RER Renewable Energy Research  français
   Overview       Expertise       Publications       Resources        Photos    
    All       By Topic       Articles       Reports       Non-technical       Presentations    

Snow and Ice Accumulation on Photovoltaic Arrays:
An Assessment of the TN Conseil Passive Melting Technology

Michael M.D. Ross
CANMET Energy Diversification Research Laboratory

Full Text of Report (5 MB, 273 pp.)
Link to CETC-Varennes

Note on Authorship:

This technical report was authored by the principal of RER Renewable Energy Research when he was an employee of the CANMET Energy Diversification Research Laboratory (now known as CETC-Varennes-- see link above).

Acknowledgements:

Funding for this project was provided by CETC-Varennes through the Panel on Energy Research and Development (PERD), Canada's Green Plan, and the Science Institute of the Northwest Territories (SINT) as part of the joint NRCan/SINT "PV for the North" programme.

Citation:

Ross, M. M. D. Snow and Ice Accumulation on Photovoltaic Arrays: An Assessment of the TN Conseil Passive Melting Technology, report # EDRL95-69(TR), Energy Diversification Research Laboratory, CANMET, Natural Resources Canada, Varennes, September 1995, 273 pp.

Summary:

Background: In Canada, photovoltaic (PV) systems are usually located at remote sites, where their low-maintenance, high-reliability characteristics make them the least-cost power supply option. However, during winter, snow and ice can accumulate on PV panels, thereby reducing electrical output for extended periods of time. This can result in the technology being unable to meet the load's power requirements, undermining the PV system's reliability. Such a failure often necessitates a visit to the site, which can be prohibitively costly at the more remote sites, especially those with helicopter access. In order to lower the probability of system outages, battery banks for PV systems are currently oversized, adding considerably to the cost of the system.

The TN Conseil Technology: Working in conjunction with the EDRL and the Institut de recherche d'Hydro-Québec (IREQ), the Montréal firm TN Conseil has developed a passive technology that melts snow and ice from the face of a PV panel. This technology consists of a black absorber foil bonded to the rear of the panel, a one cm air cavity which insulates against heat loss, and a clear Lexan back cover sheet. The absorber foil collects the solar radiation that is incident on the rear face of the panel, and in combination with the back cover, elevates the temperature of the panel, promoting snow and ice removal. In this way, the technology is somewhat like a solar thermal stagnating air collector, utilizing the radiation which is reflected off the snowcover on the ground behind the panel.

In this report, the problem of snow and ice accumulation on PV panels is reviewed, and the solution proposed by TN Conseil is investigated.

Snow and Ice Accumulation on PV Arrays: Snow and ice can accumulate on a PV array in a number of physical forms:

Snow, rime, glaze, and hoarfrost are very unpredictable, both from year to year and from site to site. Whether they are a problem at a particular PV site depends on many site-specific factors that can not be deduced from commonly available meteorological data. Simple models for the accumulation of rime and snow are presented in this report, mainly to identify for the reader the principal factors affecting the rate of accumulation. Qualitative information on snow and ice accumulation is available, on the other hand, from experts in the field of atmospheric icing of structures and from PV installers and users, eighteen of whom were contacted for this report. Even so, several PV installers pointed out that it is difficult to know whether snow or ice accumulation is a problem at remote PV systems, since oversized battery banks power the load even when the panels fail for an extended period of time.

Rime causes serious problems at many PV installations, particularly those, such as telecommunication systems, located on ridges and mountain tops at elevations greater than 650 m. Rime may accumulate in the late autumn and not melt off until the spring, effectively "turning off" the PV power system for the winter. Riming of PV panels can occur in mountainous regions throughout Canada, including Devon Island, Ellesmere Island, the Yukon, northeastern Labrador, the Gaspé, the north shore of the St. Lawrence, Cape Breton Island, and the Coastal, Cariboo, Selkirk, and Stikine Mountains of British Columbia. It also occurs in Arctic areas prone to winter fogs.

