Memo Date: Tuesday, November 17, 2015 Project: To: From: Subject: Lindley Avenue to Balboa Boulevard Vibration Study LA Metro Elliott Dick and Tim Casey Vibration Measurement Results Summary of Vibration Study and Findings HDR completed a vibration study with the goal of collecting ground-borne vibration measurements at ten locations to characterize existing vibration levels during train pass-by events. The Federal Transit Administration (FTA) Noise and Vibration Assessment Manual (2006) provided the guidance for performing this study and this memorandum presents the findings. The Federal Railroad Administration also uses FTA s methods to assess train noise and vibration. Following FTA s guidance, key terms used in the vibration study include: GBV, Ground Borne Vibration, includes perceptible movement of building floors, interference with vibration sensitive instruments, rattling of windows, etc. GBN, Ground Borne Noise, typically perceived as a low frequency rumbling sound. VdB, vibration decibels. The FTA differentiates vibration-sensitive land uses into three categories. Category 2 includes residences and buildings where people normally sleep. Based upon the frequency of train traffic in this corridor, the limits for Occasional Events and Category 2 land-uses are 75 VdB. The background vibration level in typical residential areas is usually 50 VdB or lower, well below the threshold perception for humans, which is around 65 VdB. Existing vibration levels in the project area were gathered at ten receptor properties; nine of the properties were adjacent to the rail corridor at distances ranging from 65 feet to 275 feet from the existing track. The vibration measurements were taken at multiple locations on each property for different train events (Amtrak, Metrolink, and freight). The measured vibration at the primary residence on each property ranged from 52-80 VdB. While there is a lot of variation due to differing speeds and soil types, most of the vibration measurements decrease with distance from the train tracks. The highest measured vibration was often due to freight trains. A regression line was developed using measurement data at each residence. This line was used to estimate ground-borne vibration levels at a distance that is representative of the distance to the proposed double track. At those distances vibration levels are projected to change approximately 1-2 VdB due to vibration energy naturally dissipating in the soils. HDR also measured train-induced ground-borne vibration at a location in Glendale, CA. (outside of the study area). The goal of these measurements was to quantify ground-borne vibration levels when two trains passed each other. Two trains did not pass each other while HDR 1
performed those measurements; therefore that data are not included in this memo. Those measurements lasted approximately six (6) hours. The results of the vibration measurements indicate that at many of the adjacent land-uses, the existing conditions already exceed the FTA Category 2 impact levels at the residence on each property. Because existing vibration already exceeds the limits, the allowable increase-overexisting vibration level is no more than 3 VdB. The projected increase in ground-borne vibration due to the proposed double track varies from no increase to 2 VdB, and therefore it is not considered a vibration impact according to FTA. Introduction As requested by the Los Angeles County Metropolitan Transportation Authority (LA Metro), HDR measured vibration levels along an existing rail line in the Northridge neighborhood of Los Angeles, California. The rail line is shared by Metrolink, Amtrak passenger trains, and Union Pacific Railroad (UPRR) freight trains. LA Metro is proposing to convert the rail line from single track to double track. Ground-borne vibration measurements were performed at ten residences to characterize vibration levels during train pass-by events. At each residence, HDR measured ground-borne vibration levels at several locations (different distances from the rail line); these measurements provided insight on existing and future ground-borne vibration levels if the proposed double-track is constructed. Project Description In brief, the proposed project would add a second mainline track to one side of the existing single mainline track. Both the existing and new mainline tracks would be used for passenger and freight rail. The proposed second mainline track would be located to the south of the existing mainline between just west of De Soto Avenue and approximately 1360 west of Louise Avenue. The proposed track then transitions north of the existing track until approximately 340 east of Louise Avenue where the second mainline track will remain to the north of the existing track through Woodley Avenue. Ground Borne Vibration Vibration consists of rapidly fluctuating motions. However, human response to vibration is a function of the average motion over a period of time, such as one second. The root mean square (RMS) amplitude of a motion over a one-second