Fundamentals

EbsIdent serves for identifying the quantities appearing in your cycle. In general, one observes any target quantity, which depends on certain input values. A target quantity can be any quantity, which is available in the cycle after the calculation, such as

• a property of a component, such as efficiency of a turbine or a k*A of a heat-exchanger
• a thermodynamic quantity calculated by EBSILON^®Professional (pressure, temperature, steam content, ...) at any point in the cycle
• a characteristic value, which results from other quantities, such as cycle efficiency or specific heat rate
• an arithmetic expression, such as the sum or the mean value of certain quantities

 

In most cases, EbsIdent is used for component identification, i.e. the target quantity is a component property in this case.

Input quantities may be any quantities available in the cycle, such as

• measured or calculated values (pressures, temperatures, steam contents, ...) at specific points in the model

• values that are given from outside (variables)
• arithmetic expressions like differences, ratios and others

 

The following simple example should illustrate this:

Assume that the temperature on a pipeline called "A" (in EbsScript notation: "A.T") is determined by the pressure on line B and on the mass flow on line C:

A.T := f (B.P, C.M);

Note that this approach is the most important step in your identification. It takes a certain amount of time to figure out a suitable set of variables that determine the behaviour of your target quantity. EbsIdent can calculate with nearly any approach, but the results will be more or less worthless, if a suitable approach is not selected.

EbsIdent determines the relationship f. This is done as follows:

For f, you must define an approach that is a sum of expressions that depend on the variables and (optionally) on additional parameters, e.g.

            A.T := c1 + c2 * B.P + c3 * C.M + c4 * B.P * B.P + c5 * C.M * C.M + c6 * B.P * C.M;

EbsIdent is able to calculate the set of coefficients c1 ,..., c6 that give the best fit for f.

For the calculation, EBSILON^®Professional requires many value sets (B.P, C.M, A.T), at least 6. The more value sets you supply, the more realistic is your fit.

If more than 6 value sets are specified, it is generally not possible to find a set a coefficients that fulfils the equation for A.T. exactly. What EbsIdent determines is the set of coefficients c1, ..., c6, where the mean squared error over all value sets is as small as possible. The polynomial, from which this set of coefficients results, is known as "reference polynomial".

What can you do with these coefficients now? If your approach is good enough and the mean squared error is small enough, there are two possible applications:

• The calculation of the target quantity
• The monitoring of the target quantity

 

If the target quantity is a component specification value, it is possible to use the result of the identification to improve the calculation of that component within the EBSILON^®Professional calculation kernel. This can be achieved by an adaptation polynomial, which is entered for the component. In other cases, you can use the reference polynomial for calculation within EbsScript.

For monitoring the target quantity, EbsScript offers a function to calculate quality factors. Alternatively, you can perform these calculations explicitly by applying the polynomial directly. In any case, you should be aware of the limits of this adaptation:

• A deviation is significant only if it is greater than the mean squared error of the adaptation
• Results are not reliable if the values of your input quantities are beyond the range of identification (especially if you use polynomials of higher grades)