Fullscreen HTML5Med Res (??MB)HTML4

Solar PV Update Are batteries worthwhile? By Dr Alan R. Wilson Image source: https://pixabay.com/photos/solar-system-roof-power-generation-2939551/ I have had a solar panel array installed for over 10 years now, and I have a pretty good set of data on how it has performed over those years. As I suspect the generous feed-in tariff will go away soon, I have been considering whether it would be worth adding a battery to the system and, if so, what type and how large. This article describes how my system has performed and the research I have undertaken. I wrote an article detailing my experience with an urban 5kW solar photovoltaic (PV) installation (May 2015; siliconchip.com.au/Article/8555). My array consists of 27 panels mounted on a north-facing roof with a 5.2kW inverter. With the system now around 10 years old, it is an excellent time to revisit the situation and consider adding batteries to the installation. To date, the system has performed flawlessly. A contributing factor might be the shade panel I constructed to shield the inverter from direct sun, mounted on a north-facing wall. One problem that I spotted late in 2019 was the growth of lichen on my solar panels. I caught it early enough because I could remove it using a long pole, a scraper and soapy water. Lichen can be a big problem, and there is a fair amount of ‘chatter’ on the internet about it. Here’s an example of lichen beginning to form on a solar panel. You can find more extreme examples online. 38 Silicon Chip It is critical to ensure algae does not grow on the panels because lichen is a symbiotic partnership of a fungus and an algae; any hint of green on PV panels and it is time to clean it off. The overall performance for the last 10 years is shown in Fig.1, including the daily (averaged) exported and imported energy. Clearly, more energy is exported than imported. There are two points, indicated by the black markers, where the crossover for these curves drops when circumstances changed, and the system started to export more energy. These correspond to when I installed an evacuated tube solar hot water system (1) and when one of my adult children left the house (2). Until a couple of years ago, I was not recording the amount of energy provided by the solar panels themselves. Doing so gives a greater insight into the pros and cons of Solar PVs, and this data can be used to determine whether batteries are a good option or not. Fig.2 shows the solar PV energy generated, the energy used by the household (both of these with a cosine curve fitted to them, see below), and the excess energy which is available for use by the grid, again presented as daily usage. The household uses about 11kWh a day in summer and 18kWh in winter. My house has neither electric heating nor cooling but does have an off-peak electric storage hot water system. Taking one year within this span, the system generates 6661kWh, with 5028kWh used, giving a net yearly excess of 1633kWh. Australia's electronics magazine siliconchip.com.au However, since much of the consumption is overnight, the actual yearly exported energy is 5105kWh with 3471kWh imported. This is currently a good position for me because I am the lucky recipient of the Victorian Government Premium Feed In Tariff which is significantly higher than my usage tariff. My electricity provider (and the taxpayer) pay me rather than me being faced with a yearly bill of $1609 (5028kWh at 32¢/kWh) if I did not have solar panels. This will change in the not-too-distant future, and the question is whether it is worthwhile to install batteries and a new inverter/charger. A bonus would be the capability of independent operation as insulation against power drop-outs, particularly in summer. However, how the house would be disconnected from the grid to allow this is an open question. Rather than diving directly into consideration of a solar PV + battery system, first I will assess the performance of my current PV system, followed by an analysis of commercially available batteries. Comparison with expected performance The Bureau of Meteorology provides a large amount of public data related to many aspects of the climate, including monthly measured kWh/m2 insolation values – see www. bom.gov.au/climate/data/index.shtml?bookmark=203 For December 2019 and June 2020, these values were 6.8kWh/m2 and 2.1kWh/m2 respectively (be careful to select the correct units when looking at the website). My PV array produced an average of 26kWh and 11kWh per day during these two months. With an area of approximately 28m2, this equates to conversion efficiencies of 14% and 19% for December and June respectively. The figure for December looks low, but it is a good demonstration of two effects: 1) the sun passes behind the solar panels, and 2) they run hotter and are less efficient in summer. My north-facing panels