How good was conversion? When management questioned the veracity of routine yield calculations (i.e., how much feedstock was converted to saleable product), it became clear that the end users could use some insight into the rudiments of the measurements they were using.
Years ago our shipping and logistics manager, who came from a petroleum refining background, insisted on converting a product tank’s level in percent into gallons (volume) before calculating the pounds of product that had been produced or shipped. In the fuel business, transactions are based on volume, or so seems the tradition, which is curious since the less dense/more volatile hydrocarbons and ethanol have a lower energy potential than their higher molecular weight counterparts. Still, we buy gasoline by the gallon irrespective of whether the “energy per unit volume” or miles-per-gallon can vary more than a few percent. If you live in a cold climate, you may notice your mileage is better in summer; the fuel has a lower proportion of “light ends.” The science (art?) of maximizing the cheap components in fuel is worth many millions to the average refiner, and so it’s a passion.
“I can tell you pounds per percent,” I told Mr. D. “It will be more accurate.” This was because the level was measured with differential pressure (DP) transmitters, which measure level by “weighing” the liquid head above the lower vessel or tank tap (or process connection). A foot of heavy material will consequently produce a higher DP and hence a higher apparent level than a foot of lighter stuff. Your end users—operators, engineers, accountants, managers—may be surprised to learn that the “50%” they see on a graphic or a spreadsheet may not in fact represent a height in feet and inches above some reference point with great certainty. It may, in fact, be quite a bit different.
Consider a boiling liquid, a boiler steam drum or perhaps a hydrocarbon vaporizer. The weight of the liquid head creating a differential relative to the upper tap is not only influenced by the material and the way its density changes with temperature. It’s also reduced by the vapor bubbles when the vessel is in service. So the “idle” level measurement will not equal the “hot” level, which will differ from the “boiling” level, even though a sight glass might show it’s the same. But wait, what if the liquid in the sight glass isn’t the same density as the liquid in the drum? If it’s cooler—as an external sight glass typically would be—the level in the glass will be proportionately less than the level of the less-dense, liquid-vapor mix inside.
Refiners and their ilk have been trying to improve mass and energy balances for decades, and some have invested significantly in Coriolis flowmeters. Unlike DP technology, a Coriolis flowmeter measures mass directly—innately—as the phase shift inlet-to-outlet of the oscillating tubes is directly proportional to mass flow. There is nearly zero concern about assumed density, straight run or turbulent flow. But if one’s end users are using volumetric flow (barrels per day, for example) when it’s coming from a Coriolis meter, they’re subjecting their analyses to an assumed or inferred density. Why don’t we provide not only flowing density to 3-4 decimal places, but also the native mass flow? If you have access to your Coriolis transmitters via HART or fieldbus, I’d recommend bringing in those secondary and tertiary variables, and informing your users of their availability.
Even if your process engineering end users are trying to use DP flow measurements (orifice meters) to reconcile plant productivity, you might have some helpful insights, like a DP level that can be used to validate a flow that’s in question. You know, delta-level equals flow (the derivative, if you will), and for a DP level we know it’s really measuring a change in mass.
Next time the work orders roll out to zero or calibrate instruments in the hopes of a tidier accounting of feedstocks consumed and products produced, a discussion about measurement principles and potential sources of errors and unneeded uncertainty may be in order. By using the units that are closest to the fundamental principles, unnecessary or arbitrary assumptions can be eliminated, and bona fide losses revealed.