PATREC Project1 5Nov17 · 2019-08-14 · Objecves • Assist/provide guidance on how to keep [email protected] moving (meeting growth), while being sustainable and liveable (alleviating congestion

Project  1  Addressing  Future  Uncertain0es  of  Perth  @3.5  Million:  What-­‐If  Scenarios  for  Mass  Transit      

 Precinct  Typology  as  Input  to  Inform  the  LU  and  PT  Planning      

Research  team:  Simon  Moncrieff,  Gary  McCarney,  Tristan  Reed,  Yuchao  Sun,  Cate  PaCson,  BreE  Smith,  Sharon  Biermann,  and  Doina  Olaru  

ObjecIves  •  Assist/provide guidance on how to keep P&[email protected] moving

(meeting growth), while being sustainable and liveable (alleviating congestion and promoting PT and active travel

•  Development of a typology to Inform  the  LU  and  PT  Planning  (Stage  1)    

•  Optimise use of the (integrated) transport network and best cater for accessibility and growth

•  Combination of various secondary data sources •  Review of several case studies across Australia •  Combination of data analysis techniques

–  Clustering of Place, Node, Background Traffic Functions –  Mapping all station precincts (and other locations) –  Exploring associations between boardings/alightings and the three

functions –  Quick calculators (using JTW) for stations with low patronage

Data  and  Analysis  –  Stage  1  

Place, Node, Background

Traffic

Clusters and Factors (based on indicators)  

Mapping station precincts on P, N,

BT functions  

Quantify patronage = f

(P, N, BT)  

Place  FuncIon  -­‐  Unidimensional  Mciver  

Claisebrook  

Maylands  Victoria  Street  

Glendalough   Mosman  Park  

East  Perth  Mt  Lawley  Victoria  Park  

Subiaco  

Daglish  North  Fremantle  

Meltham  

Carlisle  

West  Leederville  

Shenton  Park  

Clarkson   Claremont  Swanbourne  

Leederville  Oats  Street   Grant  Street  

Bayswater  

Showgrounds  

Currambine  Queens  Park  

Loch  Street  

Canning  Bridge  Bull  Creek  Warnbro  

Bassendean  

City  West  Butler  (formerly  Jindalee)  

Greenwood  WhiYords   CoEesloe  

Thornlie  

Cockburn  Central  Warwick  Murdoch  

Armadale  

Cannington  Rockingham  

Sherwood  

Gosnells  Burswood  SIrling   Joondalup  KarrakaEa  

Fremantle  Success  Hill  Beckenham  

Ashfield  Edgewater  Mandurah   Wellard  

KelmscoE  Challis  Welshpool  

Seaforth  Maddington  

Kenwick  East  Guildford  Midland  

West  Midland  (Woodbridge)    Guildford  

Kwinana  0  

10  

20  

30  

40  

50  

60  

70  

80  

90  

100  

0   10   20   30   40   50   60   70  

Only  46%  variance  Incomplete  data  for  the  8  cases  

Australian  Case  Studies  –  top  of  the  ladder  -­‐  Wolli  Creek,  Kelvin  Grove,  Chatswood  -­‐  Footscray,  Northshore,  Box  Hill    -­‐  Albion  Mill,  Yeerongpilly  

Place  Score  (1.6km  buffers)  vs    Distance  from  the  CBD  

0  

10  

20  

30  

40  

50  

60  

70  

80  

90  

100  

0   10   20   30   40   50   60   70   80  

Place  score  (rescaled  to  100

)  

d  from  CBD  (km)  Armadale  (1.6km)   Fremantle  (1.6km)   Joondalup  (1.6km)   Midland  (1.6km)  Mandurah  (1.6km)   Linear  (Armadale  (1.6km))   Linear  (Fremantle  (1.6km))   Linear  (Joondalup  (1.6km))  Linear  (Midland  (1.6km))   Linear  (Mandurah  (1.6km))  

Score  re-­‐scaled  0  to  100!  

Entropy  

Lower  entropy  for  Armadale  and  staIons  on  the  Mandurah  line  

Cluster 1

Cluster 4

Cluster3

Cluster2

Place  vs  Node  

Bassendean  Bayswater  

KelmscoE  

Oats  Street  Cannington  

Mandurah  

Claremont  

Clarkson  

Edgewater  

City  West   Claisebrook  

Warnbro  

Cockburn  Central  

Kwinana  

Subiaco  

Maylands  

Midland   Murdoch  

Bull  Creek  

Glendalough  

SIrling  

Greenwood  

Wellard  

Leederville  

Currambine  

McIver  

0  

10  

20  

30  

40  

50  

60  

70  

80  

90  

100  

0   10   20   30   40   50   60   70   80   90   100  

Place  

Node  >=1,000/day   <1,000/day  

Node  ‘dominant’  

