DWSIM Development Blog

Build 4513 Update Patch

nospam@example.com (Daniel Medeiros) — Fri, 11 May 2012 08:04:42 -0300

An update patch from Build 4503 to Build 4513 was released yesterday. It contains some general fixes and enhancements for the flash algorithms that increases their reliability and stability.

Click here to download the Windows patch

Click here to download the Mono Edition patch

Please note that you must apply this patch over Build 4503 only. It may not work with earlier builds. The Windows patch also works with DWSIM Portable Edition.

The source code is also up-to-date and can be acessed through the Git interface.

DWSIM and FPROPS, part 2

nospam@example.com (Daniel Medeiros) — Sun, 29 Apr 2012 20:32:58 -0300

FPROPS integration is done. It will be available in both DWSIM versions (.NET and Mono) and also exposed as a Property Package through the CAPE-OPEN and Excel interfaces. Some screens:

DWSIM and FPROPS PP

DWSIM Mono Edition and FPROPS PP

DWSIM / FPROPS PP working with COCO Simulator

DWSIM and FPROPS

nospam@example.com (Daniel Medeiros) — Sat, 28 Apr 2012 19:29:04 -0300

One of my current development activities is the integration of FPROPS with DWSIM.

FPROPS is a free open-source C-based library for high-accuracy evaluation of thermodynamic properties for a number of pure substances. It makes use of published data for the Helmholtz fundamental equation for those substances. It has been developed by John Pye and others for the ASCEND Project, and will happily function as standalone code.

Currently FPROPS supports calculation of the properties of the following substances: ammonia, nitrogen, hydrogen, water, carbondioxide, methane, carbonmonoxide, ethanol, acetone, carbonylsulfide, decane, hydrogensulfide, isohexane, isopentane, krypton, neopentane, nitrousoxide, nonane, sulfurdioxide, toluene, xenon, butane, butene, cisbutene, isobutene, transbutene, dimethylether, ethane, parahydrogen, isobutane, r41, r116, r141b, r142b, r218, r245fa.

The properties that can be calculated are internal energy , entropy , pressure , enthalpy and Helmholtz energy , as well as various partial derivatives of these with respect to temperature and density.

FPROPS reproduces a limited subset of the functionality of commercial programs such as REFPROP, PROPATH, EES, FLUIDCAL, freesteam, SteamTab, and others, but is free open-source software, licensed under the GPL. Contributions are most welcome!

I managed to compile FPROPS with small modifications to its source code in order to expose the calculator functions for external P/Invoking in DWSIM. I'm still working with John in order to better understand FPROPS' working behavior but the current results show a lot of promise:

DWSIM Mono Edition and IPOPT

nospam@example.com (Daniel Medeiros) — Sat, 28 Apr 2012 19:19:19 -0300

After trying really hard for almost three days, I finally managed to get IPOPT working with DWSIM Mono Edition on OS X and Ubuntu.

I created 32-bit libraries for OS X 10.7.3 and Ubuntu 11.10 that will ship together with the next build of DWSIM Mono Edition. This will allow me to bring the new Gibbs Minimization flash algorithm to DME while unleashing the power of IPOPT to the already present Data Regression and Multivariate Optimization utilties.

Now I can go back to Battlefield 3.

Phase Equilibria calculation by Gibbs minimization

nospam@example.com (Daniel Medeiros) — Sun, 15 Apr 2012 20:51:50 -0300

In DWSIM you can find two different approaches for phase equilibria calculation. The first of them is the Nested Loops approach, where we have a basic PT Flash based on the Rachford-Rice equation in a sucessive substitution procedure. PH and PS Flashes are done in a "nested loop" fashion, where in the external loop we change the temperature and in the internal one, we do a PT Flash, comparing the calculated enthalpy/entropy at P, T with the specified one.

The second is based on the Inside-Out approach from Boston and Brill for two phases (VLE) and in a modification by Boston and Fournier for handling of two liquid phases (VLLE). It involves a change of the iteration variables by converging an inner loop with simple, composition-independent equations while their parameters are updated in the outer loop by using the rigorous models.

Both of them have advantages and disadvantages. The Nested Loops approach is more reliable and robust but takes longer to converge, specially when the mixture is close to the critical point. The IO approach is very fast but can be unstable depending on the mixture conditions. It also seems to be insensitive to poor initial estimates (and this is a good thing).

There is also a third approach, not very common in process simulators: Gibbs energy minimization by second order methods.

Sometimes it is advantageous to treat the flash calculation as a minimization problem, using G/RT as objective function and the vapor amounts as the independent variables. Since the gibbs energy is minimum when the mixture is at equilibrium, we only need a good math optimizer to find it.