can only receive sunlight for at most 12 hours a day, but the sun is up from 5:55am to 8:42pm on December 22, nearly 15 hours (see www. timeanddate.com/sun/australia/melbourne). Considering Fig.3, sunlight before 6am and after 6pm contributes little to the energy received. More important is that the panels run hotter in summer. An increase in temperature from 20°C to 80°C can decrease performance by up to 30%, and 19 less 30% is about 13, as observed. The next parameter I considered was panel placement. Melbourne is 37.8° South, and my roof has a slope from the horizontal of about 32° North. This gives an incident angle to the sun of about 17.6° mid-summer and 29.2° mid-winter (the tropics are at 23.4°). The question is: should the panels be aimed more at the winter sun to gather more energy when it is needed? Fig.4 was obtained at midday a few days before the winter solstice in 2019 for a clear, cloudless sky and a more typical overcast day with around 80% cloud cover and the sun covered. The figure shows the percentage drop from the maximum power received by a small solar cell mounted under glass as a function of the angle with the northern horizon. Here, panels on a flat roof correspond to 0°. The clear day maximum power occurs at 58°. This is very close to aiming directly at the sun (37.8° + 23.4° = 61.2°), as expected. 20° either side of this point (38-78°) decreases siliconchip.com.au Fig.1: the energy I imported (blue) and exported (red) over the last 10 years. The two black triangles indicate the points when circumstances reduced the household’s energy usage. Fig.2: the energy generated by the 5kW PV array, the energy used by the household and the excess energy available for the grid over a 14-month period. The thin curves are fitted cosine functions used for my later modelling. Day d=0 is July 1. Fig.3: the relative strength of the incoming solar radiant power collected by flat PV panels as a function of time of day, at the start of summer (from www.eia.gov/ todayinenergy/detail.php?id=18871). Australia's electronics magazine January 2022  39 Fig.4: the percentage drop from the maximum solar energy detected by a small solar cell as a function of angle (0° corresponding to facing straight up) close to mid-winter for a clear day (blue triangles) and during an overcast day (grey triangles). Fig.5: the results from a simple model based on realworld experience for a 5kW PV array, showing the energy generated, day and night power usage, the energy stored (thin lines) and passing through (dotted lines) a 5kWh, 10kWh and 15kWh battery, and the energy exported and imported. the power by less than 2%. However, on an overcast day, peak power is at 38°. As the tilt angle increases, the solar cell sees less of the sky. The clouds scatter a large proportion of the incoming energy on an overcast day, so less of this is collected as the angle increases. 20° either side of 38° decreases the power by 10% in this case. The two sets of points suggest the optimum mounting angle is somewhere in the range 30-43°. This angle improves collection on overcast days but has a minimal effect when there is little cloud cover, being within the 38-78° range for a clear day. Fig.4 gives an idea of how this will affect the operation in summer. A panel angle of 30° maps to 76.8° in Fig.4 for a clear sky in summer, resulting in less than a 2% decrease in power. The figure implies it is best to keep the effective summer angle to less than 80°, suggesting that the ideal angle to mount panels facing North is 30-34° (80° - 46.8°). My panels are at 32°, so I do not need to do anything. Remember, this is for Melbourne. Further North, I expect the ideal angle to be lower, with 0° best at the equator. Thus Brisbane at 27.5° South would have an ideal panel angle of around 22-25°. The thin red line in Fig.2 is a good-looking cosine curve fit to the energy generated: 18.24kWh + 6.79kWh cos(π + 2π × [d + 15] ÷ 365) where d is days from July 1, and the peak occurs on December 16. We can use the same approach to give another fit to the energy used: 13.98kWh + 2.73kWh cos(2π × [d + 4] ÷ 365), with the peak occurring on June 27. Determining the total energy used is only half the solution; it must be split into day and night contributions to assess the flow to/from a battery. Unfortunately, my situation is complicated by the solar evacuated tube hot water system. In summer it uses no electricity, while in winter it uses off-peak electricity at night. It is preferable to divert the day-generated power into the hot water system before charging the batteries. To include this in the model requires an estimate of the energy used by the hot water system. My summer drop in consumption after the installation of the solar hot water system is 4.8kWh. But I decreased the water temperature by 10°C, which in a 250L tank corresponds to 0.7kWh, reducing this to 