Note: special thanks to Jarrett Walker for pushing for a better analysis and some quick feedback.
So the other day, I was talking with Jarrett Walker on Twitter about the Orange Line, which I’ve previously written about, saying that there’s no need to worry about converting to LRT. True to form, Jarrett said that LRT would offer savings on operations costs, and that to know if it makes sense, you’d have to figure out how long it would take to recoup the capital costs through savings on operations.
Real analysis based on numbers? Oh I’ll play your game, you rogue. Let’s crunch some rough numbers and see how long it would take to recoup the capital costs of converting the Orange Line to LRT through operations savings. (Note: the savings is basically labor for drivers; LRT has more specialized infrastructure, so maintenance should cost more.)
First, let’s set the ground rules for this analysis:
- Assume that during the peak periods, the Orange Line will run 4 minute headways with two buses operating in tandem; this is effectively a 2 minute headway.
- Off-peak, the Orange Line will run existing headways, but never less frequent than every 10 minutes. (The service today is much less frequent at night, but to level the playing field, let’s bump it up to 10 minutes.)
- The capacity of an Orange Line vehicle, per LACMTA policy, is 74 riders.
- An LRT option will run 3-car trains, with capacity of 399 riders, again per LACMTA policy.
- Therefore, to provide the same capacity, during peak periods the LRT will operate at 10 minute headways.
- After 7pm, the LRT will run 2-car trains. (This will tilt things in the LRT’s favor. LACMTA said they were planning to do this on the Expo Line when they went from 20-minute to 10-minute evening and late night headways, but I have yet to actually see any shorter trains.)
- Off-peak, the LRT will operate at the same headway as the existing Orange Line, since the controlling factor will be the desire to provide frequent service, not capacity.
- Analysis based on most recent NTD vehicle revenue mile operating costs. I multiplied the reported bus cost by 1.3 to account for the larger buses used on the Orange Line.
- Assume 290 weekday equivalents per year.
- Assume that it would cost $1.5b to convert the Orange Line to LRT.
- Assume a discount rate of 3%.
The operating periods, headways, number of trips, revenue vehicle-miles, and costs are summarized in the table below.
Alright, so we aren’t saving any operating costs with those assumptions! Note that we’re providing lots of capacity on the LRT even when we don’t need it. This is one of the strengths of bus – it’s much easier to pare back capacity without having the headway get too long.
Let’s assume ridership goes up and we have to double the amount of bus service we’re providing. Let’s also relax the off-peak headway on the LRT to 15 minutes so that we’re not throwing out so much unused capacity on that option.
Now, the LRT only costs half as much to run as the bus – operational savings of $50m a year, which is some serious money. However, at a capital cost of $1.5b and an interest rate of 3%, it will take over 75 years to pay off the capital costs through operations savings. That’s probably too long a time frame, because we’re going to have to make some major capital investments in the thing during that time – things like replacing signals and traction power substations, which agencies usually put under capital costs, not maintenance. So it’s still a no go.
Here’s another shot at bumping up the frequency of the bus. In this case, we’re basically providing the same capacity in both options – there’s no wasted capacity for the LRT.
Now we’re saving $88m a year, and we can pay off the capital costs in 24 years, which is definitely a winner. This last option assumes some considerable gains in ridership in both peak and off-peak periods – we’re providing more than five times the peak period capacity that we have today!
However, if things reach that point, we’re pushing the limits of a bus system. For the Orange Line, which intersects major arterials at grade, it’s probably a stretch to say that the bus platoon headways could really go below 2 minutes. LADOT isn’t going to want to make the light cycles that short. It’s also a stretch to say we could berth 80 buses per hour at a typical Orange Line station, and passenger circulation is going to be an issue. (Try to remember berthing, passenger circulation, and dwell times the next time you read Randal O’Toole telling you that a bus lane can run 1200 buses an hour.)
In other words, for the Orange Line, it looks like we’ll hit the physical constraints of the system at about the same time it makes sense to upgrade to LRT to save on operating costs. As I said in my original piece, I don’t think we’re at capacity on the Orange Line, so I think the conclusion of that piece holds up pretty well: for the time being, we should spend our capital dollars elsewhere.
Notes
- Before any LRT haters jump on this, let’s note that the analysis here is a peculiar case of replacing an existing BRT with LRT, so we have to earn back the entire capital cost of the LRT through operating savings. A much more common analysis would be comparing BRT and LRT at the outset, in which case you have to earn back the difference in capital costs. The Orange Line cost $377m for the initial piece and $215m for the extension, all in 2012 dollars, for a total of $589m, call it $600m. Say that the Orange Line from scratch would cost $1.6b and the difference is $1.0b. The LRT starts to make sense sooner in that case. (Side note: good lord, why did the Orange Line extension cost twice as much per mile as the initial piece?).