This weekend I tried to get this approach working in DWSIM by using the IPOPT solver. Calculating the Gibbs energy is easy because it is a function of the fugacity coefficients, and so are its derivatives with respect to the molar amounts. I finally made it a few hours ago, after fighting a lot against some weird numerical errors - the key here is the problem setup, i.e. variable bounds/scaling. It doesn't seem to have any advantage over the other approaches for all pratical simulation purposes, since it can be very slow depending on the number of compounds, but in theory it should always give the correct result, even for difficult VLLE cases.

The screenshot below shows the difference in calculation times for a PT Flash of an equimolar mixture containing from C1 to C9 and H2O at 270 K and 100 kPa. They all give the same results (two liquid and one vapor phases), but the Gibbs minimization approach took 3.7s to converge, while the IO one took only 0.1s to do the same. The other two times (0.4 and 0.5s) are from a hybrid IO/NL method that I`ve also implemented, where the three-phase PT flash is calculated with the IO approach and the 2p-PT/PS/PH ones are done by the traditional "nested loops" one. I call this one Hybrid NL/IO, also capable of handling two liquid phases.

New features in Build 4452 for Windows

nospam@example.com (Daniel Medeiros) — Sat, 10 Mar 2012 16:18:04 -0300

First of them is the new Petroleum Assay Manager, which allows the user to store, load, import and export petroleum assay data. The new manager is fully integrated with the characterization tools, and from its own interface the user can edit assay data and export it to a file for use in another simulation. The data is independent of the characterization methodology, that is, only the assay data is saved, not the user choice of correlations nor number of pseudos that will be generated. This is good when you wan t to test different correlations for a single assay and check which ones give the best results on your simulation.

The Petroleum Assay Manager can be acessed from the SImulation Settings > Components > Petroleum Characterization panel or from the Tools menu:

It has full integration with the characterization tools (Bulk and Dist. curves):

Also new to this build is a method for ordering items in a Master Table. The user can choose from one of the current selected properties to be shown and if it will be ordered in a ascending or descending way.

Finally, I added two new features to the PFD: Zoom All and Pan, the latter being activated by clicking on the PFD with the left mouse button while holding Shift... and then scrolling it.

Download Build 4452 for Windows directly by clicking here.

Introducing the Master Property Table

nospam@example.com (Daniel Medeiros) — Tue, 28 Feb 2012 23:38:33 -0300

I think that this will be the final "product". The nice thing about the MPT is that you are not limited to Material Streams - you can insert as many tables as you want to show grouped properties for all unit operations and streams in a flowsheet.

I forgot to mention in the previous post that I've also added a feature to save the current flowsheet view to a PNG image:

New tiny tiny features

nospam@example.com (Daniel Medeiros) — Mon, 27 Feb 2012 21:42:24 -0300

I'm working on some suggestions given by the user Max on the forums. First of them was making the current system of units editable:

This will work only for user-defined systems. Next was to make available some extra properties on a stream table, like bubble and dew points:

Also, it will be possible to change the items displayed in the table by changing the display status directly on the property grid:

The next one was actually my idea. It will be possible to edit the composition of a stream by pressing Ctrl+E with the stream selected on the flowsheet. or by clicking on the corresponding context menu item:

In the next days I'll work on a new item for the Flowsheet: the Master Property Table. It will be something similar to stream tables that we can find on printed flowsheets, with properties from different streams in a single table. The only difference is that you'll be able to group properties from other unit ops instead of only from material streams.

I hope you like these new features. If you have more good ideas, please let me know.

Build 4418 for Linux and Mac is out

nospam@example.com (Daniel Medeiros) — Sun, 05 Feb 2012 23:36:38 -0200

The Binary Data Regression utility is now available for Linux and Mac OS X! Grab your copy of the latest DWSIM Mono Edition here.

