implementation of the system in a real truck. Implementation, would in the case of using the Volvo system, require an extension of the Dynafleet system to include 3D maps, instead of the 2D maps used today. An amount of software including the rules developed in the controller part of this project should also be implemented in the onboard computer of the truck. 95

108

109 Chapter 9 Conclusion The objective of the project was to develop a system for automatic gearboxes to mimic the behavior of an experienced driver. This objective was intended to be achieved using road information to improve the control of the gearbox. The project has focused on two main areas, road information and a simulation model. The road information part includes a feasibility study of controlling the gearbox, using an improved GIS/GPS system, extended with a 3D digital map as a reference to the road profile ahead of the truck. The result of this work shows that it is possible to influence the behavior of an automatic gearbox with the GIS/GPS system. However the largest task lies in procuring a sufficient 3D map. A number of different map solutions has been investigated, from simple available maps to GPS measurement based maps. This part of the system consists of the parts that should be installed in a real vehicle, to make it possible to use the system for reading the road ahead of the vehicle, to decide what action to take to use the optimal sequence of gearshifts for climbing a certain gradient. The other part of the project, the development of a simulation model, includes a number of controllers and a highly detailed simulation model of a truck. The model includes a sophisticated engine model based on the MVEM, an automatic gearbox, including shifting sequences and synchronization, a dynamic vehicle model and a number of different controllers. The result of the work concludes that it is possible to increase the average cruising speed of a truck with more than 1%, dependent of the road profile. As expected the system shows a tradeoff between increased speed 97

110 9. Conclusion and fuel economy. However, it has proven to be possible to achieve both, if the speed is only increased with 0.5%, it is possible to save more than 1% fuel, decrease gearshifts by 25% and increase the minimum climbing speed on gradients with more than 8%, dependent of the road profile. 98

111 Chapter 10 Future Work A number of ideas for the further work with this project which has not been implemented during the thesis work. GIS Procuring a 3D map has proven to be a large task, future work could include development of 3D maps by means of data or by means of an online updating system, based on GPS measurements, as described in section Controller A number of different controllers has already been implemented, however more rules or other types of controller are likely to be investigated. Tests has indicated that it is hard to find an optimum for a controller, the result is dependent of a lot of factors. The future work could include the implementation of a MPC controller which is capable of finding the optimal solution in regards to a number of weight parameters. A lot of the results from the advanced controller tests can be used as background to determine these weight parameters. Truck Implementing this system on a real truck will be the ultimate test and validation of the system. Driving downhill The system could easily be adapted to calculate downhill gradients too, to determine if the truck is capable of keeping the speed down with the engine brake. If it is not possible it could force a shift down to increase the performance of the engine brake, and thereby decrease the wear of the ordinary disc brakes. 99

112

113 Nomenclature Abbreviations ACC DEM DGPS EGR GCM GIS GPS IVSS KMS MPC MVEM Adaptive Cruise Control Digital Elevation Model Differential-GPS Exhaust Gas Recirculation Gross Combination Mass Geographical Information System Global Positioning System Intelligent Vehicle Safety Systems Kort & Matrikelstyrelsen Model Predictive Control Mean Value Engine Model NAVSTAR-GPS NAVigation System with Timing And Ranging Global Positioning System PID RPM SCR TIS UTM VCADS VEB Proportional-Integral-Derivative controller Revolutions Per Minute Selective Catalytic Reduction Transport Information System Universal Transverse Mercator Coordinate Vehicle Computer Aided Diagnostic System Volvo Engine Brake 101

114 Nomenclature Constants α br Cruise control parameter 0.01 α cc Cruise control parameter 0.01 η r Rearaxle efficiency 0.97 η c Turbo efficiency 0.73 κ Adiabatic constant, c p /c v 1.4 µ d Dynamic friction coefficient 0.4 σ Air density 1.19kg/m 3 τ dbr Cruise control - differentiation time constant 3s τ dcc Cruise control - differentiation time constant 3s τ engine Engine time constant 0.5s τ ibr Cruise control - integration time constant 10s τ icc Cruise control - integration time constant 10s τ turbo Turbo time constant 1s τ V EB Engine brake time constant 0.5s A Vehicle front area 9.75m 2 a d Drag constant Nm RP M b d Drag constant 51.51Nm c w Vehicle drag coefficient 0.6 f Rolling resistance coefficient g Gravity 9.82m/s 2 H u Heating value of fuel 43kJ/kg i r Rearaxle ratio 1 : 3.10 J w Wheel inertia 32.9kgm 2 k b Brake constant Nm K pbr Cruise control - proportional gain 0.2 K pcc Cruise control - proportional gain

