1 Developing Item Multiple Scales with Summing Indicator S&T Items 3100-1, 3200-5, and 3600-2Submitted by Ross Andersen, NY (retired) NCWM Certification Coordinator, and Participant in 1990 decision Revised Sept 8, 2017 One issue but three agenda items for three code changes, one to the General code, one to the Scales Code and one to the Fundamental Considerations. I encourage you to look at the supporting documents for the Vmin issue in the 2016 NCWM meeting archives, comments on this proposal, and also the text of S&T Item from 1990. A little personal anecdote to begin. Early in my career, I chatted with a county official who bragged his gas pumps were the most accurate in the state. His explanation, he held all pumps in his jurisdiction to acceptance tolerance at all times, not just in the first 30 days. He claimed they did not need the extra tolerance, so he didn’t allow it. Keep that in the back of your mind as we proceed. View as a slide show to hear narration.

2 Purpose of PresentationQuestion whether the 1990 NCWM decision (S&T item ) regarding multiple scale arrays is supported by HB44. Revisit important points made in SMA presentation on Vmin. Apply them more broadly to other parts of 1990 decision. Explain proposals necessary to clarify the proper application of the code. In 1990 the NCWM S&T Committee made an interpretation of HB44 (item ) as it applies to the multiple scale array. It has become clear to me in recent years that the decision in 1990 was flawed. Also know that I was there in 1990 and supported the S&T position. The key question that I will ask is, where is that stated in the code? Particularly, this became an issue when addressing questions for the professional certification exam and in considering the Vmin issue that the S&T addressed in 2015/16. I will draw heavily on the points made by the SMA and supported by vote of the NCWM. I will present some revisions to my initial proposals, and address how this will impact the scales presently in the marketplace. All will be supported by code citations. I’m going to hurry through the narration of the presentation. I encourage you to go back and study my notes at a slower pace without the narration.

3 In 1989 NTEP Weighing Sector reached impasse on the question: Multiple Scale Array with Summing Indicator In 1989 NTEP Weighing Sector reached impasse on the question: Is this system three separate scales, or is this system three separate scales and one combined scale at the same time? The most common form of the Multiple Scale Array with Summing Indicator is comprised of three independent scales. Each of these scales has a load receiver, A to D converter, and primary weight display. The three indicators are located in a single enclosure which also includes a fourth display of the sum of the three primary weight displays. The NTEP Weighing Sector addressed this in 1989 attempting to decide how to evaluate such a system for NTEP Certification. They reached impasse. Essentially they framed the question on the slide. The record shows that most scale manufacturers favored the first option and the public sector favored the second. Unable to find consensus, the Sector turned to the S&T Committee for resolution.

4 S&T Answer to the QuestionFrom Report of 1990 NCWM (S&T Item p 157) - underline added The Committee is aware that a number of jurisdictions test each scale as an independent scale, not as a single system. The significance of this interpretation is that not only must each independent weighing device meet the requirements of Handbook 44, but the entire weighing system must meet all requirements that would apply if the device were a single scale. The S&T position supported option two. Remember that only W&M officials (public sector) vote at NCWM. The significance of the decision is shown in the underlined text. The system is three scales and one scale at the same time. We’ve been using this ruling for 27 years. So what is wrong with this decision?

5 Problem 1: Lack of Legal CitationsFew legal citations and no parsing of text for meaning Editor’s Note at end of cites lack of Code references (adds a list) but doesn’t address lack of text analysis. Where is code citation for 3 & 1 at same time? When preparing Large Capacity certification exam SME’s could not cite code to justify answer The S&T item fails to cite the legal basis for its decision. The item is a legal opinion, but lacks references to specific code sections. In addition, a legal opinion usually examines the text of applicable sections to show the specific meaning of the code as it applies to the question being asked. I’m not the only one who notices this. The NCWM Executive Secretary places an editor’s note at the end of the item in the Report identifying this weakness in the item. The editor also supplies a list of code sections that he felt applied. However, the Editor’s list is not technically part of the S&T report. Both sides in the debate agreed the individual scales in the array must meet the applicable code requirements. The critical difference between the two options identified by the Sector is the combined scale, what I will call the “and one” scale. Specifically, where is the code that requires the three devices used in combination to comply with the code as a single device? In preparing the large capacity scale exam, the group of subject matter experts working with me could not cite code to support the S&T interpretation. We do not include a question on the exam unless we can support the correct answer with a specific code reference.

