Specialist water and wastewater services provider Hunter Water Corporation purchased InfoWorks WS in 2009 to enable advanced real-time modeling of its water networks. Its experience of building complex models from basics has been extremely positive, with the process being made considerably easier than anticipated by the solution’s powerful data import, inferencing and operational tools.
Hunter Water’s Lower Hunter Valley base in New South Wales, around 170km north-west of Sydney, is the sixth largest urban area in Australia. The state-owned corporation is the second biggest water utility in New South Wales after Sydney Water itself.
There are six local government areas (LGAs) within the region: Newcastle, Lake Macquarie, Cessnock, Maitland, Dungog and Port Stephens. The corporation has around 500,000 customers, 220,000 connections to its water networks, and is responsible for around 80 reservoirs, a similar number of pump stations and 4500km of water mains.
The utility began modeling in the late 1990’s with Pipes ++. Although this proved satisfactory it lacked some key functionality, probably the most critical of which was the absence of a GIS framework.
As a result, Hunter Water began extensive research into the available solutions, which resulted in the purchase of eight licences for MWH Soft’s InfoWorks WS. The main deciding factors were the advanced multi-user environment, the easy audit trail and simple version control.
The utility had previously experienced problems in determining which version of a model was the most current. InfoWorks WS’ database structure, which identifies the latest version of a model, was found to be an ideal solution to this issue.
One key consideration for the utility was whether to convert its 15 existing Pipes ++ models into InfoWorks WS. It was quickly established that this would not be appropriate because of the original software’s lack of GIS functionality. It was therefore decided that the utility should start afresh with its new solution.
The new modeling exercise
To begin this task, the utility established a team of three external consultants and an internal project manager. Two of the external consultants were from Urban Water Solutions and the third from Hunter Water Australia, which is the consulting arm of the Hunter Water Corporation. The utility wanted to establish a more collaborative approach than is traditional, so that the team could work effectively to determine the scope of the project, the delivery plan and project implementation.
The approach has proved very successful to date, with the team achieving its target of completing model build before Christmas 2009. The project began in August, with the first two weeks devoted to determining how best to import data into InfoWorks. This process was followed by the model build, which the team originally estimated would take a month. This process proved considerably quicker than anticipated - by the third week the modelers had already begun to import and clean up the data.
As a first step the team imported the background layers using the layer control dialogue box, including the asset data and cadastral aerial images. This task was undertaken via the utility’s GeoMedia-based GIS, with all of the data exported into InfoWorks as Shape files. Not all of the data owned by the utility was held within the GIS network, so a considerable amount of additional information had to be input from a number of different locations. The modelers found that inputting pump curves, reservoir parameters and other asset-related information into InfoWorks WS via the Data Import Center was a simple process.
Hunter Water had also recently acquired a number of Digital Terrain Models (DTMs), which were imported into the model to enable the elevation of the elements within the model to be set and a 3D perspective of the overall area to be obtained. The team then imported demand information and customer data held within the utility’s Customer Information System (CIS), which is part of the GIS network. InfoWorks WS’s Demand Allocation Wizard was used to smooth this process.
The team imported all of the data into one large network to ensure a consistent approach. Six models were then separated out from this parent model (one for each LGA) so that work could begin on each individual network area.
Model verification and controls
Once the task of importing the data was completed, the next challenge was to clean up the model data, resolving issues such as missing or incorrect connectivity. This can be a significant challenge, but the tools within InfoWorks (such as the Connectivity Trace tool, Boundary Trace tool, Proximity Trace tool and data flagging) considerably simplified the process.
The final step in the model build process was the addition of controls. Originally the team hoped to undertake this task as an automated process, but this proved challenging so the team manually input controls according to the system used in the utility’s SCADA.
Once the models were running, it was necessary to check them against real data, so a verification process was undertaken. There are various levels of verification that can be adopted, and Hunter Water opted for a full calibration using live data from its existing SCADA telemetry. The next stage will be to undertake further field testing.
