Parallel Hydronic Branches: Equal Pressure Drop Does Not Mean Equal Flow
A hydronic drawing can show two apparently similar branches between a supply header and a return header. That visual symmetry does not establish equal flow. Each branch responds to its own hydraulic resistance and the pressure difference available across its endpoints. A useful review therefore separates three questions: how the total flow divides, what pressure drop the branch group requires, and what additional loss occurs in piping shared by both streams.
1. Define the two common nodes
Label the branch-group inlet A and outlet B. For this simplified analysis, every branch starts at A and ends at B, with negligible pressure variation within each idealized header. Assume steady, single-direction liquid flow and comparable endpoint elevations and velocities, so the stated pressure difference represents the branch loss. These assumptions make the two branches share one pressure-drop boundary.
Real headers are distributed pipes. Branch takeoffs at different positions may see different available pressures because intervening header sections have their own losses. If those losses matter, represent each segment and junction explicitly. Do not apply one measured header differential indiscriminately to every remote terminal on a large network.
Caleffi’s circulation guide, in its discussion of parallel circuits, explains that branches with higher hydraulic resistance receive a smaller share of total flow. It also distinguishes branch-group resistance from resistance in common piping. The following numbers are an original teaching example, not a reproduction of its pipe correlation.
2. Declare the resistance equation and its units
Assume each branch follows Δp = kQ² over the range considered. Here Δp is in kPa, Q is in L/s, and k is in kPa/(L/s)². Branch A has k = 20; branch B has k = 80. These invented coefficients represent the complete losses within each branch, with its components in a fixed condition. They are not coefficients for a particular pipe size or product.
The square-law model is an approximation. Pipe friction factors and component behavior can change with flow and fluid properties. Caleffi’s cited guide uses an exponent of 1.75 in its particular pipe-based correlation. A coefficient derived for that equation cannot be inserted into this quadratic example. Preserve the equation, units and reference conditions together whenever a resistance coefficient is passed between calculations.
At an available branch differential of 20 kPa, rearrange the assumed equation to Q = √(Δp/k). The result follows directly from each branch’s resistance:
Branch A: Q = √(20/20) = 1.00 L/s.
Branch B: Q = √(20/80) = 0.50 L/s.
Total through the common supply and return: 1.00 + 0.50 = 1.50 L/s.
The pressure drop is equal because the endpoints are common. Flow is unequal because resistance is unequal. A fourfold resistance produces half the flow in this particular square-law model. Dividing 1.50 L/s equally between the branches would produce incompatible pressure drops: 11.25 kPa in A and 45 kPa in B at 0.75 L/s each.
3. Replace the group with an equivalent resistance
For parallel branches sharing Δp, total flow is √Δp multiplied by the sum of 1/√k for every open branch. Therefore k-equivalent = 1/(Σ1/√k)². This result comes from adding flows at a common pressure, rather than adding the individual resistance coefficients.
For the example, k-equivalent = 1/(1/√20 + 1/√80)² = 8.8889 kPa/(L/s)². Check it by returning to the known operating point: 8.8889 × 1.50² = 20.00 kPa. Equivalent resistance is lower than either individual branch resistance, which is consistent with providing additional parallel paths.
Another check starts with a specified total of 1.20 L/s. The equivalent model predicts 8.8889 × 1.20² = 12.80 kPa. Individual branch flows then become √(12.80/20) = 0.80 L/s and √(12.80/80) = 0.40 L/s. Their sum matches the specified total. This reverse calculation is useful for catching unit or spreadsheet-reference errors.
Original quadratic example: branch flows add; parallel pressure losses do not. The total circuit loss is 35 kPa when common piping adds 15 kPa at 1.50 L/s.
4. Add common losses once
Suppose the common piping loses 15 kPa at the example total of 1.50 L/s. The complete modeled circuit requires 15 + 20 = 35 kPa at that flow. It does not require 15 + 20 + 20 = 55 kPa. A stream traverses the common piping and one branch path; it does not pass through both parallel branches in series.
If the common piping also follows a square law, its coefficient is 15/1.50² = 6.6667 kPa/(L/s)². Total equivalent resistance becomes 6.6667 + 8.8889 = 15.5556. At 1.20 L/s, the modeled total loss is 22.40 kPa, comprising 9.60 kPa common loss and 12.80 kPa branch-group loss. The original 15 kPa is tied to its original flow, not a constant to reuse everywhere.
5. Distinguish a fixed boundary from a pump operating point
These calculations either specify the available branch differential or specify total flow. They do not independently determine what a real pump will deliver. That requires the pump characteristic and the full network characteristic at the relevant operating condition. Changing one branch can shift the network curve and consequently change the available pressure and both branch flows.
For example, removing branch B from the mathematical network leaves k-equivalent = 20. At a genuinely maintained 20 kPa across A and B, branch A would still carry 1.00 L/s. With a different pressure boundary, its flow could change. Avoid using the fixed-pressure result as a universal prediction of how an installed system responds.
6. Make the calculation reviewable
Mark exactly where the pressure-drop boundary starts and ends.
Keep each resistance coefficient beside its equation, units and assumed component condition.
Check that branch flows sum to the common-pipe flow.
Verify that parallel paths produce the same endpoint pressure difference.
Evaluate common-pipe losses at total flow and branch losses at individual flow.
Record omitted header losses and the conditions under which the simplified model needs refinement.
For related pump-duty calculation work, review the paid Pump Head Calculator Design Suite v1.0. A clear node diagram and a reproducible resistance model remain the foundation of the review: add flows across parallel paths, then add losses along the common series path.
Comments