Numerical solution of multi-variable cell population balance models. III. Finite element methods

Nikolaos V. Mantzaris, Prodromos Daoutidis, Friedrich Srienc

Research output: Contribution to journalArticlepeer-review

79 Scopus citations


We have developed finite element approximations to the generalized multi-variable cell population balance problem under conditions of changing substrate concentration. We considered different sets of basis functions consisting of 1st-, 2nd-, 3rd-, 4th-, 5th- and 6th-order polynomials. For the time integration of the nonlinear system of ordinary differential equations that result from the finite element discretization in space, we implemented both time-implicit and time-explicit algorithms. The set of basis functions consisting of 4th-order polynomials was found to be the most appropriate for all test problems considered. It offered more than a 10-fold reduction in required CPU time, when compared to the linear basis functions. Time-explicit methods were found to be preferable to time-implicit methods in terms of computational efficiency. Despite the fact that the required CPU time of the best finite element algorithm for single-variable simulations is comparable to that of the best spectral and finite difference method, the comparison is highly unfavorable for the finite element method in the case of multi-variable simulations.

Original languageEnglish (US)
Pages (from-to)1463-1481
Number of pages19
JournalComputers and Chemical Engineering
Issue number11-12
StatePublished - Nov 15 2001

Bibliographical note

Funding Information:
We thank the Graduate school of the University of Minnesota for awarding N.V.M. a dissertation fellowship. Financial support by the National Science Foundation through the grant NSF/CTS-9624725, NSF/BES-9708146 and NSF/EES-9319380 is also gratefully acknowledged.


  • Cell growth
  • Cell population balance
  • Finite element methods
  • Numerical solution
  • Substrate consumption


Dive into the research topics of 'Numerical solution of multi-variable cell population balance models. III. Finite element methods'. Together they form a unique fingerprint.

Cite this