DF Development of a LiDAR Cleaning System for Autonomous Trucks A Product Development Project at Volvo Trucks Master’s thesis in Product Development CARL GUSTAFSSON FREDRIK KARLSSON Department of Industrial and Materials Science CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2019 Master’s thesis 2019 Development of a LiDAR Cleaning System for Autonomous Trucks A Product Development Project at Volvo Trucks CARL GUSTAFSSON FREDRIK KARLSSON DF Department of Industrial and Materials Science Chalmers University of Technology Gothenburg, Sweden 2019 Development of a LiDAR Cleaning System for Autonomous Trucks A Product Development Project at Volvo Trucks CARL GUSTAFSSON FREDRIK KARLSSON © CARL GUSTAFSSON, FREDRIK KARLSSON, 2019. Supervisor: Marcus Leidefeldt, Special Vehicles Development Examiner: Erik Hultén, Department of Industrial and Materials Science Master’s ThesisDepartment of Industrial and Materials Science Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Cover: Rendering of final design proposal. Typeset in LATEX Printed by Chalmers Reproservice Gothenburg, Sweden 2019 iv Development of a LiDAR Cleaning System for Autonomous Trucks A Product Development Project at Volvo Trucks CARL GUSTAFSSON FREDRIK KARLSSON Department of Industrial and Materials Science Chalmers University of Technology Abstract Autonomous driving is emerging as the future of transportation. For autonomous driving to be safe and reliable the perception sensors need sufficient vision in some- times challenging operating conditions including dust, dirt and moisture. LiDAR perception sensors used in certain autonomous driving solutions require both a clean and dry sensor screen. The purpose of this thesis is that through developing and testing of concepts, provide Volvo with both knowledge and a baseline design to help with future product development of sensor cleaning systems in similar applications. The developed concept for an improved cleaning system for the 2D-LiDAR sensors mounted to autonomous Volvo trucks are used in a mining operation. Emphasis in the concept is on the resource efficiency of the fluid available on the truck during operation. A cleaning head placed above the LiDAR does not compete for space with other systems around the LiDAR, allowing for greater design freedom. The prototypes indicate high potential in cleaning performance and as they were 3D printed in PA 12, proved a promising manufacturing method allowing complex ge- ometries. The design combined with 3D printing also enables flexibility, allowing quick design changes to conform to different LiDAR shapes. Comparative testing against the currently implemented cleaning system is done in this thesis with dirt scenarios from intended to use-cases. The result provides clear evidence of improvements against the current solution. Following an exploratory design and testing phase, it was concluded that great improvement in resource efficiency was achieved through the use of fluid specific nozzles, drastically reducing the fluid consumption of previous cleaning systems used by Volvo. Testing also gave evidence of existing knowledge gaps due to the novelty of the LiDAR cleaning task. This convinced the thesis workers that more research is needed and that standardized test methods need to be established to facilitate comparability between cleaning systems developed and tested in the future. Keywords: Autonomous Driving, Autonomous Trucks, LiDAR, Sensor, Cleaning, Nozzle, 3D printing, Product Development v Acknowledgements The thesis workers would like to express sincere gratitude to the people who have helped us throughout this thesis work. Special thanks are given to: Marcus Leidefeldt, the thesis workers’ supervisor at Volvo, for his guidance and professional support during the project. Associate Professor Erik Hulthén, the thesis workers’ supervisor at Chalmers, for his guidance throughout the project. Additional thanks are given to: The entire team at Special Vehicles Development at Volvo for welcoming us into their group and for their support. The personnel at A-Hallen testing facility for their help and expertise during the testing phase of the project. Finally, the thesis workers want to thank people in any way involved in the progress of this thesis work for their contributions, both big and small. Carl Gustafsson and Fredrik Karlsson, Gothenburg, June 2019 vii List of Abbreviations AD Autonomous Driving ADAS Advanced Driver-Assistance System ADS Automated Driving System BSPP British Standard Pipe Parallel BSPT British Standard Pipe Tapered CAD Computer Aided Design FOV Field of View FMCW Frequency-Modulated Continuous-Wave GTA Global Transport Application IR Infrared Radiation LiDAR Light Detecting and Ranging MEMS Micro-Electro-Mechanical-System PA Polyamide PC Polycarbonate R&D Research and Development SLS Selective Laser Sintering ix x Contents Abstract vi Acknowledgements viii Abbrevations ix 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.4 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.5 Report Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Methodology 5 2.1 Workflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.3 Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3 Theory 11 3.1 Levels of Driving Automation . . . . . . . . . . . . . . . . . . . . . . 11 3.2 LiDAR Sensor System . . . . . . . . . . . . . . . . . . . . . . . . . . 12 3.3 Cleaning Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 3.4 Fluid Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.5 Spray Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.6 Laws and Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.7 Polyamide and Selective Laser Sintering . . . . . . . . . . . . . . . . 17 4 Problem and Requirements Analysis 19 4.1 Market Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.2 Project Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.3 Description of the LiDAR System . . . . . . . . . . . . . . . . . . . . 22 4.4 Specification of Requirements . . . . . . . . . . . . . . . . . . . . . . 28 5 Concept Development 31 5.1 Concept Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 5.2 Concept Evaluation and Selection . . . . . . . . . . . . . . . . . . . . 35 xi Contents 6 Design of Cleaning Head 41 6.1 Design of Fluid Distribution . . . . . . . . . . . . . . . . . . . . . . . 41 6.2 Selection of Nozzles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 6.3 Spray and Blow Coverage . . . . . . . . . . . . . . . . . . . . . . . . 47 6.4 Variants of the Prototype . . . . . . . . . . . . . . . . . . . . . . . . 49 7 Testing and Evaluation 53 7.1 Test Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 7.2 Leak Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 7.3 Pressure Drop Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 7.4 Volumetric Water Flow Rate Test . . . . . . . . . . . . . . . . . . . . 60 7.5 Performance Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 8 Design Recommendations 75 8.1 Design Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 8.2 Final Design Proposal . . . . . . . . . . . . . . . . . . . . . . . . . . 77 9 Discussion 79 9.1 Remarks on Workflow . . . . . . . . . . . . . . . . . . . . . . . . . . 79 9.2 Remarks on Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 9.3 Remarks on Testing and Results . . . . . . . . . . . . . . . . . . . . . 80 9.4 Relevance of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . 81 9.5 Ethical, Economical and Societal Aspects . . . . . . . . . . . . . . . . 81 10 Conclusion 83 Bibliography 85 A Factors Affecting Cleaning I A.1 Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I A.2 Soil Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I A.3 Soil Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I A.4 Mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II A.5 Rinsing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II B Company Guidelines III B.1 Transport Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III B.2 Vehicle Utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . III B.3 Operating Environment . . . . . . . . . . . . . . . . . . . . . . . . . III C Specification of Requirements V D Function Means Tree VII E Morphological Matrix IX F Testing and Evaluation XIII xii Contents G Stuctural Analysis of Design Proposal XV xiii Contents xiv 1 Introduction This chapter aims to provide the reader of this report with an understanding of the background to the problems creating a need for this project. The chapter also aims to provide the reader with basic knowledge and understanding of the company and the sub-division of the company where the thesis work is done. Sections with accounts of the aim and delimitations of the thesis work follows. The chapter concludes with an outline of the report. 1.1 Background Volvo Group is a leading manufacturer of trucks, buses, construction equipment and engines and its headquarters is located in Gothenburg. The company was founded in 1927, is active in more than 190 countries and has more than 100 000 employees. Furthermore, Volvo Group is comprised of nine business areas. Four of which are re- lated to trucks; Volvo Trucks, UD Trucks and JVs, Renault Trucks and Mack Trucks. The different brands allow the Volvo Group to better match different customer and market segments globally. The remaining business areas are; Volvo Construction Equipment, Volvo Buses, Volvo Penta, Volvo Financial Services and Arquus. Special Vehicles Development in Gothenburg is a department within Volvo Trucks. The department’s mission is to develop customized vehicles to customers where the number of vehicles is at least multiple and about 200 at the most. The projects generally last between a few months up to two years. It is together with Special Vehicles Development that the thesis work is conducted. Volvo Group is making great efforts to develop the transport solutions of tomorrow, with significant resources invested in R&D. New solutions are needed to meet the changing and growing demands on transport from a wide range of industries. Volvo Group realizes the potential benefits of automated systems within the transport in- dustry with benefits such as increased resource and flow efficiency. Autonomous vehicles have the potential to increase productivity and flexibility in a sustain- able manner by simultaneously optimizing traffic management and energy efficiency. Therefore, it is of great importance for Volvo Group to continue the development of autonomous vehicles to offer competitive solutions and bring the benefits of au- tomation to the market. 1 1. Introduction Automation in commercial vehicles is forecasted to increase in the future. Both Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD) will have perception sensors as input for the automation system. A range of different sensor technologies is used, such as cameras, radar and LiDAR to collect needed data. Autonomous vehicles are classified based on their level of automation, from level 0 to level 5. Volvo Trucks will in 2019 deliver their first commercial autonomous vehicles to a limestone mine in Norway. The vehicles are fully automated. However, the transports are taking place within a limited area resulting in the trucks having automation level 4 [1], which means that the vehicles are capable to handle a defined use case without requiring human interaction . Autonomous vehicles often use a multitude of sensors with different functions to perceive and react to the surrounding environment. Regardless of the sensor type, when installed on the vehicle it will be affected by the surrounding environment. To ensure the required level of performance from the vehicle, the sensors need its working conditions to meet certain requirements. A major contributor to the deteri- oration of the working conditions, followed by a decrease in the sensor’s performance, is external factors such as dirt and/or rain- or snowfall. These external factors can either obscure sensors’ intended field of view or inhibit signal inputs and/or outputs needed for the sensor to function satisfyingly. To, from a societal perspective, offer a safe autonomous solution the system needs to be redundant to the degree that the system allows for certain lowering in performance for periods of time. To reach desired levels of automation the system needs to have the capability to manage the disturbing external factors and re-establish working conditions that enable desired system performance. 