Sigma, Sigma.. Wherefore Art Thou, Sigma? A Discussion of Freeness Calibration

It’s a given that your Drainac and your lab evaluations of freeness won’t agree. This is just the statistical nature of the beast, so to speak – and it’s true, I might add, for each and every instrument in your mill.

“Why”, you ask, and it’s a good question.

In the case of freeness, you have to remember that lab evaluations of freeness are completed using some version of the TAPPI 227 method. I say “some version” because it’s a reality of daily mill life that almost no-one actually executes the TAPPI method according to the instructions in the method. Most labs will use some kind of shortcut. I can’t blame them, really… the TAPPI 227 method has a lot of steps in it and it takes forever to finish a test if you do it the way they say you should.

So, most labs will use a version of the TAPPI method and the Drainac uses something else. We don’t utilize the TAPPI 227 method in our instrument because, well, the TAPPI 227 method really isn’t that good of a test (this is true, incidentally, for other manual methods of assessing stock freeness, like the Schopper-Riegler approach). If you’ve ever heard of the skepticism that most paper makers have for the reliability of manual freeness tests, you know why.

There are a lot of advantages if you don’t use the TAPPI method. It’s because we do things differently that we can analyze samples so quickly (30 seconds!) and why we don’t have any moving parts in our system (which makes us the easiest system to maintain) and why we are the simplest system available in the world, but I digress.

Because we don’t do TAPPI 227, the Drainac doesn’t produce a CSF number. Instead, we produce a number that reflects how fast your stock will drain. This “Drainage" number is proportional to CSF freeness and, because it is proportional, you can build a mathematical relationship between the two independent assessments of freeness. This is the “Calibration” that everyone is always talking about.

If you take your time to do a calibration properly, not only do you arrive at an estimate of the

Because the lab and the Drainac use two different methods to assess stock freeness, you can bet that there will always be some difference between the two. The statistics associated with a calibration tells you just how much deviation you can expect to see on a day to day basis and when you should get excited about that difference.

More importantly, the statistics tell you when you shouldn’t get excited at all.

What do I mean by that? Well, every calibration regression produces an estimate of the average error inherent in that calibration, based on the data that was used. This is the “Sigma” (σ) that you may have heard of, and it is a very handy number to have in hand.

Assuming you’ve built your calibration properly, the statistics say that about 95% of the time, the deviation between the Drainac and your lab should fall within plus or minus two sigma (2σ) of each other. That means that 19 out of 20 times, you shouldn’t get excited if you see a difference of up to 2σ. It also says that 5% of the time, or once out every twenty samples, you can expect the deviation to exceed 2σ, and you should still not get excited.

mathematical relationship between your lab results and the Drainac output, you also get statistics which tell you how much faith you can put into that relationship. This last point is probably the most important aspect of any calibration, but it is also, unfortunately, what most people forget to take into account when comparing their lab numbers to the instrument.
People who get excited want to change things and the statistics tell when you’re justified to start thinking about making changes and when you really should leave things well enough alone.

This is the way it usually goes: The lab runs a sample and somebody notices that the Drainac and the lab are saying different things. “Oh, boy!” says the Somebody, “We can’t have that!” and immediately calls up the overworked E&I team. “The Drainac isn’t matching the lab results”, the Somebody says. “Get out there and make it read right, would you?”.

The overworked E&I team member dutifully trudges out to the Drainac and makes an adjustment to the system to make it match the current lab result and…

He really, really, really shouldn’t have done that.

Let me say that again.

It’s premature for the E&I tech to make any adjustments to the Drainac at this time. What the E&I tech should have done is to first check the deviation of the Drainac and Lab against the calibration 2σ.

As I mentioned earlier, assuming the calibration was done properly, 19 times out of 20, the deviation will be under 2σ. If the deviation is, in fact, under 2σ, then the E&I tech is finished. Everything is working as it should and nothing should be changed at all. Even if the deviation happens to exceed 2σ, the calibration statistics tell you that should expect to see that about once out of twenty times.

Really Important Rule #1

Most of the time, you should wait to make a change only when you see that the deviations consistently exceed 2σ.

Really Important Rule #2

When deviations do exceed 2σ, don’t assume that the Drainac is the problem.

Remember when I said earlier that the TAPPI 227 isn’t a great test? Well, it isn’t. The repeatability of that test stinks, and that’s when you do it according to the published method. It stinks even worse when you take shortcuts and frankly, most people do.

Just how bad is it for your case? That depends on what your lab people do. I always recommend that you run a Total Error of Variance test (TEV) to determine just how much faith to put into their lab results. The TEV is a sort of a poor man’s Six Sigma evaluation. It gives you an idea of just how much variability you can expect from your lab on any given day.

