EXECUTIVE VIEWS The Quest for the Perfect Flowmeter

Gases & Instrumentation had the oppor- tunity to speak with Dr. John G. Olin, Founder and Chairman of the Board of Sierra Instruments about the evolution of thermal mass flow meters and his latest product in this technology.

Gases & Instrumentation: You amount of heat carried away dispersion, and the other is have been engaged in ther- by the molecules provides a the capillary-tube type. The mal mass flow technology for measurement of the mass flow one that I have been refer- a number of years. You must rate. I was using that technol- ring to so far is the thermal- have seen a lot of change. ogy back around 1964 to 1966. dispersion type where the Dr. John G. Olin: Yes, there have And then I joined a company heated element is inserted in been tremendous changes in called TSI, Inc., in Minnesota the flow, and the amount of using thermal technology to that manufactured thermal heat loss is a measure of the measure flow. I started out 50 anemometers. But I was inter- mass flow rate. The capillary- years ago, while I was working ested in getting into the indus- tube is the same basic idea, on my Ph.D. thesis at Stan- trial marketplace because it but the flow goes through a ford, where I majored in fluid was much bigger. That’s when small capillary tube, maybe mechanics and heat transfer. I started Sierra Instruments, Inc. with an internal diameter of I was working with hot-wire That was in 1973. We built an ten thousandths of an inch. It anemometers. The principles industrial sensor and placed has two identical heated plati- involved then are the same the heated element inside a num RTD wires wound around basic ones that we still use. In small diameter stainless steel its outside diameter; the heat thermal anemometers a resis- tube. The tubular sheath is very is again carried away by the tance temperature detector strong. It can be inserted into flow; and again it measures (RTD) element immersed in any pipe of flowing gas and not mass flow rate. However, since the flow streamis self-heated be broken. That was a big big it is just a small capillary tube, with an electrical current to a breakthrough. It occurred in to handle larger flow rates you fixed temperature above the the 1970s. have to incorporate a bypass where most of the flow goes temperature of the gas. The G&I: I understand there are two molecules of air or another through the bypass and only basic types of thermal mass a small fraction goes through gas get heated as they pass flow meters. through the boundary layer the capillary tube. around the heated element. JGO: Correct. One is an immers- G&I: How did you arrive at your cur- Since it is the molecules of ible thermal mass flow meter, rent products, the QuadraTh- the gas that bear its mass, the sometimes termed thermal erm 640i and the 780i? www.gasesmag.com January/February 2013 9 EXECUTIVE VIEWS

