Download Wireless MFI Monitoring Network Proposal

WIRELESS MFI MONITORING NETWORK
PROPOSAL
For Fiberglass Pipes and Tanks
Abstract
Used as a permanent, low cost, wireless system to monitor your fiberglass installations and
check for structural problems due to chemical permeation of the inner lining.
Be able to monitor the expected degradation of your system at the click of a button, log the
data and store it to your computer without climbing a single ladder.
Andrew P. Bishop
[email protected]
Introduction:
A wireless MFI monitoring network is a special type of electronic system that is mounted to
fiberglass pipes and tanks. It serves as a convenient means of performing the MFI testing procedure
developed by BRER Technical, allowing you to obtain and archive the data on all your fiberglass
hardware in a matter of minutes. The procedure involves two different tests that will be run one after
the other, the data wirelessly transmitted to a hand held control unit used by the operator, then
listed and saved to a removable memory card. This will significantly cut down on the time required
normally to collect this data, which can usually take days. All this effort will now be unnecessary with
the wireless network. With a click of a button, all the units throughout the area will compile their
data, work together to combine it over the network, then send it to the control unit. Making the MFI
testing method fast, accurate and easy.
About The Design:
This system will involve two different types of units. One is the main control unit that the
operator will use to control the system. It will include a series of buttons, an SD card slot, an antenna
and a LCD screen. Its purpose is to wake up all the nearby units whose serial numbers it has
registered in its memory, instruct each unit on its required function, receive the data from each unit,
then save it all to the SD card in a format that can be analyzed by a computer program and saved to a
data base.
The second unit is the slave unit that is able to perform two tests when the control unit asks
it to. The design of this unit will require both an analog and digital portion of the circuitry. The analog
portion will be for the data acquisition and battery management process. A significantly high voltage
will be sent across each MFI output in turn, and the resistance across each will be measured using a
voltage divider and ADC. This will be the first of the two tests that these devices must be able to
perform for each of its five channels. After it has acquired the resistance, the voltage input will be
switched off and a transient analysis will then be performed to sample the voltage drop over time.
This can then be used to analyze the capacitance of the MFI dielectric (or what’s left of it).
So what is an MFI? It may be hard to understand these test procedures if you don’t fully
understand the nature of how they’re applied. An MFI is essentially a small tungsten filament or wire
that is drilled a specific depth into the side of a fiberglass tank or pipe. Since fiberglass does not
conduct and the fluid inside of the pipe or tank does, you can usually expect resistances between the
tungsten and the liquid to be in the MΩ to GΩ range. I have yet to calculate the common capacitances,
but I’m expecting to have to use a very high input voltage in order to obtain enough resolution to see
their transients accurately. These resistances and transients change overtime as the pipes interior
wall wears down. This can give you an account of their rate of deterioration, which can then allow
you to make predictions on how much longer these piping installations can be expected to last, and
when it is a good time to replace them. This is a decision that the managers of this type of equipment
take very seriously as the cost to replacing the equipment can be in the hundreds of thousands of
dollars. The filament is never drilled completely through the pipe or tank wall. Usually about halfway
through, but maybe less if there are indications that the interior wall has been deeply permeated by
the fluid.
There will need to be a circuit built into this system that will manage the battery as well. This
unit must retain battery power for at least five years, or time that would otherwise be saved by using
this system, would instead be wasted by the user running around changing batteries. It would be
unacceptable for each unit to have their batteries changed whenever the operator needed to use
them. In fact it nearly makes the whole system obsolete. I intend on using a five to ten year
rechargeable battery that is measured and charged using a separate IC. If I run into problems with
this, if my battery life falls short of my requirements. I may be interested in applying a little solar cell
that should, over the period of a year. Keep each unit fully charged and ready to be used without any
concern for battery life. These units are not continuously used. A BRER technician will visit the plant
in which they are installed, and gather data from them usually on a 6-month to annual basis.
For the slave unit, there is also a digital aspect. The two tests that have been performed are
sampled by an ADC and from that point on, the data is in digital form and will remain so. In order for
all the devices to communicate and assemble data, I intend to create a digital wireless network that I
can hopefully use Zigbee protocol to implement. This will depend highly on the microprocessor that I
choose to use. If not Zigbee, than perhaps a less-than 1GHz carrier frequency. This network will also
have to take into account that there may be over 100 slave units testing and communicating
simultaneously. For a microcontroller to place inside this unit, I had to consider some very important
options. I need this unit to have a battery life expectancy of 5 to 10 years, otherwise BRER employees
would end up spending an unrealistic amount of time replacing batteries every time they needed to
use each unit. This was unacceptable, so I needed the smallest power consumption rating I could find.
