Text version of Images

Figure 19.

Block diagram showing elements of energy data analysis. Energy Data Analysis -> What is energy data analysis?
-> What is energy Data ->Why do I need to analyse data? What is energy data analysis?
-> Breakdown use and costs -> Pie charts, Profiles, Contour pilots
-> Calculate performance indicators -> Efficiency measure, Specific energy use, Average use
-> Understand variability -> Frequency distribution, Scatter plots, Regression, Visualization, Data Mining, Calculate targets, Modeling (what if? and optimization) (Visualization -> [Trend] Response plots, 3-D plots; Data mining -> [Rule induction] Parallel plots, Other statistical methods; Calculate targets -> [Regression] Averages, Data mining, Data modelling; Modelings [what if? and optimization]
-> Data mining, Neural nets, etc.)
-> Statistical process control
-> Data preparation

Figure 20.

Breakdown of energy use. By department (pie chart 1): Raw Materials Preparation (approximately 25%), Crushing and Grinding (approx. 15%), Reactor (approx. 25%), Separations Process (approx. 15%), Drying (approx. 6%), Compaction and Finishing (approx. 3%), Warehouse (approx. 2%), Offices (approx. 3%), Other (approx. 6%). By plant (pie chart 2): Steam Generation (approx. 35%), Compressed Air (approx. 8%), Refrigeration (approx. 25%), Cooling Towers (approx. 3%), Pumps (approx. 10%), Fans (approx. 5%), Agitators (approx. 2%) Other (approx. 12%).

Figure 21.

Breakdown of utilities' use and cost. Pie chart 1: Gas (50%), Oil (approx. 30%), Electricity (approx. 20%). Pie chart 2: Electricity (approx. 60%), Gas (approx. 30%), Oil (approx. 10%).

Figure 22.

Monthly breakdown of electricity use showing day and nighttime units (bar chart). Along the X axis, the time is indicated in single month increments ranging from July 1998 to June 1999. Along the Y axis, electricity use is expressed in kilowatt hours per month (kWh) ranging from 0 to 25 000. July-98: day approximately 17 500, night approx. 19 000 ; Aug-98: day approx. 17 500, night approx. 19 000; Sep-98: day approx. 17 800, night approx. 19 500; Oct-98: day approx. 18 000, night approx. 20 000; Nov-98: day approx. 18 500, night approx. 20 500; Dec-98: day approx. 16 500, night approx. 18 500; Jan-99: day approx. 18 000, night approx. 20 500; Feb-99: day approx. 17 500, night approx. 19 500; Mar-99: day approx. 18 500, night approx. 21 000; Apr-99: day approx. 17 000, night approx. 19 000; May-99: day approx. 17 000, night approx. 19 000; Jun-99: day approx. 18 000, night approx. 20 500.

Figure 23a.

Typical half-hourly demand profiles (as a line graph). Along the X axis, the time is expressed in 1/2 hour increments ranging from 00:30 to 24:30 hours. Along the Y axis, the demand is expressed in kilowatts in increments of 100, ranging from 0 to 700. A jumble of multiple lines running along the X axis appear in dark blue, medium blue, grey, black and white and range from just under 250 kW to just under 400 kW at the 1 hour mark before beginning a sharp rise starting at the 06:00 hour mark up to a peak at the 15:30 hour mark ranging from approximately 250 kW to 600 kW. The lines gradually descend to a low at the 24:00 mark ranging from just under 250 to approximately 395 kW.

Figure 23b.

Typical half-hourly demand profiles (as a contour plot). Along the X axis, days in the month of January are indicated in increments of six days from 01-Jan to 31-Jan. Along the Y axis, demand is indicated (although not stated, kilowatts is the presumed unit of measure based on Figure 23a) in increments of 100, ranging from 0 to 500. Along the Z axis appears the following sequence: S8, S15, S22, S36, S43, S50. The legend on the bottom right side of the image shows the plotted demand profile categories by the following colours, which appear in order from bottom to top: white = 0-100, dark blue = 100-200, dark grey = 200-300, light blue = 300-400, light grey = 400-500, medium blue = +500. In this order, along the Y axis these colours rise in bars that run in parallel to the Z axis starting from the 0 mark at the bottom left of the image upwards to above the 300 mark, then start to curve forward and continue to rise across the X axis in a wavy arc to a high point near the 500 mark at 19-Jan, then curve downward towards 31-Jan near the 400 mark. Running parallel to the Z axis, the demand profile category colour bands vary in width. White remains a constant width at a demand of approximately 50. Dark blue starts off narrow at a demand of approx. 75 at S8 then increases in width gradually towards S22, then levels off at approx. 225 until S50. Dark grey starts off wide at approx. 200 at S8, rises sharply from S15 to S22 reaching a high point of 250, then falls sharply to approx. 200 at S29, then levels off at just above 200 after S29 to S36, then falls sharply to 200 at S36 and stays close to this demand until S50. Light blue starts at a demand of just above 300 and appears to remand constant until approx. S36 where it falls gradually to 300 until S50. For light grey it is difficult to discern approximate demand figures due to the three-dimensional nature of the graph but the band appears to start off at approx. 425, then gradually fall towards 400 at S22, then gradually rise to a peak of approx. 475 at S43, then gradually fall to approx. 450 at S50. Medium blue presents the same difficulties in discerning approximate demand figures but appears to start off at approximately 500, then very gradually rise towards 550 at S50.