- The Orange Line is 18 miles long and has 16 stations. The Expo Line is 15 miles long and has 17 stations, and cost about $2.5b. It has occasionally been argued that the cost of upgrading the Orange Line would be less because of the previous investment, but I don’t see how this could be the case. The only thing that could possibly be reused is the at-grade traffic signals, but even those will require modifications. The existing Orange Line stations and guideway will have to be completely demolished and rebuilt. Upgrading the Orange Line to LRT would be cheaper than Expo because the ROW is intact, there is no street running, and the Expo Line has some big time grade separations.
- The combination of notes 1 and 2 means that if you’re choosing BRT over LRT, you’d better be pretty sure that you’re not going to hit the capacity of the BRT during the capital depreciation period, say 30 years. The tragedy of upgrading the Orange Line to LRT would not be the time needed to pay back capital costs through operations savings, but the wasted capital expenditure on BRT where we didn’t get our money’s worth. Of course, 30 years is a long time and predicting the future is hard!
- This analysis implies that the savings on operating costs could be used to pay down capital costs. Realistically, in most transit agencies the capital and operating sides are separate, so operating savings don’t accrue to the capital budget. That’s really just a matter of accounting though, and not an argument against making smart capital investments that yield long-term savings on operations. In fact, if the operating savings could pay back the capital costs on a reasonable time frame, the agency ought to be able to issue bonds for the capital costs, backed by the operating savings. I’m not aware of an agency that’s tried that though.
- Since I obviously did this analysis in Excel, you could easily modify it for a project that interests you. If you want the workbook, get in touch with me and I’ll send it to you.
Let's Go LA 
Good analysis, but there’s some elements missing from the conversation; Impacts of future corridors such as Measure R, timing and realizing the right-of-way is already paid for.
The impacts of the future Metro corridors, such as a Van Nuys Blvd – Sepulveda Pass Line and other Measure R corridors that are being built such as the Westside Subway and Expo Line and how they will impact the corridor’s ridership. Where this valley corridor will have forecasts to see even more usage out of it.
It’s smarter to be pro-active rather than re-active to save further capital costs to having to figure out as part of the capital cost of the upgrade, how many MORE buses you would need to operate to transport these large number of passengers while you’re upgrading the right-of-way that you’ve already paid for and can do what you wish compared to maybe now needed to find another right-of-way to build an upgraded corridor on. The same is true about the physical upgrades needed to the Orange Line for busway expansion to handle the 30 second headways and bus platoons needed for that kind of ridership density, this will need to be factored in to ANY upgrade that is needed.
One irony is by the time you need to upgrade the busway to light rail it would be ready to upgrade further to a heavy rail subway so the timing is critical because a $1.5B problem can easily escalate to a $6B problem because we wanted to squeeze a little more out of it.
Think of the issue of freeway capacity, How many cheap freeway capacity expansion projects can you think of that are being built in an already congested corridor?
This makes sense, if there weren’t dramatically diminishing returns for buses beyond about 4 minute headways. If you had a fully grade-separated line you may be able to get away with 120 second headways. Maybe. But with intersecting roadways, you’d need perfect signal preemption to keep any gaps from forming (beyond those that naturally occur from load imbalances). And with a bus requiring a signal to trip every 60 seconds, you’re playing with fire. If you can somehow keep buses moving through intersections on the same phase every 2 minutes, it’s workable. But inevitably, you’ll have a bus arrive at an intersection 15 seconds after the light changed, and it can’t change back because it has a pedestrian phase and requires 30 seconds to clear that phase. So that bus stops, and by the time it’s accelerated out of the light, it’s 30 seconds behind schedule. It then arrives at the next station to find 25% more passengers than normal, so it fills up past it’s capacity. This increase dwell times, so pretty quickly you have an overcrowded bus with the one behind it quickly gaining.
If humans were rational beings, some would wait for the second bus, but they’re not, so they’ll continue crowding the first. This bus loses time, and the buses become bunched, decreasing operating speed and overall capacity. This occurs on many frequent routes, even those with 6 or 8 minute headways. When you drop below 3 minutes, it’s inevitable.
This is why single-lane BRT systems generally don’t operate beyond 4 minute headways (or have capacities beyond about 2500 people per hour). A three-car light rail train can carry that many people operating once every 10 minutes and not run in to the same problem. When BRT is given passing lanes at stations, it can attain much higher throughput, but this requires 65 feet of right-of-way, something which is rarely found in long segments.
The Orange Line doesn’t have passing lanes at all the stations but it does have platforms long enough for two buses to berth at the same time, so you could run platoons. Of course you are hoping that the platoon manages to stay together because you don’t want them to get split up at a traffic signal, and this can be tricky if something causes one bus to have a longer dwell.
Because of the Valley’s phenomenal arterial grid and the Orange Line running in an old rail ROW, you really only have to preempt a signal every half mile. By looking further upstream, you might be able to predict when the bus will arrive at the intersection and improve the preempts. This would let you start the ped countdown sooner, and many of the crossings of the busway are very short. There’d certainly be some challenges at the places with very long ped crossings and where the busway crosses two streets at once diagonally. But I think it should be possible to improve on the 4 minute headway.
As some other blogger once said, organization before electronics before concrete 😉