So, you have this great new LED, generated a wonderful 130 lumens per watt. Life is good. Is this representative of real energy savings? What about Power Factor and its impact on the demand it places on the energy grid? What’s up with THD and what does it mean? Do either of these detract from the energy savings of new technologies using electronic power supplies?
Simple definitions to work from:
Volt-Amps (VA) = the product of a lighting products current (Amps) multiplied by the voltage applied. Also known as Apparent Power, as this represents the amount of current drawn from the electrical system during operation. Apparent Power is what utility companies must generate in order to support the loads placed on power grids and systems.
Watt (W) = a measure of work accomplished by a system equivalent to the theoretic volts of a system multiplied by its current (VA), with no other loss, inductive or reactive factors involved. Also known as Active Power, the assumption in all Watt metric statements is that there is perfect Unity between Active Power and Apparent Power, which exists in pure resistance loads. An example of a product that presents this simple load, where a 60W lamp draws .5A from a 120V power source – 60W = 60VA. Active Power is what residential electric meters measure, as this is a simple and cheap metering strategy.
Power Factor (PF) = the difference between Apparent Power and Active Power. PF = W/VA. The loss of energy from systems with Power Factors less than 1.0, is sometimes referred to as Reactive Power.
When a system has a Power Factor of .50, this means that there is a difference in the measured watts of the load and demand it places on the current capacity of the power supply of 50%. For example, a 60W load with a power factor of .50, produces a 120VA demand on the power grid. Since VA is what is produced by the utility system, for the purposes of the impact of a lighting system on the power grid, VA is an important metric, as it represents the current demand placed on generating plants, which increases losses throughout the supply grid.
Total Harmonic Distortion (THD) = Distortion of the relationship between line power and load current draw.
How Much Additional Energy is Actually Consumed:
This is where the issue gets sticky. For any given load, the watts consumed represents the power consumed by a product. Power factor actually has a smaller impact on the energy consumed than it has on the impact it has on the efficiency of a power grid or generating facility. Using the following formula (from the DOE EERE Information Center) you can calculate the loss factor change from power factor correction using capacitors (motor/inductive load example):
(1-(PF low/PF high)²) x 100%
This means that if you improve a product from .50PF to .90PF, the loss factor improves by (1-(.50/.90)²) x 100% = 69%. If the total losses on the circuit feeding the product is 2% (I²R) the energy savings can be calculated as (69%/100) x 2% = 1.38%. This is not a significant amount of energy savings. Further, if distribution losses are lower, so will be the energy savings. For this reason, power factor itself is not a source of direct energy reduction, as it relates to power consumption, especially for very small loads, and on circuits operating well below their maximum capacity (resulting in lower (I²R) distribution losses.
But this is just at the local level. If we extend this from the local to the grid itself, and consider all losses involved, we get a different picture. The 2% loss at the local circuit level assumes that the supply side to that point is 100% efficient, which it is not. In fact, total losses involved with transmission and distribution from the power source to the outlet are closer to 8%-10%. If we assume 9%, the energy saved from the previous improvement in power factor are 6.21%, which is a more significant number.
Load Scale Impact:
For low power systems, or lighting loads that represent a small portion of total lighting load, like task lighting, or accent lighting, the savings from power factor correction are unlikely to be significant. For example, in a 20,000 s.f. facility, designed to 1.2W/s.f., the total lighting load would be 24kW. If small loads comprise 15% of this, the total load would be 3.6kW. A savings here of 6.21% is just 223W. However applying the gains to the remaining load does produce a saving of 1,266W. However, since this is total electrical grid system load, which is not included in local metering, this only translates to a savings in cost of $88 per year, while the smaller load represents just $16. Because of this, interest in spending extra money to improve power factor is not always strong, requiring standards be established to force the issue. If we factor in the dynamic that most accent and task lighting loads are connected to circuits that are over-sized, the gains are even less in both energy and money saved. For this reason, small loads, those under 25W for example, are often set aside from compliance, while larger loads are required to comply to a higher PF requirement.