115 Nomenclature L th Stoichiometric air/fuel ratio 14.7 n r Number of crankshaft revolutions per stroke 2 n cyl Number of cylinders 6 n max Engine maximum speed 2500RP M p a Ambient pressure 1.013bar R Gas constant r w Wheel radius 0.52m R cl Clutch disk radius 0.3m s i Air charge per stroke constant 1.05 T a Ambient temperature 293K T V EBmax Maximal brake torque at maximum RPM 1360Nm V d Engine displacement 9.4L V ref Reference engine displacement 1.275L y i Air charge per stroke constant 0.13 Variables α Longitudinal road gradient rad ω Angular acceleration rad/s 2 η g Gearbox efficiency [ ] η i Indicated efficiency ω w Wheel speed s 1 Φ Fuel/Air equivalence ratio τ c Turbo time constant s τ d Time delay for fuel injection s a Total acceleration of vehicle m/s 2 B Brake control signal [0; 1] Cl Clutch factor [0, 1] e v Volumetric efficiency 103

116 Nomenclature F Total force acting on vehicle N F n Force on clutch N F w Total force acting on wheels Nm F ae Total Aerodynamic resistance N F cl Total climbing resistance N F dl Total force generated by the driveline N F ro Total rolling resistance N F rr Total opposing force (rolling, air and climbing resistance) N f c Fuel consumption liter/s g end End of gradient m g start Beginning of gradient m i g Gearbox ratio [1 : ] J Moment of inertia kgm 2 K engine Engine load [0; 1] k step Engine brake step [0; 3] K throttle Position of throttle [0; 1] m Total mass of vehicle kg m f Fuel mass flow kg/s m ap Port air mass flow kg/s n Engine speed rpm n g Gearbox rotational speed rpm p Truck momentum kg m/s P b (n) Power supplied to load kw p c Compressor outlet air pressure bar P f (n) Loss of friction kw p i Intake manifold air pressure bar P exhx Power to drive turbo kw 104

117 Nomenclature p rc p c /p a s Distance m T Torque Nm t Time s T c Compressor outlet air temperature K T e Total engine torque Nm T g Torque delivered from gearbox Nm T i Intake air temperature K T cl Torque transmitted by clutch Nm T drag Engine drag torque Nm T normmax Maximal normal engine torque at certain RPM Nm T norm Normal engine torque Nm T turbomax Maximal turbo torque at certain RPM Nm T turbo Extra turbo torque Nm T V EB Engine brake torque Nm v Vehicle speed m/s 105

118

119 Bibliography [COWI, 2006] COWI (2006). Height models and laser scanning. Ingeniøren - ing.dk. [Dakai Yang, 2003] Dakai Yang, Baigen Cai, Y. Y. (2003). An improved map-matching algorithm used in vehicle navigation system. IEEE, (pp. 5). [Glielmo & Vasca, 2000] Glielmo, L. & Vasca, F. (2000). Engagement control for automotive dry clutch. Dipartimento di Informatica e Sistemistica, Universita de Napoli, (pp ). [Gustafsson, 2006] Gustafsson, N. (2006). The use of positioning systems for look-ahead control in vehicles. Master s thesis, Linköping University. [Hellström, 2005] Hellström, E. (2005). Explicit use of road topography for model predictive cruise control in heavy trucks. Master s thesis, Linköping University. [Hendricks, 1989] Hendricks, E. (1989). Mean value modelling of large turbocharged two stroke diesel engines. SAE Technical Paper No [Hendricks et al., 1984] Hendricks, E., Blanke, M., & Andersen, J. S. (1984). On modelling large two stroke diesel engine: Simulation and model evaluation. Internal Report, Servolaboratoriet, Technical University of Denmark. [Hendricks et al., 1997] Hendricks, E., Chevalier, A., Jensen, M., & Sorenson, S. C. (1997). Modelling of the intake manifold filling dynamics. SAE Technical Paper No [Hendricks et al., 2004] Hendricks, E., Fam, M., Magnussen, S., Bertelsen, A., & Sorenson, S. (2004). A simply calibrated turbodiesel mvem. Institute for Automation, Ørsted Institute, DTU, Sixth Milepost Report, Toyota/DTU Research Cooperation. [Hendricks et al., 2005] Hendricks, E., Sorenson, S., Magnusson, S., & Bertelsen, A. (2005). Compact and accurate turbocharger modelling for engine control. SAE Technical Paper No

