Water Supply System Design — Population Forecasting & Demand

1. Components of a Water Supply System

2. Population Forecasting Methods

The design of a water supply system is based on the future population the system must serve. Population is forecasted to the end of the design period (typically 30 years for transmission mains; 15 years for treatment plants) using historical census data.

2.1 Arithmetic Increase Method

2.2 Geometric Increase Method (Geometric Growth)

2.3 Incremental Increase Method

2.4 Decremental Rate of Growth Method

Used for cities approaching saturation (stable/declining cities). The growth rate decreases each decade by a constant amount. Not commonly tested in GATE CE.

2.5 Logistic (S-Curve) Method

2.6 Graphical Method

Plot historical population data and extend the trend curve graphically. Subjective but useful when mathematical methods are inconsistent. Can also use a “comparison city” whose growth pattern at a similar stage of development is projected onto the study city.

2.7 Summary — Method Selection

3. Water Demand — Per Capita and Categories

3.1 Per Capita Water Demand (IS 1172:1993)

3.2 Total Per Capita Demand

4. Peak Demand Factors

Water demand is not uniform — it varies by hour of day, day of week, and season. The distribution system and storage must be designed for peak demand, not just average demand.

4.1 Design Demand for Different Components

5. Fire Demand

Fire demand is the additional water supply required to fight a major fire while maintaining minimum pressure (7 m or 10 psi) at all hydrants simultaneously. Though infrequent, fire demand can significantly exceed normal demand and must be accommodated in system design.

5.1 Fire Demand Formulas

6. Design Period

The design period is the number of years for which a water supply component is designed — the time horizon over which the system should adequately serve the forecast population without major upgrades.

7. Sources of Water

7.1 Surface Water Sources

7.2 Groundwater Sources

8. Storage — Balancing Reservoir and Overhead Tank

8.1 Service Reservoir (Balancing Reservoir)

A service reservoir (ground-level or elevated) stores water to balance the difference between the rate of supply from the treatment plant (uniform pumping) and the variable demand throughout the day.

8.2 Overhead Tank (OHT)

8.3 Emergency Storage

In addition to balancing storage, 1–2 days of emergency supply is stored to cover pump failures, pipeline breaks, or treatment plant shutdowns. CPHEEO recommends: total storage = balancing storage + fire storage + emergency storage ≈ 1–2 days of average daily demand.

9. Distribution Network — Types and Analysis

9.1 Types of Distribution Systems

9.2 Hardy Cross Method (Pipe Network Analysis)

The Hardy Cross method (see Civil_45 — Pipe Flow) solves the pipe network for flow distribution and head losses iteratively. For water distribution networks, the same method applies with head loss computed using Hazen-Williams formula instead of Darcy-Weisbach:

9.3 Pressure in Distribution System

10. Worked Examples (GATE CE Level)

Example 1 — Arithmetic Increase Method (GATE CE 2022 type)

Problem: Census data for a city is as follows: 1991 = 1,00,000; 2001 = 1,20,000; 2011 = 1,45,000; 2021 = 1,70,000. Forecast the population for 2041 and 2051 using the arithmetic increase method.

Answer: P₂₀₄₁ ≈ 2.17 lakh; P₂₀₅₁ = 2.40 lakh

Example 2 — Geometric Increase Method (GATE CE 2021 type)

Problem: Using the same census data as Example 1, forecast the 2041 and 2051 populations using the geometric increase method.

Answer: P₂₀₄₁ ≈ 2.42 lakh; P₂₀₅₁ ≈ 2.89 lakh (geometric method)

Example 3 — Incremental Increase Method (GATE CE 2020 type)

Problem: Using the same census data (1991: 1.00L, 2001: 1.20L, 2011: 1.45L, 2021: 1.70L), forecast 2031 population using the incremental increase method.

Answer: P₂₀₃₁ ≈ 1.96 lakh (incremental increase method)

Example 4 — Design Water Demand and Daily Volume (GATE CE type)

Problem: A town has a design population of 2,50,000 in the year 2051. Per capita demand = 135 LPCD (domestic); add 25% for industrial, commercial, and institutional demands; and 15% for losses. Find (a) average daily demand, (b) maximum daily demand, and (c) maximum hourly demand.

Answer: ADD = 48.5 MLD; MDD = 87.3 MLD; MHD = 5.46 MLD/hour

Example 5 — Fire Demand (GATE CE type)

Problem: A town has a population of 1,00,000. Find the fire demand using (a) Kuichling’s formula and (b) Freeman’s formula. Express in litres per minute.