1.2 Problem Description The trucks used at the limestone mine use, among other perception devices, four 2D LiDAR sensors for positioning of the truck to its surroundings. The harsh conditions at the mine, predominantly a dusty and dirty environment, leads to the need for regular cleaning of the sensors. Volvo Group sees the potential in an improved cleaning system, especially regarding performance and reliability, that could better withstand the conditions of the site to offer a competitive autonomous solution. Volvo has stated that improved perfor- mance regarding the release of the fluids used to clean the LiDAR will enhance the overall system competitiveness. Since the truck has constrained space for fluids, the amount of cleaning fluid that is brought is limited. Therefore, the regular cleaning of the LiDARs needs to be very efficient and optimized to achieve the needed perfor- mance. The timing of the release, the direction of release and the volume of fluids released is of significance to improve the solution. Excessive release of fluids leads to premature emptying of the tank holding the cleaning liquid, which is filled before every shift, leaving the system unable to perform further cleaning cycles during that shift. Improvement in cleaning performance and reliability would yield a system with better up-time and better cost-efficiency. 2 1. Introduction 1.3 Purpose The purpose of this thesis work is to develop a functioning and reliable sensor clean- ing system compatible with the 2D LiDAR sensors and the truck model used at the limestone mine. Achievement of this purpose involves a design meeting the require- ments set on the cleaning system due to the site-specific conditions present at the mine. The developed cleaning system must have the ability to clean the sensors entire Field of View (FOV). It should also fit within the geometrical and architec- tural limitations of the truck while fulfilling requirements set by Volvo and current legislation. The system must be able to provide cleaning for the four considered 2D LiDAR sensors without the need to take the truck out of service during a shift, under reasonable external conditions. Consequently, it is of interest to analyze the basic and specific parameters regarding what affects the trucks sensor vision, such as frequency and duration. This will enable reasonable optimization of the resources available during a shift for cleaning the sensors. Another criterion is that the sug- gested design should be of value for Volvo’s future work. The design should exhibit sufficient flexibility allowing the entire system or specific components to be adjusted for other applications. It should also be designed to facilitate easier production due to potential future increases in production volume. 1.4 Delimitations To encapsulate the core problem of the thesis and keep complexity and workload at manageable levels, the following limitations will be made. The system will be designed for the road and environmental conditions present on the site of the project. This delimitation includes that meeting the specific legal requirements for transportation within the mining area will be of concern. However, the development process are to be carried out with the multitude of environmen- tal scenarios in mind enabling the knowledge and the certain design aspects to be transferred to designs for other environments. The thesis will regard a sensor cleaning system for the truck model currently used on the mining site. Since the project is ongoing at Volvo and design changes continues to occur, the design will be done for the truck layout presented to the thesis workers no later than 2019-02-28. The thesis workers have the authority to disregard changes by Volvo done at a later date, if they are not easily implemented in the design. The design will mainly regard the hardware needed to achieve a functional cleaning system. The required system control and potential software will be checked for fea- sibility, but not developed within the thesis. Also an in-depth investigation chemical compositions of cleaning liquids will not be a part in this thesis. 3 1. Introduction The work focuses on developing the functionality of the system, leading to optimiza- tion of system performance. Cost and environmental impact will be considered as having lesser importance for the purpose of this project. To simplify data collection and validation, artificial conditions similar to that of the mining site will be accepted for testing. 1.5 Report Outline The report will follow the process of developing a prototype for a LiDAR cleaning system. Initially it will give a brief introduction to the core problem and the purpose of this project allowing the reader to gain understanding about the relevance of the project. This is accompanied with limitations set upon the thesis from the thesis workers and Volvo. The introduction is followed by a description of the methods used to complete the tasks in the project. The method chapter is followed by a theoretical chapter explaining the supporting systems, technologies and the requirements of the project. After the theoretical chapter follows a chapter of problem and requirement analysis. The report is then continued with a thorough walk-through of the concept development phase of the project, including results from the different development phases. The concept development chapter is followed by a chapter regarding the testing phase of the project, presenting both execution and results. Following the chapter on testing is a chapter presenting the final design proposal. This chapter gives a description of the final design, with potential design changes from the testing phase taken into account. The following chapter contains a discussion of the product in general. This regards the results throughout the project and factors affecting the results of the project. The last chapter contains a conclusion of the project, with final remarks of the project and its results, as well as guideline for Volvo’s future development of sensor cleaning systems for applications similar to the one in focus of this project. 4 2 Methodology The following chapter aims to provide the reader with insight into the intended work- flow as well as the theoretical and practical methods used within the thesis work. The chapter also provides the reader with information about the tools intended to be used during the thesis work, both in the form of software and hardware. 2.1 Workflow The workflow of the thesis will use the Generic Product Development Process as a guideline and the steps taken within each phase will take inspiration from literature from Ulrich and Eppinger [2]. Within this thesis, the workflow is planned to end before Production Ramp-up since the production of the system is not considered in this thesis. The phases within this thesis are described in the sections below. 2.1.1 Planning The planning phase will principally consist of two components, research and resource allocation. The research is done to provide the group with sufficient knowledge about the technologies involved in the core problem. Surrounding technologies will be re- searched if considered relevant. The knowledge gathering will come mainly from researching current solutions on the market, searching for limiting or beneficial IP from patents etc. and documenting information from knowledgeable resources in- house. The knowledge of in-house sources will be used for help in deciding the tools to be utilized during the design process. Some steps in the planning phase in this project usually occur during the Concept Generation-phase, but are moved to the planning phase. This change is done be- cause knowledge about the system usually gained from steps taken in the Concept Generation-phase is needed to plan the work efficiently. 2.1.2 Concept Development In the Concept Development-phase, the knowledge gained during the planning phase will be used to create a functional decomposition to break down and make the prob- lem manageable. After that, potential solutions are generated to the sub-problems on an aggregated level. These solutions will be combined into conceptual solutions 5 2. Methodology and checked both against stated requirements and investigated for feasibility. Dur- ing this phase, concepts will be given an industrial design and one or many feasible concepts meeting the requirements will go through testing for proof of concept. Data gathered from the testing will be used for the final selection of the concept, and the most promising concept will advance for further development. 2.1.3 System-Level Design From the findings in the Concept Development-phase, system and product archi- tectures will be developed. This includes defining necessary interfaces and the sub- systems of the product. During the architectural design of the system-level design phase, close contact with affected members of the project group at Volvo is required. This is due to the implications the architectural design has on the cleaning system and other systems affected by the cleaning system. During this phase, refinement of the concept design and preliminary component engineering will be done. 2.1.4 Detail Design The next step will be to finalize the design of the concept. The component engi- neering will be concluded by determining the final geometry and dimensions of the components. The components will be defined and modeled in CAD software. Simul- taneously, the modeled components are to be assembled as assemblies and mounted to the truck model to geometrically assure compatibility and functionality. It is also during this phase that the material selection is planned to occur. Furthermore, standard parts from suppliers that will be used are evaluated and selected. 2.1.5 Testing and Evaluation The detailed design will be transformed into a prototype during the exploratory Testing and Evaluation-phase. The phase will begin by designing a test plan to guide the process of testing the prototype. The test plan states the test methods that will be practiced. The focus will be on key requirements together with the overall performance. Finally, the results from the testing will be used as a basis for the evaluation of the design. The evaluation will be used to give recommendations for design changes for a final design proposal. 2.2 Methods In this section, selected methods that will be used in the thesis are described in greater detail. In the description, both the approach and reasoning for achieving reliable results are presented. 2.2.1 Information Search and Analysis When collecting the information, sources will be both primary and secondary to com- bine the two complementary ways of gathering the information. In the thesis work, 6 2. Methodology mainly experts, suppliers and customers will be interviewed according to strategies described by Denscombe [3]. Interviews will be held one-to-one and in groups de- pending on what is possible to undertake and the level of discussion wanted to be achieved. The interviews will be held in a semi-structured fashion with questions prepared in advance. The approach ensures a focus on topics while allowing for flexibility and further probing. Mediating tools in the form of sketches, pictures, CAD-models and prototypes will be used to express questions and describe prob- lems as well as support the evaluation of solutions and requirements. By preparing questions and bringing mediating tools, the interviews will be more productive and increase the knowledge gained. It will also ease the documentation during the inter- views and make sure that the prepared topics are covered. Before each interview, the interviewees are contacted for an agreement about the time, place and duration of the interview to increase the convenience for both parties and ensure enough time is set aside. Documenting the answers along with asking confirming questions will aid in avoiding recalling false statements, reducing errors and bias handling of the information. The process of secondary information search contains five steps starting with setting the scope of the search. After that, the search will be planned by preparing working documents as well as identifying information sources, search terms, questions and organizations of high relevance. When the plan is completed, the systematic search will take place to thoroughly search for information within the set scope. The information gathered will then be screened to reduce the data and determine the level of quality and relevance. Lastly, the remaining information will be categorized and summarized to do a final assessment. Emerging questions and gaps in knowledge that might still exist will lead to additional information searches and analysis. By redoing the process, knowledge from previous searches and progress in the project can be used to improve the process. 