Systems Operating at High DC:AC Ratios

Transcription

1 Systems Operating at High DC:AC Ratios WHITE PAPER Abstract Recent trends in the PV industry have been towards an increasing DC:AC ratio, where the AC conversion capacity attached to a PV system is significantly smaller than the nameplate DC value. As the cost of the DC components continues to decrease, and the need to reduce grid disturbances due to the intermittent behavior of conventional PV systems, this trend will likely accelerate even further. For this trend to continue significantly beyond the DC:AC ratios of 1.4 being used today however, several hardware limitations on conventional systems will have to be resolved. This paper describes a cost competitive, and highly redundant electrical topology where these hardware limitations have been eliminated. In addition, for bi-directionally oriented systems using dual MPPT or microinverter based solutions and high DC:AC ratios, a second major issue results from preferential clipping due to the combination of fixed AC limits and the imbalance of DC energy presented by the system. This results in additional energy losses at high DC:AC ratios. The tenk electrical topology presented within also eliminates this source of energy loss. tenksolar 9231 Penn Avenue South Minneapolis, MN Toll free: Simply More Energy

2 Introduction The most important point to observe when considering DC:AC ratios when sizing PV systems, is that two productivity measures exist in the industry. The DC productivity (kwh AC/kW DC) is the AC energy production normalized to the DC system nameplate sizing. If there is no AC power limit on the system, this is often the critical metric used. The AC productivity (kwh AC/kW AC) is the AC energy production normalized to the AC system sizing. For cases where the AC system power is limited (typical), this is often the most critical metric, depending on the value placed on the kwh AC, the fixed project costs, etc. Due to the intermittency of PV, the relationship between the DC and AC productivity is very different from conventional energy sources. Figure 1 is an illustration of the relationship, based on various DC:AC ratios. At the right side of the graph, the DC productivity is limited by the total amount of sunlight available, so in this case is limited to a capacity factor of about 14%. Because the system is limited by the total radiation available, a low DC:AC ratio (more AC capacity) cannot increase the net system output, thus the AC capacity factor falls with DC:AC ratio. As the DC:AC ratio is then increased, the AC capacity factor then begins to climb, at the expense of the DC capacity factor, since the under-sizing of the AC conversion capacity on the system is basically limiting the AC output power whenever the DC power available exceeds the AC output (ignoring DC:AC efficiency in this example). In PV, as the upper limit of the DC output is capped with a higher DC:AC ratio, the impact on DC productivity is initially very little, since the only capping is near solar noon on a few of the best days. As the DC:AC ratio continues to increase, the impact on DC productivity is larger, however the AC capacity factor can be driven up very quickly to over 30%, when moving toward the left of Figure 1. At the extreme left, this is basically a very large PV array (large DC nameplate) with a very small AC conversion capacity. The sun comes up in the morning and it turns on at its maximum, and shuts off in the evening, limited only by the total hours where some sunlight is available (extremely stable, but very expensive). Since the AC productivity is the (or one of the) critical parameters for many systems, selecting a high DC:AC ratio is an advantage. However when optimizing for the lowest cost of energy, the total system cost is also critical so having an extreme DC:AC ratio is not economical, as installing a very large DC array and clipping most of the DC energy away has a poor economic value (unless the PV DC costs drop much further). Thus, optimizing financial returns based on the entire system economics, at various DC:AC ratios will define the optimum design point. Effect of DC Array Configuration (i.e., Module Tilt, Positioning) When considering various DC:AC ratios as design options for optimizing system economics, it is also important to realize that some particular DC array configurations are preferred over others (orientation and placement of the modules). As noted previously, if one is to select a high DC:AC ratio, the lost DC energy will be mostly near solar noon (for a south facing array) and during direct beam illumination (i.e., summer sunny days in the northern hemisphere). To minimize the DC productivity losses with higher DC:AC, one should select a DC panel configuration that has a flatter daily profile (early to rise, late to fall, low peak) and responds strongly to diffuse illumination. To illustrate, Figure 2 is an example of a shading tolerant panel configuration where the array is comprised of 25 facing south modules and 15 facing north modules. Observe that in the summer, the north facing modules produce 90% of the energy of the 25 tilted south modules but only rise to 70% of the power, due a much higher productivity in the early AM and late PM. Thus one could use a much higher DC:AC ratio with the north facing modules in the summer, with much less clipping of the DC power. Likewise, during diffuse periods, the north modules have nearly the same productivity as the south facing modules. Higher tilt, east/ west configurations also have a lower daily energy profile and are becoming much more common in Europe. Higher tilt, south facing modules generally have a shorter day, higher peaking profile, making this configuration less desirable for high DC:AC applications. Redundant Array Electrical Topology Before detailing out the limitations of existing systems with high DC:AC ratios, an introduction to a tenk PV electrical topology is required. The electrical architecture of the tenk system is illustrated in Figure 3. tenk modules as shown interconnect all cells in both series and in parallel, such that any cell-level shading or soiling, broken cells or interconnects allow current to flow around a partially contributing cell (as opposed to it being a constraint). A set of highly redundant DC:DC converters are used on each module along with a very low cost current return backsheet to form a complete module circuit where current can take many paths through a module for superior reliability and minimal cell-to-cell current dependencies. Figure 4 is an illustration of an extreme shading test, showing the effectiveness of the design as it relates to cell-to-cell shading. Each module is then connected in parallel to a common low-voltage DC bus. Also residing on the DC bus are a set of inversion units also operating in parallel (using a proprietary DC voltage control method to avoid communication between units). Each inversion unit may pull energy from the DC bus and deliver it to the grid, depending on the individual settings and DC voltage on the bus. Low illumination efficiency is greatly improved since energy from any module can flow through any inverter (and only the inverters needed are operated at low illumination). This electrical topology also yields a very efficient module due to the low parasitic losses and minimal edge spacing due to the low voltage of each module, and also a module that can tolerate variable illumination between Page 2 tenksolar, Inc., 2015

