Recently, the Lambert W function has arisen in the analysis of many systems including a restricted class of time-delay systems. An alternative approach to this analysis, based on the well-established root locus method, is shown here to contain the Lambert W function as a special case.As a purely illustrative example of the equivalence between the Lambert W function and analytic root locus a system comprising a Proportional controller with a time-delay process is analysed. Controller designs based on rightmost eigenvalue location and the dominant eigenvalue method are described.K e y w o r d s: Lambert W function, root locus, time-delay system, stability, eigenvalues
ORIGIN OF THE LAMBERT W FUNCTIONJohann Heinrich Lambert (1728-1777) was a multitalented scientist and philosopher who produced important work in fields such as number theory, optics, meteorology, and astronomy -to name but a few. Some notes on Lambert's life and work may be found in [1]. For further biographical information and details of Lambert's contributions to the mathematical theory of perspective see [2].In 1758 Lambert derived a series solution to the trinomial equationIn 1779 Euler studied the following transformed version of (1)When considering a special case of his series solution to (2) Euler introduced a function w that satisfiesA related function W (z) that solves
ANALYTIC METHOD FOR DRAWING ROOT LOCUS PLOTS FOR SYSTEMS WITH TIME DELAYIf the process G(s) has time delay L > 0 the characteristic polynomial of the system in Fig. 1 iswhere N (s) and M (s) are polynomials of degree n and m respectively, with m ≤ n, s = σ + jω and the gain k is a real number. The roots of (5) are the eigenvalues of the system. In root locus terminology, the poles and zeros of the root locus are given by the roots of N (s) = 0 and M (s) = 0 , respectively. The root locus of (5) consists of the paths traced out in the (σ, ω) plane by the roots of (5) as k varies and these paths are called the root locus branches. As for the delay-free case, the root loci for time-delay systems are symmetrical about the real axis, but unlike the delay-free case, the number of root locus branches is infinite. Setting k = 0 , these branches start at the poles (roots of N (s) = 0 ) and at σ = −∞; setting k = ±∞, these branches terminate at the zeros (roots of M (s) = 0 ) and at σ = +∞. Root locus branches that do not terminate at a zero approach σ = +∞ along asymptotes. These asymptotes are infinite in number and parallel to the real axis. A geometric method, consisting of rules for drawing the root locus of (5) when L = 0 may be found in standard texts such as [12]. An extended set of these rules for drawing the root locus of (5) with L > 0 may be found in [13][14][15]. Alternatively, when L = 0 there is an analytic method [16,17] for drawing the root locus of (5). This method consists of deriving an equation for the root locus curves. A root locus equation for L > 0 will now be derived. In the following the real and imaginary parts of the polynomial P are written as ReP and ImP ...