Double-Glass PV panels

[JohanVanR]

DanKegel #16: Thanks for taking the time for testing. Yep, a Watt or so more according to a quick spreadsheet check. This can be attributed to LG's higher (absolute) TCoP. In turn, this is probably largely related to its front glass / EVA / cell / backsheet thickness. You could measure that sum by measuring the offset from the frame top/bottom to the glass/backsheet and subtracting the offsets from the frame thickness of 40mm (1.57"). I estimate it to be ~5mm.
Anywayz too much detail analysis I guess. The main point is that dual glass does not 'by design' perform worse than conventional panels at elevated temperatures relative to STC/PTC.

One disadvantage of dual glass *could* be its 'hermeticity' [1] → EVA outgassing / acid content cannot diffuse through glass → bubbles and cell corrosion. However, I would guess that such phenomena then show up at certification testing [2].

[1] http://www.glass-international.com/c...ar_for_web.pdf
[2] http://tuvamerica.com/services/photo...standingPV.pdf

[DanKegel]

TCoP... is probably largely related to its front glass / EVA / cell / backsheet thickness. You could measure that sum by measuring the offset from the frame top/bottom to the glass/backsheet and subtracting the offsets from the frame thickness of 40mm (1.57&quot. I estimate it to be ~5mm.

How is TCoP related to thickness? I'm missing something. You may need to draw me a picture

One disadvantage of dual glass *could* be its 'hermeticity'

Also said to be an advantage in damp areas.

[JohanVanR]

DanKegel #18:
I assume that for obtaining the specified TCoP, a thermocouple was used and attached on the module back (outer) surface [1]. This implies that the TCoP does not correspond directly to the cell temperature anymore. In effect, the substrate 'insulates' the thermocouple from the cell. If the cell heats up (T_cell↑↑↑), then the cell intrinsically has a large power reduction (P↓↓↓). Simultaneously, the 'insulated' thermocouple records only a relatively small temperature increase (T_tc↑). In effect, this yields

TCoP = P↓↓↓ / T_tc↑ (large number divided by small number),

which is relatively large compared to that of a naked cell with a thermocouple attached directly *on* it,

TCoP = P↓↓↓ / T_cell↑↑↑

Hence, for an increasing substrate thermal resistance (thickness), the TCoP increases. If the specified TCoP was determined using IR, then my story may not hold. Correct me if I am wrong.

[1] http://www.osti.gov/scitech/servlets/purl/548687

[J.P.M.]

IMO only, the recent postings to this thread demonstrate a lack of understanding with respect to the engineering aspects of heat transfer.

I'd respectfully suggest a review of the basics, starting with the idea that thermal conduction is probably not the controlling mode of heat transfer involved here, followed by a review of the emissive and reflective characteristics of the materials involved, particularly as f(wavelength), especially for glass.

Take what you want of the above. Scrap the rest.

[DanKegel]

Understanding thermal performance in enough detail to predict whether a glass backsheet is a win or not is way beyond the basics.

[JohanVanR]

J.P.M. #20: Thanks for monitoring this discussion. Could you elaborate that a bit more with (references to) quantitative information in context of conventional VS dual-glass modules? For example including the radiation heat transfer component of the total system, assuming a one-dimensional model for the sake of simplicity? Simplified models are often useful and joyful to gain understanding, for example to get a feeling about the orders of magnitudes, or to have a rough reference to compare more complex models against, especially when experimental data is lacking. Plus, the *ability* to concisely explain something complex to a non-specialist using acceptable simplifications can be a sign of deeper understanding, beneficial to both the specialist and student as a teaching and learning experience, respectively.

[J.P.M.]

J.P.M. #20: Thanks for monitoring this discussion. Could you elaborate that a bit more with (references to) quantitative information in context of conventional VS dual-glass modules? For example including the radiation heat transfer component of the total system, assuming a one-dimensional model for the sake of simplicity? Simplified models are often useful and joyful to gain understanding, for example to get a feeling about the orders of magnitudes, or to have a rough reference to compare more complex models against, especially when experimental data is lacking. Plus, the *ability* to concisely explain something complex to a non-specialist using acceptable simplifications can be a sign of deeper understanding, beneficial to both the specialist and student as a teaching and learning experience, respectively.