In general, snow melts off, slides off, or blows off PV panels fairly quickly, and thus presents less of a threat to PV installations than rime. A wet snowfall followed by very cold weather appears most likely to result in a long-lived deposit of snow on the array. Large panel tilt angles are quite effective in preventing snow accumulation; nevertheless, PV installers reported some snow accumulation problems on the Prairies and in Québec.

Glaze does not appear to be a serious problem in the operation of PV arrays in Canada, since it transmits considerable light and, in most places, is followed by warm temperatures. An exception to this may be Newfoundland, and the Avalon region especially, where glaze is more severe, more frequent, and longer-lasting than elsewhere. It is highly improbable that hoarfrost poses a threat to the winter operation of PV panels.

Presently, many panels in rime-prone areas are mounted vertically and located at the most windswept sites possible, with the expectation that this will encourage shedding and wind removal of rime. However, these two measures also maximize the flux of supercooled water droplets onto the panels, thus maximizing rime accumulation. Experts on rime accumulation suggested that, when possible, less windy sites should be chosen. Additionally, lower panel tilt angles should be experimented with.

The problem of snow and rime accumulation has been approached in a number of ways. The most common practice is to drastically increase the capacity of the battery bank. It is not uncommon for rime-prone systems in British Columbia to have "three month autonomy", or the ability to power the load for three months from batteries alone. This increases the capital costs of the system, but installation, operating, and maintenance costs may be far more significant at remote sites accessible only by helicopter. Soltek Solar Energy Ltd., of Victoria, British Columbia, has developed a solar panel for flush mounting directly onto comshells. Wind, carrying water droplets, may flow around, rather than into, the comshell. Users report that this reduces, but does not eliminate, rime accumulation. Icephobic coatings, which purportedly reduce the adhesion of ice and snow to glass, wear off quickly and are unanimously dismissed by researchers and field personnel alike.

Pertinent Properties of Snow and Ice: The classification of snow and the properties of snow pertinent to the problem of snow accumulation on PV panels are reviewed in this report. Properties examined are density, effective thermal conductivity, latent heat of fusion, albedo, solar radiation transmission and emissivity. The transmission of solar radiation through snow can be modelled by the Bouger-Lambert law; thus, the level of radiation penetrating a layer of snow decreases with thickness in an exponential manner. The properties of rime are poorly documented and must be inferred from those of snow.

Electrical Performance of a Snow or Ice Covered PV Array: Snow or ice cover "shades" cells in the array and thus limits their ability to generate electricity. The electrical performance of a uniformly covered array will be linearly related to the level of solar radiation penetrating the snow cover. However, the electrical performance of a nonuniformly covered array will be nonlinearly related to the average level of solar radiation penetrating the snow cover.

In a series string of cells, the current passing through the cells is limited by the most severely shaded cell. The less-shaded cells force this cell to operate at a current above its short circuit current, and thus the cell becomes reverse-biased, acts like a resistive component, and dissipates electrical energy as heat.

To reduce the degredation in electrical performance associated with shading a small portion of a module, most modules have one or more bypass diodes in parallel with the cells. These bypass diodes shunt current past a string of cells if the string becomes seriously shaded. Thus, the bypass diodes limit the effect of shading on the module. Nevertheless, shading by snow or ice can have a serious impact on the module output: in an experiment on a Siemens M55 and an Astropower APC4716 module, shading one cell in the module by 60 to 75 % reduced the irradiance of the panel by only 2 % but reduced the peak power output by 40 to 50 %.

A model for the effects of partial shading of the PV array would provide a better basis for studying the effect of snow or ice on the array's electrical performance, but was not developed for this study.

Solar Energy Incident on the Rear Face of a Panel: In order to predict the thermal performance of the TN Conseil technology, the radiation incident at the rear face of the panel must be modelled. For most arrays the beam radiation incident on the rear of the panel during the winter will be insignificant; the diffuse radiation on the rear face of the panel can be modelled with the Perez model or a modified Temps and Coulson model with Klucher modulating function. However, the principal component of the radiation on the rear face will be ground-reflected radiation. A model for this, accounting for the shadow of the array (for both beam and diffuse radiation) is developed.

