Geometrisk modellering av veger i 3D
Abstract
There will always be a need for road engineers to have access to suitable applications in order to execute road constructions. An application has to be both flexible and efficient, while at the same time describing the road system in detail according to the rules and standatds of road construction. Construction and maintenance are both very expensive and thus important to the whole society. Good administration of the infrastructure implies also that the basic data from planning process can be used directly in the building process and later to maintenance the road system. Planning data are therefore a valuable resource that it is important to take care of.
The purpose of this thesis is to learn more about how road geometry behaves during 3D modelling of roads in an object-oriented context. The work has resulted in a development of a methodology in constructing roads in 3D, based on existing conditions of how road construction had to be done. In this work the 3D model includes all road lines limited to the road shoulder and the road surface. Ditches and terrain areas are not included. However, it is easy to enlarge the 3D model. This model can be used as a basis to produce drawings and other illustrations, and to carry out impact analyses, etc.
The work is based on the computer program KryssWin. KryssWin is an application that is developed for Windows and executes feometric calculatons of road lines in intersection areas. The application is easy to learn and easy to use. It gives good results quickly. It includes three modules:
• T-intersection (intersection where three roads meet)
• X-intersection (intersection where four roads meet)
• Roundabout (3,4 or 5 roads)
The three modules are based on the following principles:
• KryssWin supposes that the reference alignment to the roads is calculated outside the application and then saved in different files with a specified format (10-table format)
• The degree of channelization to the different roads in an intersection is user-defined in KryssWin
• The parameters used for designing the road lines in an intersection are user-edited in KryssWin
The experiences with KryssWin have been good and the following results have been reached;
• The work with geometric design of intersections has become much more efficient.
• it takes only a few minutes to calculate a complete design of a given intersection with support of KryssWin.
• It is very easy to update a solution of an intersection.
At the same time the use of KryssWin has identified a need for several of improvements.
These are;
• One-sided use of parameters to design the intersection is a limitation to the user. Usually Krysswin is used to design the basic geometry and then the final result is edited in a CAD-application.
• KryssWin calculates horizontal geometry only. The methods used in KryssWin are developed to calculate 2D geometry (the horizontal dimension). It is complex to extend to 3D methodology because of the internal relation between all the road lines in an intersection. Such methods are not developed in any exisitng applications.
• KryssWin calculates on intersection at a time. It is a difficult and time-consuming process to design many intersections in such a way that they describe a continuous geometric solution. Two KryssWin-calculations do not know anything about each other. Any method of connecting two adjacent KryssWin-calculations has to be done manually by the user.
Based on these limitations and required improvements of KryssWin, work with developing a new and improved methodology of geometric design of roads was begun. This work has also become one of the main issues of the thesis. The following items are defined;
• All roads lines must be calculated in 3D (horizontal- and veritcal curvature)
• A better possibility for the user to adjust the geometry to local conditions
• Connection between two adjacent intersections has to be handled through the methodology
A prototype has also been developed. The work with the prototype has given good results in many areas. Firstly, a several methods for calculating dereived 3D geometry have been developed. All these methods have properties that make it possible to calculate most of the geometric situations that perform and have to be solved in connection with multi-road construction. Secondly, methods have been developed that handle the dataflow from the initial reference aligmement until the calculation of the final result consisting of a 3D model as a network of a multi-road construction.
The 3D model includes many road objects. These objects can be composed into a seamless geometric whole. Such a solution is shown in Figure 3-3, pp. 3-11. The model consists of several types of road objects, all calculated with the basis of the reference alignments describing a network. Every road object calculates different road lines according to the internal structure where these road lines are described as derived 3D geometry.
To work out the calculation of the 3D model, tow class of models have to be defined;
• The object model
• The data model
The object model describes the data that is included in the final result of the 3D model. This means all the different types of road lines that are included in every performing road object. The object model is divided into several levels of groups. This division is illustrated in Figure 3-4, pp. 3-12. Road lines are the lowest level in this object model and are included in all structures in the 3D model.
The data model describes the data used in the calculating process of generating the different road objects. These data do not describe the final solution directly, but data that are needed to execute the entire calculation. Also these data appear at several levels. Therefore, the main model demands many methods to realize the calculation of the 3D model.
The fundamental principle of multi-road construction is to start the designing process of a 3D model with one or several reference alignments. The designing process is closed when the 3D model is successfully generated, fulfilling the items in Chapter 3.2.1. The designer is allowed to edit the final model continuously. This is done by editing the different data in the data model (the flow-data) and recalculating the 3D model according to the edited data.
As mentioned, the flow-data occur at different levels of details. The technique is to define data values at a superior level of details. Then, more detailed data will be generated supported with many suitable methods (e.g. the calculation of super elevation diagrams). Finally the data flow process is finished with a data format which is used to calculate every single road line included in the specific road object. Different methods can be linked to this data format.
When a road line is calculated, different geometrical combinations have to be solved depending on the specific situation. This is done by dividing the road line into segments, with a method linked to each segment. By this principle it is possible to develop new methods if necessary. A road line consists of either one or several segments. An analogous connection from one road to another needs methods that solves the specified situation. A method is linked to the connection between two sequences. These methods are described in Chapter 5 and Chapter 6. A road object has a standard definition og methods that can be used to calculate the 3D model.
This involves a 3D model being a line model. This means that the entire geometry that is calculated is road lines. In the 3D model the lines will appear as knuckle lines between the different surfaces of which the 3D model implicity consists. Each method is primarily developed to calculate 3D lines.
To understand all the geometric requirements and conditions, the calculation is divided into tasks that are specialized to solve different structures included into the 3D model. Such structures consist of defined road objects such as T-intersection, Roundabout etc. these road objects are then put together into a seamless geometric whole. To this purpose “ connection sections” are used to deal with the road lines at the outer edge of the road object.
