A solver reports very small residuals for its momentum equations, but the summed velocity fluxes from individual cells remain noticeably nonzero. What is the best conclusion?