period is commonly used to predict human response to vibration. For convenience, decibel notation is used to describe vibration relative to a reference quantity. The Federal Transit Administration (FTA) has adopted the notation VdB (for vibration decibels), which is decibels relative to a reference quantity of one microinch per second (10-6 in/s). Ground-borne vibration (GBV) can cause annoyance to building occupants or residents, or cause disruption at facilities that are vibration-sensitive, such as laboratories or recording studios. The effects of ground-borne vibration include perceptible movement of building floors, interference with vibration sensitive instruments, rattling of windows, and the shaking of items on shelves or wall hangings. Additionally GBV can cause the vibration of room surfaces 2
resulting in ground-borne noise (GBN). Ground-borne noise is typically perceived as a low frequency rumbling sound. In contrast to airborne noise, ground-borne vibration is not an everyday experience for most people. The background vibration level in residential areas is usually 50 VdB or lower well below the threshold of perception for humans, which is around 65 VdB. Levels at which vibration interferes with sensitive instrumentation can be much lower than the threshold of human perception, such as for medical imaging equipment or extremely high-precision manufacturing. Most perceptible indoor vibration is caused by sources within a building such as the operation of mechanical equipment, movement of people, or slamming of doors. Typical outdoor sources of perceptible ground-borne vibration are construction equipment, steel-wheeled trains, and traffic on rough roads, though in most soils GBV dissipates very rapidly. Vibration Evaluation Criteria This study only characterizes the existing vibration conditions. However the existing vibration levels are compared to some vibration criteria in order to provide a reference to how the measured existing levels could potentially be perceived. The FTA sets limits for ground borne vibration, as shown in Table 1 below. The FTA vibration impact criteria are used to predict future vibration impacts from transit operations. FTA identifies separate criteria for both ground-borne vibration (GBV) and groundborne noise (GBN). Ground-borne noise is often masked by airborne-noise; therefore, groundborne noise criteria are primarily applied to subway operations in which airborne noise is negligible. The FTA differentiates vibration-sensitive land uses into three distinct categories (similar but not identical to the noise-sensitive land-use categories). These categories are one factor for setting the vibration impact threshold. Category 1: Buildings where vibration normally imperceptible to humans would interfere with interior equipment or operations. Typically includes vibration-sensitive research and manufacturing facilities, vibration-sensitive research operations, and some vibrationsensitive areas of hospitals. Category 2: Residences and buildings where people normally sleep. Non-residential land-uses in this category include hotels and hospitals. Category 3: Institutional land uses with primarily daytime use. Typically includes schools, churches, other institutions, and quiet offices that do not have vibration-sensitive equipment, but still have the potential for activity interference. The basis for evaluating FTA vibration impact thresholds is the highest expected root mean square (RMS) vibration levels for repeated vibration events from the same source. The thresholds vary for different types of vibration sensitive land uses and the frequency of the events. 3
Land Use Category Table 1. FTA s General Assessment Criteria for Ground Borne Vibration and Ground Borne Noise Frequent Events a GBV Impact Levels Occasional Events b Infrequent Events c Frequent Events a GBN Impact Levels Occasional Events b Infrequent Events c Category 1 65 VdB 65 VdB 65 VdB N/A d N/A d N/A d Category 2 72 VdB 75 VdB 80 VdB 35 dba 38 dba 43 dba Category 3 75 VdB 78 VdB 83 VdB 40 dba 43 dba 48 dba Source: FTA, Transit Noise and Vibration Impact Assessment, 2006 a Frequent Events: More than 70 vibration events per day. Most rapid transit projects fall into this category. b Occasional Events: Between 30 and 70 vibration events of the same source per day. Most commuter trunk lines have this many operations. c Infrequent Events: Fewer than 30 vibration events per day. This category includes most commuter rail branch lines. d Vibration-sensitive equipment is generally not sensitive to ground-borne noise. There are currently 34 daily train events on the existing single track that will continue to use the proposed double track. Therefore, for discussion purposes, vibration measurements shown below can be compared with the Category 2 limits for Occasional Events (75 VdB) shown above. Existing Vibration Levels FTA provides specific guidance for evaluating future impacts where vibration is present due to an existing active rail corridor. The train traffic associated with this project