Why Choose PD-PLUS?
No simulator is right for everyone or every purpose. Choosing one is a matter of your needs, the way you like to work, and what you can afford. There are perhaps three classes of steady-state chemical process simulators. In one group, there are several huge and complex pieces of software, integrated collections of various tools, some of which were formerly separate programs. They are loaded with features, of course, but they also carry a very high price tag. They also are rather slow in execution; if you have used nothing else, you may not think so, until you try PD-PLUS and note the startling difference. Speed by itself can be an important factor and is discussed in more detail below.
There is a middle group of programs that also have a great many features, probably all most process engineers need for routine things. Their prices are considerably lower, but still higher than what many consultants and occasional users can justify.
Then, in a class by itself, is PD-PLUS. While it does not have as great a collection of features, it does have everything most process engineers normally need for steady-state process simulation. Bear in mind that some licensees use it extensively for modeling whole processes, creating highly detailed heat and material balances for proposals and process design. It is serious software.
So, what indeed are the reasons why PD-PLUS just may be the right choice for you? There are several reasons:
Cost - Simplicity - Speed - Robustness
In building PD-PLUS, the intent always has been to provide a tool that will handle most needs of most process engineers, yet (unlike the others) still be affordable to consultants and occasional users. It is a matter of simple economics that to provide every conceivable feature to please all users means an enormous development and support effort that must be paid for in high cost to all users. A single copy of a high-end simulator may cost you $20,000 or more - per year - depending on what you can negotiate with the vendor. Obviously, a great number of companies, especially the large ones, have no problem spending this amount of money. A simulator from the middle grouping may cost $7,000 to $20,000 initially and perhaps another 15% per year for support and upgrades. This is much more affordable to some, yet still beyond the reach of many.
In contrast, PD-PLUS can be licensed for a one-time fee of just $1,995 and then just $100/year for support and updates. For some users who need a good tool for process calculations but can't justify the high cost of the others, PD-PLUS is really the only choice left.
One of the expensive simulators could be the correct choice for you. Still, you need to ask yourself what it is you need a simulator to do for you. Unless you really need some of the specialized features offered by a more expensive program, choosing one of them is like subsidizing the needs of other users - paying for capabilities you can't use. Many companies have licensed several different simulators, including PD-PLUS, to address varied needs and customer requirements. This lets them minimize the number of copies of the expensive software and match the tool to the task at hand.
Many users of PD-PLUS have commented favorably about the simplicity of using it. In comparing PD-PLUS with other simulators, one difference you will note right away is that PD-PLUS uses a keyword-based input language. Although the various aspects of using the program are controlled through a Windows menu window, it doesn't (yet, anyway) have a menu-based system for preparing input for a model, as do other simulators. That style is appropriate for many engineers, and, quite honestly, some will not even consider a program lacking that type of interface. However, engineers experienced in the subject often argue otherwise; they see menu-based interfaces for building and maintaining models as getting in their way.
The keyword interface of PD-PLUS lends itself well to quick assembly of a model, whether starting with an empty file or using parts copied from other related models. A text file containing the paragraphs defining the various parts of a model is nearly incorruptible. With other simulators, the engineer must save his work in a backup file often to avoid total loss of it if the program crashes or the file becomes corrupt through complexity and accumulated changes. The other advantage to using a good keyword interface is that it permits rapid progress through the repetitive cycle of making changes to the process model and evaluating those changes.
Chemical process simulation itself is not an easy subject to master, and the notion that a mouse-driven interface provides an intuitive point-and-click path to success is a fallacy. With some simulators, attending a multi-day training session is really required in order to be able to use the program at all. That says something about the impact on ease of use imposed by the shear bulk and complexity of such a program. In contrast, most users of PD-PLUS start with no formal training in its use and develop fluency while working with it on real projects.
The earlier keyword languages of the major simulators were rather cryptic when compared with that used by PD-PLUS. They really did need a different interface, and the paradigm of other Windows programs certainly has been applied fairly well. Still, it's not the best interface for everyone. You should decide what works best for you.
