Appendix 4.12 Validation of Rail Noise Assessment Models

Model Validation CadnaA Model

Model validation has been conducted based on the noise measurement results as presented in Appendix 4.11. Model picture is presented in Drawing 1 below.

Drawing 1 Model Picture

The noise measurement point as shown in Appendix 4.11 has been selected for the model calibration. The following table summarized the measurement and prediction results. As shown in the results table, thedifferences between the measurement and prediction results would be 1 to 2 dB(A).

Measurement Result Prediction ResultDifferences

betweenPrediction andMeasurement

Differenceswithout 3dBTrack wearcorrection

Event

MeasuredLeq

passby [1]Sample

Duration

MeasuredTrainSpeed

MeasuredSEL

MeasuredLeq, 30min

ReceiverPoint Heightfrom Track

ReceiverPoint

Setbackfrom Track

No. ofTrain in30min

No. ofTrainin 1hour

ModelledTrain

Speed

ModelSourceTerm

SEL [2]Predicted Leq,

30mindB(A) Sec km/h dB(A) dB(A) m m Set/30min Set/hr km/h dB(A) dB(A) dB(A) dB(A)

Train No.1 75.0 41.0 19 91.2 58.6 7.5 9.0 1 2 19 89.0 63.7 5 2Train No.2 75.4 43.3 18 91.8 59.2 7.5 9.0 1 2 18 89.0 63.3 4 1Train No.3 74.4 45.2 16 90.9 58.4 7.5 9.0 1 2 16 89.0 62.2 4 1All Trains Log sum of SEL= 96.1 63.5 7.5 9.0 3 6 19, 18, 16 89.0 67.8 4 1

Notes:

Receiver Point

[1] Measurement results refer to Test of Airborne Noise Train Passby of TCL/AEL in Depot.[2] Model source term SEL 89.0dB(A) is reference to train speed at 22km/h.[3] Prediction includes number of car adjustment from 8 cars to 10 cars for AEL train and 3dB Track wear correction.

As shown in above table, the differences between the model results and measurement are only 1 to 2 dB(A). A further review has been conducted to establish the validation criteria and the table below summarizesthe differences between the prediction and measurements reported in various standards/ reports/ journals.

Reference Differences between Prediction and Measurement, dB(A)

Railway rolling noise prediction: field validation and sensitivity analysis[1] <2Using the railway noise model (RWNM) for detailed noise analyses at some interesting receptor locations[2] 0.4 1.5Acoustics Attenuation of sound during propagation outdoors Part 2: General method of calculation[3] < 1 for mean height of the source and receiver between 5m and 30m at a separation distance less

than 100m< 3 for mean height of the source and receiver is less than 5m at all separation distance

Railway applications Acoustics Measurement of noise emitted by railbound vehicles[4] <2Experimental validation of the TWINS prediction program for rolling noise, Part 2: Results [5] <2Development of prediction model for environmental noise for Korean railway[6] <3

[7] <3Notes:[1] 127 (2013)[2] W. Wu & ation Research Board, A1 f04 Summer Meeting (2000)[3] Attenuation of sound during propagation outdoors Part 2: Gene -2 (1996)[4] Acoustics[5] 137 147 (1996)[6] ternational Union of Railways (2006)[7] tment for Environment Food and Rural Affairs (2007)

Based on the various international standards / reports / journals as presented in above table, the differences between measurement and prediction results of only 1 to 2dB(A) indicate that the rail noise model asadopted in the EIA study is considered to be valid approach to predict rail noise levels at NSRs.

Appendix 4.12 - Validation of Rail Noise Assessment ModelModel Validation -An in-house train noise model as developed by Ove Arup & Partners Hong Kong Ltd

MTR Corporation Limited Proposed Comprehensive Residential and Commercial Development atop SiuHo Wan Depot

G:\ENV\PROJECT\248118\12 REPORTS DELIVERABLES\017 EIA REPORT\03 DRAFT FINAL V1\CH4 - NOISE\APPENDIX\WORKING\APPENDIX 4.20\DISTANCE CORRECTIONVALIDATION.DOCX

