Browser Versions Supported
You could try using the following links to download a newer browser version:
* TESLAWeb will work with Internet Explorer 8, but you may see performance issues when rendering the data. We suggest only displaying a day or two of information at a time.
Browser Features Supported
TESLAWeb supports 'zooming' in your browser. I.e. if you'd like to see more at once, Hold down Ctrl and scroll your mouse wheel down (or press the '-' key) to zoom out. Scroll your mouse wheel back up (or press '+') to zoom back in. This will shrink up font sizes and scale the web page accordingly.
On tablets, you can scroll around the spreadsheet simply by dragging your finger around on the spreadsheet itself. The graph display also supports pinch to zoom.
There are several ways to explore your forecast by selecting the different tabs in the main display. Depending on what model view you have selected (see below), certain tabs may be hidden. I.e. when you choose 'Customer Numbers' view, you won't see the 'Graph' tab.
Spreadsheet - Shows a spreadsheet view of the data. Highlighting of min and max values can be toggled in the Settings section.
Graph - Shows a graph/chart, allowing you to graphically see information. The graph can be zoomed in by dragging a rectangle, or pinching in to zoom on a tablet. Also note that by clicking on the different series in the graphs legend, you can temporary toggle hiding the series that are displayed on the graph. The graph can also be printed or exported in a variety of ways, the graph legend can be hidden with the Toggle Legend Display and the graph tooltip can be hidden with the Toggle Tooltip option, simply click on the three bars in the top right corner of the chart to show the chart context menu.
Dual Display - For larger displays, it may be more useful to see the spreadsheet and graph beside each other at the same time. Select this tab to enable that.
Statistics - When 'Baseline Forecast' view is selected, this tab will show statistics about the model.
The Model Display page allows you to view your TESLA model, in a variety of ways if the model supports the views, e.g. the Baseline Forecast, Comparable Days (US only), Temperature Steps, Customer Numbers and Weather Decomposition.
The model section allows you to select the model you're interested in viewing and the type of view on that model.
Selecting a model
Click the downwards pointing arrow to the right of the current model name to see a list of models. If you have many models, you can start typing to filter the view. Click on the model name to select it.
Types of model views
Please note that the different views on the data may not be available if the underlying data isn't there. I.e. if a model was created without using customer numbers, then the 'Customer Numbers' choice will not be available.
- Baseline Forecast - Displays the TESLA forecast for the selected model, as well as any observed load values and any selected historical forecast. In Spreadsheet view you can display the associated weather variables, such as temperature, humidity, etc. You can toggle to a temperature only view in the 'Spreadsheet Options' section.
- Comparable Days - Retrieves load and weather data for the five days most closely resembling the selected day. The Comparable Days section will also be displayed where you can alter the criteria you want the search algorithm to use if necessary.
- Temperature Sensitivity - Allows you to display 'quick scenarios', where you can view the effect that varying the temperature variable only has on the TESLA forecast. Select checkboxes in the 'Temperature Steps' section to view the different forecasts. You can view a spreadsheet or graph view of the forecasts.
- Customer Numbers - A spreadsheet only view of the customer numbers that the TESLA system used when producing the forecast.
- Weather Decomposition - Weather effects decomposition consists of comparing the load experienced under actual or forecast weather with what would occur under some alternative weather conditions (generally, seasonal norms), and further decomposing the difference into that portion due to the individual effects of the differences in temperature, humidity, cloud cover, wind, precipitation, and the interaction among these weather variables.
When unticked, the Stage 1 evaluation is displayed. This is the model output made without use of the most recent observed load information and is roughly equivalent to model error two or more days ahead. When ticked, the Stage2 evaluation is displayed, where the system is allowed access to the most recent load data. This reduces model error, especially over the next few hours and up to 48 hours ahead.
Allows you to select the date and horizon (days to display) of interest. Also allows you to easily page through days/weeks
Days to Display (Horizon)
A pull down allowing you to adjust the number of days to display at once. Depending on the period of the data, the options will change to allow a larger horizon (i.e. with daily data).
Clicking the date control allows you to select any date quickly. You can type in a new date or select from the calendar control.
Note that you can:
- Page through months by clicking the arrows at the very top left and right of the control.
- Click the 'Today' button at any time to change the month display to the current month, where you'll see the current day highlighted. Note that you will then have to click on the highlighted date to display todays date.
Clicking the '<<' and '>>' buttons will page through weeks.
Clicking the '<' and '>' buttons will page through days.
Historical Data Section
If 'Baseline Forecast' is selected (in the 'Model' section) then the 'Historical Data' section will display. Allows you to bring up a TESLA forecast that was recently produced, such as the forecast for the current day made early in the morning, or the forecast produced yesterday (the '1 day ahead' option) at a certain time.
Historic Load Forecast
Checking this box will bring up the historic forecast in both the Spreadsheet and Graph view, based on the selected day and time.