6 Problem 2: 1990 Decision is Arbitrary and Capricious1990 decision states: “and 1” device is subject to all applicable requirements? 1990 S&T Item – Why make a list of 5 criteria applied to these systems? Is this cherry picking? 6th criteria (Vmin) added 1993, applied across all three scales (Revised not to apply to summed indication) see SMA presentation Problem 2 is that the 1990 decision is arbitrary and capricious, one of the common reasons that regulatory actions get overturned by the courts. The standard for arbitrary and capricious is action taken without reasonable grounds or adequate consideration of the circumstances. The 1990 decision says the “and one” scale must comply with all requirements of the class as if it were one scale. Why then does the S&T report create a list of 5 criteria applied to the summed display. You make a list if you are cherry-picking only those sections that you want to include, while you ignore other sections. As further evidence, we see another item gets added to the list in 1993 when Vmin is initially added to the Scales Code. In applying the Vmin formula to the “and one” scale, the number of load cells included the combined three platforms. However, in , SMA questions this application of the code and brings a proposal forward. NCWM agrees and takes Vmin out of the list in 2016.

7 Arbitrary & CapriciousS&T was asked (both Steve Langford and I) to revisit entire 1990 decision in 2016, but limited its action to Vmin Why is this combination of measurements singled out for special treatment as a single device? Weigh-in-Weigh-out combines multiple measurements from multiple devices – one net weight Three scale array w/no summing indicator ABWS sums multiple measurements same device Cherry picking is arbitrary and capricious as it does not consider all relevant requirements. Steve Langford of Cardinal Scale and I both commented to the S&T while considering Vmin that they should look at the 1990 decision in its entirety. They chose to only address the Vmin proposal and leave the other requirements for a later day. The Committee suggested we create a new proposal. This proposal continues the item to its natural conclusion. The Committee’s decision in 1990 is also capricious in that is treats one combination of multiple measurements as different from others. Why? We see many cases in commerce where multiple measurements are combined, but the code requirements are not applied to the sum in these other cases, only to the individual measurements. Also, the multiple devices used in combination are not combined into a single device for compliance purposes with specifications or user requirements. Consider the following examples. Weigh in weigh out systems use measurements from two scales in combination but they are not considered one scale for code application (specs or tolerances). They don’t have to meet tighter tolerances. What about a three scale array with no summed indicator. The system does the exactly same thing but the sum is manually calculated. ABWS scales are used in transactions that very often exceed the Nmax of the accuracy class. Shouldn’t use of these scales be limited to Nmax of the class as the 1990 decision tried to do to the multiple scale array?

8 2016 SMA presentation showed summed indication is not a scale(Slide 10) “Total Weight Display Sum of individual load receiver readings. It is NOT a fourth scale. It has no measuring function in the calculation of the total weight. It is simply the mathematical summation and nothing else. Because there is no measurement function, there is no scale and therefore not subject to Vmin requirements.” More on the arbitrary and capricious claim. The 1990 decision says the “and one” scale must comply with “all requirements that would apply if the device were a single scale.” Yet the SMA has clearly shown that what the S&T report in 1990 considers to be a scale, is not a scale at all. (Review the slide points) By passage of the 2016 item on Vmin the NCWM agreed with the SMA. If the combination is a scale, it must comply with the Vmin requirement. Since summed indication isn’t a scale, Vmin is not applicable. (Find this stated in General Code G-A.1.) But take note, you cannot say the combination is a still scale, but exempt it from Vmin. If you understand what Vmin is, you know that’s not how load cell based scales work! A scale can’t comply with performance requirements if it doesn’t meet Vmin requirements.

9 Further Evidence the Summed Indication is not a ScaleSummed indication can’t always comply if test loads are distributed over all three platforms! That’s the only fair way to evaluate the combined scale. S – requires AZT not re-zero a load more than 3 divisions on vehicle scales At zero add 50 lb to each platform all at once All current scales fail test – 150 lb (7.5 d) tracked off T.N.7.2. – Discrimination Test, must see 2 d change for 1.4 d added from just below ZU Load scales and get one scale just below the ZU add 9.33 lb to each scale all at once Test results 1 d ~30% , 2 d ~50%, and 3 d ~20 % All Scales fail test ~50% of the time I will address two additional code sections where the combined scale will not comply. To test the individual scales for AZT and discrimination, you will apply the loads only to the individual scale. To test the combined scale, the “and one scale,” you must distribute the test loads over the combined scale platform, essentially treating it as one scale per the S&T interpretation. AZT is tested by starting at stable zero condition and adding a load that exceeds the AZT limit. I will work with an assistant and we will apply one 50 lb weight to each of the three platforms at one time. That’s 150 lb or 7.5 d applied all at once. I believe all scales, as presently configured, will track off the distributed 150 lb, failing this test. Discrimination is tested by applying a 1.4 d load from a point just below the zone of uncertainty. Begin at maximum test load with all three platforms loaded. Now add a 2 lb wt to one scale and by coincidence get a 1 d increase. Now remove the 2 lb weight. You are just below the zone of uncertainty. Now apply the 1.4 d/28 lb load distributed (9.33 lb per scale) and results will be either 1 d, 2 d, or 3 d change. Computer simulations of this scenario show 50% of the tests will produce a 2 d change, 30% will only get 1 d change and 20% will get 3 d change. 1 d definitely fails and 3 d should also fail the test. There may in fact be more sections where the combined scale fails, as I did not fully consider the sections dealing with motion detection for semi-automatic zero or printing.

