Engineering Rome

Aqueducts – equity and public health perspective



The ability to provide clean and drinkable water is essential for the growth of a city in both the modern age and antiquity. No other civilization was better at this than the Romans. During the height of Roman power 11 aqueducts fed over a million cubic meters of water a day to the ever growing city of Rome. This water was vital in the Roman economy, entertainment, and health. This paper will focus on the processes that provided water to Rome. Starting with determing the source and ending with the disposal of waste into the Tiber River.

Water Sources

It all begins with locating a viable source of water. Finding a clean source of water wasn’t always easy. Often times water came from springs, streams and lakes. If this was the case all a roman engineer had to do was determine the quality of water. This was done through a variety of ways. According to De architectura, written by Vitruvius in 30 B.C, the engineer was to inspect the clarity, taste and flow of the water. Engineers were also told to inspect the locals who consumed the water from a potential source. If the locals looked frail and weak they knew the water was of poor quality; however, if the locals looked healthy and strong,there was a higher chance the water was pure and clean. The soil and rock composition also gave clues to the water quality. Regarding the impact of soil and rock on water Vitrivus said “Clay soil is generally a poor source, since the water will be scarce and unsavory, whereas water found around red tufa will tend to be copious and pure”. (Aicher, 1995). This was not always the case though. Springs are often underground and difficult to locate. A famous example of this problem is depicted in the Trevi Fountain, which shows a virgin girl showing clueless military engineers the location of underground springs that would become the source of the Aqua Virgo. Vitruvius suggests examining soil type, vegetation, and the geography. Certain types of plants such as willows and rushes are a good indication of an underground stream. Table 1 shows the various sources of water for the aqueducts, it shows that the Romans used a wide varity of sources with ranging qualities.

Trevi Fountain.jpg
Figure 1. The Trevi Fountain depicting the source of the Aqua Virgo.

Table 1: Aqueduct Water Sourcesa (Kosonen, 2017)
Water sources.JPG

The aqueducts relied upon gravity to deliver millions of gallons of water a day to Rome, because of this a constant slope was necessary to facilitate a constant flow. According to Vitruvius the minimum grade should be 0.5% (5 meters vertically over 1000 meters horizontally); however, this was not always the case and the average lies between 0.15% to 0.3% (Aicher, 1995). The steeper the grade the faster water will flow, and with a higher velocity comes problems such as erosion of the channels and lower stability around the curves of the aqueduct. Roman engineers had several tools to ensure the proper gradient was being made. Vitruvius calls the chorobate, drawn in Figure 2, the most accurate, a 6 meter horizontal beam supported by legs and cross braces. The beam had a water trough that when filled acted as a level. If there was no wind a plumb line could be used with markings on the legs/braces.

Figure 2: Drawing of Chorobate

Water Treatment

The quality of water depended upon the source. Sediments carried in the current from rivers and lakes was a problem dealt with by piscina limaria ,settling tanks in English. These settling tanks worked roughly the same way as they do today. By slowing the velocity of water, large enough particles would settle out. There were no locations specifically for settling tanks and they have been located at the source, part way through the aqueduct, or at the terminus of the aqueduct. There were several different types of settling takes, the simplest were just a widening of the aqueduct, thus increasing the cross-sectional area of the flow which would lower the velocity assuming constant flow. There were more substantial tanks such as the one pictured below in Figure 3 which was added at the end of the Aqua Virgo that forced the flow through several chambers and 2 levels.

Settling tank.JPG
Figure 3: Settling tank of the Aqua Virgo

Taking water directly from a river created problems that settling chambers and grates couldn’t fix. For example, Anio River Aqueducts consistently delivered sediment rich water. This was addressed when Trajan changed the source of the Anio Novus upstream to a dam above Subiaco. This area had a pristine watershed and the large artificial lake behind Nero’s dam acted as a large settling tank (Aldrete, 2008). I believe that aqueducts that were sourced from rivers provided cloudy water due to the silty sediment layer on the bottom of the rivers. Due to erosion, rocks and fine sediments are deposited at the bottom of rivers versus natural springs that tend to be mostly rock lined. Because of the low density of the silt and sand and the naturally high velocity of the river, the sediments are carried into the aqueducts where larger debris like rocks can settle out.