Historic Weather Forecast
Checking this box will bring up the historic weather forecast in both the Spreadsheet and Graph view, based on the selected day and time.
Allows you to select the forecast that was made in the past. Selecting 'Intraday' will allow you to display a forecast that was produced in the current day. The other options allow you to select previous days. E.g. selecting '2 day ahead' will display the forecast made two days ago for the currently displayed date.
You can type in a chosen time (if you are using TESLAWeb in a different timezone to the model, the time used in data retrieved will be in the same timezone that the model is in), or click the clock icon to allow you to page through hours and minutes.
Historic Rolling Forecast
Checking this box will bring up Age of historic forecast values time selection. Select the time you'd like to display the historic rolling forecast for. This means that every historic rolling forecast value for every given time, will be the forecast that was produced for that time, X hours and minutes ago (X based on the Age of historic forecast values time selection). E.g. if Age of historic forecast values is set to 2 hours, then the forecast value at 10am will be the forecast value produced at or before 8am. For 11am it will be the forecast value produced at or before 9am, etc.
Click the Refresh button after changes have been made to update the display.
Comparable Days Section
When the 'Comparable Days' option is chosen, this section will be displayed. Allows you to change the criteria you want the search algorithm to use when finding historical days to compare against the date you're currently looking at. After altering sliders, please click the 'Refresh' button in the 'Model' section to retrieve comparable days based on your new settings.
By default, the greatest selection weight is placed on weather; to change the weights of the weather factors.
- Adjust the slide control labelled Overall weight on weather to select a value between 0 (weather not considered) and weather given 50 percent more weight than the default value.
Adjust sliders in the Weather relative weights group to fine-tune the balance between afternoon high temperature, morning low temperature, and relative humidity/dew point
- These weights must sum to a fixed number, so when you move one slider to set its weight, the other two will also move to indicate the new overall balance.
Adjust the Penalty weight for weekday slider to change the weight placed on being the same type of day.
- Saturdays can only be matched to Saturdays, Sundays only to Sundays.
- The control adjusts the weight given to differences among weekdays.
Adjust the Penalty for age slider to change the weight place on age
- Move the slider to the left to make selection of older observations more likely, or move it to the right to make selection of new observations more likely
Adjust the Penalty for season control to change the weight placed on nearness in season to the selection
- Nearness in season refers to how close in the annual calendar two observations are. E.g. 1 June 2010 and 1 June 2011 are perfect matches in season, although they are a year apart in age
Temperature Steps Section
Temperature steps show how the load forecasts behaves when the temperature is raised or lowered by a number of degrees. I.e. 'Plus 2' shows the load forecast when the temperature is two degrees warmer than the supplied weather data.
When the 'Temperature Sensitivity' display option is chosen, this section will be displayed. Allows you to choose what steps in temperature should be displayed. I.e. checking 'Plus 1' will display the TESLA forecast that is produced when the weather station temperature(s) are all increased by one degree.
Weather Decomposition Section
Weather Decomposition is where the effects of weather on the load are removed. The remaining "weather-adjusted load" measures true load growth, which allows you to adjust revenue for transient weather and calendar effects. It also is crucial in long term infrastructure planning.
You can also see the effect that each individual weather variable has on the load and/or load forecast.
Checking this box allows you to select a weather station to plot a given weather variable for.
Show Weather Effects
Checking this box shows the effect that each individual weather variable has on the historical load or future load forecast.
Show Total Weather Effect
Overlays a line showing the overall effect (i.e. adding/subtracting all the individual weather effects)
Weather Risk Section
Our methodology for long-range forecasting begins with the same model that we build and maintain for short-range forecasting. TESLA models are designed to capture the complex relationships between weather, clock, and calendar variables and demand for electricity. To account for slow-moving trends caused by such factors as economic growth, population growth, and increases in energy efficiency, among other things, we use a latent variable to econometrically estimate the trend in the data.
Long-range forecasting requires that those slow-moving trends in the historical data be projected forward through the desired forecast horizon. TESLA accomplishes this by projecting the values of the latent variable over the next few months using an autoregressive moving average of the values over the last few months of the sample. That short-term trend line will damp to one established by the same methodology but employing a larger sample of historical observations, thereby establishing a long-term trend line. We believe that this methodology is implicitly consistent with a base, or business-as-usual, economic scenario.
TESLA load forecasting models are dependent on a weather forecast input. For long-range horizons where traditional weather forecasting methodologies cannot be relied upon, we employ a Monte Carlo simulation using long histories of actual weather data to generate long-range weather inputs for our models.
