Energy auditing and recommendation for energy saving is still rudimentary. Most of the auditors are using Fluke like meter or downloading a CSV file from Smart meter connected to the server. Then they either have proprietary software or Excel Macros to look at some common aspects - such as "Turn-ON", "Turn-OFF" habits, Power Factor, compare warmer vs colder days, etc. Basically, a few tricks, rules, patterns they have learned to save energy. But they are very ill-equipped when you look at the overall potential to save energy:
1. Fluke and other common energy meters that can store the data for 6-12 months and the auditor can download that data - are still expensive ( $5000-10,000). Ideally, the auditor should have around 5-10 energy meters within $5-15k range, which they can retrofit anywhere. They should be able to deploy it within 5 minutes in any location and be able to get all the patterns (signature) of energy they want, at the time interval they want. Ideal data resolution should be less than a minute, although these days they work with 10 min to 1 hour resolution.
2. Second major issue - most of them don't get the "Time Stamped" data of temperature map of various rooms, presence of people, etc. that’s required for understanding utility usage.
3. There have been significant work on deployment of Machine Learning in Energy saving studies. But those works are not very useful for practical deployment. There are several reasons. Real life optimization requires weather data from weather API, people occupancy data, temperature gradient plot of the rooms and so on. And then there must be a real time dynamic implementation at a price which is commensurate with saving from energy optimization. Let's say the yearly energy bill of a high school in US is close to $100-200k. 5% saving would be around $5-10k max. While if we have to implement at least 3 smart energy meters and 50 connected temperature sensors then total Capex will be around $50-60k. Then cloud subscription cost will be close to $1000-2000 per year minimum. Probably it will take 10 years or more to recover the investment.
However, with the advent of low cost sensors and edge computation which eliminates high cloud bill, it is increasingly feasible to design an IoT system for energy optimization taking care of all the essential variables. In ideal business world, IoT system installation cost should be recovered within 2-3 years.
For more details, please read : Energy Saving & Optimization