LECTURE 2-Design of Energy-related Production Benchmarks.pptx

Topics on Engineering Management course

1. Design of Energy-related Production Benchmarks Mr G. Gope

2. Energy-related P roduction Benchmarks Correlating Energy Use with Production Consumption and Production Data Analysis CUSUM Analysis

3. Production Benchmarks The major purpose of a management act or process is to set goals and to ensure that achieved and realised in an efficient manner. This purpose of management is not any different for the process of energy management. To set such goals and targets in an energy management program requires, that the energy cost functions ( and not just the raw material cost functions ) for different products and services in a company/firm or organisation must be well understood.

4. Production Benchmarks In order to fully understand energy consumption , and, in particular factors that influence it, so as to develop energy cost functions for products/services, a correlation of energy consumption with some measure of facility activity is needed. If the facility were a Commercial building or Hotel, building occupancy or hotel room occupancy could pr be used as the key indicator of the facility’s activity. In the case of Industry , the key indicator of the facility’s activity is usually the P roduction at the industrial plant.

5. Benchmarking Processes Six Steps of Benchmarking Process The Benchmarking Process

6. Consumption and Production Data Analysis A through plant/facility analysis for energy purposes will generate data on production and consumption. The data can be plotted in one of three (3) basic ways as shown below for a typical plant whose production figures and energy consumption are listed as on the table on the right. Month Production( tonnes ) Energy Consumption(kWh) x 100 January 11.00 807.00 February 298.00 5174.00 March 374.00 6912.00 April 378.00 7375.00 May 381.00 9072.00 June 291.00 6274.00 July 229.00 3551.00 August 475.00 5357.00 September 569.00 8884.00 October 507.00 7556.00 November 377.00 7118.00 December 319.00 5855.00 Sample Production Figures and Energy Consumption (kWh ) for a specific Process at a Facility

7. (1)Time Series Plot of Production and Energy Consumption Energy Consumption and Production plotted against time (days, months, years e.t.c .) gives a graph as shown below. This is a time-based pattern, which does not tell us the quantitative relationship between consumption and production.  

8. (2) Scatter Plot Energy consumption plotted against production begins to explore how one depends on the other. This generates a scattering of points and the degree of the scatter may indicate the degree of energy consumption and production control within the manufacturing facility. This plot tells us a little more than the previous plot, but still does not have a quantitative relationship between consumption and production to use for analysis.  

9. (3) Linear Regression The technique of linear regression analysis can be applied to the data to generate the desired relationship as shown on the diagram below. Through the use of a trend line based on regression analysis of the energy consumption data and production data a relationship can be extracted form the data which can give us some information on the nature of production process. For instance it can give us an indication of the base-load consumption at the plant. This base-load consumption might be energy consumption that occurs even if there is no production. This is energy consumption for non-production purposes and often offers opportunities for energy saving. The use of regression analysis gives us information rather than data . This information may allow us to start asking questions on why certain energy consumption variations occur even if the production is constant. This usually leads to further investigations in trying to find answers to these questions. It may be that answers yielded by these investigations will take the form of savings opportunities . However, the Cumulative Sum of difference regression technique (CUSUM) will provide us with more useful information as we will see later and also provides the basis for creating control charts for processes.

10. (3)Linear Regression Technique

11. CUSUM for Energy Consumption CUSUM ( Cumulative SUM of difference) is a technique for monitoring energy consumption based on a comparison of real measured consumption and predicted values derived from regression analysis of historical trends. This technique is sensitive to any changes in conditions, such as process changes or equipment malfunctions ( e.g failure of capacitor banks), CUSUM also provides a basis for creating control charts that can be applied in the various process or accommodation units.

12. Steps/Method for CUSUM analysis : Energy consumption is plotted against production A linear regression analysis is done for those points that are determined to be the “baseline” and the functional relationship between consumption and production determined. The equation from the regression analysis is used to calculate “predicted” values of energy consumption for given production values. The difference between the real or measured energy consumption and the predicted value is calculated. The cumulative sum of those differences is calculated , and plotted against time.  

13. Interpretation of CUSUM On the CUSUM chart, the critical points are significant changes in the slope of the line. A downward trend in the line is indicative of savings in energy consumption in comparison to the base line performance, conversely, an upward trend is indicative of increased consumption in comparison to the baseline. CUSUM is used to analyse historical performance as well as to monitor and control present and future performance. The base line in the case of historical analysis is, theoretically, the performance pattern before any change occurred to impact to on energy consumption.

