set of variables around a process, line, or machine. These EnPIs are useful for tracking
performance in “one isolated dimension.” However, in complex process industries, there
are usually several interrelated influences at play at any given time. For example, if EnPIs
include multiple heat flows into and out of a process area, or differing energy densities of
different products, or influences on the quality or condition of raw materials, the best way
to look at how each of these influencing factors affects each EnPI is to use MVR to study
the correlation between these factors. This provides a “multidimensional” view of each
EnPI, and this can be used to predict or forecast the likely or expected performance of an
area, unit, or machine.
How do single and multivariable regression analyses compare with each other?
Don’t they both provide a mathematical model that can be used to predict the expected
result of an EnPI? The short answer is yes. The real question is how well does a single
variable explain the operation of a complex process? Single-variable regression is a very
simple and straightforward method of comparing a dependent variable with how it
corresponds to one independent variable. For example, if one is looking at the energy
consumption of a chiller used for human comfort or space conditioning, it would be
meaningful to look at “cooling degree days” as that one single independent variable.
Manufacturing processes, however, are often much more complex and finding one single
independent variable that explains the variation in the dependent variable is nearly