Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
H-1 APPENDIX H. EVALUATORSâ COMMENTS FROM ROUNDS 1 AND 2 In both rounds of the evaluations of alternative design concepts, the evaluators had the option to provide comments along with their ratings and many did. Section H.1 compiles the evaluatorsâ comments on the 5 design concepts from Round 1. Like the scoring, the evaluators expressed their views of the Datum design compared to each of the alternatives. Section H.2 compiles the comments on the two designs evaluated in Round 2. H.1 Evaluatorsâ Comments from Round 1 H.1.1 Datum vs. Alternative #1 Datum vs. Alternative #1 Current Preferred Design with Build Sequence Build on AEDT â Because the AEDT method would initially incorporate agency-approved computations for rail/road, I think that agency acceptance would be much higher for AEDT than for the Datum. H.1.1.1 Agency Acceptance â In addition, if the Datum's Screening Tools were added to this AEDT alternative (easy to do, quick, multimodal), then this alternative would have multimodal capability very quickly. I assumed this will be added in my score here, since so easy to do. â In addition, I don't think that DoD's ultimate goals are very important to this current DOT effort. â It is building upon a current DOT model (FAA), but it will result in multimodal capabilities much later than Datum. â This is a difficult criterion to judge, as it begs the question -- what agency. One could argue that for the FAA, building on AEDT is better than the datum. But that would not be the case for FHWA, or even DOD. This criterion may need some discussion for this alternative. As such, I've rated it neutral. â Cons of integrated models by some agencies, noted. Likewise, the lack of inherent fidelity improvement in Alt#1 also noted. However, establishment of an enterprise system that is further supported by a data clearinghouse offers significant currency for agencies looking to demonstrate results to their constituents. The datum progression offers quick results but on a narrow base unless significant resources are applied to establish a broad archive of simulation based results. â FAA would undoubtedly accept an expansion of AEDT, but other DOT agencies might be less willing. â FAA would be OK with Alternative #1 (in this regard). However, other agencies might be reluctant to adopt this model as their standard. For one thing, they might get resistance from their "user community", who would need to undergo a sizeable conversion effort. â Because the AEDT method would initially incorporate agency-approved computations for rail/road, I think that agency acceptance would be much higher for AEDT than for the Datum. H.1.1.2 Technical Feasibility â In addition, if the Datum's Screening Tools were added to this AEDT alternative (easy to do, quick, multimodal), then this alternative would have multimodal capability very quickly. I assumed this will be added in my score here, since so easy to do. â In addition, I don't think that DoD's ultimate goals are very important to this current DOT effort. â Similar technical challenges exist with both approaches, but the sequencing in different.
H-2 â While the ability to run individual analyses is probably easier in the datum alternative, the configuration management of the software, databases, and results as well as the information management of the input and output data is significantly improved making larger and more complex analyses more feasible. â Datum approach has higher probability of success (at least in the short term) due to less ambitious scope. â Seemingly easier to implement and build upon existing methodologies or those already in development. â FICAN opinion was based, I believe, on the need for aircraft audibility computations for aircraft. Simulations of rail/road were not considered by FICAN, nor are audibility computations needed for rail/road/marine. H.1.1.3 Analytical Proficiency â AEDT approach to motor vehicles would be greatly improved under this alternative, thus eliminating the fourth negative bullet for AEDT. â In addition, the third negative bullet for AEDT applies really just to TA, which is not needed for road/rail/marine. â Integrated approach has reached its computational limits. â Would require the use of the same metric. If FHWA would consider changing its metric, it would have to through the Federal Register rulemaking process. â The end state of the two approaches is essentially identical. â The proficiency of the two alternatives is roughly equivalent. However in particular settings there are individual advantages. Specifically in the datum, a broader class of problems can be addressed, e.g. from screening as noted to high fidelity simulations, which is a positive. However in Alt #1, the system would be more adept at fully analyzing the somewhat narrower field of problems to which it is tailored which is a positive. Likewise, in establishing a system which covers the "fat" part of the curve enumerating the multimodal problems of interest and not the tails on either end, there is reduced development and maintenance cost which is also a positive. â It will not advance the state of the art â In the short term, the 'loosely integrated' Datum approach allows one to take advantage of progress in developments of individual models (TNM, NOISEMAP, AEDT). With Alternative #1, it is harder to incorporate future progress. â Both designs are too complicated for small projects, but at least Datum builds in screening tools. H.1.1.4 Scalability â Both approaches equally scalable. â Formal information management of the underlying structure enhances analyst capabilities and reduces work load. Computational scalability would be about the same as the datum. â The basic architecture is established and might be difficult or expensive to change significantly. The datum has the flexibility for scaling for different size projects. â For large-scale projects, Alternative #1 would be preferable, achieving an economy of scale through an integrated approach. For small-scale projects, Datum approach is preferable - especially if one is interested primarily (or exclusively) in one mode.
