Final calibration will be the next stage of the modeling process. It is believed that Miami-Dade will then have the largest water distribution model in use anywhere in the world - though it will not hold the record for long. InfoWorks WS is also being used to build an even larger model in Shanghai, China, that already has 400,000 pipes but is not yet fully calibrated. Creating such large models has been made possible through advances that enable the use of commercial database solutions for storage of the vast amounts of geographical data required. Users would be unable to develop such large models without the ability of InfoWorks WS to allow the choice of Oracle or Microsoft Sequel Server in addition to MSDE and Microsoft Access databases in place of the traditional file-based approaches.
The remarkable modeling task has now been completed and the InfoWorks model contains everything from the smallest 1” (25mm) supply pipes to major treatment works serving Miami-Dade County in Florida, USA. The raw water supply is also being added to the model.
Representing something of this scale has meant overcoming a succession of challenges, such as finding out information about pipes laid a century ago. Calibration work and field research is now taking place to fine-tune the model in readiness for its use in a wide range of planning and operational tasks for the Miami-Dade Water & Sewer Department.
Greater depth of information
Modeling has been carried out in great detail. The task initially appeared daunting: the statistics are impressive. The model encompasses 5,600 miles (9,000km) of pipes, ranging from 1' (25mm) to 120' (3m) in diameter. There are 218,000 nodes and 236,000 pipes in the model, as well as facilities including 49 treated-water pumps, 19 storage facilities and 98 raw-water wells.
Miami-Dade’s primary pressure zone is supplied from two major treatment plants, each with its own well system supplying raw water. There are also two other zones - a low-pressure one to the north and a small separate zone in South Dade.
The model, built in InfoWorks WS from Wallingford Software, replaces an older, more basic representation of the system. The County’s original hydraulic model, built using another software system, was limited both in its coverage and its functionality. 'It was only a single step simulation and couldn’t do extended-period modeling,” explains Mr Eerkes. Only the trunk system was modeled, which served well for a planning level analysis, however could not be used for detailed local distribution analysis. “It also had a simplistic modeling of the pumping and storage facilities,” he adds. Only the main pressure zone was modeled. In addition to the smaller pipes that were not included, other exclusions were the raw water network and the well system.
In contrast, the InfoWorks WS model includes every pipe in the entire system and carries out extended-period simulations. All facilities are now represented in detail, including reservoirs and every pump and pump curve. The model is designed to include the raw water systems as well as the interfaces between raw and treated water. All three pressure zones have been fully incorporated.
Earth Tech has now built the base model and preparations are taking place for the next steps including calibration and field testing. The raw-water aspect of the model is under development.
'One of the biggest challenges was that we had not finalized the software selection process before developing the model,” says Mr Eerkes. InfoWorks was chosen by Miami-Dade County, with the help of Earth Tech through a Modeling Software Selection Process. One of the reasons for selecting InfoWorks was its capabilities in terms of extended-period simulation on extremely large models. This was an essential feature for the new Miami-Dade model, as the engineers needed the ability to study how the system operates at different time steps and to explore aspects such as water quality and operations modeling.
The modeling process was divided in two streams - distribution and facilities - before being brought together into a single steady-state base hydraulic model. Aspects such as demand analysis, diurnal patterns and pump control rules were then taken into account to create the base extended-period simulation hydraulic model. 'Because of the short time frame we’ve had to gather as many resources as we could within Earth Tech,” says Mr Eerkes. “It required a lot of coordination between offices to meet the schedule.”
The model draws on a wide range of inputs. Miami-Dade’s initial GIS data proved to be fairly robust, says Mr Eerkes. 'Once we had completed the model, we developed a data-gap report to pinpoint areas in need of further investigation,' he adds. 'But, considering the amount of GIS that we received, these gaps were few and far between.' Field testing will confirm areas of doubt and the results can then be fed back to the GIS, as well as being used in the hydraulic model.
Another key step was to develop a digital elevation model of the system. The previous model had simply given every node an elevation of 5’ (1.5m), whereas in reality they range from 5’ to 20’ (1.5m to 6m). “We had to go in and do some manual checks,' he says.