This caused me to consider the ARM M0+ core and since I needed to use an RF transceiver as well. It
made sense that a MCU with both of these features would be the best choice. These considerations
lead me specifically to the Freescale Kinetis KW01.
Required Parts and Standards:
I am eying the Freescale Kinetis KW01 microcontroller due to its onboard RF transceiver, its
16 bit ADC and especially its M0+ ARM core. It is the only chip I can find that has both the M0+ and
transceiver onboard the same chip. Making it a very viable option. However, this chip hasn’t yet been
available for purchase from Freescale since it’s fairly knew. If it isn’t available in time for this project
I might have to consider another ARM M0+ MCU and a Zigbee transceiver that is on a separate IC.
I might have to use a separate, five channel minimum, 16 bit ADC that will communicate via
SPI. This decision isn’t final as I haven’t yet designed my analog circuit and don’t know whether an
external ADC is necessary. The on-board one my KW01 MCU has may be sufficient.
My Nokia 5110 84x48 LCD screen will communicate with the MCU via SPI.
I intend to use Zigbee protocol for my wireless network. Though depending on my chip, I
may be reduced to using a sub 1 GHz carrier signal instead. This will hold true if I confirm the KW01,
since its transceiver doesn’t support Zigbee protocol.
An DS18B20 digital thermometer using one wire communication.
Project Background and Benefits:
The MFI testing concept was originally developed and patented by BRER Technical Inc.
Originally, the testing process required that the technician walk around the whole plant, climbing
ladders, drilling and testing MFIs, writing down the data then typing it into a computer program.
With the new wireless system, all this can be done in the blink of an eye. Sure, the system will still
have to be installed, and this will be an even bigger job than before given the increased number of
MFIs per site, but it will all be a one-time event. Once installed, a single operator can conduct several
days’ worth of work in only a few seconds. He’ll never have to worry about climbing ladders or
drilling into pipes full of hazardous chemicals at that particular plant ever again, unless of course
there are technical difficulties with any of the devices. It isn’t really possible to compare this product
to any competing products because there really are none. MFI is a patented technology developed by
BRER and can’t be used by anybody else without their permission. There are other methods of
testing composites of course, but they usually rely on expensive technologies such as infrared or
ultrasound. These methods work well, but they take a lot of time to conduct and can’t be cheaply
implemented into a wireless system like MFI can because of the expensive components required.
Products Impact on Industry:
The most valuable part of the MFI test is that it’s able to be conducted on a composite system
without interfering with that systems primary function. In other words, it’s passive. You don’t need
to shut down the mill you’re working in to find out whether the piping needs repair. You don’t need
to cut into the fiberglass and take core samples to make sure the inner wall isn’t degraded too far. My
senior project takes one of the safest, most profitable composites tests and makes it easy, and
instantaneous. The potential of this product is that every mill in the United States and Canada may
want these devices installed on their equipment. With that kind of potential, my project could very
well mean a lot of big business for BRER Technical Inc. Not to mention a possible future job for
myself as well as an excellent addition to my résumé.
Project Development and Demonstration:
The manufacturing of a product this complex requires extensive planning and research. The
project will be divided into two different modules that have to communicate on the same private
wireless network. I intend to start my development process by working on the slave module first. It is
the most complex of the two and will require most of my time and effort to complete. The steps I
intend to follow for development are these:
1.
2.
3.
4.
5.
6.
Choose a microcontroller:
Program the LCD to temporarily work with this microcontroller so that it can provide
feedback to me as I attempt the next step.
Design the analog circuit for the MFI test. Start with using the MCUs onboard, 16 bit ADC. If
this doesn’t perform as required, then consider a better external ADC with 5 channels at
least.
Attach an 18B20 digital thermometer and work out its temperature conversion into
fahrenheit.
Implement a battery monitoring and charging system with possible solar panel. This may be
problematic as the RF transceiver will not be setup yet and can’t yet be measured in the
current draw.
Workout the RF transmitter parameters, but wait until the control unit is finished before
trying to set up the network.
After the slave unit is complete, the control unit has to be designed by following these steps:
1.
2.
3.
4.
Add the LCD and use for feedback.
Implement the touch sensitive buttons. They must be touch sensitive since it is in the best
interest of the product to have the container sealed in the dirty, corrosive environment it
will be used in.
Implement the RF transceiver and wireless network.
Implement SD card writer. Save this for last as it isn’t necessary for the project presentation.
It will require an SPI port on the MCU to implement which is readily available on the KW01.
If I’m unlucky, the SD may have to share this port with another device. If so I’ll have to use a
chip select.
When all of it is finished, I should have several slaves and one control unit ready to present. Once I
have one slave, it shouldn’t be difficult to make many. I intend on bringing in some composite
samples and test my units right there in class. I’ll show the data on the control unit LCD if I don’t have
time to implement the SD card.