Settling Calculations
When I read that the Anio River Aqueducts carried sediment rich water to Rome I was curious if it was possible to remove these sediments by settling, the only method the Romans had. To do this I took particles ranging in size from large sand to silt, and temperatures from 0 degrees Celsius to 20 degrees Celsius to account for the changing viscosity of water due to temperature. We can calculate the settling velocity using an equation introduced in my Environmental Engineering class shown in Figure 5. Using that velocity the required cross sectional area for a certain size particle to settle in the given flow of 190,000 cubic meters a day for the Anio Novus (Aldrete, 2008) can then be calculated. For 10 degree water the settling velocity for sand is 10.05 meters per second, for silt it is 0.0000112 meters per second. If we assume a depth of 1 meter (and cross sectional area of 1 meter squared). It would take 0.0995 seconds for large sand to settle. For silt it would take 89285 seconds or 1.033 days to settle. For a velocity of 190,000 meters a day (assuming area is 1 square meter. This number is an observation from inside the aqua Claudia) it would take 0.46 days to travel the length of the aqueduct (87 kilometers). Thus the aqueduct would have to be roughly twice as long to remove silt particles through settling. This is most likely the main cause for the poor quality of water coming from the Anio Novus.

Picutred below, Figure 4 shows the relationship between Settling Area and Particle Diameter. Further, these results suggest that any settling tanks that the Romans would have built were ineffective in removing useful amounts of sediments. Larger particles like sand would have settled naturally along the path of the aqueduct without the need of a settling tank; however, particles like silt need several hundred thousand square meters to settle, making all but the largest settling tanks useless.

Below, Table 2 shows the calculations for determining Settling Velocity varied with particle size and water temperature.

Table 2: Calculations for Settling Velocity

settling vel.JPG
Figure 4: Graph of Particle diameter vs Settling Area

Figure 5: Settling Velocity and Flow equations.

Another factor that helped maintain the cleanliness of Rome’s water was aeration provided from traveling through aqueducts and fountains. Related to modern techniques called “cascade aeration” this helped Rome’s water supply precipitate minerals, taste better, and remove odors. It was unlikely that they knew of the benefits aeration provided and was most likely a unintentional product of the practice of unfilled channels being used for transporting water.


After being collected and possibly travelling through a settling tank the water entered the specus, the channel.. When people think of Roman Aqueducts today they think of towering arches traveling through the countryside like the Pont du Gard in France; however, aqueducts like that were few are far between. Over 80% of Roman aqueducts were built below ground (Aldrete,2008). Above ground aqueducts were used sparingly to cross low valleys and approach Rome such as the Aqua Claudia and Anio Novus. Subterranean aqueducts had several advantages compared to surface structures. Tunnels used less material than arches and less skill. It was much cheaper to have slaves dig a hole in the ground than to pay someone to design a arch and build it. An underground channel would either be excavated with a tunnel or a trench from the surface and then covered. A tunnel would be used when they couldn’t go around a mountain or if the source of a spring was too deep for a trench which was the case for the Appia and Virgo. Trenches could be used when the channel followed a river bed or along the contour line of a hill instead of through hills. The size of channels on average were 1 meter wide by 2 meters tall. This was to make it easier for those excavating the channel and for future maintenance. Every 50 meters the tunnels were connected to the surface with a vertical shaft called a puteus. This shafts had several purposes. First they allowed construction to progress simultaneously at several points compared to just the two opposing rock faces at ends of the tunnel. Each shaft would tunnel down then tunnel horizontally until they met up with the group from the other vertical shaft. Shafts could also be used to drop a plumb line down to check the gradient. After construction these shafts were used for maintenance in the channels. This was necessary due to the high amount of calcium carbonate in the water which caused lime to precipitate onto the sides of the aqueducts as shown in Figure 6. The water was high in calcium carbonate because it trickles through limestone underneath Mt. Autore. These deposits could impact flow if left unchecked; therefore, periodically crews would come and chip off the deposits and remove them through the vertical shafts. The removed minerals were left by the openings of the shafts which helped archaeologists locate many of the underground aqueducts. If the channel was excavated through solid rock no further construction was needed. If the channel passed through areas with gravel or clay, a floor, walls,and ceiling had to made. When stone blocks were used the joints were sealed with cement, concrete walls were common too. To reduce water loss a fine mortar was applied to the walls. There were several methods of constructing above ground aqueducts which was needed whenever the path of water passed through a valley.
A valley would force engineers to create elevated channels. These could be supported by a wall (substructio) or arcades. Walls were primarily used leading up to arcades to a height of 2 meters. Any channel taller than 2 meters arcades were used. Arcades had several advantages compared to a wall. Less material was used and arcades disrupted drainage and traffic much less than a wall. The term Aqueduct bridge has been given to a series of arches that carries the channel across a valley or ravine for a short distance.