So, let’s say that you’ve done all that. You have a calibration with a defined 2σ. You’ve run a TEV on your lab and you know what to expect from them. Now, you’ve got a series of numbers that tell you that something is going on and you’ve concluded that the Drainac is the problem. It’s time to make some adjustments to the calibration, right?

Nope.

It’s time to check the Drainac. Clean it. Inspect it. Look for problems. Check the various operating pressures. Do all of the things that you need to do to make sure that the Drainac is functioning properly.

Let’s say that you’ve done all of that, too, and that there is still an unacceptable deviation between the Drainac and the lab. Probably the best thing to do at this time is to rerun the calibration, but if you don’t have the time to do that at the moment, you can simply shift the output of the Drainac by adjusting the BIAS of the Compensation Variable.

The Compensation Variable (COMP VAR) is a general input that you can use to apply a correction to the Drainac for some external process parameter. Most people never use this feature of the Drainac, but the BIAS component of the compensation variable is a handy way to add or subtract a number from the freeness output of the Drainac.

To adjust the bias, enter configuration mode and scroll to the COMP VAR parameter. Once there, scroll through the various attributes to get to the BIAS screen. Enter your desired offset value, either positive or negative, in the same units that the system is configured to output.

Your Drainac system will now adjust its output. Keep in mind that this is a stopgap only, you should plan to rerun your calibration if your deviations persist above 2σ.

Also, please remember that TECO is a resource for you. We are always happy to regress your data for you and tell you what your 2σ is for your calibration. We can also send you a spreadsheet which you can use to run a TEV on your lab.

Expanding the Sweet Spot: Measuring Pulp & Paper Stock Consistency Properly

If you want to measure consistency properly, it’s important to remember that all consistency transmitters have their so-called “application window”.  The application window consists of all those process parameters that have to be within a certain range in order for a particular transmitter to work properly. Which ones are relevant to you depend on the technology behind the consistency transmitter in question.

As long as you’re within a particular range for each of these parameters, you have a good chance that the instrument is reporting stock consistency reasonably well. Of course, if you get outside of that range – and unfortunately, it’s not always obvious that you have exceeded the limits – then the transmitter output can start to deviate from reality, sometimes in a really big way.

So, it pays to pay attention to the application window for an instrument - the “Sweet Spot” - if you hope to get the most out of your measurements. 

When it comes to mechanical transmitters, the process conditions you need to consider include production flow rate, furnish types and, oddly enough, stock consistency itself. 

We’ll start this discussion by asking the question:  Why is production flow rate important? 

Simple passive mechanical transmitters like blades respond to the “apparent viscosity” of the process.  Apparent viscosity is just a fancy way of referring to how thick the process slurry is.  As you would expect, the higher the consistency, the “thicker” the process is. 

Blades, however, don’t really measure the thickness of the stock directly.  Instead, they respond to changes in force as stock moves past the blade (that’s why, incidentally, they are called shear force systems).  The stock imparts a force to the blade as it moves – or shears – across the blade surface.  Stock motion, however, is the key point – the stock has to be moving past the blade.  A blade transmitter immersed in stationary stock would register zero, irrespective of what the consistency is. 

What isn’t always obvious, however, is that the force that the blade is responding to isn’t merely a function of consistency.  It has a flow-rate component to it as well.  As the flow-rate goes up, the force imparted to the blade will also go up.  This is perhaps one of the most important aspects of blade systems and it is also, one of the things that is most often overlooked by mills.  Simply put,

Blade Force =  Consistency Force + Flow-Rate Force

So how do you deal with the flow rate component?  Some manufacturers will publish flow velocity-consistency graphs for their designs.  The implication here is that if your process stays within the valid region as defined by the manufacturer, the measured force will be consistent with changes in consistency.  This is a reasonable approach if flow-rate variability is kept to a minimum,  but it is not suitable for applications with highly variable flow regimes.  Under these circumstances, you must compensate for variable flow-rates if you hope to get a useful consistency measurement.

That said, there are two ways you can compensate for highly variable flow rates.

You can measure the flow rate and mathematically subtract out the flow rate component from the force signal and/or you can select a sensor geometry which has a flow rate response which minimizes the impact of flow rate for your application.

When it comes to TECO’s StockRite® line of consistency transmitters, you can get both.

TECO’s C6000 consistency transmitters are shipped with automatic flow-rate compensation built-in.  All you need to do is to land a flow-rate signal on the transmitter and the flow-rate component is automatically removed from the consistency signal in real time.  You can drop our C9700 blade into your existing blade application – our systems fit our competitors process connections, by the way – and automatically compensate for flow rates which vary from 0.5 to 12.0 fps.

That’s what I call expanding the sweet spot.