The Problem with Lighting and SSL Loads – Harmonic Distortion, THD and dimming effects:
While motor loads can generally be PF corrected using capacitors, including applying PF correction strategies to building power systems upstream of individual loads, power factor issues caused by lighting loads using electronic switching cycle power supplies can create high harmonic distortion power factor losses not as easily resolved. For this reason and others, Total Harmonic Distortion (THD) may be as important, or more important than the resulting PF. The relationship between PF and THD are roughly tied together as such: THD = 1-PF². So, the demand for low THD and low PF are actually addressing the same issue. For example, in the previous example, a .50PF product will have a THD of roughly THD of 75%, while a .90PF product will have an estimated THD of 19%. This translates to load on the system in a similar manner that PF does.
While the impact is the same on the power generation end, due to the wide range of individual product characteristics involved, correction of THD losses can only be effectively accomplished at the individual load. THD also has other implications on other systems within the building, so there are other reasons to control this specification as well. However, as it is with PF, the impact of THD on the building and its power system is proportional to the loads involved.
Further, THD is a trickier issue than PF, as there are so many variables in frequency, distortion shape, etc.. that interaction between products can create compound issues on a circuit and building system as a whole, if not fully resolved. For this reason, THD must be corrected at each individual load, to avoid expanding its impact beyond.
Dimming adds a dimension to PF and THD that is rarely defined. For the most part, line voltage dimming will almost always increase THD, reduce efficiency of electronic power supplies and drivers, and decrease PF. While the load is lessened, and light levels drop, not all of the load reduction will be realized as energy saving. How great the effect will depend on how Efficiency, PF and THD are controlled in the product. Products with active PF correction are best, resulting in less drop in PF and less gain in THD, and often less droop in efficiency under dimming. However, this is rarely applied to line voltage dimming products – generally reserved for 0-10V or PWM dimming architectures, where the output to the LED is controlled, separate of input power characteristics. This is not to say that there are no line voltage dimmers or dimmable LED products with active PF correction that generate consistent PF and THD performance when dimmed, there are. They are, however, more expensive and less common, requiring careful shopping to find.
Lumens per Watt – Electrical Grid Energy Impact:
The majority of solid-state lighting products are powered by power supplies with Power Factors of between .50 and .90. That means that an SSL product advertised to consume 20 Watts, may actually be placing a demand of anywhere from 22 to 40 VA on the grid, even if this is not realized as a power cost at the local level. If the THD of a product is high, a similar effect will exist. Since the real goal of energy efficient lighting is to reduce demand on the power grid, it is critical to consider both PF and THD in selection of large lighting loads, and products making up the largest part of a lighting system. While we use energy saving ROI to convince customers to install efficient products, the real imperative is to cut our national grid energy demand as a whole, a far greater issue than saving building owners a few dollars on their power bill.
For this reason, simplistic lm/W metrics do not fully describe the impact the lighting system will have on the goals of reducing power system demand. For example, if an existing lighting system compiled presents a load of 12kW @ 18lm/W, at a PF of 100 (halogen) at 120V, or 100A, is replaced by a new system with a load of 4.8kW (40A) at 45lm/W (savings of 60% at the power level) but has a total lighting system PF of .45 due to a lack of consideration of THD, the resulting gains at the local level in power grid load is 51A, for a savings of just 49%. This reduces the reduction of impact on the grid, so effectively reduces the realized gain in efficiency by 22%, so cuts the realized efficacy from 45lm/W to 35lm/VA. Further, the reduction in energy saved as measured in watts is also reduced by roughly 6% over system PF of 85%, so even at the watts load level, efficacy is reduced from 45lm/W to 42 lm/W at the local level.
