In industrial and scientific practice, great emphasis is placed on the reliable computation of high-fidelity solution curves of nonlinear dynamical systems. However, prohibitively small step sizes, non-convergence, and unintentional path reversal are commonly encountered when computing solution curves of industry-grade models. The sequential nature inherent to continuation and limited robustness present a challenge that has rarely been addressed in the literature. Here we present a recently published method for obtaining the targeted solution curves in a parallelized way [1]. The point of departure is a low-fidelity model of reduced order and/or with selected nonlinear or coupling terms neglected. The simplifications must be sufficiently drastic to allow the computation of an approximate solution curve with reasonable effort. A subset of relevant points along this approximate curve is then selected, and, starting from these initial guesses, points on the targeted solution branch of the high-fidelity model are computed iteratively. It is important to note that this iterative correction towards the high-fidelity solution curve can be performed independently for each point and is therefore ideally suited to parallel computation. The proposed generic concept is demonstrated for the computation of nonlinear frequency response curves using the Harmonic Balance method. We first exploit the single nonlinear mode theory to efficiently obtain a low-fidelity approximate solution. We then refine this prediction such that the assumptions underlying the single nonlinear mode theory no longer need to be strictly satisfied. The benefits and limitations of the proposed approach are investigated for an industry-grade model of a low-pressure turbine bladed disk featuring multiple contact interfaces. The method is available, along with a selection of numerical examples, as a branch PEACE (Parallelized Re-analysis Of Solution Curves) of the open source tool NLvib [2].