In the initial step, the equations of motion of the bridge–soil and the track subsystems, as shown in Eq. (
2) and Eq. (
10), are rewritten as
$$\begin{aligned}&\left[ \begin{array}{cr} \rho A_\mathrm {b} &{} 0 \\ 0 &{} \rho A_\mathrm {r} \\ \end{array} \right] \left[ \begin{array}{c} \ddot{w}_{\mathrm {b}}(x,t)\\ \ddot{w}_{\mathrm {r}}(x,t)\\ \end{array} \right] + \left[ \begin{array}{cr} EI_\mathrm {b} &{} 0 \\ 0 &{} EI_\mathrm {r} \\ \end{array} \right] \left[ \begin{array}{c} {w}_{\mathrm {b},xxxx}(x,t)\\ {w}_{\mathrm {r},xxxx}(x,t)\\ \end{array} \right] \nonumber \\&\quad = \left[ \begin{array}{rr} -k_\mathrm {f} &{} k_\mathrm {f} \\ k_\mathrm {f} &{} -k_\mathrm {f} \\ \end{array} \right] \left[ \begin{array}{c} {w}_{\mathrm {b}}(x,t)\\ {w}_{\mathrm {r}}(x,t)\\ \end{array} \right] + \left[ \begin{array}{rr} -c_\mathrm {f} &{} c_\mathrm {f} \\ c_\mathrm {f} &{} -c_\mathrm {f} \\ \end{array} \right] \left[ \begin{array}{c} \dot{w}_{\mathrm {b}}(x,t)\\ \dot{w}_{\mathrm {r}}(x,t)\\ \end{array} \right] + \left[ \begin{array}{c} 0\\ f_\mathrm {r}\\ \end{array} \right] . \end{aligned}$$
(26)
The series approximations of the response variables
\(w_\mathrm {b}\),
\(w^{(\mathrm {b})}_{\mathrm {r}}\), and
\(w^{(\mathrm {f})}_{\mathrm {r}}\), shown in Eqs. (
21), (
23), and (
24), are also written in a compact manner,
$$\begin{aligned} \left[ \begin{array}{c} {w}_{\mathrm {b}}(x,t)\\ {w}_{\mathrm {r}}(x,t)\\ \end{array} \right] = \left[ \begin{array}{c} {w}_{\mathrm {b}}(x,t)\\ w^{(\mathrm {b})}_{\mathrm {r}}(x,t)+w^{(\mathrm {f})}_{\mathrm {r}}(x,t)\\ \end{array} \right] = \varvec{\Lambda } \mathbf {h_\mathrm {B}}, \end{aligned}$$
(27)
with
$$\begin{aligned} \varvec{\Lambda } = \left[ \begin{array}{cr} \varvec{\Phi }_\mathrm {b}(x)^\mathrm {T} &{} \mathbf {0} \\ \varvec{\Psi }_\mathrm {r}(x)^\mathrm {T} &{} \varvec{\Phi }_\mathrm {r}(x)^\mathrm {T} \\ \end{array} \right] ~,~~~ \mathbf {h_\mathrm {B}} = \left[ \begin{array}{c} \mathbf {y}_{\mathrm {b}}(t)\\ \mathbf {y}_{\mathrm {r}}(t)\\ \end{array} \right] . \end{aligned}$$
(28)
The matrix
\(\varvec{\Lambda }\) is composed of the vector
\(\varvec{\Phi }_\mathrm {b}(x)\) of the complex eigenfunctions of the bridge–soil subsystem, the vector
\(\varvec{\Phi }_\mathrm {r}(x)\) of the complex eigenfunctions of the track subsystem, and the vector
\(\varvec{\Psi }_{\mathrm {r}}(x)\) of the shape functions,