This process involves forecasting future periods using historically observed weather data. As an example, to forecast energy demand for the period January 1, 2015 through December 31, 2015, we transplant actual weather from January 1, 2000 through December 31, 2000 into the forecast horizon and generate a load forecast. We then shift that span of weather backward seven days and forward seven days, one day at a time, generating fifteen independent forecasts using just the weather data from the year 2000. This allows us to capture the effects of different actual weather patterns occurring on different days of the week. We then perform the same task using weather from January 1, 2001 through December 31, 2001, and so on. In total, we would generate 150 independent load forecasts for the period January 1, 2015 through December 31, 2015 using ten years of historical weather data.
TESLA organizes the forecasts into 10th, 20th, 50th, 80th, and 90th percentiles. For each observation in the 50th percentile hourly forecast, there is a 50% chance of observing a higher forecast, and a 50% chance of observing a lower load forecast.
The TESLA methodology is useful because it makes use of historical data for weather patterns that actually occur in a given geographical area. The alternative of using seasonal normal weather for forecasting discounts the inter-relationships between weather variables and ignores the fact that seasonal normal weather is never actually observed. There is also the advantage, in the case of hourly forecasts, of being able to identify peaks and aggregate energy totals over any desired interval. Lastly, percentile forecasts give a more descriptive account of the level of uncertainty surrounding long-range forecasting.
Display Historical Forecasts
Checking this box allows you to select a historical year, which will overlay all 15 forecasts that were produced for that year that went in to calculating the overall percentiles.
Select All / Median Only
Will check/uncheck the percentile checkboxes for you.
Spreadsheet Options Section
This section is displayed when the Spreadsheet tab is chosen.
Will reduce the number of columns displayed. For each weather station associated with the chosen model, will display just the temperature.
Will display all the weather variables for all the weather stations. I.e. you will see a column of data for each weather variable for each weather station.
Clicking this button will download a CSV file (which you can read with any spreadsheet editor, like Microsoft Excel). The default filenames can be overridden in the 'Settings' page if desired.
Graph Options Section
This section is displayed when the Graph tab is chosen.
Will plot on the graph the associated weather station(s) that make up the selected model.
Note that observed weather will be plotted in a darker colour than the forecast weather.
Weather Station to Plot
Allows you to select just one weather station to plot when many weather stations make up the model.
Weather Variable to Plot
You can choose what weather variable to display. Typically temperature is chosen.
These options are displayed when the Weather Decomposition Graph display is chosen:
Show Weather Effects
Displays a stacked column graph of the different weather effect variables and how they have impacted load
Show Total Weather Effect
Displays a line graph of the total effect weather has had on the load.
Graph menu (click three bars top right of the chart)
Allow you to print and download the chart. Also allows you to toggle off the legend (Toggle Legend Display) as well as toggling off the tooltip (Toggle Tooltip) which displays when you hover over points.
This page allows you to reset your password and customise your TESLAWeb experience.
The first line shows the id that you are currently logged in as.
To change your password:
- Enter your existing password in to the 'Current Password' text box
- Enter your new password in the 'New Password' and 'Confirm New Password' text boxes.
- Click the 'Save' button
Allows you to change how numbers are displayed in the spreadsheet and graph views.
The number of decimal places to display for each number. E.g. '2' will show numbers as 0.00
For ease of reading, numbers with many digits before or after the decimal mark may be divided into groups using a delimiter. Choose that delimiter here (there's an option for no delimiter).
A decimal separator is a symbol used to separate the integer part from the fractional part of a number written in decimal form. Select your preference here (generally based on your locale).
Shows how numbers will look in the spreadsheet and graph after updating the formats.
If the example number doesn't refresh, click this button to do so.
Click Save to remember your changes.
Allows you to change how dates and times are displayed and chosen in the spreadsheet and graph views as well as throughout TESLAWeb.
Check this box if you would like to display a time period instead of or in addition to the time. If this boxed is checked, you will also see the following two options.
Period Column Name
You can have any custom name for the period. Default is 'SP' for 'Service Period'.
Period Numbering Starts
If 'Midnight' is selected, then the first period of the day will start at midnight. If 'After Midnight' is selected, then the first period of the day will start at the first value after midnight.
Check this box to display time in addition to period. Note that if 'Display Period' is unchecked and this checkbox is hidden, it will default to being checked.
You can define how you want your date format to look here. Typical formats are
Note that 'M' must be capitalised. A lower case 'm' means minute (see time format).
You may use any of the following to define date formats:
The day of the month, from 1 through 31.
The day of the month, from 01 through 31.
The abbreviated name of the day of the week.
The full name of the day of the week.
The month, from 1 through 12.
The month, from 01 through 12.
The abbreviated name of the month.
The full name of the month.
The year, from 0 to 99.
The year, from 00 to 99.
The year, with a minimum of three digits.
The year as a four-digit number.
The year as a five-digit number.
You can define how you want your time format to look here. Typical formats are
- h:mm tt
Note that 'm' must be lower case. A capital 'M' means month (see date format).
