Updated: Apr 29
Over the last weeks, it has been discussed how social distancing will be applied in theatres and venues. As articles have hinted that social distancing may be here to stay, venue professionals from all around the world began printing their seating maps and started highlighting seats. The conclusion from the first articles was that capacity will be very low, resulting in a lot of pressure from the venue holders – and that’s not even taking into account the complexity of getting people in and out of the venue (or to the restroom).
However, if there is one thing that we are certainly learning from Covid19, it is that the dynamics are changing all the time. Over the last 6 weeks, restrictions have been gradually lifted in the Nordic countries, with the expectation that it could result in an increase in new cases – but, so far, it hasn’t. This has caused more restrictions to be lifted, and includes that social distancing requirements are now 1 meter (3 feet) as opposed to 2 meters (6 feet). On an even more positive note, in Italy, concerts are returning, which is a huge step for them and gives organizations a lot to consider.
When applying social distancing to the equation, how can the hall be set to get the highest capacity? Well, if you ask the CTO of Activity Stream, Helgi Helgason, that’s for an algorithm to find out.
So, he extracted the hall structure data for His Majesty’s Theatre in Aberdeen and had the head of business transformation, Leon Gray, take a few measures (to translate x/y pixel coordinates into actual measurements), in order to give the algorithm the necessary information about the physical placement of seats.
For context, the venue is extremely closely seated, with seats less than 50 cm / 20 inches wide, and with less than 90 cm / 35 inches between rows, so it was expected that results would be discouraging, but it was interesting to see if an algorithm could allocate seats very differently than the “manual” attempts.
And it would.
The first run was done with a requirement of distancing by 2m (6 ft 7 in), resulting in a capacity of 18%, and looked like this:
Notice how large groups are the preference, given that every person ‘triggers’ empty seats all around them.
By limiting the maximal group size to 4 results in a capacity of 17,5%, so not much lower:
The interesting part was seeing how capacity would change with changes in distancing requirements, so more experiments were run with 1,5m, 1,2m and 1,0m (but keeping the group limit to 4), and the capacity increased from 17,5% to 27,6%, 29,5% and to 37% – the low numbers caused by the fact that the hall has rows less than 1m from each other.
The optimal seating plan for the highest capacity while following the safety guidelines looks like this:
The interesting part is seeing the mix of group sizes – if no limit is placed on group size, the capacity increases to 44% (!) but with a lot of seats allocated to groups of 6-10, and the back row filled out with a group of 17 people.
And…the small details matter even when comparing the exact measurements of meters to feet. For example, if the social distance is measured in meters, 1-meter distance reaches a capacity of 37%, while a 3-feet requirement results in a capacity of 44%.
The algorithm has been set up to be able to handle various restrictions and can be applied to seat map data with x/y coordinates. Activity Stream plans to run the algorithm on a few more seat map samples and will update as restriction scenarios play out.