will be moved from the existing track to a second parallel track, consistent with the Moving Existing Tracks scenario described in the FTA manual. This scenario has several considerations when determining impact. Predicted future vibration levels lower than existing vibration levels represents a project benefit, and so is not an adverse effect. If the existing vibration does not currently exceed the thresholds identified above, then predicted future project vibration levels are compared to the thresholds above to determine impact. If existing vibration already exceeds the thresholds identified above, then vibration is not assessed as an impact unless the predicted future project vibration levels create more than a 3 VdB increase over existing vibration levels. Approach This section describes the equipment used to gather the data and the measurement locations. Equipment The vibration measurement equipment consisted of accelerometers (sensors) connected to a digital data acquisition unit. The accelerometer was a piezoelectric transducer with a ground 4
spike to effectively couple the sensor to the ground, maximizing its ability to measure groundborne vibration. The data acquisition unit provided power and signal conditioning and was connected to a laptop to record the dynamic vibration acceleration signal from each sensor. The vibration acceleration signals were analyzed in HDR s office in the computation software, MATLAB. HDR set up MATLAB to integrate the raw acceleration signal into dynamic vibration velocity, and also filter the vibration signal in one-third octave bands from 0.8 Hz through 315 Hz using a slow meter setting (exponential average with a time constant of 1,000 ms). The maximum vibration levels are reported for each train pass-by event. Locations Ground-borne vibration has potential to induce vibration within a building structure which may be noticeable to the building occupants or interfere with particularly sensitive equipment, such as high-powered microscopes, medical imaging machines, or nanoscale manufacturing. The structures which may be affected by vibration due to this project are primarily residential structures, which do not generally house sensitive equipment. Human reactions to vibration indoors are generally in response to an event which produces vibration levels that are noticeably higher than usual background levels. Residents are most sensitive to vibration events during sleeping hours. HDR measured vibration at 10 residences whose owners agreed to allow measurements on their property. These measurement locations are listed in Table 2 and shown in Figure 1, below. Table 2. Measurement Location Summary Measurement Location ID ML1 ML2 ML3 ML4 ML5 ML6 ML7 ML8 ML9 ML10 Distance to the Existing Track a 114 ft. 128 ft. 108 ft. 1463 ft. 115 ft. 65 ft. 275 ft. 75 ft. 65 ft. 220 ft. a Approximate distance from existing track centerline to nearest residential structure on property. 5
Figure 1. Vibration Map hdrinc.com 801 S. Grand Avenue, Suite 500, Los Angeles, CA 90017-4633 6
The measurements reported in this study occurred outdoors in the ground, as representative vibration before it travels into the building structure. These measurements can be compared to the overall FTA s General Assessment criteria shown in Table 1. Data Analysis The vibration acceleration signals were analyzed in HDR s office in the computation software, MATLAB. HDR set up MATLAB to integrate the raw acceleration signal into dynamic vibration velocity, and also filter the vibration signal in one-third octave bands from 0.8 Hz through 315 Hz using a slow meter setting (exponential average with a time constant of 1,000 ms). The maximum vibration levels are reported for each train pass-by event. As the vibration travels through the ground, the levels dissipate with distance from the train tracks. HDR used the maximum vibration levels from the train pass-by events and the distances of the sensors to find the rate of dissipation through the soil at each measurement location. The rate of dissipation is the slope from a simple linear regression of the vibration versus distance data. Then HDR used the regression slope to determine the vibration at the foundation from the proposed set of train tracks in the corridor. Results This section presents results at each measurement location. Vibration Measurements at ML1 The discussion below presents the vibration measurements which were gathered at Measurement Location No. 1. The proposed second track is located 20 feet farther away from this receiver than the existing track. Figure 2 shows measurement results at this location and a regression line. Figure 2. Train Pass-by Measurements vs. Distance at ML1 Source: HDR Engineering, Inc. Many of the homes in this area are very close to the rail line right-of way line. However, due to the construction type of these homes, they are unlikely to have any significant foundations, 7