$$\begin{aligned} \begin{aligned} \varvec{\Phi }_{\mathrm {b}}&= \left[ \Phi _{\mathrm {b}}^{(1)},\Phi _{\mathrm {b}}^{(2)},\ldots ,\Phi _{\mathrm {b}}^{(N_\mathrm {b})},\bar{\Phi }_{\mathrm {b}}^{(1)},\bar{\Phi }_{\mathrm {b}}^{(2)},\ldots ,\bar{\Phi }_{\mathrm {b}}^{(N_\mathrm {b})}\right] ^{\mathrm {T}}, \\ \varvec{\Phi }_{\mathrm {r}}&= \left[ \Phi _{\mathrm {r}}^{(1)},\Phi _{\mathrm {r}}^{(2)},\ldots ,\Phi _{\mathrm {r}}^{(N_\mathrm {r})},\bar{\Phi }_{\mathrm {r}}^{(1)},\bar{\Phi }_{\mathrm {r}}^{(2)},\ldots ,\bar{\Phi }_{\mathrm {r}}^{(N_\mathrm {r})}\right] ^{\mathrm {T}}, \\ \varvec{\Psi }_{\mathrm {r}}&= \left[ \Psi _{\mathrm {r}}^{(1)},\Psi _{\mathrm {r}}^{(2)},\ldots ,\Psi _{\mathrm {r}}^{(N_\mathrm {b})},\bar{\Psi }_{\mathrm {r}}^{(1)},\bar{\Psi }_{\mathrm {r}}^{(2)},\ldots ,\bar{\Psi }_{\mathrm {r}}^{(N_\mathrm {b})}\right] ^{\mathrm {T}}. \end{aligned} \end{aligned}$$
(29)
The vector
\(\mathbf {h_\mathrm {B}}\) combines the two vectors
\(\mathbf {y}_\mathrm {b}(t)\) and
\(\mathbf {y}_\mathrm {r}(t)\) of the subsystem modal coordinates,
$$\begin{aligned} \begin{aligned} \mathbf {y}_\mathrm {b}&= \left[ y_{\mathrm {b}}^{(1)},y_{\mathrm {b}}^{(2)},\ldots ,y_{\mathrm {b}}^{(N_\mathrm {b})},\bar{y}_{\mathrm {b}}^{(1)},\bar{y}_{\mathrm {b}}^{(2)},\ldots ,\bar{y}_{\mathrm {b}}^{(N_\mathrm {b})}\right] ^{\mathrm {T}}, \\ \mathbf {y}_\mathrm {r}&= \left[ y_{\mathrm {r}}^{(1)},y_{\mathrm {r}}^{(2)},\ldots ,y_{\mathrm {r}}^{(N_\mathrm {r})},\bar{y}_{\mathrm {r}}^{(1)},\bar{y}_{\mathrm {r}}^{(2)},\ldots ,\bar{y}_{\mathrm {r}}^{(N_\mathrm {r})}\right] ^{\mathrm {T}}. \end{aligned} \end{aligned}$$
(30)
In the second step, in the combined equations of motion of the bridge–soil and the track subsystems (Eq. (
26)) the response variables are substituted by the series approximation specified in Eq. (
27). Subsequent pre-multiplication by
\(\varvec{\Lambda }^\mathrm {T}\), and integration over the track length
\(-L_0\le x \le L_\mathrm {b}+L_0\) leads with the relation between the modal accelerations and velocities [
20,
28],
$$\begin{aligned} \left[ \begin{array}{l} \ddot{\mathbf {y}}_\mathrm {b} \\ \ddot{\mathbf {y}}_\mathrm {r} \end{array} \right]= & {} \left[ \begin{array}{l} \mathbf {S_\mathrm {b}}\dot{\mathbf {y}}_\mathrm {b} \\ \mathbf {S_\mathrm {r}}\dot{\mathbf {y}}_\mathrm {r} \end{array} \right] \end{aligned}$$
(31)
$$\begin{aligned} \mathbf {S}_\mathrm {b}= & {} \mathrm {diag}\left[ s_{\mathrm {b}}^{(1)},s_{\mathrm {b}}^{(2)},\ldots ,s_{\mathrm {b}}^{(N_\mathrm {b})},\bar{s}_{\mathrm {b}}^{(1)},\bar{s}_{\mathrm {b}}^{(2)},\ldots ,\bar{s}_{\mathrm {b}}^{(N_\mathrm {b})}\right] ,\nonumber \\ \mathbf {S}_\mathrm {r}= & {} \mathrm {diag}\left[ s_{\mathrm {r}}^{(1)},s_{\mathrm {r}}^{(2)},\ldots ,s_{\mathrm {r}}^{(N_\mathrm {r})},\bar{s}_{\mathrm {r}}^{(1)},\bar{s}_{\mathrm {r}}^{(2)},\ldots ,\bar{s}_{\mathrm {r}}^{(N_\mathrm {r})}\right] \end{aligned}$$