Further challenges were encountered once the GIS was imported. Anomalies then became apparent. There were instances where there appeared to be an 8' (200mm) pipe in the middle of a section of 30' (760mm) diameter, recalls Mr Eerkes. Some critical information was missing, such as pipe ages and materials, and the status was unknown for some of the major valves. There were also connectivity errors and confusion arising from lack of a unique GIS identification numbering scheme. 'We could pick these issues out in the initial runs of the InfoWorks hydraulic model,' he adds. In particular, pipe age and material were critical to the modeling. 'We then had to go through as-built record drawings and talk to operators.' Useful information came from discussions with people who drew on personal recollections dating back many years.
C-factor graphs were developed for estimating the roughness values for the various sizes and ages of pipe. The curves extend to 100 years, reflecting the age of parts of the network.
A major challenge involves confirmation of the status of valves. Some 800 valves have been identified as closed by the operators, and these will need investigation in the field. Others that might be expected to be closed are shown as open. A group of pipes was shown as totally disconnected from the system; elsewhere the high pressure system appeared to be draining straight into the low pressure system.
The next stream of work involved developing the model for the wells, pumping and storage reservoirs. 'A lot of the locations were identified in the GIS but we had to add a lot more detail into the InfoWorks model itself,' says Mr Eerkes. There were also challenges in modeling the facilities. 'We lacked detailed information - they required an extensive field review,' he adds. In some places, there was nothing more available than a hand-annotated drawing. Elevations tended to be missing for the reservoirs. 'Another problem we had was that different benchmarks had been used over the years, so different drawings would have a different datum for the elevation,' he says.
Taking the next steps
'From the drawings, we developed detailed sketches of how the model was going to be built,' he says. Numbering and naming conventions were devised. 'The goal was to model as much detail as possible.'
An extensive process was carried out to develop the demand model. It involved allocating retail and wholesale demand, developing diurnal demand patterns and estimating unbilled water. The scale of the work was considerable. The billing records showed 445,000 meters that had to be allocated to demand nodes. A high success rate of 95% was achieved in matching up with property addresses. In total, 21 diurnal demand patterns were developed for residential and industrial/commercial/institutional use.
Although the model is so detailed, some limited areas did not require individual pipes to be included as Miami-Dade supplies water to 15 wholesale customers, which in turn carry out the final distribution. However, diurnal information is still needed. Some comes directly from SCADA data and flow logging will be needed for others.
Research has been needed as wholesale patterns vary considerably, depending on factors such as whether there is storage in the system. Some major users have highly variable demand patterns, such as the Port of Miami which fills the tanks of cruise ships.
Making use of the findings
The next step will involve further calibration and field testing, including monitoring of trunk main flows and pressures, as well as hydrant flow monitoring. “One of the challenges we have is that some of the trunk mains are very large in diameter so they operate at relatively low velocity and head loss,” says Mr Eerkes. “Estimating roughness is going to be difficult for those.' Another challenge is at the opposite extreme, with the multitude of very small diameter, older pipes of 1, 2 or 3 inches (25mm, 50mm or 75mm). Furthermore, the status of many valves is unknown.
'Another challenge is defining how the facilities are operated,' he adds. There are no fixed rules on how pumps are operated. 'It may depend on the experience and system knowledge of the operators, however that doesn’t work in a hydraulic model.'
Once complete, the model will be used by the County for a host of purposes, including development approvals as many people choose to retire to the Miami area. “The County wants to use the InfoWorks model to determine what capital works are required in order to allow the projects to be built,' says Mr Eerkes.
It will also be used for evaluating potential storage locations and in pressure management, to reduce leakage. The information from the model enables demand to be broken down, showing the retail (residential, industrial, commercial and institutional) and wholesale supply. “We can then find out what the unbilled volumes are,” he says. The county’s own studies had already provided information about leakage levels as well as usage in hydrants and irrigation.
Another role for InfoWorks is in the monitoring of water quality parameters such as chlorine residuals or THMs. Further plans include using the model to help optimize operations in terms of aspects such as pumping.
'The County also wants to look more closely at its raw water system - nobody has ever modeled it before,' adds Mr Eerkes.
This article is based on a paper presented by Earth Tech senior hydrotechnical engineer Eppo Eerkes at the Wallingford Software International User Conference, 2007.