which will reduce the vibration coupling into the residence structure. Table 3 shows the vibration level measured at the home, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above. Train Description Table 3. Overall Train Vibration at ML1 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Amtrak 68 68 Metrolink 65 65 a Based upon regression line shown above. Vibration Measurements at ML2 The discussion below presents the vibration measurements which were gathered at Measurement Location No. 2. This is a single-family residence adjacent to the rail line right-ofway. The proposed second track is located 19 feet farther away from this receiver than the existing track. Figure 3 shows measurement results at this location and a regression line. Figure 3. Train Pass-by Measurements vs. Distance at ML2 Source: HDR Engineering, Inc. For many of the measured train pass-by events, the vibration levels are greatest at the residential structure, and a little lower at the rail-line right-of-way. There could be many different reasons for this, including differences in topsoil, unseen differences in the soil strata under the ground, or unseen underground boulders. For this property, the different vibration levels are very likely caused by the presence of retaining walls and concrete fences. The foundations of the fences or retaining walls could be shielding the sensors from the direct effects of the train vibration. In contrast, the larger foundation of the residential structure will catch more of this effect. Table 4 shows the vibration level measured at the home, and the level at a location that 8
is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 3). Train Description Table 4. Overall Train Vibration at ML2 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Amtrak 76 76 Metrolink 80 80 Freight 80 80 Metrolink 78 78 a Based upon regression line shown above. Vibration Measurements at ML3 The vibration measurements performed at Measurement Location No. 3 are provided below. This is a single-family residence adjacent to the rail line right-of-way. The proposed second track is located 19 feet closer to this receiver than the existing track. Figure 4 shows measurement results at this location and a regression line. Figure 4. Train Pass-by Measurements vs. Distance at ML3 Source: HDR Engineering, Inc. As is typically expected, the vibration levels are generally higher nearest to the tracks, and reduce with increasing distance from the tracks. Table 5 shows the vibration level measured at the residence, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 4). 9
Train Description Table 5. Overall Train Vibration at ML3 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Metrolink 71 73 Metrolink 74 76 Metrolink 75 77 Amtrak 78 79 a Based upon regression line shown above. Vibration Measurements at ML4 The vibration measurements performed at Measurement Location No. 4 are provided below. This residential structure is located nearly 1,500 feet from the tracks. The proposed second track is located 19 feet farther away from this receiver than the existing track. Ground-borne vibration levels are quite low at this location. Figure 5 shows measurement results at this location. Figure 5. Time-Trace for ML4 Source: HDR Engineering, Inc. The vibration peaks shown in the time-trace above were all associated with vehicles driving past the residence. It was not possible to discern a train pass-by event from these data. This is not 10
unexpected, as it would be extremely unusual to experience ground-borne train vibration almost 1,500 feet from the rail line. Vibration Measurements at ML5 The vibration measurements performed at Measurement Location No. 5 are provided below. This is a single-family residence adjacent to the rail right-of-way. The proposed second track is located 19 feet closer to this receiver than the existing track. Figure 6 shows measurement results at this location and a regression line. Figure 6. Train Pass-by Measurements vs. Distance at ML5 Source: HDR Engineering, Inc. As is typically expected, the levels are generally higher nearest to the tracks, and reduce with increasing distance from the tracks. Table 6 shows the vibration level measured at the residence, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 6). Train Description Table 6. Overall Train Vibration at ML4 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Metrolink 72 73 Amtrak 75 76 Metrolink 72 73 Amtrak 75 76 a Based upon regression line shown above. 11
Vibration Measurements at ML6 The vibration measurements performed at Measurement Location No. 6 are provided below. This is a single-family residence adjacent to the rail line right-of-way. The proposed second track is located 19 feet closer to this receiver than the existing track. Figure 7 shows measurement results at this location and a regression line. Figure 7. Train Pass-by Measurements vs. Distance at ML6 As is typically expected, the levels are generally higher nearest to the tracks, and reduce with increasing distance from the tracks. Table 7 shows the vibration level measured at the residence, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 7). Train Description Table 7. Overall Train Vibration at ML6 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Metrolink 77 78 Amtrak 79 80 Metrolink (decelerating) 70 71 Amtrak 79 80 Metrolink 78 79 a Based upon regression line shown above. There was one Metrolink train which stopped on the tracks just beyond the measurement location, so it was decelerating as it passed the measurement location. Therefore measured vibration velocities were low during this event. 12
