This is an interesting queston, but I need to ask/eduacate as well.
Energy Kilo-Watt-hours. Depends on Current, voltage and time.
AC power is difficult to measure. For this discussion, I'll assume sine wave voltage, current and a phase angle. This isn't true for a few items such as electronic ballasts, computer power supplies.
Power = voltage * Current;; for resistive loads (e.g. a light bulb at steady state)
Power (watts) is also voltage (Volts) * Current (Amps) * PF
Power for 3 phase is a little more complicated. I'll add that if necessary. Nameplate data is OK.
The PF is the power factor and is defined as cos(theta) and varies from -1 to +1; generally 0 to 1. Motors reduce the power factor. The PF is usually on the nameplate.
Motors draw a large start-up current and the load varies based on what's it's cutting etc.
So, you nned to know how long the machine is operating during the day and at what load. A tall order.
You can estimate using nameplate data.
A device called Killawatt.
Kill-A-Watt Electric Usage Monitor Review ? The Gadgeteer
This device can monitor any corder appliance and compute a cost based on the cost per KWH. If you have electric rates that are baseed on say 15 minute peak demands all bets are off.
This wont help you with the lights or the 240 V devices or things that draw more than 15 A at 120 V, but it's a start and it's cheap. Note that a reviewer said, it won't work on a modified sine wave inverter.
I bought a 3 phase energy monitor and it cost about $600, if I remember right. We used it to measure single phase power of a non-sinusoidal voltage of continually varying voltages around 30 VAC nominal. It was made by this company.
Schneider Electric United States - Solutions, Products and Services in Electrical Distribution and Automation and Control before they were gobbled up. The current transformer used selects the current range.
tkrussell has a better link with various products.
Generally these take current transformers to measure current (you pass a wire through the core and it's non-contact) to measure current and the then attach power to the monitor.
The monitor I had was able to be powered by 120 VAC, but measure much lower voltages of non sine wave power.
In light of thie, there will be inaccuraces based on what technique you use:
Some possibilites:
1. Nameplate data
2. Measure at the panel; sub-panel for the shop and get a nice average.
3. Measure individual devices or group of devices at the panel.
4. Use Kill a watt is the loads are not very odd
5. Measure V and I of motor loads and estimate the PF of the motor at 0.8 while cutting, not unloaded.
There are plenty of other devices on the market. Most will use a clamp on meter or torroid coil. The current can be measured using a hall effect sensor or a current transformer.
If you measure currents, it's important to use a Trms meter. It will remove some of the effects of waveform distortion.
Some idea of what you want and why? Cost per widget and overhead expenses. And what you have: 460 V equipment, 3 phase, 120 V? All part of a single sub-panel or a shop attached to a house.
I leave you with this link for now:
Power meters, watt meters, kWh meters - PowerMeterStore.com