Few engineering achievements have saved as many lives as clean water and sewerage. In the middle of the nineteenth century, cholera and typhoid killed people in large numbers in the world’s richest cities, because the water they drank was contaminated by the waste they produced. Within a few generations, engineered systems for treating water, removing sewage and storing supply had made those diseases rare in cities that could afford them. Today the systems are so reliable that most people never think about them until something goes wrong.
The history of urban water is also one of the clearest sources of lessons about controlling risk in any operation. It shows why independent barriers matter, why a control that is working is not the same as a result that is being achieved, why systems should be designed for the day something fails, and why improving the source of a problem is often better than treating its effects. These lessons apply as much to quality control in a factory or food safety in a kitchen as to a city’s water supply.
This article traces how urban water engineering developed, from Roman aqueducts through the sanitary reforms of the nineteenth century to disinfection, elevated storage and Australia’s desalination plants, and then draws out lessons for managers of operations, quality and risk. It is general information, not advice on designing or operating water systems.
Gravity and distance: the Roman approach
Rome’s aqueducts carried water from distant hills into the city by gravity, along channels built with gentle, carefully surveyed slopes. Where valleys intervened, the Romans built arched bridges such as the Pont du Gard in southern France, and in a few places they used pressurised pipes that dropped into a valley and rose on the other side. The approach rested on two ideas that remain central to water engineering: find a good source, and use gravity rather than power wherever possible.
After the Roman period, many European cities drew water from wells and nearby rivers, often the same rivers that received their waste. As cities grew during the industrial era, that arrangement became deadly.
1854: linking water to disease
In 1854, during a cholera outbreak in London’s Soho district, the physician John Snow mapped the deaths and found them clustered around a public water pump on Broad Street. The investigation, together with a wider study of deaths among customers of different water companies, provided strong evidence that cholera spread through contaminated water rather than through bad air, the prevailing theory of the time. The handle of the pump was removed, and Snow’s work became a founding case of epidemiology.
Four years later, in the hot summer of 1858, the smell from the sewage-laden River Thames became so bad that it was called the Great Stink. Parliament approved a major scheme, and from 1859 the engineer Joseph Bazalgette built intercepting sewers that collected sewage along the river banks and carried it downstream, away from the city centre. The embankments that carry those sewers still line the Thames.
Supply and disposal are one system
Cities that drew water from the same rivers or lakes that received their sewage learned that the two halves of the water system could not be designed independently. Chicago discharged its sewage into the Chicago River, which flowed into Lake Michigan, the source of its drinking water. Raising the city’s streets to install sewers in the 1850s and 1860s helped drainage but did not solve the conflict. In 1900, the city completed a canal that reversed the flow of the Chicago River, sending sewage away from the lake.
The underlying lesson is that dilution is not treatment. A large body of water reduces the concentration of contaminants but does not eliminate pathogens, and currents connect intakes and outfalls in ways that are hard to predict. The engineering answer was separation, of outfalls from intakes and of waste from water, followed by treatment.
Treatment: barriers in sequence
Water treatment developed as a series of barriers, each addressing a different kind of contaminant:
- Screening removes coarse solids.
- Coagulation adds a chemical such as alum, which neutralises the electrical charge that keeps fine particles suspended so they can clump together.
- Flocculation gently mixes the water so that the clumps grow.
- Settling lets the clumps sink in slow-moving basins.
- Filtration through sand or membranes removes most remaining particles and many microorganisms.
- Disinfection inactivates surviving pathogens.
The order matters. Particles shield microorganisms from disinfectants, so removing solids first makes disinfection far more effective. Plants that struggle with water clarity often have a problem at the coagulation stage rather than at the filters.
Disinfection and the value of a residual
Chlorination of public water supplies spread in the early twentieth century; Jersey City in the United States began continuous chlorination of its supply in 1908, and disinfection of filtered water became standard practice in the following decades. Typhoid deaths fell dramatically wherever it was adopted.
Chlorine is still widely used, not because it is the strongest disinfectant, but because it leaves a residual that keeps working in the pipes after the treatment plant. A distribution network includes many kilometres of pipe, joints and repairs. A measurable chlorine residual at the far end of the network is continuous evidence that nothing has entered the system along the way. Ozone and ultraviolet light are powerful treatments but leave no protection in the pipes, so they are usually combined with a residual disinfectant.
