Every once in a while we hear about circulating pool currents, and how that they created unfair racing conditions. The latest problem that I’ve heard of is the Windsor International Aquatic and Training Centre in Windsor, Ontario – host to the 2018 Eastern Canadian Swimming Championships in April, 2018.
Because I like stats, I decided to take a deeper look at this competition.
Two days before the meet, a “FINA appreciable current test” was run that showed there were no surface currents. However, as you’ll see below, it was quickly evident that there was a strong circulating current affecting races. A second FINA test still confirmed there were no surface currents. This suggests that either surface currents do not affect performance, or the FINA tests are useless.
So here’s the situation.
- On Day 1, the pool circulation system was working at 100% normal levels. Problems were quickly noticed
- It was most likely before Day 3 heats when the pool circulation system was reduced to 60% normal levels. Problems were still noticed.
- On Day 4 the system was reduced to 20% normal levels – the lowest level that pool maintenance staff were willing to go to ensure public health
My analysis uses analytical methods developed at Indiana University, and most popularly used by Barry Revzin in his analysis of the Rio Olympics for SwimSwam (see his articles here and here) . They include the following:
- Distance events: ignore first and last 100s, and then calculate the differences between outgoing and incoming 50 splits for each 100. A circulating current will make one direction faster than the other direction.
- 50 m races: on the assumption that everyone is at 100% effort in both heats and finals, changes in times based on the lane will help to illustrate the impact of any circulating currents.
I also added an analysis of the impact of swimming in the outside lanes in a 10-lane pool.
Distance events were held on Days 1 and 4. I analyzed 5 swimmers in each lane in each event, and then averaged those 5 results.
It’s quite clear that with the pool circulation at 100% on Day 1, there is a huge difference in outgoing and incoming 50 splits – a difference of up to 2.0 seconds for some lanes. Also, notice that the values are highest for lanes 1 and 8, with values slightly lower for lanes 0 and 9. I suspect that’s because the outside walls present boundary conditions for the circulating currents.
On Day 4 we see that the 50 split differences have pretty much disappeared. Strangely, all values are now very slightly negative, meaning the outgoing 50 of each 100 was slightly faster than incoming 50. Not sure why that is so consistent.
50 m Races
The idea behind the analysis of the 50 m races is to note race time differences based on the different lanes used by a swimmer for heats and finals. If circulating currents exist, then changing lanes should create a difference in race times. There’s an underlying assumption here that both heats and finals swims involve 100% effort.
The charts below show results for Day 2 with pool circulation at 100% (50 BK, 50 FL), Day 3 at 60% (50 BR), and Day 4 at 20% (50 FR).
The analysis works like this. A swimmer who swam heats in Lane 0 and finals in Lane 7 would have a lane shift of -7 (0 minus 7). We can see from the charts that this swimmer would have added ~0.5 seconds to their 50 time on Day 2, ~0.2 seconds on Day 3, and virtually no change on Day 4. This nicely corresponds to the pool circulation rates.
In a similar way, a swimmer going from Lane 7 heats to Lane 0 finals would have a lane shift of +7 (7 minus 0), and would have dropped ~0.5 seconds, ~0.2 seconds, or no change in their finals time, depending on the pool circulation rates. Clearly, dropping the pool circulation down to 20% has fixed the problem.
Think about these results. Imagine adding or dropping a half second to your 50 time just based on your lane.
By the way, these Day 2 time differences are about 40% higher than Revzin’s Rio analysis of the sprints.
A way of verifying the above results is by checking the odds of qualifying for finals out of different lanes. If pool circulation changed sprint times, then we should see the odds change dramatically between Days 2 and 4.
Sure enough, we see the lower lane numbers dominating Day 2 qualifiers, and then shifting over to a relatively even distribution on Day 4.
It was only at FINA’s 2009 Congress that 10-lane pools were approved. The primary reason for this was simple: analysis showed that swimmers in the outside lanes of 8-lane pools experienced more turbulence due to the nearby wall, and therefore had to expend more energy to keep up with the other lanes. Using only 8 lanes in a 10-lane pool just makes things more fair.
However, we are now using all 10 lane pools in many meets, and the extra 2 lanes are really helping to speed meets up. But the problem is still there. Swimmers in outside lanes will still experience more turbulence, and still be at a disadvantage.
My analysis consists of determining the percentage of swimmers who beat their seed times for different lanes. The two charts below show results for Day 1 and 4 distance events, Day 1 200 FR, and Day 1 100 BK.
This is pretty conclusive. In each case, the outside lanes had a smaller chance of beating their seed time, with the chances ranging from 3% lower to 15% lower.
NOTE: I was actually looking to analyze Day 4 events and not Day 1 events, but the rate of beating seed times was unworkably low. As an example, there were only 32 swimmers in the Day 4 200 IM to beat their seed time, and another 170 who didn’t. Roughly a 16% overall rate.
Some things became very clear.
- significant circulating currents can seriously impact times, with the severity depending on the lane
- the impacts can be devastating in the 50 sprints
- the impact can be minimized by severely reducing pool circulation flows
- outside lanes are slower than the other lanes
- FINA does not yet have an accurate or useful test for identifying impactful circulating currents
But the scariest thing to me is that this 4-year old pool, with no history of circulating currents, suddenly became a nightmare pool. This implies that it could happen at any pool.
It took a while for experts to investigate, and they found clogged inlets, one deliberately blocked cold-water feed, imbalanced feeds, and other minor problems. The good news is that these experts, including specialists from Myrtha pools, were able to completely fix the problem.