1 VEHICLE/BRIDGE INTERACTION MODELING AND VALIDATION FOR

2 SHORT SPAN RAILWAY BRIDGES.

3
4
5
6
7 Anna M. Rakoczy, Ph.D.
8 Transportation Technology Center, Inc. (TTCI)
9 55500 DOT Road, Pueblo, Colorado 81001 USA
10 Telephone: 719-584-0782. E-mail: Anna_Rakoczy@aar.com
11 (Corresponding author)
12
13 Xinggao Shu, Ph. D.
14 Transportation Technology Center, Inc. (TTCI)
15 55500 DOT Road, Pueblo, Colorado 81001 USA
16 Telephone: 719-584-0540. E-mail: Xinggao_Shu@aar.com
17
18 Duane Otter, Ph.D., P.E.
19 Transportation Technology Center, Inc. (TTCI)
20 55500 DOT Road, Pueblo, Colorado 81001 USA
21 Telephone: 719-584-0594. E-mail: Duane_Otter@aar.com
22
23
24
25 Word count: 3,458 words text + 16 table/figures x 250 words (4,000) = 7,458 words
26
27 Submission Date: July 19, 2016
28 ABSTRACT
29
30 Railway bridges are critical in the transportation network and vital to the profitability of the
31 industry. Thousands of bridge spans over 50 years old are still in service. Transportation
32 Technology Center, Inc.’s (TTCI) current work under the Association of American Railroads
33 (AAR) Strategic Research Initiatives Program on bridge life extension focuses on the effects of
34 increased axle loads, extending the safe service life of existing steel bridge spans, and onboard
35 inspection of bridge structural integrity. The program includes various tests at the Bridge
36 Deflection Test Facility (BDTF) at the Transportation Technology Center (TTC), Pueblo,
37 Colorado, as well as vehicle/bridge interaction modeling.
38 This paper presents simulation and test results of a freight car and locomotive running on a
39 railway bridge located at the BDTF. Simulation and test results of an onboard system installed on
40 the Instrumented Freight Car (IFC) indicated that the IFC is a useful tool for identifying some
41 bridge condition issues. The BDTF provides adjustable bridge strength and geometry conditions.
42 Various tests were conducted on the BDTF to investigate the potential for using onboard
43 technology to detect bridge impairment or changes in bridge behavior. Test results were used to
44 validate the NUCARS® three-layer track model of the BDTF. The experimental and analytical
45 case studies were conducted to develop onboard systems for dynamic inspection of bridges under
46 varying loads.
47
48 Keywords: Railway Bridges, Bridge Impairment, Bridge Deflection, Bridge Settlement, Bridge
49 Condition Monitoring, Structural Health Monitoring, Trestle, Vehicle-based Bridge Monitoring,
50 Onboard Systems, Multi-body Dynamic Simulations,. Testing and Validation
51
52
 Rakoczy, Shu, & Otter 1