Of course, wouldn’t it be nice if you had a sensor design which was immune to variability in flow rate in the first place? I’m happy to say that there is one available:  Our C3000 Probe design has a flat flow-rate response for production flow rates up to 3.0 fps.  That means that the C3000 has a zero flow-rate component for all flow rates below 3.0 fps.  Put another way, you could have production rates of over 1000 GPM in a 12” line and never have to worry about flow rates disrupting your consistency signals ever again.

If you’re having trouble with your consistency measurements, give us a call.  We’ll really good at helping our customers get the most out of their consistency measurements.

Pulp & Paper Stock Consistency Transmitters True Cost of Ownership

One of the objections I hear regarding passive mechanical consistency transmitters is the high cost of ownership that these systems purportedly have.

The thinking goes something like this.  Mechanical transmitters typically have a sensor in the line that protrudes into the flow in the line.  Sooner or later, that sensor will get hit and damaged and will need to be replaced.  Thus, to keep a a mechanical sensor operational requires that the sensor be replaced and represents an ongoing expense.  The alternative, a rotary transmitter is typically installed such that its sensor is wholly contained within a stilling chamber and is thus not likely to be hit and damaged.  Its cost of operation must be lower, right?

While there is some truth to this, it’s not the whole story – not by a long shot.
It’s true that passive mechanical systems do get hit from time to time and their sensors will need to be replaced.  It’s also true that rotary systems don’t often get damaged because their sensors are offset from the flow.   That said, what is not true is the notion that the cost of ownership for a rotary is far less than that for a mechanical. It isn’t.

Let me illustrate this with an example using my company’s C3000 sensor:

The TECO C3000 Consistency Sensor

A rotary system will cost you somewhere in the neighborhood of $30,000. Let’s assume that it will last five years before it will need to be replaced.  A complete TECO C3000 mechanical system, on the other hand,  will typically cost you somewhere under $7k. Let’s say you have to replace the C3000 sensor once per year.  Your annual cost, including the trade-in credit for the original sensor core, is under $2k per sensor.

Over five years you’ll pay less than half of what you’d paid for the rotary initially.  Let me say that again – you’d pay less than half of what you’d pay for the rotary.

Don’t get me wrong, rotary consistency transmitters are cool devices and they certainly have their positive points, but they ain’t cheap.  Passive mechanicals are way, way less expensive and you can use them to measure mostly the same consistency range that you would use a rotary to measure.  Properly applied, the TECO C3000 sensor will give you way more bang for the buck than any other system available in the world today. 

Sampling v2.0

I want to take another look at proper sampling because it so key to a good calibration.   While there are statistical tricks to get the most out of anything you produce calibration-wise, if you don't have good sampling, you are, in the best case, creating big problems for yourself.

We want to collect samples such that they are representative of the process.  Samples that are representative have an average that is very close to the average of the whole process at that moment in time. Samples that are not representative will have averages that are not at all similar to the process. 

Collecting representative samples isn't difficult, but you do have to follow certain rules. 

1) Collect samples from lines where the flow characteristic is known to be stable, i.e., in plug flow.  Stable flow means that you will likely not have any turbulence in the line that might de-water your stock or otherwise introduce non-representative sampling.   The easiest way to ensure this is to find a straight length of pipe that is at least seven pipe diameters long, and without any bends or obstructions in it.  

2) Make sure the pipe is full.  No, really, make sure the pipe is full.  Choose lines that are horizontal, or vertical lines with flow going up.  Choosing a vertical lines with flow going down is asking for trouble.  Do not take samples from chests if you can avoid it.

3)  If you are planning to use your data to build a calibration for an instrument, you should make sure that the sample port is close to the instrument in question.  There is no point in running analyses if the instrument is in another line or on the other side of the mill.

4) The sample port should have an internal extension that protrudes roughly to the center of the stock line.Use proper sampling valves, if you can, and avoid ball valves that have been installed on the side of a pipe.  The image below illustrates how variable things can get as they move through your stock line.  As you can see, it can sometimes be a challenge to get that "representative" sample.  That said, your best chance will be to take samples from the center of the pipe as opposed to the sides.

Variability in a stock line

5) Open the valve and let the stock run freely for a few seconds to ensure that all the stock from the last sample is fully discharged from the sampling line.

6) Collect a large quantity of stock (i.e, a gallon or two at minimum - five gallons is better).

7)  When back in the lab, agitate your large volume of stock and take at least two small samples.  Analyze each according to your favorite method and average the results.  This will yield you one data point.

6) If you haven't done so before, run a Total Error Variance (TEV) to estimate the quality of your sampling and analytical technique.  TEV's are sort of a poor man's six sigma.  They will provide you an estimate of how much variability in your analyses is attributable to your sampling and how much is due to your technique.

