Utility-Locating Technology Development Using Multisensor Platforms (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

6Research Approach State-of-the-Art Summary This chapter presents the approach used in addressing the challenges faced by utility detection and locating systems across versatile cultural and environmental conditions. The approach focused on the development and advancement of common reliable near-surface geophysical detection technologies coupled with enhancements to data acqui- sition, data processing, data interpretation, and mapping software. Multichannel GPR, TDEMI, and preliminary high-frequency shear wave seismic applications are the geophysical detection methods studied and designed under this research, along with ongoing software development and improvements to Semiautomated Process and Detect (SPADE) and other commercial geophysical software. Extensive work in the experimentation and development of modeling software was also conducted with an aim to develop finite difference time-domain software capable of simulating propagation of acoustic and elastic waves in realistic soils. The current UIT system and the improvements imple- mented under this research are meant to contribute to the determination of utility locations beyond QL-B, which means providing a depth as well as horizontal location. This depth determination is the key feature behind 3-D products designed for mapping subsurface utilities. The GPR and seismic components of this research are to offer a rendering of depth via the data processing used to produce the 3-D images obtained. Accurate depth estimates can also be attained through advanced analysis of TDEMI data. Most other technologies deployed for SUE map- ping provide a 2-D product, such as a contour map, from the measurements. The following sections briefly describe each of these elements of the R01B research work aimed at gen erating accurate 3-D results to subsurface utility investigations. Technologies Explanation The SHRP 2 R01B project can be divided into separate defin- able features of research work. The technologies addressed under this research include • Multichannel GPR system (UIT); • SPADE software enhancements (Sagentia, Ltd.); • Seismic development (OES, Psi-G, and Bay Geophysical); • Time-domain electromagnetic development (SAIC); and • Modeling software development (LTU). Multichannel Ground-Penetrating Radar GPR data gathering processing and interpretation is a preferred method for utility mapping when the depth-to-targets infor- mation is important and/or when unknown utilities are sus- pected to exist on a project. However, clay soils do not allow for the penetration of GPR waves, diminishing their effectiveness on many projects. When GPR waves can sufficiently penetrate the subsurface, they can provide unprecedented 3-D location accuracy on targets, such as pipes, with differing electrical prop- erties from the soil. GPR can detect metal and nonmetallic anomaly sources and provide accurate depth estimates to utili- ties. The systems work by transmitting a radar signal into the ground and receiving a reflection back at the antenna. An array of GPR antennas widens the detection swath offered by single- channel systems and provides the capability to produce detailed 3-D images of the subsurface; from those images, utilities and other targets of interest may be interpreted and mapped. Multi- channel GPR systems have been developed by UIT and others in the past for performing Subsurface Utility Engineering work. Over the long term, improvements to GPR hardware systems are expected to come from the manufacturing and user com- munities. Those incremental gains in capabilities resulting from hardware configurations were not the focus of this program, C h A p T E r 2

7 however. SHRP 2 R01B efforts concentrated on improving soft- ware functions needed to manage, organize, process, and inter- pret GPR data from UIT’s multichannel GPR system, known as TerraVision II. TerraVision II consists of two banks of seven antennas, each with a fixed spacing of 0.4 ft (12 cm) between each antenna module. (See Figure 2.1.) Data acquired by each of the 14 channels is spaced at 0.08 ft (2.5 cm) in the direction of travel. The central frequency and approximate bandwidth of each GPR antenna element is 400 MHz. Geophysical Survey Systems, Inc. (GSSI), the GPR antenna manufacturer, con- structs the TerraVision units with all of the antennas oriented at 45° to the direction of travel of the cart and with each mono- static antenna having relative orthogonal polarization to the one adjacent. (See Figure 2.2.) During the download process, data sets from the 14 channels are sorted and tagged with the appropriate precision survey (surface position) information. Multiple survey swaths, geo-referenced to the desired coordi- nate system, are assembled into a composite 3-D data block of the project area and subsequently loaded into SPADE for analysis. The data analysis is performed by experienced geo- scientists subsequent to field operations, a process which can be time consuming. The new software features developed under SHRP 2 R01B research are aimed at increasing the efficiency of data analysis processes and improving the accuracy of the 3-D interpretation for utility mapping projects. The follow- ing section describes improvements to UIT’s GPR processing and interpretation software (SPADE) that fulfill this