I saw a recent game in which a Masters player started distance mining to his natural rather than have more than 16 harvesters on his main. That struck me as probably wrong, so I did a quick experiment to check. The easy conclusion is that you should fill a base with 24 harvesters before distance mining to another mineral line: harvesters from 17 to 24 on one mineral line collect minerals twice as fast as harvesters that travel to another mineral line.
You can download the game I played if you want to check my calculations. I was on Derelict Watcher, as that’s the map I saw the mistake on. I also had a very easy Protoss AI in the game as well, so I spent some minerals building defences. However, I accounted for those in the calculations below.
Stage One: Six Harvesters Remote Mining
Between 1:00 and 4:30 in the game, six harvesters collect from the natural. That’s 6 * 3.5 = 21 SCV-minutes of harvesting. I also spent 100 on a supply depot and 50 on an SCV in this time. Stored minerals increased from 80 to 125; adding on the 150 spent giving 195 minerals collected. 195 minerals / 21.5 SCV-minutes => 9 minerals per minute per SCV.
Stage Two: 16 Harvesters in the Main
Between 14:00 and 16:00, 16 harvesters collected in the main, for 32 SCV-minutes of collection. Stored minerals went from 251 to 1551, an increase of 1300. 1300 minerals / 32 SCV-minutes => 41 minerals per minute per SCV.
Stage Three: 24 Harvesters in the Main
Between 16:45 and 18:30, 24 harvesters collect in the main, for 42 SCV-minutes of collection. Stored minerals increased from 1795 to 3195, an increase of 1400. Now 1400 minerals / 42 SCV-minutes => 33 minerals per SCV minute overall.
But we can also look at this in marginal terms. From stage two, we know that the first 16 SCVs collect an average of 41 minerals per minute. For the average to fall to 33 when the next eight SCVs are added, the contribution of the last eight SCVs must be (3 * (1400 / 42)) – (2 * (1300 / 32)) => 19 minerals per minute per SCV.
Stage Four: 16 Harvesters in the Main, Eight Mining Remotely
Between 19:30 and 22:30, 16 harvesters mined in the main, and eight mined in the natural. Stored minerals increased from 3071 to 5301, an increase of 2230. As with stage three, we first calculate the overall mining rate. 2230 minerals / 72 SCV-minutes => 31 minerals per minute.
Now we calculate the marginal rate for the eight remote mining SCVs, using the 41 minerals / SCV-minute figure we calculated in stage two for the average collection rate of 16 SCVs collecting in the main: (3 * (2230 / 72)) – (2 * (1300 / 32)) => 12 minerals per minute.
The distance an SCV travels from the natural to the main obviously makes a difference to how fast it can collect minerals when remote mining. The distance on Derelict Watcher seems about normal to me, but it may be shorter on maps like Korhal City.
There was some uncertainty in the results for remote mining. I put this down to the fact that only a small number of mining cycles were completed in the experiments. Remote mining could also be made more efficient by manually steering SCVs onto the closest patch on the remote mineral line. I don’t expect this would improve the rate by more than 15%-20%.
Lastly, remote mining carries an additional risk of losing SCVs if they have to travel across areas where they can more easily be attacked.
I’ve confirmed, as others have, that the first 16 SCVs on a mineral line collect faster than the next eight. The figures I got were in line with those on Liquipedia. I also found a rough figure for mineral collection rate when distance mining. It’s not exact, but it’s definitely lower than the value for the 17th to 24th SCVs on a mineral line.
First 16 SCVs on a mineral line: ~41 minerals / minute
17th to 24th SCVs on a mineral line: ~19 minerals / minute
SCVs remote mining to the natural: ~9-12 minerals / minute