Optimising bus networks is like a game of pick up sticks. Trying to shorten, extend or divert one route has implications for many more, especially if your budget is limited. The number of interdependencies can change what was a minor network change involving two or three routes into something that affects 10 to 20 across a large region.
Greater Dandenong hasn't had an implemented full scale bus network review for decades. Though there was a large mostly unimplemented review done about 15 years ago. And there were more recent smaller reforms to some Dandenong routes. Notably (i) routes from the Cranbourne area in 2016, (ii) Endeavour Hills a few years later and (iii) some Keysborough changes involving deleting the old 815 to form the new 816 and boost services on 813 a bit later again.
Mostly though it's been in the too hard basket. The tendency instead has been to layer any new routes over an existing unchanged network, such as happened when Route 709 was extended from Waterways to Noble Park via Keysborough. And the area's last bus change, the major upgrade to Route 800 along Princes Hwy, was a timetable initiative only.
Nearly a year ago I discussed 7 day upgrades for the next lot of Dandenong area buses. Largely based on patronage productivity the 802, 804 and 815 were the three most deserving, edging out routes 844, 857 and 885. Upgrades are possibly just by finding $1-2m per year to run the existing routes 7 days. Alternatively if you want higher frequencies for a given amount of money you could merge some routes. I discussed merging routes 802, 804 and 862 (5, 6 and 7 day per week respectively) into two 7 day routes back in 2020 here. For this reason I won't cover them here.
As common as missing Sunday service is low frequency. Dandenong has a lot of routes operating at every 60 minutes or worse, even during peak times on weekdays. Examples include routes 811, 812, 814, 848 and 857. More make the list when weekends are considered including 813, 824, 828, 844, 850, 862, 885, 890 and 892.
Just like with the 802/804 there's a choice between just throwing more service kilometres on existing routes or thinking about how the network could be simplified to get better value from what's spent. Doing both would be even better with wider benefits. Especially in a catchment like Greater Dandenong where you can put on a bus almost anywhere and expect above average ridership.
Mostly though it's been in the too hard basket. The tendency instead has been to layer any new routes over an existing unchanged network, such as happened when Route 709 was extended from Waterways to Noble Park via Keysborough. And the area's last bus change, the major upgrade to Route 800 along Princes Hwy, was a timetable initiative only.
Nearly a year ago I discussed 7 day upgrades for the next lot of Dandenong area buses. Largely based on patronage productivity the 802, 804 and 815 were the three most deserving, edging out routes 844, 857 and 885. Upgrades are possibly just by finding $1-2m per year to run the existing routes 7 days. Alternatively if you want higher frequencies for a given amount of money you could merge some routes. I discussed merging routes 802, 804 and 862 (5, 6 and 7 day per week respectively) into two 7 day routes back in 2020 here. For this reason I won't cover them here.
As common as missing Sunday service is low frequency. Dandenong has a lot of routes operating at every 60 minutes or worse, even during peak times on weekdays. Examples include routes 811, 812, 814, 848 and 857. More make the list when weekends are considered including 813, 824, 828, 844, 850, 862, 885, 890 and 892.
Approaches to network simplification
The Greater Dandenong bus network map shows several areas where routes are closely spaced or (even better for this exercise) overlapping. These may indicate opportunities to adjust routes so that you have fewer but they run more frequently. In Dandenong's case I wanted to test whether it was possible for some of these to go from every 40-60 to every 30 minutes, at least on weekdays.
The first exercise was drawing all routes that went through Dandenong (and some adjacent stations) to get a network route kilometre number. I then drew a more simplified network. That reduced the route network by about 70 kilometres without significant loss of bus stops. That more economic network would likely buy more frequency and/or operating hours on one or two of the reformed routes. The main trade-off involved would be that for some trips you may need to change buses.
A more aggressive approach would have reduced network kilometres further. That would permit even higher frequencies. But then you're increasing walking distances to (say) 600 or 800 metres as opposed to 400 metres which has been a historical design aim. This has a higher justification in areas where buses are poorly used as patronage versus coverage trade-offs are more acute.
But it is perhaps harder to argue in places where even indirect routes enjoy well above average usage and a strong case exists for more network service kilometres overall (ie not robbing Peter to pay Paul). Most of Greater Dandenong falls in that category. Still, if you're in a time that balks at spending $5m on a local bus network upgrade but routinely accepts $5b cost blow-outs on major projects then you may have to consider trade-offs to get anything done with buses.
