By Joe Jancauskas, Senior Electrical Engineer at Castillo Engineering
While moderately oversizing your solar panel cables can ensure fire safety as well as enable you to meet your voltage drop criteria, vastly oversizing your cables and strictly adhering to a voltage drop mandate could be unnecessarily reducing the long-term profitability of your solar projects. In this second part of our PV cable sizing series, we take a look at why exactly PV cables are so oversized and how you can better calculate cable sizes to ensure safety while also maximizing project returns.
Why are PV cables so oversized?
As a primer for understanding the reasoning behind why cables are so oversized, you should be aware that the dc input wiring to the inverter is generally split into two terms by National Electrical Code (NEC): the PV string wiring is referred to as the “photovoltaic source circuits,” while the output wiring from the combiner boxes is referred to as the “photovoltaic output circuit.” If a recombiner is utilized, its output wiring is referred to as the “inverter input circuit.”
First, part of the reason why PV cables are so oversized starts with the fact that the NEC assumes that PV is a continuous load. This is often a conservative assumption since a variable power source like the sun is not often at full output for more than three hours per the NEC definition of continuous load. Second, in addition to the normal 125% sizing factor for continuous loads, an additional 125% sizing factor is added to account for PV output occasionally being greater than nameplate for those rare/lucky irradiance and temperature combinations that are better than Standard Test Conditions, for a resulting 156% sizing factor applied to the full load current of the photovoltaic output circuits.
This solar cable sizing in accordance with the National Electrical Code (NEC) is what my colleague Vince termed “the great copper conspiracy.” For those of you that enjoy conspiracy theories, it is as if the copper suppliers banded together and then influenced enough NEC members to make 156% the default factor for PV in order to boost the sales of copper. This would be a collusion similar to the real-life Phoebus Light Bulb Cartel in the 1920s and 1930s, which by many accounts existed solely to drive up profits by standardizing light bulb life at a relatively low 1,000 hours. This cartel collusion also involved fines being issued to manufacturers whose lamps exceeded 1,500 hours of life, thereby creating lucrative product demand through wide-scale planned obsolescence. More information and links about the cartel can be found here.
Another reason why PV cables are significantly oversized is because PV module ratings are based on 1,000 W/m2 of solar irradiance, which is only ever exceeded on rare occasions in terrestrial environments. As a result, many often think that this infrequent incident must be planned for as proper worst-case design engineering, but is this truly always necessary? A lot of design concerns come from a fixation on ‘nameplate’ ratings of equipment, even though many of those ratings can end up being not that relevant in real-world scenarios.
In order to calculate a ‘nameplate’ rating for a piece of electrical equipment, you must establish a specific set of conditions, such as 100% load for 40 years at 30 oC (86 oF) ambient temperature. This combination of conditions, however, almost never happens, which is why a lot of utility transformers and cable systems are still in service long after their initial 40 years of design service life. The important thing to recognize is that real-time ‘nameplate’ ratings are not really fixed values but fluctuate with changes in ambient and loading conditions, which can be either to the detriment or the benefit of the rating. For transformers and cables, the biggest concern regarding aging and consequent end of life is the degradation of their organically-based insulation materials. Let’s look at some of the rating conditions for the major PV project elements of transformers, cables, transmission lines, and PV Modules.
PV transformers (without Battery Energy Storage (BESS)) of course, cannot be loaded to 100% all of the time, so there can be quite a bit of margin in their design, and short-time minor overloads should not be an issue. Even IEEE Standard C57.91-2011, “IEEE Guide for Loading Mineral-Oil-Immersed Transformers and Step-Voltage Regulators” recognizes that short-term overloads of up to 200% of nameplate rating can be possible under certain conditions without significant loss of life. Emergency 4-hour overload ratings of 200% over nameplate have been adopted by some major utilities since the capital cost of providing double the equipment rating, which would rarely be utilized, is cost prohibitive.