4.1kWh. Ignoring the slight offset between the energy generated and energy used, this leaves 12.6kWh (16.7kWh less 4.1kWh) used in winter compared to the 11.3kWh minimum use mid-summer. In Melbourne there are 9.5 hours of daylight mid-winter and 14.5 mid-summer. After removing the contribution by the hot water heating, the simplest thing to do is apportion the energy use according to the number of daylight hours. Yes, lights are on at night and not during the day, you might object. However, people sleep at night and use other electrical devices during the day. I am assuming these roughly balance. Table 1 shows the expected peak and trough (mid-winter and mid-summer) energy consumption figures after allowing for the number of daylight hours and shifting hot water heating to the daytime in winter. Because the model is a simple sinusoidal wave with a constant offset and known period and phase, the maximum and minimum are all that is required to determine the waveform. The estimates in Table 1 give: Are batteries worthwhile? When I lose my Premium Feed In Tariff, I am considering adding batteries to the system. With a battery, I could store energy during the day and use it at night, rather than exporting it during the day and importing it at night. But will that be worthwhile? The model I developed to analyse the financial aspect is based on real-world experience and can determine the optimal battery for a solar PV array. To do this, we need to determine how much energy is generated, how much is used when the sun is up, how much is drawn overnight from the battery and how much is imported and exported. A simplistic model for energy usage over the year is to assume it is due to the variation of daylight hours, and use a cosine function with a winter maximum and a summer minimum. 40 Silicon Chip Australia's electronics magazine siliconchip.com.au Fig.6: the modelled yearly operational income (negative is a cost) versus battery capacity for 1.5kW, 3kW, 5kW and 8kW PV arrays (red lines), and a 6kWh battery against PV size (blue line, same horizontal axis). It is clear that large batteries do not pay for themselves, but more PV panels do. The 0kWh point corresponds to PVs with no battery. Fig.7: a similar graph to Fig.6 but with air conditioner (A/C) and heating loads included for 5kW and 8kW PV arrays. As expected, due to the greater energy use, the yearly income decreases. The useful battery capacity also increases due to the larger throughput. Night use = 6.0kWh + 1.6kWh cos(2π × [d + 4] ÷ 365) Day use = 7.95kWh + 1.15kWh cos(2π × [d + 4] ÷ 365) Again, d is days from July 1. The power available to be stored in the battery is the PV energy less the day use, plus any not used the previous day, but only up to the battery’s capacity, the rest being exported to the power grid. The power used overnight is simply the night use figure. This is most easily calculated using the above expressions in a spreadsheet. We can also use the spreadsheet to determine the yearly kWh throughput for the battery, the amount of energy imported and exported and thus the annual cost of operating the system. This spreadsheet will be available to download from: https://alanrwilson.com/solar-batteries/ As a reality check of the model, the predicted total solar PV energy generated is 6658kWh, within 1% of the measured 6661kWh, and the total night + day consumption is 5092kWh, within 1.5% of the measured 5027kWh. Fig.5 is an example for a 5kW PV system using the models above. While it looks complicated, it encapsulates the results from the model and gives some immediate insight into the effect of battery capacity. The determined charge at the end of each day (Battery Charge) and Battery Throughput for 5, 10 and 15kWh batteries show that all the charge is used in winter due to the low PV energy available, irrespective of battery capacity. The energy Exported and Imported is only shown for the 10kWh battery; however, the graphs for all battery capacities are essentially the same, with a decrease in the total annual values of 147kWh exported and 127kWh imported moving from the 5kWh to the 15kWh battery. The amount of PV energy available is the governing factor, not the size of the battery. The only advantage of a bigger battery is for energy storage in case of blackouts. For my situation with a 5kW PV array, this suggests that the optimum battery size, including some latitude for estimation errors and a decrease in capacity, is about 7kWh. That’s assuming such a battery can deliver the power required. The ongoing operating costs can be calculated from the imported and exported energy figures. Fig.6 shows how the modelled yearly operational income (negative is a cost) varies with battery capacity for 1.5, 3, 5 and 8kW PV arrays. These are calculated with a feed-in tariff of 12¢ and a usage tariff of 32¢. Clearly, large batteries will not pay for themselves, but