Does it make sense to compare two distances computed with Dynamic Time Warping?

Assume we are measuring the temperature $f_i(A,T_k)$ of the engine $i \in \{1,2\}$ of a given boat $A$, at $f_s = 1\text{Hz}$, for timesteps $T_k$ and some trips $k \in \{1,2,...,n\}$.

Denoting $d_{1,2}^k(A)$ the dynamic time warping between $f_1(A, T_k)$ and $f_2(A, T_k)$, does it make sense to compare $d_{1,2}^a(A)$ and $d_{1,2}^b(A)$ for $a\neq b$ ?

The dynamic time warping is based on an optimal path which depends on the two time series we want to compute the distance on. As a result, I have some doubts about the idea of a comparison. I am not able to interpret this.

Does someone have some ideas? Any help would be appreciated.