10 Why Sum Can’t Pass DiscriminationWhy can’t the combined scale pass? Consider the graphic. The vertical hash marks on the lines for scales 1-3 represent the ZU for each of the individual scales. They are uniformly spaced 20 lb apart for each of the individual scales in the array, producing scale divisions with uniform 20 lb increments. The sum however, has to resolve three pairs of ZU’s such that scale increments cannot always be uniform. Start the test at blue line and addition of 9.33 lb x 3 yields 60 lb/3 d change. Start the test at red line and addition of 9.33 lb x 3 yields 20 lb/1 d change. With each new load applied to the array, the alignment of the ZU’s for the three independent scales will vary such that the sum can only pass the discrimination test about 50% of the time. 50% is non-compliant! Scale divisions of the “and one” scale are not uniform in size because there are 6 ZU’s to resolve. They align differently with every loading of the array.

11 Can comply if Total is a fourth scaleSMA concern that proposal prohibits a 4th scale. Not true. If the total uses internal counts and creates a unique measurement, it can comply with those sections (and all others). Then it would be a 4th scale and not 3 & 1 at the same time. A 4th scale means there will be times when A + B + C  Total (3 scales ~33% of time, 4 scales ~41 % and some differences reach 2 d) But, is it desirable? Users will not like this! Also, how do we know if it’s a fourth scale or a sum? SMA expressed concern that the developing proposal prohibits them from making a 4th scale. I have assured them I believe it doesn’t. My proposal reverses the 1990 decision and applies the code only to scales that measure something. The summed indication, as the SMA and I have shown, is not actually a scale. If the 4th scale is truly a “measuring device” then it must conform. If manufacturer sums the internal counts of the three scales, they can create a separate 4th scale indication. That 4th scale should comply fully with AZT and discrimination requirements (and others) where the summed indicator can’t. However, the 4th scale will sometimes result in a total not equaling the sum of the three independent scales. From random simulations: Scales % agree %1d diff %2 d diff This may not be a desirable solution for the users. I believe they want the sum to match up with the individual scales. Also how will the inspector determine whether the 4th indication is a sum or a true scale? We’ll come back to this.

12 Opposing View and ResponseH. Oppermann, Comments in Opposition to S&T Item – ( ) underline added Quoting the DES checklist paragraph “Sum only indication. The summed display shall be evaluated as an individual scale and must meet appropriate requirements. The indicator may provide a display for each load-receiving element, but the only display that will be considered “legal for trade” will be the summed display. In this case, the total number of divisions for the system shall not exceed for Class III and IIIL.” This is clearly based on the 1990 S&T decision. What HB44 code section would they site to support this checklist item? There is also opposition that came out in the Vmin discussions in 2016 and in response to this proposal from Henry Oppermann. The NTEP checklist section he quotes assigns legal for trade status only to the summed display. This is clearly following the 1990 S&T decision. However, what code section supports this decision? Why are the individual scale readings NOT considered legal for trade? Didn’t both sides at the Sector, and the NCWM in the 1990 report, agree that the individual scales had to fully comply? On its face the decision to only accept the sum is arbitrary, as it say’s a compliant scale is not legal for trade! In addition, the combined device, the “and one” scale, is considered legal for trade. However, it does not conform to some code requirements, and it is not really a scale as both SMA and I exposed in previous slides. These are the effects of making up your own rules.