You're welcome. As for elaboration, respectfully, no. Reason: I'm not a teacher and, IMO only, this forum is too limited a venue to do justice to the subject.

Any good undergraduate text in heat transfer should be a good start.

Meant in a respectful but serious way, my point was that you and Dan seem to be out of your depth discussing a subject. Fine to me, but IMO, you are both doing little more than wasting electrons, are mostly unaware of the subject matter or methodology necessary, writing about things that you know little about and are thus only confusing the issue more.

[inetdog]

IMO only, the recent postings to this thread demonstrate a lack of understanding with respect to the engineering aspects of heat transfer.

I'd respectfully suggest a review of the basics, starting with the idea that thermal conduction is probably not the controlling mode of heat transfer involved here, followed by a review of the emissive and reflective characteristics of the materials involved, particularly as f(wavelength), especially for glass.

Take what you want of the above. Scrap the rest.

I suspect that although radiation may be significant, at the temperature panels operate conduction through the glass with transfer to the ambient air will still be a large (or majority) part of the heat transfer total.
One problem with "off the cuff" analysis of this sort of thing is that we are conditioned to think of "double pane" glass as window glass designed to minimize conductive heat transfer and so provide better insulation for a house in winter.
The space between the panels is narrow enough to minimize convection in whatever gas fills the inner space and the thickness is chosen to provide high resistance to heat transfer using only an inert gas fill.
The panels described here sound very different, with the double pane designed more for its mechanical properties than for its thermal properties.

I have not looked at the manufacturers technical details, but am willing to be convinced.

[J.P.M.]

I suspect that although radiation may be significant, at the temperature panels operate conduction through the glass with transfer to the ambient air will still be a large (or majority) part of the heat transfer total.
One problem with "off the cuff" analysis of this sort of thing is that we are conditioned to think of "double pane" glass as window glass designed to minimize conductive heat transfer and so provide better insulation for a house in winter.
The space between the panels is narrow enough to minimize convection in whatever gas fills the inner space and the thickness is chosen to provide high resistance to heat transfer using only an inert gas fill.
The panels described here sound very different, with the double pane designed more for its mechanical properties than for its thermal properties.

I have not looked at the manufacturers technical details, but am willing to be convinced.

For starters, and still respectfully, I'd suggest not confusing a PV modules "temperature coefficient of power", that is, how much a module's efficiency decreases as f(temp.), with a module's operating temp. as may be determined by an energy (heat) balance.

If an energy balance is done on two side/side modules that are identical except that one has a glass backing and the other one has, say a tedlar/plastic backing, under quasi steady state conditions the glass backed module will probably have a higher temp. This is due to the back glass of the dual glass module having a lower external surface temp. That lower temp. is due to less thermal energy being transmitted through the back glass than through the tedlar/plastic of the other module. Reason: Less energy will be transmitted into the back glass than into and through the tedlar backing of the other module because of the higher reflectivity of the glass in the IR range as f(wavelength). Less energy will make it past the inner surface of the back glass because of the IR reflectance of the glass. That's the same reason why window glass is superior to other materials of widow glazing with respect to insulating or heat transfer characteristics.

To a first approx., assuming the same environmental conditions of wind vector, ambient temp. and irradiance (but, it is to be noted, not back surface temp.) for the side/side modules, the convective film coefficient (in units of W/m^2*deg. C.)of the back glass will be about the same as the tedlar backing of the other module. However, because the outer surface of the glass will be at a lower temp. for the reason given above, the overall heat transfer from the glass to the ambient air will be less. That will result in less overall energy (heat) transfer per degree of (entire) module temp., requiring a higher module "internal" temp. to maintain an energy balance.

