Measuring Performance

The explicit identification of measures (definition (2)) that are changed (or intentionally maintained) by functions is advocated as a standard practice here.  They are often referred to as “Measures of Performance” (MoP).  MoP can generally be categorized as one of two general types: “Effectiveness” and “Efficiency”.

Measures of Effectiveness

An effectiveness measure is one that is observed to determine whether a function has been adequately effected.  If the suggested Functional Analysis practices have been observed (at least in principle), effectiveness measures will have been identified during that exercise.

Aside

“Effect” literally means “something that is produced by an agent or cause: something that follows immediately from an antecedent: a resultant condition”, a “purpose, intention [or] end”, “the result of purpose or intention”1.  These concepts all speak to how “function” is defined here.

An example of effectiveness is the cumulative ability of a rocket to produce thrust during operation.  It is commonly measured as “impulse”, which is the integral of thrust over time2.

Measures of Efficiency

Efficiency addresses the amount of a run-time resource used to perform a measured function3.  Most efficiency measures are formed as a ratio, although the presence of a ratio does not necessarily mean the measure is one of efficiency.  The resource is usually addressed in the denominator4.   Common resources include (but are not limited to) fuel, raw material, catalysts, thermal sources, thermal sinks, munitions, logistical carrying capacity, spare parts, and various forms of time.  Efficiency measures tend to be more commonly used when run-time resources are tightly constrained.

Although efficiency measures are usually ratios, many instances “reduce” or “simplify” as the units are resolved5.  An example with reduced units is specific impulse, which measures impulse per weight of fuel; the units are “seconds”6.  Resolving the dimensions obscures the real nature of the measure.  Ensuring the end-to-end consistency and correctness of usage is a good reason for System Engineers to carefully observe dimensional analysis throughout the project.

Efficiency measures are often related to other parameters, not all of which directly address the performance of functions.  For example, we might (or might not) measure a petrol-fueled vehicle’s performance with “range” and “mileage”.  Together, they’ll size the fuel tank; the size of the tank is a Physical Characteristic, not a function.  Alternatively, we might size the tank directly and discard the range (but retain the mileage as a characteristic).  A third alternative might keep the range but constrain the total contained energy in the tank (which effectively constrains the fuel tank size and drives the fuel efficiency).  Whether specifying an item or merely describing one that exists, it is important to consider such relationships in order to avoid the potential for internal contradiction7.  See also the commentary on orthogonality.

Related Examples

Footnotes
  1.   Merriam-Websters Dictionary for Android, v. 3.4.217.26599, 2014[]
  2.   Sutton, George P. and Donald M. Ross, “Rocket Propulsion Elements, 4th Ed.”, John Wiley & Sons, New York, 1976[]
  3.   Recall that not all functions are measured![]
  4.   but counter-examples are not hard to find[]
  5.   That is, the ratio is implicit rather than explicit.[]
  6.   Specific impulse is more directly associated with a rocket engine than with the entire rocket.  Variants using the mass of the fuel are also used, but the units are different.  In reality, they are two different variables that measure the same thing using different calculation algorithms, units, and (therefore) values.  It is unfortunate that they use the same term. []
  7.   Some modeling environments make, for example, a clear distinction between “attributes” and “derived variables” for that very reason.[]