In this paper, we study a primal and dual relationship about triangles: For any graph G, let ν(G) be the maximum number of edge-disjoint triangles in G, and τ(G) be the minimum subset F of edges such that G \ F is triangle-free. It is easy to see that ν(G) ≤ τ(G) ≤ 3ν(G), and in fact, this rather obvious inequality holds for a much more general primal-dual relation between k-hyper matching and covering in hypergraphs. Tuza conjectured in 1981 that τ(G) ≤ 2ν(G), and this question has received attention from various groups of researchers in discrete mathematics, settling various special cases such as planar graphs and generalized to bounded maximum average degree graphs, some cases of minor-free graphs, and very dense graphs. Despite these efforts, the conjecture in general graphs has remained wide open for almost four decades.In this paper, we provide a proof of a non-trivial consequence of the conjecture; that is, for every k ≥ 2, there exist a (multi)-set F ⊆ E(G) : |F| ≤ 2kν(G) such that each triangle in G overlaps at least k elements in F. Our result can be seen as a strengthened statement of Krivelevich's result on the fractional version of Tuza's conjecture (and we give some examples illustrating this.) The main technical ingredient of our result is a charging argument, that locally identifies edges in F based on a local view of the packing solution. This idea might be useful in further studying the primal-dual relations in general and the Tuza's conjecture in particular.Clearly, Conjecture 1 is a consequence of the Tuza's conjecture and it implies the standard fractional version of Tuza's conjecture that was resolved by Krivelevich. Krivelevich's proof is based on induction and only implies that τ * k (G) ≤ 2ν(G) for some (very large) k ∈ N. 2 In this paper, we resolve the above question in the affirmative.Theorem 2. For all integer k ≥ 2 and any graph G, τ * k (G) ≤ 2ν(G). Moreover, we can efficiently find the triangle packing solution and the k-multi-transversal that together achieve this bound.The proof of this theorem is inspired by the local search technique. Based on an optimal packing solution, we define a collection of edges in the covering solution via the local view of the packing solution. Our proofs are constructive in nature. Starting with any packing solution, we can either find a multi-transversal or improve the current packing solution. Because the packing solution can be improved at most O(n 3 ) times, hence we get an algorithm to find the triangle packing solution and the k-multi-transversal that together achieve this bound.We believe that the structural insights from this work would be useful in attacking the Tuza's conjecture in the future.