120 BIBLIOGRAPHY BIBLIOGRAPHY [Hendricks & Sorenson, 2006] Hendricks, E. & Sorenson, S. C. (2006). Private Communication, Elbert Hendricks, Ørsted Automation and Spenser Sorenson, MEK. [Heywood, 1988] Heywood, J. B. (1988). Internal Combustion Engine Fundamentals. p. 722: McGraw-Hill Book Company. [Ivarsson et al., 2006] Ivarsson, M., Sahlholm, P., Blackenfelt, M., Johansson, K. H., Jansson, H., & Nielsen, L. (2006). Vehicle Control Using Preview Information. Technical report, Scania CV AB, Södertälje, Royal Institute of Technology, Stockholm, Linköping University, Linköping. [Jacobson, 2004] Jacobson, B. (2004). Gearshift sequence optimisation for vehicles with automated non-powershifting transmissions. Machine and Vehicle Systems, Chalmers University of Technology, Göteborg, Sweden, (pp. 22). [Jansson et al., 2006] Jansson, H., Kozica, E., Sahlholm, P., & Johansson, K. H. (2006). Improved road grade estimation using sensor fusion. Proceedings of the 12th Reglermöte in, (pp. 5). [Johansson, 2005] Johansson, K. (2005). Road slope estimation with standard truck sensors. Master s thesis, Kungliga Tekniska Högskolan. [Kiencke & Nielsen, 2000] Kiencke, U. & Nielsen, L. (2000). Automotive Control Systems. Linköping Universitet: Springer-Verlag Berlin. [Lattemann et al., 2004] Lattemann, F., Neiss, K., Terwen, S., & Connolly, T. (2004). The predictive cruise control - a system to reduce fuel consumption of heavy duty trucks. SAE Paper [Magazine, 2006a] Magazine (2006a). Daily work evaluation, Volvo FH12. Technical report, Trailer Magazine. [Magazine, 2006b] Magazine (2006b). Testdrive, Scania R620. Technical report, Trailer Magazine. [Mangan et al., 2003] Mangan, S., Wang, J., & Wu, Q. (2003). Longitudinal road gradient estimation using vehicle can bus data. IEEE, (pp. 6). [Müller, 1997] Müller, M. (1997). Mean value modelling of turbocharged spark ignition engines. Master s thesis, MEK, DTU. [Müller et al., 1998] Müller, M., Hendricks, E., & Sorenson, S. C. (1998). Mean value modelling of turbocharged spark ignition engines. SAE Technical Paper No [Wingren, 2005] Wingren, A. (2005). Fordonsreglering med framförhållning. Master s thesis, Linköbing University. 108

121 BIBLIOGRAPHY BIBLIOGRAPHY [Zogg, 2002] Zogg, J.-M. (2002). GPS Basics, Introduction to the system, Application overview. u-blox. 109

122

123 List of Figures 2.1 I-shift gear lever Flowchart of system implementation intended here including Road Information, Truck Model and, Controller Predictive Cruise Control Application for GPS/GIS aids in vehicles, according to the amount of detail available in digital maps Road Information part of the project Holux GM-210 GPS receiver Program for testing GPS Program for testing GPS and GIS TOP10DK 2D Highway theme of Søllerød municipality Part of TOP10DK DEM theme, of Søllerød municipality GIS 3D Highway of Søllerød municipality GIS 2D Data of Sealand GIS 3D Data of Sealand D GIS and GPS from E47/E55 test D Profile GIS and GPS from E47/E55 test D Profile of map created from GPS measurements at E47/E E4, Linköping - Jönköping, road profile Map matching algorithm. Raw GPS data Map matching algorithm. Matched data Map matching algorithm Test of the map matching algorithm Description of the road sequence in front of the vehicle Program calculating the road sequence ahead of the vehicle Truck part of the project Volvo FM9-380 Strawtransport Volvo FM9-300 Test truck at Volvo Truck Center Denmark A/S Volvo FM9-300 Turbo pressure. Dynamic response

