Skip to main content

Hardy Cross correction derivation and network

Short answer

The correction factor is ΔQ = –Σ r Q₀|Q₀| / (2 Σ r |Q₀|). Applying Hardy Cross to the given network yields a final discharge in pipe FG of 2.84 units, flowing from G to F.

Derive an expression for correction factor for discharge while performing Hardy Cross method of solving network problem. A water supply distribution network is developed in a ward of Kathmandu municipality as shown in the figure below. The water supply distribution features a cross fitting of diameter 0.3 m at junction D. The inlet and outlet quantities are given in the figure. Determine the discharge through pipe link FG if all pipes are manufactured with steel (f = 0.010 for 0.3 m dia pipes and 0.012 for 0.25 m dia pipes). [Figure: network of nodes A to G with inflow/outflow at nodes marked 2, 2, 4 and 1; the link table lists r values: AB 50, AC 50, BD 100, CD 80, DE 100, EF 100, FG 50, DG 50, ...; the scan is cut off after DG]

Derivation of the correction ΔQ\Delta Q (Hardy Cross). Head loss in a pipe is hf=rQnh_f = rQ^n (with n=2n = 2 for the Darcy-Weisbach law, r=8fLπ2gD5r = \dfrac{8fL}{\pi^2gD^5}). For a correct distribution the algebraic sum of head losses around every closed loop is zero (clockwise flows positive):

∑rQ∣Q∣n−1=0\sum rQ|Q|^{n-1} = 0

Take assumed discharges Q0Q_0 which satisfy continuity at the nodes but not the loop condition. Let the true discharge be Q=Q0+ΔQQ = Q_0 + \Delta Q, where the same correction ΔQ\Delta Q applies to every pipe of the loop (so continuity is preserved). Then

∑r(Q0+ΔQ)n=0  ⇒  ∑rQ0n+n ΔQ∑rQ0n−1+n(n−1)2ΔQ2∑rQ0n−2+⋯=0\sum r(Q_0 + \Delta Q)^n = 0 \;\Rightarrow\; \sum rQ_0^n + n\,\Delta Q\sum rQ_0^{n-1} + \frac{n(n-1)}{2}\Delta Q^2\sum rQ_0^{n-2} + \dots = 0

ΔQ\Delta Q is small compared with Q0Q_0, so the terms in ΔQ2\Delta Q^2 and higher are neglected:

ΔQ=−∑rQ0∣Q0∣n−1n∑r∣Q0∣n−1  →n=2  ΔQ=−∑rQ0∣Q0∣2∑r∣Q0∣\Delta Q = -\frac{\sum rQ_0|Q_0|^{n-1}}{n\sum r|Q_0|^{n-1}} \;\xrightarrow{n = 2}\; \boxed{\Delta Q = -\frac{\sum rQ_0|Q_0|}{2\sum r|Q_0|}}

The numerator is the algebraic sum of head losses in the loop (signs by direction), the denominator is the sum of absolute values. Apply the correction to every pipe (add it to flows in the clockwise direction, subtract it from anticlockwise ones, and apply both loops' corrections to a common pipe), and repeat until ΔQ\Delta Q is negligible.

Network discharge through FG

Reading of the figure. The scan is cut off at the bottom, so the part below B and D is assumed to be: A (inlet) joined to B and C; B-D, C-D; D-E, D-G; E-F and G-F. Nodes B, E, F, G have the outflows shown (2, 2, 4, 1) and the inlet at A must therefore be 2+2+4+1=92 + 2 + 4 + 1 = 9. Resistances rr (hf=rQ2h_f = rQ^2) as in the table: AB 50, AC 50, BD 100, CD 50, DE 100, EF 100, FG 50, DG 50 (the ff and DD values are already contained in rr). Flow units are those of the figure.

Initial distribution (satisfying continuity): AB 4, AC 5, BD 2, CD 5, DE 4, EF 2, GF 2 (G to F), DG 3.

Loop 1 (A-B-D-C, clockwise: AB, BD positive; CD, AC negative) and loop 2 (D-E-F-G, clockwise: DE, EF positive; GF as F to G, DG as G to D negative).

Correction in each iteration, ΔQ=−∑rQ∣Q∣2∑r∣Q∣\Delta Q = -\dfrac{\sum rQ|Q|}{2\sum r|Q|}:

| Iter | ∑rQ∣Q∣\sum rQ|Q| (L1) | 2∑r∣Q∣2\sum r|Q| (L1) | ΔQ1\Delta Q_1 | ∑rQ∣Q∣\sum rQ|Q| (L2) | 2∑r∣Q∣2\sum r|Q| (L2) | ΔQ2\Delta Q_2 | |---|---|---|---|---|---|---| | 0 | -1300.00 | 1800.00 | 0.722 | 1350.00 | 1700.00 | -0.794 | | 1 | 26.08 | 1872.22 | -0.014 | 63.06 | 1541.18 | -0.041 | | 2 | 0.01 | 1870.83 | -0.000 | 0.17 | 1532.99 | -0.000 |

After the third iteration the corrections are below 0.001. Final discharges:

LinkABACBDCDDEEFGFDG
Q4.7084.2922.7084.2923.1651.1652.8353.835

Check at the nodes: D receives 2.708+4.292=7.0002.708 + 4.292 = 7.000 and sends 3.165+3.835=7.0003.165 + 3.835 = 7.000; F receives 1.165+2.835=4.0001.165 + 2.835 = 4.000, equal to its outflow of 4.

Answer: The discharge in link FG is 2.842.84 units (flowing from G to F), ≈2.8\approx 2.8.

All Hydraulics chapter 3 questions →