Dairy Ration Software and NRC 2001 Compliance an Overview

Formulate2 dairy ration software is fully compliant with the NRC 2001 prediction and evaluation model. All aspects of the NRC model are implemented with no shortcuts or compromises.

During development, the evaluative software supplied with the NRC 2001 Dairy publication was used as benchmark software to determine the accuracy of Formulate2's implementation of the 2001 model. Rations were evaluated with both software packages.

While applying this approach certain anomalies in the NRC software became apparent.

The image below shows the values found in the NRC program for ground corn. Note that the values for TDN1x is 88.71 and the value for Lignin is .90.

TDN1x is a calculated value while Lignin is an input variable to the TDN1x equation. When the value of an item used in the TDN1x equation is changed, the TDN1x value is automatically recalculated. However, if the existing value is re-entered there should be no change in the TDN1x value since the value of equation variable has not changed.

In this case, re-entering the existing values changes the TDN1x value as shown below.

This anomaly is further complicated by the fact that, apparently the recalculated value is not saved.

If the Ration toolbar button is clicked to display the ration screen, then the Feeds button is clicked to redisplay the feeds grid and ground corn is displayed again, the value for TDN1x is the original value of 88.71not 88.69.

The question then becomes, which value was used internally to calculate the analysis of the ration?

This anomaly was found to be general with respect to TDN1x.

While it may have little practical impact in evaluating diets, this inconsistency limited the usefulness of the NRC software as a benchmark.

In addition, the NRC program uses single precision floating point variables for calculations while Formulate2 primarily employs double precision floating point variables.

Consequently, when the differences in ration evaluation between the NRC software and Formulate2 became sufficiently small enough to fall within a range easily accounted for by either the NRC program anomaly or the difference in variable types employed, further reconciliation became impossible.

Of course, since we are talking about very small values, there is little practical impact here. However, re-entering an existing value for a TDN1x equation variable, when using Formulate2, doesn't change the calculated TDN1x value for ground corn which is 88.6882134894442