Page 1

Rail Airborne Noise AssessmentModel Validation

A.1.1.1 A total of three measurement points at Ta Pang Po Offtake andPigging Station and Sunny Bay Salt Water Pumping Station havebeen selected for the model calibration. Table below summarizes themeasurement and predicted results.Table A1.1 Railway noise validation

Location IDMeasurement[1] Prediction Difference,

dB(A)Leq 30 mins, dB(A)[2]

1 62.9 65.6 2.72 63.4 67.6 4.23 65.7 70.8 5.1

Note:[1] The measurement at Ta Pang Po Offtake and Pigging Station and Sunny Bay Salt

Water Pumping Station has been taken in free field condition near a straightballast track section whereas the measurements data include TCL and AELtrains running at about 120-130km/h.

[2] Both measurement and prediction are in free field condition.

A.1.1.2 It can therefore be seen from the above table that, the differencevaries from 4 to 6 dB(A). A review has been conducted to establishthe validation criteria and table below summarizes the differencesbetween the prediction and measurements reported in variousstandards/ reports/ journals.Table A1.2 Differences between Prediction and Measurement

ReferenceDifferences between Prediction and

Measurement, dB(A)Railway rolling noise prediction:field validation and sensitivity analysis[a] < 2

Using the railway noise model (RWNM)for detailed noise analyses at someinteresting receptor locations[b]

0.4 1.5

Acoustics Attenuation of sound duringpropagation outdoors Part 2: Generalmethod of calculation[c]

< 1 for mean height of the source andreceiver between 5m and 30m at aseparation distance less than 100m

< 3 for mean height of the source andreceiver is less than 5m at all

separation distanceRailway applications AcousticsMeasurement of noise emitted byrailbound vehicles[d]

< 2

Experimental validation of the TWINSprediction program for rolling noise, Part2: Results[e]

< 2

MTR Corporation Limited Proposed Comprehensive Residential and Commercial Development atop SiuHo Wan Depot

G:\ENV\PROJECT\248118\12 REPORTS DELIVERABLES\017 EIA REPORT\03 DRAFT FINAL V1\CH4 - NOISE\APPENDIX\WORKING\APPENDIX 4.20\DISTANCE CORRECTIONVALIDATION.DOCX

Page 2

Reference Differences between Prediction andDevelopment of prediction modelfor environmental noise for Koreanrailway[f]

< 3

Additional railway noise source terms

[g]< 3

Note:

prediction: field validation and sensitivity analRail Transportation, Vol. 1, pp. 109 127 (2013)

noise analyses at some interesting receptor locations, Transportation ResearchBoard, A1 f04 Summer Meeting (2000)

Attenuation of sound during propagation outdoors Part 2:-2 (1996)

Acoustics Measurement of noise emitted byrailbou

Sound and Vibration, Vol 193, pp. 137 147 (1996)[f]

(2006)[g] A.E.J. Hardy, R.R.K. Jones & C.E. Wright, Additional railway noise source

Food and Rural Affairs (2007)

A.1.1.3 From Table A1.2, the difference between measurement andprediction is about 3dB(A) according to various internationalstandards / reports / journals. As the current difference varies from 3to 5dB(A), the train noise model is considered a conservativeapproach to predict train noise levels on NSRs as a larger noisesource term has been adopted compared with the measured one.

Barrier Correction Validation(Extracted from Tung ChungNew Town Extension(TCNTE) (AEIAR-196/2016))

Note:Retrofitting work has been carried out on the existingbarrier, the retrofitted barrier is 7m above the rail head and itis not practical to carry out noise measurement at boundarybetween shadow zone and illuminated zone. Therefore,barrier correction validation has been extracted from theapproved EIA report for Tung Chung New Town Extension(AEIAR 196/2016)

Civil Engineering and Development Department Tung Chung New Town Extension

Rail Airborne Noise Assessment

Model Validation

A.1.1.1.1 Two locations with a total of five measurement points in North Lantau(One in the reclamation area of Tung Chung East and the other betweenTa Pang Po Offtake and Pigging Station and Sunny Bay Salt WaterPumping Station) have been selected for the model calibration. Tablebelow summarizes the measurement and predicted results.