Figure 6: Underneath the Aqua Claudia, the discoloration on the sides show the build up of Calcium Carbonate.


When a valley was over 50 meters below the channel level an aqueduct bridge wouldn’t be used due to the size of the required arcade. Instead Siphons were used to convey water, through pipes, to nearly the same elevation on the other side of the valley. In the siphons, water from the gravity fed channels would fill a basin and then be converted into an airtight pressurized system. Through a series of lead pipes the water would be carried to the valley floor and then be forced up to nearly the same elevation on the other side of the valley by the pressure of the water. The water can’t reach the same elevation due to friction losses in the lead pipes. Siphons were made of lead and generally ran in series to accommodate the large flow of the Roman aqueducts. The main issue with siphons wasn’t a lack of technical understanding, rather an economical issue. Siphons were prohibitively expensive. Large quantities of lead were needed to construct siphons; however, lead was normally not located near the locations of aqueducts like stone was. Instead they had to bring lead to the aqueducts at great expense. Soldering the thousands of pipes in some siphons would have been time consuming too. Another factor that increased the cost was calcium deposits inside the pipes. In a channel you could wait 20 years until the accumulation led to reduced flow, pipes on the other hand would have needed much more maintenance because they were so much smaller.

Water Losses In Aqueducts

Bribing of those who built the aqueducts and theft resulted in large amounts of water being lost before it ever reached the city. The Aqua Claudia reportedly had a flow of 185,000 cubic meters of water a day. Below I will use Bernoulli’s principle and open channel flow equations to calculate the actual flow if there had been no theft. If we assume Vitruvius’s grade of 0.5% to be an accurate representation of the Claudia’s slope we can use the following equations and assumptions.

Figure 7: Bernoulli’s Equation

Figure 7 shows Bernoulli’s equation written in terms of head. This equation can be further simplified by assuming that Pressure at points 0 and 1 are zero because they are exposed to the atmosphere. The velocity terms can also be negated if a constant velocity is assumed, leaving Figure 8,

Figure 8: Simplified Bernoulli’s Equation


head loss.JPG
Figure 9: Head Loss

Figure 9 means that the head loss due to friction is equal to the change in elevation of the Aqueduct. Figure 10 shows that in open channel flow head loss due to friction equals

head loss 2.JPG
Figure 10: Open Channel Flow equation for Head Loss

Where f is the friction coefficient, equation provided in Figure 11, L is the length of the channel, and RH is the Hydraulic radius (Area divided by wetted perimeter).