plan. SPADE Software Enhancements (Sagentia, Ltd.) SPADE is UIT’s primary software package that accepts final interpretation-ready GPR data. The technical approach for the development of algorithms to extract features from GPR, and eventually seismic data, builds on the project team’s experience and the existing capabilities of SPADE. The soft- ware is designed to automate certain interpretation proce- dures used in picking utility targets and to make it easier for interpreters to manipulate GPR data for visualization. SPADE also currently has the ability to incorporate geophysical sen- sor data, imagery data, and geo-referenced feature data within a centralized single software platform. With this capability, interpreters are easily able to correlate geophysical data sets with each other and with as-built drawings and other site data for fast and accurate comparison and verification. A primary R01B research objective was to improve SPADE’s automated feature extraction capability and 3-D migration functions for multichannel GPR data sets, with those advance- ments ultimately transitioning smoothly to processing of 3-D seismic data sets. In a 3-D graphical environment the data analyst has the ability to view and refine utility-like features that are automatically extracted from the acquired geophysical data. This software-driven “reduction” of geophysical infor- mation is applied after field data acquisition and uses a set of parameters defined by the geophysical data analyst. Three- dimensional target picking functions are also incorporated into the SPADE software so that utility information can be interpreted, depicted, and delivered with horizontal and verti- cal accuracy. Figure 2.3 shows a graphical window example of data analysis in SPADE. UIT subcontractor Sagentia, Ltd., a technology and product development company in Cambridge, United Kingdom, was the project entity responsible for executing the software devel- opment plan and producing SPADE software enhancements. UIT data analysts implemented, reviewed, and tested the Figure 2.1. UIT’s Multichannel GPR system, the TerraVision II. Figure 2.2. Internal arrangement of a single bank of TerraVision II GPR antennas.

8software upgrades and provided user-sustained feedback. The three key technical elements of the Sagentia approach were • Algorithms for segmenting GPR and seismic data; • Algorithms for extracting features from segments; and • A user interface that enables the user to efficiently and flex- ibly process the whole data set. Algorithmic elements were developed in a modular fash- ion; activities included building initial GPR segmentation algorithms, initial user trials, software prototypes, implemen- tation in SPADE, and validation trials. Both project and archive UIT data sets were used in the experimental testing of the software. The range of algorithms developed was spe- cific to GPR data sets. Seismic and TDEMI data sets were managed through separate commercial software. The Sagentia focus of this research was based on a “data segmentation” strategy for handling GPR data sets within a common soft- ware interface with capabilities for both data visualization and feature extraction. Other data manipulation features of the software were also studied, such as the effects polarization caused by the arrangement of antennas in the TerraVision multichannel GPR unit. Sagentia has achieved its major deliv- erable for SHRP 2 R01B, which is the implementation of seg- ments functionality in the VTK version of SPADE. Seismic Development (Owen Engineering Services and Psi-G) Given the limitations of GPR in certain soil conditions, an alternate approach that offers 3-D geophysical data sets is needed to provide information on both the horizontal and depth locations of underground utilities in clay conductive soils. This calls for the development of an acoustic-based geo- physical system. The initial idea of the SHRP 2 R01B seismic reflection system—proposed by Owen Engineering Services with support from Psi-G—was an unconventional design comprising a horizontal shear wave sensor array configured to produce zero-offset (monostatic) down-looking target illumination and reflections. This seismic system was envi- sioned to respond in near-real-time when scanned over util- ity targets of interest in the same way that, for example, GPR might respond, showing hyperbolic halo-like images of dif- fractor targets in raw-data displays. With data-derived migration analysis, the halo images would be processed to present images indicating the approxi- mate localized target depth and position. The advantages important to the success and acceptability of this method were 1. Operation using a proven “pure” horizontally polarized shear (SH)–wave radiating source and matching SH-wave sensor array to use transverse-polarized low-velocity shear waves to best advantage; 2. Vibrator source operation using controlled frequency sweep excitation in the 300–1,500-Hz range and at an appropriate energy level to achieve efficient high-resolution utility target illumination and reflections in soils and backfills without unwanted lower frequency interference; 3. Vertical down-looking directional operation to minimize the detrimental effects of vertical velocity gradients char- acteristic of near-surface soils and backfills and to avoid interfering lateral reflections; Figure 2.3. Screen capture of GPR data analysis in SPADE.

