Optimising system size from 30min interval consumption data, available area, etc

[Jordn]

Hello everyone,

I'd like to talk about optimising solar pv system sizes for commercial scale projects.

When there is 30min interval consumption data, and information on constraints such as; available area and export limits - planning the right balance between system size and financial payback/ROI can be a slow and iterative process. The way I see optimising solar pv system sizes for commercial scale projects is by making simulation, after simulation, after simulation of different system sizes in an effort to arrive at the golden system size.

I was wondering if there are any methods via PVSol, or external to PVSol, that can speed up the process of arriving at the golden system size?

I guess this begs to ask what makes a golden system size? In this case, the projects are commercial financially-single-minded projects, I would suggest optimal payback/ROI would be the measure...

[developer_fw]

Hi Jordn,

thank you for the input! As always this is of utmost value for us. There is no optimization process so far, but within our team this topic comes up from time to time. I'll put this on our agenda.

Best regards,
Frederik

[Jordn]

Thank you @developer_fwany updates on this please? @developer_mh

Is there a proven method to find the golidlocks system size using HHD data?

[RaW]

Updates to this would be appreciated.

 

As mentioned, multiple simulations are required to find the sweet spot between system size and financial payback/ROI.  Is there an easy way to do this in PVSOL, or if not easy, what is the recommended steps to ensure success.

 

Maybe there is a better way external to PVSOL?

 

Thanks for your help with this.

[RaW]

Is this something that is likeley to be answered anytime soon?  I ask because we are nearing the end of our trial..

 

We have got to grips with the software we think, but knowing how to fundimentally size solar PV arrays along with their batteries using PV SOL is still witchcraft.

 

If it helps you (I'm sure it would help others too using PV SOL), I'm happy to detail a current job we are modelling here so the process can be documented.  NOT how to use the software, but how to get the right 'balance between system size and financial payback/ROI'.

 

Looking forward to hearing from you soon.

[developer_mh]

Dear RaW, dear Jordn,

in order to be able to understand exactly what kind of optimisation process you are looking for, it would be very helpful if you could provide an detailed example of how you manually do those ROI optimisations right now.

Kind regards,

Martin

[RaW]

Hi Martin,

 

Thanks for the reply.  To be honest we don't.  We give a best guess but are normally lead by the client who comes to us to fill their roof with solar panels or to compete costwise with a similar design and cost they already have.

 

We are looking to actively change this for our clients to ensure their best ROI, and not to just line our pockets with the biggest array we can install.

 

We have a sizeable opportunity that is lead by investors, to size, install and maintain commercial solar PV systems along with battery storage where required.  We want our advice to be the best it can be in regards to ROI and have a sound arguement if our recommendations are questioned.

 

How best to proceed to ensure Valentin's expertise and 25 years in the market helps us to perfect the ROI process for our clients?

 

Thanks.

[RaW]

Would me detailing an up and coming job here so we can discuss it in more detail help to iron out a best ROI process?

 

Thanks.

[developer_mh]

Hi RaW,

thank you for the reply. I am sorry, but I don't understand that sentence:

14 minutes ago, RaW said:

Would me detailing an up and coming job here so we can discuss it in more detail help to iron out a best ROI process?

Do you mean you would present an upcoming job here so that we can discuss the best solution for the customer here? That would be great, I guess!

Also, I doubt that pure ROI can be a suitable target for an optimization process. The ROI will always be high if your investment is relatively small, for a small PV installation that produces so little energy that everything can directly be used to cover the loads. Say you have a consumption on site of 25000 kWh and you install a 1kWp PV system for around 1300 €, your ROI will be extraordinary (in percent), but the cash balance in € will be negligible.

Kind regards,

Martin

[RaW]

Hi Martin,

I'll get some information together and post it here for some feedback and discussion..

I fully understand your point in regards to a tiny system on a large load.  I'm talking more about finding the drop off point where increasing the array size doesn't make sense costwise (even longterm) and finding the sweet spot for investment cost over energy cost savings.

I'll detail an example energy usage scenario here soon and fill the available roof space with soalr PV panels and hopefully raise some feedback.

[RaW]

So, here is the most basic of data to get the ball rolling.  This is the type of enquiry we are getting more and more of. 

