GreenQ Partners analyzed natural gas consumption at Weston Bakery Sudbury. This article presents partial simplified results of this analysis. Gas was metered at oven level at uneven monthly billing intervals; production was reported every 28 days; weather data was provided by NASA on a daily basis. We visited bakery to link numbers to actual production procedures and processes.

Weston-foods logo 500by324
This project had several goals:

  • Increase accuracy of costing and gas bill budgeting,
  • Quantify energy saving opportunities,
  • Identify operational inefficiencies.

Through statistical analysis of data and in close contact with client’s operational and energy management we were able to produce valuable results towards each of these goals.

We started by looking of three years of gas consumption data at once and found gas consumption to demonstrate a near perfect correlation with production levels: R2 is 97% through this period. This indicated no major troubles.

Natural Gas consumption forecast model

Natural Gas consumption forecast model

Nevertheless, deeper analysis has revealed a variety of deviations and saving opportunities. For further analysis 2010 was selected as a base period. Linear model based on this period forecasts actual consumption data with R2= 98%, though consumption varied substantially between winter and summer:

  Winter Summer
Production load per kg, per day 0.038 (±0.004) 0.028 (±0.002)
Non-production load, per day 76 (±30) 168 (±26)

 

Similar consumption patterns were observed in later years. Mathematically linking consumption with production was already a breakthrough for costing and budgeting, which previously used constantly changing monthly averages. Further we used CUSUM tool to identify deviations from the 2010 baseline. Results are presented on the next graph.Seasonal variation of natural gas consumption at Weston Bakery Sudbury

This result was somewhat expected: if we take winter level as a baseline, then lower per kg summer consumption presents ‘savings’ manifested by negative gain of CUSUM. Surprising was the absence of horizontal periods in winters. More accurate analysis has uncovered creeping growth of winter consumption year after year.

Winter natural gas consumption creep

Winter natural gas consumption creep

These results presented several business benefits to the bakery: These numbers represent difference between actual and forecast gas consumption. Though they are rounded and approximate, they represent a consistent consumption increase trend. Assuming that the data is correct this increase can result from either a change in SKU mix or from equipment efficiency loss. Operation manager indicated that SKU mix has not changed.

  • Statistically reliable consumption forecast can now be used for costing and budgeting instead of “average per kg gas consumption”, which was unpredictably different every month
  • We quantified the difference between winter and summer consumption resulting from difference in intake air temperature. This finding provided an economic basis for consideration of intake air preheating
  • The difference in summer and winter non-production gas consumption has invalidated the operational decision to keep oven always on in summer to “save gas on reheating”
  • Growing per kg winter gas consumption  indicated a possible operational or technological deficiency