Place  ‘dominant’  

5/11/17  

Example  Output  Regression  

R2-­‐adj  =0.759    DV = AM boardings Generator stations

    b   Std.  Error   Beta   t   Sig.   Tolerance   VIF  (Constant)   -­‐275.303   573.573       -­‐0.48   0.637          

IRSAD  percenIle   3.517   3.745   0.138   0.939   0.36   0.336   2.973  

%  Access  to  Jobs  45min  by  PT   23.724   7.695   0.433   3.083   0.006   0.371   2.698  

Bus  Services   4.034   1.335   0.323   3.022   0.007   0.641   1.561  

PnR  Supply   0.748   0.263   0.469   2.847   0.011   0.269   3.722  

Commercial   -­‐185.543   115.313   -­‐0.237   -­‐1.609   0.125   0.337   2.969  

EducaIon   101.798   134.785   0.168   0.755   0.46   0.147   6.808  

Health   -­‐191.503   180.462   -­‐0.27   -­‐1.061   0.303   0.113   8.889  

ResidenIal   -­‐815.288   752.822   -­‐0.237   -­‐1.083   0.293   0.153   6.527  

Retail  Outlets   249.651   199.107   0.137   1.254   0.226   0.611   1.637  

Natural  Elements   -­‐401.162   642.52   -­‐0.082   -­‐0.624   0.54   0.426   2.347  

Walking  score   -­‐201.641   611.69   -­‐0.051   -­‐0.33   0.745   0.302   3.315  

Bike  route  (km)   34.763   41.25   0.1   0.843   0.41   0.515   1.943  Pop.  Density  (‘000  people/km2)  

37.705   149.225   0.048   0.253   0.803   0.202   4.955  

MulIcollinearity  Effects  

Unstandardised  Coefficients  

Standardised  Coefficients  

Sig.  

Unstandardised  Coefficients  

Standardised  Coefficients   Sig.  

B   Beta   B   Beta  (Constant)   -­‐207.541       0.404   -­‐468.14       0.113  IRSAD  percenIle   -­‐0.103   -­‐0.004   0.963   5.513   0.239   0.019  Bus  Services   2.197   0.194   0.027   3.427   0.302   0.004  PnR   0.809   0.535   0   0.78   0.516   0  Walk  score   -­‐128.233   -­‐0.039   0.676   -­‐412.614   -­‐0.125   0.258  Bikeroute  (km)   -­‐0.686   -­‐0.002   0.974   15.389   0.055   0.543  Entropy  Place   -­‐245.463   -­‐0.064   0.445   251.686   0.066   0.491  PopulaIon  1.6km  radius   -­‐0.005   -­‐0.035   0.711  

 0.034  

 0.224  

 0.018  

%  Access  to  Jobs  45min  by  PT   34.721   0.469   0  R2-­‐adj=0.781   R2-­‐adj=0.679  

CauIon  when  fiCng  models!  

5/11/17  

Another  Example  Output  Regression  

R2-­‐adj  =0.690    DV = AM boardings Attractor stations

    b   Std.  Error   Beta   t   Sig.   Tolerance   VIF  

(Constant)   -­‐682.473   1167.756       -­‐0.584   0.59          IRSAD  percenIle   -­‐27.509   17.606   -­‐0.364   -­‐1.562   0.193   0.44   2.273  

Bus  Services   4.542   1.669   0.881   2.722   0.053   0.227   4.399  

PnR  Supply   -­‐6.711   2.691   -­‐0.789   -­‐2.493   0.067   0.238   4.205  

ResidenIal   2,250.827   2,384.005   0.242   0.944   0.399   0.362   2.762  

Retail  Outlets   654.687   289.032   0.698   2.265   0.086   0.251   3.988  Walk  score   3,322.618   2,622.356   0.426   1.267   0.274   0.211   4.743  Bikeroute  (km)   -­‐118.606   190.605   -­‐0.188   -­‐0.622   0.567   0.262   3.819  Popula2on  1.6km  radius   0.097   0.269   0.122   0.362   0.736   0.211   4.734  Employment  1.6km  radius   -­‐0.122   0.039   -­‐0.997   -­‐3.139   0.035   0.236   4.237  

Simple  Excel  Calculator  

StaUon  Name      

Corridor      

Daily  patronage  weekdays  

Daily  boardings  

AM  boardings  

Current  populaUon  1.6km  radius  

Current  total  workers  

JTW_train  Gen.  

