Using microsimulations of vehicular dynamics, we studied the effects of several proposed infrastructure developments to the mean travel delay time~\(\Delta\) and mean speed~\(\Sigma\) of vehicles passing a busy three-road fork, particularly in the non-signalized roundabout junction of Lower Bicutan, Taguig City, Metro Manila. We designed and implemented multi-agent-based microsimulation models to mimic the autonomous driving behavior of heterogeneous individuals and measured the effect of various proposed infrastructure developments on~\(\Delta\) and~\(\Sigma\). Our aim is to find out the best infrastructure development from among three choices being considered by the local government for the purpose of solving the traffic problems in the area. We created simulation models of the current vehicular traffic situation in the area using the mean travel times~\(\tau\) of statistically sampled vehicles to show that our model can simulate the real-world at a significance level of \(\alpha=0.05\). Based on these models, we then simulated the effect of the proposed infrastructure developments on~\(\Delta\) and~\(\Sigma\) and used these metrics as our basis of comparison. We found out that the proposed widening of one fork from two lanes to three lanes has the most improved metrics at the same \(\alpha=0.05\) compared to the metrics we observed in the current situation. Under this infrastructure development, the~\(\Delta\) increases linearly (\(R^2=0.98\)) at the rate of 1.03~\(s\), while the~\(\Sigma\) decreases linearly (\(R^2>0.99\)) at the rate of 0.14~\(km/h\) per percent increase in the total vehicle volume~\(\mathcal{V}\).