Figure 11: Friction Coefficient Equation

N = 0.012 for finished concrete [3]
RH= ⅓ due to rough observations inside the Aqua Claudia
roughness 2.JPG

Since we know the length and the slope we can find the change in elevation which ends up being 345 meters. With this and f we can solve for the average velocity and then convert to Flow. This ends up with a flow of 244,757 cubic meters a day. This number is a rough calculation and is subject to error. The largest error comes from the manning’s roughness coefficient and the slope. These two factors can easily change the end result; however, this value calculated suggests that the flow one would expect after accounting for losses due to friction is significantly higher than the flow that was reported to be delivered to Rome. This number suggests roughly 60,000 cubic meters of water a day was diverted for other uses.
After the aqueducts, water entered castellums, large tanks meant to hold and distribute water throughout the city. Frontinus and Vitruvius are the only sources on how the distribution system in Rome worked, a model of the Roman Distribution system is shown in Figure 12. A main Castellum would be located at the highest point in an area to ensure delivery of water was possible, the main castellum would divide the water and send it towards secondary castellums. The secondary castellums further divided the water into pipes that would send the water to public fountains and private users. Vitruvius provides a simple example of how a castellum worked. His example divided the water into three compartments, one for private use, bathes, and public fountains. This design gave priority to public fountains by placing the chamber for public fountains in the middle. This design made it so any extra water from the other chambers would overflow into the compartment for the public ensuring the public had a constant flow of water. In Rome the system was much more complicated due to the number of aqueducts and the differing quality of water. For example Trajan decided to use the water of the Anio Vetus for irrigation, industry, and the flushing of the sewers while reserving the cleaner water of the Marcia and Virgo for drinking water. The Roman distribution system was also made more complicated by the piecemeal development over the centuries. Once reaching the Castellum the gravity dependent system could be placed under pressure in pipes and be used to deliver water to locations at roughly the same elevation as the pipes origin.

Figure 12: Overview of Roman Water Distribution


Pipes were connected to Castellums with a calix. A calix is a nozzle that fit against the inside of the castellums. It was cast in bronze so people couldn’t take advantage of the malleability of lead and create a wider opening to the pipe and thus increase the amount of water they received. Calix came in several official sizes, the smallest called a Quinaria with a diameter of 2.3 cm. Pipes were generally made from baked clay or lead; although, other materials were used too such as oak in england and germany due to high supply (Aldrete, 2008). Vitruvius warned against the use of lead in drinking water pipes due to the taste and health concerns. He warns of the health risks due to workers getting sick after casting lead pipes.. Even with his concerns lead was used frequently in the creation of pipes. Though lead was used the ill effects were limited for several reasons. First of all, the Roman water system was designed on constant flow; therefor, water seldom had the chance to become stagnant and gather lead in harmful concentrations. Secondly, due to the high concentration of minerals in Rome’s water the pipes quickly became coated with calcium carbonate, further reducing the risk of the water’s contact with lead.

Since calcium built up on the lead pipes quickly there was less of a chance of lead poisoning. Besides this there were pros and cons to the buildup of calcium in pipes. Calcium is is mineral that is good for bones and teeth and is regularly added to water; however, this build up led to regular maintenance for pipes. Lead pipes came in several different sizes ranging from 13 millimeters to 574 millimeters. A safe estimate of calcium buildup is 1 millimeter a year on all surfaces. If a pipe was 574 millimeters wide it would take 287 years for the calcium to build up and block flow. But the smallest pipes would only take 6.5 years.