Although there are questions that we ask to expand upon their request (planned growth of business, funded or CapEx, etc.) if we can go through the process to find the sweet spot with the information we have, that would be great.

 

Commercial property in the Midlands - UK:

 - Energy usage attached (3 phase, NOT half hourly, showing day/night usage)

 - Roof space for maximum 544 panels, shade free

 

Client is asking for designs and costs for the appropriate array size that will give him the best energy savings balanced with the cost of installation.

They would like:

 - 1 x design of solar PV array

 - 1 x design solar PV array and battery

 

Currently, for current stock and availabilty reasons we are modelling using:

DMEGC Solar panels - DM410M10-B54HBB

Solax T-BAT-SYS-HV-5.8T58 Batteries

Solax X3-Hybrid-G4 Inverters

Presume £700 per kW install costs

 

Thanks.

Edited January 20 by RaW
attachment

[RaW]

To reignite the discussion, how best to proceed to advise the client on the best ROI for them using the information we have only?

Based on the information given, what is the sweet spot for them (between 0 - 544 panels) investment-wise (@£700 per kW)?

Where would this sweetspot be if we were to introduce batteries into the equation?

I know, as a company, you have over 25 years experience in this field modelling solar PV arrays for commercial clients, so thank you in advance for your insight into the above.

[RaW]

Request to continue this please.

[developer_mh]

Hi RaW,

in order to make things easier, would it be possible for you to provide the consumption as Excel file (or csv).

Thanks and kind regards,

Martin

[RaW]

42 minutes ago, developer_mh said:

Hi RaW,

in order to make things easier, would it be possible for you to provide the consumption as Excel file (or csv).

Thanks and kind regards,

Martin

Hi Martin,

 

Thanks for the quick reply.  Excel attached.

Electic bills 2021-22_1.xlsx

[RaW]

Hi Martin, I hope you had a good weekend. 

Is there anything else I can help with in regards to demonstrating the recommended 'sweet spot' process using PVSOL please?

Thank you.

[developer_mh]

Hi RaW,

thanks a lot for the data. I had a look into it and made a simple series of simulations in order to find a good optimization target. As pointed out in my previous post, the ROI is not a very good measure as it is highest for small PV installations and decreases continously when the installation size increases. The accrued cash flow however might be a good starting point to optimize for.

For my small experiment I created a project with your data and simulated the same project over and over (automatically, which is possible from our code), with varying PV system size. The installation costs are in £/kWp, so they scale automatically. I also tried different load profiles, and scaled each load profile once for 49 MWh and once for 100 MWh per year.

Of course this is only a quick and very sketchy analysis, but it turns out that the optimum point is somewhat stable in the range between 0.6 to 2 kWp solar per MWh of load. That means if you have 49 MWh of annual consumption you could go for a PV system size of 30 to 100 kWp.

So, in the example of the load profile you provided, with 49 MWh of annual consumption, DMEGC 410 Wp modules, Solax hybrid inverters (and the arbitray location Manchester), the optimum PV system size is 180 panels (= 74 kWp). If the annual consumption doubles to 100 MWh, the PV system size will double, too.

If the type of load profile varies, the sweet spot will move, too. It is always specific to the project where the optimum system size will be. But knowing that there is an optimum somewhere, presumably around 1 kWp/MWh, you could start iterating around this point, back and forth. Finding the sweet spot like this should be relatively fast.

This all is without taking into account battery storage, this would be too extensive for me right now, but I guess the process would roughly be the same.

I'd be happy if you could take a minute and see if this kind of optimization process is what you were looking for. Especially if the accrued cash flow is the right optimization target. I could then talk about that with my colleagues and think about ways of implementing this into a feature.

Kind regards,

Martin

 

 

 

[RaW]

Thank you very much, I really appreciate the time taken so far.

If you don't mind, I'll go through the thread asking some questions and hopefully not only will it help me, it will help others now and historically - a boon for PVSOL too.

Quote

thanks a lot for the data. I had a look into it and made a simple series of simulations in order to find a good optimization target. As pointed out in my previous post, the ROI is not a very good measure as it is highest for small PV installations and decreases continously when the installation size increases. The accrued cash flow however might be a good starting point to optimize for.

I understand this will benefit larger installations as opposed to smaller PV arrays.

Has this table been created using PVSOL?  Is it avialble in the trial version?  If so, could you highlight where to enter this information and where to output the graphs please?