H-3 â I do not believe that the simulation end state of the Datum alternative gives FHWA, FTA or FRA anything that they need. Instead, only the Screening Tools of that alternative would be of use to these three administrations. And development of those screening tools could easily be incorporated in the AEDT alternative, as well. H.1.1.5 Responsiveness â Since I don't believe that simulation is at all responsive to these three administrations, then most all of the Datum alternative is wasted for them. I see much less "waste" under AEDT, so therefore have rated it higher than the Datum. â AEDT project has not demonstrated much responsiveness to the user community. â With a multimodal capability being the focus, the datum gets us there more rapidly. â More formal structure will inherently have more development and analysis inertia but is a trade with the credibility criteria. â By virtue of the build sequence, the Datum could be considered the most responsive. Also, AEDT may not allow for new noise metrics â Datum approach facilitates responsiveness due to its 'loosely-coupled', highly modular approach. â Better because its core components are currently accepted worldwide. H.1.1.6 International Credibility â I am aware that other countries are using the FHWA TNM, but they are not my main concern. My primary customers are the State DOTs, therefore I gave this a "same" listing. â Both approaches equally credible. â More structured information technology support and accessibility across a wider stakeholder base. The data clearinghouse in particular would establish credibility. â EC has already identified modeling processes more advanced than these and is in the process of implementing them â So far, the international community has not been overly willing to accept AEDT â Both approaches are roughly equal in this regard. â I don't think the development costs here will be as low as anticipated, since the propagation algorithms for road/rail will have to be very much more complex than those for aircraft (to gain administration approval). And so the algorithm for ground shielding and ground attenuation cannot be shared among these modes. H.1.1.7 Cost Implications â However, if sharing in enforced for propagation, then the aviation computations will have to bear the "computation burden" required by the ground modes--a gross increase in computation time compared to what is truly needed for aviation computations. â Note that this also will require entire different input datasets for topography---for ground-mode vs. aviation computations. â In spite of all this, AEDT will still be much cheaper in my view. â Assumed the Datum funding went all the way to the end state.
H-4 â If all capabilities are adopted the overall costs will likely be very similar when comparing the two approaches. But, I prefer the priority-driven approach, which by definition will lead to investment only if the need is there. â Initial Alt#1 implementation higher but maintenance and life cycle cost lower. â It will continue and expand an ongoing project. As such, past investments will continue to see returns. â Cost comparison depends on how far along the build sequence the Datum would go. Alt 1 would be less expensive to achieve the same end state. â Alternative #1 would probably have lower development costs over the entire cycle, but higher upfront costs and higher risks (due to tightly-coupled integration). User costs: probably greater for Alternative #1 in short term (large, single model). As well, maintenance costs for Alternative #1 is probably more (any advances in sub-models would have to be incorporated into large model). â Some tools already under development. H.1.2 Datum vs. Alternative #2 Datum vs. Alternative #2 Current Preferred Design with Build Sequence Build on Existing Simulation Models â Agency acceptance will not be known till after the effectiveness of early training is known--and probably not even then. H.1.2.1 Agency Acceptance â Lack of an intermediate screening method, in the meantime, is a shortcoming here. I am assuming that the Datum's screening task will be added to this alternative, as well. â Although I've rated this as "5," it's not as high as rating as the "5" for Alt.1. â FAA large R&D effort on AEDT does not include or plan for simulation. â Multimodal capability in the near term is more important than the simulation capability in the near term since many of the advantages of simulation modeling cannot be realized until adequate source data are available. â Commonality of first principle algorithms and roughly equivalent existing body of results. â Agencies will be hesitant to scrap ongoing projects, but should recognize the superiority of simulation modeling. â Agencies would prefer a step-by-step build sequence to re-evaluate progress versus needs. â Alternative #2 would require all agencies (and their respective user communities) to accept this tool. â Because this end state is essentially the same (potentially) as that of the Datum, I have rated it technically the same. H.1.2.2 Technical Feasibility â In my view, the second "negative" of this alternative (to the left) is equally shared by the Datum. â [Equally feasible] Based on the similar end states.