53 INTRODUCTION
54 In an effort to provide better indication of the current state of a bridge, Transportation Technology
55 Center, Inc. (TTCI) has been investigating the potential for using onboard technology to detect
56 bridge condition or changes in bridge behavior. (1)(2) The motivations for developing such a
57 technology are twofold. First, bridges often show behavior under dynamic train loading that is
58 difficult to observe otherwise. Second, an onboard system has the potential to provide observations
59 for numbers of bridges, as opposed to a structural health monitoring system that is fixed and
60 capable of monitoring only a single bridge.
61 As part of the Association of American Railroads (AAR) Strategic Research Initiative
62 Program, TTCI is investigating the feasibility of using vehicle-based onboard systems to assess
63 bridge condition. Research has focused on using existing onboard systems, rather than developing
64 new ones. Three existing onboard detection systems studied to date include:
65  Revenue service type vehicles (locomotives or freight cars) equipped with sensor
66 packages — typically accelerometers and sometimes also spring displacements,
67  Track geometry measurement vehicles, and
68  Vehicles designed to measure track deflection under vertical load.
69 The findings indicate that each of these systems, with appropriate post-processing of the
70 data, was able to identify the general location of the test bridge. (3)(4) Vehicles designed to
71 measure track deflection under vertical load have shown the most promise for identifying changes
72 in structural behavior of the test bridge. Work is ongoing to move this technology towards revenue
73 service implementation. The most challenging technology at this point is the use of revenue service
74 vehicles such as vehicle-track interaction (VTI)-equipped locomotives and freight cars. However,
75 because many of these vehicles are already operating on several railroads in revenue service trains,
76 it is highly desirable to continue investigation of using VTI-equipped vehicles to detect bridge
77 impairment as well. In any case, it is not anticipated that these systems will replace manual
78 inspection, but they should provide early notification of the need to have an inspector check a
79 particular bridge. The systems should also provide an indication of the types and locations of
80 defects to inspect. Trending of historical readings might be needed to track changes and predict
81 when action might be required.
82 Instrumented Freight Car (IFC) and instrumented locomotive vehicle-track interaction
83 technologies are relatively inexpensive and rugged, with a high availability potential as
84 compared to track geometry and track deflection/modulus measurement systems. Therefore,
85 there is a high motivation for improving IFC capabilities. As a part of this effort, NUCARS®
86 modeling was conducted to predict performance of an onboard system installed on an IFC. The
87 modeling was also used to help determine the measurements best suited for detecting changes in
88 bridge performance.
89 NUCARS®, a registered trademark of TTCI, is a multi-body vehicle/track/bridge dynamic
90 simulation program. The NUCARS® track model allows simulation of flexible track/bridge
91 structures under moving vehicles. A track with ballast and subgrade foundation is usually modeled
92 as either a single-layer (flexible rails) or two-layer (flexible rails and flexible ties) flexible track
93 model. NUCARS® allows more layers and more complex track structure simulations to include
94 features such as bridges and track slabs. In order to determine the dynamic performance of bridges
95 under a moving load, flexible bridge spans were added to the track model to predict the dynamic
 Rakoczy, Shu, & Otter 2

96 responses of the vehicle and bridge including wheel/rail interaction. The NUCARS® multi-layer
97 track model provides essential tools for evaluation of dynamic bridge performances under multiple
98 rail vehicles.
99 Based on the findings from previous tests and NUCARS® modeling, new transducers are
100 proposed to be installed on the IFC in order to detect changes in bridge condition or response. The
101 additional accelerometers installed on the carbody and sideframes of the IFC were investigated in
102 the latest tests on BDTF performed at the end of August 2014. The new test results are presented in
103 this study. With further refinement of processing algorithms, sideframe displacement from an IFC
104 has potential to be used for identifying changes in bridge conditions.
105 Testing on the BDTF
106 The BDTF is a 42-foot, three-span, T-rail trestle. The bridge can be configured to provide variable
107 support, in terms of stringer strength as well as pier top support geometry. The bridge generally
108 follows plans for an open deck T-rail span (sometimes called rail top) bridge with modifications to
109 provide the adjustments desired. The BDTF is located on a lightly used section of track, so it is not
110 subjected to high levels of degradation.
111 In 2012, TTCI performed tests with several onboard systems at the BDTF to determine their
112 feasibility for use in onboard bridge condition assessment. (5) In addition to testing onboard
113 systems, wayside measurements were collected on the bridge. Vertical deflections and bending
114 strains were measured on the left and right sides of each span. Figure 1 shows the three BDTF test
115 configurations:
116 • Test 1: Normal condition – all bridge stringers have normal strength.
117 • Test 2: Weak center span – both stringers in the center span are weakened by about 40
118 to 50 percent.
119 • Test 3: East side of center span weak – only one stringer in the center span weakened
120 40 to 50 percent.
121 The results of the tests were utilized for validation of the NUCARS® simulation model.

122
123
124 FIGURE 1 Illustrations of BDTF Test Conditions
 Rakoczy, Shu, & Otter 3

