High or Ultra Performance LC, a matter of particle size.
Comparison of columns, and particularly the benefits of the use of smaller particle sizes is less straight forward as it looks at a first glance. Fair comparison should take into account other factors than the parameter plates/cm like there time, financial and technical constrains.
Long columns packed with 5um particles may perform equally well or better compared to shorter column packed with (sub) 3 um particles. One of the prominent differences is that (sub) 3 um column can provide more separation power ( plates, resolution) per unit of time, however at the payback of an higher backpressure. High column pressure drop require the use of a more expensive  high pressure pumps. High pressures may also be the cause of technical difficulties like leaking or bursting connections.
The table below summarizes some LC characteristics associated with particle size. It is obvious that longer columns require longer runtimes. However long columns posses a higher capacity, larger dynamic range and yield more peaks per unit of time. [c.f extremities: 100 peaks in 22 min on a 5cm 1 um column or ~5 peaks/min versus 504 peaks in 60 min on a 120cm 5 um column or ~8 peaks/min]
particle size backpressure max. column length plates/cm max # plates Max peak capacity Typical runtime*
(um) (bar/cm column) at 400 bar limit (relative) at 400 bar at 400 bar limit (20 min @ 3%B/min)
1 83 4.8 500 2400 100 22
1.7 28 14 294 4116 171 25
3 9.2 43 167 7181 299 35
5 3.3 121 100 12100 504 60
Runtime calculated for max column lengths (Lmax) at 400 bar operated at a linear velocity of 6 cm/min, an elution time window of 20 min using a gradient of 3% B/min and a column reconditioning period that equals the  length of the column divided by the linear velocity of the mobile phase (6 cm/min).
Small particle sizes yield lower plate heights. The relation is linear. Also, smaller particles create an higher pressure drop across the column. This relation is quadratic. When comparing two columns packed with different particle sizes, dp1 and dp2, the plate height and pressure drop compares according to:

HETP = dp2/dp1
ΔP = (dp1/dp2)2
dp HETP N pressure drop
um um plates/cm bar/cm
10 10 1000 1
5 5 2000 4
2 2 4000 25

However,  the observed  peakwidth (σpeak) comprises three dispersion components: σ2peak2injection+ σ2column2detector. Hence, if peak broadening occurs during the injection or in the detector (esi source) or both and are of the same order to the dispersion in the column than one would get little or no benefit from using small particles. Rather, one would suffer from the disadvantage, increased pressure drop. This is particularly true for  <2 um particles (6-7 times higher pressure drop compared to 5 um particles).

 

Comparison of the peak width obtained on two otherwise similar columns except particle size. Top: 3 um dp, middle and bottom, 5 um dp Biosphere C18. The top and bottom separations are obtained at the corresponding optimal column head pressure (175 and 100 bar for the 3um and 5 um column, resp). The middle bar represents the separation on a 5um column operated at a pressure similar to the 3 um column. Peak widths are about 8 s (4σ ) or 2.5-3 s (2σ)  for both 3 and 5 Ám.