H-5 â I am severely concerned about the impossibility of modeling intersections - FHWA project very often involve intersections. If this is not a possibility, then this option is much worse then datum. â Both technically feasible, but the question comes down to sequencing, and due to data availability and other reasons, having the multimodal capability before the simulation capability makes more sense. â Demonstrated framework of Alt#2 for multimodal analysis offers more confidence in success of the system â Starting directly with a bottom's-up design approach using simulation techniques will likely be easier than evolving legacy approaches of fixed formats. I do not regard lack of source data as a problem, as simply converting existing data to a simulation-compatible. Format of equal fidelity will provide the opportunity for more advanced propagation techniques to be used off-the-bat. â On one hand, Alternative #2 is very ambitious (in that it is creating a program from scratch). On the other hand, it is not encumbered by any legacy issues. On balance, though, the Datum approach has the edge in technical feasibility. â In my view, all of this alt's negatives are equally shared by the Datum. H.1.2.3 Analytical Proficiency â Based on the similar end states. â I believe the "+" outweigh the "-" in this case. Most of the "-" are common with any type of sophisticated model. â The end state in terms of analytical capability is very similar if not identical. â Although individual modal simulations in Alt#2 are more sophisticated, bringing the individual pipelines of results together only at presentation stage mitigates some of that benefit. With the lack of screening ability the higher fidelity simulation fidelity of Alt#2 would only benefit well endowed stakeholders with the resources to conduct such analyses. â As a program that is designed from the start to be a comprehensive state-of-the-art simulation tool, Alternative #2 can be expected to provide better results. â Datum not conducive to small projects (except screening) while building from scratch offers opportunity to incorporate appropriate scalability. H.1.2.4 Scalability â The end state in terms of scalability is very similar if not identical. â Pipelining simulations within modes affords a very scalable architecture both for the analyst in constructing studies and computationally for the system carry out the evaluations. â The Datum's building-block approach will provide a means for scaling not only algorithmic complexity but also a means for legacy algorithms to be used. â Alternative #2 has the potential for increased scalability due to its 'clean-sheet' design. However, this would require a disciplined development effort to ensure that the scalability potential is realized.
H-6 â Same very bad responsiveness as the Datum, in my view. H.1.2.5 Responsiveness â Abandonment of legacy approach seems to abandon/ignore the regulatory requirements that necessitated the legacy tools. â Both approaches equally responsive. â Although on balance the same, again there are competing strengths and weaknesses of the datum and Alt#2. For some categories of stakeholders the Datum system will be more responsive, e.g. compliance oriented users. For other categories, say, those seeking state-of-the-art models Alt#2 are more so. â Going straight to simulation is committing to an all-or-nothing approach â Lack of a build sequence â Alternative #2 is more of a labor-intensive effort, especially in the short term. Because of this, there might be pressure to freeze the design as early as possible, and thus be less responsive to evolving needs. â Seemingly more difficult to modify to respond to changes in regulations/agency requirements. â Alternative demonstrates more openness to input from others (international research community). H.1.2.6 International Credibility â I am concerned with national credibility, then international credibility, therefore this will be the same for all alternatives. â Both approaches equally credible. â Going straight to simulation puts analysis techniques on-par with those used in Europe â As a model that is simulation-based from the ground up, Alternative #2 can be expected to have more international credibility than the Datum approach. â Full simulation modeling seen as more "internationally accepted". â Lower than Datum only because it leaves out all the intermediate steps in getting the End State, which is essentially the same as the Datum? H.1.2.7 Cost Implications â However, leaving out some of these intermediate steps is not good, in my view. So this cost savings is really "less product for less money." â Little or no leverage with AEDT development means that FAA R&D expenditures to date would have been wasted. â You get what you pay for. Although this will be extensively more expensive, it would be strongly advancing the noise and air modeling capabilities in the U.S. â Because of the staged approach, the datum gets to a multimodal design within the first year. Thus the costs are very predictable. Whereas the costs to move straight into the simulation realm will likely be higher. It may turn out that the costs to get to the end state in each case will be similar. â If these two models were developed independently beginning right now then I believe simulation modeling would be cheaper. However, many funds have been expended on existing models'
