Battery energy storage systems (BESSs) are often used for demand charge reduction through monthly peak shaving. However, during economic analysis in the feasibility stage, BESSs are often sized, and BESS revenue is quantified based on 1 h load and/or solar output data for one year. To quantify the error in the demand charge from coarse-resolution modeling, the effect of two temporal resolutions, 15 min and 1 h, on peak load reduction is compared across a battery rating space defined by the power capacity and energy capacity. A linear program of the system optimizes the peak of the net load and the associated demand charge assuming perfect forecasts. Based on the 15 min load profile of a particular day, a critical power (CP) and critical energy (CE) can be defined, yielding a critical point in the power-energy space. Based on the difference of demand charge (DoDC) across the two load profiles at different temporal resolutions for a real building, the battery rating space is divided into three different regions: oversized region, power-constrained region, and energy-constrained region, which are separated by CP and CE. The DoDC in the power-constrained and energy-constrained regions is explained by time averaging effects and the load sequence at high resolutions. In the power-constrained region of the battery rating space, the difference between the original 15 min peak and the 1 h average peak persists in the optimized net load until the battery power capacity is sufficiently large. In the energy-constrained region, averaging may change the peak period duration, which depends on the sub-hourly sequence of the original load data. Through artificial load data and reordering of real load data, we demonstrate that the sequence effect causes energy-constrained batteries to underestimate peak shaving and demand charge reduction. Demand charge savings were especially sensitive to the BESS power capacity: for a ≈50 kW load, demand charge errors were up to $53 for power-constrained batteries and were an order of magnitude smaller for energy constrained batteries. The power capacity of the battery should be carefully considered when interpreting results from optimizations at low resolutions.