Figure 24.

Example of frequency distribution. This graph shows a typical example of the specific energy cost of a process operation. Energy Cost is indicated in blue bars of different lengths along the X axis. The height of these blue bars is indicated as Frequency along the Y axis. No units of measurement or increments are specified. The bars steeply rise in height from the bottom to the top of the Y axis at the midway point along the X axis, then fall steeply until 3/4 of the way across the X axis, then continue to fall variably and gradually towards the bottom of the X axis.

Figure 25.

Energy use vs. production. In this graph, along the X axis Production Rate is indicated in increments of 50 ranging from 500 to 1000. Along the Y axis, Energy Use is indicated in increments of 500 and ranging from 3000 to 6000. No units of measurement are specified. A large jumbled cluster of medium blue dots appear in the plot area starting with a small concentration near the 4000y/600x mark and expanding at 700x into a larger concentration spreading from neaer the 4500y/650x mark gradually rising upwards to near the 5500y/1000x mark. The bottom of the cluster spread runs from 4000y/600x rising variably and gradually to 5000y/1000x. Near the top left side of the graph inside the plot area, the following equation appears: y = 2.7x + 2430.6

Figure 26.

Example of three-dimensional plot, showing energy use vs. production and feed quality. Along the top of the plot area runs the Energy Use (z) axis for which no increments are specified. Along the bottom of the plot area runs the Food Quality (y) axis in increments of 1.25 ranging from 4 to 9.0. Along the left side of the plot area runs the Production Rate (x) axis in increments of 100 ranging in black text from 600 to 1000. Immediately to the left of the 1000 black text appears 3000/ in blue text, and above that point the increments change, appearing as follows (also in blue text): 3625, 4250, 4875, 5500. No units of measure are specified for Z, Y and X. A large jumbled cluster of blue dots appear scattered primarily within the following area of the plot (roughly in the shape of a dog running towards the right of the plot area): top near 5500x/6.5y, bottom near 900x/6.5y, left near 3625x/5.25y and right near 3625x/9.0y. A small scattering of light blue dots also appear across the following area of the plot:.between 800 and 900x and between 5.25 and 9.0 y.

Figure 27.

Multiple scatter plot. This plot shows five rows of five square cells, within which the shaded cells identify the axis labels for each variable that is plotted. The first non-shaded cell in the top row plots production rate on the y-axis vs. energy use on the x-axis. The first cell on the bottom row plots the controllable value "Control 1" on the y-axis vs. "production rate." High-energy-use data is coloured light blue. As an example of the patterns that may be spotted in this multiple two-dimensional scatter plot, the relationships between energy use and feed quality are clear. Another example of high energy use is associated with lower values of the controllable value "Control 2."

Figure 28.

Simple decision tree. A simple decision tree is shown as follows: All -> Solvent Temperature < -24.4260864257813 -> Solvent Temperature > -24.4260864257813 a… -> Solvent Temperature > -23.1069660186766 a… -> Solvent Temperature > -14.0492753982544 A screen shot shows the title "Attributes" at the top left. Underneath this title are two tabs running horizontally: "Totals" and "Histogram". Histogram is selected and shows the followiing table:

Figure 29.

-> Liquid Flow Dc <= 2052.8359375
-> Liquid Flow Dc <=1745.64721679688
-> Liquid Flow Dc > 1745.64721679688 a…
-> Liquid Flow Dc > 1822.14733886719 a…
-> Liquid Flow Dc > 1973.91.296386719
-> Liquid Flow Dc > 2052.8359375 and <...
--< Reagent Use C <= 0.11189890652895
--> Regeant Use C > 0.11189890652895 a…
--> Regeant Use C > 0.139309987425804 …
--> 0.198505671525002
-> Liquid Flow Dc > 2160.28857421875 a…
--> Reagent Use C <= 0.119573406875134
--> Reagent Use C > 0.119573406875134 …
--> Reagent Use C > 0.131518706679344 ...
--> Reagent Use C > 0.201007053256035
-> Liquid Flow Dc > 2224.15991210938
--> Regeant Use C >= 0.122352823615074
--> Regeant Use C < 0.122352823615074
… --> Regeant Use C < 0.164333060383797
… -> Regeant Use C > 0.122352823615074 and < 0.164333060383797
--> Regeant Use C > 0.225805848836899

A screen shot shows the title "Attributes" at the top left. Underneath this title are two tabs running horizontally: "Totals" and "Histogram". Histogram is selected and shows the followiing table:

Figure 30.

