The role of friction on heave-induced motion of offshore drilling systems: modelling, analysis, and numerical experiments

The estimate (?) together with (?) implies that, for sufficiently small \delta \in \mathbb{R}_{>0}, the mapping \bm{\Phi} : \mathcal{B}_{\rho}(\bm{Y}_0) \to \mathcal{X}_\delta defined as in Eqs. (?) and (?) is a strict contraction on \mathcal{B}_{\rho}(\bm{Y}_0). By Banach’s fixed-point theorem, \bm{\Phi}(\bm{Y}) admits a unique fixed point \bm{Y}=(\bm{X},\widetilde{\bm{w}}) \in C^0([0,\delta];C^0([0,1];\mathbb{R}^{n_{Y}})), which is the unique generalised solution of the transformed system (?). Since the exponential change of variables (?) is an isomorphism of C^0([0,1];\mathbb{R}^8), the original system (?) is also locally well-posed in C^0([0,\delta];C^0([0,1];\mathbb{R}^{n_{X}+8})) \cong C^0([0,1]\times[0,\delta];\mathbb{R}^{n_{X}+8}). Moreover, repeating the procedure above with ICs \bm{Y}(\cdot,\delta) = (\bm{X}(\cdot,\delta),\widetilde{\bm{w}}(\cdot,\delta)) \in C^0([0,1];\mathbb{R}^{n_{Y}}) satisfying the zeroth-order compatibility condition at the initial time \delta, a new solution may be obtained on an interval [\delta,\delta+\eta], where \eta \in \mathbb{R}_{>0} depends only on the IC \bm{Y}(z,\delta).

Blow-up alternative. Let now t \ensuremath{_\textnormal{max}}\in \mathbb{R}_{>0} be the maximal existence time of the solution. If t \ensuremath{_\textnormal{max}}<\infty, then \lim_{t\uparrow t \ensuremath{_\textnormal{max}}} \norm{\bm{Y}(\cdot,t)}_{C^0([0,1];\mathbb{R}^{n_Y})}= \infty, since otherwise there would exist a sequence t_n\uparrow t \ensuremath{_\textnormal{max}} such that \norm{\bm{Y}(\cdot,t)}_{C^0([0,1];\mathbb{R}^{n_Y})} \leq R for all n and some R \in \mathbb{R}_{\geq 0}. From the arguments above, this would imply that, for each t_n close enough to t \ensuremath{_\textnormal{max}}, the solution \bm{Y}(z,t) obtained on [0,t_n] can be extended on [0,t_n + \eta] with \eta \in \mathbb{R}_{>0} independent of t_n, and t_n + \eta > t \ensuremath{_\textnormal{max}}, contradicting its maximality. Finally, recalling the exponential transformation (?), it may be easily concluded that \lim_{t\uparrow t \ensuremath{_\textnormal{max}}} \norm{\bm{Y}(\cdot,t)}_{C^0([0,1];\mathbb{R}^{n_Y})}= \infty is equipollent to Eq. (3.1).

Consider t \ensuremath{_\textnormal{max}}, the maximal existence time of the local solution. From the blow-up alternative already proved in Theorem 3.1 , either t \ensuremath{_\textnormal{max}}=\infty, or Eq. (3.1) holds. However, under the stated assumptions, the a priori estimate (?) shows that \norm{(\bm{X}(\cdot,t), \bm{w}(\cdot,t))}_{C^0([0,1];\mathbb{R}^{n_{X}+8})} remains bounded on every finite interval [0,T], therefore excluding finite-time blow-up and implying necessarily t \ensuremath{_\textnormal{max}}=\infty. In particular, the solution satisfies (\bm{X},\bm{w})\in C^0(\mathbb{R}_{\geq 0};C^0([0,1];\mathbb{R}^{n_{X}+8})) \cong C^0([0,1]\times \mathbb{R}_{\geq 0};\mathbb{R}^{n_{X}+8}).

The proof starts by noting that, under the stated assumptions, Theorem 3.1 ensures the existence of a unique, maximal generalised solution (\bm{X},\bm{w}) \in C^0([0,t \ensuremath{_\textnormal{max}});C^0([0,1];\mathbb{R}^{n_X+8})) for some t \ensuremath{_\textnormal{max}} \in \mathbb{R}_{>0}. To prove that the solution is global and bounded, the variables \mathbb{R}^2 \ni \bm{\xi}_i(z,t) = [\xi_i^+(z,t)\; \xi_i^-(z,t)]^{\mathrm{T}}, i \in \{1,2,3,4\}, are first introduced, with