 

Quote

For my small experiment I created a project with your data and simulated the same project over and over (automatically, which is possible from our code), with varying PV system size.

Where can I find this feature in PVSOL?

Quote

The installation costs are in £/kWp, so they scale automatically. I also tried different load profiles, and scaled each load profile once for 49 MWh and once for 100 MWh per year.

Can you explain why 49MWh and 100MWh were used specifically please?

 

Quote

Of course this is only a quick and very sketchy analysis, but it turns out that the optimum point is somewhat stable in the range between 0.6 to 2 kWp solar per MWh of load. That means if you have 49 MWh of annual consumption you could go for a PV system size of 30 to 100 kWp.

Can you please go into a little more details in regards to how this optimum point was found please?  Also, 30 to 100kWp is quite a difference panelwise?

Quote

So, in the example of the load profile you provided, with 49 MWh of annual consumption, DMEGC 410 Wp modules, Solax hybrid inverters (and the arbitray location Manchester), the optimum PV system size is 180 panels (= 74 kWp). If the annual consumption doubles to 100 MWh, the PV system size will double, too.

Perfect, again, if you could go into a little more detail in regards to how these figures were ascertained that would be fantastic.

Quote

If the type of load profile varies, the sweet spot will move, too. It is always specific to the project where the optimum system size will be. But knowing that there is an optimum somewhere, presumably around 1 kWp/MWh, you could start iterating around this point, back and forth. Finding the sweet spot like this should be relatively fast.

I think once the above questions are answered, having any kind of process that can be shown to the client that outlines your reasons as a designer of suggesting PV array sizes is great.

Quote

This all is without taking into account battery storage, this would be too extensive for me right now, but I guess the process would roughly be the same.

When you have time, maybe adding this process (battery) to the thread would be great.

Quote

I'd be happy if you could take a minute and see if this kind of optimization process is what you were looking for. Especially if the accrued cash flow is the right optimization target. I could then talk about that with my colleagues and think about ways of implementing this into a feature.

The accrued cash flow seems an honest way of designing larger PV arrays to me - thank you.  One of the reasons for starting this thread is that I have seen over and over again where the array size isn't really thought about by designers (competitors) and I want to make sure any process we are involved in DOES what is best for the client, and can some way to explaing WHY an array size is suggested.  I like PVSOL and would like to use it as a one stop shop in the near future for all of my solar pv needs.  This undestanding goes a long way to making sure any software investment we make is sound.  Thank you so much for your time so far, it really is appreciated.

I look forward to hearing from you.

[developer_mh]

Hi RaW,

On 1/30/2023 at 6:08 PM, RaW said:

I understand this will benefit larger installations as opposed to smaller PV arrays.

Yes, but it will not favour PV installations of unlimited size. As you can see in the blue curve, there is an optimum - that is what we will want to aim for.

On 1/30/2023 at 6:08 PM, RaW said:

Has this table been created using PVSOL?  Is it avialble in the trial version?  If so, could you highlight where to enter this information and where to output the graphs please?

No, this is not yet available in the official PV*SOL version that our customers can buy. I have generated these figures and graphs directly from our code. Our goal would be to fully understand the optimisation process and then come up with a solution that is helpful to our customers and doesn't require a complete rewrite of our software.

 

On 1/30/2023 at 6:08 PM, RaW said:

Can you explain why 49MWh and 100MWh were used specifically please?

The 49 MWh came from the load profile you provided (48884 kWh to be precise). The 100 MWh was just an arbitrary choice to roughly double the consumption.

On 1/30/2023 at 6:08 PM, RaW said:

Can you please go into a little more details in regards to how this optimum point was found please?  Also, 30 to 100kWp is quite a difference panelwise?

In my simulations, I can go from 10 to 500 panels, simulate each system and write down the results. The aim would be for our customers to be able to do this automatically in the future. At the moment, as this feature is not yet available in PV*SOL, the user would have to iterate manually to find the perfect system size. My point was that once you know that there is an optimum point somewhere between 0.6 and 2 kWp/MWh, you can speed up the process dramatically.