These are certainly overly simplistic examples intended to illustrate the point that overall system PF and THD are important factors in understanding what energy savings is actually realized, both at the local and grid level, from applying SSL products. Calculating the finite impact of all these effects as it relates to grid loading of a lighting system is almost impossible to do with complete accuracy, so the best approach is to be cognizant of the issues and select the highest performing products appropriate to the application and assume the downstream impact is positive as a result.
Saving Energy, Impact on Circuit Loads, vs. Saving Money:
When it comes to energy savings – the issue of ROI, and reductions in metered power costs are the key issues discussed. This may mask PF and THD issues in several ways:
First, if the existing system has a very poor PF, like that from old ballasts, or high THD products, not including these factors in the ROI may increase payback periods. It may be worth considering the existing lighting system in this detail when attempting to realize the fastest possible ROI.
Second, if the new system has a poorer PF and THD than the existing, for example replacing incandescent products with SSL products with modest PF and THD levels, the realized savings metered at the building may not match the rough calculations assuming no PF or THD effect. It may be worth including this effect when providing customer presentations, to avoid disappointing final results.
Third, when designing new lighting systems, don’t forget to consider PF and THD impact on connected Amp loading per circuit. While watts is the energy consumed, circuit breakers and fuses are sized to the load Amps. Use only max Amps to design circuit loading. While you are there, don’t forget to factor inrush current (starting Amps). Many LED products have inrush currents significantly higher than load watts might indicate. This is independent of PF and THD, and related to how capacitors are charged within the driver and power supply. The inrush current is not always proportional to the load applied, so some care is needed here. Some low watt LED products have inrush currents as high as 35A, which means they present a load of 35A for a few cycles at start up. It is entirely possible to install lower power consuming products that trip circuit breakers not designed to tolerate this momentary load.
Approaching energy efficiency with the goal of reducing demand on the power grid is a more sophisticated and accurate way to think of energy savings. Further, when PF and THD is properly included in product evaluations, the differences between good and poor performing products is often made more visible – an important factor in a market often overflowing with low grade products, and in justifying higher performing products that cost a little more, over low end products that sacrifice these factors to produce lowest initial cost.
While ultimately the utility company is most concerned with PF, the greatest impact on customers is in PF tariffs on commercial customers to offset the additional costs involved with pushing the missing Amps through wire. Residential customers, who are metered by kWh exclusively, will see very little to no advantage of low PF products on their energy bill. However, as we move more and more toward energy saving electronic products, it is only a matter of time before the impact is felt at the grid end, resulting in utilities looking at recovery of lost capacity and income, resulting in application of PF correction or tariffs on all electrical loads.
For small loads (under 25W), there is no real advantage of high PF products, other than to achieve minimal THD, which can have a small detrimental effect on other products on the same circuits. A PF of >.50, and THD <70 seems a reasonable expectation, with efficiencies of .80%. For larger loads, and lighting products used for the larger general illumination uses, an efficiency of >85%, a minimum PF of >.90 and THD <20% seems the best prescription. This produces the most efficient system at the local circuit level through the power grid.
Micheal Poplowski from PNNL contacted me to point out that PF and THD are not tightly bound to one another. THD is a measure of distortion only, and can be measured in either voltage or current distortion, each having a different effect on current drawn by a system. He recommends additional documents for those interested in diggging deeper into this topic:
To investigate this further, I purchased a proper power analyzer, that measures both Voltage THD and Current THD, as well as calculates PF using a computer. In measuring several power supplies, I have found that any time the current THD is high, the PF is low. However, as Micheal points out, while their is a general correlation between them, with the current THD at least, deriving PF from THD or THD from PF is only roughly corollary. The relationship between voltage THD and PF is weaker, as is the relationship between voltage and current THD.
None of this detracts from the general content of the connection between PF and power consumption, or the discussion of THD and its reflection of and impact on PF as it relates to energy consumption. However, it does illustrate that a brief article on the topic falls well short of providing the depth of understanding required for designing electronic systems or a PF correction strategy for a building or grid system. But this was not the purpose of this article.