WaterSight - Numerical Models

Product(s): WaterSight
Version(s): 10.00.
Area: Documentation

Overview

This administration page is used to upload the hydraulic model and set the options for the automated real time simulation. Sensors measurements will be used as boundary conditions for the hydraulic model runs, contributing to increase the confidence in the model calculated results. The sensors measurements that can be used to override the initial conditions of the hydraulic elements are the following:

  • Tanks: level or hydraulic grade
  • FCV: flow setting
  • PRV, PSV, PBV: pressure or hydraulic grade
  • Pumps: relative speed or pump status

The uploaded WaterGEMS model will automatically run in WaterSight, at a given frequency defined by the user and also considering the hindcast and forecast period defined (more detail below). If the WaterGEMS model contains more than one scenario, only the active scenario will be considered in WaterSigth. 

The command buttons above the table allow to add a model, edit an existent model or delete the model. 

Once a new model is uploaded and the user wants to replace it by a new one, it is necessary to remove the existent model before uploading the new one.

Recommended Model Requirements

In order to successfully run the hydraulic model in WaterSight, the user should first check if the EPS model is successfully running in WaterGEMS. In case there are errors, the user should first correct them using WaterGEMS. 

To take full advantage of WaterSight capabilities, the user should check if the WaterGEMS hydraulic model is updated with all relevant network elements (all linear and vertical assets), and that:

  • customer meter elements are updated and the billing ID field in the model matches the ID provided in the Customer Meter and Billing configuration files (customers will be used in the Pipe Break, Fire Event and Pump Shutdown workflow),
  • all sectorization valves are updated (will be used in the Pipe Break workflow),
  • all SCADA elements are created and mapped to the respective model elements 
  • when using the automatic demand adjustment option of WaterSight (see below for more information) it is required that the zones name in the WaterGEMS hydraulic model are associated to the nodes and match the names given to the zones in WaterSight.

Note: The model that is uploaded in WaterSight needs to have all the SCADA elements created and mapped to the respective model elements, however it is not required that specific model has access permissions to the sensors data sources. To assure that WaterSight is being fed by real time data, it is required that a copy of the WaterGEMS SQLite with access permissions to the sensors data sources is saved inside the On-Premise tools folder. For more information about this, please go here

WaterGEMS Model Setup

If the WaterGEMS model contains more than one scenario, only the active scenario will be considered in WaterSigth. SCADA or other telemetry data will be used to set the initial conditions for the model run when they are correctly configured in the uploaded WaterGEMS model. This means:

  • SCADA elements are created;
  • SCADA elements are configured:
    • associated to the model element;
    • field defined;
    • SCADA signal defined;
  • The signal SCADA tag configured in the model must match the one configured in the sensor administration page of WaterSight
  • The calculation type needs to be "SCADAConnect Simulator" and the simulation model needs to be "Historical";
  • Control overrides configured in WaterGEMS are inherited in WaterSight if set to True in WaterGEMS;

For more information about how to create and configure SCADA elements in WaterGEMS, including how to connect with the data sources, please read this article (read the Setup, Data Sources and Assigning Data parts of the article).

Model Domain

Internal model ID generated by the system once the model is uploaded. 

Model Type 

Model input format. For now it is only possible to upload WaterGEMS models.

Spinup (h)

Please consider zero. Time period at the start of a model run where the results are considered unstable and are not displayed. This is particularly important for model runs where initial inertia is important. For water supply models spinup can be relevant when simulating water quality. 

Hindcast (h)

The Hindcast is a time period of time leading up to the current time where past model calculated results can be compared with historical sensor values. 

Forecast (h)

The time period into the future, starting from the current time, for which model results will be calculated. Results of each model run are automatically saved in the Previous Simulations page.

Run Frequency (h)

Define the automatic run frequency of the model. For example if 1 hour is defined, this means that the model will automatically run every hour. 

EPSG code

EPSG (European Petroleum Survey Group) code of the WaterGEMS model. This is required in order to have the model displayed in the correct position in the map. After defining a EPSG code is not possible to edit anymore. To add a new one, the model domain must be deleted and a new one should be uploaded.

Demand Adjustment

Select the demand adjustment options that will be automatically applied by WaterSight:

No Update: 

This means that  WaterSight will use the same exact demand values defined in the WaterGEMS uploaded model, including all base demands and unit demands, demand multipliers (patterns) and other demand adjustments defined by the user. 

Update Demand & Patterns (Global):

WaterSight will automatically compute a global pattern for the entire system, and will automatically apply this global pattern to all nodes of the network. The total demand for each node is obtained by adding up all base demands and unit demands (unit demands multiplied by number of units) defined in the WaterGEMS model and then by multiplying this final node demand by the global pattern automatically computed by WaterSight. 

The global system pattern is automatically calculated by the software and takes into consideration the following:

  • The global system zone flow corresponds to the sum of all inflows, subtracted by all outflows, and by taking into account also the storage variation inside the system;
  • Last month of historical zone flow data is used to calculate a pattern for the global system zone (dimensional values);
  • Global pattern, including a one week forecast, is calculated using a machine learning algorithm, together with advanced data analytics methods. Bad data, including outliers and spikes are removed from the analysis and a different pattern for week days and weekends is computed; 
  • Global pattern with dimensional values is transformed into a global multiplier (or dimensionless pattern) and applied to each node of the model;

To assure that a global pattern is calculated, all inflows, outflows and storage sensors for the system total zone must be defined (information configured in the Zones configuration page) and transmitting data into WaterSight (make sure sensors are defined in the Sensors configuration and that the On-Premise tool (OFOSC) is working).