Let's look at this in more detail, let's say we have a consumption of 49 MWh, load profile G2 (which corresponds to the yellow line in the diagram above):

1. You start with 0.6 kWp / MWh = 29.4 kWp (approx 72 PV modules of 410 Wp each)
2. You simulate and get an Accrued Cash Flow of around 32 000 ₤
3. Then you go with 0.7 kWp / MWh = 34.3 kWp (approx 84 PV modules)
4. You simulate and get 32 600 ₤
5. Now you already know that the optimum will be at a point larger than 0.7
6. You repeat the same for the two points 2 kWp / MWh and 1.5 kWp / MWh which will give you 21 700 ₤ and 28 000 ₤ respectively
7. This will give you these for points:
   
8. You can now calculate (or estimate) the point where the two lines through the pairs of points will intersect. In this case it is 0.95 kWp/MWh:
   
9. Now you calculate the cash flow for 0.95 kWp/MWh, and you get approx. 31 900 ₤ and the diagram looks like this:
   
10. Now you know your optimum is between 0.7 and 0.95 kWp / MWh. Determine the next point, 0.77 in this case is where the two lines intersect:
    
11. Simulating with 0.77 kWp / MWh (i.e. 37.7 kWp or 92 PV modules), the Accrued Cash Flow is around 32 800 ₤, which is already a good optimum:
    

 

Of course, this still involves some manual work, but you don't have to iterate through all the possible system sizes. In this case, six simulations are sufficient. When you write up the Accrued Cash Flow results for your client and then show them the diagram, you will have a really convincing way of explaining why you chose this size of system and not a larger or smaller one.

We will also discuss how and when we can implement a first simple version of the optimisation in PV*SOL. But as we already have a lot of topics on our list, I can't give you a date at the moment.

I hope this helps, best regards,

Martin

 

 

 

[RaW]

Thank you Martin, this is really appreciated.

Quote

No, this is not yet available in the official PV*SOL version that our customers can buy. I have generated these figures and graphs directly from our code. Our goal would be to fully understand the optimisation process and then come up with a solution that is helpful to our customers and doesn't require a complete rewrite of our software.

It's one I'm going to keep an eye on for sure as I think this aspect is key in designing.  From my limited knowledge this DOESN'T happen currently with the majority of PVSOL designs submitted.

Quote

The 49 MWh came from the load profile you provided (48884 kWh to be precise). The 100 MWh was just an arbitrary choice to roughly double the consumption.

The load profile I submitted was for 98692 kWh (at £14,755.68) for the year.  There is a mention of 48934 kWh on the file which is confusing, but is just their internal estimate before the correct readings were ascertained.  Sorry if this has confused the issue.

Quote

In my simulations, I can go from 10 to 500 panels, simulate each system and write down the results. The aim would be for our customers to be able to do this automatically in the future. At the moment, as this feature is not yet available in PV*SOL, the user would have to iterate manually to find the perfect system size. My point was that once you know that there is an optimum point somewhere between 0.6 and 2 kWp/MWh, you can speed up the process dramatically.

Let's look at this in more detail, let's say we have a consumption of 49 MWh, load profile G2 (which corresponds to the yellow line in the diagram above):

1. You start with 0.6 kWp / MWh = 29.4 kWp (approx 72 PV modules of 410 Wp each)
2. You simulate and get an Accrued Cash Flow of around 32 000 ₤
3. Then you go with 0.7 kWp / MWh = 34.3 kWp (approx 84 PV modules)
4. You simulate and get 32 600 ₤
5. Now you already know that the optimum will be at a point larger than 0.7
6. You repeat the same for the two points 2 kWp / MWh and 1.5 kWp / MWh which will give you 21 700 ₤ and 28 000 ₤ respectively
7. This will give you these for points:
   
8. You can now calculate (or estimate) the point where the two lines through the pairs of points will intersect. In this case it is 0.95 kWp/MWh:
   
9. Now you calculate the cash flow for 0.95 kWp/MWh, and you get approx. 31 900 ₤ and the diagram looks like this:
   
10. Now you know your optimum is between 0.7 and 0.95 kWp / MWh. Determine the next point, 0.77 in this case is where the two lines intersect:
    
11. Simulating with 0.77 kWp / MWh (i.e. 37.7 kWp or 92 PV modules), the Accrued Cash Flow is around 32 800 ₤, which is already a good optimum:
    

Thank you, the process is clearer.  Can I break it down into some smaller questions please?