Stages of an initial data-mining analysis. The left side of the image shows the following flow for these stages (bulleted text appears to the right of each stage, as listed in brackets): Understand Process Operations (- workshop, - discussion with engineers and operators) -> Define Performance Objectives (- workshop with engineers and managers) -> Identify and Collect Historical Operating Data (- study data and historians, - acquire historical data) -> Data Analysis (- pre-process data, transformation, - data visualization, - data mining) -> Interpretation (- workshop to interpret patterns, - identify opportunities) -> Develop (- develop opportunities, - quantify savings, - assess practicalities) -> Implementation Strategy (- implementation and workshop).

Figure 31.

Boiler manifold steam pressure (graph). Time is indicated along the x-axis (no units of measure or increments are specified). Manifold Pressure is indicated along the y-axis in increments of 5 ranging from 1450 to 1480 (no units of measure are specified). Data oints are plotted in a zig-zag line parallel to the x-axis starting at approximately 1460, rising sharply to nearly 1470, descending to 1465, rising mid-way between 1465 and 1470, rising gradually from this point to between 1470 and 1475, continuing at this point then rising to just below 1475, then sharply descending to mid-way between 1465 and 1470, then sharply rising to 1470, then gradually descending to mid-way between 1460 and 1465, then sharply rising to just above 1470, then sharply descending to mid-way between 1465 and 1470, then gradually rising to just below 1475, then descending to between 1470 and 1475, then sharply rising to just above 1475, then gradually descending to 1460, then sharply rising to 1475.

Figure 32.

Impact of manifold pressure on operating cost – partial decision tree. The partial decision tree is shown as follows. From the top right of image and fllowing down and towards the left: = 3.1

Figure 33.

Cost vs. manifold pressure (plot). Manifold Pressure is indicated along the x-axis in increments of 5 ranging from 1450 to 1480 (no unit of measure specified). Cost is indicated along the y-axis (no increments or units of measure are specified). Half-way up the y-axis, a thick band of scattered dark blue dots runs across the plot area parrallel to the x-axis.

Figure 34.

Actual vs. target performance (graph). Time is indicated along the x-axis in increments of seven ranging from 1 to 92. Manifold Pressure is indicated along the y-axis in increments of 0.2 ranging from 3.0 to 4.0. No units of measure are indicated. The legend in the top right corner of the image shows a light blue line "Expected" and a dark blue line "Actual". Expected and Actual are similarly plotted in zig-zag lines across Time, starting with a Manifold Pressure ranging from 3.6 to 4.0 between 1 and 8 on the Time axis, then dropping sharply to a range of between 3.3 and 3.5 between 8 and just before 22, then rising sharply to just above 3.6 at 22, then dropping sharply to a range of between 3.3 and 3.5 at 29, then rising sharply to 3.6 just past 29, then dropping sharply to 3.5 at 36, then rising sharply to just under 3.6 and dropping sharply to approximately 3.4 just before 43 where Expected and Actual start to diverge significantly.

Expected rises sharply to above 3.6 at 43, then fall to 3.5 between 43 and 50, then rise to a peak of 3.9 at 50. The words "Faulty operation" appear with an arrow pointing at this peak. After this peak, Expected declines to 3.7 just after 50, then rises to just above 3.6 and falls sharply to just below 3.6 at 57, then rises sharply and variably to just above 3.8 before 64, then falls sharply to 3.6 at 64, then rises sharply to 3.8 and falls sharply to below 3.2 at 71, then rises sharply to 3.4, then falls to just below 3.4 and rises sharply to 3.7 at just before 78, then falls sharply to 3.5 before ralling sharply to just above 3.2 then just below 3.2 just before 85, then rises and falls variably between 3.3 and just below 3.2 until 92.

Actual rises to 3.5 at 43, then drops sharply to above 3.3 between 43 and 50, then rises sharply to above 3.5 at 50, then dips slightly before rising again to 3.5 and sharply falling to 3.3 just before 57, then rising to just below 3.6 and dropping sharply to 3.3 at 64, then rising sharply to 3.5 and dropping sharply to below 3.2 at 71, then rising sharply to just below 3.6 and dropping sharply to above 3.3 at 78, then continuing to slightly rise and fall variably between just under 3.4 and just over 3.2 to 92. The top right of the screen shot shows the minimize, restore down and close boxes typical of most Microsoft Windows-based applications.