(32)
finally to the following coupled set of equations track–bridge–soil subsystem in terms of its modal coordinates,
$$\begin{aligned} \mathbf {A}_\mathrm {B} \dot{\mathbf {h}}_\mathrm {B} + \mathbf {B}_\mathrm {B} \mathbf {h}_\mathrm {B} = \mathbf {f}_{\mathrm {B}}. \end{aligned}$$
(33)
The members of the diagonal matrices
\(\mathbf {S}_\mathrm {b}\) and
\(\mathbf {S}_\mathrm {r}\):
\(s^{(m)}_{\mathrm {b}},\bar{s}^{(m)}_{\mathrm {b}}, m=1,\ldots ,N_{\mathrm {b}}\);
\(s^{(n)}_{\mathrm {r}},\bar{s}^{(n)}_{\mathrm {r}}, n=1,\ldots ,N_{\mathrm {r}}\), are the complex natural frequencies of the bridge–soil and track substructures and their complex conjugate counterparts, respectively (see Appendices A and B). The system matrices
\(\mathbf {A}_\mathrm {B}\) and
\(\mathbf {B}_\mathrm {B}\),
are composed of several sub-matrices, which are explained in the following. The two diagonal sub-matrices
\(\mathbf {A}_\mathrm {b}\) and
\(\mathbf {B}_\mathrm {b}\) contain the coefficients for the orthogonality conditions of the eigenfunctions of the bridge–soil subsystem
\(a_\mathrm {b} ^{(m)}\),
\(\bar{a}_\mathrm {b} ^{(m)}\),
\(b_\mathrm {b} ^{(m)}\),
\(\bar{b}_\mathrm {b} ^{(m)}\),
\(m=1,\ldots ,N_\mathrm {b}\),
$$\begin{aligned} \begin{aligned} \mathbf {A}_\mathrm {b}&= \mathrm {diag}\left[ a_\mathrm {b} ^{(1)},a_\mathrm {b} ^{(2)},\ldots ,a_\mathrm {b} ^{(N_\mathrm {b})},\bar{a}_\mathrm {b} ^{(1)},\bar{a}_\mathrm {b} ^{(2)},\ldots ,\bar{a}_\mathrm {b} ^{(N_\mathrm {b})}\right] ~,\\ \mathbf {B}_\mathrm {b}&= \mathrm {diag}\left[ b_\mathrm {b} ^{(1)},b_\mathrm {b} ^{(2)},\ldots ,b_\mathrm {b} ^{(N_\mathrm {b})},\bar{b}_\mathrm {b} ^{(1)},\bar{b}_\mathrm {b} ^{(2)},\ldots ,\bar{b}_\mathrm {b} ^{(N_\mathrm {b})}\right] , \end{aligned} \end{aligned}$$
(35)
described in detail in Appendix A. The corresponding matrices for the stand-alone track subsystem,
\(\mathbf {A}_\mathrm {r}\) and
\(\mathbf {B}_\mathrm {r}\), read as
$$\begin{aligned} \begin{aligned} \mathbf {A}_\mathrm {r}&= \mathrm {diag}\left[ a_\mathrm {r} ^{(1)},a_\mathrm {r} ^{(2)},\ldots ,a_\mathrm {r} ^{(N_\mathrm {r})},\bar{a}_\mathrm {r} ^{(1)},\bar{a}_\mathrm {r} ^{(2)},\ldots ,\bar{a}_\mathrm {r} ^{(N_\mathrm {r})}\right] ~,\\ \mathbf {B}_\mathrm {r}&= \mathrm {diag}\left[ b_\mathrm {r} ^{(1)},b_\mathrm {r} ^{(2)},\ldots ,b_\mathrm {r} ^{(N_\mathrm {r})},\bar{b}_\mathrm {r} ^{(1)},\bar{b}_\mathrm {r} ^{(2)},\ldots ,\bar{b}_\mathrm {r} ^{(N_\mathrm {r})}\right] . \end{aligned} \end{aligned}$$