2.2.2 Patent Search The patent search is based on the approach established by Haldorson [4]. To begin with, the scope with the aim of the patent search and its confines will be determined. This is followed by designing the patent search and accordingly identify classes, assignees and keywords to limit and optimize the search. The keywords are generated by answering a few questions stated about the invention. In the answers, describing words and corresponding synonyms are then selected as keywords. The keywords are used to create search strings using operators that will be used together with the relevant classes identified. The patent search based on these premises is to be carried out to find patents of relevance. The relevant patents will be analyzed and the process documented. The main patent database to acquire the information will be Espacenet, to account for patents in many countries. 7 2. Methodology 2.2.3 Specification of Requirements To visualize the functional criteria for the intended product a Specification of Re- quirements will be constructed. They will be categorized as requirements and de- sires, to provide information about the importance of each criteria. The require- ments will be given a weight to further provide information about the importance of achieving the criteria. Each criteria is given a target value, visualizing the functional ability needed for the product to meet the criteria. Each criteria is given individual means to evaluate/verify the performance and a justification of the need for the individual criteria. The specification of requirements will be utilized to coordinate the project and will be set early in the development of the product. Due to the build-up of knowledge and the potential change of requirements during the project the specification will be open for change during the project. The use of the specification in this thesis is due to its benefits in communicating targets in the present context [5]. 2.2.4 Function-Means Modeling To break down the functional requirements of the product designed in the thesis work and to facilitate the generation of sub-solutions, Function-Means Modelling will be done with the help of a Function-Means tree. This is done in a top-down approach through hierarchical levels by first stating the functional requirement on the highest level, followed by its possible solution or design parameter. There is a possibility that one functional requirement has several solutions. In a concept generation phase, it is beneficial to generate as many possible solutions to a functional requirement as possible. After the generation of solutions is done, the best sub-solutions for solving the functional requirement are combined. After a combination is chosen, the process is continued by generating sub-solutions on a lower level to the functional requirements that arise from the sub-solutions on the level previously handled [6]. Within the thesis work, certain limitations were set upon the solution, limiting the design space. This made certain parts of the tree to be out of scope and the approach of iterating the Function-Means Tree unnecessary. Instead, the tree was used as a solution generation and visualizing tool where the results were transferred into a Morphological Matrix. The transfer is done since the Morphological Matrix provides a better overview of the different solutions path of sub-functions and sub- solutions. 2.2.5 Morphological Matrix A morphological matrix was used since it provides a structured way of combining solutions to sub-systems in the overall solution to the problem at hand. It also provides a scheme of tested combinations, lowering the risk of unnecessary work being done due to repetitive evaluation of similar concepts. In combination with a thorough evaluation, the morphological matrix can provide knowledge of a sub- solutions ability to stand alone or which other functions the specific sub-solution is 8 2. Methodology dependent on. This information can prove vital in understanding the design space of the system. 2.2.6 Pugh Matrix The Pugh Selection matrix, often called Pugh Matrix, was used in the concept screening phase of the thesis. The matrix is commonly used in an iterative manner evaluating generated concepts, in this thesis concepts from the Morphological ma- trix. The Pugh matrix evaluates concepts on criteria formulated from information gathered in the research of the problem. The concepts will be evaluated against a reference concept, based on if it is better, equal or worse than the reference. The matrix can also be designed with weights, giving certain criteria a bigger impact on the screening [2]. In the thesis, no numerical weighting was given to the criteria. Since one category of criteria was the most important for the solution, many criteria in that category existed in the matrix. This worked as a weighting of the matrix, giving concepts excelling in criteria of that category a better potential for a high score. 2.3 Tools The tools that will be used to accomplish the thesis work is presented in this section. 2.3.1 PTC Creo At the location where the thesis work is performed, the CAD-software Creo is used when 3D modeling everything but the cab of the truck. Since the system of interest for this thesis is fitted to the chassis, Creo will be used when modeling for com- patibility reasons. Creo also allows for lighter simulations of the model for proof of concept. 2.3.2 ANSYS To provide a baseline for design decisions and to speed up the testing phase of the project, the software package ANSYS 19.2 will be used during the project. For static structural analysis Workbench 19.2 will be used. The structural analysis involves mainly stress and deflection analysis. For fluid analysis, AIM 19.2 will be used. The fluid analysis involves mainly getting initial information about high and low pressure points in regions with fluid-structural interaction as well as different fluid profiles. 2.3.3 KOLA Konstruktionsdata Lastvagnar (KOLA) is a Product Data Management (PDM) Sys- tem utilized by the Volvo Group. The system is foremost used in product devel- opment to document, manage and access information regarding variants and items 9 2. Methodology of the trucks. The PDM system will help to collect and work with documents containing information of the truck on which the solution will be installed on. 2.3.4 Teamplace Teamplace will be used as the online storage of all working documents in the the- sis. The setup enables the sharing of documents between the two thesis workers and others involved in the project. Also, most documents related to the project is accessible in Teamplace. 10 3 Theory This chapter aims to provide the reader with appropriate information about the tech- nologies relevant to the thesis. The chapter starts with theory regarding different levels of driving automation. A description of LiDAR sensor systems follows. The theoretical description of the aspects of cleaning is also investigated. Relevant as- pects of fluid mechanics, followed by theory regarding spray nozzles are also exam- ined. This is followed by a section regarding relevant laws and regulations affecting the product. The chapter concludes with descriptions of the material and technology used for prototyping of the product. 3.1 Levels of Driving Automation The American National Standards Institute manages a standard that recommends practices regarding driving automation referred to as SAE J 3016-2018 The stan- dard is also known as Taxonomy and Definitions for Terms Related to Driving Au- tomation Systems for On-Road Motor Vehicles. The document presents six levels of automation, commonly known as SAE levels in the industry. The levels are presented and described below [7]: Level 0 - No Driving Automation Level 1 - Driver Assistance Level 2 - Partial Driving Automation Level 3 - Conditional Driving Automation Level 4 - High Driving Automation Level 5 - Full Driving Automation The levels of automation gradually increase with every level [7]. In the very first level, there is no driver assistance. For both level 1 and level 2, ADAS support the human driver who still has to monitor the driving environment. The automated systems of a vehicle with Level 3 only needs the human driver to be prepared and intervene when alerted. In level 4, the Automated Driving System (ADS) is capable of driving the vehicle in certain conditions without interventions from a human. Fi- nally, level 5 corresponds to full driving automation that is not confined to certain conditions [1]. In the case of cars, the most advanced driving automation systems available reach automation level 2. This includes systems like the Tesla Autopilot and Volvo Pilot Assist II. To achieve this level of autonomy, multiple systems that use perception sensors are combined. Example of assisting systems are adaptive cruise control and 11 3. Theory lane assist [8]. 3.2 LiDAR Sensor System Light Detecting and Ranging (LiDAR) is a system that uses active remote sensing to perceive and map its surroundings. The light that is generated and emitted as pulses is in the form of light amplified by stimulated emission of radiation (laser). The emitted laser pulse reflects on a surface and the time it takes until the laser pulse returns to the LiDAR is registered [9]. There are a few different ways of measuring the distance to the objects on which the light is reflected. Time-of-flight refers to the most common technique where the time for the laser pulse to reach the object, reflect and return is measured [10]. The distance to the surface can after that be calculated with the knowledge that the laser pulse travels at the speed of light [9]. The formula to calculate the distance is the following [11]: Distance = Speed of Light × Time - of -Flight 2 (3.1) There are four main approaches to directing the laser beam of the LiDAR [10]. A spinning LiDAR directs the laser beam by having the sensor system rotate around its axis. The fact that the LiDAR can rotate continuously results in coverage of 360◦ [10]. Mechanical scanning LiDAR redirects the laser beam by using a Micro-Electro- Mechanical-System (MEMS). The system revolves around a very small mirror that redirects the laser’s beams in the wanted direction. The small size of the mirror enables rapid movements and consequently fast scanning of an area. The redirect- ing functionality allows for scanning patterns where certain areas are scanned with higher concentration. This, in turn, facilitates the identification of objects in the area that might be problematic to detect due to size or distance [12]. Optical phased array LiDAR directs the laser beam by adjusting the phasing be- tween the arrayed lasers. This means that the lasers’ phase relative to each other, changes the direction of the beam without any moving parts [10]. Flash LiDAR diffuses the laser beam to produce a wider scan, like a flash of the scene. The LiDAR is then equipped with multiple sensors to capture the returning laser beams [10]. 3.3 Cleaning Parameters The result from a cleaning process is mainly dependent on four parameters affecting the result of the process [13], see Figure 3.1. The parameters are time, action, temperature and concentration. Considering the case of equal occurrence of soiling and use of the same cleaning agent, these parameters can be altered while still yielding the same result as an increase in one parameter can compensate for the 12 3. Theory decrease of another. Other factors exists, which influence the required level of the parameters, such as surface, soil levels etc. [13]. A description of the parameters is found in the sections below. A description of the factors affecting cleaning can be found in Appendix A. Figure 3.1: The four main cleaning parameters. 