It's also useful to map key network issues before reaching for the felt pens. For example noting corridors that need more service and others that may have too many infrequent routes overlapping. A non-exhaustive map is below (click for a better view).
The first exercise was drawing all routes that went through Dandenong (and some adjacent stations) to get a network route kilometre number. I then drew a more simplified network. That reduced the route network by about 70 kilometres without significant loss of bus stops. That more economic network would likely buy more frequency and/or operating hours on one or two of the reformed routes. The main trade-off involved would be that for some trips you may need to change buses.
A more aggressive approach would have reduced network kilometres further. That would permit even higher frequencies. But then you're increasing walking distances to (say) 600 or 800 metres as opposed to 400 metres which has been a historical design aim. This has a higher justification in areas where buses are poorly used as patronage versus coverage trade-offs are more acute.
But it is perhaps harder to argue in places where even indirect routes enjoy well above average usage and a strong case exists for more network service kilometres overall (ie not robbing Peter to pay Paul). Most of Greater Dandenong falls in that category. Still, if you're in a time that balks at spending $5m on a local bus network upgrade but routinely accepts $5b cost blow-outs on major projects then you may have to consider trade-offs to get anything done with buses.
It's also useful to map key network issues before reaching for the felt pens. For example noting corridors that need more service and others that may have too many infrequent routes overlapping. A non-exhaustive map is below (click for a better view).
The reason why I say it's non-exhaustive is that it does not include weak termini (these exist on routes 814, 816, 844 and 848) or low weekend frequencies (on nearly every route). It is however still useful to identify opportunities for improvement and how you might resource them.
Limited service points 1 and 7 as well as the significant overlap at 9 are addressed in my look at 802, 804 and 862. Thus they do not need to be covered here.
Studying the remainder can help with some tentative ideas, especially if there are synergies between the green and red circles. Two really jump out:
Limited service points 1 and 7 as well as the significant overlap at 9 are addressed in my look at 802, 804 and 862. Thus they do not need to be covered here.
Studying the remainder can help with some tentative ideas, especially if there are synergies between the green and red circles. Two really jump out:
* 811/Heatherton Rd
This is only every 60 min, even in peak times. The portion between Springvale and Dandenong gets high usage, justifying at least a 30 minute service. See 3 on map.
Route 811 is paired with Route 812. From Brighton they head east together providing that 30 minute service as far east as Dingley. Then they split off with 812 going to Dandenong via Keysborough. Route 812 contributes to significant overlap between Dandenong, Keysborough and Dingley at 11, 12 and 13 on the map above.
The top map below gives a rough schematic of the existing 811/812. I used this to illustrate an east-west service divide. That is Brighton got double the frequency of Heatherton Rd despite the latter having higher social need, greater bus patronage propensity and strong actual usage.
This is only every 60 min, even in peak times. The portion between Springvale and Dandenong gets high usage, justifying at least a 30 minute service. See 3 on map.
Route 811 is paired with Route 812. From Brighton they head east together providing that 30 minute service as far east as Dingley. Then they split off with 812 going to Dandenong via Keysborough. Route 812 contributes to significant overlap between Dandenong, Keysborough and Dingley at 11, 12 and 13 on the map above.
The top map below gives a rough schematic of the existing 811/812. I used this to illustrate an east-west service divide. That is Brighton got double the frequency of Heatherton Rd despite the latter having higher social need, greater bus patronage propensity and strong actual usage.
How to fix this? Most obvious is to make 812 like 811 so it runs along Heatherton Rd to give it double the buses (on weekdays) between Dandenong and Springvale (closing the inequality with the Brighton end). The result is that the routes become identical except in Dingley where they diverge to retain coverage.
You'd want the routes to be of similar length to keep gaps between trips at an even 30 minutes at both ends. Springvale, Dandenong Markets, Dandenong Hospital, Chisholm TAFE, Oasis Pool and Heatherton Rd itself are just a few of the destinations that would benefit from this corridor getting a frequency upgrade from every 60 to every 30 minutes.
Rerouting the hourly 812 from Keysborough removes a bus from a part of Corrigan Rd that needs it. That can be addressed by a minor rerouting of 824 (which at every 40 min is more frequent). Most other parts of Keysborough are near Route 813 which would also gain frequency as discussed next. There are different ways in which the sections around Springvale South/Dingley can be revised with this important to permit Route 814 upgrades also covered next.