125 Three vehicles with onboard measurement systems were operated over the BDTF to
126 determine their feasibility for use in onboard bridge condition assessment. The data was gathered
127 using the following systems:
128 • Instrumented locomotive (VTI),
129 • Instrumented freight car (IFC),
130 • Track Loading Vehicle (TLV), and
131 • Track geometry system (mounted on the TLV).
132
133 The TLV used vertical loading to provide measurements related to track deflection and
134 stiffness under various loads. The TLV track geometry system provided the measurements of track
135 surface, alignment, cross level, gage, and related parameters.
136 The VTI and IFC measurements were primarily acceleration-based, which have proven
137 valuable in finding relatively short wavelength defects especially, related to rail surface conditions.
138 Data from the IFC was collected for the three test conditions on the BDTF. The results indicated
139 that of the various IFC measurements available, carbody acceleration and sideframe acceleration
140 have the highest potential to detect changes in behavior of a bridge. Due to high frequency ambient
141 vibrations, it is not possible to detect even the location of the bridge from the raw acceleration data,
142 which has much less changes. It was necessary to develop a post-processing algorithm to obtain
143 signals that might help to identify bridge conditions. Figures 2 and 3 show examples of vertical
144 acceleration and displacement measurements for the IFC as it travelled over the BDTF at a typical
145 revenue service operating speed of 40 mph. The results are presented for the normal condition of
146 the bridge (blue line) and the weak stringer (red line). With the post-processing algorithm, the
147 general location of the test bridge became evident. The BDTF is located between coordinates 300
148 and 400 on the x-axis. Changes in the acceleration signal with change in structural behavior were
149 not very distinct, however.

150
151
152 FIGURE 2 Vertical Carbody A-end Acceleration (2 top histories) and Displacement (2
153 bottom histories) - IFC at 40mph
154
 Rakoczy, Shu, & Otter 4

155
156
157 FIGURE 3 Vertical Sideframe Acceleration (2 top histories) and Displacement (2 bottom
158 histories) – IFC at 40mph
159
160 Summarizing all the results, in some cases the deflections of the three individual spans
161 could be distinguished. However, the distinction between different conditions of the bridge were
162 not very evident. The previous configuration of the carbody accelerometers in the IFC did not
163 provide enough information to distinguish differences in displacements of the bridge from one side
164 to the other, making it difficult to identify changes in behavior of only one side of the bridge.
165 Although the results were not strong, the sideframe accelerometers showed more promise since
166 they were installed on the left and right sides. Further work focused on changing the location of
167 accelerometers and using noise-cancellation to improve the signal.
168
169 NUCARS® Simulations Overview
170 NUCARS® software was used to model typical freight train equipment, a track system, and a
171 bridge.
172 A model of the test train vehicles was created based on the characteristics and properties of
173 the IFC (a loaded 110-ton hopper) and instrumented locomotive (VTI on Electro-Motive Diesel®
174 EMD-SD70M) used by TTCI. The rail vehicle model is presented on Figure 4. Many details about
175 the model can be found in the publication by Wilson. (6) This particular report includes
176 characterization tests to verify the suspension characteristics in model and track tests with
177 instrumented wheelsets. The model presented in the report was developed for an EMD GP35
178 locomotive using Version 1.0 of NUCARS® with ballast and truck modifications to represent an
179 EMD GP40 locomotive with springs and dampers for higher speed operation. Many updates were
180 implemented to take advantage of features in more recent versions of NUCARS® not available in
181 Version 1.0. There are a few differences in models used for the simulation and presented in the
182 report, but the concept is the same.
 Rakoczy, Shu, & Otter 5

183
184
185 FIGURE 4 The Railcar Model
186
187 The track system model included the left rail, right rail, and ties. Each span of the BDTF
188 was modeled in NUCARS® as a pair of beams (left and right). Each of the beams was modeled as a
189 flexible element with parameters representing a real bridge. The parameters of the center span were
190 modified to match two cases of a weakened bridge: 1) a case in which the left beam of the center
191 span was weaker and, 2) a case in which both beams of the center span were weaker.
192 NUCARS® Three-Layer Track Model
193 The NUCARS® track model allows simulation of flexible track/bridge structure under moving
194 vehicles. It represents actual track structures as a multi-body system. The track is usually defined
195 as either a single-layer (flexible rails) or two-layer flexible track model (flexible rails and flexible
196 ties). However, more layers and more complicated track simulations to include features such as
197 bridges and track slabs are allowed.
198 This study used a three-layer track model developed to analyze vehicle-bridge interaction.
199 Figure 5 presents the connections between individual members.