3.3.1 Time A cleaning process yields better results if the soil experiences long exposure time to the action, temperature and cleaning agent.[13] But in some cases a long exposure time can be problematic. If the area being cleaned is exposed to cleaning for ex- tended periods of time, the mechanical force, chemical agent or high temperature can impose damage on the area being cleaned. In many applications, such as the one present in this project, time is restricted due to the need for low downtime of the application being cleaned during the cleaning process. 3.3.2 Action Action is the shear force acting on the soiled surface, such as a brush, a water jet or an air jet. Higher levels of force result in a higher probability of breaking the bond between the soil and the surface being cleaned. Higher forces also allow for the potential residue of soil, water or cleaning liquid to be removed from the surface [13]. One form of action is the impact generated by a nozzle that for instance spray water, see Section 3.5.3. 3.3.3 Temperature Increased temperature has a positive effect on the cleaning process. An increase in temperature has effects such as higher solubility of the cleaning agent and faster reaction between the cleaning agent and the soil. High temperatures also decrease the viscosity of the soil, allowing for better penetration of the cleaning agent and re- moval from the applied force. Increased temperature can have negative effects on the cleaning process with an increased rate of corrosion of components and evaporative losses in the application of water and cleaning agent [13]. 13 3. Theory 3.3.4 Concentration Concentration refers to the concentration of a cleaning agent used during the clean- ing process. Generally higher levels of concentration improve the result of the clean- ing process as the cleaning agent aids in lowering the surface tension of the soiled surface. The environmental and safety constraints are the main limiter of the level of concentration [13]. 3.4 Fluid Mechanics The following section will describe certain relevant theory on fluid mechanics used during the thesis. The theory was used during the design of the product as well as for flow calculations and understanding of certain behaviour during testing. 3.4.1 Flow Rate The volumetric flow rate is the volume that flows through a cross-sectional area over time. The flow rate can also be calculated using the velocity of the fluid. The velocity is the length of the fluid divided by the time, a term that is given when referring to the volume as the cross-sectional area multiplied by the distance [14]. Q = V t = A × v (3.2) Q = The volumetric flow rate V = Volume of the fluid portion t = The time it takes the fluid portion to flow through its length A = Cross-sectional area v = Velocity of the fluid at the section The flow rate of a nozzle at a specific pressure can be obtained if the flow rate of the nozzle at another pressure is known, see Equation 3.3. As the difference between the pressures P1 and P2 increases, the deviation from the theoretical flow rate increases due to not accounting for several aspects. For instance, the equation only regards laminar flows, whereas turbulent flows most likely are present. Also, with increased velocity, friction losses increase. Furthermore, the amount of energy that is used to achieve a certain spray angle and pattern differs between nozzles [15]. Q2 = Q1 × √ P2 P1 (3.3) Q1 = The nozzle flow rate at pressure P1 Q2 = The nozzle flow rate at pressure P2 P1,P2 = Nozzle pressure energy 3.4.2 Losses Losses occur in pipe systems, and parts of those losses are categorized as friction losses and minor losses. Friction losses are mainly a function of the system geome- 14 3. Theory try, fluid properties and flow rate. Important factors increasing friction losses in a pipe system is long pipe lengths, small diameters of the pipe, high flow rates, high relative roughness of the pipe, the pipe cross-section and the Reynolds number [16]. Minor losses often contribute to a large portion of losses in a system. They occur at pipe entrances and pipe exits, gradual and sudden contractions or expansions of the pipe and different fittings and valves. The minor losses are summarized in a system and may account for a larger pressure loss than long pipes [17]. 3.5 Spray Nozzle Spray nozzles, also referred to as water nozzles, function as a way to increase the speed of a fluid and break it into droplets propelled by pressure energy. The process comprises of two steps, separating the fluid into droplets and directing the fluid. There are several applications of spray nozzles, among them is washing [15]. 3.5.1 Spray Pattern The choice of water nozzle design influences how the water is distributed by changing the characteristics of the spray. There are different designs of nozzles to achieve the desired performance for a specific application. Typical spray patterns produced by different nozzle designs are shown in Figure 3.2 [18]. Figure 3.2: Spray patterns produced by different spray nozzles. To obtain a Full Cone spray pattern, the water is distributed very evenly in droplets with spray angles between 30◦-170◦. The Hollow Cone spray pattern is in many re- gards similar to the Full Cone spray pattern, although the distribution is condensed around the cone’s circumference. The Flat Fan spray pattern concentrates the dis- tribution of water in a line with spray angles between 15◦-145◦. The spray pattern with the highest concentration of the water distributed is the Solid Stream. In con- trast, the mist spray pattern is created shortly after the water departs from the nozzle and forms a uniform mist with little to no force [18]. Trigonometry is used to compute the theoretical spray pattern and thereby the surface area of the spray pattern. The calculated theoretical spray pattern is less accurate further away from the nozzle as the spray is affected by several factors. 15 3. Theory The actual spray pattern caused by the following factors; gravity, viscosity, flow rate, pressure and nozzle design has a smaller spray coverage [18]. 3.5.2 Droplet Size There is a relationship between droplet size and the surface area of the spray. By doubling the surface area, the mean droplet size is reduced to half. Hence, a wider spray angle results in smaller droplets, partly because there is less chance of the droplets recombining. A Hollow Cone nozzle produces the smallest droplets while a Solid Stream nozzle does not separate into droplets. In cases of moving flows in the proximity to the spray, such as wind, smaller droplets are more affected and might be redirected and miss the targeted area, called spray drift [18]. In general, higher pressure creates smaller droplets by atomizing the spray. The same is true for lower flow rates if the pressure remains the same [19]. 3.5.3 Impact The impact of a spray, the measure of impact force divided by the surface area [19], is important when it comes to cleaning. An increased pressure affects the impact by increasing the internal energy of the fluid. The type of nozzle alter the proportions of energy that is used to atomize the spray and increase the impact of the spray. Consequently, a nozzle that produces a Solid Stream is very efficient in transferring the internal energy to impact force as it does not atomize the fluid [18]. The impact increases at the same rate as the pressure when ensuring that the flow rate remains the same. It is also possible to increase the impact, as long as the pressure is kept, by increasing the flow rate [19]. Flat Fan and Solid Stream spray pattern is often utilized in washing operations to ensure a high impact to remove residue from a surface. These spray patterns are however limited to the area that the spray is distributed and require motion to clean an area [20]. As the relation of impact force and surface area state, the impact is reduced with an increase in surface area. By increasing the distance between the nozzle and the surface, the surface area sprayed by the nozzle enlarges leading to reduced impact[19]. 3.6 Laws and Regulations The general maximum allowed total width of a motor vehicle is 2550 mm, including the type N3 used at the mining site [21]. Apart from rear-view mirrors, devices and equipment are generally not allowed to have a total protrusion adding to the width larger than 100 mm. LiDARs are to be considered as watching and detection aids, and are therefore not included in the regulation [22]. The maximum allowed protrusion of mounted equipment in front of the truck in the forward direction of the truck is 250 mm. Units and equipment mounted on the truck, both in the front and in the rear of the vehicles, must not protrude more than 750 mm combined. 16 3. Theory 3.7 Polyamide and Selective Laser Sintering The solid polymer Polyamide (PA) 12, often known as Nylon 12, is referred to as PA 2200 when in powder form. PA 12 is regularly used for rapid functional prototypes and small productions of 300-1000 parts [23]. Table 3.1 presents important material properties regarding PA 12 [24]. Table 3.1: Material properties of PA 12. Measurement Value Unit Standard Density 0, 95 ± 0, 03 g/cm3 Tensile Strength 48 ± 3 MPa DIN EN ISO527 Tensile Modulus 1650 MPa DIN EN ISO527 Heat Deflection Temperature 86 ◦C ASTM D648 @ 1.82MPa Chemical Resistance Yes [25] Selective laser sintering (SLS) is an additive 3D printing technology. The process starts with a thin layer of powder that is applied to the printing surface. The laser selectively heats the desired cross-section of the part to less than the melting point of the material. The heat fuses the particles of the powder. The process is repeated for each layer, forming the complete part layer by layer. SLS can create complex geometries without the use of support structures as the un-sintered powder acts as support during the printing [23]. The residual powder is thereafter removed, unveil- ing the part with a surface texture that is optional to finish [24; 25]. The accuracy of the SLS manufacturing process is ±0, 3 % with the lower limit of ±0, 3 mm [25]. When manufacturing a part in PA 12 with SLS, several guidelines direct and affect the design. The un-sintered powder in internal channels might be difficult to re- move. It is therefore recommended to have a diameter larger than 3 mm for internal channels in the part. For larger wall thicknesses, thicker than 9 mm, it is suggested to hollow out the solid part to counter tendencies of deformation. Warping is an- other kind of deformation that occurs if the part has a flat plane that is too large. Furthermore, wall thickness greater than 1 mm is recommended [26]. 17 3. Theory 18 4 Problem and Requirements Analysis This chapter aims to provide the reader with information about the problems and requirements. The chapter begins with a section containing a market analysis of the market on which the product acts on. The chapter is continued with relevant project and system information enabling a better understanding of future design decisions. Lastly, the specification of requirements is compiled. 4.1 Market Analysis The market for self-driving vehicles is increasing and the number of companies com- peting for market shares in automotive-grade LiDARs is increasing. LiDAR has in later years been a crucial component in self-driving vehicles and experts see LiDARs as being a crucial component in self-driving vehicles in the foreseeable future [12]. As a part of preparing for and aiding the concept generation, a patent search was performed during the planning phase of the project. The reason for the search was to find where the risk of infringements on intellectual property was present and how this affected the design space of the thesis work with respect to its aims. In the patent search, no limiting patents were found that affects the design of the cleaning head. From the market search and patent search is was also evident that few clean- ing systems for sensors exist in automotive applications in general, and for LiDARs in particular. Conceptual designs exist for sensor cleaning systems, predominantly for cars. However, a majority of the cleaning systems on the currently under devel- opment are made for significantly different conditions than the conditions present at the project site. Since no other comparable system could be found on the market the projects current cleaning system will be used for benchmarking during the thesis. 