It is also odd that more PV designs do not take advantage of utilizing a set of cooling fans in order to purchase a lower-rated transformer and save significant capital costs. For example, when Florida Power & Light is designing an 85 MVA system (75 MW PV + 10 MVAr of capacitors), they only purchase a 51 MVA transformer. The first stage of added fans takes the rating to 68 MVA and the second set of fans takes the rating to 85 MVA. The fans are a lot less expensive than essentially buying another 34 MVA worth of transformer, and each stage of fans gives you a boost of ~33%. The transformer should be located within a fence to avoid public exposure to the moving fan blades, which typically do not have “finger-safe” guards.
When sizing cables for a continuous load, the NEC requires a 125% factor to be applied to the rating, with one exception. Per NEC, if the feeder is only supplying transformers, then you just need to size the cables for “the sum of the nameplate ratings,” with the assumption apparently being that the transformer size already incorporates the 125% factor from all the loads that they were sized to serve.
Just like transformer ratings, the widely used NEC Table 310.15(B)(16) is based on continuous 100% loading, and using it for all applications of cables in either raceway or direct buried is definitely conservative, as many PV systems have actual “load factors” of around 40%. One way to possibly drop down one cable size is to utilize a specific table provided by IEEE, which provides voluminous tables for both 100% and 75% load factors, with the 75% load factor option generally giving a cable size reduction from the 100% tables which closely match the tables in the NEC.
Using a messenger wire system such as a CAB System lets you use a higher cable ampacity, but this higher value can get negated if the cables have to transition to the underground for any significant distance.
Transmission Line Ratings
Large utility transmission lines have been adopting ‘dynamic ratings’ based on actual ambient and loading conditions measured by sensors placed around the line conductors themselves. Utilities are applying technology to save money and increase ratings; why not PV owners?
PV Module Ratings
The rating of the PV panels is based on 25 oC (77 oF) operating temperature at 1000 W/m2 irradiance. It is important to note that as the temperature gets hotter, PV modules produce less power. Although counterintuitive, it is an understood axiom in the renewable energy world that PV modules hate the hot sun and wind turbines hate high wind. The highest PV output is often on cool, windy days in late Spring when the temperature conditions will be far less than the high temperatures built into the nameplate assumptions of the other electrical equipment such as transformers and cables.
The PV module international rating standard provides good consistency for comparing module ratings, with only one slight drawback: depending on geographic location, the conditions that define STC almost never occur in the real world. A big reason why STC occurs so rarely is that the temperature parameter is 25 degrees Celsius cell operating temperature. This is the operating temperature of each solar cell behind the glass front of the module, not the ambient temperature. When bright sunlight shines on something for a while, it naturally tends to get hot. So, for an individual cell to be operating at 77 oF, it means that the ambient temperature would likely have to be closer to 32 oF. This does, of course, depend on numerous variables such as how close to the roof the module is mounted, how much cooling airflow the module is receiving, etc.
Several years ago, we worked with a community college in Ohio and obtained 1-minute irradiance data for an entire year from the weather station that they had set up as part of the NABCEP training which they offer. It ends up that the “Standard Test Conditions” on which the modules are rated were only present for about 12 minutes out of the entire year. So there goes that blanket “STC rating” assumption. It was interesting that a few data points almost hit the tremendously high value of 1,400 W/m2, likely from ‘cloud lensing,’ but it was only for a minute. More importantly, we couldn’t find any interval above 1,000 W/m2 that was present for more than 6 minutes, which was definitely less than an NEC three-hour period for continuous loads. Also, a fair percentage of high irradiance minutes are at temperatures above 85 oF, meaning that the power reduction from high temperature means that you should next expect maximum output for those minutes.
For every PV module, there are three temperature adjustment factors given on their data sheets: power in relation to temperature, the voltage in relation to temperature, and short circuit current in relation to temperature. The first two are negative factors, with only the short circuit current increasing with temperature.
So, let’s take an independent look at where that extra 125% factor might have come from. For starters, as the temperature goes higher, the short circuit current goes up. Assuming a very conservative factor of 0.6%/oC, moving from the rated conditions of 25 oC (77 oF) to 50 oC (122 oF) yields an increase of 1.5%, which is not very much of an increase and not close to 25%. The rest of the 23.5% would be assumed to mostly come from irradiance, but that still seems conservative to me. We have not found any documentation that describes how that extra 25% factor was derived.
