# The Coming Bus Apocalypse

## Electric driverless taxis will compete with many bus routes by offering high convenience transport at a fraction of current costs.

Once electric driverless cars can provide transport as a service (think Uber with electric cars and no driver) the cost of personal transport will fall so low that most bus services will become non-viable. What follows is a detailed description of one bus service to illustrating this view. There are a lot of numbers and calculations in this description so I have included that detail at the end of the article if you want to delve down and test our numbers and assumptions.

As there is no rail or light rail access to Melbourne Airport in Australia, there is a **Sky Bus**. This service operates from Southern Cross Station in the CBD to the airport. A trip of approximately 22km. The cost for the service is $18 one way for an adult by themselves or with children (children travel free). The service provides a cost-effective alternative to a taxi which costs **$55-$65** including airport fees and tolls. It is particularly cost-effective for a single passenger. Especially one who wishes to connect to urban or regional rail services, which depart from Southern Cross Station. It is less cost-effective for two adult travelers who want to go somewhere else in the city or the Northern suburbs (closer to the airport). From 6am until midnight the service runs every 10 minutes. The rest of the time it runs every 15 minutes. Its price per km one way is 81 cents for every adult passenger. This is about 10x the price for a **Melbourne to Adelaide bus trip**. Of course, there is much more downtime per km travelled on the return route to the airport. Tolls and airport fees also need to be considered.

Looking closely at this service allows us to compare driverless bus costs with driverless car costs. I have created a table of these costs using the **Freight Metrics Bus Cost Calculator**. Some of the assumptions have been verified using the **Australian Transport Assessment and Planning Site**. Our estimates are just that. Reasonable cost assessments that are useful for comparing low cost alternatives. They are not intended to be accurate to the last cent per km.

The following table shows the cost comparison based on the detailed assumptions we have provided at the end of this post. We have compared a current diesel bus model with an imagined electric driverless bus:

*Note: the electric driverless bus travels less distance annually because downtime to charge batteries reduces available travel hours. This favours the diesel bus in the comparison as its fixed costs are spread over more kilometres*

The conclusions from this are:

1/ Electric driverless buses will replace standard bus services once the technology is viable. Even if some of my numbers turn out to be underestimates the differences are staggering. I believe that I have been too conservative, especially on depreciation.

2/ Electric driverless buses can compete with electric driverless cars in a transport as a service model in limited circumstances. If we look at the cost per passenger km for various passenger numbers we get the following:

If we go back to our previous costings on **driverless electric cars as a service** then we calculated a shared operating cost of 4.4 cents per kilometer per passenger for a car carrying four passengers and 17.6 cents for a single passenger. This ignores issues such as taxes, interest and profit margins (but so do our bus calculations).

For the journey to the airport of 22 km we get the following operating costs per passenger:

The bus will beat out most of the car combinations on a pure cost basis. The problem for buses is that cost is not the only consideration and becomes less of a consideration as the raw price points fall. If we assume that all the non-vehicle costs for both services are similar per passenger, then at a customer level then differences start to become irrelevant. The difference in price offered to the consumer in a perfect market is the difference in operating costs.

We have included a detailed analysis of our assumptions of non-vehicle costs in the appendix at the end of the post. The overall conclusion must be that these cost differences will make very little difference in the costs to the customer.

If the difference in cost you offer a customer for a trip as a bus provider is $3.61 less than what a car trip for a single passenger costs (*bus with 20 passengers versus customer as a single passenger in a car: Vehicle cost difference $2.89 + Extra fuel tax 39 cents = 2.42 cents per km for 22km versus 14.74 cents in total. GST of 10% on cost differences*), you are in big trouble as a business. To take up that offer the customer must find their way to the pick-up point. In most cases, this will incur extra travel costs. It will also cost them time and inconvenience.

If the car service averages 2 passengers per car the difference becomes only $1.19 (*bus with 20 passengers versus customer as passenger in a car with two passengers. Vehicle cost difference $0.96. Extra fuel tax 12 cents as described in the assumptions. GST of 10% on cost differences*), making it a no-brainer. The actual business models will be more complicated than this. Many more models and price points will emerge. However, this will also favour the car service. Once operating costs become so low, the percentage differences matter much less than the total dollar difference. Convenience and experience start to win over percentage discounts. A personal car that takes you to the airport sounds much more attractive than a bus.

Of course there will be some extra costs incurred for the car in picking up passengers from their home, office or hotel. The problem for the bus services is that at these low operating costs that charge will be minimal compared to the convenience.

I am not privy to the actual running costs of the Skybus or its capacity utilisation numbers. I suspect the actual loadings are very low early in the morning and late at night. This means that to have the concession the business must run some services that lose money. Despite this problem, plus the freeway and airport tolls, I suspect that it is currently a pretty good business to run. The price the Skybus charges is positioned in the marketplace as cheaper than taxis and airport parking, but not as cheap as a regular bus fare. How much the government gets of that in concession fees is not clear. Once driverless electric cars provide much cheaper car services to the airport, that price point becomes non-viable. As a business model, I cannot see it surviving in the medium to long term.

The new competitive model will probably be a combination of smaller buses plus personal cars from the normal transport as a service. The small buses can be employed just at peak travel times and redeployed for other purposes at other times. At non-peak times there will be more cars available and at a cheaper price point. This means it will be impossible for a bus company to supply the service as a standalone business. A more comprehensive transport as a service company that owns buses and cars will be able to supply a far more flexible and cost-effective service.

