Title: Mobile LiDAR Approach for Evaluating Tunnel Clearance and Alignment on the BNSF s Cascade Tunnel

Transcription

1 Title: Mobile LiDAR Approach for Evaluating Tunnel Clearance and Alignment on the BNSF s Cascade Tunnel Authors Names: Marc A. Cañas, GISP/MBA joined J.L. Patterson & Associates 20 years ago. He is currently Vice President and Director of Class I Services. Marc was project manager for the Cascade Tunnel Study. He holds a Geographic Information System Professional (GISP) certification from Cal Poly Pomona, a B.S. in Computer Information Systems from Chapman University and a Master of Business Administration from the University of Southern California (USC). Marc is an active member of AREMA Committee 14 Yards and Terminals Nathan Ortega, P.E. joined J.L. Patterson & Associates 5 years ago. He is a Project Manager for the firm and was the project engineer for the Cascade Tunnel Study. Nathan holds a B.S. in Civil Engineering from the University of Arizona. Bryan Philips is director of business development for SAM, Inc. a leader in providing geospatial data solutions such as land and hydrographic surveying, airborne, mobile, and terrestrial LiDAR, aerial mapping, GIS, SUE, utility coordination and construction phase services. Number of Words: 3198 Abstract: Even though railroads have been assessing and resolving clearance issues for a long time, some situations introduce enough variables to render traditional solution methodologies ineffective. Such was the case with clearance issues plaguing BNSF s Cascade Tunnel in western Washington. The tunnel is 7.8 miles long and it was constructed between 1925 and Traditional data collection and survey methodologies for a tunnel such as this are problematic and ineffective due to high traffic, limited access windows, relatively tight clearances and tunnel length. In 2012, BNSF contacted J.L. Patterson & Associates (JLP) for assistance in solving the problem. JLP together with Surveying And Mapping, Inc. (SAM) developed a unique approach that combined traditional survey methods, Mobile LiDAR, and innovative engineering solutions, particularly track to tunnel alignment analysis to identify the causes and recommend solutions for the clearance issues. This paper recaps the challenges and circumstances initially encountered; outlines the process for developing the solution, reviews highlights of the technology employed, and summarizes the recommendations developed for the project. The technology employed includes sample point clouds and the tools used to extract the data utilized as input to the engineering analysis. It also includes sample data sets and representations of the track alignment analysis completed to locate the precise problem areas and provide recommendations for minimally invasive solutions for resolving the clearance issues. AREMA Image Cascade Tunnel Map 1

2 In summary, this paper provides an overview of the projected cost/effort reduction including the decrease of cut material from 177 yd 3 to 8.3 yd 3 by leveraging technology and innovative analysis methods. Introduction: The Cascade Tunnel is the longest railroad tunnel in the United States and, at the time of its construction, it was the longest tunnel in North America. It is 7.8 miles long and was constructed between 1925 and 1929 near Stevens Pass in Washington on the Skykomish to Wenatchee route, Scenic Subdivision, operated by BNSF Railway. When it opened, it reduced the existing route length by 8.7 miles and eliminated 21 miles of treacherous grades of two percent or steeper. The tunnel is at an elevation of 2,894 feet above sea level at the east portal and it has a grade of 1.57 percent rising from west to east. The tunnel is single track and its alignment is tangent along its entire length (more about this later). In 1993, the tunnel was listed as a US Engineering Landmark by ASCE and is described as follows in the ASCE register. The area of Stevens Pass contains the first and second Cascade Tunnels and the switchbacks which carried Great Northern Railroad trains over the pass since The pass is named for John F. Stevens, (1927 ASCE President), who located it in Stevens was made chief engineer of the Great Northern in 1895, and while in this position he supervised construction of the first Cascade Tunnel between 1897 and The second Cascade Tunnel was the longest tunnel in the Western Hemisphere from 1929 to The Cascade Tunnel serves as a crucial route for BNSF in hauling varied cargo to and from the Port of Seattle; it is also used by Amtrak on its Empire Builder scenic route. Though the tunnel has aged well as a consequence of its sound engineering design and construction, the structure itself does Image 2 Survey High Rail Vehicle with LiDAR equipment require upkeep to combat environmental factors as well as changes in the rail industry. In 1956, a ventilation system was installed to allow diesel locomotives to pull straight through the tunnel, eliminating the need to couple and uncouple electric engines at each end. In 1984, a crown cutting project was completed to create notches in the crown of the tunnel to accommodate additional height requirements of double stack container cars. This brings us to 2012 when BNSF began to assess the need for alignment and clearance envelope improvements in the tunnel. However, due to the corridor s voluminous traffic, engineering and survey crews are limited to no more than one or two hour access windows at any given time. With these access constraints it would take weeks just to survey the track using conventional survey method, to say nothing of modeling the interior of the tunnel which was required to determine clearance optimization plans. AREMA