If you don't have a TEV in hand, send me an email and I'll send you a copy of our spreadsheet that you can use.

The 64 Dollar Question

So, how accurate can a transmitter be, anyway?

This is a question I frequently hear from both my customers and my prospects.  While I understand why my customers ask this question, the real question they should be asking is, “How repeatable is your transmitter?”.

What’s the difference?  Glad you asked.

Repeatability refers to how closely something – an instrument, for example  – will reproduce a measurement given the same test conditions.

Accuracy, on the other hand, refers to how well that same something measures up to a different assessment of the same thing.  When it comes to consistency measurements, accuracy typically refers to how well a particular transmitter measures up to a lab assessment of the same stock. 

The lab assessment could be anything, but it is usually some variant of the TAPPI 240 method and this is where the problem comes from.  The TAPPI 240 method specifies a repeatability of 10% for that test, which means that 95% of the time, the lab test, if executed as specified, will yield results within 10% of each other. So, for a nominal test of 4.0% consistency, a second, properly executed test of the same stock sample should yield a number between 3.6 and 4.4%.  Of course, the repeatability statement also says that 5% of the time , or once out of twenty tests, you could get a number that’s worse than that 10% limit.

What makes this really scary is that very few laboratories actually execute the TAPPI 240 test as described in the procedure.  Many labs take short cuts – I once saw one guy try to squeeze dry a sample by stepping on it, for example - which means that the repeatability for the manual lab test may actually not be as good as 10%.

That’s the reason why we manufacturers prefer to talk in terms of repeatability rather than accuracy.  While we can never be sure how accurate our transmitters will be relative to the procedures your lab uses, we can be very certain about how our transmitters will respond, given the same stock conditions.  In the case of the TECO StockRite series of consistency transmitters, that repeatability is 0.0025 of the full scale range.  So, if your transmitter is set to read from 2% to 6%, for example, the TMC6000 system will repeat to within no greater than 0.01% (-4.0 * 0.0025).

Which is pretty doggone repeatable, if you ask me. 

So the correct answer to the question “How accurate is your transmitter?” is that we are highly repeatable.

The C5000 Sensor - Our Workhorse System

You may have heard me mention a thing or two about our retractable consistency sensor, the C5000. 

I know, I go on and on about this, but this sensor is really a big deal for our customers, and if you’ve haven’t tried one yet, it could probably be a big deal for you , too.

Why is it such a big deal? 

It’s retractable.  This means you can pull the sensor body out of the line at any time.  You don’t have to wait for a shutdown or isolate the line in order to service the system.

Because we made it to be retractable, we also made it easy to retract.  Here’s a short video I posted on YouTube on how to do it.  As you can see, you can extract a sensor in less than a minute.

Because it is retractable, we also made it hot-swappable.  That means you can swap out a sensor with another one without having to go through a recalibration.  Just auto-zero the new sensor and you can use your existing calibration without further adjustments.  All TECO sensors, by the way, have this “auto-zero” feature.  You can swap any of them at any time without having to recalibrate.  The nice thing about the C5000 is that you can do it without having to wait for a shutdown.

The C5000 also has what I like to call a superior flow characteristic.  What do I mean by that?  Well, you’ve heard me before mention that all consistency sensors – the mechanical ones, anyway -  are sensitive to changes in flow rate.  This means that shifts in flow rate will be perceived by the sensor as a shift in consistency.  If you’re not careful with your consistency setup, you might find yourself going in circles chasing what looks like a consistency problem but is really a flow rate problem.   

The C5000 avoids this problem altogether because it is immune to shifts in flow rate below 3.0 fps.  Put another way, there is no measurable impact on the consistency measurement for flows below 3.0 fps.  You can go from nothing to 3.0 fps and back again and it won’t have any impact on the consistency signal.  

 Of course, once you get above 3.0 fps, it’s a different story.  The C5000 will start to react to changes in flow rate.  Because we know that, we have provided on-board velocity compensation for all of our sensors in our transmitter.  Just land a flow signal on the transmitter and you can automatically compensate for variable flow rates up to 6.0 fps (up to 11 fps with our C9700 sensor body).

We’re low cost.  That means we won’t charge you an arm and a leg to buy one of our systems.  Also, when it comes time to swap sensor bodies, you don’t have to buy another whole system.  You can just buy the sensor body part.  We’ll even give you a credit when you send us the old sensor back (we do that with all of our consistency sensors, by the way, not just the C5000).

Finally, we’re local.  We fabricate and assemble everything in our factory in New Orleans.  We keep plenty of spare parts on the shelves, so we can ship you the things you need overnight.

If you haven't tried one of our C5000 sensors, please give me a call.  I'd be happy to talk to you about it.