The above 811/812 change is like a Stage One. Legibility could be increased further if the 811/812 was consolidated into a single route with some other solution found to maintain coverage in Dingley Village but that's a lower priority than what's discussed next.
You'd want the routes to be of similar length to keep gaps between trips at an even 30 minutes at both ends. Springvale, Dandenong Markets, Dandenong Hospital, Chisholm TAFE, Oasis Pool and Heatherton Rd itself are just a few of the destinations that would benefit from this corridor getting a frequency upgrade from every 60 to every 30 minutes.
Rerouting the hourly 812 from Keysborough removes a bus from a part of Corrigan Rd that needs it. That can be addressed by a minor rerouting of 824 (which at every 40 min is more frequent). Most other parts of Keysborough are near Route 813 which would also gain frequency as discussed next. There are different ways in which the sections around Springvale South/Dingley can be revised with this important to permit Route 814 upgrades also covered next.
The above 811/812 change is like a Stage One. Legibility could be increased further if the 811/812 was consolidated into a single route with some other solution found to maintain coverage in Dingley Village but that's a lower priority than what's discussed next.
* 813/814 Keysborough, Springvale and Mulgrave areas
Both are indirect but popular routes serving catchments well away from trains. 813 is currently every 40 minutes on weekdays and 60 minutes on weekends. 814 is hourly on its limited hours Monday to Saturday morning timetable. Both overlap and treat a portion of Springvale Rd as a through stop. That's important as it relieves pressure on the often overcrowded Route 902 for short trips.
It would benefit the 902 (eg faster loading) if 813 and 814 run at a 7 day frequency attractive enough for many of the large numbers of passengers who alight at Springvale to prefer those over the 902. 20 minutes 7 days per week would be the ideal but even a 30 minute service would be preferable to current irregular frequencies. The rest of this item will assume 30 minute frequency on weekdays.
As you can see under 'existing' below, 813 and 814's indirectness means that no one would ride either route end to end. Also 814's Springvale South terminus is weak. Both characteristics make it easier to simplify the route as may be needed to increase frequencies on key portions.
It would benefit the 902 (eg faster loading) if 813 and 814 run at a 7 day frequency attractive enough for many of the large numbers of passengers who alight at Springvale to prefer those over the 902. 20 minutes 7 days per week would be the ideal but even a 30 minute service would be preferable to current irregular frequencies. The rest of this item will assume 30 minute frequency on weekdays.
As you can see under 'existing' below, 813 and 814's indirectness means that no one would ride either route end to end. Also 814's Springvale South terminus is weak. Both characteristics make it easier to simplify the route as may be needed to increase frequencies on key portions.
Where can routes be shortened or merged to allow the higher frequencies needed? There are several candidates, including:
(i) Spring Rd, Springvale South. Currently a weak terminus for the 814, this could be served by rerouting 811 or 812. This delivers 7 day service to the area and frees up the Springvale end of the 814. This is numbered 6 on the Dandenong bus issues map.
(ii) Police Rd, Mulgrave. This is a busy road that with few pedestrian crossings is hard to cross. Four routes operate along sections of it including 631, 813, 814 and 848. Arguably that's excessive. Scope exists for the 813 and 814 to be merged to create a more frequent 7 day route. See 10 on the issues map.
(iii) Noble Park North to Dandenong. Route 814 follows this path from Waverley Gardens via Jacksons Rd. It overlaps the 848 closer to Dandenong. Apart from the 814 not running evenings, Saturday afternoons or Sundays a major issue is that it does not connect Noble Park North to its nearest station at Noble Park. The trade-off here is that weekday and Saturday morning frequencies at some stops closer to Dandenong are reduced unless Route 848's frequency is increased.
(iv) Route 885 between Springvale and Glen Waverley is another part-time route. It gets good usage but adds little unique coverage. There may be scope for most of its stops to be served by an upgraded and rerouted Route 814. This would preserve the Springvale connection and deliver new 7 day service to some stops. This is an example of a trade-off common with bus network reform when aiming for higher frequencies.
(i) Spring Rd, Springvale South. Currently a weak terminus for the 814, this could be served by rerouting 811 or 812. This delivers 7 day service to the area and frees up the Springvale end of the 814. This is numbered 6 on the Dandenong bus issues map.