3 cells. For reliability and minimum field service, complete redundancy is achieved in the entire system from the cell to the grid. PV System: DC vs. AC Capacity Factors AC Capacity Factor 50% 40% 30% 20% 10% Hours in the Day Limit (On/Off PV System) Incident Radiation Limit DC:AC Ratio 0% 0% 5% 10% 15% DC Capacity Factor 0 AC CapFactor DC:AC Ratio Figure 1. The relationship between DC (kwh AC/kW DC) and AC (kwh AC/kW AC) productivity (stated as annual capacity factors) in Minneapolis, MN. As the DC:AC ratio is increased and some of the DC energy is clipped at the high productivity periods, the DC productivity falls, however the AC productivity increases. As the DC:AC is increased further, the amount of DC energy lost increases, causing the rate of increase in AC productivity to slow as the DC:AC ratio is increased further. A DC:AC ratio of 2.5 corresponds to a DC capacity factor of 12% and an AC capacity factor of 30%. Designs for Tolerating High DC:AC Ratios For a conventional PV system, the design requirements are that any component included on the DC side of the array, must be able to withstand 156% of the short-circuit current of the array. Consider as an example, where a DC:AC ratio of 2 is to be considered for a 1 MW AC array (i.e., a 2 MW DC system). This requirement forces the entire array to be able to support 156% times the short-circuit current of a 2 MW DC array, including all DC wiring capacity as well as the 1 MW AC inversion unit must have a greatly oversized set of DC circuits to withstand an abnormal short-circuit event in the inverter. All of this oversizing of the DC side vs. the allowed maximum AC power limit adds costs to the system. Page 3 tenksolar, Inc., 2015

4 Figure 2. Examples of various DC energy production profiles to illustrate how to optimize systems for high DC:AC ratios. The upper graph is a summer trace for PV modules tilted at 25 south and another group at 15 north. Observe that the 15 north modules can produce 90% of the energy of the 25 south facing modules, but with only 70% of the peak DC power. Thus, if the system is capped at 70% of the peak power, the south facing modules would effectively produce less energy than the north facing modules. The second graph illustrates a diffuse day, where the 15 north facing modules produce the same amount of energy as the 25 south facing modules, since the sky is nearly uniformly illuminated under these conditions. RAIS PV Architecture V OC = 0 ; J SC = 0 DC Bus Optional Storage Redundant Inversion Grid Figure 3. The RAIS electrical architecture, illustrating the modules with cells connected both in serial and parallel, the highly redundant electronic DC:DC converters integrated into each module circuit, the low-voltage parallel DC bus, and the redundant inversion units operating on the same DC bus. No communications is required between any of the units on the DC bus, for greatly improved reliability. An optional storage system is also shown which can also operate directly on the DC bus. Page 4 tenksolar, Inc., 2015