The model for ground-reflected radiation was compared with four measurements of the radiation incident on the rear face of the 20 kW roof-top array at the EDRL when there was snow on the ground. For this array, the average error for the model that does not account for the shadow of the array was 90 %, whereas the model that accounted for the shadow was, on average, 25 to 30 % in error.

A sensitivity analysis was performed on the model. On the rear face of the array, the incident diffuse radiation was only 5 to 15 % of the intensity of the incident ground-reflected radiation. The radiation incident on the rear face of an array was found to be 8 to 34 % of magnitude of the total radiation incident on the front face of an array, with 24 % a typical value. For most small and medium-sized arrays (up to one kW), the shadow of the array does not significantly affect the radiation incident on the rear face of the array.

Thermal Performance of a PV Panel with the TN Conseil Snow Removal Technology: The thermal performance of the TN Conseil technology without snowcover was studied in two ways. First, two panels, equipped with the TN Conseil technology and installed in the rooftop array at the EDRL, were monitored. Second, mathematical models of the steady-state thermal behaviour of a Siemens M55 panel, with and without the TN Conseil technology, were developed. When compared with the monitored data, the models tended to overestimate the monitored readings by three to four ºC, with a standard deviation of about four ºC. The models appeared to be less accurate at lower windspeeds and higher insolations, but equally accurate at all ambient air temperatures.

As expected, both the models and the monitored data showed that the TN Conseil-modified panels operate at temperatures significantly higher than those of unmodified panels. With moderate to strong insolation levels, the monitored TN Conseil-modified panels operated at temperatures 15 to 30 ºC higher than the unmodified panels. The models showed that the operating temperature is principally affected by the insolation on the front and back of the panel, the ambient air temperature, and the windspeed, which has a nonlinear effect.

The models were used to compare various options for the absorber foil and the plastic back cover. The standard TN Conseil technology, as described above, was found to have the highest temperature, the unmodified panel the lowest temperature, and all panel configurations with a plastic back cover operating at higher temperatures than all panels without such a back cover. However, the models predict only the average temperature of the panel, and the monitored panels revealed the existence of significant temperature variations at different locations on the TN Conseil-modified panel. Free convection results in the top half of the panel being up to 10 º C warmer than the bottom half, and the panel's aluminum frame acts as a heat sink and depresses the temperature of the panel in the vicinity of the frame.

Snow and Ice Removal from Photovoltaic Panels: Mathematical models were used to investigate the the rate of snow and ice removal from a Siemens M55 panel with and without the TN Conseil technology. For a given windspeed and insolation level, the TN Conseil modified panel initiates snow or ice removal at an ambient temperature much lower than that required for melting by an unmodified panel. For example, the model indicates that with peak sun and a 10 m/s wind, a 10 cm thick rime or dense snow accumulation will start melting at an ambient air temperature of -23 ºC on the TN Conseil modified panel but only at -3 ºC for an unmodified panel; the TN Conseil panel performs even better at lower windspeeds and with thinner accumulations. Furthermore, the TN Conseil-modified panel has a significantly higher melting rate than the unmodified panel. For example, for an ambient air temperature of -5 ºC and peak sun insolation, the modelled melting rate for rime on the unmodified panel is about 0.25 cm/hr whereas the melting rate for the TN Conseil modified panel is about 0.7 cm/hr.

The models reveal that the contributions of the Lexan back cover and the black absorber foil to the snow removal performance of the TN Conseil technology are very different. The Lexan cover limits heat losses and thus raises the temperature of the panel to the snow's melting point. The black absorber foil maximizes the energy gains and increases the melt rate. A system with one or the other component, but not both, would not function well, especially for accumulations that are well-attached to the panel.