200
201
202 FIGURE 5 NUCARS® Three-Layer Track Model of the BDTF
203
204 The top layer includes left and right 119 RE rails. The track is modeled to be 310 feet long.
205 The second layer contains 7-inch by 9-inch by 8-feet 6-inch timber ties. The weight of the ties used
206 in the model is 26.25 lbs./ft. (225 pounds) and the Young’s Modulus of timber is 16,000 ksi. The
207 third layer is the bridge stringers modeled as prismatic steel beams. The Young’s Modulus for steel
 Rakoczy, Shu, & Otter 6

208 is assumed to be 30,000 ksi and the normal condition beam is built from six 136 RE rails per span.
209 Other inertial properties used in modeling are listed in Table 1. The rails, ties, and bridge span
210 were modeled as Euler-Bernoulli beams.
211
212 TABLE 1 Body Inertial Properties NUCARS® Track Model
Body Mass Roll Pitch Yaw
Properties lbs.-sec^2/in lbs.-sec^2-in
Rails 31.772 40.48 0.0 0.0
Ties 0.578 503.41 6.26 504.95
Bridge Beams Normal
9.855 549.15 23,693.60 23,213.50
Conditions
Bridge Beams Weak
4.928 81.62 11,655.00 11,607.90
Conditions
213 The connections within the track model have vertical stiffness k1 and k3 of 1 × 106, k2
214 equal 0.8 × 106, and k4 of 2 × 108. Vertical damping of 400 lbs.-sec./inch was used for all
215 connectors. The lateral stiffness of 2.0E5 and lateral damping of 100 lbs.-sec./inch was used along
216 with vertical parameters.
217 Simulation Results versus Measurements
218 The simulations were run with test train speeds ranging from 10 mph to 40 mph. To better simulate
219 actual conditions, track irregularities were added to the model. The NUCARS® output included
220 vertical rail positions and deflections. The NUCARS® model shows reasonable agreement with the
221 test data, indicating that the model should be useful for predicting responses and outputs for a test
222 train with a track deflection test car.
223 Figures 6 and 7 compare simulations of the test train over the BDTF with test data for Test
224 1 and Test 3, respectively. The shape and magnitude of each of the test results and simulations are
225 very similar. The discrepancy between measurements and modeling may be related to unknown
226 conditions of soil properties and track geometry on the approach to the bridge and stiffness of the
227 bridge supports. Changes to the stiffness of the bridge supports influences the overall deflection of
228 the bridge. The parameters used in this study were adjusted based on the track modulus
229 measurements from previous research studies on this topic. (7)
230
231
 Rakoczy, Shu, & Otter 7

Deflection vs Time
Field Measurements - Exterior Span
Field Measurements - Center Span
0.1 NUCARS Modeling - Exterior Span
NUCARS Modeling - Center Span
0

-0.1
Displacement, in

-0.2

-0.3

-0.4

-0.5

-0.6
100 150 200 250 300 350

Time, sec
232
233 FIGURE 6 Deflection Histories for Center Span – Test 1
234
Deflection vs Time
Field Measurements - Exterior Span
Field Measurements - Center Span
0.1 NUCARS Modeling - Exterior Span
NUCARS Modeling - Center Span
0

-0.1
Displacement, in

-0.2

-0.3

-0.4

-0.5

-0.6
100 150 200 250 300 350

Time, sec
235
236 FIGURE 7 Deflection Histories for Center Span – Test 3
237
238 Application of the NUCARS® Model
239 NUCARS® was used to model typical freight train equipment as well as a track and bridge system.
240 The previous section showed that the predicted dynamic responses of bridges from the NUCARS®
241 model closely matched test results. (1)(2) The NUCARS® simulations can also be used to predict
242 the dynamic performance of the vehicle on the bridge.
 Rakoczy, Shu, & Otter 8

243 Therefore, simulations were performed to predict results of additional accelerometers

244 installed on the NUCARS® model of the IFC. The simulated transducer outputs were analyzed to
245 predict changes in bridge performance. The analysis provided information on how the additional
246 onboard transducers might help detect changes in the bridge. More precisely, the analysis may
247 provide potential locations of transducers on the vehicle that are likely to detect bridge conditions.
248 Figures 8 and 9 present example acceleration and deflection histories for the lead sideframe of the
249 IFC simulated in NUCARS®. The results showed noticeable differences between the normal
250 condition of the bridge (blue line) and the weak stringer (red line). The acceleration data showed
251 differences in both amplitude and frequency content while the displacement data showed only
252 large differences in displacement. Note that with an actual onboard system, vehicle motions would
253 probably be measured with accelerometers. Displacements would need to be derived from the
254 acceleration data. Results of the modeling were analyzed to identify the most promising locations
255 to install transducers on an onboard measurement system.