9 4. Concomitant transmission and reflection gains in detec- tion signal-to-noise ratio associated with vertically ori- ented narrow-beam operation; 5. Elimination of the various conventional data processing steps normally applied to seismic measurements, such as ground roll removal, velocity analysis, spatial averaging trace stacking, and removal of interbed multiples, all of which require a relatively large number of source shot points and sensor records together with off-line processing to achieve their results; 6. Ability of seismic waves to penetrate to much greater depths than GPR in practically all types of soils and back- fills to provide a more universally applicable utility detec- tion and mapping capability; and 7. Simplified sensor array operation that eliminates the need for multichannel data recording and analysis by directly summing all sensor element signals in the 2-D array to yield a single-output down-looking reflection response. Figure 2.4 illustrates the envisioned seismic system plat- form. Testing and equipment requirements were detailed by subcontractor Owen Engineering Services. All of these fac- tors contribute to a prodigious undertaking, so the SHRP 2 R01B project focused primarily on proof of concepts to meet the research objectives behind such prototype development. To achieve practical mobility of the conceived seismic sur- vey system, proven methods of attenuation compensation and ground coupling must be determined. To that end, the project team conducted a series of preeminent and individu- alized seismic soil properties tests at various locations across the United States. The results of that testing is discussed fur- ther in Chapter 3. Time-Domain Electromagnetic Development (SAIC) Time-domain electromagnetic induction (TDEMI) systems are another geophysical methodology used to support Qual- ity Level B SUE activities. These systems can be operated as single sensor units or as an array of synchronized sensors. They detect subsurface utilities constructed of any type of metal. Detection capabilities depend on the depth to target, composition of target, size and orientation of target, sur- rounding geology, and the amount of cultural debris or inter- ference from other surface cultural features. TDEMI systems used for digital geophysical mapping (DGM) provide multi- ple measurements of the decay of the secondary magnetic field associated with any metallic object. With the current industry standard TDEMI DGM system (Geonics EM61-MK2), data from as many as four monostatic Figure 2.4. Schematic of the production prototype seismic system.

