The Autonomy Tool and Direct Drive Systems

Dear Steve, After further consideration and some computer simulations we have come to the conclusion that the autonomy tool is not appropriate to use for direct drive systems. Analysis of a direct drive requires hourly data to analyze the performance. The autonomy tool was developed using data for total daily insolation (solar radiation), not hourly data. Currently almost all, if not all, direct drive refrigerators incorporate a Danfoss compressor. This compressor requires about 60 watts to start and consumes about 60 watts when running. If the output of the solar array is less than 60 watts the compressor will not run. If the output of the array is greater than 60 watts the compressor will still consume about 60 watts. With a battery powered system the energy addition to a battery is directly proportional to the intensity of the incident insolation; the energy accumulated in a battery is proportional to the total daily insolation. This is not true for the direct drive refrigerators incorporating Danfoss compressors. Currently the number of hours a day a direct drive refrigerator will operate depends not only on the total daily insolation, but on the distribution of solar radiation during the day. Two days could have the same total daily insolation, but the number of hours the compressor operates could be different. That is the primary reason daily total insolation cannot be used to calculate the run time of the compressor. Hourly data is difficult to obtain for developing countries. However, we found one station in Illorin, Nigeria through the WRMC where data is recorded every 3 minutes. The data is recorded at the University of Illorin through their Physics Department. We used data from Illorin to illustrate the importance of knowing the daily temporal variation in insolation. Two days were compared, both received 3.2 wh/m^2/day of insolation. We assumed the solar array connected to the refrigerator was 140 watts and the compressor required 60 watts to run and start, as represented by the horizontal line on the attached graph. As illustrated in the accompanying graphs, the compressor on 8/20/93 ran for 1.45 hours and the compressor on 7/23/93 ran for 2.66 hours, which is 1.8 times longer. When insolation levels are low the difference in run time for different days which have the same level of daily insolation could be larger. This is all when accuracy is most important to insure reliability. Total daily insolation data could be used for a direct drive system if the speed of the compressor could be varied so that the compressor power requirements range from 0 to 140 watts. At the beginning of the day the motor would spin slowly then midday at full sun it would spin at full speed and consume all 140 watts. With a battery powered system if the batteries are not full almost all the power produced by the array can be captured under both full sun and cloudy conditions and hence hourly data is not needed. In conclusion to determine the number of hours per day the compressor in a direct drive system will run requires a method of analysis which incorporates solar data showing the temporal variation of incident solar radiation during the day. The autonomy tool uses for its analysis only total daily insolation and cannot be used to determine the performance of a direct drive system. Attached Graph: Solar-Insolation-Graphs.pdf

I read the recent test report on a direct drive refrigerator installed in Vietnam. The refrigerator was powered by an array which is at least 360 watts and the minimum power required to run the compressor is about 60 watts. With this size array a compressor should run when the incident sunlight is over 160 watts/m^2, or about 1/6 of full sun conditions. The refrigerators installed in Northern Vietnam maintained appropriate temperature when total daily insolation was very low. During the period of low isolation from 12/29/11 to 1/12/12, data is presented for total daily insolation. With this data there is no way of predicting how many hours the compressor would run per day. To determine the runtime each day, data showing the variation in insolation during the day would be required. For example, the total insolation for a cloudy day may be about 800 watt hours/m^2/day (a sunny day is about 6000 watt hours/m^2/day). For this 800 watt hours/m^2/day there is no way of telling if the insolation did not peak out over 160 watt/m^2 for several hours allowing the refrigerator to operate, or if the insolation stayed below 160 watts/m^2 all day not allowing the refrigerator to run. Either scenario is possible. The autonomy tool would not help size the system because it only relies on total daily insolation and could not be used to determine compressor run time. Field test data does not guarantee the performance of the refrigerator at a new location even if the values of total daily insolation are the same. One location might have uniform cloud cover all day and a second location may have a clear morning followed by heavy clouds. The run time would probably be different in both cases even though they have the same total daily insolation. If temporal data were available the performance of the system could be more accurately analyzed. With a more accurate performance simulation for a given probability of failure, the size of the solar array could be reduced. If temporal data is not available perhaps a hypothetical solar day could be modeled which would represent the worst case solar distribution. For the tropics this day could be 10 hours, but the hourly distribution of insolation would be different than the current PQS test day. If the system were then designed for this worst case distribution, the system could work at any location with the same total daily insolation.

