We develop and analyze a system of equations describing a molten linear polymer being extruded in a capillary rheoineter which is operating under the controlled condition that the mean velocity at the capillary inlet is maintained at a constunt iial,ue. Slipping of the melt at the pipe wall is permitted and an evolution equation for the boundary slip parameter is postulated. The combined system governing the mean cross-sectional velocity and slip parameter is
S~O I L~I Ito exhibit relaxation oscillations similar to those observed in actual rheometers.
Capillary EquationsOur modelling effort accounts for what happens in the capillary tube and not in the reservoir. We assume the flow in the capillary is axisymmetric and that all quantities depend only on T , z and t . We let p denote the polymer density, p denote the pressure, and assume that the velocity fieldThe governing equations are the continuity and balance of momentum in the directions e,, es, and e3. At the boundary, r = R, we assume thatwhere F 2 1/4 is a dimensionless quantity depending on z and t and ~~( z , t ) = T(R, z , t ) is the wall shear. We further assume that the polymer is slightly compressible and that the following equation of state holds between the density and pressure:Here, po and po are reference values of the density and pressure and 0 < ~1 << 1 is a dimensionless small parameter.To assess which terms in this sytem are important and which may be neglected we cast the system in dimensionless form. We let, 2E1171 POE2 r = Rrl, z = L z l , and t = T t l and (4) where 0 < ~1 = f << 1. With these scalings the shear stress becomes where A syst,ematic perhrbation analysis where we assume that 0 < E I << 1, 0 < ~2 << 1. and 0 < << 1 514 yields ZAMM . Z. angew. Math. Mech. 76 (1996) S4 and The continuity equation, averaged over the cross-section, reduces to dt, 8Pl -8 (FZ) = 0 , o < z1 < 1 with boundary conditions 8Pl 8z1 -F(O,tl)-(O,tl)= q and p(1,t)l) = 1.
Modelling of the Wall Parameter -FFor molten polymers one typically observes two distinctly different steady flow regimes. In each regime the (dimensionless) mean velocity GI, (dimensionless) pressure gradient 2, (dimensionless) wall shear T: , and slip parameter F are independent of z1 and t l as shown in Figure 1. The first flow regime exists for mean flow rates 6 1 5 2. In this range the capillary flow is Poiseuille with no-slip at the boundary; that is F 5 1/4. The second regime persists for (dimensionless) mean velocities satisfying GI 2 F2c2 > 2 and corresponds to a capillary flow with constant slip parameter F2 > 1/4 at the capillary wall. Hence and Experimental evidence indictes that the wall shear c1 corresponding to the mean flow % exceeds the wall shear c2 corresponding to the mean flow F~c z .In general one finds no stable steady flows for incoming mean flow rates G l ( 0 , t ) q E (2, F2c2) but one observes oscillatory flows. Figure 1 515Motivated by t,hese observations, we assume the existence of a switch curve (scaled as shown in Figure 1) and assume that at. each point tl th...