We often see extremely simplistic approaches used to forecast utility usage in complex manufacturing or refining processes – estimating the amount of electricity you’re going to use over the next month is a tricky endeavour and forecasting usage for the next 12 months can seem almost impossible.
Many external factors can affect the forecast; market conditions, product planning, maintenance, environment/ambient conditions and so on.
A seemingly logical strategy is to look over the last 3-4 years worth of data, add up all of the electricity used and divide by the number of months. Then you have the average monthly usage, so sending that to the utilities company seems quite reasonable. We’ll use around that amount every month – how wrong can it be?
The main issue with this is that utility companies don’t bill annually; they look at your usage every month. Very few processes have a constant usage throughout the year and with maintenance downtime and changing external temperatures the usage can fluctuate drastically on a monthly basis. Most utilities incur penalties if you are over OR under your predicted usage, so these discrepancies can add up to a hefty sum at the end of the year.
Utilities such as electricity and gas are as important as feedstock to a process – in some processes, they are the feedstock – but they are often viewed completely separately. Where a process engineer may be able to give quite accurate forecasts for product rate per month, they often struggle to do the same for electricity, gas, or other utilities.
At Sabisu we have developed a method to derive the utility usage from these accurate product foreacasts by analysing historical data.
By looking at the historical utility usage Sabisu will find the non-linear relationship between utility usage and production rate so that the engineer’s production estimates can be used to generate accurate utility forecasts.
As can be seen in the image above, Sabisu’s electricity usage forecast for an ethylene cracker is extremely close to the real values. Importantly, the forecasts follow the rises and dips in production which are not accounted for by constant average forecasting.
Using these profiles as the submitted forecasts for the utilities providers would lead to significant savings by greatly reducing the amount of penalties incurred when compared to the average usage forecasts that were previously used.
We’re always interested in hearing from you with any comments or suggestions, feel free to get in touch.
- Join us for our next Sabisu User Group on 14th June 2018.
- Sign up to our newsletter: Sign Up Form
- You can always contact us at email@example.com
- For a How to Guide on any Features go to http://wiki.sabisu.co/
- Our contact web form is here
- Our LinkedIn group is always open for business
- Join our Beta Test Group: Find out how