Enhance your Inventory Management with Meddic Metrics for Inventory

How to create outlook signature

yeah thanks aha um I'm uh Jeff Rutherford director of research and development uh now at Highwood it's been almost exactly one year since I uh since I graduated uh from Stanford um and um I'm now uh metacore alumni um so yeah it's it's kind of uh it's it's interesting kind of strange giving a uh giving a talk to meta from a different from a different desk um and I guess that's not just switching uh switching seats from an institution it's going from an academic uh very academic uh area at Stanford to Highwood which is um uh very much more um industry industry facing um you look at problems a lot differently uh so this is going to be a pretty highlevel talk uh some of it'll probably be um material will be new to some people some of it'll be really uh very familiar old news um for some people but for me it's trying to uh put the pieces together and connect the academic way of looking at uh measurement informed inventories to where really where we're going with uh regulatory um and voluntary initiatives um so yeah I called this talk uh yeah the first steps on a long journey cu I think we're really at the start of our um uh incorporating uh measurements um into uh into our inventory approaches uh measuring methane emissions um but we know where we uh know where we need to go um so yeah with that being uh without further Ado I'll get started here so um I want to cover a few um seemingly basic questions uh why first of all um why are we doing this why our inventor is important um uh what is uh a measurement uh a measurement informed inventory a a simple but profound question um what is uh what is reconciliation are we talking about the same thing there or are they different um and then just a a quick example at the end of uh some of the some of the technical technical hurdles that we're uh that we're talking about um so I've had the I've had the pleasure of working a little bit with um uh CD uh cdph um uh uh Ben uh around the same time as I transferred from uh um Stanford to Highwood uh transferred from uh EDF working with uh Steph Rucker at cdph I think that's um Colorado um uh as usual is a is a leader in this space um and uh I think their Journey uh provides a really good example of why inventories are important and uh how we're starting our long journey to measurement informed inventories um in 2019 Colorado committed to achieving a 50% reduction in overall emissions by 2030 and a 90% cut uh by uh by 2050 um this was a as a part of setting out their uh greenhouse gas emissions road map um and uh so it's one of it's one thing to set these targets um what about how do you know that your uh actually tracking on them um and as we all know um tracking some of these things is a lot harder um than uh you might originally think um so first uh in 2021 um Colorado introduced a revision requiring oil and gas operators to submit greenhouse gas intensity plans um and then uh in in in line with that uh cdph Department of Public Health and environment uh was tasked with developing plans to check the progress um of those uh uh of those intensity plans um specifically for uh um oil and gas operators uh a greenhouse gas intensity ver verification program um tied to Colorado's annual emissions reporting program uh was established and cdph was also required to evaluate um aerial and groundbased Survey Technologies and as well as reviewing the effectiveness availability and reliability um of uh of those uh Technologies so that is important and relevant for us because it says two key things it says that um well Colorado is taking uh by in introducing this greenhouse gas emissions roadm road map they are taking the inventories very seriously because the inventories um are what is going to be used to track those Pro track the progress um second second um Colorado has recognized that measurement is critical to establishing a credible emissions estimate by introducing this uh intensity verification program so for everybody on this call-in meta this is uh this is a big deal um now how did we get here um to this place where uh Colorado Regulators are creating this uh intensity verification program um The Meta Community um this slide will be no news to anybody here um the medic Community has been emphasizing the importance of measurement for a long time so it's really exciting to see this work uh make it into the regulatory sphere um it's been about a decade coming now but um uh research has uh pretty much established that um in many and probably most cases uh our bottomup uh invent traditional bottomup approaches are poor a poor model for the on the ground real ity of methane uh emissions estimates um I highlight some of the work here but there are many other examples um uh this uh on the left is work by um uh my old Stanford colleagues Julia Chen um Evan Sherwin um in their analysis of the uh uh peran Basin um highlighting a substantial disagreement um between the uh inventories and their uh best estimate based upon aerial data um but also um uh Matt Johnson and Bradley Conrad have done a lot of great work uh in western Canada um publishing estimates for uh Alberta and British Columbia um this is based upon collaborations with Bridger and these are the disagreements are less but there's still uh from the from the perspective of a credible greenhouse gas inventory there um they're still substantial uh so in general these updates these updated estimates are based upon top down approaches and we are comparing them with the more traditional bottom up approaches um top- down approaches leverage uh technologies that measure um at the site level uh Basin level or Regional scale um so you're it's sort of a trade-off you're trading off the granularity um that you're able to achieve with uh um the bottom up estimates with these uh these capabilities that you're able to achieve from surveying at a higher level and seeing the emissions in a different light um bottom up approaches uh so well top