4.2 Project Information During the information search of the thesis, knowledgeable employees at Volvo were contacted and interviewed in different setups, see Section 2.2.1. This was done to get a nuanced view of the problems and requirements. Studies of literature, mostly 19 4. Problem and Requirements Analysis online, were performed to gain information that was not attainable in interviews. Of- ten the two approaches were used as a complement to each other. Certain company guidelines for Volvo’s development process regarding different applications were also investigated to better understand requirements set on the product by factors such as operating environment and transport mission, see Appendix B. 4.2.1 The Mine The mine is an open-pit limestone mine situated in the central part of Norway. The mine extracts limestone that is transported from the extraction point in the open-pit mine through a series of tunnels, see Figure 4.1, to an unloading site on the edge of a fjord. The trucks unload the limestone into a processor that crushes the limestone which is then transported by conveyor belts and loaded onto ships waiting in the fjord, see Figure 4.2. Figure 4.1: Autonomous Volvo truck driving in one of the tunnels.[27] 4.2.2 Volvo’s mission Volvo’s role in the project differs from Volvo’s normal products. Volvo’s normal operations have its foundation in selling trucks to customers and offering after-sale services. In this project Volvo sells a result-oriented service based on a functional result [29], in this case transporting raw material from A to B, instead of selling the 20 4. Problem and Requirements Analysis Figure 4.2: Autonomous Volvo truck offloading at the crusher.[28] product. This provides new challenges for Volvo since aspects of the operation that previously required little or no consideration now gets a high priority. One aspect of this is cutting the costs occurring during the on-site operation. Volvo aspires to do this through automation of transportation, removing cost associated with the drivers and enabling high efficiency through optimized routes and driving behaviour. This operational strategy provides new challenges for Volvo by increasing the requirement on up-time and utilization of the trucks on the site and through that minimizing the cost per unit distance. 4.2.3 Site Conditions The mine is an open-pit limestone mine. The method of extraction, loading, un- loading and transporting of the limestone leads to fine dust being spread into the surrounding environment. The roads at the sites are maintained dirt roads, which especially under dry conditions, leads to increased spreading of dust to the surround- ing environment. The mines geographical position leads to the mining site being classed as having a warm and temperate climate [30]. The average annual temperature is approxi- mately 5.5 ◦C and varies with around 15 ◦C throughout the year. The site has four months with an average temperature below 3 ◦C, meaning that the site experiences perception both as rain and snow. The annual average precipitation is around 1400 mm. 21 4. Problem and Requirements Analysis The external conditions present at the site in combination with the type of roads at the mine leads to the trucks, and also the LiDARs, being exposed to a substantial amount of mud splashes and swirling dust. 4.3 Description of the LiDAR System In the following section, the current cleaning system and supply systems associated with it will be described. A description of LiDAR currently used is also given in the section. 4.3.1 LiDAR Helmet The LiDARs are installed in multi-functional helmets, see Figure 4.3, that are mounted to the truck with bracers. The bracers are fixed to the truck’s chassis to not be exposed to the additional dampening of the cab. To enable the multiple LiDARs’ horizontal scan plane to be aligned, the positioning in the helmet is ad- justable. Furthermore, the interface between the helmet and LiDAR is suspended to counteract minor vibrations. Another important function of the helmet is to protect the LiDAR from precipitation and contamination present during operation. The helmet is therefore designed with protective covers with a narrow horizontal gap to allow the LiDAR to operate. These multiple functions require certain geometrical space that impacts where and how the cleaning system is possible to be integrated. However, the LiDAR helmet is subject to a major re-design concurrently with the thesis work. For that reason, the development of the new cleaning system considers the new helmet when designing. Simultaneously, the new design should also con- sider production aspects to efficiently accommodate a potential future increase in production volume. During operation, the trucks are exposed to vibration due to a combination of speed and the road conditions present at the mine. The vibrations are particularly harmful to parts and systems with a long lever from its center of mass to its fastening point, such as the LiDAR helmet. The long lever increases the moment generated by the acceleration of mass-produced during vibration and by that the stress in affected parts. For this reason, it is beneficial to minimize the weight of the helmet, cleaning system and LiDAR. This becomes evident when investigating an S-N curve, where it can be seen that a reduction in stress amplitude leads to an increasing number of load cycles a component can withstand [31]. 22 4. Problem and Requirements Analysis Figure 4.3: Current LiDAR helmet installed in the project. The picture is cropped.[32] 4.3.2 LiDAR There are several 2D LiDARs mounted on the autonomous trucks at the mining site of the model SICK LMS111-10100, see Figure 4.4. SICK’s 2D LiDARs redirect the laser with a mirror that rotates [33]. One LiDAR is placed on each corner of the truck’s cab. The horizontal angle of the scanning range is 270◦ [34], see Figure 4.5. The 90◦ that is out of the scanning range faces the cab. 23 4. Problem and Requirements Analysis Figure 4.4: 3D rendering of a SICK LMS111-10100 2D LiDAR. Figure 4.5: The horizontal scanning range angle of the LiDAR. 24 4. Problem and Requirements Analysis The positioning of the LiDARs results in the LiDARs’ FOV overlapping, see Figure 4.6. This allows up to two LiDARs to be blind at any one time while the truck con- tinues to operate. However, there are situations such as both front LiDARs being blind that limits the FOV too much and thereby require cleaning for the vehicle to be operational. Figure 4.6: The positioning of the four LiDARs and their FOV overlapping The LiDAR of interest emitts a short pulse of laser light that diverge with an angle of around 0, 43◦, see Figure 4.7. The size of the beam diameter is calculated with the following formula [34]: Beam diameter = (distance [mm] × 0, 015rad) + 8mm (4.1) Moreover, Figure 4.7 also illustrates that the reflected beam requires a vertical gap 25 4. Problem and Requirements Analysis of a minimum of 15 mm with a horizontal gap of a maximum of 200 mm from the center axis. This is required for the protective helmet not to interfere and block the receiving pulse. Not following these guidelines might drastically reduce the ability of the LiDAR to acquire accurate results. The same is true for surfaces that reflect sun glare at the LiDAR optics cover causing blind spots [34]. Figure 4.7: The divergence of the laser beam is visualized to the left. The minimum vertical gap at the maximum horizontal gap for the receiving laser beam is visualized to the right. The LiDAR has a built-in solution to determine if and how obstructed the optics cover is. Below the LiDAR’s optics cover there is a shelf in which seven IR-sensors are located, positioned 38, 6◦ apart from each other in a radial pattern, see Fig- ure 4.8. These seven IR-sensors send beams through the optics cover to thereafter be reflected to receivers that determine the level of cleanliness of the optics cover [34]. The optics cover is made out of Polycarbonate (PC) with a surface area of 6, 48249× 10−3m2. The optics cover is protected with a PHC 587 silicon coating and additional additives to achieve both protective and hydrophobic properties [34]. According to the technical data sheet of a comparable coating, it is resistant to chemicals and solvents such as oils, fuels and washer fluids. The protective layer is also abrasion- resistant and mar-resistant to withstand for instance sand and dirt [35]. However, the operating instructions of the LiDAR state that the maintenance in the form of cleaning the optics cover should be cleaned in a certain manner. No aggressive detergents or abrasive cleaning agents should be applied when cleaning. Also, the manufacturer recommends a soft brush and damp lens cloth to remove dust and wipe the optics cover clean [34]. 26 4. Problem and Requirements Analysis Figure 4.8: Description of the 2D LiDAR SICK LMS111-10100 with important features and dimensions indicated. 4.3.3 Current Cleaning System The project is the first project of its kind that Volvo has undertaken at this scale. It is case-specific, leading to very specific requirements on the truck and its systems. The specific nature and novelty of the project creates a lack of set requirements since there are limited prior knowledge about the application and no comparable products to benchmark against. The truck has a limited amount of cleaning liquid leading to Volvo wanting an improvement of the fluid distribution in the new solution in comparison with the current solution. The truck carries one 25 liter tank carrying the cleaning liquid while the compressed air is generated and stored in a separate tank. To control the distribution of each fluid, solenoid valves are installed along with each hose close to the cleaning system of each LiDAR. The hoses of the two fluids are merged directly after the solenoid valves. Consequently, one hose is connected to the single inlet of the current cleaning system. The current cleaning system directs both the cleaning fluid and the air from below the optics cover of the LiDAR. The fluids are distributed through the same nine holes situated around the LiDAR, except at the back which faces the cab. The hole design creates a spray pattern similar to the Solid Stream, see Section 3.5.1. 4.3.4 Air Supply The air in the cleaning system is supplied from the central air system installed on the truck. This system serves several vital systems on the truck, including the brakes and often parts of the suspension. The air pressure is created by a compressor sit- uated close to the engine in the engine bay of the truck. The maximum pressure of the air system is 12, 5 bar. Several vital systems on the truck use air from the same 27 4. Problem and Requirements Analysis supply leading to different systems being given different priority. To be able to uphold vital functions of the truck, such as breaking, the cleaning system is of lower priority. For this reason the cleaning system can only be used when the pressure in the system is above the cut-off level, set at 8, 5 bar. This is a safety feature as a truck must always have the ability to stop. The LiDAR system is designed to allow collaboration between the LiDARs so that not all LiDARs must be clean at all times, see Section 4.3.2. The air is distributed through a regulating block on the frame and distributed to the cleaning system through hoses running along the frame of the truck. 4.3.5 Cleaning Liquid For the washing of the LiDARs, an application-specific tank containing cleaning liquid is installed on the trucks at the mine. The tank can hold 25 liter of liquid and is installed on the frame of the truck. The liquid that is used is a mixture of water and regular washer fluid. The tank is pressurized by compressed air supplied by the main air system, which propels the liquid through hoses to the LiDAR in need of cleaning. The liquid is directed to the correct LiDAR through a distribution block mounted on the frame of the truck. 4.4 Specification of Requirements 4.4.1 Preface to Specification of Requirements Due to the novelty of the cleaning system and its application, Volvo and other stakeholders were shown to have no prior requirements set upon the product devel- oped in the project. However, different tests of the truck at the mining site lead to Volvo having perceptions of potential needs and requirements. The group therefore worked to compile acquired data and information about the involved technologies to translate these into a Specification of Requirements. Included is a limited number of statements gathered from personnel in the project that have been translated into requirements and desires. The novelty of the product causes the acquired specifi- cation to be better used as recommendations for the group to work towards in the development process. Since the requirements were specified throughout the entire duration of the project, the specification was developed over time. New information was constantly acquired and new inputs led to changing perceptions, even to the most important requirements. 4.4.2 Utilization of Specification of Requirements The Specification of Requirements, in its various levels of completeness, was used throughout different stages of the project. It was used for setting the baseline when generating and evaluating concepts. During the design phase, it was used as a check- list to ensure that important design parameters were met and the desired function- ality could be expected from the design. The Specification of Requirements was also 28 4. Problem and Requirements Analysis utilized when designing the tests and analyzing the results. The approach ensured that the tests were designed to acquire answers to the most important knowledge gaps in the design’s functionality. The analysis of the tests was also checked against the Specification of Requirements to map where the design met requirements and where more testing or possible design changes were needed. 4.4.3 Requirement Specification The Specification of Requirements, see Table C.1 in Appendix C, is divided into six main areas. This was done to aid a better overview of fulfillment based on func- tional categories. The Specification of Requirements contains seven columns. First, the criteria order number is stated. Secondly, the criteria is stated followed by a target value in the third column. The fourth column states whether the criteria is a requirement or a desire and the fifth column states if the criteria is a desire, the weight of importance of the desire. The weights are based on a scale of 1 - 5, where 5 denotes the highest rate of importance. The sixth column states the method of evaluation or verification of the criteria. The seventh column states the justification of criteria. From the complete specification of requirements, twelve criteria are regarded as being more important to develop a successful cleaning system, see Table 4.1. At an early stage, the importance of limiting the consumption of liquid was understood. The criteria 1.5 showcase this with the target value of less than 0, 25 liter per cleaning sequence. The amount corresponds to 100 cleaning sequences with a liquid tank of 25 liter. Criteria 1.6 states that the air is limited at times, even though it is generated on the truck. Criteria 1.7 - 1.9 are importance since they are requirements for the LiDAR to function as intended, see Section 4.3.2. The flexibility regarding the time is set with criteria 1.12 and 1.13. Criteria 1.15, 1.18 and 1.19 all focus on the result to be achieved by the cleaning system. Finally, the optics cover of the LiDAR should not be damaged over time, which criteria 2.4 defines. Table 4.1: The twelve most important requirements and desires. 29 4. Problem and Requirements Analysis 30 5 Concept Development This chapter aims to exhibit the progress and results obtained through the concept development phase of the thesis. The concept development phase followed the re- search phase, with the results of each step in the development phase working as the foundation for the next step. Certain steps of the concept development generated a lot of results, such as generated concepts, which will not be fully accounted for in the report. Some results obtained, with informational relevance for the final product, will be used as comparisons to clarify other relevant results and as proof of concept. This is done in an attempt to keep the report as concise and informative as possible for the reader. 5.1 Concept Generation 5.1.1 Functional Mapping To support the concept generation of the LiDAR cleaning system, a functional mapping is constructed with a Functional Means Tree, see Figures D.1 and D.2 in Appendix D. During the mapping, the entire system is considered in order to understand the various functions involved and to not limit the potential solution space. To support the functional mapping online searches and patent searches were performed to expand the number of solutions by taking inspiration from solutions to similar problems. The primary function Clean LiDAR sensor is decomposed into two sub-functions, Remove obstructions and Ease cleaning. Remove obstructions is further decomposed into the sub-functions Water-, Air- and Mechanical systems as means of removing obstructions. Furthermore, as shown in orange in Figure 5.1, seven sub-functions that are in large separated from the supporting systems and consequently more connected to the primary function of Clean LiDAR sensor is ini- tially chosen to be in focus. The focus enables sub-solutions with greater flexibility as they have a lesser impact on other parts of the system. For every sub-function, several sub-solutions is generated as the possible means of solving the function. The seven sub-functions are described below: • Distribute water/air: The fluids are needed to be distributed to the sur- face of the LiDAR optics cover to remove obstructions and thereby clean the surface. • Direct water/air: The fluids can be applied to the LiDAR optics cover from 31 5. Concept Development various directions. Several aspects are affected by the direction of the fluids. The LiDAR optics cover is tilted causing certain angles to be achievable only from one direction. Due to the shape of the LiDAR body, the accessibility around the LiDAR also differ. • Move distributor of water/air: The distributors of the fluids can either be fixated, move in a horizontal or vertical plane or both. • Detach obstructions: With a mechanical system, the obstructions can be detached and removed by physical means. Figure 5.1: Function Means Tree of the focus area containing the seven sub- functions and their potential sub-solutions. Although several sub-functions related to the complete system is excluded at this stage, they are still to be considered throughout the project as they impact designs of the focus area. For instance, the functions of fluids being supplied and transported need to be considered since they act as inputs to the chosen functions. 5.1.2 Generation and Description of Concepts The generated sub-solutions, 27 in total, are arranged in a Morphological Matrix where they correspond to the seven chosen sub-functions, see Figure E.1 in Appendix 32 5. Concept Development E. By combining the different sub-solutions, seven concepts of total solutions are generated. Figures E.2 to E.8 in Appendix E show the selected sub-solutions for each concept. A sketch and a short description of each concept are shown below. The number of concepts generated is sufficient to ensure that most sub-solutions are represented and therefore possible to evaluate further on. Top Plate The concept is positioned above the LiDAR, see Figure 5.2 Hoses distribute the fluids to multiple nozzles that are mounted on a plate, shaped to direct the fluids as desired. Figure 5.2: Sketch of concept Top Plate. Fixed Ring Below The concept is a fixed ring that wraps around the LiDAR to direct the fluids from below, see Figure 5.3. The fluids are distributed in the ring with slots that are connected to multiple nozzles. Figure 5.3: Sketch of concept Fixed Ring Below. 33 5. Concept Development Solid Ring Above The concept is a fixed ring that is positioned above the LiDAR to direct the fluids from above, see Fig- ure 5.4. The water is distributed to multiple noz- zles with slots. However, the air is distributed to an open slot located closer to the LiDAR to create an air knife around the edge of the LiDAR. Figure 5.4: Sketch of concept Solid Ring Above. Spinning Ring The concept is a ring that wraps around the Li- DAR to direct the fluids from below, see Figure 5.5. A few nozzles are mounted to an inner ring with blades that rotates by the flow of the fluids. Figure 5.5: Sketch of concept Spinning Ring. Spinner The concept consists of two arms positioned above the LiDAR, see Figure 5.6. A nozzle to distribute the fluids is mounted at the end of each arm. Fur- thermore, the arms are rotated by the flow of the fluids. Figure 5.6: Sketch of concept Spinner. 34 5. Concept Development Wiper In the concept, a wiper is positioned above the Li- DAR to wipe the optics cover and shelf, see Figure 5.7. A motor at the top rotates the wiper 360◦. Multiple nozzles are mounted to a plate to dis- tribute the fluids from below. Figure 5.7: Sketch of concept Wiper. The Clash The concept’s water distribution is positioned above the LiDAR, similar to concept Top Plate, see Figure 5.8. However, the air distribution is performed from below with separate nozzles that are mounted on a plate. Figure 5.8: Sketch of concept The Clash. 5.2 Concept Evaluation and Selection 5.2.1 Elimination of Concepts The generated concepts are initially screened in a Elimination Matrix to exclude any concept that does not fulfill the criteria set up in the matrix. The eight screening criteria encompass among others; that the concept solves the main problem, fulfills the requirements and is feasible to develop within reasonable cost and within the project scope. Six out of the seven concepts passes the screening and will continue to be further evaluated. However, concept Spinning Ring is eliminated due to not fulfilling all requirements. The concept is designed as a spinning ring that is rotated by the force from the fluid flow. The design is assessed to use high amounts of cleaning fluid to drive the rotating mechanism. Furthermore, the concept deemed as not being space efficient and adding unnecessary complexity that makes it of a high risk of not fulfilling additional requirements and not being feasible. The criteria and result of the screening is shown in Figure 5.9. 35 5. Concept Development Figure 5.9: Elimination Matrix. 5.2.2 Selection of Concept The concepts are evaluated in a Pugh Concept Selection Matrix to compare their performance regarding 18 criteria based on requirements and desires. The criteria are categorized into three groups; Cleaning, Production and Maintenance as well as Miscellaneous, see Figure 5.10. Concept Top Plate is established as the reference in the evaluation. An explanation of each criteria applied in the Pugh Concept Selection Matrix is presented below. • Water and air usage: The amount of fluid that is used during the cleaning process are limited resources. Furthermore, multiple simultaneous LiDAR cleaning is only possible with low resource usage as the supply of fluids by supporting systems is restricted. • Length of cleaning sequence: The time it takes to perform the cleaning. The possibility of a shorter cleaning sequence increases the up-time and flexi- bility of the cleaning and drying. • Risk of additional cleaning sequence: Predictable cleaning performance reduces the risk of additional cleaning. • Clean optics cover and IR-sensors: The coverage of the cleaning to ensure the needed surfaces are cleaned. • Risk of removed obstructions re-emerging: Risks of deposits that con- 36 5. Concept Development tain fluids or dirt that re-emerges. • Risk of damaging optics cover: The optics cover is resistant to a degree, however, there is a risk of abrasion from certain forms of cleaning. • Cleaning reliability: The cleaning system’s reliability to function as de- signed over time due to wear. • Ease of changing parts and system maintenance: The accessibility and ease of changing parts to clean and maintain the cleaning system. • Assembly and installation: The total time of assembly and installation should be kept low. Reduced complexity is achieved with fewer and easier steps during the process of assembly and installation. • Production: The available manufacturing methods affects the flexibility, time and cost of the production. • Number of parts: In general, fewer parts is desired as it reduces complexity and lowers costs during the whole product life cycle. • Environmental robustness: The ability to withstand temperature and pre- cipitation. • Robust against change: The development of LiDAR technology is rapid, changing the dimensions, shapes and performance of LiDARs. Adaptability to these changes is crucial to prevent the cleaning system from becoming obsolete. • Ability to Utilize Volvo’s existing products: Standard parts, especially those already in use at Volvo is preferred as they are tested and proven already. • Product development complexity: The complexity of the product devel- opment is affected by the difficulty of optimizing the concept as well as adding complementary systems. 