It is good to try some analysis - at least to get a good review of the theories involved. Unfortunately, analysis is not always very accurate with heat exchangers (HX). If you were trying to size, for instance a brazed flat plate cross flow liquid heat exchanger (a pretty simple system to analyze relatively speaking) for a particular job, talking to suppliers, you would learn that they do not have curves for many applications, because the curves are nearly all empirical. Someone hired them to prove the function of the HX before they bought one (or many), that is the only reason why the curve exists. They do the testing because analysis is inaccurate. HX manufacturers won't comprehensively test their product line, because to do it well, it would take an incredible amount of resource. I was looking for just this sort of HX a few years ago, and I was surprised to learn they had analysis software they were using and writing, but it wasn't easy for them and they really didn't like to base much on it - the analysis results did not correlate well to the real world. This is coming from people who really know what they are doing. Let this be a warning to you, not to commit much capital to a design based solely on analysis - do some prototyping. This is the common sense part. Your system is going to be more uncertain than a brazed flat plate HX. How you duct the air, how the fan design works (a science and art unto itself), how the mounted environment changes the behavior of the system - compared to lab or prototypical use, how well you really understand the boundary conditions of the system. Do you really have a good model for the system around the HX at least for the more extreme conditions? Common sense will be revealed in how you look at the overall picture. You want to try and bracket the situation. What are the extremes of inputs and environment that will affect the HX. You can save the details in between the extremes for later when trying to optimize. But a first you don't want to waste effort on a system design that is just inappropriate. You look at similar applications already in existence as a frame of reference. If your system seems radically different from existing systems, you need to start asking questions about why. With HX you want to pay particular attention to approach temperatures. This is where the two flows have a small delta T. The ability of the HX is greatly diminished near the approach temperatures. You need to make sure you cover this situation. Your liquid flow needs to be turbulent or the HX will drop considerably. You need to look at the Moody Chart. Common sense comes into the compensation factor you apply - how much you jack up the transfer area (and cost) to make sure you are covered for unforeseen occurrences. THere is no perfect answer, you have to think in terms of risk and benefit. Finally, pay very close attention to the mounting of the HX. You might get the HX part perfect and it falls apart in the field. This is where being observant, surveying existing apps, and talking to manufacturers is very helpful. Follow best practices. Do not discount simple prototype failures as trivial. Be observant, take good notes, and don't forget what you saw. Don't be among the many who omit this step.