Recently-released FEFLOW 6.2 includes FePEST, a convenient graphical user interface for using PEST with FEFLOW models. One important capability of this new FePEST is the so-called [b]IFM-Implemented functionality[/b]. Although FePEST includes most relevant FEFLOW parameter types and observation types for being associated with PEST, the IFM-implemented functionality provides an additional range of possibilities. Clearly, the operation of this functionality requires certain knowledge of IFM Programming in FEFLOW.
The intention of this post is to provide a general overview of the workflow. In this simple example, a plug-in computes first the depth to groundwater in the form of [b]user nodal distribution[/b] in FEFLOW and second it writes a text file (ObsIFM.DAT) with average depth for the entire model domain. Text file contains a single row. This average depth to groundwater will be subsequently used as an observation definition in FePEST in [b]Prediction mode[/b] for the evaluation of worst-case scenario (maximum aquifer drawdown). General steps for IFM-implemented observations are described below. Notice that these steps may vary depending on information retrieved with IFM plug-in.
1) IFM plug-in should be loaded and attached to the FEFLOW fem file for later used in FePEST.
2) Create a new FePEST problem (see help system for further information) and define parameters and observations (normal workflow).
3) Define a new observation definition using [b]#IFM-Implemented#[/b] as entries for both [b]definition[/b] and [b]source[/b].
4) Save the FePEST problem and create three PEST main files (and pilot point files, if needed) with [b]CHECK[/b] routine button and/or [b]RUN[/b] button in FePEST. Notice that a name of IFM-implemented observation is required. This name will appear in the PEST Instruction file (*.INS) and follows same regulations than PEST classical run.
5) Until here, FePEST creates all needed information. This can be reviewed any time from the work directory (menu [b]Estimation – Open work folder in explorer[/b] in FePEST).
By default, FePEST cites [b]IFM-implemented observation[/b] in the INS file, together with all other observations defined (e.g. hydraulic head, temperature, concentration, groundwater age, etc. specified in observation points). Therefore, IFM plug-in should be also able to write other observations in its output file when FEFLOW runs. Otherwise, it is highly recommended to create a second INS file containing only IFM-implemented observation (this option is explained below).
6) Open PST file from FePEST working directory and modify [b]NINSFILE[/b] variable. This is placed by default in [color=red][b]Line 5 and Column 3[/b][/color]. [b]NINSFILE[/b] refers to the number of INS files used in the PEST problem.
7) In the PST file, go to section [b]model input/output[/b], here the link between PEST Instruction file (*.INS) and results written by IFM Plug-in should be indicated. Write a new third row, e.g. [b]“ifm.ins” “ObsIFM.dat”[/b], where file named ifm.ins should contain as many rows as IFM-implemented observations. Make sure that output file from IFM plug-in should be written in FePEST work folder; otherwise full file path should be indicated in this section.
8) In FePEST, click on [b]Open work folder in console window[/b] from Estimation menu.
9) Run check routine for PEST control file: [b]pestchek fepest_name.pst[/b].
10) If errors are not found in item 9, run FePEST (or PEST from console window with [b]run_pest.bat[/b]).
The IFM-Implemented functionality in FePEST is a convenient manner to get access and use FEFLOW User Data in PEST optimization problems.
Carlos Rivera