Generally speaking, a multi-road construction consists of many roads and will be thus be composed of many reference alignments. The final calculation of a multi.road constrction describes, therefore, a 3D model for the given road system.
There are endless possibilities to how a reference alignment can be designed. To make the situation even more complex, a network can be put together of as many reference alignments as the designer wants. By combining the different road objects to the structural requirements, it is possible to design 3D models that solve every existing situation and those that will occur.
The technique of building a 3D model by put together road objects where each is specialized to different structural situations, is a basic principle in this thesis. Figure 3-4 illustrates different levels of objects in constructing the 3D model. In this way different problems can be isolated and solved according to specific requirements. By defining an interface between the one object to another. It is also possible to implement new objects to the 3D model if required.
In multi-road consruction every reference alignment that describes the network has to be administrated. A lot of geometrical requirements have to be fulfilled before the construction of the 3D model can be executed.
A road construction can consist of many different geometrical structures. These structures are defined according to the geometric design of the construction. This depends on the traffic situation and the solution desired. A road object is a such structure and is rather complex. It contains a lot of road lines. Every road object gets a special method. This method handles the consistence of the road object and ensures that the road lines fulfil the internal geometric relation, 6 road objects for the 3D model have been developed.
• Straight road
• T-intersection
• X-intersection
• Roundabout
• Acess ramp
• Bus-bay
The main task of this thesis is to describe how a 3D model is designed of different components. These components have different levels of complexity. A road object is the highest level of complexity and includes many lower.level components that hasndle different local tasks combined to the geometric solution. Tp design a 3D model it is important to develop road objects that can be included in the 3D model description. Without these road objects, the 3D model can not be edited. It is therefore important that the system is organized in such a way that new road objects can be developed continuously as required.
The development of the road design system has to be started today. There will always be discussion about how this must be done. The next generation of a such road design system has to be object-oriented. The whole process will be digitalized. This means that the result will contain a 3D model that describes every road line with a high degree of accuracy. The 3D model is a digital copy of the final product. Målsettingen med avhandlingen har vært a få mer kunnskap og bedre innsikt i de geometrisk sammenhenger som er avgjørende når en skal beregne veger i 3D. Arbeidet har resultert i utvikling av en medtodikk som bærer preg av en sterk forankring i de vegfaglige premisser og berep som gjennom lang tid er innarbeidet innen fagfeltet.
Avhandlingen tar utgangspunkt i programmet KryssWin som undertegnede har vært sentral i utviklingen av. KryssWin ble utviklet på Windows-platformen og utfører geometrisk beregninger av veglinjer i kryssområder. Programmet er enkelt å ta i bruk og gode løsinger lar seg raske beregne. KryssWin består av tre moduler som beregner T-kryss, X-kryss og rundkjøring.
Alle tre modulene fungerer etter samme grunnprinsipp;
• KryssWin forutsetter at referanselinjene i tilfatrene er linjeberegnet utenfor programmet og lagret på hver sin fil på 10-tabell format
• Kanaliseringsgrad i de ulike armene gis i KryssWin
• Parametre for utforming av veglinjene i krysset gis i KryssWin
Erfaringene med KryssWin har i stor grad vært gode. Arbeidet med geometrisk utforming av kryssområder er effektivisert i betydelig grad. En kan raskt få beregnet et utkast til kryssutforming ved bruk av KryssWin og det er enkelt å utføre endringer i kryssets utforming.
Samtidig har bruken av KryssWin vist at det er behov for forbedringer av programmet. KryssWin beregner kun horisontalkurvatur uten høydedata. Begregningsmetodene som er brukt i KryssWin er utviklet for å kunne håndtere 2 dimensjoner (2D). Overgang fra 2D til 3D er komplisert siden det krever at samtlige veglinjer i krysset får begregnet vertikalkurvatur (høydedata) med en innbyrdes geometrisk sammenheng. Slike beregningsmetoder finnes ikke i dagens geometri-og prosjekteringssystemer.
KryssWin beregner ett kryss i gangen. Det er vanskelig og tidkrevende å beregne flere veger og kryss slik at de sammen beskriver et sammenhengende vegprosjekt. To tilstøtende KryssWin-bereginger kjenner således ikke til hverandre og eventuell sammenkobling mellom dem må håndteres manuelt av brukeren.
Med utgangspunkt i de generelle bergrensninger ved dagens prosjekteringssystem samt de forbedringer som er ønsket ved KryssWin ble det startet et arbeid med å utvikle en ny og forbedret metodikk for geometrisk utforming av veger. Dette er også blitt en av hovedoppgavene ved denne avhandlingen. Målet for arbeidet er således gitt ved følgende hovedpunkter;
• Alle veglinjer skal beregnes i 3D i form av horisontal-og vertikalkurvatur.
- Dette innebærer at modellen er en linjemodell. Selv om en beskriver tverrprofiler med flater, vil likevel den endelige løsningen av modellens geometri bestå av veglinjer. Disse opptrer som knekklinjer mellom de ulike flatene som modellen implisitt består av.
• Det skal være større mulighet for brukeren å tilpasse geometrien til lokale variasjoner.
• En skal kunne håndtere sømløse overganger mellom flere kryss (flervegskonstruksjon).
I avhandlingen er modellen begrenset til å omhandle den geometri som ligger innenfor skulderkantene og begrenset til vegoverflaten. Dette betyr at terrengdata, grøfteområder og underliggende lag i vegkonstrukson ikke inngår i denne modellen. Det er imidlertid enkelt å utvide modellen senere dersom en ønsker det.