• Construction Noise
• Measurement
• Noise Barriers
• Noise Compatible Planning
• Regulation and Guidance
• Resources
• Tire Pavement Noise
• Traffic Noise Model
• Training

Ground and Pavement Effects using FHWA’s Traffic Noise Model® 2.5

3. TNM Version 2.5 Specific Pavement and Ground Type Results and Analysis

In this first set of modeling, specific pavement types were used but ground types not updated. The uncalibrated and calibrated results for this modeling are presented in Section 3.1. In the second set of modeling, the specific pavement types were used and the ground types were also updated for Sites 01MA, 02MA, 05CA, 09CA, 13CA, 15CA, 16MA, and 17 CT. The uncalibrated and calibrated results for this modeling are presented in Section 3.2. Analysis and discussion of the results are given in Section 3.3.

3.1 Specific Pavement Type

The uncalibrated results for the specific pavement modeling are shown in Figure 2. In this figure, the abscissa indicates the level of the measured 15-minute LAeq while the ordinate indicates the level of the predicted 15-minute LAeq. Each 15-minute datum is indicated by an orange x. A dashed blue line indicates the linear fit and solid green lines show the 95% confidence band for the linear fit. The solid black diagonal line indicates perfect (1 to 1) agreement between TNM predicted levels and measured data. Data above the black line indicate over-predictions while data below the black line indicate under-predictions.

Figure 2: Specific Pavement Study – Uncalibrated Results

The specific pavement modeling data were calibrated by using reference microphones as described in Section 2.3. The range of calibration values for each 15-minute datum and the average calibration value are given for each site in Table 6. A positive calibration value indicates an over-prediction and a negative calibration value indicates an under-prediction.

The calibrated results for the specific pavement modeling are shown in Figure 3. In this figure, the abscissa indicates the level of the measured 15-minute LAeq while the ordinate indicates the level of the predicted 15-minute LAeq adjusted by the calibration level. For further details about the graph parameter see description for uncalibrated results. Analysis and discussion of these results will be presented in Section 3.3, where results from modeling both specific pavements and updated ground types will also be discussed.

Figure 3: Specific Pavement Study – Calibrated Results

3.2 Combined Specific Pavement and Ground Type

In the second set of modeling, in addition to using specific pavement types for the models, select models had ground types updated as well. These are indicated in Table 5. The uncalibrated results for the specific pavement modeling are shown in Figure 4. The specific pavement and ground type modeling data were calibrated by using reference microphones as described in Section 2.3. The range of calibration values for each 15-minute datum and the average calibration value are given for each site in Table 7. The calibrated results for the specific pavement and ground type modeling are shown in Figure 5. For further details about the graph parameters see Section 3.1. Analysis and discussion of these results will be presented in Section 3.3, where results from modeling specific pavements (without updated ground types) will also be discussed.

Figure 4: Specific Pavement and Ground Type Study – Uncalibrated Results

Figure 5: Specific Pavement and Ground Type Study – Calibrated Results

3.3 Analysis and Discussion

Sections 3.1 and 3.2 presented the uncalibrated and calibrated results for two sets of modeling in TNM v2.5, 1) models with specific pavements and 2) models with specific pavements and updated ground types. This section compares the performance of these two sets of modeling with the 2004 addendum, which used Average pavement and had the original ground types.

3.3.1 Comparison of Calibration Factors

Emission levels for TNM were developed by averaging measurements for each of five vehicle types from several sites for each pavement type. The averaging process minimized the effects of site specific biases, such as site-to-site variation in vehicles, pavement, and meteorological effects (e.g. temperature inversions, wind, etc.). The use of these standardized emission levels, therefore represent an average level, but each modeled site may have site specific biases which cause the measured levels to deviate from the average, thus calibration can be useful for improving performance.