14. Determining the Baseline A manual technique can easily be used to establish the baseline data from any historical data as demonstrated by the following example . Example A manufacturing plant records its energy consumption and production over a 36 week period as given below:

15. Example: Production and Energy Consumption Week Production (T) Energy Cons.(kWh) 1 80 9669 2 85 9558 3 90 10527 4 87 10043 5 100 11790 6 110 12198 7 115 12837 8 95 10956 9 100 11320 10 105 11984 11 130 13453 12 135 14409 13 150 15756 14 145 15007 15 150 15556 16 132 13709 17 125 13361 18 107 11114 19 97 10416 20 98 8716 21 102 9765 22 100 9165 23 89 9017 24 91 9017 25 96 10366 26 115 12312 27 117 12612 28 108 12014 29 82 7819 30 88 9018 31 87 8418 32 89 9117 33 95 9516 34 96 9766 35 101 10365 36 105 10814

16. Production and Energy Consumption: Time Series The production data and energy consumption data can be plotted against the week of production to produce a too familiar graph in many plant operating reports. This is a very interesting graph which only serves to confirm the obvious case that energy consumption varies with production. The graph , however fails to quantify the relationship between energy consumption and production , and therefore fails to give us useful information for use in controlling the energy costs.

17. Production and Energy Consumption: A Scatter Plot The plot of consumption against production yields a scattering of points. It is important to note that in a plant/facility various changes could occur over a given production period/season which can impact on the consumption of energy ,either to increase the amount of energy wasted or used for non-productive purposes or to achieve energy savings through an energy conservation program initiated during the production period under review . A common graphical technique in plant operating reports is to draw a best-fit line or simple linear trend line through the scatter of points . Although this can be used in closely controlled production plants in which the scatter of points is very small, it is usually not adequate in production environments in which the scatter of points is not small as a number of changes could have occurred in the plant to produce such a scatter.

18. Best Fit-Line Drawing a best-fit line in such circumstances would simply be trying to average the energy consumption over the production period under review. This can be misleading and may not bring out fundamental changes that have occurred during the historical period under review. There is therefore a need to establish a reference consumption pattern that can be used to compare future performance or performance over the rest of the production period.

19. Baseline This reference consumption pattern is known as the baseline and needs to be established first. During the baseline period a relationship can be established between consumption and production which will then be used as a reference in order to detect if any changes have occurred in the production/manufacturing plant that has lead to the increase or decrease in the consumption pattern. The baseline is by definition the consumption pattern when the rate of consumption was stable, i.e ,before anything has happened to the plant to increase or decrease the energy consumption rate.

20. Baseline This baseline consumption pattern would then be used to compare the consumption rates during the other periods ,so as to see whether there has been an increase/decrease in the consumption pattern per production unit. Further investigations can then be done to establish why there has been this change in the consumption rate. In most cases this investigation can identify some energy conservation/saving opportunities.

21. Determining the Baseline On the Consumption vs Production scatter graph the graph points are identified and marked in chronological order by week starting from week 1 (oldest historical data) until all of the remaining unmarked points are either below and to the right of those that are marked or are above and to the left of those that are marked. In this example that situation reached after the first ten weeks. This implies that something happened after the tenth week which changed the consumption pattern. Therefore the period from week 1 up to week 10 can be used as baseline period. By plotting the consumption against production for this period a baseline relationship can be established between consumption and production which can then be used for comparison purposes with the rest of the production period under review.

22. Determining the Baseline According to the analysis , the baseline relationship between energy consumption and production during the baseline period is given by the equation: Energy Consumption(kWh) = 97.411x Production (T) + 1668.5 However the relationship can be used to predict the energy consumption during the baseline period within the production range covered by this baseline period. It can not for instance be used directly to predict the energy consumption if the production is 20 tonnes or 40 tonnes or even when it is zero because this is outside the production range considered during the regression analysis . Whether the relationship established applies to production ranges outside the real measured production figures used in the regression analysis depends on nature of the production process itself, so this relationship can not be generalized.

23. Predicted Energy Consumption Using this baseline relationship , energy consumption values can be predicted weekly for the actual production over the rest of the production period under review. These predicted consumption figures based on the baseline can be recorded against the actual measured consumption figures and the signed difference between actual consumption and the predicted consumption is recorded on a weekly basis as shown on the table below.

24. Predicted Energy Consumption Week Prod (T) Actual Energy Cons.(kWh) Pred Cons.(kWh) Diff CUSUM 1 80 9669 9461 208 208 2 85 9558 9948 -390 -182 3 90 10527 10435 92 -91 4 87 10043 10143 -100 -191 5 100 11790 11410 380 189 6 110 12198 12384 -186 4 7 115 12837 12871 -34 -30 8 95 10956 10923 33 3 9 100 11320 11410 -90 -86 10 105 11984 11897 87 1 11 130 13453 14332 -879 -878 12 135 14409 14819 -410 -1288 13 150 15756 16280 -524 -1812 14 145 15007 15793 -786 -2598 15 150 15556 16280 -724 -3322 16 132 13709 14527 -818 -4140 17 125 13361 13845 -484 -4624 18 107 11114 12091 -977 -5601 19 97 10416 11117 -701 -6303 20 98 8716 11215 -2499 -8802 21 102 9765 11604 -1839 -10641 22 100 9165 11410 -2245 -12886 23 89 9017 10338 -1321 -14207 24 91 9017 10533 -1516 -15723 25 96 10366 11020 -654 -16377 26 115 12312 12871 -559 -16935 27 117 12612 13066 -454 -17389 28 108 12014 12189 -175 -17564 29 82 7819 9656 -1837 -19401 30 88 9018 10241 -1223 -20624 31 87 8418 10143 -1725 -22349 32 89 9117 10338 -1221 -23570 33 95 9516 10923 -1407 -24977 34 96 9766 11020 -1254 -26231 35 101 10365 11507 -1142 -27373 36 105 10814 11897 -1083 -28455

