The pioneering work of Gupta and Kumar  has inspired many subsequent studies on asymptotically achievable per node throughput and delay analysis for wireless networks. In this paper, we evaluate the delay and maximum achievable per node throughput of a heterogeneous wireless mesh network (HWMN). We derive an analytical model which incorporates diffusion approximation method  and then compare with the results of per node information theoretic capacity done by P. Li and Y.G. Fang . The network topology consists of n normal nodes and m more powerful helping nodes in a rectangular area. Results show that higher magnitude of exponent w gives higher absorption probability which consequently reduces the delay. We found that the derived maximum achievable per node throughput is inversely proportional to the average number of hops traversed by packets, node density, and number of squares in the network. Also, we are able to obtain the result stated in  i.e., maximum achievable per-node throughput is according to ΘW1/√nlogn, where W1 is normal node transmission bandwidth.