Vibration Measurements at ML7 The vibration measurements performed at Measurement Location No. 7 are provided below. This is a single-family residence adjacent to the rail right-of-way. The proposed second track is located 19 feet farther away from this receiver than the existing track. Figure 8 shows measurement results at this location and a regression line. Figure 8. Train Pass-by Measurements vs. Distance at ML7 Source: HDR Engineering, Inc. As is typically expected, the levels are generally higher nearest to the tracks, and reduce with increasing distance from the tracks. Vibration from one of the Metrolink trains is nearly 20 VdB lower than the other Metrolink train as shown in the graph above (the data between 50 and 60 VdB on the bottom of the graph). This is most likely due to a slow travel speed as it passed, and influenced the slope of the regression line in Figure 8. Table 8 shows the vibration level measured at the home, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 8). Train Description Table 8. Overall Train Vibration at ML7 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Metrolink 70 69 Metrolink 52 52 Freight (1 locomotive, 3 cars, empty) 68 68 a Based upon regression line shown above. 13
Vibration Measurements at ML8 The vibration measurements performed at Measurement Location No. 8 are provided below. This is a single-family residential property adjacent to the rail right-of-way. There are two residences on this property. The primary residence is almost 200 feet from the tracks. An additional freestanding residential rental unit is located towards the back of the property, less than 100 feet from the tracks. The proposed second track is located approximately 15 feet closer to this receiver than the existing track. Figure 9 shows measurement results at this location and a regression line. Figure 9. Train Pass-by Measurements vs. Distance at ML8 Source: HDR Engineering, Inc. The graph shows the same trend of decreasing vibration vs. distance that was found at most other measurement locations. Although not labeled on Figure 9, the highest levels shown are due to freight trains. Table 9 shows the vibration level measured at the residence, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 9). Train Description Table 9. Overall Train Vibration at ML8 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Amtrak 76 77 Metrolink 77 78 Amtrak 78 79 Metrolink 76 77 Freight (4 locomotives, 53 cars) 80 81 Metrolink 77 78 a Based upon regression line shown above. 14
Vibration Measurements at ML9 The vibration measurements performed at Measurement Location No. 9 are provided below. This is a single-family residential property adjacent to the rail right-of-way. The primary residence was over 200 feet from the tracks. An additional guest-residence building is sometimes occupied. This secondary residence is located at the back of the property within a couple feet of the property line, which puts it approximately 65 feet from the tracks. The proposed second track is located approximately 15 feet closer to this receiver than the existing track. Figure 10 shows measurement results at this location and a regression line. Figure 10. Train Pass-by Measurements vs. Distance at ML9 Source: HDR Engineering, Inc. Table 10 shows the vibration level measured at the home, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 10). Train Description Table 10. Overall Train Vibration at ML9 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Metrolink 76 77 Freight (2 locomotives, 3 cars, empty) 75 76 a Based upon regression line shown above. Vibration Measurements at ML10 The vibration measurements performed at Measurement Location No. 10 are provided below. This is a single-family residential property adjacent to the rail right-of-way, where the primary residence was over 200 feet from the tracks. The proposed second track is located approximately 15 feet closer to this receiver than the existing track. Figure 11 shows measurement results at this location and a regression line. 15