down approaches these include uh aerial remote sensing space-based remote sensing um vehicle vehicle based uh downwind plume Point sensors uh could also include continuous monitors um the uh bottom up approach es that underpin our traditional inventories are generally based upon some direct uh sampling of sources with uh high flow Samplers or bagging uh but this could also include inline inline flow sensors for measuring emissions uh uh flows from pneumatic devices and uh and flares so the key difference um and I'm going to uh pretend I'm Evan Sherwin for a moment here uh but the key difference here is that topown Technologies uh survey more faster um and uh I would say for me at least um working with Evan the uh updated estimates that we're generating are just simply uh based upon more data there's obviously a lot more there there's uh additional additional Nuance but original emissions factors um if you dig into the the epa's inventory these are um essentially based upon a handful of measurements that are used to generate the emissions factors um in the 2010s um when we uh uh we were looking at edfs uh EDF surveys that they were doing with the uh downwind vehicle based Samplers and uh Omar's work uh alvare the Alvarez work these were based upon uh order hundreds of measurements um and then starting in uh 20 uh uh 2019 and onwards with uh work from carbon mapper um and estimates generated uh from uh by Evan and Julia and then Matt Johnson's work these are order thousands of measurements and in some cases tens of thousands and now with uh Evans um distributions paper we're looking at 1 million measurements each time we scale that up our emissions get higher um so just to kind of uh highlight or put a finer point on this um this is a cumulative uh distribution function um from uh the uh peran 2019 uh distribution in Evans uh in Evans paper um and just looking at uh each of these uh orders of magnitude for uh for the uh emissions rate so if we're looking at emissions that are higher than 10 kilograms per hour um this is about 4% of sites but that con constitute 8 constitutes 85% of your total emissions for this particular distribution emissions greater than or equal to 100 kilograms per hour sight level emissions that is um we're looking at 0.9% of SES but 76% of total emissions um if you scale that up even more one thou emissions greater than or equal to 1,000 kilograms per hour just 0.1% of sites but we're still talking about 32% um um 32% of your total emissions um so you can really see uh See here what the the magnitude or the um uh amount of emissions that you're missing if you aren't sampling uh if you aren't sampling enough um and really highlighting that it's really only been in the past few years that we've been able to achieve sampling on the order of thousands um so I think it's it's it's diff uh it's important to emphasize that here um and thinking just thinking about the tradeoffs that we have between uh these two different types of measurement or ways of generating inventories top down approaches versus bottom up approaches there is a tradeoff with top down approaches you're losing that granularity but um and I think this is a uh I remember this from one time I heard Evan presenting this material um it's better to be roughly correct than precisely wrong if you're missing one of those large emitters that captures that long tail of the distribution um that uh that will have significant effects on the accuracy of your total emissions estimate so now I'm going to sort of um uh pivot uh this talk a little bit and talk about how we are applying these estimates into measurement informed inventories via the process of reconciliation so the way I think about reconciliation um there isn't really a single um there isn't really a single definition in the literature or it really depends where you look um but uh Thomas has a great blog post on this um and the way he describes it as the comparison of two or more different emissions estimates and the following investigation into why they into why they differ so the the work that we presented on the first slides those are examples of reconciliation herian paper is an example of reconciliation Matt Johnson's work is an example of reconciliation you're taking your bottom up inventory estimate you're comparing it to a new estimate based upon sight level top- down data and you're looking at why why they're different and how they're different um this could include digging into the engineering calculations the emissions factors of your bottom up inventory it could involve uh looking at the composition of sources within your site level estimate or in some cases as we're seeing with uh more recent work looking at supplemental data to better understand the intermittency of these uh of these large sources that you're detecting with your site level estimates um now what I've been spending a lot of time thinking about is how reconciliation is deployed and used under different programs um we're in a space right now where there are uh are a lot of different bodies a lot of different institutions that are thinking about how best to um voluntarily give operators the opportunity to demonstrate um low emissions or better emissions performance um via a number of different programs and also regulatory bodies that are looking at how to enforce this uh at a at a state or country level um I'm going to be uh rattling off a few different names here including the uh uh gas technology institute's Veritas program um the uh oil and gas methane partnership um miq um but Highwood has a excellent um voluntary initiatives report that um I keep open on my desktop as I um scramble to catch up with uh Thomas and his team and learn about all these um so part of uh my sort of Journey towards understanding these programs has been to try to understand generally the common aspects that each of them uh that each of them share and these aren't a whole lot different from uh the academic papers but it's really the details and the motivations um that differ and we'll get there um but in general um I like to think of these in a few uh sort of simple steps um it really starts with measurement measurement is at the core of this um but you you know you might say it's also the hardest part because it it involves selecting amongst amongst a vast uh array of technologies that are out there out there right now Each of which has their uh strengths and their limitations it involves determining how much you should be surveying um when you should be surveying um and uh and then next in the step of um what I call data consolidation and Analysis it's taking those uh taking those measurements and uh organizing them performing some initial analysis to come up with site level totals um or event level totals um so each site will probably be covered multiple times um uh and with some technologies sites are going to be covered continuously throughout the year um but from that data you need to come up with uh event level magnitudes um sometimes event level durations and then those need to be extrapolated to the entire year um so there's a lot going on there in that step next is the reconciliation step and this is really where um things start to diverge depending on the program um and I'll I'll talk about this in the next few slides but uh in some cases it's a simple comparison um but in others you're you appending or you're adding these large emitters to your to your bottomup inventory uh in some cases it's replacement of bottomup inventory uh with your uh uh with your uh measured totals and then also this is yeah this is all sort of overlaid on top of your uh initial your initial inventory and then um depending upon how these programs are constructed what the overall motivation is there is uh decision- making that happens on top of your uh on top of your measurement informed inventory there's adjustments to it you're making changes um uh key point being that this is this is an iterative cycle we're starting somewhere uh we're starting from the very beginning and the part of the process and part of the way that these programs are geared towards is is Improvement every year start with uh select a few Technologies start using them introduce them introduce them into your inventory and make progress on that uh make progress on that every year so I'm not going to um uh I don't have the time or the uh wherewithal right now to go through these in a tremendous amount of detail um but I just wanted to uh highlight some of the uh key similarities and differences um across across each of these programs um I've tried to bundle them into into similar into similar categories um and also highlight what data is used in under each of these programs um so um starting with the epa's uh proposed update to the uh greenhouse gas reporting program for oil and gas uh oil and gas gas systems um what they're doing is a supplemental or additive approach um so the EPA has introduced a new category for uh other large sources um and this this category of other large sources is being put on top of the uh of the traditional bottomup inventory and there's also some changes there in terms of adding new source level um source data but overall this is additive um now uh the oil and gas uh oil and gas methane partnership um is a different approach from that um under uh ogmp 2.0 um and for those not familiar this is a a reporting and mitigation commitment framework um that uh close to 153 companies have uh or sorry 150 it's around 147 something like that right now 147 companies have signed on to um this is more of a comparative uh comparative style framework where uh the company will come up with a level four inventory um that is based upon uh Source level uh Source level measurement uh data and bottomup calculations and they will validate that level four inventory with um uh with site level data in level five so um the way I can the way I put this together might be a little bit misleading because the level five inventory is still composed of the uh Source level measurement data and bottomup calculations but it is compared compared to uh site level measurement total um and it is uh it is validated once these two once these two match um so that's the uh ogmp 2.0 process for uh reconcil iation um miq is uh similar to uh EPA subpart W in the sense that um as they describe it um uh I believe it's unintended uh unintended or additional sources are added to your bottomup inventory um if they are considered to be outside of the typical um operating or process conditions um so this is another kind of additive Style inventory that that is composed of all different kinds of data including uh the sight level emissions and Source level measurements and bottom up uh calculations finally uh Veritas um Veritas is a standardized methodology um this is not a a commitment or a certification Veritas aims to come up with an approach to determine your best estimate of uh best estimate of true measured emissions Veritas is probably the closest to the uh the way that um reconciliation has been done in some of these um academic papers where you're you you take your sample of measurement data and then you use that and you extrapolate that in different ways to best understand your total emissions over the course of the entire year um there are multiple pathways in baritas where uh pathway one is measurement only and is strictly comprised of your uh uh site level measurement data pathway two involves uh distinguishing sources as either best measured or best calculated and using the measurement data uh in the categories where the operator deems it to be to be the most appropriate um so yeah that was a that was sort of a really uh quick run through of all of the different Frameworks um but just highlighting uh how measurement data is used in each of these different cases how reconciliation um is conducted and how some of these are similar and some of them some of them are different so uh like I said at the beginning where in uh we've taken the first steps in a long journey um and there are