It is our experience that simulations can focus thinking, improve productivity, and reveal new insights about a chemical system. We are integrating them more and more into our work.

Q. How do you use simulations to develop a mechanism?

A. Modelling is a time-honored way to develop a mechanism for a chemical reaction. This is usually done inductively, starting out with a very simple scheme (1-2 steps in many cases) and comparing the simulation results to experiments. The scheme is refined by adding more steps as needed to flesh out the details and continuing to test its predictions by experiments.

Q. If you need to know the mechanism in advance, what does CKS get you?

A. Running a simulation allows you to see if the mechanism is correct, and evaluate where changes need to be made. CKS is a fast and simple way to test your ideas.

Q. If you do not know the reaction rate parameters, where do you start?

A. CKS can be used at many levels. If you can guess the order of magnitude, that should be enough to try out a proposed mechanism. You can then compare the results to your experimental data and start refining your guess (we do this routinely). If you want to be physically realistic from the start, the literature is full of kinetics studies, critical reviews and compendia that may describe exactly the chemistry you wish to model, or a close analog. You can use the literature as a starting point since rate constant parameters tend to be characteristic of reactivity of a class of similar molecules, rather than varying dramatically from molecule to molecule.

Q. Is CKS a substitute for integrators?

A. Yes. The stochastic method is rigorously accurate and can be used wherever integration would be used. In fact for some systems it is more accurate than integration because no approximations are made during the solution process.

Q. Have you benchmarked CKS against integrators?

A. Yes, we have done this with several popular integrators. The output of the two types of simulations is identical, but the computation time tends to vary depending on the characteristics of the mechanism. The methods are completely comparable for non-stiff problems with no partial equilibria. The stochastic method is significantly faster for stiff problems, and the integrator is faster for systems with partial equilibria. This last case is due to a characteristic of the way that the stochastic algorithm works - in systems with partial equilibria the simulator will spend a lot of time maintaining the equilibria, and not enough to propagate the rest of the chemistry. We have developed an emulation routine that alleviates the inefficiency, but does not completely eliminate it.

Q. Can CKS handle stiff problems?

A. Yes - in fact this is one of the stochastic method's big advantages over integration of coupled differential equations. Stiff problems are no more difficult to simulate than non-stiff problems.

Q. Is the stochastic method effectively a Monte Carlo simulation?

A. Yes, it is a type of Monte Carlo calculation, but differs from that most commonly used in the physical sciences. It does not track the behavior of individual molecules or atoms, but of ensembles of them at local thermal equilibrium. This allows a very large dynamic range of time scales to be accessed, unlike more conventional Monte Carlo simulations.

Q. Is CKS written in FORTRAN?

A. No, it is written in C++ to make it portable across platforms.

Q. Can I add my own routines to CKS, or interface it to other packages?

A. No, this full-featured package is self-contained, and designed to be used without additional programming. The simulation data can be exported in a format that is compatible with spreadsheets and graphics packages.

Q. Is CKS more of a research tool, or applicable to industrial problems?

A. CKS is very versatile. In IBM it has been distributed to manufacturing, environmental, development and research groups, and it has also been used in university research and classroom settings through joint studies.

Q. Where does CKS fit with respect to chemical engineering simulation packages such as Aspen?

A. CKS is a dynamic simulator, and complements the capabilities of codes like Aspen. It can only model chemistry in a single reaction volume, so cannot be used to simulate reactions in a system of coupled reactors, for example.

Q. Can CKS automatically extract rate constants from experimental data?

A. No. CKS is not a curve fitting package, it calculates time behavior from mechanistic input. We do use CKS routinely to extract rate constants from complex kinetics data (that is, reactions too complex to analyze algebraically) by iteration.

Q. For what classes of systems does CKS not apply?

A. You can't use CKS for systems in which reactions are coupled to diffusion of matter or complex fluid flow in general, although it is possible to simulate reactions involving laminar flow and simple gas diffusion using it. CKS will not handle coupled reaction - heat flow - mass transport problems.