(36)
Details on the coefficients
\(a_\mathrm {r} ^{(n)}\),
\(\bar{a}_\mathrm {r} ^{(n)}\),
\(b_\mathrm {r} ^{(n)}\),
\(\bar{b}_\mathrm {r} ^{(n)}\),
\(n=1,\ldots ,N_\mathrm {r}\), in these matrices are found in Appendix B. The sub-matrices
\(\Delta \mathbf {M}_\mathrm {b}\),
\(\Delta \mathbf {C}_\mathrm {b}\), and
\(\Delta \mathbf {K}_\mathrm {b}\) read as
$$\begin{aligned} \Delta \mathbf {M}_\mathrm {b}= & {} {} \rho A_\mathrm {r} \int _{-L_0}^{L_b+L_0}\varvec{\Psi }_\mathrm {r}\varvec{\Psi }_\mathrm {r}^{\mathrm {T}}\mathrm {d} x, \nonumber \\ \Delta \mathbf {C}_\mathrm {b}= & {} {} - c_\mathrm {fb}\Bigg ( \int _{0}^{L_\mathrm {b}}\varvec{\Phi }_\mathrm {b}\varvec{\Psi }_\mathrm {r}^\mathrm {T} \mathrm {d} x+\left( \int _{0}^{L_\mathrm {b}}\varvec{\Phi }_\mathrm {b}\varvec{\Psi }_\mathrm {r}^\mathrm {T} \mathrm {d} x\right) ^\mathrm {T} \nonumber \\&- \int _{0}^{L_\mathrm {b}}\varvec{\Phi }_\mathrm {b}\varvec{\Phi }_\mathrm {b}^\mathrm {T} \mathrm {d} x\Bigg ) \nonumber +\int _{-L_0}^{L_\mathrm {b}+L_0} c_\mathrm {f} \varvec{\Psi }_\mathrm {r} \varvec{\Psi }_\mathrm {r}^\mathrm {T} \mathrm {d} x,\nonumber \\ \Delta \mathbf {K}_\mathrm {b}= & {} {} - k_\mathrm {fb}\Bigg ( \int _{0}^{L_\mathrm {b}}\varvec{\Phi }_\mathrm {b}\varvec{\Psi }_\mathrm {r}^\mathrm {T} \mathrm {d} x+\left( \int _{0}^{L_\mathrm {b}}\varvec{\Phi }_\mathrm {b}\varvec{\Psi }_\mathrm {r}^\mathrm {T} \mathrm {d} x\right) ^\mathrm {T} \nonumber \\&- \int _{0}^{L_\mathrm {b}}\varvec{\Phi }_\mathrm {b}\varvec{\Phi }_\mathrm {b}^\mathrm {T} \mathrm {d} x\Bigg ) \nonumber + \int _{-L_0}^{L_\mathrm {b}+L_0} k_\mathrm {f} \varvec{\Psi }_\mathrm {r} \varvec{\Psi }_\mathrm {r}^\mathrm {T} \mathrm {d} x \nonumber \\&+ EI_\mathrm {r}\int _{-L_0}^{L_\mathrm {b}+L_0}\varvec{\Psi }_\mathrm {r}\varvec{\Psi }_{\mathrm {r},xxxx}^\mathrm {T} \mathrm {d} x, \end{aligned}$$
(37)
and the sub-matrices
$$\begin{aligned} \begin{aligned} \mathbf {M}_{\mathrm {br}}&= \mathbf {M}_{\mathrm {rb}}^\mathrm {T} = \rho A_\mathrm {r} \int _{-L_0}^{L_b+L_0}\varvec{\Psi }_\mathrm {r}\varvec{\Phi }_\mathrm {r}^\mathrm {T} \mathrm {d} x,\\ \mathbf {C}_{\mathrm {br}}&= \mathbf {C}_{\mathrm {rb}}^\mathrm {T} = \int _{-L_0}^{L_\mathrm {b}+L_0} c_\mathrm {f} \varvec{\Psi }_\mathrm {r} \varvec{\Phi }_\mathrm {r}^\mathrm {T} \mathrm {d} x - c_\mathrm {fb}\int _{0}^{L_\mathrm {b}}\varvec{\Phi }_\mathrm {b}\varvec{\Phi }_\mathrm {r}^\mathrm {T} \mathrm {d} x \end{aligned} \end{aligned}$$
(38)
result from coupling of the bridge–soil and the track subsystems.