You may use any of the following to define time formats:
The hour, using a 12-hour clock from 1 to 12.
The hour, using a 12-hour clock from 01 to 12.
The hour, using a 24-hour clock from 0 to 23.
The hour, using a 24-hour clock from 00 to 23.
The minute, from 0 through 59.
The minute, from 00 through 59.
The first character of the AM/PM designator.
The AM/PM designator.
After updating your date and/or time formats, click the 'Refresh Examples' button to update the example date and time text.
Click 'Save' after changing settings to record them.
Allows you to alter the spreadsheet display
Highlight Daily Min & Max Values
Check this checkbox to highlight in pink, daily high values and in blue daily low values in the spreadsheet display.
Customer Number Column Format
If your models use customer numbers, you may override the format of the column names here if desired.
Allows you to alter the graph display
Display Data Points
Check this checkbox to show all the data points on the line graphs. When unchecked, points only show up when you hover the mouse above them.
Check this checkbox to make the graph automatically select the minimum value on the y-axis, based on the data being displayed. When unchecked, the graph y-axis starts at zero or whatever the minimum value is if negative.
Spline point interpolation
Check this box if you want smooth (curved) lines connecting data points on the graph. Otherwise, straight lines will join the data points.
Allows you to specify what statistics you want displayed in the Statistics tab in 'Model Display'.
Model Level Statistics to Display
Specifies the statistics for the model which compares the currently displayed forecast against currently displayed actual data.
Daily Statistics to Display
Specifies the statistics for the load forecast you want to display. These are done on a daily basis.
If you would like the option to display models at a different granularity (e.g. half hourly as hourly or daily), check this box. You will then see an "Aggregation" pull-down appear in the "Model" section on the "Model Display" tab. You can then select the type of aggregation you wish to see.
Automatically refresh forecast
When checked, the display will automatically refresh to display new data when the load forecast is updated
Display message briefly after refresh
When checked, a message will show for 10 seconds after refreshing the data.
Display message if significant change to forecast
If checked, then a section will appear where you can define what a significant change to the forecast is (either by specifying a value or % change).
Variation for alert
Here you can specify what a 'significant' change in the forecast is. You can type in a percentage or value difference and select the type from the pull-down. E.g. if you select 1%, then if a new forecast has any values that are more than 1% different to the currently displayed forecast, a message will be displayed (which you can dismiss by clicking 'x').
Show difference between fcst and previous
When viewing the baseline forecast and displaying a historical (previous) forecast, in spreadsheet view, you can choose to show a column which displays the difference between the latest TESLA forecast and the displayed historical (previous) forecast. Check this box to display the 'Fcst Delta' column in spreadsheet view when 'Baseline Forecast' display is selected on the 'Model Display' page.
forecast - previous
select to subtract the previous forecast figures from the forecast figures.
previous - forecast
select to subtract the forecast figures from the previous forecast figures.
Display stop date rather than horizon
If selected, will allow you to select a start and stop date to specify the amount of data to display on screen at once.
Display horizon rather than stop date
If selected, will allow you to select a start datea and horizon to specify the amount of data to display on screen at once.
Shift Historical Forecast date for every day displayed
If selected, when displaying "Historical Load Forecast" for several days in the past, you will see the date and time that the historical data is pulled from incremented for every displayed day. I.e. this will have no impact when viewing a single day, or future forecasts.
Data Export Filenames
Allows you to override the default file naming when exporting spreadsheet on the 'Model Display' (when viewing a Spreadsheet).
The link to this page is only visible if you are logged in as an administrator.
This section allows you to add/delete/edit user accounts.
Select User - If you wish to edit or delete a user, select them from this drop down. If there are many users, you can start typing to filter the list.
Delete - Click this button to delete the selected user. You will be prompted first.
Account Tab - Allows you to edit the selected users login details, or enter in the details of a new user (click 'Clear Form' first if you wish to add a new user).
Personal Details Tab - Where you can record the Name, Organisation, Job Title and Phone Number of the user if desired.
Models - Where you can select what models in the TESLA system the current user is able to view. Check the check boxes next to each model name to let the user view and select them in the 'Model Display' page.
Allows you to select models and display a message when a user chooses them in the 'Model Display'. The message will display above the spreadsheet/graph tabs.
Select models to assign message - Check what model(s) you wish to assign a message to
Message for selected Models - Enter the message you would like displayed when a user chooses the selected model.
Message valid to - enter a date and time when the message will expire. After this point in time, the message will no longer be displayed. You can click the calendar icon to bring up a calendar to select the date. Click the clock icon below the calendar (when it's brought up) to graphically edit the expiration time.
Used to tell the server to throw exceptions. This is only used when testing the TESLA installation, checking that correct logging and behaviour of errors happens.
Allows you to bring up the TESLA log so you can track when users log in to the system and what models they are viewing.