Building monitoring

Energy Display Scripts Outline version 1.0

1. Java scripts harvest data directly from Modbus
2. Perl scripts collects this data and put it in the Postgress database "energy" under the table "electrical_energy"
   • electrical_energy attributes:
     • area - what building.
     • preal - kW hours for that instant.
     • date - date of harvest.
3. A final perl script (display/production|development/js-file-gen.pl) accesses this data and generates a .png graph using the Google graph API
   • This script takes one parameter for either the production or development branches.
4. The images generated (day, week, month, year) are put in a html frame on a page to display the data.
   • This page refreshes every 5 minutes

Energy Display Scripts Outline version 2.0

Postgres db

• Data stored in databse 'energy' under table 'electrical_energy'
• 'electrical energy' has the following attributes:
  • area - which building.
  • preal - kW hours for that instant.
  • date - date of harvest.

data_gen.py

• generates a JSON file filled with data from a sql call to the database.
• requires Python 2.5
• file run by cron

• Main(building_file)
  • Called from main loop.
  • Calls CreateBuildingList(), ConnectToDatabase(), GetAllData(), EncodeData(), DisconnectFromDatabase()

• CreateBuildingList(building_file)
  • Called by Main().
  • For each building in building_file add to a python list.
  • return that list.

• ConnectToDatabase()
  • Called by Main().
  • connects to 'energy' databse in Postgres
  • returns database connection and the current name of database.

• GetAllData(db connection, building list)
  • Called by Main().
  • makes a python dict of all data
  • calls GetBuildingData()
  • replace unwanted dates from older buildings
  • calls EqualizeDataValues()

• GetBuildingData(dbconnection, building)
  • called from GetAllData()
  • calls SetSQLQuery()
  • returns rows from sql call and row count

• SetSQLQuery(building)
  • called from GetBuildingData()
  • Makes a sql call to the databse energy from the table electrical_energy
  • returns rows from sql call and row count

• EqualizeDataValues(all_data, dates)
  • fills null values for dates of newer buildings.
  • returns all data

• EncodeData(all_data, dates)
  • Called by Main().
  • sorts all_data by date
  • dumps data in JSON file, note: columns are organized alphabetically.
  • initializes gviz metadata
  • calls LoadData().
  • writes JSON file which JS reads from.

• LoadData()
  • Called by EncodeData()
  • gviz api call
  • encodes data to JSON file with headers.

• DisconnectFromDatabase()
  • Called by Main().
  • disconnects from database 'energy'.

main.js

• non-function
  • creates several global variables.
    • whichBiuldings: an array of booleans signifying which buildings get drawn
    • buildingColors: an array of colors for each of the buttons and colors on the graph, change them here and only here.
    • startDate, endDate: range on the graph, by default they are one week ago and today respectively, if the page was reached through a permalink these get modified.
    • lineChartJSONData: this is the json file which data_gen.py makes, it's placed as a global so it doesn't have to be reloaded everytime a building is checked or unchecked.

• flipAndRedraw
  • called when a button is pressed
  • flips values in the global whichBuildings[]
  • calls drawLineChart() provided whichBuildings has at least 1 true value.

• flip(val)
  • takes a boolean and returns its opposite.

• redrawVis(values[])
  • Takes an array of booleans
  • calls DrawLineChart()

• drawLineChart(buildingArray[], colors[])
  • called when index.html loads and when RedrawVis() is called.
  • takes an array of booleans processed by RedrawVis()
  • creates a Dashboard
  • creates a data table with the data source as the json file data-gen.py makes.
  • creates a chart wrapper, this is where you can adjust basic settings about the main chart.
  • creates a control wrapper, this creates the controling bar on the bottom. This is where you can adjust the starting range for the control wrapper.
  • creates a listener for the 'statechange' event on the controlWrapper, when the control range is updated the dates are saved. When a building is unchecked and the graph redrawn, the selection stays the same.
  • creates a listener for the 'ready' event on the chartWrapper. When chartWrapper is ready (all drawn) resizeControls is called.
  • deletes columns in data table depending on the booleans that are in values[], the more columns that are deleted the better the performance.
  • binds the chart wrapper and the control wrapper together, draws the chart.

• equalArrays(arr1, arr2)
  • checks if the contents of arr1 are equal to arr2, returns true if they are, false otherwise.

• parseDate(urlArg[])
  • takes the array of strings from a permalink and returns them as a date object

• parseInts(strArray[])
  • takes the array of strings from a permalink and returns them as an array of integers.

• parseButtons()
  • Takes button arguments from the permalink and unchecks the buttons on index.html accordingly using jQuery's removeClass medthod.

• removeIndices()
  • processes and array to return containing indices of which columns delete and therefore not draw.

• genPermalink()
  • generates a permalink to the chart's state and which buttons are checked. The format of the permalink is (url)/?(startDate, yyyy-mm-dd)%(endDate, yyyy-mm-dd)%(buttons, 0123)
  • writes permalink to text box on index.html

• redrawControls()
  • Accesses where the svg paths for the control are using getElementByTagName() and sets them using jQuery's attr method.
  • This is called every time the graph sends a 'ready' event.

HTML+CSS

• index.html, the file people load.
• includes jquery.minm.1.7.2, for gviz api
• main.js, draws graphs
• bootratp-button.js, enables buttons to behave like checkboxes. Toggling these buttons updates a global array of booleans held in main.js, and then redraws the graph depending on which are selected.
• main.css, styling (some css3).

Improvements and Bugs

• Performance
  • When all buildings (residential and academic) are monitored the graph will be unsuably slow as it will be reading from an even more massive file. At time of writing performance is already somewhat sluggish. One big issue is reading from the JSON file everytime, keeping that file in memory or a closer-to-cpu cache could help.
  • More resolution in the graph will come at the expense of performance, right now points on the graph are averaged values from each hour. It would be better for graph readability if points were every, let's say, two minutes, however the graph would be unusable.
• UI/Usability
  • dragable buttons in the control bar are too small for a touch interface, they are generated svgs from the gviz api, RedrawControls() in main.js is trying to fix this.

Building EGX100 IPs

• 159.28.165.100 - Bundy
• 159.28.165.101 - Barrett
• 159.28.165.102 - Wilson
• 159.28.165.103 - Warren
• 159.28.165.104 - Mills*
• 159.28.165.105 - Olvey Andis
• 159.28.165.? - Main switch gear (PM8)

* - CT magnets are not installed yet and/or no EGX100 on network

Building Amperage Correction Factors

• Bundy - 31.25
• Barrett - 15.625
• Wilson - 62.5
• Warren - 62.5
• Mills* - ?
• Olvey Andis - 31.25

Source: load the schneider doc, amps, reference page

Documentation

• MIB for the PM8 (device mounted on the main switchgear)
• Register list for the Enercept/EGX100s (devices mounted in each of the buildings)