I'm sharing a question from a teacher who has recently jumped into SWI with her students in Toronto. This teacher and her students noticed the spelling  <travelog> did not have the spelling they would expect if it was based on <travel> and <log>. It was the single <l> in <travelog> and the <ll> in a word like <travellers> that grabbed their attention. They wrote: 

Hi Peter!
I've got a question that came up today in SWI.
We're making word matrix for travel. Couldn't figure out why travelog (1 L) vs. travellers (2 L).

Here is my response:

Well done to you and your students!

I’ve never noticed the spelling <travelog> before. The obvious first hypothesis is that this word would be built on <b> plus <log>. But the word sum reveals a problem with hour hypothesis:
travel + log —> *travellog
While there are two attested spellings for this word (<travelog> and <travelogue>) I do not know of an attested spelling with <ll>. 
So the first lesson here is to remember to let the word sum falsify a hypothesis no matter how obvious that hypothesis seemed at first. 
I went to Etymonline to see if they had a story to help me understand. They cite the British spelling, but points us in a helpful direction for understanding the single <l>.
travelogue (n.)
"a talk on travel," 1903, a hybrid word coined by U.S. traveler Burton Holmes (1870-1958) from travel + Greek-derived -logue, abstracted from monologue.
Interesting that we actually know who coined this word. And we see that the coiner of this neologism in 1903 was thinking about <travel> and <logue>. Notice also that it is described as a “hybrid word”.
The term “hybrid” describes a word formed from elements of different origins. The <travel> goes back through <travail> to a Latin origin. By contrast the <logue> goes back to Greek. 
But we also see this description that ths word is “abstracted” from “monologue”. From time to time you’ll see descriptions like “abstracted from” or “on the model of”. The way I understand that is that people notice a parallel word formation that informs a new word. 
Notice that <monologue> is not a hybrid, its two bases are both of Greek origin. When a US traveler (traveller) wanted to use the idea of that word to form a word that was about writing about travel, they just used this model.
But the actual analysis of <monologue> does not have any issue of how many <l>s. 
mono + logue —> monologue
or the more full analysis 
mone/ + o + logue —> monolouge
But with <travelogue> the sources of this word <travel>  and  <logue> raises an interesting issue. The final letter of the first and initial letter of the second are both <l>.
But the coiner of the term was “abstracting” from <monologue> with only one <l>.
So how can we understand the structure of <travelogue>?
There is another term for word formation when morphemes are blended one into the other for some particular reason. 
You probably know of words like this. A classic example is the word “brunch”.
This word blends the <br> of <breakfast> and the <unch> of <lunch> to form the word <brunch> meaning a meal later than breakfast, but earlier than lunch.
The term for words that purposely violate morphological structure to form a word is very logical. They are called “blends”. 
This has NOTHING to do with the term often used in schools for the <pl> in “play” or the <st> in “stop”. 
So my thinking is that when this person wanted to make a new word, he wanted the single <l> of <monologue> so he just created a blend <travelouge> and had the <l> of <travel> overlap — or blend with the <l> of <logue>.
Now that you’ve encountered the concept of the <blend>, you and your students could keep their eyes out for such words. A clue will be when you seem to recognize meaningful parts of two words, but a word sum cannot be used to show that construction without violating the spelling of a morpheme. 
Your second question is about <travellers> with <ll>.
Notice this entry in my Oxford: 
traveler (British traveller) 

So there is yet another question to explore in your question.
There is a consistent difference in how US and British based spellings treat the letter <l> for consonant doubling. That difference is reflected in the old Real Spelling “Big Suffix Checker” 
For some reason, British spelling has evolved to treat the <l> differently in consonant doubling. 
For most consonant doubling, the following criteria need to be in place for doubling to occur:
  • You are adding a vowel suffix
  • The base or stem to which the suffix is added ends in a single consonant
  • There is a single vowel letter immediately before that final, single, consonant of the base or stem
  • The primary stress of the completed word is where the suffix is added. 
For some reason, British spelling adds one other criteria. If the final single consonant of the base or stem is an <l>, it is doubled regardless of where the primary stress falls. 
This is such a rich vein of orthographic learning you and your students have uncovered by noticing spellings that appeared to violate the word structure conventions you’ve been learning. 
Please keep emailing with your questions/discoveries. Perhaps we can find a time to Zoom to discuss such things with your students.