H-7 development so ignoring these investments and building a model from scratch effectively wastes these monies. â If the Datum end-state is achieved, the build sequence in the Datum would increase costs compared to a single build. However, the learning curve developed in the build sequence could reduce overall costs â The Datum approach will likely have greater overall costs (because of all the intermediate steps). However, Alternative #2 will likely have higher up-front costs, and will likely involve greater risk. My judgment is that on balance, the total overall cost of each approach is about equal. H.1.3 Datum vs. Alternative #3 Datum vs. Alternative #3 Current Preferred Design with Build Sequence Federal Adoption of Commercial Software â Same very bad prospect for agency acceptance as for the Datum. H.1.3.1 Agency Acceptance â (This alt's page A-1, Road Noise): FHWA has not approved SoundPLAN computations. I was part of FHWA's evaluation of SoundPLAN and am well aware of SoundPLAN's failure to match TNM results (and SoundPLAN's designer's refusal to change their code to make a match). â In addition, the propagation method within SoundPLAN is inherently unable, in my opinion, to account for at least two essential aspects of highway-noise propagation. â Cannot imagine regulatory agencies agreeing to depend upon a commercial entity. â Coming from a Federal agency, this option provides a loss of control. This is not a turf issue, but rather a regulatory compliance issue. The FHWA TNM is incorporated by reference and therefore required by, 23 CFR 772. If FHWA endorses a commercial model as the primary model, then it would revise 23 CFR 772 through the Federal Register process. The other issue is the cost to the users. Currently FHWA provide TNM free of charge to it primary stakeholder, i.e. State DOTs. â The risk is far too great here should the vendor go out of business, unless licensing could be arranged to include the source code. â Volatility of solutions in a market-based dynamic equilibrium of Alt#3 is a distinct negative. â Some will be reluctant to scrap ongoing projects, and yet others will be happy to. â Agencies would be unlikely to rely on the market to dictate performance and costs - particularly costs to the users â Alternative #3 would require all agencies to abandon their own noise/emissions models, and would mandate their user communities to purchase commercial software (and a foreign one at that). These are serious barriers to acceptance. â US-approved code in TNM (planned for Version 3) will not fit into the modular propagation routines within SoundPLAN and CadnaA. They will have to redesign their products substantially. H.1.3.2 Technical Feasibility â Lack of US source data key factor. â Lack of source data is pretty much a wash between the two.
H-8 â This is a difficult criterion to rate, as it is unclear what, if any, domain knowledge the software firm may or may not have. For this criteria, all the risk shifts to the software firm. To what degree the Government could influence development is completely unclear. â Prioritization of problems of interest allows the selection of "winners" across the field by problem category. The integration step however is non-trivial and it is unclear where that responsibility would lie. â It takes years and $millions to develop a federally accepted model, whereas the commercial software exists and only policy would need to change. â With Alternative #3, there will be some tension between the commercial software developer and the agencies regarding direction of development. For example, a commercial software developer might not develop a fully-integrated product (i.e. the desired end state) if it deems that there is not enough of a demand for it to make a profit on that effort. â US-approved code in TNM (planned for Version 3) will not fit into the modular propagation routines within SoundPLAN and CadnaA. They will have to redesign their products substantially. Same as Datum, more or less, but much worse than possible with AEDT (once supplemented by the DLLs that will emerge from current TNM 3.0 development). H.1.3.3 Analytical Proficiency â These models were not based on a simulation foundation, which is where we need to go. One would have to be developed. In essence these tools implement the procedures within tools like INM, TNM, etc. So, they essentially have a similar proficiency with what we have today, and comparable complexities to moving towards simulation, if this is even planned. â Highly tailored solutions that make the system easy to use enhances Alt#3's proficiency. There would likely be compromises on the underlying fidelity/accuracy as there is a lack of fundamental research and long-term development in market-based system solutions. â The datum's modular approach is based on domestic models, whereas modules available within a commercial software package will include algorithms both domestic and non-domestic. â Alternative #3 could turn out better or worse or the datum, depending on how well the software developers can work with the regulatory agencies. On balance, I deem the analytical proficiencies scores equal. â [Alternative #3 more scalable] Based on a cursory look at the types of projects that have applied either SoundPlan or CadnaA. H.1.3.4 Scalability â Both approaches should be equally scalable. â Scalability depends a lot on the design of the base software. I will assume that the two approaches are equal in scalability. â Mainly because so much is out of the control of US agencies. H.1.3.5 Responsiveness â Commercial developers could give up any time they wished. And I think they would do that "soon," once they truly understood how different the US would want matters, compared to the products they already have developed.