Table A1.1 Railway noise validation

LocationsMeasurement[1] Prediction

Difference, dB(A)Leq 30 mins, dB(A)[2]

Reclamation Area at East of Tung Chung Town#1 70.7 74.8 4.1#2 62.6 67.0 4.4#3 63.9 69.5 5.6#4 64.4 70.1 5.7

Note:[1] The measurement in the reclamation area at east of Tung Chung Town has

been taken in free field condition near a straight ballast track section whereasthe measurements data include TCL trains running at about 100-110km/h.There is a 4.4m barrier between the track and the measurement locations.

[2] Both measurement and prediction are in free field condition.

A.1.1.1.2 It can therefore be seen from the above table that, the difference variesfrom 4 to 6 dB(A). A review has been conducted to establish thevalidation criteria and table below summarizes the differences betweenthe prediction and measurements reported in various standards/ reports/journals.

Table A1.2 Differences between Prediction and Measurement

Reference Differences between Prediction andMeasurement, dB(A)

Railway rolling noise prediction: fieldvalidation and sensitivity analysis[a]

< 2

Using the railway noise model (RWNM) fordetailed noise analyses at some interestingreceptor locations[b]

0.4 – 1.5

Acoustics – Attenuation of sound duringpropagation outdoors – Part 2: Generalmethod of calculation[c]

< 1 for mean height of the source and receiverbetween 5m and 30m at a separation distance lessthan 100m

< 3 for mean height of the source and receiver isless than 5m at all separation distance

219844-REP-121-00 | |\\HKGNTS22\acoustic\env\project\219844-70\12 Reports Deliverables\43 EIA Report\03 Final Draft \Appendices\4. Noise\Model files\Appendix 4.20 model validation text.docx

Page 1

Civil Engineering and Development Department Tung Chung New Town Extension

Reference Differences between Prediction andMeasurement, dB(A)

Railway applications – Acoustics –Measurement of noise emitted by railboundvehicles[d]

< 2

Experimental validation of the TWINSprediction program for rolling noise, Part 2:Results[e]

< 2

Development of prediction model forenvironmental noise for Korean railway[f]

< 3

Additional railway noise source terms for“Calculation of Railway Noise 1995”[g]

< 3

Notes:[a] S. Jiang, P.A. Meehan, D.J. Thompson & C.J.C. Jones. “Railway rolling noise prediction: field validation

and sensitivity analysis”, International journal of Rail Transportation, Vol. 1, pp. 109 – 127 (2013)[b] W. Wu & S. Rosen. “Using the railway noise model (RWNM) for detailed noise analyses at some

interesting receptor locations, Transportation Research Board, A1 f04 Summer Meeting (2000)[c] “Acoustics – Attenuation of sound during propagation outdoors – Part 2: General method of calculation”,

International Standard ISO 9613-2 (1996)[d] “Railway applications – Acoustics – Measurement of noise emitted by railbound vehicles”, British

Standards BS EN ISO 3095 (2005)[e] D.J Thompson, P. Fodiman & H. Mahe. “Experimental validation of the TWINS prediction program for

rolling noise, Part 2L Results”, Journal of Sound and Vibration, Vol 193, pp. 137 – 147 (1996)[f] J.H. Cho, J.C. Kim & H.I. Koh, “Development of prediction model for environmental noise of Korean

railway”, International Union of Railways (2006)[g] A.E.J. Hardy, R.R.K. Jones & C.E. Wright, Additional railway noise source terms for “Calculation of

Railway Noise 1995”, Department for Environment Food and Rural Affairs (2007)

A.1.1.1.3 From Table A1.2, the difference between measurement and predictionis about 3dB(A) according to various international standards / reports /journals. As the current difference varies from 4 to 6dB(A), the trainnoise model is considered a conservative approach to predict train noiselevels on NSRs as a larger noise source term has been adoptedcompared with the measured one.

219844-REP-121-00 | |\\HKGNTS22\acoustic\env\project\219844-70\12 Reports Deliverables\43 EIA Report\03 Final Draft \Appendices\4. Noise\Model files\Appendix 4.20 model validation text.docx

Page 2

Project: Tung Chung New Town ExtensionProject no.: 219844-70Title: Leq at measurement locations