larger PV arrays do. Any battery larger than where the curves flatten out is not a good financial investment. It is also apparent that batteries are not helpful with a small PV installation: there is simply not enough energy to store. The variation in cost to operate a 6kWh battery with PV size is also shown (in blue) and more dramatically demonstrates the significant increase in income with a bigger PV array. Essentially, all the extra capacity of the larger PV array is generating more revenue in summer. This is for my situation and does not include heating or cooling, which is common in many houses. The spreadsheet can be modified to include both these cases, modelled as a half-sinewave with a start date and end date (zero to zero). Fig.7 is like Fig.6 but includes a hypothetical 3kW air conditioner, operating (cooling) for a peak of 0.6 hours during the day, 1.4 hours at night, from November 1 to April 1, using a total of 580kWh and 1590kWh of heating using a 3kW heat pump for one hour during the day and 2.9 hours at night, from April 1 to October 30. These figures are based on Law et al., “Energy consumption of 100 Australian residential air conditioners”, Ecolibrium November 2014. What has not been included is siliconchip.com.au Table 1 – estimates of energy used to split between day and night Winter peak (12.6kWh total) Summer trough (11.3kWh total) Day Night Day Night Portion of use 5.0kWh 7.6kWh 6.8kWh 4.5kWh Hot water +4.1kWh Total 9.1kWh 7.6kWh 6.8kWh 4.4kWh Energy used Australia's electronics magazine January 2022  41 the possible impact of limited power delivery by the battery, which would result in higher Imported energy costs. As expected, the operating costs increase, but it again is apparent that the size of the PV array is the most critical factor. With the extra energy throughput, the point at which any increased battery capacity has no effect is moved a little to the right. All of the operating costs above are well below the $1629 (my house) or $2324 (with heating and cooling) figures for PV or batteries at 32¢/kWh, and some even generate income. But what is missing is the up-front cost. Most systems on the market have a 10-year warranty. $25,000 spent on a system that fails after 10 years effectively costs $2500 a year just to cover the purchase price (ignoring opportunity costs associated with not having that $25,000). If the system does not generate a positive return, it will be more expensive than simply pulling power off the grid. The following sections consider a range of battery technologies, their pros and cons, how much they cost and whether they will pay for themselves. Battery choices Lithium-ion based batteries currently dominate the solar/ renewable energy market; however, one of their claims to fame (lightness) is not a consideration for static installations. Still, a quick survey of commercially available batteries offered for solar PV installations yielded 36 lithium-ion, one graphene super capacitor solution, one flow battery and one lead crystal battery. The faithful lead-acid battery is not considered in the race, primarily due to a low cycle lifetime. Nickel-iron batteries are another old and proven technology. They are robust and long-lived, but I will not consider them because they suffer from a number of disadvantages, including the evolution of hydrogen gas, low efficiency, low charge/discharge rates and a wide operating voltage needing special inverters. Irrespective of the type of battery, the parameters I consider important are: 1. It needs to have sufficient capacity for the requirement. This is rather obvious but beware, usable storage can be significantly less than nominal storage, and overdischarging a battery can significantly impact its useful life. The results in Figs.6 & 7 must also be kept in mind to not needlessly over-specify the battery capacity. A bank of A602 2V gel cells which is used to store energy from a 4kW solar array. Source: www.flickr.com/photos/ stephanridgway/14141342129 42 Silicon Chip 2. Many batteries suffer from a drop of capacity with use. Some warranties are for 10 years but at 60% of the original capacity. Other manufacturers using the same base technology make no upfront mention of reduced capacity. It is worth checking the fine print in the specifications; it might be that the initial battery capacity must be over-specified to ensure it is fulfilling the requirements at end-of-life. 3. The ability to deliver the power you need. High capacity does not necessarily mean the battery can deliver enough power. An 8kWh battery is not as useful if it can deliver at most 3kW and the household needs 6kW peak to run, say, an air conditioner (3kW) and cook dinner (3kW) at the same time. However, if peak powers are transient, it is worthwhile pulling power from the grid for a short time rather than installing a big, expensive battery. Off-grid use requires a big battery and/or a really smart energy management system that prioritises certain circuits and/or a lifestyle change. A small petrol or diesel-powered generator is not very expensive and could be a solution in these cases, if low available solar energy is a transient or rare event. 