(assumed  AM)    

JtoW  by  train  (%)    

JTW  train/AM  

boardings    

AddiUonal  boardings  required  

Assumed  mode  share  by  train  for  new  workers  

AddiUonal  people  required  

 Armadale   Armadale    1,825     917   322.43    10,583      2,991.2     225.1   7.5%   0.70   83   7.5%    1,103    Beckenham   Armadale    1,010     566   251.23    7,461      1,528.2     279.0   18.3%   1.11   434   18.3%    2,377    Carlisle   Armadale    733     432   146.3    10,339      1,983.0     299.1   15.1%   2.04   568   15.1%    3,766    Challis   Armadale    344     193   71.28    5,505      1,643.5     154.4   9.4%   2.17   807   9.4%    8,591    CoEesloe   Fremantle    1,096     592   163.74    4,413      1,729.5     160.7   9.3%   0.98   408   9.3%    4,390    Daglish   Fremantle    944     536   162.29    10,477      5,308.9     244.6   4.6%   1.51   464   4.6%    10,070    East  Guildford   Midland    530     296   108.38    2,491      741.1     79.8   10.8%   0.74   704   10.8%    6,535    East  Perth   Midland    994     660   80.69    5,725      3,636.1     141.6   3.9%   1.75   340   3.9%    8,731    Gosnells   Armadale    1,760     995   359.61    10,584      1,296.6     252.4   19.5%   0.70   5   19.5%    26    Grant  Street   Fremantle    426     232   113.76    5,682      1,532.1     193.1   12.6%   1.70   768   12.6%    6,093    Guildford   Midland    835     484   187.36    1,221      1,342.7     44.6   3.3%   0.24   516   3.3%    15,528    KarrakaEa   Fremantle    428     247   74.29    2,672      2,113.0     41.8   2.0%   0.56   753   2.0%    38,104    Kenwick   Armadale    806     459   213.69    4,991      1,083.7     153.5   14.2%   0.72   541   14.2%    3,820    Loch  Street   Fremantle    417     217   79.06    4,357      1,465.6     112.3   7.7%   1.42   783   7.7%    10,215    Maddington   Armadale    1,486     841   323.25    7,569      4,426.5     159.3   3.6%   0.49   359   3.6%    9,974    Meltham   Midland    799     498   204.02    9,870      1,378.5     402.1   29.2%   1.97   502   29.2%    1,721    Mosman  Park   Fremantle    780     427   156.06    5,259      1,202.3     275.1   22.9%   1.76   573   22.9%    2,504    Mt  Lawley   Midland    494     307   91.5    9,937      3,574.2     260.0   7.3%   2.84   693   7.3%    9,525    North  Fremantle   Fremantle    828     457   212.25    4,897      2,414.9     173.8   7.2%   0.82   543   7.2%    7,547    Queens  Park   Armadale    1,146     647   196.19    8,608      4,511.4     333.1   7.4%   1.70   353   7.4%    4,781    Seaforth   Armadale    202     115   40.78    4,535      1,167.4     111.5   9.6%   2.74   885   9.6%    9,263    Sherwood   Armadale    554     319   117.37    9,606      1,352.2     216.5   16.0%   1.84   681   16.0%    4,254    Success  Hill   Midland    221     132   55.27    4,152      805.4     131.1   16.3%   2.37   868   16.3%    5,331    Victoria  Park   Armadale    1,256     709   207.69    14,385      5,177.3     312.0   6.0%   1.50   291   6.0%    4,829    Victoria  Street   Fremantle    713     398   181.7    4,360      592.4     219.5   37.0%   1.21   602   37.0%    1,625    

12%  (av.)  of  the  labour  force  living  within  the  0.8  km  staIon  buffers  travel  to  work  by  train!  

Conclusions  -­‐  Regression  •  Good  Place  (7D,  including  ì  entropy)  enhance  amenity  &  

liveability,  enable  creaIon/development  of  good  acIvity  centres  (thus  good  job  accessibility  for  the  local  residents),  this  may  not  lead  to  substanIal  increase  in  transit  ridership!    

•  DisIncIon  between  primary  and  secondary  benefits  of  balanced  Place-­‐Node  precincts/TODs  is  necessary  

•  Node  dominant  -­‐  may  benefit  more  from  densificaIon,  if  aligned  with  greater  city-­‐wide  access  to  employment    –  High  %  train/workers  è propensity  to  commute  by  PT  

•  Place  dominant  -­‐  in  general  low  patronage  and  low  raIos  of  train  use  for  commuIng,  as  well  as  of  train  ridership  relaIve  to  jobs  

•  Mixed  (not  necessarily  Balanced?)  Place  and  Node  –  would  benefit  from  increased  densiIes  (jobs)  combined  with  PnR  access    

