This is a difficult use-case to address as Algolia is a keyword-based search. We perform a normalization process at query-time that removes all non-alphanumeric characters.
The single and double-quote characters often used to infer inches, feet, etc. will be removed. You can add them back into consideration via the "separatorsToIndex" configuration, which may help. However it still will likely not produce the best relevancy when using these characters due to tokenization.
With regards to text abbreviations (e.g. "in" or "mm"), it should find matches if the phrase is a valid word AND is contained in a searchable attribute on your records. However, again there may be difficulties as it may interpret the query incorrectly without a space (e.g. "6in" may be interpreted as a typo, whereas "6 in" will likely not.) In addition, measurement abbreviations that are not actual words (e.g. "mm") will likely be interpreted as typos. This can be prevented by adding them to the disableTypoToleranceOnWords configuration setting.
In short, measurements and abbreviations can be difficult for keyword-based search solutions, including Algolia. The most effective and consistent way to address this is to force/funnel the appropriate terms via the front-end UX. In other words, if your records are appropriately indexed with well-defined and well-formatted measurement data, you can then have a specific section on the front-end that gears your end-users towards typing that information in a specific area (like a "measurements" text field.) This allows the search to be performed more accurately, as you can refine the results based on the information typed in those user-defined fields.