3 Image 3 Image Capture from Tunnel Surveillance Video The Challenge: BNSF has received reports of container damage due to apparent strikes by trains traveling through the Cascade Tunnel. This damage was observed on the top containers of double stack trains. Some physical investigations followed the observation of the damage, including painting the roof of the tunnel black, to allow for visual inspections and identification of strike points, where the paint had been scraped away. This visual inspection was followed by the use of cameras at the locations that appeared to be the most severe areas of conflict. This camera inspection did indeed support the strike locations, in a dramatic fashion. The observation method of conflict identification had some limiting factors, one of which was that cameras could only be placed at accessible areas of the tunnel and there was not any guarantee that there were not additional conflict points throughout the non-accessible portions of the tunnel that did not have cameras. Additionally, the owner desired to ensure that these strikes could be avoided in the future, as track and tunnel maintenance continues throughout the years. With this challenge at hand, BNSF contacted J.L. Patterson & Associates, Inc. (JLP) to develop a work plan to positively identify the cause of the conflict and to develop innovative ways to collect, model, and study the tunnel in route to providing a solution for the conflicts and damage being observed. JLP partnered with our mapping subconsultant, SAM, Inc. and our ground survey partner True North Land Survey to develop a survey program combining a control survey and a LiDAR survey to map the track and inside surface of the tunnel, in effect developing a complete and highly accurate 3D model of the 7.8-mile tunnel. LiDAR is an acronym for light detection and ranging. LiDAR is a remote sensing technology that measures distance by illuminating a target, or target surface, with a laser and analyzing the reflected light. The LiDAR equipment used at the Cascade Tunnel was configured on a high-rail vehicle capable of directly accessing the track. The system utilizes two 360-degree LiDAR sensors, two 5-MP roof mounted cameras, two GPS receivers, an Image 4 Actual Point Cloud Image of the Interior of the Tunnel this is not a photographic image Internal Measurement Unit (IMU) and a Distance Measuring Instrument (DMI). This setup was specified to provide data point density of up to 150 points per square foot. This minimally invasive method of surveying can be conducted day or night with efficient acquisition of millions of 3D design points per minute. The data collection was accomplished in one hour. That was the easy part! AREMA

4 Image 5 Actual Point Cloud of Tunnel Entrance this is not a photo image Because of the length and depth of the tunnel, GPS reception was non-existent and therefore many of today s modern survey tools would have been inoperative. The model of the tunnel needed to be one homogeneous representation of the entire tunnel rather than separate cross sections. Therefore, the survey effort provided two distinctly separate challenges: (1) determining the location of the trajectory of the LiDAR sensor; and (2) determining the relative location of discrete elements of the tunnel Surface with respect to the LiDAR sensor. To address the first challenge, the JLP team established horizontal and vertical survey control. Pairs of control targets were placed at each portal and a control targets were placed approximately every 750 feet along the tunnel floor. The control targets were referenced to the North American Datum of 1983 and the North American vertical datum of 1988, using conventional survey methods. This process was effective but time consuming. Because of the frequency of train traffic, the survey crews were limited to no more than 1 hour of track time and in several days no more than 30 minutes. Setting targets and turning points in this limited time frame required orchestrating a plan to make the most efficient use of the work time allotted. The survey crews employed the help of the flagman s hi-rail and would load up as soon as the time was given. They would then place their survey equipment in front of hi-rail and set and shoot as many targets as they could. Then jump back in the hi-rail and clear the track for the next train. AREMA

