Engineering exploration of the water supply system of Constantinople

Abstract

Before this research study began, relatively little was understood of the water supply in Constantinople, particularly within the walls of the city. Archaeological work had focused on collecting details of 160 cisterns and a small number of channels and pipes were incidental finds in other excavations. Although no-one had considered the water supply in Constantinople as a whole, the evidence seemed to indicate a sophisticated water management system. With the available data fragmented, and the potential for more evidence limited to serendipitous finds associated with construction work, the only way to move the understanding of the water supply forward is to take a radically different perspective: civil engineers are well placed to envisage the water supply as a working system and make use of their modern design skills and tools to fill in the gaps between the fragmented data. This reimagining of the water supply system was driven by a key piece of knowledge: the water supply worked, and worked for many centuries. That fact, combined with the fragments of physical and literary evidence, the largely unchanged landscape and the fundamental physical laws governing gravity-fed water systems, are enough to start filling in the information to create a complete system. The core work in reimagining the water supply system has been developing an understanding of the physical infrastructure of the distribution system. Although the two most recent and comprehensive studies appeared to agree that there were about 159 cisterns in the city, close examination of the available data showed that there were actually 209 with the possibility of more. An evaluation of the aqueduct routes in previous studies highlighted inconsistencies with newly available evidence: alternative routes were designed that tied together the available evidence, providing a consistently downhill route, shorter and more straightforward to construct. Having established the number and spread of cisterns and the locations of the aqueducts, it was possible to create a network delivering water from the aqueduct channels to the cisterns for collection by the public. Consideration has also been given to what occurs at either end of this physical infrastructure. At the upstream end, quantifying and characterising the water source defines the water available to distribute and helps to indicate the purpose of the cisterns. At the downstream end, developing even a basic model of water consumption has enabled the distribution network to move from a static artefact to a system with a quantifiable purpose. The combination of the physical infrastructure, inflow data and demand assumptions in an agent-based model demonstrate that the decisions and assumption made within each element work together and allow a fourth element, management, to be considered. The agent-based model of the water supply enables consideration of a dynamic system and the exploration of a number of “what if?” scenarios. This exploration concludes that the cistern-based distribution system probably developed because of fluctuations in inflow. It may have been possible for the city to use a merged arrangement on the Aqueduct of Valens inflow, but the burden of pro-active management required to make it successful suggests that a parallel arrangement is more likely. There was likely to be an interconnection between the two main aqueducts, which would have enabled the use of water stored in the largest open-air cisterns.