13 Response (based on Code)What NTEP seems to exclude is expressly permitted in code UR Single‑Draft Vehicle Weighing. – A vehicle or a coupled-vehicle combination shall be commercially weighed on a vehicle scale only as a single draft. That is, the total weight of such a vehicle or combination shall not be determined by adding together the results obtained by separately and not simultaneously weighing each end of such vehicle or individual elements of such coupled combination. However, the weight of:  (a) a coupled combination may be determined by uncoupling the various elements (tractor, semitrailer, trailer), weighing each unit separately as a single draft, and adding together the results; or  (b) a vehicle or coupled‑vehicle combination may be determined by adding together the weights obtained while all individual elements are resting simultaneously on more than one scale platform. (underline added) Cannot be construed to be a single measurement. (UR.3.3. is pre 1986 and not in Editor’s list) In SMA Vmin presentation. I offer a response by citing UR.3.3. as a code section that clearly contradicts the Sector position. This is also a key point made by SMA in their Vmin presentation (slides 12 & 13). Critically examine the wording of UR Part 1 is a “shall” requirement to measure a vehicle as a single-draft (i.e. a single measurement), and not two non-simultaneous measurements. Then we encounter the critical word “However.” This introduces two “may” statements, which are legal alternatives to the shall requirement. Both of these recognize the acceptable use of multiple measurements (or drafts) to determine the weight of a vehicle. In part (a) the coupled vehicle combination is uncoupled and weighed in multiple drafts. In part (b), the code permits the combining of multiple simultaneous measurements to arrive at the weight of the truck. That is the code says “more than one scale platform,” not more than one load receiver. Nothing in section (b) can imply the combination is to be considered one device and one draft. The code, in UR.3.3.(b), considers them a combining of multiple, independent, simultaneous, legal measurements. Nothing in either (a) or (b) can imply that the individual scales use to produce the sum incur some additional accuracy requirements, as the OWM has suggested in their comments. To do so is to make up rules.

14 Why did S&T Act in 1990? “The number of vehicle scales consisting of three or more individual weighing elements used simultaneously to obtain a gross weight for commercial transactions is increasing.” “However, a request has been received to permit four 20,000 lb x 5 lb scales to be used in combination to provide a scale of 80,000 lb x 5 lb, essentially to circumvent the Class IIIL limit on the number of scale divisions.” From Report of 1990 NCWM (S&T Item p 157) - underline added Also consider the only specific requirement mentioned in the NTEP checklist, Nmax. (S&T Protecting hard-fought, pre-1986 battle to put 10,000 d limit on Class IIIL) All of this legal analysis begs the question, “Why did the S&T do this in the first place?” What was really going on? We need to try to understand the times in which the 1990 decision was made The 1990 report includes a good indication of what was important in The first sentence in the report indicates officials back then were faced with an increasing number of these multiple-scale arrays entering the marketplace. The quote from the next paragraph reflects a belief that these scales were circumventing the 10,000 d limit on Class IIIL. This wasn’t about accuracy! Remember footnote 4 to Table 3 added 9 years later only deals with number of divisions, and consider the only requirement actually mentioned by the Sector in the section quoted was Nmax. The concern resulted from a hard-fought battle when Class IIIL was created. Industry argued strenuously that the digitized % tolerance structure of Class IIIL required no maximum on number of scale divisions. We have no Nmax for LMD’s or other codes with % tolerances. As hammered out, the final version of the code has the 10,000 d limit. Yet in our haste to protect that victory on Nmax, we ran roughshod over some other important principles.

15 Core Principles: TransactionsIn commercial transactions, commodities may be measured using: A single measurement from a single device Multiple measurements from a single device , e.g. gross – tare, summed ABWS Multiple measurements from multiple devices e.g. gross – tare (WIWO), fleet sales on RMFD’s, splash blending, multiple scale array No misrepresentation of quantity UWML§15 We need to ensure that officials understand the proper core principles. Quantities in trade may come from multiple measurements. HB44 recognizes it is not always possible or feasible to have one device of suitable capacity and division size to measure all quantities. While a single measurement from a single device may seem to be preferred by most of us, that is not actually supported in the code. HB44 has no section expressing a preference. Multiple measurements are equally acceptable for use in commerce. Examples of multiple measurements from single device - Gross and tare on the same device and the ABWS systems combine multiple measurements from a single device. Examples of multiple measurements from multiple devices – weigh in weigh out scales doing gross/tare, RMFD are often used for fleet sales where whole month of transactions are combined, splash blending, and the multiple scale array we have been discussing. From the standpoint of the UWML you must have valid measurement data for use in transactions (HB44 compliance) and you may not misrepresent these values in a transaction as covered in sections 15 and 16.