Table 8 shows the average calibration and calibration range for each site for three modeling sets with 1) Average pavement, 2) specific pavement, and 3) specific pavement with updated ground types. In general, calibration levels are about the same for each modeling set. For Sites 01MA, 02MA, 03MA, 08CA, 09CA, 15CA, 16MA, and 17CT, the use of specific pavements reduces the magnitude of the calibration. In such cases the use of specific pavements could be useful. For Sites 05CA, 06CA, 10CA (berm and open), 11CA, 12CA, 13CA, and 14CA, the use of specific pavements increased the magnitude of the calibration. Both results are reasonable, since the change due to the use of the specific pavements is on the order of 1 to 2 dB, while the bias due to pavements within the same class can be about 10 dB (see Appendix E of [Fleming 1995]). For example, if a PCC site is already over-predicting with Average pavement due to the site either being a “quiet” PCC, temperature inversions, quiet vehicles, etc, then modeling with PCC pavement will increase the calibration level. For the ground type updates implemented, no calibration improvements were observed from site to site as is expected since the reference microphones are close to the road and see little effect from a change in the ground type.

The calibration values are also categorized by site type in Table 9 for Average pavement, specific pavement, and specific pavement with updated ground types. In general, calibrations are about the same for each set of models. These results indicate that, even with the use of specific pavements and the best available ground type classifications, it is still advisable to calibrate data by using a reference microphone.

3.3.2 Linear Fit for Calibrated Data

The purpose of calibration is to account for specific characteristics of a modeled site which do not conform to standard values in TNM, for example calibration can be used to account for traffic which has higher or lower sound levels than typical traffic. Both the linearity of the fit between the measured and predicted data and the distribution of the differences between the measured and predicted data are useful characteristics to quantify how close the model is to measured data.

The relation of the linear fit to the line of perfect agreement is examined in Table 10 along with the width of the 95 percent confidence band in Table 11 for three model sets, 1) Average pavement, 2) specific pavement, and 3) specific pavement with updated ground types. (Graphs of these fits are given in Appendix A for the specific pavement models and in Appendix B for the specific pavement with updated ground types models.) Both the average difference and the average of the absolute value of differences are given in Table 10. The average difference represents the difference between the linear fit line and the perfect agreement line. The absolute value of differences indicates how well TNM is performing as a function of the amplitude of the over- and under-predictions. In general, TNM v2.5 performs well for all three modeling sets. The use of specific pavements improves the performance for open areas with soft ground, and the updated ground types further improve the results for both open areas with soft and hard ground types.

The average, maximum, and minimum values of the 95% confidence band, respectively are shown in Table 11. If all three values are small, and the maximum and minimum values are similar, this indicates that the data shows little variation in amplitude over a broad range of sound levels; as such, a similar data set (sound levels measured and predicted under the same conditions) would provide similar results. The difference between predicted and measured results can change over distance when inappropriate ground types are assigned. When the appropriate ground type is used, differences remain more consistent and thus it can be expected that the confidence band will be smaller than if inappropriate ground types are used. It can be seen in Table 11 that the updated ground types decrease the size of the confidence band, indicating less variation in the difference between the predicted and measured results.

3.3.3 Level Differences between Measurements and Model for Calibrated Data

In addition to the linearity of the fit between the predicted and measured data, the difference between the measured and predicted data is also useful in understanding TNM’s performance. Table 12 presents the average differences between measured data and calibrated TNM v2.5-predicted data modeled with Average pavement. The results are given as a function of microphone height, distance, ground type, and shielding (with or without a barrier). Table 13 presents the same information, except that specific pavements were used in the models. Table 14 presents the same information as well, except that specific pavements and updated ground types were used in the models.

It can be seen that each modeling set performed better at some locations / conditions than the other two. This indicates that specific pavements and updated ground types alone are not sufficient to quantify all of the variation that was observed in the 2004 validation addendum, however, it can be seen that for certain locations / conditions these updates did provide improvement. Specifically, specific pavements did offer improvements over Average pavement for open areas with acoustically soft ground at far distances, some improvement for open areas with acoustically hard ground at far distances, and some improvement for sites with barriers at near distances. Similarly, also updating ground types to use a more appropriate EFR value improved results for most open area sites (both acoustically hard and soft ground as well as near and far distances). By the appropriate use of calibration, specific pavements, and the correct selection of ground types it is possible to achieve small average differences between predicted and measured sound pressure levels.