5 Figure 1 Survey Control Targets AREMA

6 This sequence continued for 8 days until all 7.8 miles of tunnel was targeted (Figure 1 Survey Control Targets). The control points and inertial navigation data were used to determine the smoothed best estimate trajectory of the three dimensional route of the mobile mapping system. The second challenge was addressed with the density and highly accurate point cloud from the LiDAR sensor and tracking equipment as it traversed the tunnel on a specially equipped hi-rail vehicle. Figure 2 Sample Engineering Cross Section, Existing Track and Tunnel Surface Engineering Work Flow: With the modeled existing interior tunnel surface information, JLP developed the following workflow to proceed with engineering and analysis of the tunnel s existing condition: 1. Process Survey/LiDAR Data 2. Analyze results from the Survey/LiDAR Data 3. Generate Existing Surfaces 4. Analyze Existing Track Alignment 5. Analyze Existing Notch Alignment 6. Propose Best Fit Alignment 7. Propose Additional Notching 8. Conclude with Best and Most Economical Overall Fit Figure 3 Actual interior triangulated surface with modeled key points JLP utilized Bentley MicroStation V8i Select Series 3 to accomplish the engineering analysis. The specific software package included: Bentley Rail Track Select Series 2, Tunnel plug-in tool, Roadway Modeler and supplemented with on-line Tutorials from Bentley s Learn Server. The design team was faced with a unique challenge, model & analyze a 3 dimensional surface of the tunnel utilizing rail regression tools that are normally used for linear surfaces as opposed to a surface that wraps on itself as is the nature of a tunnel. The initial attempts resulted in failures because the DTM (Digital Terrain Model) would connect AREMA

7 the interior surfaces as if it was a solid. The JLP team quickly realized that we needed to reach out to experts outside of the rail industry and through our worldwide forums on Bentley Learn, we reached out to a user group in Chile who recommend an additional piece of software specifically for tunnel analysis. This plugin did the trick literally unwrapping the tunnel surface in our 3D modeled environment allowing us to use the regression tools we knew we need to properly analyze the tunnel and its relative alignment to the track. Utilizing these tools, then allowed our team to develop a flattened digital 3D model of the tunnel s interior surface and extracted a precise center line of the track alignment as well as the location of each of the existing notches within the tunnel. To our surprise, while the track was tangent and within FRA class 2 alignment specifications, the tunnel notch was not exactly centered on the alignment as it stands today The cross sections showed that the 1984 crown notching was slightly off center when compared against the CL of the tunnel crown and after years of maintenance the track alignment while tangent was now slightly off center of the notching, causing unwanted and damaging interaction between train/containers and the tunnel notch in this rather restricted space. It is important to note that without a dense 3D LiDar model of the tunnel, it would have been nearly impossible to positively identify the conflict locations. Having data points at this level of detail allowed JLP to analyze and make recommendations on a foot by foot basis. Armed with this powerful information, JLP was able to execute an exhaustive evaluation process of track and notch alignment. The evaluation revealed that the track meandered through the tunnel and deviated to the left and right of a single tangent alignment by a maximum of approximately 6. The evaluation also revealed that when the centerline of notch was related to the existing centerline of track alignment there was a consistent offset in one direction (left) that averaged 2.2 inches, with maximum offsets of almost 5 inches. (See Figures 4 & 5 Left & Right Offset Report) The foot by foot analysis led to a level of detail that could pinpoint the conflict areas and effectively eliminated any wasted construction effort and, maybe most important of all, keeps train traffic interruptions due to construction at an absolute minimum and minimizes construction risk by reducing the amount of work required inside the tunnel. AREMA

8 Figure 4 Station Offset Report Left Figure 5 Station Offset Report Right The JLP team quickly realized that using conventional methods for track design in open terrain was not going to be AREMA

9 the best solution in this case. As track design engineers we know inherently that angle breaks in a tangent main track are not desirable. However this proved to be one case where a properly designed angle break was exactly the solution we needed. Well, 36 angle breaks to be exact. To best serve our client, JLP proposed a track realignment consisting of 37 horizontal tangents through nearly 8 miles of tunnel. These segments were structured to take maximum advantage of the existing notch alignment while Figure 6 Basis of Circular Curve and Angle Point maximizing each tangent length and utilizing almost negligible bearing shifts that are equivalent to curved elements with radii so large that any curve or angle break is virtually undetectable and just as important constructible. A maximum angle break of 0 o 1 23 (See Figure 6) was used to propose an alignment that would follow the notch alignments to a reasonable and constructible degree while exceeding track alignment and speed standards for FRA Class 2 track. It is important to note that while FRA Class 2 standards meet the timetable speed requirements, it is the recommendation of this analysis that the track through the tunnel be maintained to FRA Class 4 or better requirements to ensure that alignment and cross level conditions always stay within minimum clearance envelope limits established. The maximum angle break was developed due to its qualities that allow it to operate in the same manner that a very flat curve with a 250,000 radius and a length of 100 would, the curve and the tangents with a bearing difference of 0 o 1 23 result in an external distance of AREMA