Answer: Kuichling = 31,820 L/min; Freeman = 34,095 L/min; Design with higher = 34,095 L/min

Example 6 — Balancing Storage Volume (GATE CE type)

Problem: A town has an average daily demand of 5 MLD. The treatment plant pumps at a uniform rate for 8 hours per day. What is the minimum balancing storage volume needed in the service reservoir?

Answer: Balancing storage = 3333 kL = 3.33 ML (≈ 2/3 of daily demand)

11. Common Mistakes

Mistake 1 — Using n = Number of Years Instead of Number of Decades in Population Forecasting

Error: Forecasting 30 years ahead with n = 30 (years) instead of n = 3 (decades) in P_n = P₀ + n × ΔP.
Root Cause: Census data in India is collected every 10 years — so one “period” in population forecasting formulas is one decade (10 years), not one year. The increment ΔP is computed per decade.
Fix: In all population forecasting formulas: n = number of decades to forecast, not years. If forecasting 30 years from 2021 (to 2051): n = 3 decades. Verify by checking that ΔP_avg is in persons/decade — if it’s 20,000 per decade and you use n = 30 years, you’ll get 30 × 20,000 = 600,000 extra people instead of 3 × 20,000 = 60,000.

Mistake 2 — Confusing Maximum Daily Demand with Maximum Hourly Demand

Error: Designing distribution mains for MDD (1.8 × ADD) instead of MHD (2.7 × ADD/24).
Root Cause: Both are “peak demands” but for different purposes. Treatment plants and transmission mains are sized for MDD (peak in a day). Distribution mains must supply the maximum demand at any hour, which is MHD (much higher per unit time than MDD per unit time).
Fix: Treatment plant: sized for MDD. Distribution main capacity: sized for MHD. MHD in flow rate = 2.7 × ADD/24 hours. This is critical — a distribution main designed for MDD would be undersized and cause pressure drops during peak morning/evening hours.

Mistake 3 — Using Arithmetic Mean Instead of Average Percentage for Geometric Method

Error: Computing r = (r₁ + r₂ + r₃)/3 (arithmetic average) when geometric mean is more appropriate for the geometric method.
Root Cause: The geometric growth method ideally uses the geometric mean of growth rates: r_geo = [(1+r₁)(1+r₂)(1+r₃)]^1/3 – 1. However, in practice, most GATE CE problems use the arithmetic average of percentage growth rates, which is an acceptable approximation.
Fix: For GATE CE: use arithmetic average of percentage growth rates (r₁ + r₂ + r₃)/n for simplicity unless the problem specifies geometric mean. The error is small (1–3%) for growth rates below 20% per decade. If the problem asks for geometric mean, compute: r_avg = [(P_last/P_first)^1/n – 1] × 100%.

Mistake 4 — Applying Kuichling’s Formula Without Checking Units (P in Thousands)

Error: Substituting P = 1,00,000 (full population) into Q_f = 3182√P instead of P = 100 (in thousands).
Root Cause: In Kuichling’s formula, P must be in thousands. Using P = 1,00,000 gives Q_f = 3182 × 316.2 = 10,06,000 L/min — impossibly high and clearly wrong.
Fix: Always convert: if population = 1,00,000 → P in thousands = 100. Q_f = 3182 × √100 = 31,820 L/min. Similarly, Freeman’s formula also uses P in thousands.

Mistake 5 — Not Adding Losses When Computing Total Demand

Error: Computing design supply = population × domestic LPCD, ignoring the 15–25% unaccounted-for water (UFW) losses.
Root Cause: Students compute the net demand (water reaching consumers) but forget that the supply system must actually deliver more than net demand — the extra accounts for pipe leakage, metering errors, illegal connections, and operational losses.
Fix: Total supply = net demand / (1 – loss fraction). Or: add losses as a percentage on top. If net demand = 135 LPCD and losses = 15%: gross supply = 135/0.85 = 158.8 LPCD, or 135 + 0.15 × 135 = 155.25 LPCD (slightly different depending on whether % is on gross or net).

12. Frequently Asked Questions

Q1. Which population forecasting method should be used for a rapidly growing Indian city, and why?