256
257 FIGURE 8 NUCARS® Acceleration Histories for Lead Sideframe
258

259
260 FIGURE 9 Displacement Histories for Lead Sideframe
261
262 Simulation results were compared to the field measurements. Figure 10 presents histories
263 of measured displacement and simulated displacement. The displacement signal from the
264 measurements did not exactly overlay with NUCARS® simulated results; however, the overall
265 amplitudes had similar magnitude at all the major features. Taking into account that a generic car
 Rakoczy, Shu, & Otter 9

266 was used in the simulation instead of an exact model of the IFC, the comparison of results are
267 considered a good validation of the NUCARS® modeling.
268 The best locations for the onboard system were chosen based on the location of
269 accelerometers already installed on the IFC and potential improvement that could be easy to
270 implement. The onboard technology to detect changes in the bridge behavior should distinguish
271 changes in the left and right sides of the bridge. Previous tests with the IFC showed potential for
272 using carbody accelerometers and sideframe accelerometers.
Vertical Displacement of Left Side Frame
Measurements NUCARS Pier
0.8

0.6

0.4
Displacement, in

0.2

-0.2

-0.4

-0.6

-0.8
50 100 150 200 250 300

Distance, ft
273
274 FIGURE 10 Comparison of Displacement Histories for
275 Sideframe – Measurements vs. NUCARS Simulation
276
277 Many different locations of accelerometers were simulated, and the results were examined.
278 Then, new locations and methods of installation were proposed. Figure 11 shows the locations of
279 the new accelerometers. On the sideframes, two accelerometers were installed with foam to reduce
280 high frequency input and the other two are installed without foam.

281
282 FIGURE 11 Locations of New and Old Accelerometers on the IFC
283
284 Test Results from New Accelerometers
 Rakoczy, Shu, & Otter 10

285 Tests conducted in August 2014 showed the new configuration of accelerometers at sideframes
286 have the biggest potential to collect useful data. Also, the old configuration of accelerometers was
287 not as effective as the new locations, because it did not collect data from the left and right sides —
288 only from the centerline of the carbody. Figures 12 and 13 show results of carbody displacements
289 from the new transducers.
290 The displacements show that the left side of the bridge was weaker during Tests 2 and 3.
291 Although the differences were not as clear for the right side, results showed that during Test 3 the
292 center span was weaker.
293 This can be seen by looking at the relative displacement (dashed red line) between Piers 2
294 and 3.
Car Body Left A-end Displacement
Test 1 Test 2 Test 3
Test 2 - Test 1 Test 3 - Test 1 Pier
0.8

0.6

0.4
Displacement, in

0.2

-0.2

-0.4

-0.6

-0.8
100 150 200 250 300 350

Distance, ft
295
296 FIGURE 12 Comparison of Carbody, Left side, A-end Displacement for
297 Different Bridge Conditions
298
Car Body Left A-end Displacement
Test 1 Test 2 Test 3
Test 2 - Test 1 Test 3 - Test 1 Pier
0.8

0.6

0.4
Displacement, in

0.2

-0.2

-0.4

-0.6

-0.8
100 150 200 250 300 350

Distance, ft
299
300 FIGURE 13 Comparison of Carbody, Right side, A-end Displacement for
301 Different Bridge Conditions
 Rakoczy, Shu, & Otter 11

302
303 A significant improvement was noticed between results from the old configuration and the
304 new locations of accelerometers placed on the sideframe. Almost no difference was found between
305 new configurations of accelerometers with and without the foam mounting. Therefore, only
306 displacement results from the new locations without the foam are presented in Figure 14 for the left
307 sideframe and in Figure 15 for the right sideframe. The relative displacement (dashed lines)
308 between Pier 2 and Pier 3 shows that Test 2 and Test 3 have a weak span on the left side (Figure
309 14). Test 3 also shows a weak span on the right side (Figure 15).
Left Sideframe Displacement
Test 1 Test 2 Test 3
Test 2 - Test 1 Test 3 - Test 1 Pier
0.6

0.4
Displacement, in

0.2

-0.2

-0.4

-0.6
100 150 200 250 300 350

Distance, ft
310
311 FIGURE 14 Comparison of Left Sideframe Displacement for Different Bridge Conditions
312
Right Sideframe Displacement
Test 1 Test 2 Test 3
Test 2 - Test 1 Test 3 - Test 1 Pier
0.6