10 Following the analysis, JLP revisited the data to examine it for any abnormalities or additional benefits that could be extracted. With the amount of data collected at such a detailed sample rate all aspects are available for review at any time. During this process of reexamining the data for details not of primary interest in the initial study, an added bonus to our analysis was found and that finding was a particular area deep inside the tunnel that contained a segment of track with up to three inches of cross level condition (See Figure 8). This cross level condition was hard Summary of Critical Locations of 1 inch or Greater Cross Level North Rail Report Report Created: 7/17/2012 Time: 10:58am South Rail Report Report Created: 7/17/2012 Time: 1:12pm Project: Description: File Name: Last Revised: Cascade_Tunnel Project: Description: File Name: Last Revised: Cascade_Tunnel J:\Projects\BNSF\84_Cascade_Tunnel\Drawings\17084Cascade.alg rcarrillo 7/17/2012 Note: All units in this report are in feet unless specified otherwise. J:\Projects\BNSF\84_Cascade_Tunnel\Drawings\17084Cascade.alg rcarrillo 7/17/2012 Note: All units in this report are in feet unless specified otherwise. North to South Rail Offset to Offset to Elev(1/8 Feature Station Elevation Northing Easting Feature Station Elevation Northing Easting Elev(ft) Elev(in) CL CL in frac) North Rail South Rail North Rail South Rail Over 1in North Rail South Rail Over 1in North Rail South Rail Over 1in North Rail South Rail Over 1in North Rail South Rail Over 1in North Rail South Rail Over 1in North Rail South Rail Over 1in North Rail South Rail /5 Over 1in North Rail South Rail /5 Over 1in North Rail South Rail /9 Over 1in North Rail South Rail /9 Over 1in North Rail South Rail /9 Over 1in North Rail South Rail /9 Over 1in North Rail South Rail /9 Over 1in Figure 7 Cross Level Summary to believe however, it was later confirmed by the Geometry Car that coincidentally was running through this section of track just after the Lidar survey was completed. The cross level variance condition is produced when one rail is higher than the other in a segment of track. This was causing rail cars to lean and could create additional conflicts between them and the clearance notches of the tunnel that were not due to misalignment. Therefore, the project plans proposed to also included a vertical alignment correction in this segment of the tunnel to eliminate those conflicts all while avoiding extensive re-notching of the tunnel. Figure 8 Zoom View + 3 Cross Level Condition AREMA

11 Conclusion: In conclusion, JLP s engineering solution to benefit BNSF included a recommendation for modifying the track alignment to best fit the existing tunnel clearance notch. The track alignment modifications represent minimally invasive measures to address the misalignments in lieu of extensive removal of hundreds of cubic yards of material in this busy tunnel. JLP calculated that if the existing track alignment was used and the notches were cut to the current required clearance envelope then removal of approximately 177 cubic yards of material and the impacts to operations were going to cost BNSF approximately $10 million. Instead, JLP ascertained that the new proposed notching of the tunnel would amount to only 8.3 cubic yards and the cost would be decreased to approximately $1 million, an unbelievable 90% reduction from the original estimate. Through this study and the innovative use of technology, JLP realized a considerable savings in projected construction time and funding to bring the tunnel up to current clearance standards. In all, JLP analyzed several million data points amounting to more than 2 TB of point data and was able to model the tunnel without affecting BNSF operations in this busy corridor. JLP and our partners, SAM, Inc. are proud to leverage technology to benefit our clients. We understand the railroads are investing significantly to improve their infrastructure and doing it largely within the rights of way that have been in place for over a hundred years. The rail industry is now seeing a substantial resurgence because rail is a safe, reliable, and cost effective transportation alternative that plays a vital role in our nation s economy. Acknowledgements: JLP and SAM would like to thank BNSF, in particular, Mr. Glen Gaz, for the opportunity to provide an innovative solution to the clearance issues in the Cascade Tunnel. His trust and out-of-the-box thinking is much appreciated. We also would like to thank the engineers and designers at JLP who helped finding a cost effective method to achieve the results and who embraced technology to its fullest. References: Bentley Learn Server List of Images: Image 1: Cascade Tunnel Map Image 2: Survey High-Rail Vehicle with LiDAR Equipment Image 3: Image Capture from Tunnel Surveillance video Image 4: Actual Point Cloud Image of the Interior of the Tunnel Image 5: Actual Point Cloud of Tunnel Entrance List of Figures: Figure 1: Survey Control Targets Figure 2: Sample Engineering Cross Section, Existing Track and Tunnel Surface Figure 3: Actual Interior Triangulated Surface with Modeled Key Points Figure 4: Station Offset Report Left Figure 5: Station Offset Report Right Figure 6: Basis of Circular Curve and Angle Point Figure 7: Cross Level Summary Figure 8: Zoom View +3 Cross Level Condition AREMA