Machine vision systems are in essence industrial IT solutions. Often when purchasing this equipment, the project engineer is used to buying a machine, such as a coater or an extruder and categorizes the inspection system as a machine also. 100% inspection solutions are completely solid state. There are no moving parts or consumables. Let's look at the anatomy of a vision system. What are the key components?
- The linescan cameras, optics and enclosures
- The LED lighting with power supplies
- The PC processing the image data, often housed in an industrial console
- The software application for inspecting the image data
So if you have 10 lines and you are looking at purchasing for a two camera system, then that is 20 line scan cameras, 10 LED lights and 10 PCs. The project engineering costs of building a single vision system or ten systems is not a ten to one ratio. It may drop to three to one if done at the same time. The costs associated with the construction and test do not change dramatically but if the equipment is installed at the same time that can reduce the labor costs associated with commissioning and training.
The key part of the puzzle is the software. This is what the operator will be using and will determine the success of your project. The hardware purchased should be totally independent to the software. This means that if you choose to change software vendors in 3 years, the hardware installed should not have to change. That is the true test of a vision network. When looking at the software it is important to make sure that it is scaleable, designed for the 21st century and that there will be upgrades in the future.
The costs of multiple 100% inspection solutions
Lets look at an example. A standalone vision solution from a vertical integrator costs $50,000 and an additional $7,000 to commission. So allowing for insurance, shipping and some line modifications, the project costs could be $70,000 for the end user. So the purchase of 10 machines would be close to $700,000. This approach may finally cost $665,000 with a volume discount, and after one year you would have to allow for service and support. Let's allow a 2.5% cost per annum to cover this. Over 10 years the total cost of ownership would be approx $875,000.
Let's look at building a vision network. On a $50,000 system, the typical break down would be $15,000 for software, $25,000 for hardware, build and test, and $10,000 for services. So the cost of hardware and build for ten systems would be $250,000. Now let's consider the additional $13,000 that the client added to cover project and material cost. We would expect that to drop by 50% when implementing 10 lines together. So adding (13,000/2 by 10) $65,000 this would bring the total to $315,000. Now we have the commissioning and training. This should drop from $70,000 to $35,000 if installed as a package. The total network installed would have cost $350,000. Assume the same 2.5% TCO cost over 10 years, the network will cost $428,750. Finally we must add this software at a perpetual cost of $150,000, the final cost would be $578,750. This is a saving of 44%.
The figures speak for themselves. Breaking the investment of machine vision down to an industrial IT project can yield at least a 40% cost reduction over 10 years. If the client looks at a subscription model for the software, the initial CapEx costs can be reduced as much as 55%. OneBoxVision is a leader in building vision networks for roll to roll, roll to sheet, and sheet to sheet manufacturing. Access our white papers to learn more about this topic.