The inverse problem: given a chromatogram you measured, find the isotherm constants that reproduce it. A Nelder-Mead simplex varies the parameters and simulates the column at each step, minimising the sum of squared differences against your data. The fitted values feed straight into the batch simulator, which is also where the column and the step sequence are set.
Fitting to: 2 components · langmuir · 1 × 25 cm · 2 steps
Results
Load a measured chromatogram and press Fit. The optimiser reports every evaluation as it goes, so you can tell a converging fit from a stuck one without waiting for the end.
Getting a usable fit
- The column, the step sequence and the starting parameters all come from the simulator page. Get its chromatogram roughly over your data first — a fit refines a design, it does not find one from nothing.
- Fit one thing at a time. Isotherm constants and mass transfer times both broaden a peak, so solving for both at once lets the simplex trade one against the other and land somewhere that matches your data without meaning anything.
- Sample the peaks, not the baseline. The objective is an unweighted sum of squares, so a hundred near-zero points before the first peak contribute nothing while costing a simulation each.
- “Stopped at the cap” means the simplex ran out of evaluations rather than converging. Those parameters are the best it found, not a converged answer — raise the cap or start closer.
- Read N and K together, not separately. At dilute loading a Langmuir isotherm depends almost entirely on the product N·K — the Henry constant — so many N/K pairs fit the same data. Fitting synthetic data from a known column recovered N·K to 0.1% while N and K individually landed 40% off. To pin them apart you need data at concentrations high enough for the isotherm to curve.
Parameter determination
Add at least one measured chromatogram to fit against.