Methods for identifying and locating errors

The following search methods allow you to identify errors or problems in a file in several ways.

In both cases, you start with a model or file containing an error, but the cause of the error is unknown and has not been pinpointed. The aim is to pinpoint the problem using a search method that involves trial and error after eliminating groups of elements.

Sequential search

This method involves a sequential search, removing elements or parts of the model (for example, the floors) one by one until the error is found.

Example

A model of a five-storey building (ground floor, 1st, 2nd, 3rd and 4th floors). The error is on the 2nd floor, but its exact location is unknown. The following search procedure is used:

• A backup of the file is made.
• The 4th floor is removed. The error keeps appearing.
• The 3rd floor is removed. The error keeps appearing.
• The 2nd floor is removed. The error no longer appears. The error is on the 2nd floor.

Binary search

This method performs a binary search by eliminating halves of the model until the error is found.

Example

A model of a water supply system with a common service connection and two risers supplying the two parts of the system (I and II). Each riser supplies two dwellings (dwellings IA, IB, IIA and IIB). The error is in dwelling IIB, but its location is unknown. The following search procedure is used:

• A backup of the file is made.
• I has been removed. The error keeps appearing.
• The original file is recovered.
• II is removed. The error no longer appears. The error is in II.
• The original file is recovered.
• IIA has been removed. The error keeps appearing.
• The original file is recovered.
• IIB is removed. The error no longer appears. The error is in IIB.

Limitations

• The proposed methods require that the models resulting from the removal of part of the data from the file be consistent and capable of being calculated when necessary. Furthermore, they must be representative of the original situation.
• If the error is not located at a specific point in the model, these methods will not yield valid results.
• Similarly, if the error occurs in multiple locations, the procedure for identifying them may be more complex.