Flowsheet simulators play a vital role in process engineering. They allow for rapid and precise calculation for designing process plants, troubleshooting operating problems and checking if proposed solutions will work. Simulations are tools and, like all tools, have limits. These limits can lead to simulator results not matching plant performance. If the engineer is fortunate, the model doesn’t work at all or gives an answer so wrong that it’s immediately obvious. Other times, the answer is significantly wrong but not obvious. This can create severe problems. Designs don’t work. Troubleshooting doesn’t find the problem. Proposed solutions don’t fix difficulties and maybe even worsen them.
I have covered issues that can undermine simulation in a couple of previous columns: “Get Real about Your Simulations,” and “Don’t Model in a Vacuum.” In addition, several CP articles, such as “Prevent Simulation Meltdowns,” and “Succeed at Simulation,” provide valuable pointers. However, the importance of modeling certainly justifies another column about potential problems.
These problems fall into three main areas:
• limitations in the simulator and the correlations it depends upon;
• weaknesses in the data; and
• lack of simulator or system knowledge.
First, flowsheet simulators rely on thermodynamic, property and equipment design correlations. These enable modeling a unit operation and estimating its performance. Common types of equipment handled include pumps, separator drums, distillation towers, heat exchangers, fired heaters, crystallizers, etc. Some simulator packages offer very complex equipment options such as multiple-stream exchangers and wiped-film evaporators.
Assumptions in the simulator often depart from how the equipment actually performs, especially when worn, damaged or pushed to extreme operating conditions. For instance, simulators typically assume drums reach equilibrium conditions but they may not — hot-vapor bypass drums are a classic example. Drums also entrain liquid into the vapor and may carry vapor droplets into the liquid. Many other real-world issues also undermine assumptions: Fired heaters have heat losses. Distillation tower trays leak. Foam causes carryover and operating problems in multiple types of equipment.
Many engineers rely on simulator results without understanding the correlations behind them and the caveats in using these correlations. This isn’t much of a concern for finding the pressure drop of water in commercial steel pipe, where the correlation typically is very good. However, other correlations demand more caution. At the extreme are those for two-phase flow, tray efficiency and shell-side pressure drop in condensing exchangers. These correlations have significant limits, often making their results markedly wrong for a specific situation. Using a simulator doesn’t free an engineer from the responsibility for knowing calculation applicability and accuracy.
Results for stream properties and behavior depend upon the thermodynamic methods and assumptions used. Picking the right method can dramatically improve your simulation. Picking the wrong method is a sure way to bad results.
Second, simulation requires good data. You usually can measure temperatures and pressures with reasonable accuracy. However, even these measurements can have problems. Pay attention to the accuracy of flow measurements. Careful meter selection and proper installation and maintenance can get accuracy to 2% or better for most process streams. Unfortunately, many industrial flow meters often exhibit errors of 10% or more.
Compositions pose a particularly challenging area for data gathering. Is the sample taken correctly? Has it been kept conditioned and secure until the laboratory runs the tests? Did the lab conduct the tests correctly? How good are the lab measurements even if performed correctly? Do you know the repeatability and reproducibility of the results? Do you know the difference between repeatability and reproducibility? What do they mean for your simulation?
Most laboratory work adheres to ASTM (which originally stood for American Society for Testing and Materials), ISO (International Standards Organization) or DIN (Deutsches Institut für Normung) specifications and standards or those of other recognized organizations. Staff working with these bodies devote extensive effort to developing such standards and determining their statistical validity.
One common thread in unsolved problems is that the plant personnel often don’t know how good (or bad) are the results they’re using. Without correct composition or performance data, diagnosing plant problems changes from analysis with basic principles to just trying random things to see what happens.
Third, troubleshooting requires knowledge of both the process and the tools. The engineer must set up the flowsheet simulator to match the plant. Mismatches can arise because the engineer doesn’t properly understand the plant, the tool or both. Simplifying the model by ignoring things that appear unimportant often prompts problems.
One widespread error in plant setup is ignoring static head in pressure profiles. In one case, a side-reboiler added to a column didn’t work at all. Its model didn’t account for the static head of 70 ft of liquid in estimating the boiling point of the process. The heating medium wasn’t hot enough to vaporize the liquid in the exchanger at startup because the static head increased the boiling temperature.
Other difficulties arise because of defaults in setup. One commonly used exchanger routine automatically assumes zero sealing strips in detailed calculation of TEMA (Tubular Exchanger Manufacturers Association) S-type and T-type exchangers. This results in dramatically low pressure-drop estimates on the shell side and moderate decrease in estimated heat transfer. However, the software works fine if you specify the correct number of sealing strips.
Another simulator uses an obsolete method of ASTM D86-D2892 distillation conversion that gives impossible results under some situations. This method has persisted for years even though the problem is well known. A third simulator relies on an approach of using vapor and liquid leaks in its calculations to model trays having less than 100% efficiency. This can yield inconsistent results when evaluating tray efficiency effects.
Flowsheet simulators give process engineers extremely powerful tools to simplify design, optimization and troubleshooting. Nevertheless, like all tools, they have limits. The engineer must know and respect these limits to get useful results.