(ii) Police Rd, Mulgrave. This is a busy road that with few pedestrian crossings is hard to cross. Four routes operate along sections of it including 631, 813, 814 and 848. Arguably that's excessive. Scope exists for the 813 and 814 to be merged to create a more frequent 7 day route. See 10 on the issues map.
(iii) Noble Park North to Dandenong. Route 814 follows this path from Waverley Gardens via Jacksons Rd. It overlaps the 848 closer to Dandenong. Apart from the 814 not running evenings, Saturday afternoons or Sundays a major issue is that it does not connect Noble Park North to its nearest station at Noble Park. The trade-off here is that weekday and Saturday morning frequencies at some stops closer to Dandenong are reduced unless Route 848's frequency is increased.
(iv) Route 885 between Springvale and Glen Waverley is another part-time route. It gets good usage but adds little unique coverage. There may be scope for most of its stops to be served by an upgraded and rerouted Route 814. This would preserve the Springvale connection and deliver new 7 day service to some stops. This is an example of a trade-off common with bus network reform when aiming for higher frequencies.
A concept network
There are many ways to start to untangle what is a complex and infrequent network. Below is one concept. As with any idea it has pros and cons. But this needs to be weighed against the substantial issues of the current infrequent and often part-time network.
Alternatives to the 816 northern extension for Noble Park North could involve retaining the 814 in the area but terminating it at Noble Park station. Or extending the 631 south to Noble Park station, providing a new Monash University connection. Both have merit but involve more service kilometres than extending the 816 which, like the 814 it would replace on Jacksons Rd, is only hourly (though it has better peak service and runs 7 days).
The above concept has two points where road works would be needed to permit through movement of buses via a shorter route. This adds implementation complexity. This may be avoidable if less direct routes were taken. However following such a 'path of least resistance' also denies the high returns possible from making an initial capital investment now to save future service kilometres and cut trip times.
Network frequency changes
Shown below are the frequency changes resulting from the above described network. It assumes that 811/812 retain their current 30 minute combined weekday frequency (but shifted to Heatherton Rd) and the modified 813 and 814 get upgraded to every 30 minutes.
The biggest gain is along Heatherton Rd between Springvale and Dandenong. Not only is this a currently very underserved bus corridor (only hourly, even in peaks) but route density is very low, partly due to difficult road geometries. The revised network halves gaps between buses to 30 minutes with improved connections to major destinations at Dandenong and Springvale.
Also gaining frequency are neighbourhoods in Springvale, Keysborough and Noble Park with the 813 upgrade which cements its status as a key route in the area. Parts of Mulgrave north of Police Rd also have improvements from the 814 rerouting, though there is a trade-off involving the 885.
Currently both 813 and 814 overlap on a section of Springvale Rd, going from at least the shops to the Princes Hwy junction. The 40 and 60 minute frequency results in 2.5 buses per hour with uneven intervals. Through-routing with Route 813 trips forming the 814 (and vice versa) changes this to 2 buses per hour but with an even spacing. This 30 min clockface timetable should be more appealing.
As with any economical network there are trade offs. The main one is between Dandenong and Eastlink. There the 814/848 corridor is reduced to just the hourly 848. Much of this area has alternative 30 (and even 20 minute) routes including 800, 811/2, 813 and 850. However if this is still an issue scope exists for the 848 to have trips added.
Also gaining frequency are neighbourhoods in Springvale, Keysborough and Noble Park with the 813 upgrade which cements its status as a key route in the area. Parts of Mulgrave north of Police Rd also have improvements from the 814 rerouting, though there is a trade-off involving the 885.
Currently both 813 and 814 overlap on a section of Springvale Rd, going from at least the shops to the Princes Hwy junction. The 40 and 60 minute frequency results in 2.5 buses per hour with uneven intervals. Through-routing with Route 813 trips forming the 814 (and vice versa) changes this to 2 buses per hour but with an even spacing. This 30 min clockface timetable should be more appealing.
As with any economical network there are trade offs. The main one is between Dandenong and Eastlink. There the 814/848 corridor is reduced to just the hourly 848. Much of this area has alternative 30 (and even 20 minute) routes including 800, 811/2, 813 and 850. However if this is still an issue scope exists for the 848 to have trips added.