4. Warranted life: do not accept a battery with less than a 10-year warranty. This is really warranting the construction quality, the ‘nuts and bolts’, not the storage elements. 5. Warranted throughput: the lifetime warranty is usually expressed as a time or maximum kWh throughput with some capacity drop at the warranted kWh. Warranted throughput indicates the amount of energy the battery can store and deliver; it is directly linked with the gradual degradation of the chemistry/physics of the storage system. The yearly battery use needs to be determined to give the expected 10-year kWh input/output required from the battery. Ideally, the 10 years and the maximum kWh occur at much the same time. With ‘normal’ use, most modern batteries will have enough throughput to last 10 years. 6. Round-trip efficiency: this is an indication of the efficiency of energy storage and retrieval. Lithium batteries are generally in the 95-97% range. Some others are as low as 80%, which means more solar panels are required to compensate for the lost energy, but they may have other beneficial properties. 7. Off-grid capability: if you want it. Standalone batteries can be used anywhere, but some batteries come with integrated inverters and/or chargers, affecting how they can be used. 8. Compatibility: batteries and any associated parts of the system included with them must be compatible with the other elements of the system. One area to watch for is the different solar PV panels available. Some (which used to be the norm) are an array of solar cells connected in series string(s) to provide a high DC voltage. Others, which have some advantages, include micro-inverters that manage the power from each individual solar panel and generate the AC at the PV panel itself. These are all connected in parallel and provide a (nominal) 230V AC. Australia's electronics magazine siliconchip.com.au Table 2 – comparison of four battery types suitable for solar PV storage Lithium-ion, NMC (Nickel Manganese Cobalt) Lithium, LFP (Iron Phosphate, LiFePO4) Flow cell (FC) Super capacitor hybrid battery (SC) Robustness Fair Good Excellent Very Good End-of-life Capacity 60% 80% 100% 85% 27-32MWh 32-36MWh 36MWh 36-45MWh Round trip efficiency ~95% ~95% 80% >96% Available power per 10kWh 4-6kW 4-9kW 3kW (5kW peak) 13kW (33kW peak) Maintenance Requirement None None Needs period full discharge. None $8-10k $8-10k $13k $12k Energy density (lower weight). Cycle life. End life capacity. More Robust than NMC. Very robust. Full discharge. No drop in capacity. Very robust. Full discharge. No drop in capacity. (Projected long life) Large capacity drop over life. Medium capacity drop over life. Low efficiency. Low power. Mechanical pumps. Liquids. Maintenance. New technology. Warranted Life: throughput per 10kWh Cost per 10kWh Advantages Disadvantages Table 2 has details for two leading lithium-ion battery technologies and two other more novel technologies. Some of the figures presented have been factored up or down from quoted values to compare hypothetical 10kWh batteries. The cost is for a bare battery, and with the fluctuations in exchange rates and the rapid progress being made, these could very well be wrong by the time this article is published. The two common lithium battery technologies are very similar. Both technologies suffer from a gradual drop in storage capacity, with the LiFePO4 outperforming the NMC type. LiFePO4 also has higher lifetime energy storage, may deliver higher power, is a little more robust (especially if heavily discharged) and is considered safer. Both batteries may have prolonged life by reduced discharge, say to 50%, but then a larger capacity battery is needed to ensure sufficient energy is available for the requirement. They both offer good power delivery but could be challenged in an all-electric house over summer with air conditioning. They are reasonably mature technologies with a lower price than the other two. Flow cell (FC) technology is effectively like reversible electroplating. For instance, zinc-bromide systems plate out zinc in a reversible process. The FC battery is very robust, can be discharged entirely and holds its full capacity through life. Unfortunately, the round trip efficiency is only about 80% (10kWh is required for 8kWh to be later supplied), and FC has a more complex maintenance regime requiring a regular full discharge. Along with a low available power and higher cost than lithium-ion batteries, it is probably not suited for domestic energy storage. Super capacitor (SC) hybrid battery technology is much more