H-9 â While the alternative appears to more responsive technically, there is substantial risks that certain regulatory requirements would not be fulfilled if left to the determination of a commercial entity. â Of course this will be better - we are dealing with commercial models where the companies will have huge financial gain as a direct result of their responsiveness. â The risk is far too great here should the vendor go out of business, unless licensing could be arranged to include the source code. â Again highly dependent on the category of stakeholder using the system. For those users very close to the development team and sponsoring agencies the Datum will be very responsive. However, for other commercial and consulting agencies using third party support tools Alt#3 will be more responsive at adopting standards and plug-n-play capabilities. â Surely response to change through professional software development will be more efficient than existing government contracts processes. â Highly dependent on the developer commitment and the market â It is expected that a commercial software developer, placed in a competitive situation, might be more responsive. â Example products seem to be widely used, especially in the EU states. H.1.3.6 International Credibility â Since these companies are from other countries, I am sure international credibility will be better - but since I am not concerned with international acceptability, it is the same. I am concerned with national acceptability. â The U.S. would have little international credibility following this approach, as each module would be considered somewhat of a black box, unless licensing could be arranged to obtain the source code. The U.S. will not be in a position of global leadership with this approach. â Similar to Responsiveness, the Credibility would probably track with the alignment of the stakeholder with the agency. â Commercial software is already heavily used abroad. â Transparency could be a problem with a commercial product if the developer protects his product. Presumably, the contract to develop the product would be re-competed every few years. â Alternative #3 already used in a number of international jurisdictions. â My rating here is an average between: Rating of 1 for development costs. Rating of 5 for User's costs. SoundPLAN, for example, has a very steep learning curve, adding burden to its extremely high license fee. H.1.3.7 Cost Implications â All government agencies are used to having zero license fees for current models. I seriously doubt that possibility for this alternative. â Savings on future federal R&D funds is outweighed by high user fees and wasted expenditures on existing projects, such as AEDT. â It will be much more expensive for the user. State DOTs would have to purchase their own, as well as every consultant the does this work for a State DOT. These costs could adversely hinder small, minority or women lead businesses.
H-10 â Substantial individual seat costs associated with these tools render this a non-viable option â Both the recurring costs to the user base and the maintenance of accessibility to results inventories will be costly for both the agencies and commercial vendors. â Cost is higher for the end-user, but the government will save developmental costs. So it depends on your perspective. â Lower costs for DOT, but potentially high costs for users. â Smaller development cost (to the government agencies) offset by higher user costs (purchase of licenses). H.1.4 Datum vs. Alternative #4 Datum vs. Alternative #4 Current Preferred Design with Build Sequence Build on EC IMAGINE Project â I cannot image the "foreign" methodologies surviving in this alternative, however. H.1.4.1 Agency Acceptance â The 'Interim methods' idea is a good approach for the federal agencies to take, but dependency on foreign methods and entities would be hard for the agencies to swallow. â I am not viewing foreign government models much differently than foreign commercial models in terms of acceptance by U.S. agencies. â Use of foreign methods is not desired, and if source code can not be obtained this is a non-starter for the U.S. Government. â If the stakeholder base is US the adoption of some methods and the framework would be difficult in Alt#4. â If federal agencies can put hubris aside, they should recognize the enormous effort which has already been put into the EC approaches and be happy most of the legwork is already complete. â Agencies would be reluctant to accept a European approach â Alternative #4 would require all agencies (and their respective user communities) to accept this tool. â A propagation method that replaces a roadway with intermittent points cannot possibly compute correct roadway noise levels, due to two critical (and "show-stopping") limitations---which, however, are too complex to fit within this spreadsheet cell. H.1.4.2 Technical Feasibility â EC IMAGINE seems to achieve technical harmonization in noise modeling areas that face Datum. â [Alternative #4 less feasible] Since EC projects were only on noise, no air. â It is not clear if a license could be negotiated to obtain the source code and the model does not include any air quality considerations. Not clear on modularity of design, etc. â The IMAGINE end-state noise algorithms are ready and tested (advantage over Datum). Implementation of air quality will be of similar difficulty to that of the Datum.