For rapidly growing Indian cities — particularly medium-sized cities (2–10 lakh population) receiving significant rural-urban migration, IT/manufacturing employment centres, or newly declared municipal areas — the geometric increase method typically gives more realistic estimates because it assumes the percentage growth rate remains constant. If the city grew at 20% per decade historically (e.g., from 2 lakh to 2.4 lakh to 2.88 lakh), the geometric method correctly projects this proportional growth forward. The arithmetic method, by contrast, assumes the same absolute number is added each decade regardless of current size — which underestimates growth for expanding cities. However, for long-range forecasting (30–50 years), the geometric method often overestimates by projecting compound growth indefinitely — which no city sustains. In practice, Indian engineers use the incremental increase method as a compromise: it acknowledges changing growth rates and typically gives an intermediate forecast between arithmetic and geometric. For master planning purposes, multiple methods are applied and the professional judgement governs which is most credible given local economic and demographic conditions.

Q2. Why is the maximum daily demand factor 1.8 and the maximum hourly factor 2.7 in IS 1172?

These factors represent observed patterns of water use variation in Indian cities. The maximum daily demand of 1.8 × average daily reflects the fact that summer days (peak domestic use — filling overhead tanks, garden watering, cooling), festival days, or post-maintenance resumption periods can demand up to 80% more than the annual average. This factor was derived from metered demand studies across Indian water utilities and incorporated into IS 1172. The maximum hourly factor of 2.7 × average hourly represents the sharp morning peak (6–9 AM) and evening peak (6–9 PM) seen in Indian cities when all households use water simultaneously for bathing and cooking — creating a demand surge 1.5× above the already elevated maximum day demand rate. The 1.5 and 1.8 factors combine to give 2.7 (1.5 × 1.8 = 2.7). These factors are averages across many cities; actual values can range from 1.5 to 2.5 for daily peaking and 1.2 to 2.0 for hourly peaking depending on the city size, supply hours per day, and consumer storage behaviour. Cities with intermittent supply (which most Indian cities have — 4–8 hours per day rather than 24 hours) show much higher instantaneous peaks because consumers rush to fill storage vessels when water is available.

Q3. What is non-revenue water (NRW) and why is it so high in Indian water utilities?

Non-revenue water (NRW) is the difference between the water supplied into the distribution system and the water billed to consumers. It consists of two components: real losses (physical leakage from pipes, storage tanks, service connections — typically 60–70% of NRW in Indian cities) and apparent losses (unauthorised consumption from illegal connections and meter under-registration — typically 30–40%). Indian water utilities average 30–50% NRW — significantly higher than the 10–20% seen in well-managed systems. The primary causes in India are: old distribution pipes (asbestos cement, cast iron pipelines from the 1940s–1970s still in service in many cities) with high corrosion and break rates; intermittent supply creating pressure transients that accelerate leakage; poor metering infrastructure (many unmetered connections, inaccurate old meters); and underfunded operations and maintenance departments that cannot afford systematic leak detection. The Atal Mission for Rejuvenation and Urban Transformation (AMRUT) specifically funds NRW reduction programmes in Indian cities — including pipe replacement, smart meters, district metering areas (DMA), and 24×7 water supply demonstration projects which paradoxically reduce NRW by maintaining stable positive pressure (preventing vacuum-induced contamination ingress) and making leaks easier to detect and locate.

Q4. How are fire demand provisions integrated with the regular water distribution system in Indian towns?

Fire demand is integrated into the distribution system design through three mechanisms. First, the distribution pipe sizes are verified to carry normal maximum hourly demand plus fire flow simultaneously without pressure dropping below the minimum 7 m head. The pipe diameter is increased at critical nodes where fire demand would otherwise cause inadequate pressure. Second, dedicated fire hydrants are located at maximum 100 m spacing in commercial/industrial areas and 150 m spacing in residential areas, connected to mains of minimum 150 mm diameter. Third, a portion of the service reservoir storage is reserved as fire storage: Q_f (litres/minute) × duration (typically 4 hours) = fire reserve volume. For a city of 1 lakh population: Q_f = 31,820 L/min × 240 min = 76.4 lakh litres = 7.64 ML. This fire reserve is not normally used for regular supply and is maintained at full capacity. The National Building Code (NBC 2016) and CPHEEO Manuals provide detailed specifications for fire hydrant locations, flow testing, and pressure requirements. Most Indian municipalities have historically under-provisioned fire water supply, which is a significant contributing factor to the severity of urban fires.