Figure 11. Train Pass-by Measurements vs. Distance at ML10 Source: HDR Engineering, Inc. Table 11 shows the vibration level measured at the home, and the level at a location that is representative of the distance to the proposed double track; this projected vibration level is based on the regression line shown above (Figure 11). Train Description Table 11. Overall Train Vibration at ML10 Vibration at Residence Foundation, VdB Measured From Existing Track Predicted From Proposed Track a Amtrak 67 67 Metrolink 66 67 a Based upon regression line shown above. Pass-by Measurements at Double-Track HDR also performed ground-borne vibration measurements of individual train pass-by events at the end of Allen Avenue on the north side of the tracks in Glendale, California; the measurement duration was approximately six (6) hours. This location has double tracks with Ventura County Line traffic (which is the line that travels through the project area) and Antelope Valley Line traffic. The purpose was to capture a vibration event from a train on each track simultaneously. This event did not occur at this location, which could be an indication of the frequency that a simultaneous two train pass-bye will occur in the study area. These data are not presented here because it did not achieve its goal, and because the location is far from the project area the measurements of single-train events are not representative of vibration in the project area. 16
Aggregate Results by Distance HDR measured vibration levels during 63 train pass-bye events across all 11 measurement locations (the 10 receptors near the project area plus the reference measurements at Allen Avenue). The concurrent measurements at different distances in all of the 63 events accumulated to a total of 274 vibration measurements. A summary of the number of events at each measurement location is provided below. ID Table 12. Summary of Measured Train Events Metrolink Trains Amtrak Trains Freight Trains ML1 1 1 ML2 2 1 1 ML3 3 1 ML4 1 ML5 2 2 ML6 4 2 ML7 2 1 ML8 3 2 1 ML9 2 1 ML10 1 1 PB Allen Ave. 24 4 1 Total 45 14 5 All of the preceding paragraphs discussed the measurement data sorted on a per-location basis. The following page presents the measurement data sorted by train type (Metrolink, Amtrak, and freight). The figures below compares measured vibration levels vs. the distance from the train tracks. While there is a lot of variation due to differing speeds, differing soil types, and other factors, most of the data points generally follow a sloped line that indicated vibration levels decrease with distance from the source (i.e., trains). 17
Figure 12. All Train Pass-by Measurements Source: HDR Engineering 18
Discussion Train traffic on the existing rail line creates vibration levels that exceed FTA vibration criteria at homes in the study area, based on the measurement results shown above. FTA recognizes the potential for this in their discussion of existing vibration levels (FTA, 2006, 8.1.2). In circumstances where existing vibration levels exceed FTA vibration criteria, if a project shifts an existing track to add an additional track and the resulting vibration levels increase less than 3 VdB, FTA does not consider the project (addition of a second track) to cause a vibration impact. Measurement and analysis results in the tables above show that the proposed double track has potential to change ground-borne vibration levels by 1-2 VdB at each measurement location. In some locations analysis results indicate an increase; in other locations analysis results indicate a decrease. The projected increase in ground-borne vibration due to the proposed double track is less than 3 VdB, and therefore it is not considered a vibration impact according to FTA. Conclusion HDR measured ground-borne vibration at residences in the study area, during train pass-byes. Three types of trains operated on the existing rail line during these measurements: Metrolink, Amtrak, and Union Pacific Railroad freight trains. At each measurement location, HDR installed accelerometers (sensors) near the home, and at other locations on the property that were closer to and farther from the existing rail line. In this manner, HDR simultaneously measured traininduced ground-borne vibration at multiple distances from the rail line. In many cases, measurement results indicate that train-induced ground-borne vibration exceeds FTA criteria at distances from the rail line at which the residences exist. Measurement results at farther distances generally show lower vibration levels. Measurement results at closer distances generally show higher vibration levels. A regression line was developed using measurement data at each residence. This line was used to estimate ground-borne vibration levels at a distance that is representative of the distance to the proposed double track. At those distances vibration levels are projected to change approximately 1-2 VdB. Because existing vibration levels exceed FTA criteria at homes, and the projected change is less than 3 VdB, FTA does not consider the proposed double track to cause any vibration impacts. HDR also measured train-induced ground-borne vibration at a location in Glendale, CA (outside of the study area). The goal of these measurements was to quantify ground-borne vibration levels when two trains passed each other. Two trains did not pass each other while HDR performed those measurements; therefore those data are not included in this memo. References 2006 Transit Noise and Vibration Impact Assessment. Second Edition. (Federal Transit Administration Report Number FTA-VA-90-1003-06) 19