The global system dimensionless pattern (global demand multiplier) covers all simulation window period defined by the user in the Numerical model settings (hindcast + forecast). The global dimensionless pattern also differentiates between weekday, Saturday and Sunday. For example if the simulation window covers a Friday and a Saturday, the global demand multiplier for Friday and Saturday will be quite different. 

Another important aspect to mention is that the global pattern is continuously and automatically being updated in real time (whenever new data arrives into WaterSight) using always last month of historical data. This means that the global pattern for today is slightly different than the global pattern of the previous day and will be slightly different from tomorrow global pattern. 

Update Demand & Patterns (Zone-by-Zone)

WaterSight will automatically compute a pattern for each zone (according to each zone defined in the Zones configuration page), and will automatically apply each zone pattern to the nodes belonging to the respective zone. For this to happen it is necessary that the zones name in the WaterGEMS hydraulic model (and associated to the nodes) match the names given to the zones in WaterSight (zone names are configured in WaterSight by changing the zones information contained in the shapefile that is uploaded in the GIS configuration page). For the model nodes with no zones assigned or with zones different that those defined in WaterSight, patterns and demand adjustments will be ignored (only base demand and units demands will be taken into account).

The total demand for each node is automatically obtained by adding up all base demands and unit demands (unit demands multiplied by number of units) defined in the WaterGEMS model and then by multiplying this node final demand by the respective zone pattern. 

The zone pattern is automatically calculated by the software following the same methodology used for global system pattern explained above:

  • Each zone flow corresponds to the sum of all inflows, subtracted by all outflows, and by taking into account also the storage variation inside the zone;
  • Last month of historical zone flow data is used to calculate a pattern for each zone (dimensional values);
  • Each zone pattern, including one week forecast, is calculated using a machine learning algorithm, together with advanced data analytics methods. Bad data, including outliers and spikes are removed from the analysis and a different pattern for week days and weekends is computed;
  • Each zone pattern with dimensional values is transformed into a zone multiplier (or dimensionless pattern) and applied to all the nodes that belong to the respective zone;

The zone dimensionless pattern (zone demand multiplier) covers all simulation window period defined by the user in the Numerical model settings (hindcast + forecast). The zone dimensionless pattern also differentiates between weekday, Saturday and Sunday. For example if the simulation window covers a Friday and a Saturday, the zone demand multiplier for Friday and Saturday will be quite different. 

Another important aspect to mention is that zones pattern are continuously and automatically being updated in real time (whenever new data arrives into WaterSight) using always last month of historical data. This means that the zone pattern for today is slightly different than the zone pattern of the previous day and will be slightly different from tomorrow pattern. 

To assure that a zone pattern is calculated, all inflows, outflows and storage sensors must be defined (information configured in the Zones configuration page) and transmitting data into WaterSight (make sure sensors are defined in the Sensors configuration and that the On-Premise tools are correctly working - SCADA pusher and On-Site Coordinator). Besides this, to assure that a reliable pattern is computed, all zone boundary valves should be closed and zones should be as stable as possible throughout the year (frequent change of zones boundaries affect pattern reliability).  

The user can analyze each zone dimensional pattern automatically calculated by WaterSight by acceding to the Zone details graph (under Network Monitoring page). Once in the zone detail graph, the user should change the interval to "15 min" and the pattern dropdown option to "Last month"  and enable the legend by clicking on the existent button on the top right of the graph.

Then disable on the graph all pattern confidence bands (5/95 band and 20/80 band) and the real time series, and keep only the median series (P50 Average). This is the dimensional pattern that will be automatically transformed into a dimensionless pattern (zone demand multiplier) used to multiply all existent node demands of the hydraulic model, in case this option is enabled in the Numerical Model settings. The user also has the option to export the zone dimensional patterns to CSV by clicking in the Download button  located on the top right of the graph 

In case the user defined patterns in the original WaterGEMS uploaded model are not very accurate (for example the patterns are not updated or were developed based on reference values, etc) or if no patterns at all were defined in the original WaterGEMS uploaded model, then the automatic adjustment options available in WaterSight can significantly improve the accuracy of the model. 

In the example below, tank level results (modeled VS SCADA) are compared using the two different demand adjustment options available: Global Demand Adjustment and Zone Demand Adjustment. 

Global Demand Adjustment

Zone Demand Adjustment

Figure 1. An example where tank level results (modeled vs SCADA) were compared using different automatic demand adjustment options available in WaterSight.

Active

Turn the model active or inactive. If inactive, the hydraulic model will not appear in WaterSight and the Modeling section will not be available.  

Model File

Click here to directly upload a WaterGEMS model - must be uploaded only the sqlite file.  

Manage Style Configuration

Manage the color coding configuration for the model. More information here.


Recommended
Related