--------------------------------------------------- The origin of multiplicities in binary distillation --------------------------------------------------- by Sigurd Skogestad, NTNU (07Apr00) This note is to clarify some controversy in the distillation community about who first discovered multiple steady states in binary distillation. The first journal publication on multiple steady states in binary distillation is "Multiple steady states in ideal two-product distillation", E.W.Jacobsen and S. Skogestad, AIChE Journal, April 1991, 499-511 (Manuscript received Apr. 23, 1990 and revision received Jan. 10, 1991). The basis for the controversy is the following paper: "An output multiplicity in distillation: Experimental verification" by A. Koggersbøl, T.R. Andersen, J. Bagterp and S.B. Jørgensen Presented at ESCAPE conference at Rhodes in 1996 (Comp.Chem.Engng., Vol. 20, S835-S840) Here it is stated that "Nielsen (1990) observed that for certain operating conditions the signs of those model gains related to reflux flowrate where dependent on whether the flowrate was measured in units of moles or volume", and then they reproduce a Figure from her thesis showing this. They also refer to multiple steady states and state that "this has never been observed when using flowrates on molar basis". There is then a statement about the instability of the (Lw,V)-configuration where they (by the way incorrectly) claim that the sign change would imply instability under feedback control. Only after only these two paragraphs, which the reader is given the impression originates from the work of Nielsen (1990), is there is a reference to AIChE-paper of Jacobsen and Skogestad (1991) but this is only related to the stability of the multiple solutions. There are two things here which we (Skogestad and Jacobsen) find improper: 1. The reference "Nielsen (1990)" is to a thesis, and this is not regarded as a publication in the international community. 2. More seriously, the statement gives an incorrect impression about the history of these multiplicities. The reader gets the impression that they were first discovered by Nielsen (1990), and that Jacobsen and Skogestad (1991) only made some contribution with respect to the understanding of the instability. However, the fact is that Jacobsen and Skogestad (1991) were the first to publish also the results on the negative gains and the multiplicities, and the paper was even submitted to conferences and AIChE Journal before the thesis of Nielsen. So, in fact, the work by Nielsen (1990) had no effect on the history of events in this field. To be more precice this is the history: Elling Jacobsen discovered the multiplicity in simulation for water-ethanol sometime in the summer of 1988 (report from Aug. 1988). A letter was written to A. Lucia and W. Marquardt on 03 April 1989 [ see http://www.chembio.ntnu.no/users/skoge/publications/1990/ ] because the results seemed to go against some of their proofs. Elling and I then spend some time trying to understand what had happened and found a very simple explanation (dL/dLw negative) some time during the autumn of 1989. We submitted abstracts/papers for this in late 1989/early 1990 and published two conference papers on this in 1990: E.W. Jacobsen and S. Skogestad, ``Dynamics and Control of Unstable Distillation Columns'', Presented at 40th Canadian Chemical Engineering Conference, Halifax, July 15-21, 1990. E.W. Jacobsen and S. Skogestad, ``Multiple Steady-States in Ideal Two­Product Distillation'', Paper 133a, AIChE Annual Meeting, Chicago, Nov. 1990. The final journal paper to AIChE J. (published in 1991) was submitted on 10 April 1990. Now, about Charlotte Nielsens results. She submitted her thesis at the end of March 1990 - after all our results, both the discovery and explanation, were finished, and thus the thesis has had no effect on the "history of multiplicities in binary distillation". In the thesis (p 107), in a section on "Note on molar versus volumetric flows", she has a figure of a simulation which shows multiplicity. Apart from the observation itself (where she says nothing about the possible source, e.g. differences in mole weight), she only makes the following statement: "The calculations indicate that operation with molar flows should be preferred for identification and control purposes to avoid problems with the sign of the gain which can arise around certain operation points". But this statement contains two misunderstandings: 1) It is not possible to operate with molar flows, since no "molar valve" exists to my knowledge. 2) The sign change does not really affect feedback control since there is always a unique flow for each composition. I know Nielsens work well as I was the external examiner for the thesis during its defence on 10 May 1990. Indeed, I gave a seminar in the department at the same day, (S. Skogestad, ``Multiple steady-states in distillation columns'', Guest lecture at DTH, Denmark, 10. May 1990) where I presented Ellings results, and everyone agreed at that time that we with the submission to AIChE KJournal were the first to publish these results. (The first indication to the contrary came at the above-mentioned conference in 1996.) Conclusion: The work of Nielsen (1990) was never published, and in any case the thesis was written after our work was submitted for publication. According to academic standards it is then not proper to refer to this work, and at least not in a way where it is indicated that it was done before the work of Jacobsen and Skogestad (1991).