This lack of correlation to real-world weather conditions has resulted in many solar module suppliers publishing the alternative Nominal Operating Conditions (NOC) ratings in addition to STC ratings. NOC conditions are defined as an irradiance of 800 W/m2, 20 degrees Celsius ambient conditions (68 oF), and an air mass of 1.5. This is meant to give a lower, more realistic indication of expected power output. A review of the community college irradiance data, however, shows that NOC occurred for a total of only 1,306 minutes or just 0.5% of the total daylight time during the year. This is a large improvement over the 12 minutes at STC, but it’s still not a meaningful percentage of actual operating time. And just to keep things confusing, there are also Standard Operating Conditions (SOC) of 1,000 W/m2, cell temperature at NOC, air mass of 1.5, and PVUSA Test Conditions (PTC) of 1,000 W/m2, 20 degrees C ambient conditions, air mass of 1.5, and a wind speed of one meter per second. Both the SOC and PTC ratings fall in between STC and NOC ratings and have fallen out of use, leaving STC and NOC ratings as what is typically offered on PV module data sheets.
Voltage Drop Considerations
Voltage drop design criteria vary from project to project, with a relatively common criteria of 2.0%, although sometimes this is specified as a maximum value and sometimes as an average value. We have seen criteria as low as 0.5% for dc, which pretty much drives the design to #8 copper string wires and large copper combiner box output conductors.
For string inverters that take multiple individual string inputs, the standard voltage drop criteria and resulting loss calculations are fairly realistic. When combiner boxes are used to provide a single input to either a string inverter that has a single MPPT or to a central inverter with a recombiner, it is not so realistic. Each combiner box is a single electrical node and can only have one voltage – those individual varying string voltages will have to average themselves out to a single voltage because, electrically, they don’t have a choice in the matter. For those PV Owners who have rigid, ‘cannot violate’ voltage drop criteria, are you letting a worst-case string voltage drop drive an increased design cable size when that voltage practically won’t exist?
Using a Factor Lower Than 1.56
Are the NEC requirements overly cautious to prevent fires? In many cases, yes. Are they often costing additional money unnecessarily? In many cases, yes. Fortunately, there was a change in the 2017 NEC that allows you to perform an engineering analysis to determine the maximum 3-hour current for arrays over 100 kW. We have applied this to several of our clients, and it has saved them at least one cable size for portions of the array.
A low voltage drop criteria will minimize energy losses over the lifetime of the project, but does the present value of those losses even pay back with the high material prices seen today, particularly for PPAs with a relatively low $/kWh rate? We will provide some interesting calculation examples in part three of this series now that we have covered the required background discussion.
Why don’t more people use this method of reducing the 1.56 factor? The reasons can be many, such as the lack of adoption of the 2017 NEC Code in your area, lack of knowledge about the NEC change, not having an engineering firm comfortable with the various approved methodologies, AHJs not allowing an engineered design because they don’t have the background to evaluate it, and Owner’s Engineers demanding a traditional sizing approach. And then there is the “need-it-now” factor for many projects, which precludes the extra time for engineering analysis, and just doing a straight, traditional NEC design can be completed within the rush timeline. A few years when copper prices were low, it tended not to matter very much monetarily to the project, but today, with high commodity prices and supply chain issues, it is a different story. If you have a system over 100 kW, you need to get away from the 1.56 factor whenever you can.
Depending on the PV array dc/ac ratio and geographic location, the peak inverter output may not even be present for three continuous hours, or with a high dc/ac ratio on single axis trackers with bifacial modules and string optimizers, it could be present for ten hours or more. The bottom line is that you need to understand the anticipated year-long variations in the behavior of your PV system and take advantage of allowed engineering cost reductions wherever you can, and don’t double up on worst-case factors that are not going to occur concurrently.
If you have any questions about PV cable sizing, voltage drop, or otherwise, get in touch with one of our engineering experts today. Also, stay tuned for part three of our PV Cable Sizing blog post series, in which we will include insights such as more examples of cabling cost reduction ideas.