New build 4410 for Windows + PRSV2 EOS + Data Regression tutorial

nospam@example.com (Daniel Medeiros) — Sun, 29 Jan 2012 12:15:38 -0200

Build 4410 is available for Windows and includes a new Property Package with the PRSV2 equation of state. So, since Build 4258 we have the following list of changes:

- [NEW] Added the Compound Creator Utility and corresponding user database structure (work in progress)

- [NEW] Added a Binary Data Regression utility (work in progress)

- [NEW] New Property Package: Peng-Robinson-Stryjek-Vera 2 (PRSV2)

- [NEW] Added feature request #3409646 (Name given to Material Objects)

- [CHG] Enhanced Unit Set Creator to start with units from the current system

- [CHG] Changed density input from Specific Gravity to API Gravity on the Dist. Curves Characterization Utility

- [CHG] Rewritten solver logic increases calculation speed by 35% (Cavett sample)

- [FIX] Fixed infinite volumetric flow on first Material Stream calculation through the property grid

- [FIX] Fixed Adjust and Spec Ops not showing anything on the property grid

- [FIX] Fixed Lee-Kesler Cp/Cv calculation

- [FIX] Fixed Distillation Curves Petroleum Characterization Utility

- [FIX] Fixed Pipe Segment model overall HTC calculation and liquid phase volume retrieval

- [FIX] Fixed Spanish translation of the Object Palette

- [FIX] Fixed bug #3409641 (Call to Terminate missing)

- [FIX] Fixed bug #3409637 (Collection Count method called twice in a row)

- [FIX] Fixed bug #3409628 (ICapeUtilities queried twice in a row)

There is also a new tutorial on how to use the Data Regression utility => sourceforge.net/apps/mediawiki/dwsim/index.php

Build 4407 Beta

nospam@example.com (Daniel Medeiros) — Thu, 26 Jan 2012 00:25:06 -0200

Build 4407 Beta is up and you can get it here => http://sourceforge.net/projects/dwsim/files/DWSIM/DWSIM%202.0/DWSIM_bin_v21_b4407_beta.exe/download

Changes since 4404:

- [CHG] Added LLE support and initial estimates calculation using UNIFAC structure to the Data Regression Utility

- [CHG] Changed density input from Specific Gravity to API Gravity on the Dist. Curves Characterization Utility

- [FIX] Fixed solver not recalculating outlet streams

- [FIX] Fixed infinite volumetric flow on first Material Stream calculation through the property grid

- [FIX] Fixed Adjust and Spec Ops not showing anything on the property grid

I've also included some data regression samples, just check them on the "samples" folder. Feedback here please => http://sourceforge.net/projects/dwsim/forums

Data Regression, work in progress - Initial estimates

nospam@example.com (Daniel Medeiros) — Tue, 24 Jan 2012 21:03:45 -0200

This time I've added the option to calculate initial estimates for NRTL/UNIQUAC regression using UNIFAC structure information. Also added were the so important "title" and "description" textboxes, just in case you want to specify were your data came from, what and why are you doing the regression, and so on.

Data Regression, work in progress - LLE

nospam@example.com (Daniel Medeiros) — Mon, 23 Jan 2012 21:53:28 -0200

LLE data regression is just plain HARD. I had to add the possibility of estimating initial values for the interaction parameters so the flash algorithm can return two liquid phases right from the start and finish the optimization successfully. You better have good initial estimates, otherwise the regression will fail miserably.

Build 4404 BETA available for testing and feedback

nospam@example.com (Daniel Medeiros) — Sun, 22 Jan 2012 23:21:58 -0200

If you want to try the new compound creator and data regression utilities, you can download the newest beta build here - there are some fixes too. Please note that this is still work in progress and, as such, some new features are not fully working while the rest needs testing and feedback.

Changes since Build 4358:

- [NEW] Added the Compound Creator Utility and corresponding user database structure (work in progress)

- [NEW] Added a Binary Data Regression utility (work in progress)

- [NEW] Added feature request #3409646 (Name given to Material Objects)

- [CHG] Rewritten solver logic increases calculation speed by 35% (Cavett sample)

- [FIX] Fixed Lee-Kesler Cp/Cv calculation

- [FIX] Fixed Distillation Curves Petroleum Characterization Utility

- [FIX] Fixed Pipe Segment model overall HTC calculation and liquid phase volume retrieval

- [FIX] Fixed Spanish translation of the Object Palette

- [FIX] Fixed bug #3409641 (Call to Terminate missing)

- [FIX] Fixed bug #3409637 (Collection Count method called twice in a row)

- [FIX] Fixed bug #3409628 (ICapeUtilities queried twice in a row)

Data Regression, work in progress

nospam@example.com (Daniel Medeiros) — Sun, 22 Jan 2012 14:09:58 -0200

Here is a screenshot of the new data regression utility. In its first iteration, only binary equilibrium data will be supported, and I'm still trying to figure out how to calculate NRTL and/or UNIQUAC parameters from LLE data. It won't be easy because it is not guaranteed that the flash calculation will always return two liquid phases (and no vapor)...

Ths is work in progress, I'm now polishing the interface and checking if the code is working for all possible combinations of exp. data and model selection.