The above transformation is an isomorphism of C^0([0,1];\mathbb{R}^2). Indeed,

Moreover, inserting (Proof) into the PDE dynamics (?) and BC (?) provides

where the matrix-valued functions \mathbf{A}_i \in C^0([0,1];\mathbf{M}_2(\mathbb{R})), i \in \{1,2,3,4\}, read

Accordingly, defining z_i(z,t,\tilde{t}) \triangleq z_i^+(z,t,\tilde{t}) and \tau_i(z,t) \triangleq \tau_i^+(z,t), i \in \{1,2,3,4\}, as in (?), integrating Eq. (Proof) along the characteristic lines and imposing the BC (Proof) yields

where the spatial state-transition matrices \mathbf{\Phi}_i(z,\tilde{z}) \in \mathbf{M}_2(\mathbb{R}) generated by \mathbf{A}_i(z), i \in \{1,2,3,4\}, write

The next step consists of estimating upper bounds for the different components of the solution (Proof). To this end, the vectors \mathbb{R}_{\geq 0}^4 \ni \bm{W}(t) \triangleq [W_{1}\; W_{2}\; W_{3}\; W_{4}]^{\mathrm{T}}, \mathbb{R}_{\geq 0}^4 \ni \bm{W}_0(t) = [W_{0,1}\; W_{0,2}\; W_{0,3}\; W_{0,4}]^{\mathrm{T}} \triangleq \bm{W}(0), and \bm{\beta}\in \mathbb{R}_{\geq 0}^3 are first defined as

Utilising Eq. (Proof), it may be observed that the condition (3.3) implies that

and, since \mathbf{B}(\bm{\phi}_f(\overline{\bm{w}}),\bm{\eta},\overline{\bm{w}}) \in \mathbf{M}_{4\times 3}(\mathbb{R}_{\geq 0}), also

Moreover, by Eq. (3.3), \mathbf{A}(0)\succeq \mathbf{I}_4, yielding \bm{W}_0<\overline{\bm{w}} and thus \bm{w}_0 \in \operatorname{int}\overline{\mathcal{B}}_w. Consequently, by continuity of t\mapsto\bm{w}(\cdot,t) in C^0([0,1];\mathbb{R}^8), the first-exit time

with the convention \inf\varnothing=t_{\max}, is strictly positive, i.e., t_w \in \mathbb{R}_{>0}. By the definition of t_w, \bm{w}(\cdot,t) \in \operatorname{int}\overline{\mathcal{B}}_w on every interval t\in[0,T] with T \in (0,t_w); therefore, all the bounds in Eq. (?), and in particular the coefficients \phi_{i0}(\overline{\bm{w}}), \phi_{ij}(\overline{\bm{w}}), and \phi_{if}(\overline{\bm{w}}), are valid throughout [0,T].

Then, defining the functions \widetilde{\bm{g}}_i, \widetilde{\bm{h}}_i \in C^0([0,1]\times[0,T];\mathbb{R}^2) as \widetilde{\bm{g}}_i(z,t) \triangleq \bm{g}_i(\bm{X}(z,t),\bm{\xi}(z,t),z) and \widetilde{\bm{h}}_i(z,t) \triangleq \bm{H}_i(\bm{\xi}^-(0,t),\bm{U}(t)), i \in \{1,2,3,4\}, and introducing

provides, in conjunction with Eq. (?),

Moreover,

with \alpha_i(z,\tilde{z}), i \in \{1,2,3\} as in Eq. (3.3).

Hence, combining Eq. (Proof) with (?), (3.3), and (?), and exploiting the fact that the ICs vanish for t\geq \lambda_i^{-1}, it is possible to infer the existence of an arbitraty constant \varrho \in \mathbb{R}_{>0} verifying

with \mathbb{R}_{\geq 0} \ni K_{if} (\overline{\bm{w}})\triangleq K_{i3}\phi_{if}(\overline{\bm{w}}), i \in \{1,2,3,4\}. Furthermore, using the fact that the inequality

holds for every \eta_{i1},\eta_{i2},\eta_{i3} \in \mathbb{R}_{>0} satisfying \sum_{j=1}^3 \eta_{ij}^{-1}=1, i \in \{1,2,3,4\}, the following ISS estimate is obtained from Eq. (?):

Since the semigroup property holds for the solution (Proof), i \in\{1,2,3,4\}, it is possible to invoke the fading memory inequality [Krstic2, Lemma 7.1] to conclude the existence of an arbitrary small constant \epsilon \in \mathbb{R}_{>0} such that

where \delta \in (0,\varrho) depends only on \varrho and \lambda_i,K_{i1}, i \in \{1,2,3,4\}. Consequently, recalling the bounds (?) and defining the vectors \mathbb{R}_{\geq 0}^4 \ni \bm{W}^\delta(t) \triangleq [W_1^\delta(t)\; W_2^\delta(t)\; W_3^\delta(t)\; W_4^\delta(t)]^{\mathrm{T}} and \bm{\beta}^\delta (t) \in \mathbb{R}_{\geq 0}^3 as

using both bounds in (?), taking the supremum in (?), and collecting the four resulting inequalities, provides

where, to alleviate the notation, the following matrices \mathbf{\Theta}_w \in \mathbf{M}_4(\mathbb{R}_{\geq 0}) and \mathbf{B}_w \in \mathbf{M}_{4\times 3}(\mathbb{R}_{\geq 0}) have been defined:

