ACR Journal

February | March 2023 EFFICIENCY 16 Volume 9 No.2 incoming temperature or managing door openings more carefully require behaviour change, perhaps with no capital investment. That doesn’t mean that it is easy; changing operating habits is sometimes very di cult, but it is possible without high investment cost. An interesting consequence of this understanding of the factors that make a facility “best practice” is that the best practice metric is not heavily dependent on the location of the site or the weather. Compressor overhaul The third element of a good benchmarking programme is where it gets really interesting. Recent work by Star Refrigeration has shown that with daily recording of energy data and some simple manipulation can enable a relatively accurate prediction of SEC to be made, even from a very small data set. This means that changes in energy performance can be identified quickly. If they are the result of a maintenance intervention, for example a reduction in energy use following a compressor overhaul, then the annualised benefit of the work is easy to calculate. On the other hand if they are an unplanned and unexpected change for the worse then they may give early warning that something is amiss and needs to be identified and fixed. In this case the annualised benefit of the early warning can also be quantified, by calculating the additional running cost that would have been incurred until the problem was identified and addressed. The full year SEC can be calculated from the sum of 365 historic daily readings. With 270 readings (three quarters of the year) it is possible to fill in the blanks and make a prediction of the SEC in 95 days’ time. With 90 readings a prediction of SEC in 275 day’s time can be made and with as few as 10 consecutive readings a prediction of full year SEC in 355 days can be made. More readings in the calculation make the prediction more accurate but even with only 10 readings the estimate is usually well within +/-30% and this is usually good enough to show a change in the energy use trend. This is illustrated in Figure 3 which shows the data for a cold store. The long term prediction, the blue line, is based on the last 10 energy readings and shows an increase in annual SEC quickly when something changes. It is very responsive but not very accurate. The green line is based on 90 readings and so is more accurate but slower to react. The red line is the full year figure, updated on a daily basis by adding the most recent reading to the data set and dropping the oldest one. All that the operator needs to do to get a sense check of the health of his system is to see whether the blue line is above or below the red line and which direction it is going. Benchmarking, even the predictive type, can only tell you how good or bad your facility is or is going to be. The identification of suitable remedial action is sometimes di cult, even when it is known that the operation of the plant is worse than comparable facilities or is poorer than it used to be. In this regard however it is far better to know that there is an issue to be identified rather than wondering whether it is acceptable or not. This is like the di erence between “is there a needle in that haystack?” versus “there is a needle, and your job is to find it”. More to the point, even once the first needle has been identified it is helpful to know whether there are any more to be found. Taking corrective action and continuing to benchmark can answer that question relatively quickly. Various more or less sophisticated tools can be used to find the needles. Detailed performance assessment on complex refrigeration plant is di cult because each component interacts with the others, with the load required and with the ambient conditions. Often design performance information is only available for the “design condition” of maximum load on the hottest day, so underperformance goes undetected when the load is light, or the weather is cooler. It is therefore necessary to build a picture of “what good looks like” under all operating conditions and over a longer period of time. It’s not feasible to have a skilled technician watching how the plant responds for weeks on end so some form of automation is required, probably involving o -site analysis of the data. Automated dashboards The days of judging cold store performance by considering only whether the room is cold enough to satisfy the QA inspectors should be ancient history by now. Temperature compliance should be a given, but the question of whether it is achieved e ciently or not is still di cult to answer. There is a need to develop a desire for energy compliance alongside the temperature compliance regime. In this regard, any form of regular record keeping and analysis will be beneficial because it helps to create a picture of normal operation. This is helpful in justifying plant improvement initiatives and also in quickly identifying when something changes for the worse. Basic benchmarking is not expensive to implement but it can be costly to make good use of the data collected. The use of automated dashboards and analysis calculators may seem to be a more expensive path but often these initiatives pay for themselves in the savings earned and even when there is no obvious immediate improvement made, they can be considered to contribute to the financial security of the client organisation by demonstrating unequivocally that performance remains good and highlighting opportunities to make it even better. Figure 3 – Predictions of Specific Energy Consumption for a Cold Store 10 consecutive readings a prediction of full year SEC in 355 days can be made. More readings in the calculation make the prediction more accurate but even with only 10 readings the estimate is usually well w thin +/-30% and this is sually o d enough to show a change in the en rgy use trend. This is illustrated in Figure 3 which shows th data for a cold store. The long term pr diction, th blue line, is based on the last 10 energy readings and shows an increase in annual SEC quickly when something changes. It is very responsive but not very accurate. The green line is based on 90 readings and so is more accurate but slower to react. The red line is the full year figure, updated on a daily basis by adding the m st recent read ng o the data set and dropping the oldest one. All that the operator needs to do to get a sense check of the healt of his system is to see whether the blue line is above or below the red line and which direction it is going. <caption>