If you do nothing you make no money.
If you get help you may make money but there is a risk your idea may get stolen.
You need to try and weigh up the risk against the advantage.

If you have convinced yourself you really have a brilliant idea then you should investigate patents, copyright, contracts and non-disclosure agreements.  Don't make you idea public if you think you might be able to get a patent because, once you do, you won't be able to get a patent.

People seem to manage to turn very ordinary ideas in massive profits. Isn't there a billionaire who used to live in a car and survive by collecting bottles for their return deposit who then made his money with shampoo!!!!!

I would think about trying to keep things as simple as possible an minimising the distance between components.
Why not just measure the temperature at the reservoir outflow and at the return?
If the return is not several degrees cooler than the outflow there is no point in pumping.
You could use the ambient temperature and the reservoir outflow temperature to decide when to start pumping.

Attaching temperature sensors to the outside of heat conductive pipes will give a good reading.

It sounds like you have the two "ends" for your system.
The arduino can read an input and the mac can play VLC (I don't know what "terminal" is).

You need to think what can join the two.
The Arduino can send serial data so possibly instead of lighting the LED you could get it to send a message to the mac.
On the mac you have a program waiting for the message and then playing a video.

Try getting an LED to blink first.
http://www.arduino.cc/en/Tutorial/BlinkingLED

Then look around the other tutorials, add a switch and so on.

If you told me I would find 100msps ADCs in there I would happily dive.

You obviously don't visit recycling centres, they have just about everything.
It is very easy to come away with more than you went in with.

9600BPS and 8 data bits, no party and 1 stop bit

The 9600BPS is the speed (baud rate) at which data is sent, i.e. 9,600 bits per second.
Each data character is made up of a number of bits in this case 8 plus a stop bit makes 9, so 9600 means it could send about 1067 characters every second if it sent data continuously.
The stop bit helps tell where one character ends and the next begins.
Parity is a means of helping detect if a character has been corrupted, in this case parity is not being used.

In general you just setup the transmitter and the receiver to use the same settings and you don't need to worry about the bits. Whole characters are written at the transmitter and read at the receiver, the fact that the were sent as bits is normally invisible to you.

Providing the receiver is able to read the characters faster than the transmitter sends them all is well.
Things get more complicated if your receiver is busy doing other things.

Using serial data is relatively slow so you need to think about how fast your altitude will change and that the interface speed will let you take readings fast enough to get the accuracy you want. No problem in a balloon, but worth checking in a rocket.

This site lists load cells that look as if they are probably suitable for what you want
http://www.alibaba.com/product-gs/672691783/High_accuracy_load_cell_10kg_20kg.html
You should check price, minimum order quantity, how you interface to the cell and that it will do the job you want.

From a quick look the applications they mention, the weight ranges and the platform sizes all suggest they might be suitable for what you have in mind.

I think you would need a mechanism that creates water droplets at a constant rate, or perhaps a number of selectable rates.
You then need a strobe light and by varying its frequency you will be able to make the movement of the droplets appear to change.
I think you would be better with a strobe light rather than LEDs. The strobe might have an external trigger and an Arduino could be used to provide the trigger if you wanted to change the strobe rate under program control.

The specific heat of water is 4.187 kJ/kgK.
In other words you need to input 4.187kJ to raise the temperature of 1kg of water by 1 degree centigrade or kelvin.
That is a lot of energy, which explains why water and water mist are so good at putting out fires.

If you know the wattage of your heater and the mass of water you have to heat you can probably calculate fairly accurately how long the heater needs to be on for to change the temperature by a desired amount. The main thing introducing an inaccuracy will be how fast the tank loses heat, which will depend on the ambient temperature and how hot you are trying to make the water.

You are going to use a PID for the control but doing some calculations will help you tune it and confirm your results.

Ladder still rocks and is a powerful tool for process control and machine control.

Absolutely, ladder logic is a very easy to use yet powerful tool for solving many problems.
However it is not a good idea to devise a solution to your problem in ladder logic and then implement it in C/C++, which is what ash901226 seems to be trying to do.

If your solution is in ladder then implement it in ladder, not C or C++.
Likewise if your solution is devised in C/C++ implement it in C/C++, don't tie yourself in knots trying to convert to ladder. 

Your first objective has to be to get paid.
Therefore you need to investigate what hardware/software you will need to interface with systems that will provide you with payment. It also has to be very obvious and easy for the user to use your system.

In addition need to consider your profit margins. If you use gprs/gsm, and a payment system then how much will you have to pay for those services?

Security would also probably be quite a big concern. I guess vending machines are quite good at rejecting fake coins, they certainly reject enough of my cash. You will need to make sure somebody does not figure out how to make your machines pump out all their goodies and then post their hack on the internet.

the only reason for me to use PLC way of programming is that it is something i fully understand

Limiting what you do to what you already know is a bad idea, learn new things.

If you insist on programming the Arduino in Ladder logic then what you should look for is something that pre-processses ladder logic into C code.

Somebody seems to have done this here;
http://electronics4dogs.blogspot.ie/2011/07/ldmicro-and-arduino.html

However even if you were going to use something like this I would suggest you need to learn how to program the Arduino fully in C first, otherwise what is happening will just be "magic" and you will not really know what is going on.

I am a little off topic but, out of interest, how does a CO detector work?

I can figure out roughly how a smoke detector might work but telling the difference between CO and N2, O2, Ar, and CO2 seems like a neat trick.

Is is a chemical reaction, can the detector be used multiple times?

Go with the basics, then work your way up.

- good advice.

Start with one LED and photoresistor, drop/throw balls though the beam - can you detect them accurately? You might try changing the ambient light as well. If you can get the simple case working everything else falls into place.

In my experience if you build and test a project incrementally you understand everything well and the end product is solid with few bugs. On the other hand if you go straight for a final solution you will not have the same depth of understanding, it is much more difficult to test the product fully, and you may find the product does not work at all but you cannot figure out why. Even if it seems to work there will probably be more bugs in it.

I have experience with RS-485 but Lefty has made a career of using it in very bad environments

Why use RS485 though rather than RS422 if there is no need to multidrop?