Changes in route lengths
Reducing network kilometres is key to frequency increases assuming no additional resourcing. On the network mapped above the following apply:
* 816 has a large increase due to its extension to Waverley Gardens (replacing part of 814)
* 811, 812 and 824 overall have a small increase due to alignment changes.
* 813 and 814 have large decreases as they are made shorter with overlaps removed
* 885 also has a large decrease as it is entirely replaced by a rerouted 814
Overall route kilometres across the seven routes drops from about 150 to 132 km, ie about 11%.
The 60 to 30 minute frequency boost on Heatherton Rd between Dandenong and Springvale is already accounted for as both routes would continue to run hourly with Heatherton Rd gaining the 812.
Hence the returns from that 11% can go almost entirely to boosting the 813 and 814's frequency.
Let's say the aim was a 30 minute frequency on weekdays, ie 2 buses per hour. Boosting the 813 from 1.5 to 2 buses per hour is a 33% increase while the 814 uplift from 1 to 2 buses per hour is 100%. Both are much bigger than our 11% so initial hopes for much are dim.
But recall that 813 and 814 were made shorter. By finishing at Springvale 813 loses about 6.5km while 814 is 10km shorter by no longer operating between Waverley Gardens and Dandenong and Springvale South. The pair is around 45% shorter, making it cheaper to upgrade than if this was attempted on the full length of both routes.
Cost and implementation
You could reverse engineer timetables to work out exact changes in annual service kilometres. It is most unlikely you'll get an answer that makes it free. But it is very likely that a network like this will give a result that is relatively cheap, especially relative to the amount of extra 30 minute frequency added.
Further enhancements that boost weekend service and operating hours would also be cost-effective in view of the more economical, fewer route kilometre network than exists currently.
On the con side, this represents, by current DTP standards, a large bus network reform. Other cities achieve bigger bus network reforms quicker but existing processes here might entail 3 or even 4 year time-lines (unlike say level crossing removals where we do them quicker than others). Thus governments wishing to reform buses have a trade off between value for money, implementation speed and work effort with two but not three being readily possible.
Conclusion
Concepts for more frequent buses in Springvale/Keysborough have been presented. They are 'oily rag' type upgrades that nevertheless involve significant frequency gains on routes that need them.
Comments are appreciated and can be left below.
* 816 has a large increase due to its extension to Waverley Gardens (replacing part of 814)
* 811, 812 and 824 overall have a small increase due to alignment changes.
* 813 and 814 have large decreases as they are made shorter with overlaps removed
* 885 also has a large decrease as it is entirely replaced by a rerouted 814
Overall route kilometres across the seven routes drops from about 150 to 132 km, ie about 11%.
The 60 to 30 minute frequency boost on Heatherton Rd between Dandenong and Springvale is already accounted for as both routes would continue to run hourly with Heatherton Rd gaining the 812.
Hence the returns from that 11% can go almost entirely to boosting the 813 and 814's frequency.
Let's say the aim was a 30 minute frequency on weekdays, ie 2 buses per hour. Boosting the 813 from 1.5 to 2 buses per hour is a 33% increase while the 814 uplift from 1 to 2 buses per hour is 100%. Both are much bigger than our 11% so initial hopes for much are dim.
But recall that 813 and 814 were made shorter. By finishing at Springvale 813 loses about 6.5km while 814 is 10km shorter by no longer operating between Waverley Gardens and Dandenong and Springvale South. The pair is around 45% shorter, making it cheaper to upgrade than if this was attempted on the full length of both routes.
Cost and implementation
You could reverse engineer timetables to work out exact changes in annual service kilometres. It is most unlikely you'll get an answer that makes it free. But it is very likely that a network like this will give a result that is relatively cheap, especially relative to the amount of extra 30 minute frequency added.
Further enhancements that boost weekend service and operating hours would also be cost-effective in view of the more economical, fewer route kilometre network than exists currently.
On the con side, this represents, by current DTP standards, a large bus network reform. Other cities achieve bigger bus network reforms quicker but existing processes here might entail 3 or even 4 year time-lines (unlike say level crossing removals where we do them quicker than others). Thus governments wishing to reform buses have a trade off between value for money, implementation speed and work effort with two but not three being readily possible.
Conclusion
Concepts for more frequent buses in Springvale/Keysborough have been presented. They are 'oily rag' type upgrades that nevertheless involve significant frequency gains on routes that need them.
Comments are appreciated and can be left below.