16 Core Principles: DevicesHB44 specifications apply to a single device, unless specifically stated. HB44 user requirements apply to a single device, unless specifically stated. HB44 tolerance requirements apply to a single measurement of a single device, unless specifically stated (e.g. shift test, repeatability, CIM RR track scales, gross/net agreement ATC, product depletion (VTM), wheel-load weighers used in pairs I believe the proper core principles need to be stated (I recommend in Fundamental Considerations) in order to prevent future recurrences. These guidelines follow G-A.1. I would formulate them as follows: HB44 specifications apply to a single device unless specifically stated. I couldn’t find a single example where HB44 specifications are to be applied across multiple independent devices. The 1990 S&T decision tried to do it without specific code to support their action. HB44 user requirements apply to a single device unless specifically stated. G-UR.4.1. Maintenance of Equipment (predominance) applies across multiple devices at one place of business. Its one of the rare instances. HB44 tolerances apply to a single measurement of a single device unless specifically stated. Consider an increasing load test. Each test load is a separate measurement and in applying tolerances, each measurement is compared to the tolerances independent of the other measurements. There are many examples of where tolerances involve multiple measurements. We have shift test, repeatability test, agreement for RR CIM systems, ATC gross/net agreement, and VTM product depletion tests. The case of the wheel-load weighers used in pairs is interesting. Tolerances are specifically applied to the sum of the pair, but none of the specifications or user requirements apply to the pair, only to the individual scales.

17 Actions Proposed Withdraw item 3100-1, it’s unnecessaryOfficially reverse conclusions of 1990 S&T item System is three independent scales! Move forward on to remove references to multiple scale arrays from Table 3 Footnote 4. Revisit 2016 changes to S.5.4. Vmin (recommend returning to original, pre-2016 language) Based on discussions at the NEWMA meeting May 2017 and further study on my part, I propose withdrawing the proposal for the General Code change. It is already covered in UWML section 15. I believe it is vital to formally reverse the 1990 decision. I’ve provided the code analysis necessary. The NCWM should reverse it’s position and treat the array solely as multiple independent scales. This is the alternative position from the Sector in The reversal should explain the code requirements are applied only to the individual scale as per G-A.1. This requires the Scales Code amendment to remove the two sentences from Footnote 4 of Table 3. While only the second deals with the “and one” scale, I recommend removing both as the first sentence is redundant. If the individual scales must comply, they obviously must comply with Nmin and Nmax for the designated class. In 2016, the code was changed to back out Vmin from the cherry picked items applied to the array. I am recommending we revert back to the original language, pre It was correct to begin with.

18 Conclusions and ImpactWe overstepped our authority in 1990! Scales were held to a bar raised above the code. Only when a manufacturer was seriously hurt did the 1990 decision get challenged. Let’s return to the county official I mentioned at the start. He did keep his gas pumps more accurate than the rest of the state? But at what cost? Increased maintenance costs for users. More inspections he had to perform to verify devices returned to service. No significant harm done. Yet it was WRONG! It was arbitrary and capricious, and not legal. We did the same thing with the 1990 decision. We held the multiple scale array to higher standards by inventing our own rules, as opposed to following HB44. All of that is now clearly exposed, it’s out on the table. Now that we are aware, we should have the integrity to fix it. Even the U.S. Supreme Court has had to reverse previous decisions. It should never be done without careful study, but sometimes it is necessary to make things right.

19 Impact of the Proposed ChangesThe individual scales in the Multiple scale arrays in marketplace always complied fully with HB44. Lowering the bar has no impact on current devices. Cap x div marking on Sum may cause confusion with a 4th scale, but it is not non-compliant. Real impact is on the W&M official Nmax and Shift test are applied only to the individual scales Use UR for approaches, not UR It is vital we recognize that the multiple scale arrays out there have always met the code as independent scales. They were forced to go above and beyond the code by the 1990 decision. When we lower the bar, reverting back to using HB44, we see no physical changes of any kind need to be made to existing equipment. Manufacturers and users should, for the most part, be unaffected by the proposal. The only issue I see is the capacity by division on the SUM required by the 1990 decision. Doesn’t this marking declare that the sum is a scale, when it isn’t. If the manufacturer decides to make it a fourth scale, the marking will again be required and will have the meaning that the marking was intended to convey. This is a scale! Yet I see no prohibition of marking cap x div on something that is not a scale. I would like to see the markings removed if possible. The real impact is on the W&M official. We have to lower the bar back to the written code. Both Nmax and shift test requirements can only be applied to the individual scales under the code. Remember the sum is not a scale. Also, the array does not conform to the definition of vehicle scale. It is multiple, axle-load scales used in combination. Approaches must conform with UR for axle-load scales. All we need to do is some re-education.

20 Feedback comments to please include “developing item” in title I thank you for your consideration of my proposal. I am providing my address so people can direct questions to me and offer comments on this presentation. I pledge to compile those comments and provide the compilation to the Committee in any final package I submit. Thank you.