10 time gates provide geophysical data analysts information on the detection and characterization of utility anomaly sources. The earlier time gates offer improved detection for smaller tar- gets where the decay rate of the secondary field is relatively quick. Additionally, the early gates provide an increase in the response amplitude from any target, regardless of size, com- pared with later time gate measurements. Later time gate mea- surements are useful for the description of the time decay associated with any target of interest. Detailed time gate infor- mation that is recorded from various sensors at strategic orien- tations can offer data analysts a basis for constructing software models for specific target discrimination techniques. All data from the detection sensors can be easily integrated with global navigation satellite system (GNSS) and/or robotic total station (RTS) equipment data. With a maximum data collection rate of 15 records (total) per second for current systems, travel speeds up to 3 mph are preferred for optimal detection. One aspect of the R01B project was to improve on current TDEMI systems by testing and developing a new advanced TDEMI technology for locating and classifying underground utilities. SAIC led the effort, building on the company’s work with the U.S. Department of Defense in the development of improved TDEMI technologies for locating and characteriz- ing buried munitions objects. Figure 2.5 shows the TDEMI sensor array that SAIC developed collaboratively with the U.S. Naval Research Laboratory (NRL). The array, which is towed behind a utility vehicle, consists of 25 transmit (Tx) and receive (Rx) coil pairs and is 2 meters square. It has fully pro- grammable Tx and Rx parameters; and because it employs modern digital electronics, it is not subject to the drift prob- lems typically experienced with commercial EMI sensors such as the Geonics EM61 and EM63. As it is usually configured for characterizing buried objects, it measures the EMI decay from 0.04 ms to 25 ms. This system has demonstrated near perfect performance in classifying buried targets such as unexploded munitions or metallic clutter at U.S. Army test sites. The R01B project aimed to develop a prototype transient electromagnetic method (TEM) system consisting of five Tx and Rx coil pairs. The system would use SAIC’s current TDEMI processing algorithms that were developed for locat- ing and identifying compact buried objects such as unexploded munitions items. Those algorithms are based on point, rather than line, target models; and the R01B plan involved adapting those TDEMI array data processing and analysis procedures to support classification of underground utility lines, assemble a library of utility signatures, and test the processing and classifi- cation procedures using the prototype TDEMI array at a few utility test sites. The full scope of objectives set forth in the plan have yet to be realized; however, a 5 × 1 sensor TDEMI proto- type has been developed to operate advanced Tx and Rx coil pairs at variable time gate intervals on a towed instrument plat- form. The Tx and Rx parameters of the prototype are fully pro- grammable, which is useful in optimizing the detection/location of underground metallic utilities. Data from the TDEMI pro to type can be imported into Geosoft Oasis Montaj soft- ware for organization, processing, 2-D and 3-D visualization, and interpretation. Seismic Modeling Software Development (LTU) As part of the seismic element of the R01B work, researchers at Louisiana Tech University (LTU) worked to develop mod- eling capabilities useful in helping to predict and understand high-frequency shear wave seismic measurements and to indicate the amount of computational resources necessary. LTU objectives were to develop finite difference time-domain (FDTD) method software capable of simulating propagation of acoustic and elastic waves in a realistic soil, to study the properties of acoustic and elastic wave propagation through a soil, and to analyze the physical consequences of the obtained results. Based on its results, LTU was to explore the possibili- ties of developing a virtual testing laboratory for the simula- tion of the acoustic and elastic methods for the detection of buried pipes and conduits. During Phase 1 of the project, the 2-D modeling code, already developed by LTU researchers, was expanded to a 3-D model. The code was restricted to run on multiple processors on the Louisiana Optical Network Initiative (LONI) network of super- computers to achieve a reasonable run time for large models with dimensions similar to some of the proposed field tests. The numerical simulations use the most accurate material proper- ties available, representing the in situ field conditions. Along with the development of a custom-written code, other commercial software packages, like the Wave Propagation Figure 2.5. SAIC/NRL TDEMI array for buried-object location and classification.

11 Program (WPP) from Lawrence Livermore National Labora- tory, were used during the simulation exercise. WPP is a finite difference code which solves elastic wave equations in three dimensions. The 3-D elastic wave equations are a 3-by-3 system of second-order hyperbolic equations. The WPP code uses the MPI (Message Passing Interface) library for parallel computa- tions on a Cartesian grid, with variable wave speeds and density throughout the domain. The code was used for simulating both seismic wave propagation and nondestructive testing. WPP was intensively tested in the seismic case to assimilate the scale of application intended for the R01B project. In the second phase of the project, the results from the numerical models were validated against the data collected during the field experimentation component of the work. Following validation of the code, multiple simulation runs were performed, with and without attenuation, to cover a wide range of scenarios. Simulations included various soil and source types, along with multiple buried plastic pipe tar- gets and different depths. The goal was to develop a dynamic finite element method to analyze the wave propagation in the nonlinear inelastic soil media. The applications of the FDTD method in numerical simulations started relatively recently; the numerical conditions the FDTD method must satisfy require substantial computer memory and computational power. Therefore, the WPP code and simulations were run and tested on the LONI system of supercomputers.