AI timetable optimisation: key metrics for efficient public transportation
Public transportation companies now more than ever face challenges in providing efficient and reliable services due to the prevalent problems of long waiting times, bus congestions, or delays. Luckily, AI can offer innovative solutions to age-old problems - with its assistance, public transportation timetables can be optimized to avoid the aforementioned problems, ensuring flexibility and high business value. In this article, we will delve into the challenges of timetable optimisation, along with the key metrics that are detrimental to success.
Understanding the Process
While working on the project “Govtech Transport” our team had to explore effective offline timetable optimisation ways and metrics, in order to provide a robust solution to KKT (Klaipėdos Keleivinis Transportas). We found out that there are several viable approaches to optimising public transport timetables, each tailored to different requirements.
One approach involves adjusting the number of buses in the timetable, either by adding or removing them. While this method can effectively improve headways and reduce congestion, it lacks flexibility. Generally, most public transportation companies operate with a fixed fleet size, so, adding a large amount of new vehicles may not be possible due to shortage, and removing active vehicles from the schedule is just counterintuitive.
The preferred and most adaptable method of offline timetable optimisation was found to be trip shifting. By adjusting the departure times of buses within the constraints of the existing schedule, we can optimise existing schedules with minimal disruption, yet significant benefits.
Understanding the Metrics
Now that we know the base process, we need some metrics to optimise - we will look at three of them, which are expected to bring the most business value for public transportation companies.
Max Headway
Max Headway refers to the maximum time interval between two consecutive bus arrivals at the same bus stop, showing the longest potential wait time for passengers. With the help of AI-driven optimization, the goal is to minimize this wait time, so that the whole traffic flows like clockwork. This is especially important for passenger satisfaction, as long wait times are rarely a good thing for a passenger.
However, it is essential to avoid over-optimisation, which can lead to high bus congestion. This scenario involves several buses arriving at the same bus stop simultaneously, which would make the situation even worse. While exploring the possible metrics, the congestion was determined to be one of the most important ones. We will call it Bus Bunching for simplicity.
Bus Bunching
Bus Bunching occurs when multiple buses arrive together at the same bus stop, creating gaps in service. It is like a traffic jam at a bus stop, causing frustration for passengers and inefficiency in service. When this occurs, clumped buses will have to wait for others to depart before departing themselves, increasing the chance of potential delays.
To address this issue, spreading out buses by adjusting departure times or temporarily holding some buses can improve the situation. But yet again, it's important to avoid excessive optimisation, which could worsen bunching at some stops and lead to longer passenger wait times at others.
So, the main takeaway for both (Max Headway and Bus Bunching) metrics is to not overdo it - the key to effective optimisation is keeping a good balance between both metrics and avoiding excessive minimisation of one, without considering the other.
Schedule Span
The last metric to keep in mind is Schedule Span. It represents the total duration between the first departure and the last arrival of buses within a day, and it serves as an indicator of invasiveness. That is, if we see a significant deviation in schedule length (compared to the original timetable), we can probably say that our optimisation algorithm is overly aggressive, which ideally should be avoided.
Although this metric is not used for direct optimisation, it nevertheless should be taken into account - no transport company will want to use your AI algorithm if it completely rearranges the original timetable, leaving a mess behind :)
Putting it All Together
The use of metrics such as Max Headway, Bus Bunching, and Schedule Span, along with differing AI optimisation techniques, can indeed bring significant business benefits to public transportation firms. Managing these metrics in a balanced manner leads to improved operational performance and overall passenger satisfaction.
Interested in boosting your business success and improving transportation operations? Contact us to schedule the first consultation, and our experts will find the optimal solutions for your enterprise.