Q. How would CKS handle diffusion problems?

A. CKS can handle gas diffusion by dividing up the reaction zone into multiple connected zones. Gases can be transferred reversibly from zone to zone with a rate constant that depends on gas velocity and mean free path. Further details can be found in any gas kinetics monograph. Solution phase and solid phase diffusion cannot be simulated.

Q. Will CKS model a continuous stirred tank reactor (CSTR)?

A. Yes - we have included an example in the demo simulations.

Q. How would you handle an adiabatic combustion?

A. CKS is equipped to simulate adiabatic reactions through its option set. You only need to supply heats of formation and temperature-dependent heat capacity data for the species in the reaction system.

Q. How would CKS handle the need to remove heat from a system during an exothermic reaction?

A. CKS cannot handle this situation, which requires inclusion of energy transport.

Q. Can you handle multiphase systems?

A. Yes, as long as chemical reactions are confined to only one of the phases, and the others merely accumulate or are depleted. This restriction is because of the way that CKS defines reaction volumes. Examples of multiphase simulations we have included in the demos are polyimid.rxn, and accrete.rxn.

Q. How is CKS applicable to pharmaceutical problems?

A. CKS is useful for modelling synthesis: see, for example, our kinetic resolution simulation demos. It can also be used for thermogravimetric analysis simulations, also in our demos. It will not model complex pharmacokinetics problems.

Q. Can you change ambient pressure, i.e. model compression reactions?

A. This capability is not available in CKS. It does allow you to track changes in gas pressure during constant volume gas phase reactions, however.

Q. Can you model pH changes in a reaction?

A. Yes, we have included a demo simulation showing how to do this.

Q. Can you use CKS to model polymer reactions?

A. Yes, we have examples of polymer synthesis, including copolymerization and terpolymerization, polymer curing and polymer deprotection in our demo library.

Q. Can you get molecular weight distributions of polymers?

A. Yes, we show how to do this in the PMMA demo simulation set, which includes a sample spreadsheet to extract the molecular weight distributions from simulation data.

Q. How long does it take to run a simulation?

A. It depends on the mechanism - times can range from under a second to many hours.

Q. Can you overlay experimental data on simulation results?

A. Yes, a checkbox in the plot selection window lets you import an experimental data file.

Q. How much time does it take to learn and become competent with the software?

A. CKS is designed to be intuitive and very user-friendly. The experience of our user base is that they become productive immediately.

Q. How many steps can you put in a reaction mechanism?

A. If you are running the OS/2 or Macintosh versions then you can have up to 32,000 steps(!). If you are running under Windows you can have up to 160 steps.

Q. How long does it take to set up a simulation and do the calculations?

A. This is quite variable - setting up a simulation involves deciding on the input and typing it in, and the run time required depends on the complexity of the mechanism and the processes involved in it. For simple cases a few hours would be typical when you are putting together a new simulation, a few minutes for simulations that you are already working with.

Q. How do you choose the correct number of particles to use in the simulation?

A. The particles represent amounts of material, so you need enough to cover the dynamic range of concentrations present in the system. For example if you have an initial concentration of one species that is 1 mol/l, and another that is 1e-6 mol/l, you will need about 1e7 particles. CKS is limited to 4e8 particles. Usually the dynamic range of concentrations is a good guide, but subtleties can arise during a simulation that may require additional particles. To test for this, run one simulation with your best guess, then run another with 5x - 10x as many particles and compare the results. If they are identical, you have enough particles. If you want to reduce the amount of time it takes to run a simulation you can try decreasing the number of particles, but be sure that this change does not change the simulation results. The thing to remember is that you can never have too many particles - you just use more computer time than you need - but you can have too few.

Q. What is the typical simulation file size?

A. This depends on how complicated your mechanism is and how often you save the state of the system to disk during a simulation. File sizes can vary from a few kbytes to many Mbytes.

Q. If you stop the simulation prematurely, do you lose all the data?

A. No, CKS allows you to interrupt a simulation, inspect the results, and resume the calculation as many times as you like.