Optimising Renewable Energy Models: webinar recording, handout and Q&A


Download the webinar handout here, and find answers to questions raised during the webinar below.

Will you be making the model example available?
– We don’t plan to, but you can come on our classroom course to learn how to build your own or sign up to be notified when the online course is ready.

Is the shareholder IRR an input or an output? Or does it not have any effect and is just for reference? 
– Both. It’s our target and we adjust the price at P50 to achieve that. At P90, in our case, we ignore it and look at the lender’s metrics.

Could you talk us through the mechanism to find the optimal target ADSCR efficiently? Many thanks!
– This is possibly a topic for another webinar (let us know if you would be interested in attending one like this).

What ADSCR is typically required by Lenders for the P90 scenario?
– Good question and the answer is wholly specific to the region and the novelty of the project. As wind characteristics of a site / region become known, and as experience in the sector grows, then ratios are dropping. That drop is also influenced by banks’ eagerness to participate, of course. In some areas, notably the US, a target ADSCR at P90 is not the measure but rather the project is priced on P99 with an ADSCR of 1.

How do you consider pre-completion revenues (revenues generated before reaching COD)? 
– There are 2 options: as cash flow that reduces the total funding requirement thereby reducing both debt and equity contributions, or (and what we are seeing on the majority of projects currently) treating such PCRs as equity which means having them circumvent the funding requirement calculation and letting them flow directly to equity either directly during the construction phase or through escrow into a payment at the end. How this is perceived by lenders is project-specific, but you can see it serves to increase leverage to the benefit of equity providers.

Can you explain the drop in average DSCR on the graphic at the end of the project?
– See the answer on tax below.

I had a question around how you make the model dynamic enough to keep a repayment profile of 18 years in both P50 and P90 cases?
– We’re letting the DSCR float up and down under the P90 case and fixing the leverage of senior debt during the construction phase. We always try to extend the tenor to its maximum then goal seek the DSCR to a level that sees debt repaid 100%.

How do we make the model sculpting dynamic enough to keep the repayment at 18 years when we switch between P50 and P90? This is assuming the tenor of 18 is constant.
– Let’s be clear that at P50 and P90 there is no difference to the repayment. They are one and the same. From the model’s perspective we are freezing the repayment as hard coded numbers so that when we switch to P50 the model doesn’t try to adjust those repayments to the new cash flows.

Do you have a macro for freezing the debt payment when the switch is made?
– It’s very simple. We simply make the fixed line of repayments equal to the “live” line:
Range(“fixed”).value = Range(“Live”).value

Just curious: is this P90 debt P50 equity common to analysing other sectors?
– Great question. It’s common across renewables, yes, but practically unused elsewhere in project finance. If you consider what P50 & P90 are, in one sense they’re merely upside and downside cases – common terms across PF. Where P50 and P90 differ of course is that they are confidence levels / percentile ranks which are based upon statistical analysis of observations. You could easily apply such an approach to transport projects, if you were taking volume risk, or indeed any project where volume was factor in your pricing / planning considerations. We’ve used P50 and P90 concepts on a number of programme planning exercises and also in terms of analysis of recruitment to look at delivery timescales and the variance therein.

Debt cover ratio goal seems to be able to create stability in the ratio… but if most renewables power generation is highly variable, could we see the part of the model which shows how volatile revenues get converted into stable debt cover ratio?
– Volatility short term is dealt with by having the ratios measured over a sufficient time period that the high and low periods can cancel each other out. Long period cycles in wind (I’m thinking El Nino and its 11 year cycle) are typically ignored.

Can you explain why the ratios are falling drastically when the generation under P50 has increased and there is a fixed repayment profile?
– Ratios as a whole are increasing. What we are seeing is purely the impact of tax. The cash flows are effectively shielded from additional tax in the early years due to losses. When these run out earlier, due to higher earnings, the sudden larger tax payment depresses the DSCR. The same effect, smoothed due to forward looking discounting, is apparent in the LLCR.

A workaround on such sudden dips is to ensure that some cash reserving has occurred prior to the large tax payment becoming due and that releases from such a reserve count towards the calculation of CFADS within a credit agreement. Alternatively, if the tax payment doesn’t depress the DSCR below lock-up – ignore it!

I was interested that the objective in the P90 lender profile was so strongly around Debt Coverage – makes perfect sense of course but I was wondering how to reconcile it with my (perhaps faulty) understanding that most renewables have variable power generation (therefore presumably, even with some dynamic / seasonal pricing element, variable cash flows)??

One element I could see being in debt repayment events being suitably spaced to minimise variability. So six-month periods Jan-June and July-Dec, for example, would insure somewhat against oscillating solar power generation… Beyond that, I was wondering what the convention was about possibly dealing with such variability using Monte Carlo: running thousands of iterations with values that were stochastically set (within distributions defined by experts), and then producing probabilities that DSCRs would fall within given values – then defining an allowable probability.

The creation of two different scenarios with deterministic constants for how much power seems to suggest this model, at least, would not use such techniques. I wondered though what your thoughts were about such kinds of techniques applied to solve the variability question.

– We’re making some simplifications in the demonstration model but nonetheless renewables projects – and indeed any project where either or both of revenues and costs are volume linked / variable – will lead to variations in the resultant DSCR. The key thing to bear in mind with optimising is that in the first instance we are fixing the variability, and in our case we are fixing that at a known P90 level of future generation. Once the model is solved and we are notional past financial close, we can then lock the debt repayments and note how the DSCR moves up or down period for period based upon variations in generation.

Having debt service periods of 6 months helps greatly in smoothing out the seasonal and monthly variations that may occur in generation. The extent to which such a six-monthly timing helps is directly proportional to the variability of generation across the seasons (low variability for sun shine in the south of Spain vs high variability for wind speeds in the North Sea). We’ve seen instances where projects have pushed towards quarterly timing only to then leave themselves reliant on other reserving measures that seek to smooth out the variable cash flows that, once the model moves from planning to operations, invariably mean that the cyclical timing once envisaged during planning no longer fits the debt service profile.

A further means of dealing with such variable cash flows is the use of cash sweeps with triggers linked to periods of low generation. Such a sweep may be written into the contract or may equally be viewed as employing operational cash planning prudence as financial controllers ensure monies are rigorously reserved prior to any debt service / debt service cover ratio calculation period.

To answer your last point – we kept the model simple for the webinar but either scenario could be locked and the wind generation adjusted through a range of “P” values or through sensitivities that adjust the seasonality of let us call it the “base case” to that of another case.

Shouldn’t DSRA deposit come before withdrawals? 
– An initial deposit at the end of a construction phase for example should be made, however thereafter the deposits and withdrawals would happen at the same time so effectively the net movement would be passed through to the cash flow.

Please give us examples of variables that only affect one case (e.g. P50) as well as both?
– Cash sweeping and / or debt refinancing you might consider as two key variables that affect both P50 and P90 case. It’s important however to consider at what stage you might want to turn either of these two variables on. While “solving” the model, you would probably want to keep the Cash Sweep disabled and only enable it when you’ve reached a optimised solution “pre-cash sweep”. The same might be true of a refinancing solution. You might want to target a specific IRR at P50 only when the refinancing is enabled.