A DN200 line, a DN100 electromagnetic flowmeter and two reducers welded in on site look like a finished job. Weeks later the reading jumps, the zero drifts and low-flow figures stop matching batch records. The reducer itself is rarely the cause. The failures are a flow profile that never recovered, a tube that is not full, and a ground that was never equalised.
Conclusion: calculate velocity, choose a concentric conical reducer, keep enough straight pipe, guarantee a full tube, bond all potentials, mount the meter last. Here D is the meter tube bore, not the process pipe.
Bore should be back-calculated from flow rate, not copied from the pipeline. Velocity of 1 to 3 m/s at normal flow is comfortable, the practical floor is 0.3 m/s, and short peaks should stay under 10 m/s. Slurries are better held at 1 to 2 m/s to limit liner wear. Use v = 4Q / (πD²) across minimum, normal and maximum cases.
Most overlooked: the converter bore parameter must be the tube bore, and range must be recalculated after any diameter change.
Use a concentric conical reducer aligned to the meter axis, convergent upstream and divergent downstream, with a single-side contraction angle of 7.5 degrees or less and length around L ≥ 7.6 x (D large - d small). Steeper angles separate flow from the wall and no downstream length will fix it. Sudden contraction or expansion is prohibited: bolting DN200 straight to a DN100 body, or forcing the bore with an eccentric gasket, is a red-line error. Eccentric reducers are acceptable only where the manufacturer's technical file permits them.
The reducer is a disturbance; the equal-bore section between reducer outlet and meter flange lets the profile recover.
| Condition | Upstream | Downstream |
|---|---|---|
| Code minimum | ≥ 5D | ≥ 3D |
| After a reducer (recommended) | ≥ 5D, preferably 10D | ≥ 3D, preferably 5D |
| Elbow, pump or valve upstream | 15 to 20D, strictest value | ≥ 3 to 5D |
Control valves always go downstream. Where a reducer, elbow and pump discharge coincide upstream, apply the strictest single requirement, not a sum or average. Flow conditioners mitigate but never replace straight pipe at custody points.
The meter measures induced voltage in a full conductive liquid, so a half-full tube or bubbled electrodes read false.
Mechanical: flanges concentric and aligned, bolts tightened diagonally, gaskets centred, independent supports both sides, no pipe load or thermal stress on the body.
Electrical: housing, medium and flanges bonded to one potential. With lined pipe and insulating gaskets fit grounding rings at both ends, matching electrode material and tube bore with no step at the reducer. Bridge metal pipe with braided copper, keep independent ground resistance at 10 ohms or less, and never share the ground electrode with motors or VFDs. Route signal cable apart from power cable, shield grounded at one end.
Prefabricate and weld, fit a temporary spool in place of the meter, blow through and hydrotest, then install the sensor. Never weld with the meter in line. Before start-up confirm bore parameter, recalculated range, flow direction against the body arrow, a filled and vented line, zero calibration with a full tube at rest, and an empty-pipe check.
Direct DN200 to DN100 flanging causes abrupt contraction, bias flow and double-digit errors. A reducer hard against the meter flange feeds the disturbance straight into the tube. A single grounding ring gives a jumping signal and slow zero drift; an unchanged bore parameter gives a proportional offset across the range.
Standards: JB/T 9248-2015, GB 50131-2013, GB 50093-2013, SH/T 3551-2024, HG/T 20507-2014 and manufacturer manuals; values differ between them. Final decisions follow design documents and site conditions.
A DN200 line, a DN100 electromagnetic flowmeter and two reducers welded in on site look like a finished job. Weeks later the reading jumps, the zero drifts and low-flow figures stop matching batch records. The reducer itself is rarely the cause. The failures are a flow profile that never recovered, a tube that is not full, and a ground that was never equalised.
Conclusion: calculate velocity, choose a concentric conical reducer, keep enough straight pipe, guarantee a full tube, bond all potentials, mount the meter last. Here D is the meter tube bore, not the process pipe.
Bore should be back-calculated from flow rate, not copied from the pipeline. Velocity of 1 to 3 m/s at normal flow is comfortable, the practical floor is 0.3 m/s, and short peaks should stay under 10 m/s. Slurries are better held at 1 to 2 m/s to limit liner wear. Use v = 4Q / (πD²) across minimum, normal and maximum cases.
Most overlooked: the converter bore parameter must be the tube bore, and range must be recalculated after any diameter change.
Use a concentric conical reducer aligned to the meter axis, convergent upstream and divergent downstream, with a single-side contraction angle of 7.5 degrees or less and length around L ≥ 7.6 x (D large - d small). Steeper angles separate flow from the wall and no downstream length will fix it. Sudden contraction or expansion is prohibited: bolting DN200 straight to a DN100 body, or forcing the bore with an eccentric gasket, is a red-line error. Eccentric reducers are acceptable only where the manufacturer's technical file permits them.
The reducer is a disturbance; the equal-bore section between reducer outlet and meter flange lets the profile recover.
| Condition | Upstream | Downstream |
|---|---|---|
| Code minimum | ≥ 5D | ≥ 3D |
| After a reducer (recommended) | ≥ 5D, preferably 10D | ≥ 3D, preferably 5D |
| Elbow, pump or valve upstream | 15 to 20D, strictest value | ≥ 3 to 5D |
Control valves always go downstream. Where a reducer, elbow and pump discharge coincide upstream, apply the strictest single requirement, not a sum or average. Flow conditioners mitigate but never replace straight pipe at custody points.
The meter measures induced voltage in a full conductive liquid, so a half-full tube or bubbled electrodes read false.
Mechanical: flanges concentric and aligned, bolts tightened diagonally, gaskets centred, independent supports both sides, no pipe load or thermal stress on the body.
Electrical: housing, medium and flanges bonded to one potential. With lined pipe and insulating gaskets fit grounding rings at both ends, matching electrode material and tube bore with no step at the reducer. Bridge metal pipe with braided copper, keep independent ground resistance at 10 ohms or less, and never share the ground electrode with motors or VFDs. Route signal cable apart from power cable, shield grounded at one end.
Prefabricate and weld, fit a temporary spool in place of the meter, blow through and hydrotest, then install the sensor. Never weld with the meter in line. Before start-up confirm bore parameter, recalculated range, flow direction against the body arrow, a filled and vented line, zero calibration with a full tube at rest, and an empty-pipe check.
Direct DN200 to DN100 flanging causes abrupt contraction, bias flow and double-digit errors. A reducer hard against the meter flange feeds the disturbance straight into the tube. A single grounding ring gives a jumping signal and slow zero drift; an unchanged bore parameter gives a proportional offset across the range.
Standards: JB/T 9248-2015, GB 50131-2013, GB 50093-2013, SH/T 3551-2024, HG/T 20507-2014 and manufacturer manuals; values differ between them. Final decisions follow design documents and site conditions.