124 LIST OF FIGURES LIST OF FIGURES 5.5 Speed and fuel tests at Vårby - E20. Automatic - Solid, Manual - Dashed The effect of throttle and load on the gearbox strategy for shifting up st Model main parts Engine torque of a Volvo FM9, 380HP Engine brake torque of a Volvo FM9, 380HP Gear ratios of an I-shift VT2412B Gauges from the simulation model, similar to the gauges in a Volvo Truck Primary parts of the engine Fuel Function Maximum turbo pressure Engine power Engine torque Engine efficiency η i Equivalence ratio θ Engine dynamic response - Torque. It is seen that the response with the fuel time delay is not remarkable slower than the response without the time delay Engine dynamic response - Efficiency. The efficiency is kept high by using a time delay in the fuel injection Gearbox shifting sequence Gear shift sequence - shift down Gear shift sequence - shift up Controller part of the project E20, Vårby, gradient and gearshift strategy Torque output in 12th and 11th gear Economy in 12th and 11th gear [Top] Gradient of the road between Linköping to Jönköping. [Bottom] Gradients that exceeds the maximum climbable gradients of the truck in 12th gear. FM9-380HP, 40 tons Estimated decrease in speed on steepest gradients between Linköping and Jönköping, using 12th or 11th gear. FM9-380HP, 40 tons Program calculating the road sequence ahead of the vehicle E20-1st Model, 85 km/t, 40 tons E20-2nd Model, 85 km/t, 40 tons Route 21-1st Model, 85 km/t, 40 tons Route 21-2nd Model, 85 km/t, 40 tons E47/E55-1st Model, 85 km/t, 40 tons E47/E55-2nd Model, 85 km/t, 40 tons

125 LIST OF FIGURES LIST OF FIGURES 7.8 E4 Linköping - Jönköping - 1st Model, 85 km/t, 40 tons E4 Linköping - Jönköping - 2nd Model, 85 km/t, 40 tons E4 Jönköping - Linköping - 1st Model, 85 km/t, 40 tons E4 Jönköping - Linköping - 2nd Model, 85 km/t, 40 tons Variation in vehicle cruising speed from 80 km/h - 90 km/h. Driving Jönköping to Linköping, 40 tons, allowed speed drop 5 km/h Variation in vehicle weight from 16 tons - 48 tons. Driving Jönköping to Linköping at 85 km/h, allowed speed drop 5 km/h Variation in allowed speed drop from 1 km/h - 20 km/h. Driving Jönköping to Linköping at 85 km/h, 40 tons

126

127 List of Tables 5.1 Estimated maximum turbo pressure of FM Speed and fuel tests at Vårby E Test of acceleration 0-80 km/h Average fuel tests, Route tons, 85 km/h Minimum speed test, E tons, 85 km/h Acceleration test, 0 km/h - 80 km/h Power and Economy engine speed for the drive modes. Optimal shift up and down Combined results of tests, FM9-380, 40 tons

128

129 Appendix A Models SIMULINK diagrams of Models. Appendix A.1 Appendix A.2 1st Model 2nd Model 117

130 A. Models A.1. 1st Model A.1 1st Model SIMULINK diagrams of 1st Model. Blue blocks includes Acceleration, Speed, RPM. Red blocks includes Torque. Green blocks includes Moment of inertia. Appendix A.1.1 Appendix A.1.2 Appendix A.1.3 Appendix A.1.4 1st Model - Main 1st Model - Engine 1st Model - Gearbox 1st Model - Differential, brakes and wheels 118

131 A. Models A.1. 1st Model A.1.1 1st Model - Main 119