For the temperatures involved, and assuming the wind vector is not too great, and to another first approx., the convective film coefficient of the glass and the thermal radiation coefficient (as normalized to a per deg. coeff.) of the glass (to what will probably be in many/most situations mostly to a parallel roof deck under the module) will be something like the same order of magnitude. However, the total rate of thermal radiation heat transfer from the back glass will be less than the thermal radiation heat transfer rate from the tedlar backed module to the roof deck, simply because the back glass will be at a lower temp. for reasons described above. Since the radiation heat transfer rate is proportional to the 4th power of the absolute temp. diff. between two surfaces, that will be an additional reason for increased module temp.

Overall, the glass is a better insulator. Better insulators inhibit heat transfer. Less heat transfer means more temp. diff. More temp. diff for the same ambient conditions means higher temperatures for the better insulated module.

Q.E.D.

As Inetdog notes, this is all an off the cuff analysis and leaves many things out such as non linear dependence of convective film coefficients which may/may not be constant as f(temp.), or the likelihood that the front surface of the glass backed module will have higher overall heat transfer due to higher overall module temps. However, without going into it with any depth, it seems clear to me that, to a first approx., glass backed or two glass surface PV modules will have a higher operating temp. than plastic backed PV modules under the same conditions. I suppose I could be wrong, but the above is perhaps the beginning of what a surface scratch of an engineering type analysis of the situation might look like as opposed to the recent comments.

As usual, take what you want. Scrap the rest.

[sensij]

Might be worth a read, pages 25-33 dealing with thermal modeling, page 29 showing the actual energy balance equations that could be solved for any panel construction of interest.

http://www2.jpl.nasa.gov/adv_tech/photovol/ppr_86-90/FSA%20Final%20Rpt%20VII%20-%20Encapsulation.pdf

From page 26:

(2) In terms of these thermal properties, module operating temperature is primarily regulated by the IR emissivity of the front and back surfaces, and to a lesser extent by thermal conductivity. This leads to the finding that a glass superstrate design and wood substrate design may have nominally the same solar cell operating temperatures.

(3) The dominant control on module operating temperature, which can be exercised through selection of encapsulation materials, involves the use of front- and backcover materials with maximum IR emissivity (E). Transparent glass and plastic-film front covers have E values ranging between 0.85 to 0.90. Backcover materials should also have very low solar absorptivity. The two requirements for the back cover are best satisfied using a white organic (nonmetallic) material. Values of E for white organic materials can be >0.90.

[J.P.M.]

Might be worth a read, pages 25-33 dealing with thermal modeling, page 29 showing the actual energy balance equations that could be solved for any panel construction of interest.

http://www2.jpl.nasa.gov/adv_tech/ph...apsulation.pdf

From page 26:

Thank you.

Yea, that's what I think I mostly said in a slightly different way, but without the wood.

[sensij]

Thank you.

Yea, that's what I think I mostly said in a slightly different way, but without the wood.

Ha! Yes, agreed. I think trying to solve a heat transfer problem without an energy balance is going to go off the rails quickly.

If I understood right, your off-the-cuff formulation also considered heat transfer via radiation through the EVA layer, ignored by the paper I linked, but perhaps justifiable given how transparent EVA is to most of the IR band. The paper had already demonstrated the relative importance of emissivity in just the outward facing surface of the panel, including that mode as you suggest pushes their conclusions even further in that direction.

The FEM paper posted by Dan earlier in the thread found no difference between glass and Tedlar, but also appears to ignore emissivity (and heat transfer via radiation) for all materials except the silicon and silver. It does note:

The color map for the NOCT condition using EVA as the encapsulating material is also given in Fig-11. The module has the highest temperature at the cell and the lowest at the top of the protective glass cover. The temperature difference between cell and glass surface is about 1°C in all simulations.

With only a 1 deg gradient across the panel materials, it is easier to understand how outward convection and radiation can dominate. That paper doesn't state what convection coefficients are used (in fact, no form of "convect" is used in the paper at all), so it is not hard to imagine that the FEM calculated NOCT temps that are close to real world observed values were obtained by gaming the model to create operating conditions that achieved the desired outcome:

First of all, the models are verified incorporating standard (common) operating conditions where the outcome can be speculated based on literature and experience.

