Indentification of syllables is fairly simply in Khuzdul. Syllables must all start with a consonant, followed by a vowel. Syllables can optionally end (close) with a single consonant. At the end of a word, the final syllable can close with two consonants. Because of this, wherever there is a single consonant in the middle of a word, the consonant starts a new syllable. In places where there are two consonants in the middle of a word, the second consonant starts a new syllable. The only challenge comes in identifying unique consonants when at least one is is a digraph, which is a single sound represented by two letter. In Khuzdul, this would be < ph th ch kh zh sh hy gh >. As you can see, all of those have < h > in them, so the trick is to know when you are dealing with one of these versus a simple < h > or one of the sounds that doesn't have < h >, with those being < p t c k z s y g >. Generally, the only way to know is to have an idea of what the consonantal root of the word is. A good way to see this is to look at the declension chart for the given word, which is usually necessary in order to use Quasi-Khuzdul. However, the times where there may be some confusion should not be too often, so it's not a large concern.
The rules for syllabification in Khuzdul can be summed up as follows:
The valid syllable structures for Khuzdul may be defined in the following manner:
Allowable Coda Consonant Clusters
* - Clusters that include <y> or <w> end up forming a diphthong + the final consonant, rather than a "standard" consonant cluster.