Once the water left the aqueducts and entered the city it had a wide range of utilization. For the rich water could be piped directly to their homes, the poor on the other hand had to get their water from a public fountain. At the height of power Rome had 1500 fountains and 900 baths. Of the 900 baths there were 11 Thermae, large public Baths founded by the Emperors, and 856 balnea,smaller private or public baths. Baths were an important part of Roman Culture and socializing. One didn’t simply clean themselves at a bath in ancient Rome. Baths were a place to exercise, relax, learn,socialize, eat, and conduct business among countless other things. One of the most important aspects of the public baths was that they were free or only charged a token fee. This meant that anyone, the rich, the poor, could visit a bath. Balnea were often the social centers of a neighbor hood while the Thermae offered a higher variety of activities. Both types of baths would include a similar layout for bathing. First there was a dressing room were Romans would have undressed and stashed their belongings. Niches in walls have been found that suggest there may have been lockers too. Crime was rampant and theft of belongings was a problem. There was a bath attendant to watch over possessions but some brought slaves to safeguard their belongings. Baths always included a tepidarium, caldarium, and frigidarium, warm, cold, and hot water. Baths were heated by a system of furnaces that directed heated water to tanks underneath rooms. The water would heat the air underneath the floors and between the walls for a room like the caldarium. Rooms meant to be hot were constructed with a double floor. In-between the floors were columns of tiles. Hot air was forced into the space between the two floors, shown in Figure 13, thus heating the room, this is known as a hypocaust system.

Ostia Ant.jpg
Figure 13: Air Vents in Baths of Ostia Antica


In modern times drainage and sewer systems are often separated, although in seattle there are many combined sewers throughout the city. Over the centuries the Romans created a system of drains connected to natural waterways that they eventually converted into underground canals.The first Roman drainage system was designed to deal with excessive water. Situated on the banks of the tiber with many natural streams, the low valley areas between the hills of Rome were often transformed into marshes (Aicher, 1995). Because rain wasn’t a reliable enough system to wash away the waste of a million people the Roman’s relied on water from the aqueducts to wash away most of their waste into the Tiber. Some water came straight from the castella as overflow. This water came from the highest opening in the castellum to insure that only extra water was used for this purpose. Most of the water used to flush away waste had already cycled through the distribution system such as baths and public fountains before reaching the sewers, very similar to modern gray water systems. An example is large latrines used at public baths, there was enough water there to guarantee a constant stream to flush away the waste. Few houses had a direct connection to the sewers and there were few public latrines, because of this most waste was dumped directly on the streets. Direct lines were not common due to the stench associated with the sewers and the danger of gas build ups which could cause explosions.

Rome’s first drain was also its most famous. The Cloaca Maxima, the great sewer,began as an open ditch and was later converted into a underground system. The earliest version was built by the kings of Rome in the 6th century BC. It originally ran through the Forum, went through the Velabrum between the Capitoline and Palatine hills and then into the river Tiber. When fully constructed the sewers were made of concrete and stone with portions of the Cloaca being 4 meters tall and 3 meters wide. Figure 14 shows the path of the Cloaca Maxima. If you follow the Link provided below you can view different water systems and structures throughout Roman history.

Clocia max path.JPG
Figure 14: Path of Cloaca Maxima

GIS layers of Rome

The effect of Cloaca Maxima on BOD and DO

The Romans generated roughly 100,000 pounds of excrement a day (Aldrete, 2008), which was washed into the Cloaca Maxima and emptied into the Tiber river. That amount of untreated waste could have serious consequences for water quality and local wildlife. I was interested in determining the Biological Oxygen Demand (BOD) and Dissolved Oxygen (DO) along several points of the tiber, where the waste was dumped, the minimum dissolved oxygen location, and Ostia Antica. I was curious if Rome, roughly 40 km upstream of Ostia, had a major impact on the water quality of Ostia, and if so, how much.
To start off I needed to find Flow, BOD and DO concentrations, Saturated DO, and Kd and Kr (deoxygenation and reaeration coefficients), for the Tiber River and Cloaca Maxima. To do this I used lecture slides from my Environmental Engineering Class, several graphs from the European Environmental Agency, and the book Daily Life in the Roman City. The values I found are shown in Figure 15 and are rough estimates.

BOD conditions.JPG
Figure 15: Estimates of water conditions during the Roman Empire.

Qsewer was found by summing the flows of all the aqueducts into the city.
BODsewer was found by dividing the 100,000 pounds of excrement by Qsewer and converting to milligram per liter.