[DanKegel]

I'd suggest not confusing a PV modules "temperature coefficient of power".

I think you misunderstood JohanVanR's post. I think was talking about which temperature tcop refers to -- on the surface of the panel, or at the surface of the cell.

[J.P.M.]

Ha! Yes, agreed. I think trying to solve a heat transfer problem without an energy balance is going to go off the rails quickly.

If I understood right, your off-the-cuff formulation also considered heat transfer via radiation through the EVA layer, ignored by the paper I linked, but perhaps justifiable given how transparent EVA is to most of the IR band. The paper had already demonstrated the relative importance of emissivity in just the outward facing surface of the panel, including that mode as you suggest pushes their conclusions even further in that direction.

The FEM paper posted by Dan earlier in the thread found no difference between glass and Tedlar, but also appears to ignore emissivity (and heat transfer via radiation) for all materials except the silicon and silver. It does note:



With only a 1 deg gradient across the panel materials, it is easier to understand how convection and radiation can dominate. That paper doesn't state what convection coefficients are used (in fact, the word "convection" is never used in the paper at all), so it is not hard to imagine that the FEM calculated NOCT temps that are close to real world observed values were obtained by gaming the model to create operating conditions that achieved the desired outcome:

Thank you.

The easiest way to comment on this is to refer to Duffie & Beckman, chap. 3 , selected topics in heat transfer.

It's hard to go wrong using an energy balance as long as steady or quasi steady state conditions can be used and any energy storage or internal generation is accounted for.

What is accounted for is energy that crosses the "control surface" or imaginary boundary that is placed around the module/array/whatever body is under consideration. What happens inside the control volume such as heat transfer between components,while certainly important, is a somewhat separate issue provided there is no energy generation taking place inside the control volume such as from, say a combustion process.

Emissivity (and absorptivity for that matter), either wavelength dependent or gross (summed over all wavelengths and then combined) is actually a measure of how close a material approximates black body or ideal behavior. With respect to how much of the heat transfer of a body is due to convection vs. radiation, a dirty little secret is that, to a 1st approx. the entropy increase (change) from both processes will be approximately the same - that's not to say the actual quantity of heat transfer will be the same or that the convection and radiation coefficients will be numerically equal. The trick involves finding the representative temp. diff. for both processes. For example, the convective heat transfer from the front glass will be to the ambient environment, while the radiation loss from the front glass rwill be to a combination of temps. of the surrounding environment - trees, buildings. supports, etc., and the radiant sky temp. which will be different.

The convective loss from the back surface will be to the ambient temp., while the radiative loss will be mostly to the under module roof deck which will probably be at a temp. (and confirmed to me by my fouling experiment temp. measurements of under array roof deck temps.) somewhere between the ambient air temp. and the back of the module. That's where things get interesting and iterative.

[JohanVanR]

FWIW: Today I implemented the thermal model from DOE/NASA [1] (thank you sensij) in C#.NET with a non-linear solver [2] and 1e-6 precision.
I ran it for various modules/problems, including those shown in [1 Fig.18] for checking the code. Preliminary results using this model show:

• only a few degrees temperature drop from cell to module outer surfaces, corresponding to what sensij quoted,
• radiation heat transfer cannot be neglected, J.P.M. is correct about that, I follow his reasoning there (~40% @ 1m/s wind),
• the fraction of the forced-convective part varies *a lot* with the wind speed. Huge sensitivity there,
• Dan's Sunpreme (glass/glass) and LG (Tedlar(?)/glass) modules show roughly *the same* cell temperature of 58ºC @ 1000W/m2 insolation | 20ºC ambient | 1 m/s wind | 0º tilt | 0W elec. power | etc. Kind of matches Dan's IR measurement also showing a small difference.

Notes about [1]:

• Table 18 header 2 and 3 should be swapped. People not used to US customary units can get confused there .
• It seems that Figure 18 was created using T_sky = T_ambient instead of T_sky = 0.914 * T_ambient (or I made a mistake)

[1] http://www2.jpl.nasa.gov/adv_tech/ph...apsulation.pdf
[2] http://www.alglib.net