Conclusions  –  Cont’d  

•  Measures  targeIng  populaIon  and  economic  growth  vs  increasing  accessibility  (city-­‐wide  and  local)  

•  Analysis  should  be  at  the  corridor/city  level,  rather  than  staIon  precinct  –  Accessibility  is  key,  especially  PT  more  aEracIve  than  car  

•  Places  with  good  PT  access  could  deliver  maximum  value  for  patronage  

•  Same  story  for  the  good  TODs  selected  as  cases  –  Yeerongpilly  (1,500/day),    Wolli  Creek  (5,600)  –  Chatswood  (43,000/day)  the  highest,  followed  by  Footscray  (14,800/day)    

PotenIal  LU&T  SoluIons  

Cluster  1  ’Low  Access,  Node,  PnR,  SE  of  the  city’  •  Bus  services  (+),  Offices  (-­‐)  è BnR  or  Flexible  access  

soluIons            DensificaIon  

•  Cluster  2  ‘Best  Place  &  Node,  close  to  CBD’  

•  Workers  access  in  45  min  by  PT  (+),  Bike  route  (-­‐)  

è Do  nothing            Add  jobs  if  populaIon  ì  

Cluster  3  ’Interchange,  Feeder  buses,  Generators’  •  PnR  (+),  Bus  services  (+),  

Workers  access  in  45  min  by  PT  

è PnR  (transiIon)  and  separate  traffic  

è BnR  

•  Cluster  4  ‘Moderate  performance  all  func2ons’  

•  Workers  access  in  45  min  by  PT  (+),  Retail  (+),  Bus  services  (+)  

è  Mix  of  soluIons  –  BnR/Flexible  access  soluIons  è DensificaIon  outside  the  800m  buffer?  

PotenIal  Next  Steps  –  Stage  2  •  Compare  findings  with  project  4!    

–  Also,  larger  buffers  are  expected  to  have  disInct  entropy,  catchment,  etc.    

•  Corridor  analysis  would  be  useful,  in  addiIon  to  individual  case  studies  

•  Suggested  cases  –  TBD  (outliers  corridors,  new  staIons)    ExisIng  -­‐  Byford,  Challis,  Carlisle,  Cannington,  Rockingham,  Meltham,  Mt  Lawley,  Seaforth,  Shenton  Park,  SIrling,  Thornlie,  Warnbro,  Wellard,  Welshpool  new  -­‐  Aubin  Grove,  Yanchep/Alkimos,  Ellenbrook,  Karnup,  Morley,  Nicholson  Rd    

•  Use  models  for  predicIon  2051  (Place  indicators  not  available)  

•  SIll  missing  the  capacity  to  assess  performance  

Thank  you!    

Q  &  A  

Mix  of  Types  by  Corridor  

0  1  2  3  4  

0   5   10   15   20   25   30   35  

Cluster  

d  from  CBD  

Joondalup  

0  1  2  3  4  

0   10   20   30   40   50   60   70   80  

Cluster   d  from  CBD  

Mandurah  

Joondalup  

Kwinana  and  Wellard  

Rockingham   Warnbro  

Mix  of  Types  by  Corridor  

0  1  2  3  4  

0   5   10   15   20   25   30   35  

Cluster   d  from  CBD  

Armadale  Burswood  

0  1  2  3  4  

0   2   4   6   8   10   12   14   16   18   20  

Cluster  

d  from  CBD  

Fremantle  

0  1  2  3  4  

0   2   4   6   8   10   12   14   16   18  

Cluster   d  from  CBD  

Midland  

Welshpool  

Thornlie  

KarrakaEa   Mosman  Park  and  Victoria  St.    

Bassendean  

5/11/17  

Cluster Analysis (CA) •  Cluster analysis groups/classifies objects/variables on the

basis of the similarity of the characteristics they possess •  Technique used to classify entities into relatively

homogeneous groups •  Interdependence technique - numerical taxonomy using a

hierarchical clustering, followed by refined k-means clustering

•  Various types of data (binary to continuous) •  Euclidean distance as dissimilarity •  Various agglomeration approaches, algorithms, but

repeated at various spatial resolutions to check for consistency

Entropy •  Entropy measures the spread in the distribution of LU and hence,

mixed LU has higher entropy, whereas very homogeneous (one use only patches) approach 0 in entropy

•  Shannon’s entropy (SE)

•  Where p(j) is the proportion of use j in the mix at a specific location

!" = − ! ! !"#$(!)!

!

28  

Super-Clusters Variable 3 - Background

Variable 1 - Place

Variable 2 - Node

0

Station X (X1, X2, X3)

Station Y (Y1, Y2, Y3)

Euclidean d between items X and Y

Y2

X2

Y3

Y1

X3

X1