interesting with a very high available power and high efficiency. It is like a hybrid between a super capacitor and siliconchip.com.au a lithium battery. Not shown in the table is the high charge rates that are possible. Like the FC battery, it is very robust, can be completely discharged and holds most of its capacity through life. Performance curves suggest 98% after 10 years; however, the warranty still only guarantees 85% capacity. It can also provide enough power for pretty much any domestic use. The only downside is that it is a relatively new technology. Still, I will keep my eye on it over the coming years (especially since the company involved is based in Melbourne). While more expensive than lithium-ion batteries, the projected life, as opposed to warranted life, is well beyond the 10-year warranty, which could make it cheaper in the long term. The other costs besides the batteries and solar panels are the inverter/chargers. Inverter/chargers usually include a battery management system and load management, making them more expensive than a grid-tied inverter. For comparison, I am allocating these a cost of $2000/kW. I decided to investigate 5kW and 8kW PV arrays and batteries with two primary requirements: 1. At end-of-life (ten years), they must be able to deliver 5kW. The minimum capacity to achieve this is 10kWh for NMC, 7.7kWh for LFP and 3.8kWh for SC, 3.8kWh. The NMC and LFP have a quoted spread of values, so I used the means. 2. The optimum capacities indicated in Fig.6 for my application, 4kWh for 5kW PV and 6kWh for 8kW PVs. Based on Fig.7, 7.5kWh must be available at the endof-life for either a 5kW or 8kW system for the inclusion of both air conditioning and heating. Combined with the first requirement and the capacity drops specified in Table 2, the relevant batteries for my theoretical situations are shown in Table 3. Australia's electronics magazine January 2022  43 Fig.8 shows the 10-year cost for the 5kW PV array, while Fig.9 is for the 8kW array. The cost with time is simply calculated as (up-front cost) + (operating cost) × years. The results are simple straight lines; however, a visual representation of the slopes and intersections gives a quicker comparative insight than numbers in tables. Both figures include the ongoing cost with No PV and the relevant PV arrays with no battery. The systems are priced at $1500/kW in the no-battery case due to the cheaper inverter required. It is clear from both figures that the most cost-effective course for the first 10 years is to not use batteries due to their high up-front cost. Both figures also clearly indicate that the payback rate is mainly independent of the battery capacity, with the average cost over time highly dependent on the up-front cost of the system and the size of the PV array. Both figures indicate that the PV arrays without a battery system pay for themselves after ~5-6 years. The batteryinclusive systems would eventually return more than the no-battery systems; however, this takes at least another 16 years in the best case, well outside warranty periods. While the 8kW PV array with no battery takes longer to pay for itself than the 5kW PV array, it pays back more, becoming superior to the 5kW PV array after about nine years. Given that the panels should last 20 years, it is better to install the higher capacity in the long run. In both cases, the payback is earlier with the higher consumption cooling and heating case simply because more energy is being used. If heating and cooling are included, the 5kW array can never generate income, whereas the 8kW array can. Throughput All of the above considers the 10 years life rather than throughput. The modelled 10-year throughput for the NMC, LFP and SC batteries are all much the same at 18MWh for the 5kW array and 28MWh for the 8kW array. These are likely to be on the low side since it does not consider times Fig.8: modelled total cost for a 5kW PV array in my situation (solid lines) and with cooling and heating included (dashed lines) with the NMC, LFP and SC battery capacity as indicated. The No PV and No Battery cases are included to show payback times. A negative slope indicates income. 