H-11 â Despite using the results of the IMAGINE research work, Alternative #4 is more ambitious than the Datum approach, and is thus more risky. â Smaller scale projects will need screening tools, which this alternative might incorporate (assumed yes). H.1.4.3 Analytical Proficiency â [Equally proficient] Based on the similar simulation end states. â This would be a regional model, which is not good for FHWA projects. As indicated, single (and more localized) mode projects would need to use that agency's model. Since most of FHWA's projects would fall in the more localized category, the FHWA would need to maintain its TNM to keep it up to par and usable for these situations. â Although simulation-based, does not include air quality considerations. â No option for reduced fidelity modeling excludes very large portion of stakeholders in the user base. â A complex model to use â As a program that is designed from the start to be a comprehensive state-of-the-art simulation tool, Alternative #4 can be expected to provide better results. â Especially better than the Datum's End State. H.1.4.4 Scalability â Retaining existing tools for small projects and reserving multimodal model for large projects is a good idea. â Demonstrated, scalable capability. â Information management system of Alt#4 reduces analyst workload so that larger problems can be more easily executed. â IMAGINE algorithms were built with scalability in mind, and the "reference" and "engineering" models allow for variable detail. â Alternative #4 has the potential for increased scalability due to its 'clean-sheet' design. However, this would require a disciplined development effort to ensure that the scalability potential is realized. â Without knowing the availability of the source code, it is not clear that this alternative is responsive. H.1.4.5 Responsiveness â With modularity and first principles approaches being roughly equivalent between Datum and Alt#4 there do not seem to be any differentiator in Responsiveness. â Getting concurrence for changes from multiple countries could be difficult, but the US does not necessarily need to do everything in parallel with Europe - just use what they've got as a springboard. â Not certain that the source code would be made available. Would require coordination with foreign developers.
H-12 â Alternative #4 is more of a labor-intensive effort, especially in the short term. Because of this, there might be pressure to freeze the design as early as possible, and thus be less responsive to evolving needs. â But this criterion is of no important, in my view. H.1.4.6 International Credibility â Collaboration with EC IMAGINE is a major plus. â Collaboration with European developers should lead to a level of international credibility, but there will always be the question about U.S. domain knowledge working off of a European design. The U.S. will not be in a position of global leadership with this approach. â While underlying algorithms have equivalent credibility internationally, the framework of Alt#4 seems to have better potential. â Obviouslyâ¦ [Alternative #4 would be more credible because of EC IMAGINE collaboration.] â Presumably, a product based on an approved EC model would be credible. â As the product developed from the result of an EC research effort, Alternative #3 should have good international credibility. â Judged at higher cost because not clear what happens to AEDT project. If it continues at current pace, then simulation modeling becomes a new initiative requiring additional funding. Even if AEDT is retooled, some new funding is necessary for the simulation modeling effort that is not currently part of the AEDT development plan. H.1.4.7 Cost Implications â FHWA would still need to fund TNM to keep it operational. â Substantial additional costs needed to reach an end state with this approach -- although multimodal it does not include air quality considerations. â Seems that development costs preference Alt#4 but user/analyst costs slightly favor Datum. â Development costs are now spread across the globe - many users and developers should result in economies of scale. Any European investment will make costs for the US lower, but international bureaucracy will have its costs. â Working with foreign developers would be difficult and time consuming. â The Datum approach will likely have greater overall costs (because of all the intermediate steps). However, Alternative #2 will likely have higher up-front costs, and will likely involve greater risk. My judgment is that on balance, the total overall cost of each approach is about equal. H.1.5 General Comments Getting acceptance from the various Federal agencies will be a monumental task, and will be the life or death of whichever alternative is selected. I am grateful for being apart of this projectâs panel, however, I can only speak from the highway noise side of the equation. â A few general comments: (1) Alternatives 3 and 4 just seem like non-starters to me, given the uncertainty associated with availability of source code; (2) I struggled with rating Alternative 1 - going back and adjusting it several times. I see a lot of the difference being driven by how this model gets funded (assuming it ever does).