37 5. Concept Development Figure 5.10: Pugh Concept Selection Matrix. 38 5. Concept Development Table 5.1: List of advantages and disadvantages of the concepts. The evaluation demonstrate that concept Top Plate and Solid Ring Above performed better compared to the other concepts. Both concepts lack any significant weak points, however, they do not distinguish themselves in any specific area of perfor- mance either. The primary reasons for marking the concepts in the Pugh Concept Selection Matrix are compiled as advantages and disadvantages, see Table 5.1. The concepts involving rotating mechanisms, Spinner and Wiper, perform poorly in the evaluation. Their weakness is their inherent complexity and difficulty to at- tain the required reliability. The two concepts are classified as too uncertain and are consequently dismissed as concepts to develop further. To improve the perfor- mance of the remaining concepts, an assessment of is made potential changes to improve the concepts and counteract the disadvantages while enhancing the advan- tages to increase the overall ability of the concept. In general, it was recognized that separating the fluids into different channels is beneficial to achieve the best performance and efficiency of the cleaning. This was due to that in the previous cleaning system, water residuals remained in the channels and were distributed dur- ing the air-phase of the cleaning cycle. This led to a less efficient drying phase in the cleaning sequence. Furthermore, the direction of both fluids is preferred to be coming from above. This ensures the best accessibility of the LiDAR optics cover as well as the shelf on which the IR-sensors measuring the obstruction-levels are located. By having all fluid distribution from above, it is also possible to increase the compactness and the design freedom of the cleaning system. By implementing these changes to all concepts, Fixed Ring Below and The Clash become very similar to the top-performing concepts making them unnecessary. 39 5. Concept Development A combination of Top Plate and Solid Ring Above creates a new concept, Top Plate Block, see Figure 5.11 and E.9 in Appendix E. The air nozzles are mounted on a block that distributes the air in slots while the water is distributed with hoses. An updated version of Solid Ring Above creates another concept, Top Block, see Figure 5.12 and E.10 in Appendix E. In this concept, both types of fluids are distributed in slots making the distribution block to also function as the mounting point for all nozzles. In both new concepts, the air knife sub-solution is kept. However, the function is changed to be accomplished by several nozzles that when placed close together acts as a uniform air knife. Figure 5.11: Sketch of concept Top Plate Block. Figure 5.12: Sketch of concept Top Block. Throughout the concept generation and selection phase, the concepts were modeled in 3D along with the LiDAR helmet and LiDAR to better understand the geomet- rical advantages and disadvantages of each concept. This work led to the group having the ability to in better detail compare concepts and understand where im- provements could be made and how different changes in design parameters affected the surrounding components and systems. During this process, it emerged that Top Plate Block added previously overlooked complexity and required larger geometrical dimensions to be realized. The use of hoses to the nozzles for water distribution positioned with sheet metal parts is the main weakness. To connect all nozzles to hoses, several couplings and adapters are needed. When considering that multiple short hoses, couplings and adapters are needed for each nozzle, the number of in- terfaces increases drastically. It is problematic as the space available is limited and the numerous interfaces reduce reliability and increase the difficulty of assembling. Concept Top Block’s design lacks these shortcomings. The block design functions as a distributor for both water and air as well as mounting points for all nozzles, eliminating the need for unnecessary interfaces. It was decided to not perform an iteration of the Pugh Concept Selection Matrix with the updated and additional concepts. The limitations set on the thesis design space, the need to increase detail in the design of the concepts and the time needed to design and produce the prototype were the primary reasons. Accordingly, concept Top Block is selected as the concept to further develop and evaluate. 40 6 Design of Cleaning Head This chapter aims to exhibit the progress and results obtained through the design development phase of the thesis. It starts with the development of the design that enables fluid distribution. The section is followed by the selection of nozzles and their coverage. The chapter ends with two variants of the prototype being described and evaluated in simulations. 6.1 Design of Fluid Distribution Following the concept selection, compliance between the conceptual cleaning system and the existing systems of the truck were considered in the design of the cleaning head. Extensive contact with suppliers of components during the development en- sures that the cleaning system satisfies the requirements of the external components. This was important in order to achieve full functionality and performance of the out- sourced components. Due to long lead times in contact with suppliers and personnel at Volvo, the decision was taken to deviate from the intended workflow, see Section 2.1. A concurrent engineering approach in the work with System-level Design and Detail Design was taken to speed up the design process. At this stage, the vision was to develop a design that had multiple possibilities of manufacturing. It enables focus on the functionality of the cleaning head while not limiting the possibilities of producing the prototype. The desired functionality of the cleaning head could be achieved by milling the cleaning head in aluminium. Multiple iterations of designs with this manufacturing method in mind were evalu- ated. In this design-phase, the set desires and requirements continued to dictate the design together with further geometrical assurance by 3D modeling. The design of the cleaning head proceeds from concept Top Block. Henceforth, concept Top Block is named Divided Prototype. The cleaning head is divided into two parts and therefore named the Divided Pro- totype, shown in Figure 6.1. In the cleaning head, six positions for water nozzles and seven positions for air nozzles are incorporated, see Figure 6.2. When splitting the block in two, the slots that distribute the fluids are incorporated in one of the parts while the other works as a lid. With milling as the manufacturing method, a cleaning head in a single block is possible to design and manufacture. However, such a distribution block requires the channels to be in the form of drill holes that intersect and are plugged in places. Moreover, the method is limiting when design- 41 6. Design of Cleaning Head ing as the drill holes are straight and interfere with each other causing the block to increase in dimensions. Instead, the size of the cleaning head is minimized by combining drill holes and slots. The material utilization is high and the complex geometry is able to be kept to one of the two parts, partly by placing the fluid inlets at the top. The total height of the cleaning head ads up to 33 mm and the diameter is 184 mm. There are twelve holes in the lid part with corresponding holes threaded for M5 screws in the part with the slots. These screws both join the parts together and allows the cleaning head to be mounted in a test rig. Figure 6.1: Design of the Divided Prototype shown in an exploded view, seen from above. Figure 6.2: Design of the Divided Prototype shown in an exploded view, seen from below. 42 6. Design of Cleaning Head To dimension the two fluid inlets, the expertise of the nozzle manufactures was utilized. By following their recommendations, the inlets were determined to accom- modate couplings of 3/8” (9, 525 mm) for the cleaning fluid and 1/2” (12, 7 mm) for the compressed air. The cross-sectional area for the fluids is maintained in the slots that connect the nozzles in series. The inlets, outlets and the slots of Divided Prototype are visualized in Figure 6.3. Figure 6.3: Visualization of the slots for liquid and air distribution. The liquid is shown in blue (outer slot), while the air is shown in orange (inner slot). The couplings and hoses widely used on Volvo’s trucks are of the metric standard. It was therefore beneficial to use this standard on the cleaning head to the largest extent possible. The inlets were dimensioned to allow for couplings with a male M16. This in turn allowed for couplings that could accommodate hoses with an outer diameter between 8 and 16 mm increasing the flexibility by allowing testing of various hose dimensions, see Figure 6.4. Table 6.1 shows the corresponding inner diameter of the different hoses. The outlets for the nozzles are all female threaded to British Standard Pipe Parallel (BSPP). This enables the outlet to mount both parallel and tapered male threaded nozzles [36]. As the nozzles are mounted with treads, the design had to ensure that the nozzles had enough space to be screwed into the cleaning head. 43 6. Design of Cleaning Head Table 6.1: Dimensions of the different metric hoses. Hose Outer Diameter [mm] Hose Inner Diameter [mm] 8 6 12 9 16 12 Figure 6.4: M16 couplings for the hose sizes of 16, 12 and 8 mm. 6.2 Selection of Nozzles To achieve high efficiency of the fluids distributed by the cleaning head, it is de- cided to use fluid specific nozzles. Air nozzles are used to manage and direct the compressed air by reducing turbulence. The nozzles increase efficiency while also producing an air stream with more force [37]. Silvent 961 air nozzles, see Figure 6.5, are selected to distribute the compressed air. The nozzle produces a flat blow pattern similar to an air knife, especially when positioned in an array with several nozzles close together. The blow angle is 90◦ relative to the male G 1/8” connection [38]. The G thread type is also referred to as BSPP-thread [36]. This enables the nozzles to be mounted closer to the center enabling a more compact design while still achieving the desired blow direction and angle. When mounted, the nozzle protrudes 15, 5 mm from the mounting surface and the width of the strip with the multiple orifices is 23, 9 mm [38]. 44 6. Design of Cleaning Head Figure 6.5: Silvent 961 air nozzle. The maximum operating pressure of the nozzles is 10 bar. The flow rate and blow force at different pressures are plotted in Figure 6.6. The chart shows that the flow rate and blow force increases at the same rate with raised pressure. The distance between the shelf and the nozzles is approximately 50 mm with a lesser distance to the optics cover. The width and height of the blow pattern at a distance of 50 mm is 35 mm and 25 mm respectively, see Figure 6.7 [38]. Figure 6.6: Blow properties of Silvent 961 air nozzle. 