5 Figure 4. A shading test illustrating an opaque block across the module eliminating 22% of the optical energy, and 25% of the module power (this extreme block results in a 3% loss due to the electrical topology of the module, vs. nearly 100% on a conventional module under similar conditions). The thermal images illustrate the temperature uniformity (no hot spot or other deleterious effects of partial shading as exists with other solar electrical topologies). For the tenk electrical topology of Figure 3, the electronics within each module has been certified to withstand the entire DC shortcircuit of the PV system within each module (proprietary, highly redundant cell and electronic architecture within module), and also is certified to export a limited amount of current, which can be at or below the maximum PV power without consequence, as the module simply moves off the peak power point within the module if the peak output current limit is reached. This extra PV current during short-circuit conditions can never exist in the DC wiring or DC:AC conversion process. The following describes how the system of Figure 3 operates: 1) At times when the system is not limited in power output, the modules all produce the maximum current possible at whatever system voltage is presented to the modules (30-55V DC). 2) At times when the system is not limited in power output, the redundant inverters pull current to maintain the system voltage between 50-52V DC. In this case, the inversion units set the system voltage to a targeted value, not the modules. Also note the inversion units do not need to manage MPPT, since the modules deliver maximum power at any voltage. If one attempted to run MPPT with multiple inverters on a parallel bus, each inverter would not be able to distinguish changes in array power levels from changes in power level of other inverters destabilizing the entire system each time an inverter changed its peak power point. 3) A narrow but finite range of voltage setpoints are used in each of the inverters, for those inverters with setpoints low in the allowable range, they will operate at full power since they cannot maintain the system voltage setpoint. For inverters with higher setpoints, they will pull smaller amounts of current (if any), and will actively adjust the amount of current pulled to maintain a constant system voltage (50-52V DC). 4) The inverter setpoints change daily, to wear-level all inverters. 5) When the last inverter reaches a point of saturation, the system voltage will then begin to rise, at which point one or more of the modules will then reduce its current output to maintain the system voltage of 55V DC. At this point in time (when all inverters are saturated), the modules then control the system voltage, not the inverters). Figure 5 is an illustration of the daily output of a commercial 13.2 kw AC sub-array (24 redundant inverters in this sub- array, and part of a larger 99.0 kw AC array) when operating in saturation over a portion of the day. Observe the rise and fall show no transition periods when inverters are coming on or off, or the transition from when the inverters are controlling the DC voltage setpoint vs. the modules taking control when the inverters saturate. Observe that in this topology, the inversion units are never exposed to the full PV short-circuit current, only the current limited output of each module. When using very high DC:AC ratios (i.e., when all inverters are saturated), the modules limit the system current, thus allowing the wire and inversion system sizing to be designed to the AC power output limit (no cost penalty), rather than the wiring and inversion DC circuits having to match the installed DC capacity (see wire sizing comparison between tenk systems and 600/1000V DC conventional systems). Page 5 tenksolar, Inc., 2015

6 Figure 5. An example of a 24 redundant inverters (13.2 kw AC) operating with a 14 kw DC array, illustrating the rise and fall in the production through the day, and also the period of inverter saturation. At the left and right, the system voltage is controlled by the inverter group, allowing more inverters to engage and operate as the power climbs and falls (note there are no transitions visible it is not detectible when units come on or off). At the saturation point, all 24 inverters are running all-out, and the system voltage is held at 55V DC by the constant voltage operation of some or all of the modules. Observe also how smooth the cap line is as well. Clipping in Bi-Directional Arrays (e.g., East-West Arrays) Another severe issue in bi-directional arrays which are AC power limited and operating at relatively high DC:AC ratios, is related to how the AC power is summed and combined in conventional DC systems, or even in micro-inverter based systems. For bi-directional arrays where one group of panels is tilted in one direction and another group is tilted in another direction (an example would be an east-west system where a group of panels is tilted east and another group is tilted west), due to the mis-alignment each of these two groups must be tracked on a separate maximum peak power tracking loop (MPPT), using either two central or string inverters, or inverters with two DC inputs and separate MPPT loops. Another option is to use all micro-inverters, since each inverter then tracks MPPT separately. Consider the case of a 1 MW AC limited system operating at a DC:AC ratio of 2.0 (i.e., 2 MW DC nameplate). First consider the case of two separate inverters, 500 kw AC each and an east-west system. By mid-morning the sun alignment is high on east facing panels and the power may be 800 kw DC, but still poorly aligned on the west facing panels (say 300 kw DC). Since there is 1100 kw DC of total power available, the system should be delivering at the peak of 1 MW AC. However, since each inverter (or MPPT tracker clips individually) clips at 500 kw AC (ignoring DC:AC efficiency for this simple example), and the other is operating at 300 kw AC, for a total of 800 kw AC well below the maximum energy available. This same issue exists on dual-mppt tracking systems since the DC sizing is limited to about ½ of the total AC output, in order to minimize the cost of the DC side of the unit. Micro-inverters AC undersized to the same limits suffer the same issue, where the east facing inverters are saturated and the west facing inverters are not thus the total AC energy falls short. Consider now the tenk topology of Figure 3, where the DC from east and west facing modules are combined on the DC bus before the current is delivered for inversion. Since the DC sources are combined before the AC limit is imposed, current from any module can flow through any inverter to fulfill the total maximum AC level. Thus, in the example above, an equal number of east facing and west facing modules have their DC combined such that the sum total of the 800 kw DC is combined with the 300 kw DC, the entire 1 MW AC can be delivered (only limited by the sum of the AC inversion capacity). Conclusion By limiting the DC current output at the source of the PV, tenk electrical topology allows the use of much larger DC:AC ratios than are possible with conventional technologies without severe cost penalties for oversizing the DC systems. In addition, in bi-directional arrays such as east-west systems, the tenk electrical topology eliminates the early saturation that takes place when one DC current input is largely imbalanced from the other. About tenksolar tenksolar, Inc., provides a photovoltaic solar solution that delivers on the promise of the lowest cost of solar electricity, while at the same time improving power density, safety, longevity and bankability of photovoltaic systems. Since its founding in 2008, tenksolar has been a leading innovator in the delivery and implementation of photovoltaic solar systems for commercial customers. More information about tenksolar is available online at Page 6 tenksolar, Inc., 2015 TKS WP