5 ms·

An aquantance of mine wrote a peice of software that looks at satelite pictures of Prague at night and tells the city which street lights are out, so they don't

An aquantance of mine wrote a peice of software that looks at satelite pictures of Prague at night and tells the city which street lights are out, so they don't have to use a "regular replacement schedual" but only replace the bulbs that need it. Could the same thing not be done using red-light cameras and stop lights?

mattygee 10y ago
Hey, that's a good idea - exception catching for traffic lights
germanier 10y ago
Replacing a traffic light after it already failed is too late.
- londons_explore 10y ago
  Not in London. Nearly every day I see a new failed one. There is usually redundancy tho
  - lsaferite 10y ago
    That seems like the right approach. The traffic light has a backup bulb and alerts when one burns out. I mean, cars can tell when a bulb is out, right? Why shouldn't traffic lights.
    - Klathmon 10y ago
      Actually it's even simpler than that, just have a minimum of 2 lights per direction of travel. So if one fails, a bulb burns out, one gets destroyed in a storm, or anything else happens, there will be at least one full backup.
      - lsaferite 10y ago
        Having automated outage reporting still makes sense though. It keeps you from having a guy that's paid to check every light in the city manually. And since your lights are on a city-wide network (they are, right?), then adding outage reporting should be a feasible thing.
Spooky23 10y ago
It might or might not. This is a costing exercise where you need the understand the objective and cost factors. If you have a strong requirement to avoid failure, proactive replacement let's you optimize the unit cost of replacement because you can fully utilize the crews and have no wastage. If you dispatch reactively, you either have people sitting around waiting for light bulbs to burn out or lightbulb replacement pulls crews off of other work, impacting schedules.
- thaumasiotes 10y ago
  ...or you could hire contractors when you need to do bulb replacement? Then you have full utilization while you're replacing bulbs, no one is idling on your payroll while you're not, and no one gets pulled from their existing work to do replacement.
  - jdmichal 10y ago
    ... and the external contractor who you're pushing all that risk on will charge you appropriately.
    - thaumasiotes 10y ago
      Sure, but they only receive the full amount of the risk you lose if you are their only client. They're still cheaper than either solution I responded to.
  - Spooky23 10y ago
    That's an option too, but is more complex and less deterministic. The key assumption is you want low/no downtime. The problem you have is negotiating SLA penalties are difficult to get right for organizations like cities and getting good metrics are tough. The other thing, especially with government contracting where RFPs are extremely difficult to change after the fact without getting contracts disputed by the losers is that the contractor will add risk premiums for penalties.
tialaramex 10y ago
In the UK the device managing the junction can already detect most failed lamps, it's common in safety applications to have paired filaments, so when one fails the other still works but you can detect the problem. People selling a city devices to manage their traffic offer the feature of the city being informed about failed lamps. But, that costs money and saving money is important in most places, so they'd rather wait for motorists to report a failed junction and repair it after the fact versus paying extra to solve it first. As things move to LEDs it hardly matters anyway, a correctly engineered LED lamp doesn't suddenly fail like filament lamps, it just gets sort of dim gradually, so a regular schedule (e.g. replace all the LED lamps every 5 years) is the correct way to deal with it. Cheap consumer-grade LED modules sometimes fail suddenly due to over-heating or badly made electricals but the road maintenance people probably aren't buying unlabelled stuff off eBay so they're unlikely to have that problem.

mattygee 10y ago

Hey, that's a good idea - exception catching for traffic lights

germanier 10y ago

Replacing a traffic light after it already failed is too late.

londons_explore 10y ago
Not in London. Nearly every day I see a new failed one. There is usually redundancy tho
- lsaferite 10y ago
  That seems like the right approach. The traffic light has a backup bulb and alerts when one burns out. I mean, cars can tell when a bulb is out, right? Why shouldn't traffic lights.
  - Klathmon 10y ago
    Actually it's even simpler than that, just have a minimum of 2 lights per direction of travel. So if one fails, a bulb burns out, one gets destroyed in a storm, or anything else happens, there will be at least one full backup.
    - lsaferite 10y ago
      Having automated outage reporting still makes sense though. It keeps you from having a guy that's paid to check every light in the city manually. And since your lights are on a city-wide network (they are, right?), then adding outage reporting should be a feasible thing.

Spooky23 10y ago

It might or might not. This is a costing exercise where you need the understand the objective and cost factors. If you have a strong requirement to avoid failure, proactive replacement let's you optimize the unit cost of replacement because you can fully utilize the crews and have no wastage. If you dispatch reactively, you either have people sitting around waiting for light bulbs to burn out or lightbulb replacement pulls crews off of other work, impacting schedules.

thaumasiotes 10y ago
...or you could hire contractors when you need to do bulb replacement? Then you have full utilization while you're replacing bulbs, no one is idling on your payroll while you're not, and no one gets pulled from their existing work to do replacement.
- jdmichal 10y ago
  ... and the external contractor who you're pushing all that risk on will charge you appropriately.
  - thaumasiotes 10y ago
    Sure, but they only receive the full amount of the risk you lose if you are their only client. They're still cheaper than either solution I responded to.
- Spooky23 10y ago
  That's an option too, but is more complex and less deterministic. The key assumption is you want low/no downtime. The problem you have is negotiating SLA penalties are difficult to get right for organizations like cities and getting good metrics are tough. The other thing, especially with government contracting where RFPs are extremely difficult to change after the fact without getting contracts disputed by the losers is that the contractor will add risk premiums for penalties.

tialaramex 10y ago

In the UK the device managing the junction can already detect most failed lamps, it's common in safety applications to have paired filaments, so when one fails the other still works but you can detect the problem. People selling a city devices to manage their traffic offer the feature of the city being informed about failed lamps. But, that costs money and saving money is important in most places, so they'd rather wait for motorists to report a failed junction and repair it after the fact versus paying extra to solve it first. As things move to LEDs it hardly matters anyway, a correctly engineered LED lamp doesn't suddenly fail like filament lamps, it just gets sort of dim gradually, so a regular schedule (e.g. replace all the LED lamps every 5 years) is the correct way to deal with it. Cheap consumer-grade LED modules sometimes fail suddenly due to over-heating or badly made electricals but the road maintenance people probably aren't buying unlabelled stuff off eBay so they're unlikely to have that problem.