There are a number of qualities by which an engine can be judged. First, a range of equations must represent accurately the flows of water through the pipes, valves, pumps and other components of the network. Second, the management procedures of the engine must use appropriate timesteps for running the model, which means small timesteps when conditions are changing, and longer timesteps in steady state conditions. Third, the flows of water into the system, through the network, and out of the system through customer demand and leakage must be accurately represented in terms of volumes, mixes, time, and geography. Fourth, the engine must be capable of running models of all sizes, including large models: some clients need to run models of many thousands of nodes. Finally, despite all the above requirements for detail and size that point to a heavy computation load, the engine must also be fast – when modelers are building, calibrating, and then using a completed model, run speed is essential for good productivity and reducing the cost of modeling.
Unlike most other water distribution software products, which base their calculations on the EPANET engine, InfoWorks WS uses its own proprietary engine. Sharing the same computational roots as EPANET, its first release was in 1985 as part of the WESNET product, and over the 20 years since then it has been continually updated. Now known as WS SIM, the engine’s latest update was a very significant enhancement in mid-2005, incorporated in InfoWorks V6.0, that increased the already fast calculation engine by an order of magnitude.
This note outlines some of the advantages of the WS SIM engine and explains why in more complex conditions WS SIM will outperform other engines in all of the considerations outlined above – accuracy of equations, appropriate timesteps, demand modeling, size, and speed.
An example of the basic equations of InfoWorks WS SIM
WS SIM is at the center of the regular updates of InfoWorks WS, of which there are two each year, the outcome of the 30 man-years of development that are invested in enhancements to the InfoWorks suite every year.
One of the results of this is that WS SIM is acknowledged by users as the fastest solution on the market, particularly when models require high accuracy, short time steps, or both. The reason for this is the custom-built Matrix Solver of WS SIM, which is extremely fast compared to the widely used Gradient Algorithm method. As an indication of run times, notoriously difficult to compare between engines, an 80,000-node InfoWorks WS model produced results of a 24-hour simulation, at 15-minute intervals, in 6 minutes. Smaller models are of course faster.
One big difference from other engines is the size of model the engines can handle. WS SIM has a very large capacity of up to 300,000 elements. Because of this capacity, distribution zone and district models do not have to be arbitrarily chopped into pieces, or skeletonized, to fit the limitations of an engine.
For steady state calculations and extended period simulations, WS SIM uses the same equations as most other engines to compute heads at nodes and flow in links based on the boundary conditions (reservoir levels, tank levels, and demands) and system operation (pump and valve positions).
However, there are key differences between WS SIM and other products. It supports true date and time at all levels of the program, including demands and controls. WS SIM will accurately model isolated areas in a network, dry systems, and empty tanks,. This means that WS SIM can be used to closely match actual operating conditions at any specific time, and can easily be used in conjunction with time-tagged monitored data from SCADA or loggers.
WS SIM uses a wide range of industry-standard friction calculations, including the modified Colebrook-White method, which improves accuracy. It also dynamically recalculates friction factors at each timestep, to more closely represent the true behaviour of the water/pipe interface, because friction losses are not constant for all flow velocities.
Finally, WS SIM can automatically calculate friction factors from install date and pipe material, to give consistent values for uncalibrated friction factors.
It is easy to concentrate modeling effort on pipe configurations and fail to reflect the crucial importance of valves in distribution network modeling. Unusually, WS SIM models both valve operations and valve hydraulics, giving a true representation of the impact of the valve on flow and pressure at all times. WS SIM models them as objects with two elements – the valve, and the pipe to which they are connected. This has the benefit of resolving instabilities in water quality modeling.
WS SIM distinguishes between system valves and operation valves. On valve controls, most engines support all standard and logical valve operation types. WS SIM supports these, and additionally supports mathematical operations in controls and incremental valve closing or opening, WS SIM also supports PLCs (Programmable Logical Controllers).
The approach to pumps in WS SIM is very similar to that of valves. They are modeled as pump objects and pump stations. WS SIM can group pumps together as pump stations, or they can be treated separately. Both fixed and variable speed pumps are modeled,.
In modeling of controls, as well as supporting logical operations in controls, WS SIM additionally supports mathematical operations in controls and incremental pump starts, stops, and changes. This means that WS SIM can model real life pump controls, such as a pump that supplies not more than 60% of total system flow.
WS SIM handles reservoirs in the same way as most other engines, but transfer nodes, wells, and tanks are represented differently.
For transfer nodes, WS SIM produces automatic reports on total water produced, consumed, and transferred in and out of the system,.
WS SIM can represent wells as fixed head nodes or can use a drawdown curve, for which an iterative procedure is used to determine well level and pump flow. For tanks, it can model backflow valves, empty tanks, tank overflow, and free surface tank inlet.
The calibration phase of model building depends heavily on flow data, but many engines do not support SCADA or flow meters. WS SIM represents flow meters directly linked to SCADA.
Hydrants are increasingly important in water modeling, and WS SIM models them both accurately and efficiently with a minimum of data elements to represent each hydrant, so keeping the model size and complexity down. In contrast, while other software products represent hydrants, they use five data elements to model every hydrant, which can double or even triple the model size for a typical city.
WS SIM matches the industry standards in the handling of water quality, with identical functionality for constituent tracking, water age, and tank mixing. The only difference is in source tracking. Many products allow only single source tracking, whereas WS SIM offers multiple source tracking.
A key differentiator between engines is the built-in Demand Management Area tools used in WS SIM to modify demands using observed consumption records and flow meter records, allowing grouping by demand type, grouping by system zone, and scaling by a factor or a fixed value.
WS SIM also allows customer level calculations of demand, daily, weekly and monthly demand variations, and the modeling of pressure-related customer demand, and pressure related leakage.
There are a number of other features that WS SIM addresses yet some other engines do not, including Sedimentation calculations, energy cost estimates, critical link assessment, unaccounted-for water estimation and unidirectional flushing.
As the above description of features shows, WS SIM outperforms other engines in terms of the five key elements set out at the beginning – asset modeling, management of timesteps, modeling inflows and outflows, including demand and UFW, model size, and run speed.
With the regular releases of new versions from Wallingford Software, InfoWorks WS and its engine, WS SIM, is likely to continue to remain the best of all simulation engines for water distribution modeling.