If you measure the voltage and current with a true-RMS (AC+DC) meter, I don't think there is any condition where true power is greater than apparent power. But if you use other types of meters, particularly with capacitive coupling which removes the DC component, there might be such effect.
Whenever there is asymmetrical distortion on a sine wave, phase angle cannot be measured accurately, or in other words, it becomes undefined. Most phase angle meters use zero crossings to determine the measurement points, and thus a waveform with zero crossings not exactly spaced at 180 degrees will read differently when the polarity is reversed. Also when there is crossover distortion, such that there is a flat spot at the zero crossing, a phase angle meter cannot determine the point accurately.
Power can be measured accurately by using an analog multiplier circuit, such as the AD534. I used this IC to design a DC wattmeter, which can also be used on AC and distorted waveforms. The main problem was isolation from the shunt that was used to measure current. A Hall effect sensor could have been used, but an isolation amplifier did the trick.
You can also measure power using A/D techniques, especially if you use simultaneous sampling converters. You must perform instantaneous products of current and voltage, and then take an average of these sums for true power.
Measuring power in three phase systems, especially with unbalanced and non-linear loads, is even more complex. It is not so difficult in star-connected systems where you can measure each leg and the voltages to neutral, but a greater challenge in delta systems.
Paul