For related posts see:

### Driverless Buses: The Specific Medium Trip Proposition: Luxury and Space

### A Taxi to the Airport for $1.80

I am writing a book on autonomous vehicles with Dr Chris Rice . It is called **Rise of the Autobots: How Driverless Vehicles will Transform our Economies and our Communities. **Follow us here to see more excerpts as we write.

*Paul Higgins*

Come visit our website to see more of my work.

Read on if you want the detailed assumptions and calculations behind our numbers above — please feel free to weigh in and critique them. It will assist our analysis

### Appendix: Detailed Assumptions

#### Diesel Bus

The assumptions we put in the calculator were as follows:

· **Diesel price** $1.15 per litre — accessed October 10th 2017

· **Diesel fuel Rebate** 14.5c per litre — Liquid fuel rate for heavy vehicles travelling on heavy roads.

· Large Bus with 48 passengers

· Average Vehicle Burn Rate of 3km/Litre which matches a combination of flat route with slight gradients used in the ATAP figures.

· Distance travelled per day was 903 km. This was a workaround of given the bus calculator on the site is locked at 5 days per week and the Sky bus runs 7 days a week. We have assumed that the turnaround for the 44-km trip is 1.5 hours factoring in pick up times. The service runs every 10 minutes between 6 am and midnight so that equates to 12 trips in that time. The service runs every 15 minutes from Midnight until 6 am. So, we have assumed 2.67 trips in that time as some buses would be pulled out of service because of the more spread out timetable. This means a bus travels 645km a day which is 4517 km per week. To fit that into the locked 5 days in the calculator we therefore assume 903 km for 5 days.

· The bus in the calculation only runs 46 weeks of the year. That figure allows 6 weeks for downtime and service requirements. That may be during normal operating days or out for extended periods of major maintenance.

· For interest we have used 6.5% interest over five years on a $301,600 purchase price with a 25% residual at the end of the five years. The interest is averaged over the five years.

· For depreciation we have used the same numbers as the interest calculations. Assuming that the 5-year-old bus with 1.03589 million kilometres on the clock can be sold for $75,400.

· For insurance/registration we doubled the amount to account for the higher kilometres travelled.

This produces the annual costs in the following table for a diesel bus travelling 207,782 km per year:

### Electric Driverless Bus

For the comparison to an electric driverless bus we have changed the figures based on the data provided on the **Proterra** electric bus website as follows:

· For fuel we have used 1.5kWh per mile (1.60934km) at a kWh price of A12.57 cents

· For interest costs we have used a purchase price of A$600,000 with a 25% residual after 5 years. $200,000 of this was added for the electric bus costs and the rest for autonomous technology costs.

· Tires remain the same

· Maintenance and service, we have put at 64% of the diesel bus costs as per the Proterra costings.

· For depreciation we used the same capital figures as in the interest costs for the electric bus. However, we have adjusted them to reduce the depreciation in line with the km travelled as per our bus capacity calculations below. Because of downtime for charging in a service that runs 24 hours a day the electric driverless buses will travel less distance in a year than a diesel bus. Therefore, they will depreciate less. Even with this adjustment we think the depreciation calculations are very conservative given the experience with car longevity I have described in Electric Cars — Saving Real Money or Arbitrage Opportunity?

· Insurance was reduced by half based on reduced risk from driverless vehicles.

· $20,000 was included as a running cost for the technology platform for the autonomy part of the bus This includes mapping and operating system services. This is the economic number which is the least realistic as we have no models to compare it to.

· Finally, there was an adjustment for bus utilisation. In a service that runs 24 hours a day, recharging requirements for an electric bus cut into operating times. We have assumed that for a diesel bus refueling takes place as part of the normal rotation cycle. The E2 series Proterra bus has a nominal range of 251–350 miles depending on the route, and driver behavior. The route for the airport bus is relatively flat. This means that the longer range is probably applicable. We will be conservative and assume a re-charge every 240 miles. At a recharge rate of 12 miles for every 5 minutes that will is 2 hours per recharge including set up. At an average day of 645km that is 1.67 charges a day. That reduces bus availability by 3.34 hours a day or 13.9%. We have used an 86% bus availability figure to adjust the total km per year travelled downwards.

This results in the following annual costings for 178,693 km travelled.

#### Non- Vehicle Operating Cost Assumptions

There are a number of non-vehicle costs that will have a large impact on the price offered to the consumer. Let’s look at the majority of these for driverless buses and driverless cars with the assumptions we have used in our assessment:

Assumptions are:

· Freeway tolls are currently charged much higher for heavy vehicles than passenger vehicles but cap out on a daily limit. Hard to imagine that government would differentiate between the two services on a per head basis.

· Airport fees — same argument as freeway tolls

· Return on Capital — we have previously calculated that is 5 cents per km for a 13.8% internal rate of return for a driverless electric car. The buses are travelling more kilometres than the cars we used in that calculation but if cars replace the bus service then they will do similar travel distances. The bus is about ten times the cost of a reasonable estimate for an autonomous electric production car. This means that the return on capital for an average of 2 people in a car service would be the same as a bus carrying 20 passengers.

· Billing systems are unlikely to be different on a per person basis.

Fuel taxes are included on the basis that the government would want to claw back the taxes they currently get on fuel excise (A$.403 per litre) by charging electric vehicles a fee. Heavy diesel buses currently get a heavy transport fuel tax discount. It is hard to argue that cars as a transport service should be taxed differently than buses. Therefore, we have included the full rate here on a per kilometer basis. For a 2-passenger car vehicle that would be 1.21 cents per km per passenger or 26.62 cents per passenger for the trip. For a bus with 20 passengers it would be 0.67 cents per passenger kilometer or 14.74 cents per passenger trip. That is 45% cheaper but 12 cents per passenger difference