44 Silicon Chip Table 3 – minimum battery capacities for two PV arrays with and without heating/cooling 5kW, no AC 5kW, with AC 8kW, no AC 8kW, with AC NMC 10kWh 12.5kWh 10kWh 12.5kWh LFP 7.7kWh 9.4kWh 7.7kWh 9.4kWh SC 4.7kWh 8.8kWh 7.0kWh 8.8kWh when the PVs become shaded during the day and will provide power from the battery. However, even allowing for a 33% increase to 24MWh and 37MWh, all of the batteries should be able to provide this, although 37MWh exceeds the NMC specification and is close to the LFP specification. If throughput is the critical ageing parameter then, provided the other mechanical and electrical systems do not fail, the lithium batteries for a 5kW PV array could have another five years or so of useful life left, and the SC around seven years. This increases the cost-effectiveness of the systems and making them all a sound financial proposition. For a 10kWh battery, the effect of a battery on imported and exported energy is shown in Fig.10. The squares are the results from my 5kW Solar PV with no battery. Doubling the size of the battery is not helpful, as shown before, and decreases the imported and exported energy by a miserly 10kWh. Power grid stability One of the complaints against solar PV is the wild fluctuations in available energy that can occur when the sun is, for instance, suddenly shaded by clouds. A smart energy management system could be implemented whereby stored battery energy is available to smooth out these fluctuations. This is the essence of the Virtual Power Plant concept, and there are some companies already doing this and making profits by feeding battery stored energy into the grid when the spot price is high. The Australian Capital Territory is Fig.9: model results similar to Fig.8 but for an 8kW PV array. Australia's electronics magazine siliconchip.com.au An example of a solar panel setup, the smaller panel along the wall is for a solar hot water pump. planning to implement this strategy in a distributed network of battery storage in the Territory. Going off-grid If the desire is to go off-grid then no energy can be imported. With everything electrical, the model for my situation indicates a 16kWh battery is required to cover the nightly use, with a 13kW PV array to fully charge it in mid-winter. Using the SC battery, this will cost around $45,000. That sounds like a lot, but it might not be too bad if you have to pay $20,000-30,000 to have power cables laid to a remote location and then have to pay for the connection costs and electricity. The combination of battery and PV array assumes there is always average sunshine and does not allow for cloudy days. The problem is the daily use is 28kWh; it rapidly becomes very expensive trying to install enough batteries to cover the occasional 2-3 day overcast period. A more cost-effective method is to use a small generator. Fig.10: imported and exported energy values from the model calculated with 3kW, 5kW and 8kW PV arrays and a 10kWh battery. The black and red squares are the real-world results from my 5kW system with no battery. A 3.8kW PV array is enough to generate all the power for the household over a year; the problem is moving the excess energy from summer to winter. siliconchip.com.au These are reasonably cheap (3.5kW for around $500) and can be run long enough to charge the battery when needed. But personally, I do not like this option in an urban setting. Conclusions Investing in Solar PV by itself is definitely worthwhile, with a suggested payback time of fewer than six years for the 5kW, north-facing installation considered here. From a purely financial point of view, batteries are still too expensive, except possibly for the newer SC batteries with their potentially longer life and their high power-to-capacity ratio allowing the use of a smaller, cheaper battery. But adding batteries does reduce the dependence on grid power and, with the right management systems, should help reduce power supply fluctuations from renewable energy. This is something which will need to be considered in the future. A network with a large number of (highly variable) solar PV-only generators that simply attempt to deliver as much power as possible to the grid will become unstable. The model shows that increasing the allowed household installation of solar PV to 8kW with a suitable battery backing it up can bring a house in Melbourne close to self-sufficiency. The situation will be even better closer to the equator. Keep in mind that some batteries have failed Australian tests, and some companies have failed too, and are not around to honour warranties. Doing business with mainstream companies and suppliers in this relatively new market is probably advisable. Choosing a great-sounding, cheap deal could very well leave you with expensive boxes that do not function. So what battery would I recommend? From a purely financial point of view, none at all. However, there might be other reasons for installing a battery. If you must have a battery and are on the conservative side, go for the LiFePO4 option from a reputable source, but be aware that the power requirement may be the governing factor. If you are more of a betting person, look very seriously at the ‘super capacitor’ option with its ability to deliver high power with a relatively low capacity. There is also the hope the SC technology will live up to its promise of a superior lifetime. And overall, keep the information in Figs.6 & 7 in mind; big batteries are definitely not worth it unless you are going off-grid. SC Australia's electronics magazine January 2022  45