Figure 35a.

Performance reporting. A screen shot shows "Operations Report" in the top left of the image. Under this text is:

Under this text appears the bar graph Performance vs. Target. The legend at the top right corner of the graph shows a light blue line "Actual" and a dark blue line "Target". The x-axis shows the following: Margin, Output, Yield, Quality, Energy. Bars rise up from these elements parallel to the y-axis, which shows increments of 20 ranging from 0 to 100. Margin shows Actual just above 60 and Target just below 100; Output shows Actual just above 40 and Target just below 80; Yield shows Actual just above 80 and Target just below 100; Quality shows Actual at 90 and quality at 80; Energy shows Actual at 100 and Target at 80. The top right of the screen shot shows the minimize, restore down and close boxes typical of most Microsoft Windows-based applications.

Figure 35b.

Cumulative sum (CUSUM) reporting (as a line graph). A screen shot shows "Operations Report" in the top left of the image. Under this text is: Main Reactor and Separations Process - Energy Cost Last Eight Hours: 22.11.02 16:00 to 22.11.02 24:00 Energy Overspend 123456 $ per year equivalent - Improving trend - Recent performance good Under this text appears the line graph Performance vs. Target. The legend at the top right corner of the graph shows a light blue line "Actual" and a dark blue line "Target". The x-axis shows increments of 2 randing from 2 to 32. The y-axis shows 0.0, 1.0, then increments of 0.5 ranging from 1.5 to 4.0. No units of measure are specified. The lines Actual and Target run parallel to the x-axis in a variably straight line. Actual starts at 3.5, then rises to just below 4.0 at nearly 8, then drops and remains steady at 3.5 until 16, then rises to just below 4.0 and remains fairly steady until a gradual drop starting at 25 until reaching just below 3.5 and staying at this level between nearly 27 and 29, then rising slightly to 3.5 until 32. Target starts at just above 3.5 and rises steadily to 4.0 at just before 8, then drops to and remeans at just above 3.5 from 9 to 11, then rises to close to 4.0 at 12 and drops to just above 3.5 before 16, then rises to a peak of above 4.0 and variably drops to a low of 3.5 where it remains fairly steady until 32. The top right of the screen shot shows the minimize, restore down and close boxes typical of most Microsoft Windows-based applications.

Figure 35c.

Cumulative sum (CUSUM) reporting (as monthly, weekly or yearly summaries). A screen shot shows "Executive Report" in the top left of the image. Under this text is: Entire Manufacturing Facility Monthly: January to December 2002 Margin Loss 123456 $ per year equivalent

- Significant margin loss - Improving trend - Recent performance good - Worst performance January and July - Best performance November and December Under this text appears the bar graph Performance vs. Target. The legend at the top right corner of the graph shows a light blue line "Actual", a medium blue line "Target" and a dark blue line "Best Practice". The x-axis shows each month ranging from Jan to Dec. The Actual, Target and Best Practice bars rise up from these elements parallel to the y-axis, which shows increments of 20 ranging from 0 to 100. Jan shows Actual at 100, Target at 90 and Best Practice at 80; Feb shows Actual at just under 90, Target at 80 and Best Practice at 70; Mar shows Actual at just above 80, Target at 80 and Best Practice at 70; Apr shows Actual at just below 100, Target at just above 90 and Best Practice at just above 80; May shows Actual at just above 90, Target at just above 90 and Best Practice at just above 80; Jun shows Actual at 90, Target at 90 and Best Practice at just below 90; July shows Actual at 100, Target at 90 and Best Practice at 80; Aug shows Actual at just below 90, Target at 80 and Best Practice at 70; Sept shows Actual at just above 80, Target at 80 and Best Practice at 70; Oct shows Actual at just below 100, Target at just above 90 and Best Practice at just above 80; Nov shows Actual at just above 90, Target at just above 90 and Best Practice at just below 90; Dec shows Actual at 90, Target at 90 and Best Practice at just below 90. The top right of the screen shot shows the minimize, restore down and close boxes typical of most Microsoft Windows-based applications.

Figure 36c.

Cumulative sum (CUSUM) reporting. The x-axis represents an unspecified course of time. The y-axis indicates cumulative savings (loss) ranging in increments of 100 from -900 to 0. A blue line runs from 0 fairly straight across to the mid-way point in Time, then drops sharply to a low of between -800 and -700 at four fifths across the Time line, then rises to just above -700 at the end of the Time line.