45 6. Design of Cleaning Head Figure 6.7: Blow pattern of Silvent 961 air nozzle. Scaling of the distance is not consistent. Full Cone water nozzles are chosen to ensure full wetting of the surface in need of cleaning. The good coverage of Full Cone nozzles is believed to be beneficial because a large surface area is possible to clean without an excessive amount of nozzles or moving mechanics. The selected water spray nozzles are Spraying Systems’ B1/8HH, see Figure 6.8. The nozzle generates a Full Cone pattern with a spray angle of 58◦ at 1, 5 bar and a slightly narrower angle at higher pressures, 53◦ at 6 bar. In contrast to the air nozzle, the water spray nozzle is not angled 90◦ as angled water spray nozzle designs generally are unavailable with low flow rates. The same is true for nozzles with wider spray angles. The flow rate at different pressures is illustrated in Figure 6.9 [39]. The small dimensions, 12, 7 mm in diameter and a protrusion of 15, 6 mm when mounted, give the nozzle a compact design. The male connection is of the thread type R 1/8” [39], also referred to as British Standard Pipe Tapered (BSPT) [36]. Figure 6.8: Spraying Systems’ B1/8HH spray nozzle 46 6. Design of Cleaning Head Figure 6.9: Flow rate at different pressures with Spraying Systems’ B1/8HH spray nozzle. 6.3 Spray and Blow Coverage The water spray nozzles are positioned to obtain the desired coverage with as few nozzles as possible. Both the optics cover and the shelf containing the seven IR- sensors are required to be cleaned. This is achieved by distributing the cleaning fluid from an angle that hits both surfaces and partially neglect the rear facing 90◦ of the optics cover, see Figure 6.10. In order to restrain the usage of cleaning fluid, complete theoretical coverage is not assumed to be optimal. The cleaning liquid will disperse at impact increasing the coverage. In addition, complete theoretical coverage would clean unnecessary surfaces leading to waste in the cleaning process. Figure 6.10: Theoretical spray pattern projected with five B1/8HH nozzles with a spray angle of 53◦, evenly positioned over 216◦. 47 6. Design of Cleaning Head The air knife characteristics of the selected air nozzles will create a precise and powerful air curtain that will act in two stages. Firstly, it will help to remove the wetted dirt on the optics cover. Secondly, it will dry the optics cover from moisture that remains after the dirt is removed. The coverage of the air nozzles is designed to be more extensive than that of the water spray nozzles, drying potential water that flowed towards the back of the LiDAR during the water application of the cleaning sequence. The air nozzles are positioned to blow a curtain of air along with the optics cover and thereby transfer remaining dirt and cleaning fluid in one direction. Furthermore, the air stream is also directed to dry the shelf, see Figure 6.11. Figure 6.11: Theoretical blow pattern projected with six Silvent 961 nozzles, evenly positioned over 270◦. The position of the nozzles in relation to the LiDAR is shown in Figure 6.12. These angles and distances are used as the original placement of the nozzles of which the testing in Section 7.5.1 refers to. The nozzles, which corresponds to the lowest part of the cleaning head, have an approximately distance of 30 mm to the center line of the optics cover. As stated in the Specification of Requirements, criteria 1.8 and 1.9 requires a minimum vertical gap of 15 mm in both directions of the center line. The placement of the cleaning head and its nozzles is well within the requirements. Moreover, the LiDAR’s horizontal FOV is not obstructed which the criteria 1.7 requires. Figure 6.12: The placement of the nozzles’ orifices in relation to the centerline of the optics cover and the angle of the nozzle in relation to the surface of the optics cover. The air and water nozzle are illustrated in orange and blue respectively 48 6. Design of Cleaning Head 6.4 Variants of the Prototype The cost of milling the two parts of the Divided Prototype in aluminium at an ex- ternal manufacturer was 21000 SEK. Furthermore, the manufacturer reported that the design was possible to mill. To enable prototyping with shorter lead times at relatively low cost, in comparison with milling aluminium, SLS 3D printing of the prototype in PA 12 was considered. SLS is frequently used at Volvo for proto- typing components and the printing is available in-house. With the lower costs and increased flexibility, both Divided Prototype and an additional variant could be man- ufactured. The newly established variant called Solid Prototype has five positions for water nozzles and six positions for air nozzles, see Figure 6.13. Figure 6.13: Design of the Solid Prototype, seen from below. The coverage of this configuration is previously illustrated in Section 6.3. Even though the rear facing 90◦ of the optics cover is not used by the LiDAR as the scan- ning range is 270◦, the Divided Prototype has increased coverage due to the extra nozzles. Dirt and cleaning fluid that might gather on this part of the optics cover could possibly be displaced, not least when the vehicle is moving, to a cleaned area. It is a potential trade-off, where better coverage with additional nozzles might solve this issue. However, with higher resource usage as a consequence. There were uncertainties in how to remove the residual powder in the channels. Thereof, only Solid Prototype was 3D printed as one part which its name entails. To further support the investigation of the material and technology, see Section 3.7. To see if it would suffice for testing purposes, structural simulations were performed on the intended design. For the structural simulations the simulation software ANSYS was used, see Section 2.3.2. The simulations were performed as static structural simulations where the planned attachment point of the cleaning head to the helmet was modeled as fixed supports. The internal channels were assigned an internal pressure of 10 bar. The threaded surfaces for the water nozzles are assigned with 49 6. Design of Cleaning Head distributed forces. The forces of 5 N normal to the direction of the water spray simulates the force that the water spray creates. The prototype was given the material properties of PA 12, see 3.1. The outputs of the simulations were total deformation and equivalent stress. The results of deformation simulations for the Solid Prototype can be seen in Figure 6.14 and Figure 6.15. The results of stress simulations for the Solid Prototype can be seen in Figures 6.16 and 6.17. Figure 6.14: Total deformation simulation of Solid Prototype. Seen from above. Figure 6.15: Total deformation simulation of Solid Prototype. Seen from below. The results of the deformation simulation showed that the maximum total deforma- tion of the Solid Prototype is approximately 0, 0242 mm. The deformation occurs at the back of the prototype, probably due to the two fluids flowing from the inlets having a 90◦ to the channels. This deformation is considered to be within the bound- aries of the design. The reason is that the small deformation does not significantly alter the angles of the nozzles in reference to the LiDAR. The result of the stress simulations showed that the maximum stress occurs in the inner channel, shown in Figure 6.16 and Figure 6.17 by the red ”Max” marker. The maximum stress of approximately 4, 85 MPa occurs at the junction between the channels and the outlets. In other parts of the , it can be seen that relatively low stresses occur. The stress levels of the model in relation to the material properties of PA 12, see Table 3.1 in Section 3.7, are considered low. 50 6. Design of Cleaning Head Figure 6.16: Equivalent stress for Solid Prototype. Seen in cross section from above. Figure 6.17: Maximum equivalent stress for Solid Prototype. Seen in cross section from above. From the deformation and equivalent stress simulations, it is concluded that the design of the prototype and the manufacturing method is viable for testing purposes. One of the two variants of the prototype to be further tested, Divided Prototype, is shown in Figures 6.18 and 6.19 assembled with screws, nozzles and couplings. The Solid Prototype was assembled in the same manner for testing. Figure 6.18: Divided Prototype assembled, seen from above. Figure 6.19: Divided Prototype assembled, seen from below. 51 6. Design of Cleaning Head 52 7 Testing and Evaluation This chapter aims to provide the reader with information about the testing of the prototypes within this thesis work. The chapter begins with an explanation of the test setup’s design and limitations. The chapter continues with tests to performed to gain knowledge about the physical prototypes and the test rig. The following sections regard the individual tests and comparisons, providing the reader with in- depth information on the results provided by the tests. 7.1 Test Setup Before starting the tests, a Test Plan was developed to ensure that the wanted aspects were tested and evaluated. The Test Plan supported in organizing the proceedings in the tests and the test process in large. The Test Plan also functioned as support when sourcing material, searching test facility and communicating the planned process to others of interest. 7.1.1 Test Rig Setup The tests within this thesis work were conducted in the dirt laboratory in A-hallen at Volvo GTT in Lundby. To support the prototype and the LiDAR a frame system was built in the laboratory. The intention of the frame system design was to achieve a modular rig, allowing for quick changes in parameter setup while maintaining structural rigidity. The frame system was constructed with aluminium profiles. The slots in the alu- minium profiles allowed for flexible fastening options of components that were to be tested in the rig. The base consisted of a horizontal frame on lockable wheels. This gave the frame system good mobility and spread out the contact points with the ground making the frame system insensitive to horizontal loads. On the horizontal frame, a vertical frame was attached. On the vertical frame, two horizontal frame elements were mounted. These frame elements acted as supports for mounting the cleaning heads and the LiDAR. The elements were individually adjustable allowing for quick changes in height from the floor and the distance between the cleaning head and the LiDAR. The frame system also allowed for fastening of hoses reducing the risks during testing. The frame system used during the testing can be seen in Figure 7.1. The LiDAR was fastened to the lower horizontal frame element by a sys- tem of brackets, see Figure 7.2. The brackets were fastened with T-nuts positioned 53 7. Testing and Evaluation in the T-slots of the aluminium profiles. This allowed for easy sideways adjustment of the LiDAR. Slots in the brackets allowed for forward and backward adjustment of the LiDAR. The cleaning head was fastened with two brackets in the T-nuts in the T-slots on the upper horizontal frame element. The T-nuts allowed for sideways adjustment of the cleaning head, see Figure 7.2. Figure 7.1: The test rig with frame, LiDAR, cleaning head and hoses installed seen from the back. Figure 7.2: Fastened LiDAR and cleaning head on the horizontal frame elements attached to vertical frame. The rig was placed in close proximity to a built-in air supply of the laboratory. The in-house air system has the capacity of delivering 0 − 12, 5 bar. The liquids used in the cleaning heads during testing were pressurized with compressed air in a tank of 25 liter. The tank was filled up with the liquid and connected at the top to the laboratory’s air supply. At the bottom, the hose with the correct dimension for the testing was attached. Since many risk factors were present during testing, such as relatively high pressures and high noise levels, a risk analysis was made for the test rig. For each risk factor, corrective actions were taken. For instance, protective screens were placed around the test area and hearing protection was used. The planning of the rig setup was done in collaboration with the la