JGO: To start from the beginning, the JGO: Well, this is where we tackle the too much and the pressure doesn’t traditional industrial thermal dis- second problem with thermal, stem vary too much, you are fine, and persion flow meter has two sensors conduction. The main difference that’s where you get thermal mass immersed in the flow. Each one is between the QuadraTherm and tra- instruments with 1% to 5 % of full encased in a stainless steel sheath. ditional thermal mass flow meters is scale accuracy. One measures the gas tempera- that instead of having two tempera- G&I: There is also a lot of thermodynamic ture, usually just a thin-film RTD. ture sensors—one a temperature math going on, correct? The other is a velocity sensor. They sensor in the temperature probe and are usually located side-by-side the other self-heated temperature JGO: Yes, and that happens with our such that they do not interfere sensor in the velocity probe—there iTherm algorithm set found in the with each other aerodynamically. is an additional sensor in each, giving firmware. iTherm is the fourth pillar The velocity sensor has a heated a total of 4 temperature sensors. The of our technology. The first pillar RTD element in it. extra temperature sensor in the stem was the DrySense stable sensor. The of each probe measures the stem second is the Aeromax aerodynamic G&I: What are the drawbacks to this tra- conduction in each. And, if you don’t design for the sensor head. And the ditional design? account for that you are out of luck. third is QuadraTherm-- the physical layer with the 4 temperature sensors. JGO: It suffers from two problems. These First, stem conduction is a goodly iTherm solves the first law of ther- are skin resistance and stem conduc- function of the total heat transfer modynamics for thermal technol- tion. The skin-resistance problem has budget and must be accounted for. ogy very quickly. So, every fraction to do with the fact that the typical Second, stem conduction depends of a second you are getting updated construction of the velocity sensor on the ambient temperature outside values of the mass flow rate. It calcu- uses takes a wire-wound platinum of the pipe. For example, the outside lates the mass flow rate accurately RTD that is shoved down into the tip temperature gets cold at night and even when the outside temperature of tubular sheath and is surrounded heats up when the sun comes out. changes plus or minus 50 degrees by ceramic cement or other potting The amount of that stem conduction C. And the pressure of the gas can compounds. Over a period of time, varies all over the place. change, and all of these changes when there is temperature cycling, G&I: So QuadraTherm solves the stem- are taken into account through the cracks can develop in the cement, conduction problem by measuring it? math. For me, it is like a result of a and that causes a big change in the lifetime of work that I have done, heat transfer characteristics of the JGO: Exactly. With the total of 4 sensors, especially the last 5 years. velocity sensor, and you generate two temperature sensors in the tem- large errors over time. perature probe, and a heated veloc- G&I: Why is an accuracy of 0.5% so impor- ity sensor with additional tempera- tant? That’s just a half percent lower G&I: How did you overcome this problem? ture sensor in the velocity probe, we than some existing models. JGO: The solution to the skin-resistance are able to solve that problem. When JGO: Basically, the accuracy of the 640i problem is DrySense. The DrySense you have the two additional sensors QuadraTherm is a half percent of velocity sensor technology uses a you can measure the heat flux going reading, that is, a half percent of proprietary swaging process that down the stem. It more or less acts reading when you are 50% to 100% eliminates all air gaps between the like a heat flux-gauge. of full scale, which is the range where most users operate. Note that heated velocity sensor and the tubu- G&I: With other products, how do they this is “per cent of reading”, a much lar probe sheath eliminating the account for stem conduction? need for any potting compounds. tougher accuracy than traditional There is no cement at all. This is why JGO: They don’t and we didn’t but we meters with their “per cent of full we call it “dry.” No so-called “wet” do now. What it amounts to, is that scale” spec. That half per cent is very potting compounds are used. We in the field application where the important. (For more details on this, never have any drift in that sensor. thermal dispersion mass flow meter see G&I December 2012 eNewsletter As a result, it is extremely stable and is located, if the field conditions of at www.gasesmag.com). One of the why we issue a lifetime warranty. temperature and pressure and out- big applications is custody transfer. side temperature are exactly equal A lot of custody transfer happens G&I: But QuadraTherm has other major to those for which the sensor was in the semiconductor industry. We differences from the traditional con- calibrated in the lab, then it is OK. have an ultra high purity version figuration. So, if the temperature doesn’t vary of QuadraTherm used where gas

10 January/February 2013 Gases&Instrumentation EXECUTIVE VIEWS

supply companies have tanks of gas outside fab areas delivering the big 5

gases—N2, O2, H2, argon and helium. What the fab pays the gas supply company depends on the accuracy of the flow measurement. So, if the reading is high, the difference between 1% and 5% can amount to thousands of dollars over a period of months, or even weeks. G&T: Are there other reasons? JGO: Another, more abstract reason, is that the entire flow meter manufac- turing industry is pushing for higher accuracy. It’s just the world that we live in. Just as there are more and more apps for your iPhone, there’s better and better accuracy for your flow meters. Sierra needs to be part of that. The other big driver is natu- ral gas with the tremendous discov- eries and resources that America has. This means we are going to be using a lot more natural gas. Thermal flow meters are perfect for that. They are perfect for measur- ing the custody transfer at the point where a distribution line splits off from a major natural gas pipe line. Then there is the quality issue. High accuracy is really important in the chemical processing, food process- ing, and numerous other industries where the products’ quality gets better because the measurement of the ingredients that are going into that product are measured more accurately, G&I: So it took a lot of years to arrive at the QuadraTherm. JGO: For me this is my most satisfying accomplishment after working for so many years. The quest for the perfect sensor has always been my goal. The previous traditional sensor was not perfect, but it did satisfy an awful lot of need. The QuadraTherm, though, is “near” perfect, and satisfies a lot more. For more information go to: http:// www.sierra instruments.com/products/ quadratherm.html . www.gasesmag.com January/February 2013 11