0.4
Displacement, in

0.2

-0.2

-0.4

-0.6
100 150 200 250 300 350

Distance, ft
313
314 FIGURE 15 Comparison of Right Sideframe Displacement for Different Bridge Conditions
315
316 CONCLUSIONS AND FUTURE RESEARCH
317 NUCARS® was used successfully to evaluate vehicle-bridge interaction presented in this study.
318 The modeling and validation of the bridge performance was described in detail. Additional
319 simulations and tests on the BDTF that were conducted in August 2014 using an onboard
 Rakoczy, Shu, & Otter 12

320 measurement system with transducers and locations derived from the NUCARS® simulation
321 results were presented.
322 The main purpose of this paper is to present simulation and test results of a freight car and
323 locomotive on a NUCARS® three-layer track model developed to analyze a railway bridge at the
324 BDTF. A reliable, vehicle-based system to evaluate changes in bridge condition over time can
325 assist with the inspection of hundreds or thousands of railroad bridges. Test and NUCARS®
326 simulation results were used to improve the onboard measurement capability of such a system. The
327 new configuration of accelerometers provided useful results for bridge condition evaluations. First,
328 all data from accelerometers was post-processed to develop displacements from the IFC
329 accelerations. Then, the results from different bridge conditions were compared and relative
330 deflections were calculated. Sideframe displacements clearly identified a change in bridge
331 condition. Carbody displacements also showed good correlations with bridge conditions, but they
332 were less definitive than sideframe displacements.
333 The NUCARS® software can be used successfully to evaluate vehicle-bridge interaction.
334 The advantages of using NUCARS® over other software for analysis of railway bridges are as
335 follows:
336 • NUCARS® can accurately model the dynamic responses of the vehicle including
337 wheel/rail interaction.
338 • The track model includes multi-body flexible rail, ties, and bridge components as well
339 as connections between them.
340 • Track connection parameters, such as soil properties on the bridge approach, can be
341 estimated based on track modulus measurement data.
342 • NUCARS® has the potential to be used for development of an onboard structural
343 integrity inspection system for bridges. Further tests and analyses were conducted in
344 August 2014 and the findings are presented in a second Technology Digest of this
345 study.(2)
346
347 NUCARS® can only model the flexible components as simple beams; therefore, it is
348 recommended to use the NUCARS® track model for dynamic evaluation of a simple bridge
349 structure. For detailed modelling of complex flexible structures such as truss bridges, development
350 of an improved version of NUCARS® model with finite element analysis modal inputs is
351 recommended.
352
353 REFERENCES
354 1. Rakoczy, A., X. Shu, and D. Otter. March 2015. “Vehicle/Bridge Interaction Modeling and
355 Validation – Part 1.” Technology Digest TD-15-004, AAR/TTCI, Pueblo, Colorado.
356 2. Rakoczy, A., S. Shu, and D Otter. March 2015. “Vehicle/Bridge Interaction Modeling and
357 Validation – Part 2.” Technology Digest TD-15-005, Association of American Railroads,
358 Transportation Technology Center, Inc., Pueblo, Colorado.
359 3. Rakoczy, A. and D. Otter. March 2015. “Bridge Condition Evaluation Using Track Modulus
360 Measuring Systems.” Technology Digest TD-15-006, AAR/TTCI, Pueblo, Colorado.
 Rakoczy, Shu, & Otter 13

361 4. Rakoczy, A. and D. Otter. March 2015. “Bridge Condition Evaluation Using Track Geometry
362 Systems.” Technology Digest TD-15-007, AAR/TTCI, Pueblo, Colorado.
363 5. Otter, D. and R. Joy. February 2013. “Feasibility of Detecting Weak Bridge Stringers Using
364 Onboard Systems.” Technology Digest TD-13-003, Association of American Railroads,
365 Transportation Technology Center, Inc., Pueblo, Colorado.
366 6. Wilson, N.G., "Locomotive Control Compartment Safety Program, Locomotive Heavy Axle
367 Load Tests," FRA Report (DOT/FRA/ORD-93-02). June 1993.
368 7. Li, D., J. Elkins, H. Wu, H., S. Singh. May 1999. “Characterization of Track Stiffness and
369 Damping Parameters.” Research Report R-930, Association of American Railroads,
370 Transportation Technology Center, Inc., Pueblo, Colorado.
371