still um technical hurdles that need to be uh that need to be figured out in order for this um in order for these programs or um or for reconciliation and measurement informed inventories to uh to really become successful um so I um I'm highlighting here a paper that recently came out um uh lead author um lead author arbent um and they highlight a lot of these uh technical hurdles here in this paper um a few of them being that uh all of these Technologies are currently undergoing rigorous testing um how do we incorporate what we're learning from these testing uh uh what we're learning from these uh controlled release tests um into the measurement informed inventory reports so how do we improve our understanding of the uncertainty in measurement informed inventories um and then also uh based upon the strengths and weaknesses of each of these Technologies how do we uh leverage multiple of these Technologies um sort of a multi-tech framework to improve uh to improve our best estimates um and then some of these Technologies are are deployed uh for only a fraction of the year so this requires or in some cases only a fraction of the sites will be surveyed how do we properly extrapolate short duration data um to an annual ual emissions estimate uh and then finally um this is something I've been thinking about um with li based upon the nature uh of these skewed distributions um limited sampling will uncover less large emitters compared to more extensive sampling so it's sort of a situation of the more you sample the more you pull up the rug the more you're going to find um and how do you incentivize uh sampling to the greatest extent possible without without penalizing people that are uh that are you know um trying to do the best they can and use measurement Technologies the best they can so these are all these are all difficult problems not necessarily problems that have been solved but uh that we're all working on so I just wanted to illustrate um the challenges and where we've gotten with some of these uh uh technical hurs with a really quick example um so so here's a here's a situation um for a hypothetical operator um where they've they've conducted two sampling campaigns using one of the multiple aerial remote sensing Technologies um they've performed sampling in the spring they performed sampling in the fall uh this is a Time series for a single site and a single Source category and in reality that sight Source uh combination has a continuous emissions Trace throughout the year um where there'll be uh sometimes it'll be sometimes there'll be uh emissions sometimes they there won't but it's a continuous Trace um some of these per some of that uh Trace is observed by the technology and uh some of it is not so how do we um incorporate this short duration measurement data into our measurement informed inventory um answer is it depends um it depends upon the framework that or what this operator is subject to be that uh regulatory programs or voluntary programs that they're participating in so the first case um uh the first case here would be uh additive Frameworks um and this would include uh EPA subpar W uh it includes miq both of which I uh both of which I highlighted um on the earlier um on the earlier slide uh where the the base of their inventory is the uh traditional bottomup approach and what they're doing is they're uh depending upon conditions and criteria that exist in both of these Frameworks they're adding um they're adding large emitters to their inventory um so in the case of miq it's It's relatively uh relatively non-prescriptive um where the operator um based upon the uh the data that they've collected um all of the technology that they're deploying um they must provide evidence of calculating the rate and evidence of calculating uh the duration for these large emitters and depending on if the uh uh the emission event is deemed to be uh unintended or additional to their inventory they'll add it in so uh there's really no requirement for them to look at this period of unobserved um uh period of unobserved emissions it's simply what technologies have you deployed what data is available um add it to your inventory um in the proposed EPA subpart W um uh we've already talked about the new large uh other large release events category uh in this case the events added are those detected either via quad screening or the super emitter response program and uh there's details there's details in there about what constitutes additional um or other large uh other large sources but that uh that includes greater than 100 kilograms per hour um 250 metric ton uh total emission events over the course of the year so this could in the case of subpart w it's not just what you've been observing through your um through your screening programs it could also be uh public data so um did the uh methane sat is going up soon did methane sat fly over or did the new carbon mapper satellite fly over during those unobserved periods um so that's the uh there are also uh examples in the literature um that have demonstrated how multiple Technologies uh can be used under these additive Frameworks um so you know we've already talked about how um throughout through your screening campaign or through public uh data sources um you'll you'll capture you'll capture an event um you'll detect that event um and that'll provide a raid estimate um but you can also uh Implement other data sources to bound the duration um to bound the duration of those events and and basically hone in on the specific period of time uh where that emitter was occurring and and thus you know come up with a better estimate of the total emissions um that were released uh for that for that particular event uh rather than in the case of uh EPA subpart W using the default uh value of 182 days uh so uh will Daniels has a has a really nice paper where he looks at using uh continuous monitoring uh continuous monitoring data to come up with a uh time bounded uh duration for these uh emitter events and um uh that being an example