Two barriers on the same principle are one barrier
Disinfection is valuable partly because it works on a different principle from filtration. If a filter fails, chlorine still acts; if chlorine dosing fails, filtration still removes many organisms. Barriers that rely on the same mechanism can fail together, so they offer much less protection than they appear to.
A barrier that worked while the outcome did not
An established history of American engineering records that Poughkeepsie, New York, built one of the first sand filtration plants in the United States in 1872, treating water from the Hudson River. Yet a hospital continued to discharge sewage into the river a short distance upstream of the intake for decades. Typhoid deaths continued at an erratic rate after filtration began and fell sharply only after the plant was improved and the hospital’s sewage was treated from the 1930s.
The filters were in place and operating throughout. The outcome they existed to produce was not being achieved, because the contamination entering them far exceeded what they were designed to handle. A control judged on whether it is running, rather than on whether it is achieving its purpose, can appear successful for a very long time.
When barriers fail together
Modern failures show the same lessons. In 2000, in Walkerton, Ontario, contaminated water from a town well caused an outbreak of E. coli infection in which seven people died and about 2,300 became ill. The public inquiry found that chlorination had been inadequate, that monitoring had not been carried out and reported properly, and that operators lacked the training to understand the risks. Several layers of protection failed at once.
In Australia, detections of the parasites Cryptosporidium and Giardia in Sydney’s supply in 1998 led to boil-water notices across much of the city. The subsequent inquiry led to stronger catchment protection and to the creation of a separate authority to manage Sydney’s drinking water catchments. Today, the Australian Drinking Water Guidelines set out a risk-based framework that manages water quality from catchment to tap through multiple barriers, rather than relying only on end-of-pipe testing.
Reaching for a better source
Many cities chose to protect distant catchments rather than treat polluted local rivers. New York’s Catskill system, completed in the 1910s, brought water from mountain reservoirs through an aqueduct that passes more than 300 metres below the Hudson River. Melbourne protected forested catchments in the ranges east of the city, giving it water that needed comparatively little treatment for much of the twentieth century.
Improving the source is often cheaper and more robust than treating poor water, but a protected catchment remains valuable only while the land draining into it stays protected. That requires an institutional commitment that lasts for generations, not just a construction project.
Sewerage: designed for gravity
Water supply mains are pressurised pipes that must stay full and tight. Sewers, by contrast, are usually laid as channels that flow partly full, on carefully chosen grades so that gravity carries solids along. A sewer that is too flat lets solids settle, reducing its capacity and eventually blocking it; one that is too steep causes turbulence, odour and wear. Designers typically aim for a minimum self-cleansing velocity, commonly around 0.6 to 0.75 metres per second at design flow, which makes sewer design largely a matter of survey and grade.
Modern wastewater treatment follows a sequence of physical settling, biological treatment by microorganisms and, increasingly, further treatment to remove nutrients and allow recycling. Managing the sludge produced is often a large share of operating cost.
Elevated storage: designed for the failure state
A water tower or hilltop reservoir looks simple, but it performs three jobs without moving parts or energy once filled:
- It sets network pressure. Each metre of water height produces about 9.8 kilopascals of pressure, so a tank 30 metres above a service point delivers a little under 300 kilopascals before friction losses.
- It meets demand peaks. Morning and evening peaks are supplied from storage, so pumps and mains can be sized closer to average demand.
- It rides through failures. If pumps or power fail, the network stays pressurised for hours.
The last point matters more than it first appears. Every buried network leaks a little. While it is pressurised, water leaks out and nothing gets in. If pressure is lost, groundwater, which may be contaminated, can be drawn in through the same defects. That is why water utilities often issue boil-water notices after a main loses pressure. Storage buys time to restore pressure before that happens, and it keeps fire hydrants working when they may be needed most.
Manufacturing water: desalination
The Millennium Drought, which affected much of south-eastern Australia from the late 1990s until about 2009, together with a long-term drying trend in south-western Western Australia, pushed Australian cities to secure supplies that did not depend on rainfall. Perth commissioned Australia’s first large seawater desalination plant in 2006, and plants serving the Gold Coast, Sydney, Adelaide and Melbourne followed within about six years.
These plants use reverse osmosis, pumping seawater at high pressure against membranes that let water through but hold back salt. Seawater’s osmotic pressure is around 27 bar, and plants typically operate at 55 to 70 bar. The economics depend heavily on energy recovery: devices called pressure exchangers transfer most of the pressure energy in the salty reject stream back to the incoming seawater, cutting energy use dramatically. Membrane life depends mostly on pretreatment that removes particles and organic matter before they foul the membranes.
