English is currently available.
Control & Instrumentation Engineering DISCUSSION
I need a control valve for a cooling water line: DN80 pipe, normal flow 30 m³/h, minimum 10 m³/h. My first thought was to buy a DN80 globe valve to match the pipe, but I am told that control valves are sized by flow coefficient, not by pipe size.
How is the coefficient calculated for water, what pressure drop should I assume across the valve, and what goes wrong if the valve is simply line size?
For a liquid in ordinary turbulent, non-flashing flow the metric form is Kv = Q × sqrt(SG / ΔP), with Q in m³/h, ΔP across the valve in bar and SG the specific gravity (1 for water). Kv is simply the flow of water in m³/h that the valve passes with 1 bar across it. The US coefficient uses gallons per minute and psi, and Cv = 1.156 × Kv.
With 30 m³/h and 1 bar available across the valve, the required Kv is 30, or Cv about 35. You then pick a valve whose rated Kv is larger than this, so that normal flow sits at roughly 60 to 80 percent of travel.
The pressure drop is not a free choice; it is what is left after the rest of the system has taken its share. Take the pump head at the design flow, subtract static head and the friction losses in pipe, fittings and heat exchangers at that flow, and the remainder is what the valve must absorb. As flow falls, friction losses fall with the square of flow and most pumps deliver more head, so the valve sees a higher ΔP at minimum flow.
If the valve at minimum flow sees 2.5 bar, the required Kv is 10 / sqrt(2.5) = 6.3. Calculate both ends, because the valve has to control well at both.
A line-size valve is nearly always too big. A DN80 globe valve typically has a rated Kv in the region of 100, about three times what you need; the manufacturer's table gives the exact value. It would then do all its work in the first part of its travel. Small stem movements cause large flow changes there, so positioner resolution, friction and backlash are magnified and the loop tends to cycle. Throttling close to the seat also wears the trim faster.
It is normal for the control valve to be one or two sizes smaller than the pipe, fitted between reducers.
The inherent characteristic matters as well. A linear trim gives equal flow change per unit of travel at constant ΔP. An equal-percentage trim gives an equal percentage change of the current flow per unit of travel, so it is fine at low openings and coarse at high openings.
When the ΔP across the valve rises as flow falls, as in the pumped system described above, equal percentage compensates and the installed behaviour comes out closer to linear, which keeps the loop gain more constant. Linear suits cases where the valve ΔP stays roughly constant.
Two checks before ordering. First, with high pressure drops in liquids, ask the supplier to check for cavitation and choked flow; the simple formula stops being valid once the pressure inside the valve falls to the vapour pressure, and cavitation damages trim quickly.
Second, decide the fail position: on loss of air or signal the actuator spring drives the valve open or closed. For cooling water fail-open is the usual safe choice, but that should come from the hazard review of the process, not from habit.
Back to Control & Instrumentation Engineering community