Firstly I’d like to agree that it’s impossible to accurately model the performance of a Solar Direct Drive (SDD) refrigerator given only total insolation for a site. Furthermore the published Autonomy Tool is helpful only in providing guidance. The biggest danger for SDD refrigerators is that solar energy companies start treating them in the same way as a conventional, battery powered systems. Many times I have been asked for ‘Daily Power Consumption’ figures and many times I have refused to give them. Although commonly used in ‘normal’ solar power systems, power consumption is highly misleading in this context and as it is not directly linked to the size of solar array required for SDD applications. This is largely a result of the ‘threshold’ of power needed to start a compressor. Below this threshold all energy generated is wasted unless the compressor has already started. I have to disagree on the cut-in threshold stated of 60W. This is not necessarily the case and it’s misleading to make this assumption. The power (the product of current and voltage) required depends on the design of the refrigerator. What must also be considered is what happens when the light levels increase. Some designs are able to do more with increasing levels of light. Variables that determine the energy that can be used include refrigerant choice, refrigerant charge, evaporation temperature, condensor temperature, compressor controller settings, ancillary control and technology combinations. Added to this there’s a wide range of solar modules that claim the same power output. This is by no means straightforward. Low insolation can be an indication of lower ambient temperatures. This is often but not always the case. Solar Direct Drive refrigerators therefore need to be safe over a wide range of ambient temperatures if they are going to perform to the levels a secure vaccine cold chain requires. The ability of the solar direct drive refrigerator to store energy and control temperature is crucially important. A conventional battery powered fridge stores energy in the battery and uses the conventional electronic control system to maintain temperature. Both of these functions need to be adequately covered by the SDD refrigerator. Too often manufacturers have focussed on the battery replacement (the easy bit) and ignored the need to control temperature given no power and over a wide range of ambient temperatures (+5 to +43C for instance). As a result a number of designs are likely to fail in field conditions. Part of the reason why the study in Vietnam took in both Northern and Southern sites is because they were known to be so different. Hot and sunny in the South and long periods of overcast weather in the North. What has been encouraging is that during the exceptionally poor weather experienced early this year, the True Energy refrigerator faithfully maintained correct temperatures better than an expensive mains powered fridge would have done. It is the Sure Chill technology that’s inside the True Energy BLF100DC that means it’s able to control internal temperatures so well whilst enabling safe storage of energy (evidence the 10 day holdover at 43C of the BLF100AC) and delivery of unsurpassed levels of performance There is no way yet devised to accurately model all SDD refrigerators. They are all at least a little different, often a lot. Hourly data for a site is unlikely to be available so guidance on product selection should be based on the best available data. A good dose of common sense needs to be used in using published insolation (sunshine) data for a site along with local knowledge. In time a variation of the autonomy tool could be developed to aid solar power system sizing for SDD refrigerators but this will need to take account of the widely varying running characteristics of the devices. In the meantime it is necessary to either use the WHO published array size or get an approval for a change from WHO after consulting the manufacturer. It is important for WHO to be involved in this loop as claims by manufacturers need to be independently checked. To increase the levels of comfort that freezing will be avoided and that safe storage temperatures will be maintained we would suggest selecting the refrigerators that are designed to operate in the lowest sunshine levels as well as those with greatest ambient temperature tolerance. It is also advisable to avoid all chest type refrigerators as these are prone to the freezing of vaccines. The Sure Chill technology was initially developed specifically for the Solar Direct Drive Vaccine refrigerator market and is now inside two PQS pre-qualified products – True Energy BLF100DC E003/019 and the True Energy BLF100AC E003/013. The technology is significantly different from technologies used in other vaccine refrigerators, the benefits include; consistent temperatures between 4C and 5C throughout the compartment, very long holdover times and guaranteed no freezing of vaccines – even in the cool down period (an issue with ILR/Chest formats). Ian Tansley CTO True Energy Ltd & Inventor of the Sure Chill Technology

If a solar designer tried to use the Autonomy Tool to size a True Energy BLF 100DC they would have a number of obstacles. When applying the autonomy tool the first step is to calculate the size of array needed to power the refrigerator with the insolation equal to the average insolation on the cloudiest month. To make this calculation the refrigerators energy consumption is needed, for good reason Ian often “refuses to give them” because normal methods of analysis for battery based systems do not apply to direct drive systems and the PV array may not be properly sized. The autonomy tool is one of those methods which are designed for battery based systems. The next step in sizing a system with the autonomy tool is to examine a table which yields the array size for a given number of days of autonomy; the larger the number of days of autonomy the smaller the required array size. Ians’ BLF100DC has about 10 days of autonomy. The table presented by WHO for most African locations provide sizing data primarily for 3 days of autonomy. The table could be extended to show 10 days of autonomy. However, it would show an appropriate solar array size for a battery powered system. This would be several times smaller than what is needed to run a direct drive refrigerator. In other words, it is difficult to apply the autonomy tool to direct drive systems, and the results obtained would be incorrect.