The Solar Developer’s Modeling Challenge: How Much Detail Is Enough?

In the article below, Chris Lord, HeatSpring’s expert instructor, outlines when and why detail in modeling cash flows from a renewable energy project really matters.

Key Takeaways:

• Modeling forces us to confront the troubling “unknowns” about our project, and that implicates two curious aspects of human nature.
• The purpose of modeling is to help us get a handle on the numbers at any given stage of a project’s lifecycle, and help us focus on the “hard truths” that can spell the difference between success and failure.
• We should begin our modeling when we begin our project.We can use estimates initially to help fill in some unknowns, but our first iterations in the early stages of a project will necessarily be very simple affairs.

Chris Lord

In one of our recent classes, a student asked: “When does increased detail in modeling cash flows from a renewable energy project really matter?”

As the resulting class discussion brought to light, the answer depends on when and what kind of increased detail we are talking about. Is this an early stage or late stage project? Are we talking about taking time-based changes, such as refining a model from annual to quarterly or monthly re
sults? Or, are we talking about breaking down higher lever categories of production, revenue or cost into a series of more detailed sub-categories, such as separating out inverter replacement costs and insurance from a general “”O&M” category?

The first type of detail is known as “longitudinal” by data nerds because it slices across the whole term of the model. For longitudinal data, the short answer is that the breakdown from annual to quarterly or monthly can typically wait until later in the project development cycle when we are ready to talk seriously with prospective owners, investors and lenders about when and how they will get their money back.

But for the other type of detail, consisting of a finer breakdown of our production, revenue and costs within a single unit of time, also known as “cross-sectional” data, the additional detail is important. It helps us determine whether or not our project has a legitimate shot at making enough money to pay back its owners, investors and lenders. Cross-sectional data helps us identify items that have the biggest potential impact on our success or failure. Identification of these high-impact items also pushes us to focus on them earlier in the development cycle, when we have the time and budget to address them. 

Before delving into the complexity behind these answers, though, it is important to understand the context of the question within the project development process as a whole.

Just how important is modeling?

The importance of modeling is often downplayed in the early stages of project development. Sure, its importance gets a lot of lip service, but the truth is that even in later development, modeling is too often treated as a separate, technical part of getting a solar PV project built and financed. And, why not? Modeling is a pain in the butt. It requires attention to a lot of annoying detail, and looks suspiciously like math. Modeling also forces us to confront the troubling “unknowns” about our project, and that implicates two curious aspects of human nature. First, humans prefer reaffirming assurances (for a fun proof of this “conformational bias”, check out this quiz), even where a negative outcome can help us strengthen a troubled project, or abandon one before we over-invest. Secondly, humans naturally fear “unknowns.” An unknown or absence of information necessarily has a negative or adverse potential to it. These human traits come together in ways that make it easy to procrastinate on serious modeling, and not pay enough attention to the results in the rare cases that we do it.

And yet, everyone knows that modeling is important precisely because it forces us to face, earlier than later, the hard and uncertain challenges we would prefer to ignore.  In fact, it is axiomatic in successful project development that the hardest issues and greatest uncertainties must be tackled upfront. It would be a high-risk, low-payout strategy to invest in getting all of the easy things done at the expense of the harder ones, only to discover late in the game, and deep into our budget and schedule, that these thorny issues and uncertainties require a major re-orientation in the structure or nature of the project, or worse yet prove insurmountable within the remaining time or budget.

So, all that should give us a good clue as to when increased granularity makes a difference. The purpose of modeling is to help us get a handle on the numbers at any given stage of a project’s lifecycle, and help us focus on the “hard truths” that can spell the difference between success and failure. That means we should begin our modeling when we begin our project. We can use estimates initially to help fill in some unknowns, but our first iterations in the early stages of a project will necessarily be very simple affairs. Over time though, as we dig deeper and make progress against our development plan, we must steadily add layers and details to match – and even drive – our progress. For example, how many projects have priced the sale of electricity for a PPA without really analyzing the impact on cash flows and valuation? Far too many to count. (We will talk in another article about just how much you should really believe and trust results from your model – short answer: take them with a good dose of salt – but without conducting the analysis and thinking through the PPA price you may not even be in the right city, let alone the right ballpark.)

Cost and Revenue Considerations

For most projects, improving our understanding of the breakdown of our costs, and the projected model’s production and revenue numbers (together, cross-sectional data) is probably more important than breaking down annual numbers into quarterly or monthly data (longitudinal data) early in the development cycle. This is because the finer breakdown of cross-sectional data will help us uncover unknowns earlier and more easily than simply dividing up our annual numbers into quarters or months. For example, we might start with an EPC Construction Cost that is nothing more than a single number, but there will come a point when we need to know the detail behind that number. Imagine we have a sloped site. That should push us to ask sooner than later: does our EPC number include enough for site grading, or the rack-mounting system best suited to our specific site conditions? At all stages the model should help point us to places where we need more answers or creativity, to improve our chances of ultimate success.

In doing this we must also keep in mind the nature of our project. If it is a small C&I system going on an investor’s own roof or parking lot, our modeling needs might be kept relatively simple, both for longitudinal data as well as cross-sectional data. For utility scale projects, the larger dollar value and certain need for traditional project financing, probably dictates a more detailed model earlier than later. For example, in the context of a utility-scale project, longitudinal data done monthly or quarterly will be important to ensure that seasonal production matches debt payment obligations, and our tax equity’s cash flow needs. More cross-sectional detail will also be necessary early, such as accounting for state income, sales and use taxes.

Time and Season Considerations

Another place where the model might require more detail is where we have time of day and seasonal pricing. At first blush, we might think we need monthly or at least quarterly data across the full term of the Project model to see what our production and cash flow looks like when our sale price will vary within a single day and over seasons. But, we can model a single year’s 8,760-hour revenue, and then use the sum as our first operating year’s annual revenue. That annual number can then be adjusted for panel degradation and power price escalators to give us a simple – and for most purposes –adequate year-by-year view of a  20 or 25-year revenue model. This can simplify our modeling in the earlier stages of project development. When we get close to a financing, and have a feel for what the debt might look like, or need to understand a tax equity investor’s cash flow expectations, we can adapt the model easily enough to see the quarterly or monthly view. At that point our project is sufficiently real that the investment of time and effort to adapt the model is a good investment.

So, in sum, the answer to our question of when do we need increased granularity in project modeling depends entirely on the nature of our project, its status in the development cycle, and the unknowns we must overcome to complete development successfully. But there are some good rules of thumb. In the early stages of development, we should focus on improving our understanding of cross-sectional data – costs, production and revenue – to ensure that we have either a successful project in hand, or at least have a clue what major obstacles we must overcome to make our project successful. For larger projects we must dive into greater detail earlier because we are making a bigger bet. Additionally, the larger development costs for a utility scale project make it imperative that we identify and manage our critical issues as early in the process as possible. In all cases, as we progress in the development cycle, the detail in our project model should keep pace with, and perhaps even drive, our growing understanding of the project challenges and our economic expectations.

Additional Resources:

• HeatSpring Free Tool: Site Lease vs. Buy Modeling Calculator by Chris Lord
• HeatSpring Course: Financial Modeling for Solar PV Projects with Chris Lord
• HeatSpring Free Course: Finance 101 for Renewable Energy Professionals with Chris Williams
• HeatSpring Course: Solar Executive MBA Training with Chris Lord and Keith Cronin
• HeatSpring Course: Microgrid Executive MBA Training with Mahesh Bhave