The force vector
\(\mathbf {f}_\mathrm {B}\) in Eq. (
33),
$$\begin{aligned} \mathbf {f}_\mathrm {B} = \left[ \begin{array}{l} \mathbf {f}_\mathrm {b} \\ \mathbf {f}_\mathrm {r} \end{array} \right] , \end{aligned}$$
(39)
captures the effect of the vehicle interaction forces
\(F_{k}^{(j)}\),
\(j=1,\ldots ,N_c\),
\(k=1,\ldots ,4\) (for the considered four axle vehicles), separately for the bridge–soil subsystem by the vector
\(\mathbf {f}_\mathrm {b}\) and for the track subsystem by the vector
\(\mathbf {f}_\mathrm {r}\). The vector
\(\mathbf {f}_\mathrm {b}\) reads as
$$\begin{aligned} \mathbf {f}_\mathrm {b} = \sum _{j=1}^{N_\mathrm {c}}\varvec{\varPsi }_\mathrm {r}^{(j)} \varvec{\Pi }^{(j)} \mathbf {F}^{(j)}~,~~~\mathbf {F}^{(j)} = \left[ F_{1}^{(j)},F_{2}^{(j)},F_{3}^{(j)},F_{4}^{(j)}\right] ^\mathrm {T}, \end{aligned}$$
(40)
where
$$\begin{aligned} \varvec{\Pi }^{(j)} = \mathrm {diag}\left[ \Pi (t,t_{\mathrm {A}1}^{(j)},t_{\mathrm {D}1}^{(j)}),\Pi (t,t_{\mathrm {A}2}^{(j)},t_{\mathrm {D}2}^{(j)}),\Pi \left( t,t_{\mathrm {A}3}^{(j)},t_{\mathrm {D}3}^{(j)}\right) ,\Pi \left( t,t_{\mathrm {A}4}^{(j)},t_{\mathrm {D}4}^{(j)}\right) \right] \end{aligned}$$
(41)
and
$$\begin{aligned} \begin{aligned}&\varvec{\varPsi }_{\mathrm {r}}^{(j)} = \left[ \varvec{\Psi }_\mathrm {r}\left( x_{1}^{(j)}\right) ,\varvec{\Psi }_\mathrm {r}\left( x_{2}^{(j)}\right) ,\varvec{\Psi }_\mathrm {r}\left( x_{3}^{(j)}\right) , \varvec{\Psi }_\mathrm {r}\left( x_{4}^{(j)}\right) \right] , \end{aligned} \end{aligned}$$
(42)
with
\(x_{i}^{(j)}\) (
\(i = 1,\ldots ,4\)) corresponding to the axle position of each of the four axles of the
jth vehicle. The vector
\(\mathbf {f}_\mathrm {r}\) is computed according to
$$\begin{aligned} \mathbf {f}_\mathrm {r} = \sum _{j=1}^{N_\mathrm {c}}\varvec{\varPhi }_\mathrm {r}^{(j)} \varvec{\Pi }^{(j)} \mathbf {F}^{(j)}, \end{aligned}$$
(43)
where
$$\begin{aligned} \varvec{\varPhi }_{\mathrm {r}}^{(j)} = \left[ \varvec{\Phi }_\mathrm {r}\left( x_{1}^{(j)}\right) ,\varvec{\Phi }_\mathrm {r}\left( x_{2}^{(j)}\right) ,\varvec{\Phi }_\mathrm {r}\left( x_{3}^{(j)}\right) ,\varvec{\Phi }_\mathrm {r}\left( x_{4}^{(j)}\right) \right] . \end{aligned}$$
(44)
In case of a single load model, where interaction forces in
\( \mathbf {F}^{(j)}\) are assumed to be constant and of known value (i.e., the static axle loads of the train), Eq. (
33) can be solved numerically by applying the fourth-order Runge–Kutta method [
6].