1. Are these figures based on 20 or 25 years?

2. As a guide when setting Annual Average Return On Capital Employed, what should this be set to as a guide?

 

Also, as a slight aside, when running these tests is there a simple way of selecting module areas you want in the design each time?  Using this as an example, if I filled the roof with my initial design, but then want to run the above scenario, how best to select separate areas each time to model?  I can't find a selector or tick box for each area to include, or not?

 

Quote

Of course, this still involves some manual work, but you don't have to iterate through all the possible system sizes. In this case, six simulations are sufficient. When you write up the Accrued Cash Flow results for your client and then show them the diagram, you will have a really convincing way of explaining why you chose this size of system and not a larger or smaller one.

I totally agree.  The accrued cash flow graphs can be showed for each simulation to highlight why your recommendations were made - perfect.

Quote

We will also discuss how and when we can implement a first simple version of the optimisation in PV*SOL. But as we already have a lot of topics on our list, I can't give you a date at the moment.

I understand, although I do believe this feature is very important and if advertised correctly, could really help PVSOL sales..

Again, I look forward to hearing from you and really appreciate you working with me on this - thanks.

[RaW]

Hi again, any updates at all please? 

Also, is there an easy way to select which elements of the design are used for each calculation?  For example, if I create a design that fills the available roofspace with many panels over mutiple strings and then want to add/remove panels for each simulation, is there an easy way to do this?

Thank you.

[RaW]

Hi again,

Apart from the questions above (which would be great if you could answer)

So I have been testing alternative methods for ROI (accrued cash flow) using software from other companies, and although possible, I am keen to invest in PVSOL.

I really hope you release a method of doing this very, very soon. 

You have been very detailed in your replies, and for that I am grateful, but could you please help me to undertstand some of the finer points so I can test PVsol in it's current state?  Some of my questions may be very simplistic, for that I apologise.

I think some of the problems I have is because sometimes you mention MWh when I think it should be KWh and also the energy usage over the 12 months is 98692 KWh not 48934KWh.

1. The graph 'optimizing system size for accrued cash flow':

What exactly are you plotting there?  Please answer using the data I have specified if possible.
 

You then go on to state 'that means if you have 49 MWh of annual consumption you could go for a PV system size of 30 to 100 kWp'.  the 49 MWh aside (should be 99 KWh?), how have you come to this conclusion?

Quote

So, in the example of the load profile you provided, with 49 MWh of annual consumption, DMEGC 410 Wp modules, Solax hybrid inverters (and the arbitray location Manchester), the optimum PV system size is 180 panels (= 74 kWp). If the annual consumption doubles to 100 MWh, the PV system size will double, too.

Again, the obvious aside, I can't see how you garner the information of 180 panels (74kWp) from the graph? 

 

Quote

 

In my simulations, I can go from 10 to 500 panels, simulate each system and write down the results. The aim would be for our customers to be able to do this automatically in the future. At the moment, as this feature is not yet available in PV*SOL, the user would have to iterate manually to find the perfect system size. My point was that once you know that there is an optimum point somewhere between 0.6 and 2 kWp/MWh, you can speed up the process dramatically.

Let's look at this in more detail, let's say we have a consumption of 49 MWh, load profile G2 (which corresponds to the yellow line in the diagram above):

1. You start with 0.6 kWp / MWh = 29.4 kWp (approx 72 PV modules of 410 Wp each)
2. You simulate and get an Accrued Cash Flow of around 32 000 ₤
3. Then you go with 0.7 kWp / MWh = 34.3 kWp (approx 84 PV modules)
4. You simulate and get 32 600 ₤
5. Now you already know that the optimum will be at a point larger than 0.7
6. You repeat the same for the two points 2 kWp / MWh and 1.5 kWp / MWh which will give you 21 700 ₤ and 28 000 ₤ respectively
7. This will give you these for points:

 

1.  Where does the 0.6kWp come from?
2. How does 72 panels equate to £32,000?

Would love clarification on these steps too.

I really want to know this process inside out.  I understand there is no easy way to do this currently and it means remodelling each time which will take a long time, however if I can successfully do this, I am happy to do this to ensure clients get the best modelling service from me when looking at jobs for them.

Please help me understand the above - thank you.

