I have a cupcake recipe that I've doubled successfully in the past. I'd like to triple it for an upcoming party but I seem to recall reading that some ingredients don't work as well if tripled. Is that true? If so, what are they? Is there any time when tripling a cake or cupcake recipe would work?
The short answer is no; there are no specific ingredients that don't scale (double or triple or quintuple, makes no real difference).
There are, however, two pitfalls you need to be aware of when scaling any recipe:
As long as you don't change anything else - such as making individual cupcakes 3 times as large - then you should not run into any problems.
Other than issues with crowding issues (during cooking, in the mixing bowl, etc.), I can only think of one problem that I've specifically seen with baked goods, and it was directions to:
The person who made the cake followed the instructions exactly (rather than modify it to add double or triple the number of eggs per addition), which meant that the batter was being beaten for twice as long, resulting in too much gluten development and tunneling.
Yes, a cupcake recipe is probably one of the few which can't just be tripled.
All cupcake recipes I've come across are chemically leavened (they include baking powder or baking soda). With baking powder and baking soda, the leavening starts as soon as you mix the dry and the liquid parts of the batter - even before you put the batter into the oven. It produces gas bubbles which slowly rise to the surface and puff away.
That's why you need to bake the batter immediately. Once the dough starts setting in the oven, the bubbles start to get trapped within the forming sponge of the baked cake and give it volume and lightness. Even the time to pour the batter into 36 cups could be somewhat too long, provided that you can fit all of them in your oven. But I guess that you don't have such a big oven, and are planning to bake them in batches. While the first batch is baking, the leavening in the remaining batter will proceed, all the bubbles will form and pop, and then the batter will be flat, like an opened soft drink left outside for too long. When you put it into the oven, there will be nothing left to make new bubbles - the baking soda will have been used up - and it won't rise, leaving the middle dense and uncooked.
Some cakes are leavened without baking powder or baking soda, just with eggs. Maybe this is applicable to cupcakes too (although I haven't seen such recipes). But as egg foam falls with time, especially if mixed with other ingredients, you have the same problem - your batter shouldn't be spending any time outside the oven.
So the normal thing to do would be to make the recipe three times, not to make all the batter at once. If you have a kitchen scale, you can mix all the dry ingredients in one bowl and all the wet ones in another, then for each batch, mix exactly one third of each (or one half, if you have enough cupcake pans and oven space to handle the amount at once) and wait until the oven is free before mixing the next batch. Measuring by volume is risky, and estimating it by its looks is more likely to flop than to work. I've heard advice to mix everything but the baking powder and add the powder at the last moment, but getting a very small amount of baking powder evenly mixed into lots of wet batter, while it is reacting, doesn't work well. So mixing dry and wet separately is what I'd do. If leavened with beaten egg whites, I'd beat each batch of whites separately at the last moment and mix under the rest for each batch.
The other thing to consider is that your oven will likely be hotter after a batch of cupcakes than after 10 minutes of preheating. So throw a look early at all batches after the first one.
*Edit: An addition about the effect of tripling on salt and baking soda. *
I have read the answer which Aaronut linked in the comment to the question, but don't agree that a bigger size gives you more room for error - explanation see below.
As for soda and salt. You are right that these are very sensitive to wrong measurement. First, they are chemically very active, will not be able to do their important work if underdosed, and will leave a noticeable undesirable taste if overdosed. Second, because they are used in small quantities, a small absolute error in measurement will make a big relative one - using 5 gr instead of 4 is a 25% increase. Third, as kitchens seldom have exact measuring instruments with the needed sensitivity, recipes are usually rounded, and a recipe designed for the triple size will probably prescribe an amount of salt and/or soda which works much better than tripling your recipe. Fourth, because of the rounding, the amount given is probably not the optimal one, but somewhere between the optimal one and the upper or lower limit of working well, and you don't even know on which side. Meaning that your available margin of error is much smaller than of bulkier items which are (hopefully) given in a more exact amount.
But the previous paragraph all applies to bad measurement, not to multiplying the recipe while still using exact measurement. And I don't see how this can affect the tripled size, for one simple reason - when you are baking, you are not interested in absolute amounts. What you need to get right are the ratios. So even if tripling the recipe can lead to a bigger absolute error, the probability for a relative error stays the same, or even falls.
First, you should be measuring as exactly as you can. If you measure exactly at both the single and triple dosage, the recipe should work, regardless of the fact that the amount prescribed is not optimal due to rounding. Let's say that 2% salt by weight would have been optimal, but your recipe cites "3 gr", which results in 1.8% salt. Obviously, this ratio is still acceptable, or else the single size won't work. If you measure exactly for the triple size, you'll still end up with 1.8%, which will work just as well as the single size. (You could have done better if you had a recipe for a triple size which says 10 grams instead of 9, but obviously you don't have it).
But what if your measurement techique isn't exact? I hope that you can measure well enough to make roughly a gaussian distribution, sometimes measuring just a bit less and sometimes just a bit more than what you wanted. As the error should be independent of the amount measured, tripling the recipe should lead to the same absolute measurement error, which is a smaller relative measurement error in the bigger size - so you are better off. This applies if you are measuring with a scale, because its error is mostly independent of the weight measured, except at the extreme ends of its range.
If you are measuring by volume and throwing in three tablespoons instead of one, you are measuring three times instead of one (just please don't start calculating that three tsp are a tbsp or something similar and use the calculated measure instead; measure it three times). Provided that your average measurement is the one you want, in the best case your errors from the three measurements will cancel each other partially, providing a better measurement. In the worst case, you'll make a big error in the same direction three times after each other (but the probability for this is low), and then you'll be just in the same position as measuring once for the single dose and making the same big error. (So if you consistently scoop 1.1 tsp instead of 1 tsp, the ratio of soda to batter will be the same in the tripled case as in the single case).
The only case I can think of when tripling is a risk is a case where the recipe is inconveniently close to a limit (as in prescribing 3 grams of soda when 3.2 would be too much) and your average measured result is somewhat translated towards this limit (as in, your scale is not perfectly calibrated and usually overmeasures by 0.1 gram). Then you'll have a high probability of your measurement exceeding the limit. If this is the case, the probability of passing the limit after three measuring events is higher than after a single measuring event. So the probability that the tripled recipe will fail is higher. But in this case, the single recipe will also fail often, so you probably won't be using it at all.
The meaning of all this is that while using a recipe specifically designed for a big batch is the best solution, using a multiplied small-batch recipe isn't more probable to fail than using the same recipe for a small batch. If somebody finds something wrong with my reasoning, please point it out, it is a bit late in the evening to be doing math.