H-13 H.2 Round 2 Comments H.2.1 Datum vs. Alternative #1 Datum vs. Alternative #1 Progressive Build Build on AEDT â Pure AEDT much much much worse than Datum, re agency acceptance, for lack of Datum's Builds 1 and 2---that is, post-processor for existing computer programs, and screening tools. And although these items could be added to Alt.1 that would just serve to convert Alt.1 partially into the Datum, itself. Screening tools could either (1) employ lower-precision algorithms, re the existing computer programs, or (2) act only as a pre-processor to feed "simplified GUI/spreadsheet input" directly to the official computer programs for calculation, with results passed to the post processor. I believe that method (2) would gain better agency acceptance, since it will be lower precision only because input is constrained to be simpler...and the various agencies could help determine how "simple" is still adequate when they are not funding the project. What about screening for marine-vessel noise and emissions? I would suggest adding Alt.1's tasks concerning (1) Rulemaking and (2) Data Clearinghouse to the Datum, as well. I'd also suggest not "combining" sound levels from differing modes, since noise impact differs significantly by mode for the same sound level. Decibel-adding sound levels into one composite noise-contour set ignore this inherent, subjective difference. H.2.1 Agency Acceptance â With the active role of agencies in the progressive build, Datum is clearly superior. â Starting from the existing tools will help with this [Datum]. â Some like better, others not as well, so this [Datum vs. Alt#1] is the same. â This depends upon which agency, but I've answered it within the context of DOT as a whole, whereas it [Alt#1] would be 2 if FAA-centric. â The simulation model end state should be the long term goal. However, using the performance criteria description from page 16 of the first round scorebook, it is technically more challenging. H.2.2 Technical Feasibility â Not clear to me that it [Alt#1] is more or less feasible. â Need to think this through a bit more as the detailed design evolves, but currently both seem equally feasible. â Pure AEDT would likely be more technically feasible than the Datum, because the Datum's simulation development will be difficult for ground sources (huge numbers of individual vehicles). Nevertheless, I like very much the scaled-down simulation effort now in the Datum. Other comments: o If we force the same noise-propagation algorithms onto every mode, we will either (1) greatly sacrifice the required precision for noise barriers along highways, or (2) burden aviation computations with algorithms grossly more complicated than needed. o I'd suggest dropping this self-imposed requirement. Including refraction via ray tracing will upset/undermine highway-noise propagation. Instead, I'd strongly recommend transforming the vertical geometry of each transept computation to account for refraction.
H-14 o To increase technical feasibility of the Ground-Source screening tool in the Datum, I'd suggest digitized USGS maps as input, with automatic Terrain-Line generation, parallel to roadways and rail lines, at specified offset intervals from the road/rail. â This [Alt#1] is proven and development has been worked out. â Both designs will use AEDT framework for some time, but Datum is better because of the screening tools and the introduction of a multimodal model sooner. H.2.3 Analytical Proficiency â New datum goes farther towards simulation. â The datum draws the best from the best, while balancing the needs of the agencies. It may not be the most efficient way of getting to the end state, but it provides more analytical proficiency quicker. â Both will provide acceptable answers. AEDT sooner but the Datum could be better in the long term so again, [Datum and Alt#1 are] the same. â Again, the simulation end state is a better long term goal. â AEDT aspects are about the same for both alternatives, but the Datum's simulation (with sufficient choice of input precision) provides much finer-tuned scalability. H.2.4 Scalability â In addition, scalability might be improved in the Datum by utilizing a common terrain input method (digital USGS and/or digital highway-project contour maps), but allowing the user to determine the precision needed when approximating that digital terrain during computations. In highway-barrier design, the user could then choose very precise vertical and horizontal precision when very close to the highway, but lesser precision at larger distances. â By design the datum is more robust and scalable. â Datum includes screening tools and will eventually lead to simulation model; which significantly better than Alt#1. â Starting fresh gives the edge to the Datum. â Pure AEDT much much much worse much than Datum, re responsiveness, for lack of Datum's Builds 1 and 2---that is, post-processor for existing computer programs, and screening tools. H.2.5 Responsiveness â Datum includes screening tools and will eventually lead to simulation model; which significantly better than Alt#1. â [Datum is more responsive] Since the datum appears to get to a multimodal capability sooner. â As the datum draws the best from the best it is by definition more responsive. â This really depends on the timeframe of consideration. AEDT is better in the short and mid term while the Datum, if all goes right, is better in the long term. So [Datum and Alt#1 are] the same. â Pure AEDT ultimately might have less international credibility for lack of a simulation end state. H.2.6 International Credibility â Drawing form EC IMAGINE gives the advantage to Datum.