By virtue of Eq. (3.3), \rho(\mathbf{\Theta}_{w})<1, and, since the spectral radius is continuous, there exists \epsilon_0\in \mathbb{R}_{>0} such that \rho( (1+\epsilon)\mathbf{\Theta}_{w} )<1 for all \epsilon \in (0,\epsilon_0]. In particular, the matrix-valued function \mathbf{R}_w \in C^0([0,\epsilon_0];\mathbf{M}_4(\mathbb{R}_{\geq 0})), reading

is well-defined and componentwise nonnegative. Thus, premultiplying Eq. (?) by \mathbf{R}_w(\epsilon), then multiplying by \ensuremath{\mathrm{e}}^{-\delta t}, and recalling Eq. (Proof) yields

since \ensuremath{\mathrm{e}}^{-\delta t}\bm{\beta}^\delta(t)\leq\bm{\beta} componentwise. It remains to show that \varrho \in \mathbb{R}_{>0} and \epsilon \in (0,\epsilon_0] may be selected so that \bm{w}(\cdot,t) \in \overline{\mathcal{B}}_w for all times. To this end, the following function \bm{\mathcal{F}} \in C^0(\mathbb{R}_{\geq 0}\times[0,\epsilon_0];\mathbb{R}^4) is conveniently introduced:

From Eq. (3.3), it follows that

since multiplication by the componentwise nonnegative matrix \mathbf{R}_w(\epsilon) preserves the strict componentwise inequality. Furthermore, the mapping (\varrho,\epsilon)\mapsto \bm{\mathcal{F}}(\varrho,\epsilon) is continuous in a neighbourhood of (0,0), implying the existence of \varrho_0 \in \mathbb{R}_{>0}, \epsilon_1\in(0,\epsilon_0], and \mathbb{R}_{>0}^4 \ni \bm{\nu} \triangleq [\nu_1\; \nu_2\; \nu_3\; \nu_4]^{\mathrm{T}} such that

for every (\varrho,\epsilon)\in(0,\varrho_0]\times(0,\epsilon_1]. Hence, choosing sufficiently small (\varrho,\epsilon)\in(0,\varrho_0]\times(0,\epsilon_1], and using the fact that \ensuremath{\mathrm{e}}^{-\delta t}\leq1, Eqs. (?) and (?) give

Since T\in (0,t_w) was arbitrary, Eq. (?) holds for every t\in[0,t_w). Suppose, by contradiction, that t_w<t \ensuremath{_\textnormal{max}}: passing to the limit as t\uparrow t_w in Eq. (?), and exploiting the continuity of t\mapsto\bm{w}(\cdot,t) provides

implying that \bm{w}(\cdot,t_w) \in \operatorname{int}\overline{\mathcal{B}}_w. By continuity, there exists h\in \mathbb{R}_{>0} such that \bm{w}(\cdot,t)\in \operatorname{int}\overline{\mathcal{B}}_w for all t\in[t_w, t_w+h], contradicting the definition of t_w in Eq. (Proof). Thus, t_w = t \ensuremath{_\textnormal{max}}, and \sup_{t\in[0,t \ensuremath{_\textnormal{max}})} \norm{\bm{w}(\cdot,t)}_{C^0([0,1];\mathbb{R}^8)} <\infty. Consequently, recalling Eq. (3.1), it may be concluded that \sup_{t\in[0,t \ensuremath{_\textnormal{max}})} \norm{ ( \bm{X}(\cdot,t),\bm{w}(\cdot,t) )}_{C^0([0,1];\mathbb{R}^{n_X+8})} <\infty, which excludes the blow-up alternative of Theorem 3.1. Hence, t_{\max}=\infty, and Eq. (?) proves that \bm{w}(\cdot,t)\in\overline{\mathcal{B}}_w for all t\in\mathbb{R}_{\geq0}. Finally, since the bound (?) is valid globally in time, combining (?) with (3.1), and using equivalence of norms on finite-dimensional spaces, permits deducing the existence of constants \Theta_X,\Theta_w,\gamma\in\mathbb{R}_{>0} such that the estimate (3.3) holds for all t\in\mathbb{R}_{\geq0}.