One aspect of the TN Conseil modified panels melting performance that is not reflected in the models is the modified panel's tendency to form a ledge of densified snow or ice at the bottom lip of the panel. This ledge partially shades the panel and provides a foothold for further snow or ice accumulation. It is hypothesized that this ledge arises either from the temperature gradients along the panel or from the TN Conseil panel initiating melting under marginal conditions such that the ledge freezes onto the panel overnight. The use of Lexan Thermoclear, a corrugated transparent plastic, in place of the Lexan back cover sheet and associated air cavity, is proposed as a way to reduce the temperature gradients across the panel and decrease the modified panel's tendency to form an ice ledge. In addition, very simple mathematical models of the panel frame show that anodizing the aluminum black instead of silver, or more effectively, insulating the frame against heat loss, will also reduce the temperature gradients.

It is possible that rime will accumulate on the rear as well as the front surface of the panel. Rime accumulation on the rear face of the panel will decrease the radiation incident on the black absorber foil and reduce the efficacy of the TN Conseil system considerably, especially for thicker rime deposits. A solution to this problem, referred to as a "rime shield", has been proposed during the course of this investigation but is not described in this report.

Simulation of System Performance at Four Sites: A computer program that models the melting performance of the unmodified and TN Conseil modified panels at four sites across Canada using actual weather data was developed. A rime or snow accumulation of a certain thickness was assumed to exist on the panels on January 1, and the program determined how long it would take for the deposit to be removed, assuming no further accumulation occurred. For rime, it was assumed that an equally thick accumulation existed on both the front and the back of the panel. The initial accumulation was assumed to be very dense with high thermal conductivity and poor solar radiation transmission-- a type of accumulation which is rare but represents the most serious threat to PV systems.

The modelled performance for a site at Bagotville, Quebec, suggests that the TN Conseil modification is highly effective for snow, reducing the average time required for snow removal by about 95 % (if shedding occurs) to 80 % (if complete melt-off occurs). For rime, the model suggests that the TN Conseil technology is somewhat less effective. For mountainous sites in B.C. and Newfoundland, the average time to shedding was reduced by about 30 to 40 % and the average time to melting by about 20 to 30 %. The technology was less effective at a site just above the Arctic Circle in the Northwest Territories. However, if the parameters used to model the rime accumulation are incorrect, these results probably underestimate the improvement realizable with the TN Conseil technology.

The times required for melting and shedding for the TN Conseil panel with the rime shield are only slightly better than those for the TN Conseil panel without the rime shield. This is misleading, however, since it was assumed that rime would be removed from the rear face of the panel at the same rate that it was removed from the front face of the panel. In reality, however, the air cavity in the TN Conseil technology insulates the panel sufficiently well that the Lexan back cover does not heat up. This suggests that rime will not remove itself from the rear of the TN Conseil technology, and therefore the performance of the TN Conseil Technology without the rime shield will degrade over the winter as rime accumulates on the rear face.

The large number of assumptions and unknowns required to model the removal of rime and snow from the panel suggest that the results presented above can be used only as a rough comparison of thermal performance. A more accurate assessment will require experimentation, preferably in the field, at rime-prone sites.

Summer Battery Charging Performance: The TN Conseil technology will elevate panel temperatures during summer as well as winter. Hotter panels operate at lower voltages, suggesting that during the summer a single panel or parallel arrangement of panels might not be able to fully charge a 12 V battery. Mathematical models of the thermal behaviour of the panel were used to determine whether this would be a problem.

According to the models, during worst-case summer conditions of very low wind, peak sun, and high ambient air temperatures, the M55 panel and other 36 cell panels will be able to charge the battery, although the rate of charging may be lower. Under worst-case conditions, the M75 and other panels with 33 or fewer cells may be unable to fully charge the battery. However, in locations where snow and ice accumulation is a concern, it is unlikely that worst-case conditions will persist for the duration of the summer. At the EDRL, the highest monitored temperature reached by the TN Conseil-modified panels was 78 ºC. At this temperature, an M55 module would be capable of full charging.

Observed Short-term Operation of the Snow Removal Technology: The snow removal performance for March 6 and 7, 1995, of panels with and without the TN Conseil technology is examined. Results are presented in graphs of monitored panel current and temperature and a chronological series of photographs.