Step One - Bright Sparks
What you should know first about particles and visible lights is that there is a direct relation between:

1. The number of particles used
2. The size of the visible lights
3. The brightness/thickness of the light sources

All of those will, if raised (ceteris paribus), positively affect the brightness and thickness of the particle stream. The more particles there are per area unit, the more are visible (self explanatory). If the visible size of the light source is higher, you will need fewer to cover an area unit. That's the same with the thickness of the light. I really don't know any rule or formula to tell you which value to choose for either of the three above. This depends on what you try to achieve. The best advice would be to pick apart the demo scenes. Get a feeling by looking at what other people did. You can download the final scene and modify it. Freely change any value and render to see how that change affects the output.

Now, the first particle emitter we will add is the one for the main body of sparks. Fast moving, dense, small, bright and burning out rather soon. Those sparks are being created while the saw cuts through the steel plate. They differ more from each other than, for example, smoke rising from boiling water. Some chunks of metal are larger than others (this affects how long it takes to burn out). And, of course, some are moving faster than others (this affects how they interact with modifiers - a fast moving particle will be less affected by wind than a slow moving one). If you want to determine how fast a particle is to move, remember the speed of the emitter! If they move in different directions, those speeds (speed of emitter and speed of particles) will add up. On the right side you can see my settings for this particles emitter

For now you will have to trust that this will generate the desired stream of particles. Notice a couple things though. First, the emitter is a flat (2x * 1y) plane emitting in one direction. And then, size, duration and speed may vary quite a bit per particle. This emitter sets free those brightly glowing sparks. I used two light sources to simulate this - one for the bright head and one for the not quite so bright tail.

Some of you might remember this from the explosion demo by Maxon. Same idea. You will find the values for those light sources to the left (bright head) and below (tail). Notice how one of the lights is brighter and spherical while the other is less bright but longer along the z-axis.








They are grouped like they are because the tail is to follow the bright spark.