Several of these plants have operated at low output in wet years. That drew criticism, but they were bought as insurance: their value lies in availability during drought, not in constant use. Whether that insurance was priced well against alternatives such as recycling and demand management is a fair debate; judging it by utilisation alone misreads its purpose.
Lessons for operations and quality managers
Use independent barriers
In a factory, food business or service operation, layers of protection are effective only when they fail differently. Two inspections by the same method at the same stage add little. The process FMEA and control plans article explains how to design controls that address different causes of failure.
Monitor the outcome, not just the control
Poughkeepsie’s filters ran for decades while typhoid continued. Check that controls achieve their purpose, through outcome measures such as defect rates, customer returns or test results, not only that equipment is switched on. The making a risk register change decisions article covers how to track whether barriers actually work.
Sequence controls so each protects the next
Removing particles makes disinfection work. In manufacturing, cleaning before inspection, deburring before coating and calibration before measurement play the same role.
Design for the failure state
Ask how a system behaves when power, supply or a key machine fails, and provide buffers, storage or passive protection that work without intervention.
Improve the source
Fixing a supplier’s process, a material specification or an upstream step is often cheaper and more reliable than inspecting and treating the consequences downstream.
Treat dilution with suspicion
Mixing a problem into a larger volume, whether contaminated material, nonconforming parts or unresolved complaints, reduces its visibility, not its effect.
Value insurance as insurance
Backup equipment, spare capacity and reserve stock look wasteful in good times. Judge them by their value in the conditions they exist for.
A worked example
This is an illustrative example. A food manufacturer uses town water in its products and for cleaning, and it recycles final rinse water through a small treatment unit for use in early rinse stages. A quality review applies the lessons of water engineering.
Independent barriers. The recycled water passes through a filter and an ultraviolet unit, two barriers that both depend on low turbidity. The review adds a chlorine residual to the recycled loop, which works on a different principle and keeps protecting the water in the pipes.
Outcome monitoring. Previously, records showed only that the ultraviolet lamp was on. The business adds weekly microbiological testing of water at the point of use and daily checks of turbidity and chlorine residual, with set actions when results fall outside limits.
Separation. A licensed plumber confirms that the recycled loop is physically separated from the drinking water supply, installs a testable backflow prevention device where the systems meet, and arranges annual testing, as plumbing rules require.
Failure state. The plant uses about 20 kilolitres of town water an hour. To finish a production run and clean down safely during a supply interruption of up to four hours, it installs an 80 kilolitre storage tank with level monitoring and a procedure for supply interruptions.
Result. The business can show, with outcome data, that its water is safe at the point of use, and it has a plan for the day its supply fails.
Applying these lessons in an Australian business
- Map your barriers and check that they fail independently.
- Measure outcomes as well as whether controls are running.
- Order controls so earlier steps protect later ones.
- Design for failure with storage, buffers and passive protection.
- Fix problems at the source before adding downstream treatment.
- Separate clean and contaminated streams, including backflow prevention.
- Use licensed plumbers and follow relevant plumbing and drinking water requirements.
Questions worth considering
- Which of our controls rely on the same mechanism and could fail together?
- How do we know our controls are achieving their purpose, not just running?
- What happens to our operation if water, power or a key supply fails for four hours?
- Where are we treating problems that could be removed at their source?
- Which backup assets would we cut in a good year that we would need in a bad one?
Bringing it together
Clean water became reliable through engineering that separated waste from supply, treated water through independent barriers, used gravity and storage to keep systems safe when things failed, and, in Australia, manufactured water from the sea when rainfall could not be relied on. Its failures, from Poughkeepsie to Walkerton, show what happens when barriers share weaknesses or when controls are judged by whether they run rather than what they achieve. For any operation, the lessons are to use independent barriers, monitor outcomes, sequence controls, design for the failure state and fix problems at the source.
Source: KEVOS editorial notes, drawing on an earlier KEVOS engineering history series on urban water engineering, sanitary engineering and seawater desalination, together with established histories of public health and engineering. Historical figures are widely cited approximations. The worked example is illustrative. This article is general information; water treatment, plumbing and drinking water systems must be designed and maintained by qualified people in line with applicable requirements.
