9.3 Pipe Network Analysis

Pipe Network Analysis

Pipe networks distribute fluid through interconnected pipes, junctions, and loops. Analyzing these systems means figuring out how flow divides among different paths and what pressure drops result. Two conservation laws govern the analysis, and the Hardy Cross method is the standard iterative technique for solving the resulting equations.

Conservation Principles for Pipe Networks

Two fundamental laws constrain every pipe network: conservation of mass and conservation of energy. Every valid solution must satisfy both simultaneously.

Conservation of mass requires that the total flow entering any junction equals the total flow leaving it:

$\sum Q_{in} = \sum Q_{out}$

This is sometimes called the junction equation or node equation. If three pipes meet at a junction and two carry flow in, the third must carry the combined flow out. No fluid accumulates at a junction in steady-state analysis.

Conservation of energy requires that the net head loss around any closed loop in the network equals zero:

$\sum_{loop} h_L = 0$

This is the loop equation. It follows from the fact that pressure at any point is unique. If you trace a path around a loop and return to the starting node, all the pressure gains and losses must cancel. (Pipes carrying flow in the assumed clockwise direction contribute positive head loss; pipes carrying flow against that direction contribute negative head loss.)

For any single pipe segment, the energy equation between two points is:

$\frac{p_1}{\rho g} + \frac{v_1^2}{2g} + z_1 = \frac{p_2}{\rho g} + \frac{v_2^2}{2g} + z_2 + h_L$

Continuity along a single pipe with varying cross-section gives:

$Q = v_1 A_1 = v_2 A_2$

This lets you find velocity changes at expansions or contractions within a pipe segment.

Head loss equations. Friction losses in each pipe are typically computed with one of two formulas:

• Darcy-Weisbach (general, any fluid): $h_L = f \frac{L}{D} \frac{v^2}{2g}$
• Hazen-Williams (empirical, water only): $h_L = \frac{10.67 \, L}{C^{1.852} \, D^{4.87}} \, Q^{1.852}$

In both, head loss increases with pipe length $L$ and decreases with pipe diameter $D$. The Darcy-Weisbach equation uses the friction factor $f$ (from the Moody chart or Colebrook equation), while Hazen-Williams uses a roughness coefficient $C$ that depends on pipe material (e.g., $C \approx 150$ for new PVC, $C \approx 100$ for old cast iron).

Hardy Cross Method for Flow Analysis

The Hardy Cross method solves the loop equations iteratively. You start with a guess for how flow distributes, then refine it until the energy balance around every loop is satisfied.

Step-by-step procedure:

1. Set up the network topology. Identify all pipes, junctions, and independent loops. For a network with $P$ pipes and $J$ junctions, the number of independent loops is $P - J + 1$.

2. Assume an initial flow distribution. Assign a flow rate and direction to every pipe. The assumed flows must satisfy conservation of mass at every junction, but they won't yet satisfy the loop energy equations.

3. Assign sign conventions. Pick a direction (e.g., clockwise) as positive for each loop. Pipes with flow in the positive direction get positive $h_L$; pipes flowing against it get negative $h_L$.

4. Calculate head loss in each pipe using Darcy-Weisbach or Hazen-Williams based on the current assumed flow rates.

5. Compute the flow correction for each loop:

$\Delta Q = -\frac{\sum_{loop} h_L}{\sum_{loop} \frac{\partial h_L}{\partial Q}}$

For the Darcy-Weisbach equation (assuming $f$ is roughly constant), $h_L \propto Q^2$, so $\frac{\partial h_L}{\partial Q} = \frac{2 \, h_L}{Q}$. The correction becomes:

$\Delta Q = -\frac{\sum_{loop} h_L}{\sum_{loop} \frac{2 |h_L|}{Q}}$

Note the absolute value in the denominator. The numerator retains signs (positive or negative depending on flow direction relative to the loop).

1. Update flow rates. For each pipe in the loop, add $\Delta Q$ to the current flow: $Q_{new} = Q_{old} + \Delta Q$. Pipes shared by two loops receive corrections from both.

2. Repeat steps 4–6 until $\Delta Q$ for every loop falls below an acceptable tolerance (e.g., less than 1% of the smallest pipe flow).

A common mistake is forgetting the negative sign in the correction formula or mixing up sign conventions for shared pipes. Be consistent with your clockwise/counterclockwise assignments.

Once converged flow rates are known, you can convert head loss to pressure drop for any segment:

$\Delta p = \rho g h_L$

Pumping Power for Head Loss Compensation

Pumps add energy to the fluid to overcome elevation differences and friction losses. The power a pump must deliver to the fluid is:

$P = \rho g Q H$

where $P$ is the hydraulic power (W), $Q$ is the volumetric flow rate (m³/s), and $H$ is the total head the pump must supply (m).

Total head $H$ has two components:

• Static head: the elevation difference between the source and the delivery point.
• Loss head: the sum of all friction losses (major losses from pipe length) and minor losses (from fittings, valves, bends, etc.) along the flow path.

The actual electrical power consumed will be higher because of pump and motor inefficiencies:

$P_{actual} = \frac{\rho g Q H}{\eta_{pump} \cdot \eta_{motor}}$

For example, if a water distribution system requires $Q = 0.05$ m³/s against a total head of $H = 30$ m, the hydraulic power is $P = (998)(9.81)(0.05)(30) \approx 14{,}700$ W or about 14.7 kW. With a combined pump-motor efficiency of 70%, the actual power draw would be roughly 21 kW.

Optimization of Pipe Network Design

Designing a pipe network involves trade-offs. The main variables are pipe diameters, network layout, and pump placement.

Pipe sizing is the most direct trade-off: larger diameters reduce head loss (and therefore pumping costs) but increase material and installation costs. Smaller diameters save on materials but drive up long-term energy costs. Engineers typically evaluate this using life-cycle cost analysis, comparing the present value of energy costs over the system's lifetime against the upfront capital cost of larger pipes.

Network configuration also matters:

• Parallel pipes between two nodes split the flow, reducing velocity in each pipe and lowering total head loss compared to a single pipe carrying the full flow.
• Series pipes add their head losses together, increasing the total loss along the path.
• Looped networks (common in municipal water systems) provide redundancy and more uniform pressure distribution compared to branching (tree) networks.

Formal optimization of large networks uses techniques like linear programming, genetic algorithms, or particle swarm optimization. The objective function might minimize total cost (capital + operating), minimize energy consumption, or meet minimum pressure requirements at all demand nodes. Constraints include maximum allowable velocities, minimum pressures, and available pipe sizes.

These optimization approaches are applied in water distribution networks, district heating systems, gas pipelines, and industrial process piping where the scale justifies the computational effort.