of a multi uh uh multi technology framework for emissions reconciliation so the second set of um uh the second set of uh examples that I'll look at for uh uh reconciliation here would be the extrapolation Frameworks um so under these extrapolation Frameworks you need to account both for both both the measured and un unmeasured sources uh in coming up with your total annual emissions estimate um because in this case we're not using the bottomup inventory as our base for uh for our total annual emissions estimate we are using the measured data to come up with the come up with the total so we're not just looking at additional events we're looking at the whole thing um and this is a little this is a little bit more complicated um so two examples examples here would be the we've already talked about uh the Veritas approach um where um there's two Pathways first pathway is measurement only second pathway is a hybrid approach where only the best measured calcul best measured categories um are used for measurement data but still uh under both Pathways you are extrapolating short duration data sets to come up with your total annual emissions estimate in ogmp level five you're required to come up with a site level emissions total that you use to validate your um Source level uh level four emissions inventory so there are um examples in the literature about how to go about this extrapolation um and uh these um I'm going to start with the example of a single technology solution so um uh the uh ulia's Julia Chen's paper in the peran um Matt Johnson's work these deal with a aerial remote sensing um Technology Solutions for this uh extrapolation approach um in this so in this case the goal and uh the reason that I'm highlighting these uh two single observation periods um is because we don't have most often we will not have aial data over the course of the entire year you're going to have a a discret period where um whoever your vendor is they go out and Sample over the course of several weeks um what you need to do is then extrapolate those to the remainder of the year you need to say something about what is happening during that unobserved period um so uh and uh Julia has a very nice appendix in her paper where she talks about how this how you do this but basically you need to prove that the data that you're collecting during these observed periods provides an unbiased estimate of emissions during the unobserved periods um and there are there is a list of criteria in that appendix or a set of assumptions under which that unbiased assumption is valid um but the key assumption being that your distribution of emissions during those observed periods and this is a cartoon distribution here um that is that must be the same or within a certain set of bounds um similar to your emissions distribution during over the entire year or during the unobserved period and the way that they describe that is we are assuming that the distribution is stationary uh across the entire year stationary here meaning that dial weekday weekend and seasonal Trends are um relatively NE negligible um obviously the the situation is going to be more complicated than this um but these are the set of assumptions that we need to make in order to perform this extrapolation um so uh what about uh multi-technology deployment and using uh multiple different types of uh aerial remote sensing Technologies using satellites um can we add satellites in during that unobserved period um what if uh car mapper happens to have some data on their data portal that we can sprinkle in there um what about using uh continuous emissions monitors to um I mean if you have if you have one of these technology deployed they're going to tell you something about the frequency and the duration of your emitters during both The observed and the unobserved periods um I'm thinking about ways that we can that we can do this um still we're still working on it I'm sure there's a lot lot of people that are working on this but uh to me this is the frontier and this is where uh this is where we want to this is where we want to go is we want to come up with rigorous methods well documented um for how to perform these types of multi-technology deployment so um I don't have a unfortunately don't have a good citation to put in there right now um there's been uh although I would say there there is some good work uh published by Aon toos on this um I I should have put her her citation there but anyways I think this is um to me this is the frontier for uh multi-technology deployment and measurement informed inventories so that concludes my presentation um I'd say uh a a plug for Highwoods uh voluntary emissions um reduction initiatives report uh I also learned how to put uh QR codes in my slides um so please go and download um a few uh few key points to uh to highlight um from this report being it's been acknowledged I think um to me Colorado uh the uh uh Colorado um intensity verification program was pivotal um the updates to subpart W were pivotal it's it's widely acknowledged um the importance of measurement in accurate in accurate emissions reporting um this is also uh reflected in the growing participation in volunt emission reduction initiatives um and part of this is because top- down technologies have produced a step change in our understanding of methane emissions but also introduce their own challenges uh in terms of credible uh in terms of credible emissions reporting um and uh I highlighted some of these in the um in the technical hurdles that we're facing um we're not going to solve all of these at once I think these uh that's why these voluntary initiatives exist right now and we're all going to be working collaboratively and iteratively um amongst uh The Operators technology vendors Regulators academics and nonprofits all of whom are represented on this call um so thank you for listening um and thank you Sahar for uh introducing me or bringing me back to meta to share some of the stuff that I've been thinking about and working on and