[developer_mh]

Hi RaW,

Sorry for not getting back to you earlier. Short explanation/escuse: This forum here is a free service for our customers and run by us developers, in addition to our professional support team. So here we can only answer when our developer schedule allows. This very topic here, as interesting as it is, has already received quite a bit of attention, so I wasn't able to answer earlier. Hope you understand (but nevertheless I understand how urgent this topic might be for you!)

So, I will try to go through your questions from the posts above:

On 2/2/2023 at 11:11 AM, RaW said:

The load profile I submitted was for 98692 kWh (at £14,755.68) for the year.  There is a mention of 48934 kWh on the file which is confusing, but is just their internal estimate before the correct readings were ascertained.  Sorry if this has confused the issue.

Actually, the exact value of the energy consumption does not matter. What I wanted to show you is an approach that enables you to find the sweet spot yourself, for any arbitrary annual consumption.

On 2/2/2023 at 11:11 AM, RaW said:

Are these figures based on 20 or 25 years?

25 years in this example

On 2/2/2023 at 11:11 AM, RaW said:

As a guide when setting Annual Average Return On Capital Employed, what should this be set to as a guide?

This is a tough question, and no one really knows The "Annual Average Return On Capital Employed" is like the average interest rate that you would get from your bank institute for the next 25 years if you would invest the money instead of buying a PV system.

On 2/2/2023 at 11:11 AM, RaW said:

Also, as a slight aside, when running these tests is there a simple way of selecting module areas you want in the design each time?  Using this as an example, if I filled the roof with my initial design, but then want to run the above scenario, how best to select separate areas each time to model?  I can't find a selector or tick box for each area to include, or not?

For finding the sweet spot I would recommend to create a separate project in 2D (without 3D that is). Altering the number of modules and the inverter configuration is a lot faster.

 

On 2/22/2023 at 5:18 PM, RaW said:

I think some of the problems I have is because sometimes you mention MWh when I think it should be KWh and also the energy usage over the 12 months is 98692 KWh not 48934KWh.

As mentioned above, the exact value for the consumption is not important for the process of finding the sweet spot. After all, the process should work for every future project, not just for this one. Sometimes I write the consumption energy in MWh because it is shorter than the same value in kWh. Sorry if that caused confusion.

On 2/22/2023 at 5:18 PM, RaW said:

What exactly are you plotting there?  Please answer using the data I have specified if possible.

I designed to initial project, with the climate data, consumption profile, PV module type, financial parameters and so on. Then I just varied the amount of PV modules, configured automatically with the inverters, simulated and wrote down the Accrued Cash Flow. Then I took 2 modules more, configured, simulated and wrote down the new Accrued Cash Flow, and so on.

 

On 2/22/2023 at 5:18 PM, RaW said:

You then go on to state 'that means if you have 49 MWh of annual consumption you could go for a PV system size of 30 to 100 kWp'.  the 49 MWh aside (should be 99 KWh?), how have you come to this conclusion?

This is just where the maximum of the Accrued Cash Flow is in the plots. Let's say the maximum is at 0.75 kWp solar / MWh load (the yellow and grey dots in the graph above). In this example I had a consumption of 49 MWh (48884 kWh), which leads to 0.75 * 49 = 36 kWp of solar.

On 2/22/2023 at 5:18 PM, RaW said:

Where does the 0.6kWp come from?

This is the value where the maximum of the graph lies for the grey curve (approximately).

On 2/22/2023 at 5:18 PM, RaW said:

How does 72 panels equate to £32,000?

The £32000 was the accrued cash flow for this example.

 

Hope that helps, kind regards,

Martin

[RaW]

Hi Martin,

Although I thanked you via PM I thought I should do so here too.

From what you have said in regards to replying in the forum, it looks like I should be taking this up with support rather than posting here (and I will..)

With regards to all the information I have garnered from this thread and by researching, this would be a very easy feature to add to PVSOL so I can't understand why it hasn't been already?  I realise you will have an answer for this, but really?

From chatting on the internet with others there is a need for this, and it is growing (showing due diligance to clients is huge) and I know of at least one developer who is working on this to create software that creates the graphs you have shared automatically.

Just food for thought really.

Thanks again for your help, it's appreciated, especially as this is a free service that can only be answered in your spare time rather than a product support forum, which I presumed it was.