Following up on my comment, your problem can be rewritten as as Mixed-Integer Linear Programming:
Minimize z (A)
Subject to
z >= x + 6y (B)
z >= 80 - 3x - 5y (C)
x >= 0 (D)
x <= 10 (E)
y >= 0 (F)
y <= 10 (G)
x, y are integer (H)
MILP are solved using a branch-and-bound algorithm that should be faster than a non-linear solver. One such free solver is lpSolve
:
library(lpSolve)
res <- lp(direction = "min",
objective.in = c(1, 0, 0), # (A) (weights for {z, x, y})
const.mat = rbind(c(1, -1, -6), # (B)
c(1, +3, +5), # (C)
c(0, 1, 0), # (D)
c(0, 1, 0), # (E)
c(0, 0, 1), # (F)
c(0, 0, 1)), # (G)
const.dir = c(">=", ">=", ">=", "<=", ">=", "<="), # (B through G)
const.rhs = c( 0, 80, 0, 10, 0, 10), # (B through G)
int.vec = c(2, 3)) # (H)
res$solution # optimal values for z, x, y respectively
# [1] 33 9 4
I hope this helps. If not, maybe some will find it interesting nonetheless.