1- create a PostgreSQL + Postgris local DB

2- download the osm (brf format) and import in the DB
example: planet_7.63_49.178_3c305aff.osm.pbf

3- generate a temporary noise table 'noise_tmp' containing the noise_factors

SELECT     losmid, lhighway, sum(noise_factor) sum_noise_factor 
into table noise_tmp 
from (
SELECT
    l.osm_id losmid, l.highway lhighway,
    case 
    when p.highway in ('motorway', 'motorway_link','trunk','trunk_link') then     	sum(st_area(st_intersection(ST_Buffer(l.way, 30), ST_Buffer(p.way, 50)))
       	 / st_area(ST_Buffer(l.way, 30)))
    when p.highway in ('primary','primary_link') then
	sum(st_area(st_intersection(ST_Buffer(l.way, 30), ST_Buffer(p.way, 50))) 
        	/ (2 * st_area(ST_Buffer(l.way, 30))))
    when p.highway in ('secondary') then 
	sum(st_area(st_intersection(ST_Buffer(l.way, 30), ST_Buffer(p.way, 50))) 
        	/ (6 * st_area(ST_Buffer(l.way, 30))))
    end
as noise_factor
FROM planet_osm_line AS l
INNER JOIN planet_osm_line AS p ON ST_Intersects(l.way, ST_Buffer(p.way,50)) 
WHERE (l.highway is not null or l.route='ferry’)
and p.highway in ('motorway’, 'motorway_link','trunk','trunk_link','primary','primary_link','secondary')
GROUP BY losmid, lhighway, p.highway
order by noise_factor desc)
as abcd
GROUP BY losmid, lhighway
order by sum_noise_factor desc;


4- generate a file with the osm_id + noise class
(as admin \o my_noise.log)

SELECT     'osm_id', losmid, ' noise_factor',
case
  when y.sum_noise_factor < 0.1  then null
  when y.sum_noise_factor < 0.3  then '1'
  when y.sum_noise_factor < 0.6  then '2'
  when y.sum_noise_factor < 1    then '3'
  when y.sum_noise_factor < 1.5  then '4'
  when y.sum_noise_factor < 4    then '5'
  else '6' 
end as noise_factor, ' ende'
from noise_tmp y where sum_noise_factor > 0.1;


5- convert the osm map from pbf to osm format
osmconvert64.exe planet_7.63_49.178_3c305aff.osm.pbf > planet_7.63_49.178_3c305aff.osm

6- enrich the osm with new tags (example   <tag k="estimated_noise_class" v="5"/> )
==> use the perl script "add_noise_tag.sh"
(input are my_noise.log and planet_7.63_49.178_3c305aff.osm, output is 'modif.osm')

7- generate the rd5 files
standard mapcreation using 'modif.osm' as input AND the new lookups.dat
(do not forget srtm data and estimated_traffic_class by need) 
( "process_modif.sh" )

remark: By adding the "estimated_noise_class" tag the size of the rd5 files increases by nearly 2%

8- extend the lookups.dat

add in "---context:way "


estimated_noise_class;0000000001 1
estimated_noise_class;0000000001 2
estimated_noise_class;0000000001 3
estimated_noise_class;0000000001 4
estimated_noise_class;0000000001 5
estimated_noise_class;0000000001 6


9- extension in the brouter scripts (brf)

9-a add a new variable:
...
assign   consider_noise       = true   # %consider_noise% | set to true to consider costs for noise | boolean
...

9-b add a penalty for noise
example:
...
assign noise_penalty 
   switch consider_noise
     switch estimated_noise_class=   0
     switch estimated_noise_class=1  0.3 
     switch estimated_noise_class=2  0.5
     switch estimated_noise_class=3  0.7
     switch estimated_noise_class=4  1
     switch estimated_noise_class=5  1.2
     switch estimated_noise_class=6  1.5 0 0
...
assign costfactor
  add noise_penalty
...

 
start tests....example (fastbike-verylowtraffic.brf)
   /#map=12/49.8197/9.1224/osm-mapnik-german_style&lonlats=9.077396,49.867285;9.172485,49.783066

 
(consider to manually change the options "consider_noise" and "consider_elevation" in the profile to examine the impacts)



performance
=============

germany map (4 GB as pbf, 73 GB as osm)
==> create/calculate "noise" data with Postgreesql  ==> 20 minutes

==> add noise tags in osm		==> 15 minutes