Be Skeptical of Models

Engineering Tip: Be Skeptical of Models​
Models. We deal with them all the time. Frac models, reservoir models, fluid models, models of models. Lots and lots of models. But models have two major limitations:

1. Inputs - Models are only as good as the data you put into them. If you have inaccurate or misinformed data, then your model is going to be inaccurate and misinformed. Every reservoir model requires porosity as an input. But porosity can change a lot across the reservoir. Often times we are using porosity from a core sample and extrapolating across the reservoir. It's better than nothing, but we need to be realistic about the potential inaccuracy this causes.

2. Understanding of the physics - The universe is a very complex machine. We like to take natural mechanisms and boil them down to one or two variables so that it is understandable. However, most things in nature are an interconnected web of many, many variables. Let's take fracture modeling. People love to reduce fracture growth down to a simple model of pressure and injection rates. In reality, there are a gazillion things that impact fracture growth. We love simple models that show nice, parallel, predictable frac wings, when in reality, we often get a spider web of fractures that grow in many different directions.

Models will always give you an answer. Whether or not that answer is accurate is a different question entirely.

And don't get me started on climate change models...

Please don't read this and think that I am anti-modeling. Models are a useful tool, but they are a tool and not an exact science. When you use a model, be a little skeptical of the results. Ask questions and compare it to other information and intuition that you have. If the model gives you something that seems unreasonable, dig deeper and ask more questions.

At the end of the day, human intuition and historical data can be just as powerful a tool as modeling. Let's not get too carried away with models.

Don't Underestimate the Importance of Good Quality Data

Engineering Tip: Don't underestimate the importance of good quality data​
Engineering analysis and modeling are only as good as the data collected. If you collect good quality data, then your analysis and models will be more accurate. If you collect crappy data, all of the fanciest models and calculations won't matter. Collecting good quality data is the first, and perhaps the most important step when arriving at a meaningful and accurate answer. Don't underestimate the importance of good data.

I suspect that most of us are aware this. So why then is it that we often settle for crappy data? It's because the crappy data is cheaper. I'll give you an example that I see often in well testing. Well test analysis relies heavily on bottomhole pressure. The best way to get bottomhole pressure is to run gauges downhole and measure it directly. However, this costs a lot more than just estimating it from surface pressure. So what people often do is go the cheap route and estimate bottomhole pressure from surface, and settle for a less accurate answer. This is a fine strategy and there is nothing wrong with it, so long as you are honest with yourself that the results will be less accurate. We all work for companies who are in the business of making money, and we should all be mindful of costs. However, there is a cost to making decisions on low-accuracy assumptions. Just be mindful of this.

So how do we decide when good data is necessary and when it's safe to cut corners? Well, if the answer you are trying to find is really important and has large implications, then I believe that good data is worth the price. However, if your potential answer is less important and an "estimate" is all you really need, then I'd say that lower-quality data is a fine choice. Keep costs low where you can, so you have a little more to spend where it's really necessary.

Measured Data Is Always Better Than Calculated Data

Engineering Tip: Measured data is always better than calculated data.​
In this industry we are often faced with a decision of whether or not to measure a meaningful data point. It is often cheaper in the moment to calculate a data point rather than measure it. However, measuring something directly always leads to more accurate data. It doesn't matter if its pressure, flowrate, temperature, or any other physical data point - measuring it directly is more accurate than calculating it.

I'll give you an example: At FyreRok, we run a lot of downhole gauges to measure bottomhole pressure. Now, we can calculate the bottomhole pressure based off of surface pressure, but this introduces the potential for error. We always try to run downhole gauges as often as possible to ensure the best data.
​