H-15 â Beyond the aviation sector, it's not clear how important this is (hence the low weighting), but inclusion of the best from the canned international packages makes the datum better. â [Datum more credible] Since the datum starts from models that are more widely known â AEDT has already been out there. Nothing to prove as the Datum would require. â Hard to determine which will cost more/less. H.2.7 Cost Implications â Time will tell on this. It is very difficult to judge. If all the funding agencies are playing well together, there will be buy-in and the datum should cost less. However, there is risk that in an effort to keep "everyone" happy, the datum could cost more. Hence the neutral stance. â Pure AEDT would cost less for lack of simulation. However, I do not believe that the resulting Value Score of 0.93 reflects my overall comparison of the two alternatives. If I were asked to provide an intuitive Value Score of Alt.1 re the Datum, I would estimate it to be somewhere around 0.5. Numerically, instead of dividing 1.85 by 2 to get the Value Score, I would like to divide 1.4 by 2.8, to get 0.5. That would require finer-grained options than were available for scoring---options such as: o Much Much Much worse, and o Only Very Slightly Lower. â Costs are similar through the Build on AEDT stages, but development of a simulation model makes Datum more expensive. â A lot of effort has gone into the AEDT development. So, many problems have already been worked out. This is why I think the Datum should begin with the AEDT not just take from. H.2.8 General Comments â On the impacts since, the air quality and climate modeling we do in APMT already is multimodal (this is required since atmospheric chemistry is non-linear; we need to put all inventories in before doing a calculation in CMAQ for example). I don't think you need to say anything about this, but just wanted to make sure you were aware that the impacts models for emissions are already set up to take full inventories (e.g. from NEI). Then they apply standard time fluctuations to them for different sources (e.g. daily and weekly cycles). â When I think about things like the current CAEP NOx stringency analysis, I see a string of modeling tools starting with an FESG demand forecast (derived from a consultative process with input from Airbus and Boeing market projection models), going to one industry economic model (APMT-Economics), going to multiple international noise and emissions inventory estimators (AEDT, AERO2K), ending with one environmental impacts modeling effort (APMT-Impacts). Within this string, the two most influential model aspects in terms of changing the answer are arguably the demand forecast/economic modeling, and the environmental impacts modeling (all the inventory estimators give about the same answer plus/minus 10%). Like this modeling string, the plans described for Project 02-09, are heavy on inventory estimation, and light on the other two ends of the spectrum. So for example, getting movements data for aircraft, trains, autos, long- haul trucking, etc. that are consistent with a single economic forecast will be a challenging task (let alone how those movements and that forecast are influenced by policies) -- and different growth assumptions are likely to be very influential. But economics modeling and scenario generation are not discussed. Conversely, on the impacts side life may be a little easier because the air quality and climate models already incorporate all the emissions, but there are other influential aspects not given much attention in the write-up -- e.g. the extent to which the fuel-
H-16 side (especially alternative fuels, life cycle GHGs, etc.) will be addressed. Since we now have a low carbon fuel standard signed by Obama, I think we need to look ahead to how integrated environmental impact modeling might be done 5 years from now (e.g., will it include agricultural models or energy grid models?). â This [a build sequence is predicated on giving the users and agencies the tool that they need within expandable system architecture] is the idea that I really wanted to keep. But starting with individual models with a combined GUI delays this resulting in more cost and time. The related issues and suggestions are as follows: o In some cases screening tools donât exist. Where they do they are often quite different. For example, FTAâs screening tool is essentially a distance from the tracks. But in an intermodal plan, this doesnât really work since we should be looking at the combined effects of all modes in answers we provide. So the screening tool represents a significant amount of work to build unless it is not truly inter-modal. o It appears now that both Build 1 and Build 2 are to use existing models and then improve this approach in Build 2. Then we throw this out for Build 3 and start with AEDT expansion? o So why could we not just start by bringing the other models into the AEDT system by changing the AEDT GUI? In this way we don't throw away previous work but can still answer the questions you pose in the operational track and at the same time get a head start on the overall modal model. This advances us to Build 4 very rapidly.