European Environmental Agency DO

The first step shown in Figure 16, is calculating the initial dissolved oxygen at the point of mixing between the river and the waste. To do this a mixing equation is used

Figure 16: Mixing Equation

Next the Initial Oxygen deficit is calculated in Figure 17

DO deficit.JPG
Figure 17: Calculation of Initial Oxygen deficit

Next you calculate the initial BOD concentration, seen in Figure 18, using the a mixing equation like initial DO

Figure 18: Calculation for Initial BOD Concentration

You then calculate the critical time and distance depicted in Figure 19 (to minimum Dissolved Oxygen)

time critical.JPG
Figure 19: Critical Time and Distance to Minimum DO

Next find the minimum Oxygen Concentration in Figure 20

Critical D.JPG
Figure 20: Calculations for Minimum Dissolved Oxygen

I then wanted to find the dissolved oxygen and biological oxygen demand at Ostia Antica. DO is found the same way as above with a new time while BOD uses a new equation. The results are shown in Figure 21

bod ostia.JPG
Figure 21: Biological Oxygen Demand at Ostia Antica due to Waste Water from Rome

Table 3: Minimum Dissolved Oxygen Levels for Fish (Korshin, 2017)
Water Quality.JPG

The calculations show that the minimum dissolved oxygen concentration is 81 km from the cloaca maxima, meaning DO levels are still decreasing as it reaches Ostia Antica. At Ostia the dissolved oxygen concentration is 6.009 milligrams Oxygen per liter of water. According to the table above (Greogry Korshin, environmental lecture) the lowest level of DO that can safely sustain life in a freshwater environment is 6.5 mg per liter. Therefore most of the wildlife in the Tiber would have died near Ostia. The BOD levels make the water unsafe to drink too. This highlights the inequality of water resources not only in Rome but in the whole Empire. These results suggest that at the height of Roman Power the river would have been severely degraded. I have looked for quotes that illustrate that the waters around Ostia Antica were of poor quality and have not found any. This is not suprising. The results are based upon many assuptions and simplified equations from an introductory Environmental Engineering course, which could lead to errors in the results. The assumptions that would create the most error are those I took from the European Environmental Agency, specifically the Dissolved Oxygen content of the Tiber river. There was not a value for the Tiber River today, just average values for rivers in Europe depending on size. These values are also average values for the present, not 2000 years ago. Dissolved Oxygen concentrations vary tremendously based on temperature and concentration of organic matter. If the Tiber River’s initial Dissolved Oxygen was 10 milligrams per liter instead of 8 like I used, the minimum Dissolved Oxygen would be 8.28 milligrams per liter and the Dissolved Oxygen at Ostia would be 8.79 milligrams per liter, which is still low for some fish, but wouldn’t lead to vast fish deaths.

Being able to Study abroad is a privilage I will always be thankful for. This trip was an amazing experince that allowed me to combine my love of food and history in a relevent setting to my education. I loved the sites we saw, the food we ate, and the people I met. Pictured below is our group with the Roma Sotterranea on our way to touring the Aqua Marcia, and Aqua Claudia, my favorite trip of the program. Thank you Steve for being a wonderful Professor and giving us all the opportunity to experience this.



  1. ^ [1] Aicher, P. J. (1995). Guide to the aqueducts of ancient Rome. Wauconda, Ill: Bolchazy-Carducci [2] Aldrete, G. S. (2008). Daily life in the Roman city: Rome, Pompeii and Ostia. Norman: Univ. of Oklahoma Press.

    [3] Korshin, G. (2017, May). Water Quality: Dissolved Oxygen. Streeter-Phelps Model. Lecture.

    [4] Kosonen, H. (2017, August). Water in Rome. Lecture.

    [5] Manning’s Roughness Coefficients. (n.d.). Retrieved September 15, 2017, from

    [6] Total oxygen in river stations by river size. (n.d.). Retrieved September 16, 2017, from

Follow us

Don't be shy, get in touch. We love meeting interesting people and making new friends.