User Programmable Controls
The modelers had to decide whether to use User Programmable Controls (UPCs) or conventional controls within the models. InfoWorks WS offers two ways of adding controls: the first is the conventional method in which the modeler selects a preset control arrangement and then inputs the various control parameters; the second is via UPCs, in which the modeler writes a script that determines the operation of the pumps and valves.
It was necessary to decide whether to use a UPC or conventional control for each pump or valve being modeled. Both types of control have advantages and disadvantages. UPCs hold all of the control data within one tab, whereas the conventional method requires one tab to enter the type of control and another to enter the data.
UPCs can also be used to control multiple pumps with one script, whereas the conventional method requires a control for each pump. UPCs also allow use of the reservoir fill percentage rather than simply reservoir depth as the basis for triggering various actions. This proved extremely useful for Hunter Water because its SCADA system utilised fill percentages, so this was a system the operators understood well and there was no need to transpose the data into a different format for comprehensibility.
The biggest advantage of using UPCs is that they allow a high level of control complexity. For instance, the inclusion of multiple time zones allowed pumps to have a peak period, an off-peak period and even a shoulder peak period. UPCs also proved able to cope with other complex operations such as variable speed drives that were activated at a preset pressure level and deactivated at a preset pump speed.
The biggest disadvantage of the UPC is that scripts can be quite complex and particularly daunting for novice users. The sample script (Figure 1) appears to be long and complex, but much of this consists of comment lines that actually provide an excellent description of the UPC. This means that although UPCs are initially complex and may be challenging during the initial model build, once the model is built and people are familiar with the controls they are reasonably simple to operate and maintain. All that is required is to change a few variables as needed to match any changes in the operation of the real system. Hunter Water adopted UPCs for every pump and valve in its system, which adds to the models’ ease of use.
InfoWorks allows access to live data from telemetry or dataloggers, which can be imported into the model. Hunter Water has telemetry at the majority of its reservoirs and pump stations as well as a number of strategically-located flow meters and pressure loggers. Ultimately, importing SCADA data into InfoWorks proved to be reasonably simple, and around 450 telemetry points were brought into the models including reservoir fill percentage data, pump states, WPS suction and delivery pressures, flow meters and AIV states.
InfoWorks accepts a number of formats, so Hunter Water has chosen the Time Series Data (TSD), a format designed specifically to handle SCADA data, because it allows irregular polling of the sites. This format was provided to the utility’s existing SCADA programming contractor, who took less than a day to provide an output for all flow meters, reservoirs, pump states, PRV and pressure meter states that could be read directly into InfoWorks. This data was then programmed into the SCADA system and every night at around 2am the previous day’s data is uploaded onto the system so that the live data is available immediately.
The modelers are also able to directly compare the predicted and actual data from a file created midway through the current day. From this, the initial water levels for the day can be set to enable modeling to be undertaken. This has proved very helpful operationally.
The process of using live data proved far easier than anticipated, and a number of advantages were found. The first was that using live data was a considerable aid to model verification, allowing predicted and real data to be quickly and easily compared. Retaining all data within InfoWorks WS has proved much easier than exporting large quantities of data from SCADA to Excel files in order to make comparisons. The utility anticipates that this will have a considerable benefit in day-to-day modeling operations: for example, when assessing water main breaks the model could be set to match the exact conditions, enabling further analysis to be easily undertaken.
There have also been a few challenges in the process, the first being in communicating the requirements to the SCADA programmers. Although the data was returned with commendable speed, a certain amount of communication was necessary to ensure that the output was exactly as required.
There have also been issues regarding SCADA point description. On occasions flow meter data was returned with very little description, which meant it was not possible to determine the location of the meter. The modelers also experienced filename structure issues – for instance, if the asset was given too long a filename, it was found to affect the derived values. Modeling pump states against live data also provided challenges although these have been overcome.
InfoWorks WS has proved a good purchase for Hunter Water to date. The model build project ran smoothly. The live data is working well and has proved very useful, and the company is looking forward to using the solution’s more advanced features.
This article is based on a presentation by Anthony Gentle, project manager for Hunter Water, to the MWH Soft Australian User Conference.