In presenting information on PD-PLUS, no exaggerated claims are made. However, when it comes to execution speed, modesty simply has to be set aside. The simple fact is that, compared to the other major simulators, PD-PLUS is by far the speed champion. There are two points of view on the striking difference in speed. A process simulator with a good collection of calculation algorithms can be expected to take a certain amount of time to calculate the heat and material balance for any given process. PD-PLUS can be viewed as an amazingly speedy exception to a standard defined by the others. Or, PD-PLUS can be regarded as the standard defining how long the task ought to take, while the sometimes incredible extra amount of time taken by others is a shameful measure of their inefficiency.
Many frequent users of PD-PLUS take its speed for granted and become frustrated at the relatively sluggish performance of other programs on a similar task. For example, just the input translation of small to medium sized PD-PLUS models on a modern PC is seemingly instantaneous; a large model, with a hundred or more flowsheet blocks, still takes less than a second. With other simulators, input translation for just a small problem can take upwards of 15 seconds. For a single problem, 15 seconds before flowsheet calculations can even begin may sound insignificant, but few models are run only once. That extra overhead gets to be rather annoying as the model is changed and run repeatedly as the model evolves and a great many process alternatives are evaluated.
Input processing speed is one thing, but flowsheet execution speed is where the bigger difference is seen. Run time, of course, depends on the nature of the model and the speed of the PC. Distillation columns take more effort than simple flash calculations, and recycle streams and feedback controllers add a substantial convergence load to the task. PD-PLUS contains a highly optimized set of calculation procedures, particularly for distillation columns, that allow convergence of even large flowsheets in very little time.
So, just how fast is PD-PLUS? You be the judge. For the following typical problems, some solution times given are actual measurements from some years ago on a rather slow 166 MHz Pentium-based PC, except as noted. Divide by ten or so for a more current PC.
Atmospheric crude column, 20 stages, two sidestrippers, 33 components, SRK thermo: 0.5 seconds.
Aromatics separation column, 100 stages, 7 components, two composition specs, SRK thermo: 1.4 seconds
Gas plant, 12 flowsheet blocks (including 3 small columns), two recycle streams, 25 components, SRK thermo, no recycle stream estimates; nine recycle stream iterations: 0.9 seconds.
LNG plant, 27 blocks (including two LNG heat exchangers with full heat curves and two feedback control loops), 9 components, SRK thermo: 1.6 seconds.
Ethylbenzene process model, 90 flowsheet blocks (including four columns and multiple feedback controllers), three recycle streams, 54 components, mixed SRK and ideal thermo; without good initialization: 1 minute 43 seconds; with recycle streams and controllers well initialized: 2.3 seconds (233 MHz Pentium).
Integrated multi-train styrene process model, 184 flowsheet blocks (including 14 columns and multiple feedback controllers), 10 recycle streams, 36 components, mixed thermo methods; with reasonably good estimates, but requiring nine outer recycle loop iterations: 50.1 seconds (233 MHz Pentium).
In the above examples, the columns typically had product composition specifications, and all were done rigorously, with equilibrium and heat balance on every stage (PD-PLUS does not have a shortcut column model). Of course not all models execute as fast as these examples. Some large models have been known to run for a minute or more, and even longer, but they are the exceptions.
So, what does all this speed mean for you? If you do a lot of process modeling and you don't normally need some of the specialized features found only in the expensive simulators, the speed advantage of PD-PLUS will let you make better use of your time and shorten the duration of your projects. Doing process calculations only 20 or 30 percent faster would not be very compelling, but calculating models literally in seconds rather than minutes can change the way you work.
It also should be noted that the simplicity of PD-PLUS and its ability to do rigorous process calculations so fast provides an opportunity to build on-line process optimization applications quickly and inexpensively. A generic optimizer can pass a new set of process parameters to the flowsheet execution engine of PD-PLUS, have it update the converged solution, and get back selected results within a few seconds, or even in less than one second. Other simulators are too slow to respond this way and are too bulky and complex to be integrated in such applications without considerable effort and expense.
Having a long list of advertised features in a simulator doesn't help much if you spend much of your time just getting it to converge your model. Particularly annoying are frequent program crashes and failure to detect and report input errors. That is why so much attention has been given to implementing efficient convergence algorithms in PD-PLUS and checking input for errors (see "Development Philosophy"). Even a model with no detectable input errors can have specifications that lead to an infeasible solution, so all execution code checks carefully for situations that would cause a less robust simulator to crash. The result of devoting so much attention to robustness in PD-PLUS is a simulator that is typically much more "bullet-proof" than others. Bugs have been a rarity in PD-PLUS for years.