25. CUSUM Plot There are four points of interest on this graph, as denoted by changes in the slope of the line. At week 10 , something happened that caused energy consumption to decrease, such that over the next nine weeks , a total of approximately 6300kWh of energy was saved. At week 19 , something else happened such that the rate of savings increased; over the next five weeks a further 9400kWh of energy was saved. However , at week 24 ,something went wrong , the rate of savings dropped such that in the next four weeks, only a further 1800kWh was saved. However on week 28 , corrective action was taken and the rate of savings for the balance of the period continued at the rate seen before week 24.

26. Some Reflections The question that needs to be asked is, “ What happened in weeks 10,19,24 and 28 to cause these changes in the rate of energy consumption ?” The CUSUM technique highlights these points in a way that is not evident from the original consumption and production against time plots or on the original consumption against production plot. The CUSUM plots gives us more information which allows for further investigation and therefore possible identification of saving opportunities.

27. Monitoring Performance When improvements have been made in the energy consumption practices , we expect a downward trend in the CUSUM curve. Our interest becomes maintaining those savings , and, therefore, we have a new baseline to use for predicting consumption values. That new baseline would be recent consumption /production data fro which the savings measures are functioning. Regression on those points yields a new functional relationship for the calculation of predicted values in the ongoing CUSUM calculation.

28. Determination of a New Baseline From the CUSUM plot , a new trend in energy consumption was initiated in week 28 with the correction of a measure that had ceased to function. Weeks 29 through 36, the most recent trend up to today, become our new baseline. The procedure for determining the CUSUM as described earlier then repeated as follows: Regression on weeks 29 to 36 to create new baseline Calculation of new predicted values Calculation of weekly difference between actual and predicted , and CUSUM and monitored in an ongoing fashion with each week that follows.  

29. Monitoring Performance Week Prod (T) Actual Energy Cons.(kWh) Pred Cons.(kWh) Diff CUSUM 29 82 7819 7996 -177 -177 30 88 9018 8745 273 96 31 87 8418 8621 -203 -107 32 89 9117 8870 247 140 33 95 9516 9619 -103 36 34 96 9766 9744 22 58 35 101 10365 10368 -3 55 36 105 10814 10868 -54 1

30. Using CUSUM for Control Charts Control charts provide a means for plant personnel to monitor the performance of their systems , andd to take corrective action if needed. The predicted value of energy consumption becomes the “standard”, within which real performance should fall are established. This information is presented graphically in such a way that the operator can plot actual consumption variance values to see whether or not they fall within the control limits, thereby providing relatively IMMEDIATE FEEDBACK ON THE SYSTEM PERFORMANCE. Such control charts should become a normal feature of plant operations

31. Using CUSUM for Control Charts A control chart can be created with control limits from the week 29 to 36 baseline as shown below. Here the plot is variance (or difference between actual and predicted value) against week. Control limits based on knowledge about normal variability in the plant and consideration of reasonable control expectations is set.

32. Using CUSUM to set Reduction Targets In addition to providing a structure for the monitoring of performance, CUSUM provides a basis for establishing energy consumption reduction targets. We are trying to find an answer to the question, “By how much can energy consumption be reduce without adversely affecting production” A number of answers typically arise: We can replicate the best performance achieved in the past as shown by the CUSUM regression analysis of the historical data. Its more like the plant manager saying , “ Gentlemen, our energy efficiency performance was remarkably on week 29 ,can we therefore repeat that performance for the rest of the year!!” Based on the scattering of water consumption against production , we can set a target to reduce consumption by x%, where x% places the performance between current average performance and the best historical performance. We can arbitrarily set out to reduce consumption by y% per year, where y% is a modest and achievable reduction target,- perhaps one that you can easily surpass!!

33. Conclusion Whatever the target is , it now becomes the basis for new “predicted” consumption values for the ongoing CUSUM process, and new control charts. Its selection will be made now in view of a more complete understanding of how the plant has performed historically, and with some control tools in place. This might be a very useful study to undertake in line with the major challenge that we face in terms of energy efficiency performance benchmarking. The energy efficiency benchmarks are not readily available for all our situations and the state of our technology , therefore having localised or company-based performance benchmarks might be a good starting point. Good luck in your endeavors!!  

34. Ad-hoc Approach to Energy Management Structured Approach to Energy Management